International Finance

TUGAS KULIAH TEORI MANAJEMEN KEUANGAN INTERNATIONAL FINANCE

KELOMPOK 4 1. Anita Setianingsih

C2C021037

2. Diah Arum

C2C021043

3. Arfella Dara Tristantia

C2C021044

4. Jaja Abdul Jalil

C2C021045

5. Eko Harinatalistini

C2C021051

KELAS WEEKEND ( C )

PROGRAM STUDI MAGISTER MANAJEMEN PROGRAM PASCA SARJANA FAKULTAS EKONOMI DAN BISNIS UNIVERSITAS JENDERAL SOEDIRMAN 2021 1

INTERNATIONAL FINANCE (International Capital Asset Pricing Model, International Diversification Portofolio, Law of One Price) Oleh Kelompok 4

I. TEORI PENGANTAR Banyak perusahaan yang dalam menjalankan operasi berhubungan dengan pihak asing (luar negeri). Perusahaan mungkin menjual hasil produksi ke luar negeri (Ekspor), membeli bahan baku dari luar negri (impor), menggunakan dana dari luar negeri (kredit), atau bahkan melakukan penanaman modal di luar negeri (investasi). Dengan demikian perusahaan akan terbuka (expose to) terhadap risiko valuta asing, risiko tingkat bunga, dan bahan risiko politik atau negara. Oleh karena itu perlu mempelajari tentang International Finance.

A. PENGERTIAN Keuangan Internasional merupakan aktivitas bisnis, yang juga merupakan disiplin ilmu.

B. PENTINGNYA KEUANGAN INTERNASIONAL Keuangan Internasional itu penting untuk mengetahui perkembangan : 1. Ekspansi perusahaan multinasional (MNC) ke Negara-negara sedang berkembang (NSB) 2. Expansi ideology globalisasi, dan 3. Perdagangan Internasional (expor-impor). Para pemikir ekonomi liberal menyatakan bahwa expansi MNC ke Negara2 NSB merupakan Lokomotif pembangunan di NSB oleh sebab itu kehadirannya sangat diharapkan. MNC itu penting dipromosikan ideologi globalisme, tanpa MNC tidak akan ada pembangunan di NSB karena mereka perusahaan lokal kekurangan modal, ilmu, teknologi dan tenaga ahli.

C. FUNGSI KEUANGAN INTERNASIONAL Pada kegiatan bisnis dalam Era Pasar terbuka dan Pasar Keuangan yang terintergrasi fungsi keuangan internasional semakin terlihat, fungsi-fungsi tersebut antara lain sebagai: 1. Lembaga pembiayaan 2

2. Investasi 3. Sarana lindung nilai bahkan ajang spekulasi. (Lindung nilai adalah suatu strategi yang bertujuan mengurangi dampak risiko bisnis yang tidak terduga, di samping tetap memungkinkan untuk memperoleh keuntungan dari hasil investasi tersebut. Lindung nilai juga dilakukan terhadap mata uang dan dilakukan oleh para investor guna melindungi investasinya di negara lain.) Contoh lindung nilai: 1. Kontrak serah dan kontrak berjangka adalah salah satu bentuk lindung nilai untuk mencegah risiko pergeseran harga pasar. 2. Untuk mengurangi risiko kurs dilakukan lindung nilai dengan membuat kontrak forward untuk bentuk ini disebut kontrak forward yang dinamakan juga posisi beli (buy) dollar forward atau long forward dolar. 3. Kontrak future adalah perusahaan menyetujui harga barang yang akan dibeli pada tahun 2018 untuk dijual pada tahun 2019.

Era global yang berdampak terhadap perdagangan dan investasi, menyebabkan antara lain proses pembayaran melintasi batas Negara. Implikasi terhadap institusi keuangan semakin signifikan.

D. STRUKTUR Teori keuangan internasional sebagian besar mencerminkan topik dari teori keuangan domestik. Sebagai aktivitas bisnis, keuangan international merupakan multi billion business baik melalui sarana pasar uang (Maturity kurang dari 1 tahun), sarana pasar modal (Maturity lebih dari 1 tahun baik fixed income markets maupun equity markets) serta penanaman modal asing langsung ( FDI ).

E. Kedudukan Keuangan Internasional dalam Ilmu Hukum Dari sudut pandang hukum, keuangan internasional merupakan bagian dari hukum ekonomi, yang bersifat transnasional, dalam arti lingkup kegiatannya melewati batas-batas Negara, sehingga dalam mengkaji keuangan internasional, hukum Negara-negara terkait tidak dapat diabaikan. Terdapat interdependency yang kuat yang menyangkut berbagai kepentingan antar negara. 3

Selain itu, keuangan internasional harus dipandang dengan metode pendekatan yang multidisipliner, yakni melibatkan berbagai disiplin ilmu seperti ekonomi, sosial dan politik, disamping pendekatan hukum. Pendekatan hukumpun melibatkan tidak saja bidang hukum perdata, melainkan juga hukum pidana, hukum administrasi Negara, hukum lingkungan dan lain lain.

F. Ruang Linkup dan Fungsi Keuangan Internasional Ruang lingkup keuangan internasional berkembang pesat sejalan dengan kebutuhan manusia akan sumber-sumber pembiayaan, kemajuan teknologi, bahkan umber daya manusia yang menghasilkan produk-produk keuangan yang beragam. Oleh karena itu, keuangan internasional secara tidak sadar telah menciptakan hukum-hokum baru, baik melalui kebiasaan dalam praktek internasional, maupun yang secara sadar harus dibentuk oleh Negara-negara untuk mengatur masalah keuangan ini. Globalisasi melalui GATT, WTO dan APEC telah membuat dunia menjadi seamless (tanpa sambungan) dan borderless (tanpa batas). Kemajuan teknology telah membuat dunia menjadi lebih sempit. Perbedaan waktu membuat pasar berlangsung selama 24 jam, uang menjadi produktif selama manusia tertidur (overnight rate). Secara umum keuangan internasional bersumber pada: 1) Pasar Uang; yang meliputi aktifitas antara lain Foreign exchanges market, Currency Options, Currency Futures and Swaps 2) Portofolio investasi melalui Pasar Modal; meliputi Bond Investment dan Equity Investment serta 3) Foreign Direct Investment.

G. Keuangan Internasional dan Biaya Pembangunan Pembiayaan Pembangunan Indonesia pada dasarnya dilakukan melalui berbagai sumber yakni: 1. Sumber CGI ( dulu IGGI/ Inter-Governmental Group Indonesia) Forum ini mempunyai 30 anggota baik anggota bilateral maupun multilateral. Bertemu setiap tahun guna merumuskan tingkat bantuan para anggauta kepada Indonesia, dibawah koordinasi Bank Dunia (IBRD). Selain bantuan berbentuk pinjaman, ada pula 4

yang berbentuk hibah. Hibah umumnya diberikan dalam bentuk bantuan teknis guna peningkatan kapasitas (capacity building) baik sektoral maupun regional. 2. Sumber Non CGI. Baik secara bilateral (Brunei, Saudi Arabia misalnya) maupun multilateral. Keuangan Internasional dan Biaya Pembangunan 3. Sumber dana komersil. Diperoleh dari berbagai bank komersil dari seluruh dunia dalam bentuk : a. Pinjaman sindikasi: Syndicated Loan b. Pinjaman jumbo c. Pinjaman siaga d. Obligasi e. X Credit Dalam memanfaatkan sumber dana dipasar keuangan international penting diketahui salah satu elemen penting yang menjamin suksesnya kegiatan mobilisasi dana dimaksud yakni adanya Good Governance. Pasar internasional terbiasa dengan prinsip dasar RAFT yakni Responsibility, Accountability, Fairness dan Transparency. Terlaksananya prinsip dimaksud merupakan nilai tambah bagi Issuer/Emiten.

II. ICAPM ( International Capital Asset Pricing Model) A. DEFINISI • ICAPM ( International Capital Asset Pricing Model) adalah perluasan dari CAPM (Capital Asset Pricing Model) • CAPM menggunakan portopolio pasar domestic • ICAPM menggunkan portopolio pasar international ICAPM dikembangkan oleh Solnik (1974) dan De Santise dan Bruno (1997) dan barubaru ini oleh Nguyen et al. (2017). Perluasan CAPM domestik hanya dapat dibenarkan dengan tambahan dua asumsi yang tidak rasional : -

Investor di seluruh dunia memiliki pola konsumsi yang identik.

5

-

Harga riil barang konsumsi identik di setiap negara. Dengan kata lain, paritas daya beli berlaku tepat pada setiap titik waktu.

Dengan penilaian langsung, nilai riil adalah nilai nominal dikalikan rasio tingkat harga luar negeri terhadap tingkat harga domestik. X= 𝑺 × [(𝑷𝑭𝑪 /𝑷𝑫𝑪 ) di mana: X adalah nilai tukar riil 𝑆 adalah nilai tukar nominal 𝑃𝐹𝐶 adalah tingkat harga negara asing. 𝑃𝐷𝐶 adalah tingkat harga negara domestik

Nilai tukar riil berubah dalam suatu periode jika apresiasi valuta asing selama periode tersebut tidak sama dengan perbedaan inflasi antara kedua negara selama periode tersebut. • ICAPM dikembangkan dengan asumsi bahwa warga suatu negara peduli dengan pengembalian dan risiko yang diukur dalam mata uang negara mereka. • Semua asumsi CAPM masih berlaku. • Dalam ICAPM, seperti dalam CAPM domestik, semua investor menentukan permintaan mereka untuk setiap aset dengan optimasi varians rata-rata menggunakan mata uang domestik mereka sebagai mata uang dasar.

B. FOREIGN CURRENCY RISK PREMIUM (SRP) Premi risiko mata uang asing (SRP) didefinisikan sebagai pengembalian yang diharapkan atas investasi dikurangi tingkat bebas risiko mata uang domestik. SRP= 𝐸 [

𝑆1 −𝑆0 𝑆0

] − (𝑟𝐷𝐶 - 𝑟𝐹𝐶 )

a) Foreign Currency Risk Premium (SRP)-Example Tingkat bunga bebas risiko satu tahun adalah 6 persen di DC dan 3 persen di FC. Apresiasi nilai tukar yang diharapkan dari FC adalah 4 persen. Berapa premi risiko mata uang asing? -

Penyelesaian:

-

SRP= 𝐸 [

𝑆1 −𝑆0 𝑆0

] − (𝑟𝐷𝐶 - 𝑟𝐹𝐶 )

= 4% −(6% − 3%) 6

= 1% C. ICAPM: Hubungan Penetapan Harga- Risiko Penetapan harga risiko untuk ICAPM adalah bahwa pengembalian yang diharapkan atas aset i adalah jumlah dari tingkat bebas risiko ditambah premi risiko pasar ditambah berbagai premi risiko mata uang: E(𝑅𝑖 ) = 𝑅0 + ß𝑖𝑤 × 𝑅𝑃𝑊 + ϒ𝑖1 × 𝑆𝑅𝑃1 + … + ϒ𝑖𝑘 × 𝑆𝑅𝑃𝑘 di mana adalah eksposur pasar dunia dari aset dan adalah eksposur mata uang, atau sensitivitas, dari pengembalian aset ke berbagai nilai tukar (1 hingga k). 𝑅𝑃𝑊 adalah premi risiko pasar dunia dan 𝑆𝑅𝑃𝑘 adalah premi risiko mata uang. D. ICAPM: Kesimpulan Penetapan Harga Risiko Dengan satu mata uang asing, persamaan harga aset ICAPM disederhanakan menjadi: E(𝑹𝒊 ) = 𝑹𝟎 + ß𝒊𝒘 × 𝑹𝑷𝑾 + ϒ𝒊 × 𝑺𝑹𝑷𝑭𝑪

7

III. PORTOFOLIO DIVERSIFIKASI INTERNASIONAL (IDP) ▪

Portofolio Diversifikasi Internasional adalah sumber dari jenis baru kesejahteraan dunia yang diperoleh dari hubungan ekonomi internasional, berbeda dari "keuntungan dari perdagangan" tradisional dan dalam peningkatan produktivitas yang mengalir dari migrasi faktor-faktor produksi.

▪

model teoritis menunjukkan bahwa pergerakan modal internasional merupakan fungsi tidak hanya dari perbedaan suku bunga, tetapi juga dari tingkat pertumbuhan kepemilikan aset total di dua negara

▪

analisis tersebut memiliki beberapa implikasi kebijakan yang penting di dunia yang sedang berkembang di mana kebijakan moneter dan fiskal disesuaikan untuk mencapai keseimbangan internal dan eksternal. (Herbert G. Grubel)

▪

Pengurangan risiko yang lebih besar dapat dicapai dengan mendiversifikasi portofolio secara internasional.

▪

Beberapa penulis menunjukkan bahwa pergerakan harga saham di berbagai Negaranegara hampir tidak berhubungan: a. Perubahan harga pada Bursa Paris tampak independen dari harga saham fluktuasi di bursa London, dan sebagainya. b. Ketika sekuritas dari satu negara (katakanlah AS) adalah Melakukan lebih buruk dari yang diharapkan, pasar lain adalah kemungkinan akan melakukan yang lebih baik, maka mengimbangi Kerugian.

8

IV. LAW OF ONE PRICE (LOOP) A. Pengertian LOOP LOOP adalah singkatan dari Law of One Price atau dalam Indonesianya disebut Hukum Satu Harga. LOOP merupakan teori ekonomi yang menyatakan bahwa harga dari aset atau komdoditas identik pada pasar yang berbeda harus sama setelah mempertimbangkan kurs atau jika harga dinyatakan dalam mata uang yang sama.

B. ASUMSI-ASUMSI TEORI LOOP Untuk mendukung teori LOOP terdapat asumsi asumsi seperti: a. Pasar harus kompetitif, b. Tidak adanya pembatasan perdagangan, c. Fleksibilitas harga, pembeli dan penjual mempunyai akses informasi yang sama dengan biaya yang rendah. Hukum satu harga ada karena perbedaan antara harga aset di lokasi yang berbeda pada akhirnya akan diperbaiki dan dihilangkan oleh peluang arbitrase. Peluang arbitrase muncul karena pedagang dapat membeli barang di pasar dengan harga lebih rendah dan segera menjualnya di pasar lain dengan harga lebih tinggi untuk mendapatkan laba tinggi. teori menyatakan bahwa selanjutnya, kekuatan penawaran dan permintaan akan menyatukan harga di seluruh pasar dan oleh karena itu peluan arbitrase akan dihilangkan. Contohnya. Jika harga sekuritas di pasar A adalah $10, tetapi dijual dengan harga yang berbeda di pasar B yaitu $20, maka investor akan membeli sekuritas pasar A dan menjualnya di pasar B dan mendapat laba sebesar$10 tanpa risiko. Karena sekuritas di pasar A lebih murah, maka permintaan pun meningkat, sesuai dengan hukum permintaan dan penawaran, maka harga sekuritas di pasar A pun ikut meningkat. Hal ini juga yang terjadi di pasar B dimana penjualan atau penawaran sekuritas yang meningkat akan menyebabkan penurunan harga.

C. PELANGGARAN LOOP DI DUNIA NYATA LOOP memperhitungkan pasar tanpa gesekan dimana tidak ada biaya transaksi, biaya transportasi, atau batasan hukum, kesamaan nilai tukar mata uang dan tidak ada

manipulasi harga oleh pembeli atau penjual. Berikut adalah asumsi yang sering tidak berlaku atau dilanggar di LOOP: 1. Biaya transportasi Dalam transaksi komoditas atau barang fisik apapun, biaya pengangkutan harus dimasukkan sehingga harga komoditas akan berbeda di lokasi geografis yang juga berbeda. 2. Biaya transaksi Biaya transaksi seperti biaya menemukan rekan penjual atau biaya negosiasi mengenai kontrak tentu dapat bervariasi di berbagai pasar dan wilayah geografis sehingga harga barang komoditas pun akan berbedaPembatasan hukum 3. Hambatan hukum berupa tarif, kontrol modal, pembatasan upah imigrasi akan menyebabkan perbedaan harga. Misalnya, suatu negara mengenakan taris impor karet, maka harga karet domestik akan cenderung tinggi dibanding harga internasional. 4. Struktur pasar karena jumlah dan kemampuan pembeli maupun penjual untuk memasui pasar dapat bervariasi, maka konsentrasi pasar dalam menetapkan harga juga dapat bervariasi.

D. LOOP DI PASAR KEUANGAN LOOP umumnya berlaku di pasar keuangan. para ekonom percaya LOOP lebih kuat di terapkan di pasar keuangan daripada perdagangan internasional karena ada lebih sedikit hambatan perdagangan potensial di pasar keuangan sebelumnya. Dalam pasar keuangan, implikasi utama LOOP yaitu keamanan finansial harus datang dengan satu harga terlepas dari bagaimana keamanan itu dibuat. Misalnya opsi panggilan dapat direplikasi menggunakan saham dan obligasi. LOOP menyatakan bahwa harga opsi panggilan harus sama dengan harga portofolio yang direplikasi.. Komoditas tetap menjadi contoh paling menonjol dari LOOP di pasar keuangan. Komoditas di perdagangkan di berbagai pasar dunia menggunakan berbagai instrumen keuangan. Biasanya berjangka dan tidak. Meskipun demikian, haarga komoditas biasanya homogen di seluruh pasar karena LOOP.

E. PENERAPAN LOOP Dalam ekonomi internasional, dijelaskan bahwa terdapat hubungan timbal balik antara suku bunga, kurs, valuta asing, inflasi, dan premium atau diskon dari kurs forward. Hubungan-hubungan tersebut diantaranya adalah purchasing power parity (PPP), interest rate parity (IRP) dan international fisher effect (IFE) yang merupakan konsep yang membentuk LOOP. a) Purchasing Power Parity (PPP) – Paritas Daya Beli ▪ Secara teoritis, PPP merupakan model jangka panjang yang menentukan tingkat ekuilibrium nilai tukar. Hubungan ini memungkinkan terjadinya deviasi dalam jangka pendek. PPP menghubungkan kurs valuta asing dengan komoditas dalam mata uang domestik di pasar internasional. Hubungan antara kurs dengan harga komoditi adalah negatif, sehinga kurs akan cenderung mengalami penurunan dalam proporsi yag sama dengan laju kenaikan harga komoditas secara relatif asumsi yang mendasari teori PPP adalah bahwa pasar komoditas merupakan pasar yang efisien baik sisi lokasi, operasional, penentuan harga, dan informasi. ▪ Konsep PPP dibagi menjadi dua versi yaitu: 1. Versi relatif mengatakan bahwa tingkat kurs mata uang domestik dengan mata uang asing harus disesuaikan sesuai dengan perubahan-perubahan tingkat harga dari kedua negara. Perubahan tingkat harga ini dapatdilihat dari tingkat inflasi dari masing-masing negara atau dilihat dari Indeks HargaKonsumen (Consumer Price Index) dari masing-masing negara. 2. Versi Absolut, menjelaskan bahwa tingkat harga diseluruh dunia akan sama apabila menggunakan mata uang yang umum. Dengan kata lain dapat dinyatakan bahwa satu unit mata uang domestik harus mempunyai daya beli yang sama di seluruh dunia. Versiabsolut ini tidak memperhatikan atau menyampingkan dampak dari biaya transportasidalam perdagangan bebas, tarif, quota dan segala jenis pembatasan (ristriksi) dandiferensiasi produk. ▪ Pendekatan PPP menganggap bahwa komoditas tertentu cenderung untuk menerapkan hukum satu harga atau LOOP. Hal ini dimungkinakan karena

dengan tidak adanya hambatan dan biaya transportasi, maka harga komoditas atau jasa cenderung akan sama di setiap pasar. Permasalahannya adalah apabila pasar terletak di negara yang berbeda. Perbedaan negara akan mengakibatkan harga jomoditas atau jasa dinyatakan dalam mata uang masing-masing negara, walaupun

sebenarnya

“harga”nya

masih

tetap

sama.

Perbedaan

ini

membutuhkan konversi satu mata uang ke mata uang yang lain. ▪ Nilai kurs yang digunakan dalam PPP adalah nilai kurs spot (S) dimana kurs harga domestik dibandingkan dengan harga luar negeri. Jika LOOP dipegang untuk setiap komoditas, maka PPP secara otomatis akan terjadi selama referensi mengenai tingkat harga antar negara adalah sama. Namun ketika LOOP tidak dapat diterapkan dalam setiap komoditas, maka dalam konteks PPP, harga dan nilai tukar seharusnya tidak menyimpang terlalu jauh dengan harga yang diprediksikan dalam PPP. Bila harga komoditas di suatu negara lebih mahal dibandingkan negara lainnya, maka permintaan akan mata uang dan komoditas akan turun. Kondisi ini mendorong nilai tukar dan harga domestik untuk kembali sama dengan nilai PPP. Sebaliknya bila harga domestik lebih murah, maka akan terjadi apresiasi dan inflasi. Dalam konsep PPP, apabila LOOP tidak terjadi maka terdapat kekuatan ekonomi yang akan mendorong menyamakan PPP di semua negara b) Interest Rate Parity (IRP) – Paritas Tingkat Bunga Investasi yang dilakukan di dalam negeri tidak akan terganggu oleh fluktuasi nilai mata uang. Berbeda dengan investasi yang dilakukan di luar negeri yang dapat memunculkan masalah apabila nilai mata uang tersebut berubah. Dalam PPP kondisi tersebut akan berlaku di pasar barang. Sedangkan di pasar sekuritas akan muncul IRP. Pada dasarnya IRP menjelaskan bahwa tingkat return investasi dari suatu mata uang atau biaya pinjaman dalam suatu mata uang asing dengan tingkat bunga rendah akan mengalami forward premium jika dibandingkan dengan mata uang negara lain dengan tingkat bunga yang tinggi. Dalam pasar yang efisien dengan asumsi tanpa biaya transaksi maka tingkat bunga aktual akan kurang lebih sama besarnya dengan Kurs Forward.

Jika memang terjadi seperti yang digambarkan diatas tadi, terjadilah apa yang dikatakan dengan Interest Rate Parity. Ada 2 jenis Interest Rate Parity : 1) Apabila perbedaan tingkat bunga domestik dengan tingkat hedging asing (the hedged foreign rate) adalah nol maka keadaan ini disebut dengan Covered Interest Differetial. 2) Apabila hasil dari covered interest differential ≠ 0 akan terjadi arbitrage incentive yang menyebabkan uang akan bergerak dari suatu negara ke negara lain. Hal ini disebut dengan Covered Interest Arbitrage Jadi apabila ada selisih suku bunga dalam negeri dibandingkan suku bunga luar negeri yang disertai dengan forward premium atau forward discount yang tidak sama besarnya dengan selisih suku bunga tersebut maka akan terjadi arbitrage internasional seperti yang digambarkan pada uraian diatas. IRP menjelaskan kepada semua orang bahwa seharusnya selisih suku bunga luar negeri dengan suku bunga dalam negeri besarnya harus sama dengan forward discount atau forward premium. Jika ini terjadi berarti terjadi keseimbangan yang letaknya tepat di garis paritas (parity line). Apabila terdapat kasus-kasus yang menyimpang yang berarti bahwa titik-titik koordinat tersebut berada diluar parity line, akan terjadi arbitrage yang dapat berwujud arbitrage dana masuk kedalam suatu negara atau sebaliknya dana mengalir keluar negeri. Kesimpulan dari semua uraian tentang paritas tingkat bunga ini adalah sebagai berikut : Tingkat bunga yang tinggi dari suatu mata uang akan diimbangi (offset) dengan forward discount dan tingkat bunga yang rendah dari suatu mata uang akan diimbangi dengan forward premium. c) International Fisher Effect (IFE) Untuk memahami dampak dari perubahan-perubahan relatif dari tingkat bunga nominal antar negara terhadap nilai tukar mata uang nominal adalah dengan mempelajari kembali implikasi dari PPP dan Fisher Effect. Tingkat binga yang digunakan dalam transaksi finansial adalah tingkat bunga nominal oleh karena itu tingkat bunga nominal harus disesuaikan dengan ekspektasi inflasi di

masa depan. FE menyatakan bahwa mata uang dengan tingkat inflasi yang tinggi akan menyebabkan tingkat bunga Pada penelitian (Martoatmodjo, 2001), hukum satu harga atau LOOP tidak seluruhnya mutlak dapat diterapkan. Hukum ini bahkan untuk negara-negara maju sekalipun tidak mutlak berlaku serta merta. Hanya pada kondisi-kondisi tertentu hukum ini berlaku. Tetapi pada kondisi yang lain, hukum ini tidak berlaku secara relatif.

References Black, A. C. (2006). Dictionary of Economics Over 3,000 Terms Clearly Defined. London: A & C Black Publisher Ltd. Hady, H. (2004). Analisis Faktor-Faktor Determinasi Kurs Rupiah Berdasarkan Pendekatan Moneter. Jurnal Bisnis Strategi Vo. 13. Martoatmodjo, S. (2001). Penerapan Hukum Satu Harga (Law of One Price) Dalam Arbitrage Internasional. Ekuitas Vol. 5 No. 3, 238-260. Rahutami, A. I. (2011). Purchasing Power Parity: Teori dan Perkembangan Model Empris. https://www.investopedia.com/terms/l/law-one-price.asp https://corporatefinanceinstitute.com/resources/knowledge/economics/law-of-one-price-loop/

V. PORTOFOLIO DIVERSIFIKASI INTERNASIONAL (IDP) ▪

Portofolio Diversifikasi Internasional adalah sumber dari jenis baru kesejahteraan dunia yang diperoleh dari hubungan ekonomi internasional, berbeda dari "keuntungan dari perdagangan" tradisional dan dalam peningkatan produktivitas yang mengalir dari migrasi faktor-faktor produksi.

▪

model teoritis menunjukkan bahwa pergerakan modal internasional merupakan fungsi tidak hanya dari perbedaan suku bunga, tetapi juga dari tingkat pertumbuhan kepemilikan aset total di dua negara

▪

analisis tersebut memiliki beberapa implikasi kebijakan yang penting di dunia yang sedang berkembang di mana kebijakan moneter dan fiskal disesuaikan untuk mencapai keseimbangan internal dan eksternal. (Herbert G. Grubel)

▪

Pengurangan risiko yang lebih besar dapat dicapai dengan mendiversifikasi portofolio secara internasional.

▪

Beberapa penulis menunjukkan bahwa pergerakan harga saham di berbagai Negaranegara hampir tidak berhubungan: c. Perubahan harga pada Bursa Paris tampak independen dari harga saham fluktuasi di bursa London, dan sebagainya. d. Ketika sekuritas dari satu negara (katakanlah AS) adalah Melakukan lebih buruk dari yang diharapkan, pasar lain adalah kemungkinan akan melakukan yang lebih baik, maka mengimbangi Kerugian.

Tambahan: Diversifikasi adalah sebuah strategi investasi dengan menempatkan dana dalam berbagai instrument investasi dengan tingkat risiko dan potensi keuntungan yang berbeda, bertujuan untuk mengurangi tingkat risiko dan tetap memberikan potensi tingkat keuntungan yang cukup atau strategi ini biasa disebut dengan alokasi aset (asset allocation). Alokasi aset ini lebih fokus terhadap penempatan dana di berbagai instrumen investasi. Bukan menfokuskan terhadap pilihan saham dalam portofolio. Contoh senderhana dari diversifikasi international ini adalah seorang investor yang memiliki suatu portofolio investasi yang terdiri atas kombinasi

dari dua saham perusahaan yang listing di Bursa Efek Jakarta, Surat Berharga Bank Indonesia, tiga saham yang listing di Bursa Straits Time Singapura, dua saham yang listing di Bursa Nikkei Jepang, dan tiga saham yang listing di London Stock Exchange. Dalam melakukan diversifikasi, karakter instrumen investasi yang harus dipertimbangkan, yaitu: 1. Potensi tingkat pengembalian (return), Contoh antara investasi pada saham dengan deposito. Umumnya, saham memberikan tingkat pengembalian atau return yang lebih tinggi daripada deposito 2. Risiko, Risiko untuk berinvestasi pada saham cenderung lebih besar karena fluktuasi atau perubahan harga saham lebih tinggi sehingga dapat menyebabkan peluang untuk mengalami kerugian menjadi lebih tinggi daripada berinvestasi di deposito.

3. Likuiditas. Maksudnya adalah kemudahan untuk membeli dan menjual sebuah instrumen investasi, jika berinvestasi di deposito, kita tidak dapat menguangkan investasi tersebut sewaktu- waktu karena deposito memiliki masa jatuh tempo. Sedangkan jika berinvestasi di saham, kita dapat dengan mudahmenjualnya sesuai dengan keinginan kita.

Teori portofolio modern menjelaskan bahwa diversifikasi dapat mengurangi risiko portofolio yaitu dengan tidak melakukan investasi pada asset-aset yang berkorelasi sempurna. Mansourfar et al. (2010) menjelaskan ada beberapa manfaat dari portofolio internasional yaitu: 1) Besarnya persentase dari modal. 2) Investasi dalam saham asing, dimana investor akan memperoleh keuntungan akibat meningkatnya expected return. 3) Menurunnya variasi return. 4) Rendahnya korelasi return saham asing dengan saham domestik. Lessard (1976) menyatakan bahwa ada 3 (tiga) hal penting yang membedakan pasar internasional dengan pasar domestik yaitu: 1. Covariance antar aset pasar domestik lebih tinggi dibandingkan covariance antar aset pada pasar internasional.

2.Adanya berbagai macam biaya yang harus dikeluarkan seperti pajak yang tinggi, hal ini merupakan hambatan investasi di pasar internasional. Selain itu, adanya kontrol mata uang domestik dan tradisi investor pasar nasional yang tersegmentasi sehingga dapat menyebabkan harga aset domestik lebih tinggi dari harga internasional. 3. Nilai tukar mata uang antar negara berbeda sehingga dapat menimbulkan risiko mata uang pada portofolio internasional. Mengapa harus diversifikasi : Karena adanya perkembangan reksadana saham domestik dan reksadana saham internasional yang berbeda dan perkembangan bursa saham antara negara satu dengan lainnya tidak sama menunjukkan bahwa adanya peluang bagi investor untuk melakukan diversifikasi, baik dipasar modal domestik maupun internasional. Teori Markowitz berhasil membuktikan kalau risiko portofolio dapat menjadi minimum jika kedua aset itu mempunyai koefisien korelasi negatif sempurna yaitu -1. Markowitz juga menemukan bahwa diversifikasi selalu dapat menurunkan risiko portofolio sepanjang koefisien korelasi tidak positif sempurna atau lebih kecil dari satu. Markowitz (1952) membedakan risiko menjadi 2 (dua) yaitu; risiko sistematis (systematic risk) dan risiko tidak sistematis (unsystematic risk). Risiko sistematis atau risiko pasar merupakan risiko yang timbul akibat dari kejadian- kejadian di luar perusahaan seperti: perubahan sistem pemerintahan, bencana alam, perubahan ekonomi, politik, hukum, sosial, budaya dan teknologi. Risiko ini tidak dapat dieliminasi dengan diversifikasi karena risiko ini melekat dalam pasar. Risiko tidak sistematis merupakan risiko yang berhubungan dengan kejadian acak dan disebabkan oleh kegiatan-kegiatan yang dilakukan dalam perusahaan seperti: pemogokan buruh, tuntutan pihak lain, perkara hukum

serta kejadian-kejadian unik lainnya. Risiko ini dapat dieliminasi dengan diversifikasi yaitu dengan cara memiliki beberapa sekuritas tunggal dalam bentuk portofolio. Manfaat diversifikasi internasional Dengan melakukan diversifikasi internasional, investor akan memperoleh manfaat pengurangan risiko pada tingkat return tertentu. Besarnya manfaat yang akan diperoleh investor akan sangat bergantung dari koefisien korelasi, risiko dan tingkat return di masing-masing pasar modal. Dalam jangka panjang, kontribusi return melalui diversifikasi internasional yang diperoleh investor akan lebih tinggi dibanding investasi-investasi yang hanya dilakukan pada pasar modal local. .

1.

Pendahuluan jika LOOP tidak berlaku, akan ada peluang keuntungan tanpa risiko melalui arbitrase.

Dengan kata lain, barang dapat dikirim dari lokasi yang harganya rendah ke lokasi yang harganya tinggi. Namun, dalam praktiknya, sering diamati bahwa harga barang serupa gagal menjadi sama di seluruh negara. Ini bertentangan dengan gagasan arbitrase yang mendorong LOOP dan merupakan sinyal integrasi pasar yang tidak lengkap. Salah satu alasan mengapa harga barang homogen mungkin gagal untuk menyamakan di berbagai negara adalah adanya biaya transaksi yang signifikan, dan hambatan perdagangan seperti tarif atau kuota. Ketika saham ekuitas yang terdaftar silang dipertimbangkan, sebagian besar jika tidak semua biaya ini mungkin hilang, menciptakan lingkungan yang nyaman untuk menguji LOOP.

2.

Latar Belakang Teori Dan Ekonometrika

a.

Paritas Daya Beli (PPP) Absolut Vs. Relatif − Absolut memeriksa tingkat harga agregat, dan dinyatakan sbb: Pdt = Pft × Etd/f

(1)

dimana Pdt dan Pft adalah tingkat harga domestik dan luar negeri masing-masing dan Etd/f adalah nilai tukar mata uang asing. Jika persamaan tersebut tidak berlaku, maka peluang arbitrase akan muncul. − Relatif memeriksa perubahan tingkat harga agregat dari waktu ke waktu (persentasenya) secara matematis: %δpdt = %Δ Pft × %Δ Etd/f

(1*)

b) Pekerjaan Awal Dan Ordinary Least Square Untuk mendapatkan bentuk persamaan regresi umum, yang pertama tentukan persamaan dan menambahkan white noise. Selanjutnya dilakukan uji empiris, apakah paritas daya beli absolut berlaku, dan hasilnya paritas daya beli absolut berlaku. Namun untuk situasi tertentu, masalah kausalitas terbalik mungkin ada pada antara tingkat harga domestik dan luar negeri oleh karena itu dimasukkan variabel baru. Namun, studi awal menunjukkan hubungan tersebut gagal dan penyimpangan dari loop cukup signifikan dan sangat fluktuatif. Sejak itu hukum satu harga sebagian besar dianggap sebagai fenomena jangka panjang dan analisis lintas sektor telah ditinggalkan oleh para ulama. Gluschenko mencoba menganalisis perubahan integrasi pasab barang rusia dan disimpulkan disversi harga masih signifikan dan hukum gagal berlaku. Salah satu metodologi baru dan

dominan dalam litertur modern yaitu mengikuti gagasan bahwa harga komoditas homogen tidak mungkin sama di berbagai negara karena adanya biaya transaksi. Jika dua homogen dijual dengan harga berbeda, loop tidak berlaku karena tidak akan layak untuk diarbitrase jika manfaat yang diantisipasi melebihi biaya transportasi. c)

Evolusi Model Threshold Autoregressive (Tar) Dalam Literatur Paritas Daya Beli Karena kegagalan awal hubungan paritas daya beli bertahan, kerangka baru berkonstentrasi

pada nonlinier dalam arbitrase perdagangan inter sebagai akibat dari biaya transaksi. Dalam penelitian ini biaya transaksi diperlakukan sebagai pemborosan sumber daya oleh larena itu zona tanpa arbitrase diperluas dari tiap penyimpangan loop ke zona yang mencakup biaya transaksi. Tingkat perbedaan harga negara lain dan perkiraan biaya transaksi mungkin lebih lebar daripada yang tersirat oleh biaya trasnportasi. Oçonnel dan wei memperluas dengan menggunakan interpretasi yang lebih luas dari gesekan pasar yang beroperasi pada tingkat teknologi dan preferensi. Mereka juga memungkinkan adanya gesekan pasar yang tetap dan proporsional. Ketika kedua jenis biaya ini ada, mereka menemukan bahwa dua pita untuk arbitrase dihasilkan. Arbitrase kuat ketika manfaatnya cukup tinggi untuk melebihi biaya tetap. Di hadapan friksi pasar yang proporsional, penyesuaiannya kecil, dan mereka tidak membiarkan penyimpangan dari LOOP tumbuh, juga tidak membiarkan penyimpangan menghilang sepenuhnya. Penelitian dimana zona tanpa arbitrase diperkirakan dengan menggunakan model ambang batas autogresif (TAR) untuk menganalisis penyimpangan dari loop dan paritas daya beli mulai dilakukan. Hasilnya, penemuan bukti yang mendukukung dari loop ketika penyesuaian non-linier diperbolehkan. Namun ada beberapa masalah yang gagal diatasi oleh model TAR yaitu meskipun modelnya menarik untuk konteks barang individu, mungkin tidak sesuai dalam konteks agregat. Alasannya adalah bahwa biaya transaksi mungkin berbeda antar sektor dan akibatnya kecepatan arbitrase mungkin berbeda antar barang yang menciptakan efek yang tidak jelas pada tingkat agregat. Masalah lain adalah bahwa “pengembalian rata-rata dengan demikian tidak berarti pengembalian ke PPP Absolut”

d) Model VECM (Vector Error Correction Model)

Pendekatan tradisional yaitu penggunaan analisis kointegrasi dan model vecm digunakan karena adanya argumen dari Ardeni (1989) yang menyatakan adanya unit root akan melanggar asumsu model klasik karena setiap seri mengikuti random walk dan menciptakan regresi palsu. Model-model ini berfokus pada dua faktor. Pertama, model menguji apakah, dari waktu ke waktu, ada konvergensi ke LOOP. Dan kedua pada fase kecepatan konvergensi, dengan asumsi bukti konvergensi telah ditemukan di tempat pertama. Namun model ini tidak dapat memecahkan teka-teki paritas daya beli yaitu Bagaimana seseorang dapat mendamaikan volatilitas jangka pendek yang sangat besar dari nilai tukar riil dengan tingkat yang sangat lambat di mana guncangan muncul? Lembab? Dan sejak itu, model TAR telah menjadi metodologi yang dominan di pakai untuk pengujian loop dan paritas daya beli. e)

Hukum Satu Harga Dalam Literatur Keuangan Konsep LOOP adalah salah satu landasan yang menjadi dasar buku teks teori keuangan

internasional. Hasil dari LOOP, jika memang berlaku, adalah tidak adanya arbitrasepeluang. Tidak adanya arbitrase, pada gilirannya, adalah premis yang menjadi dasar hipotesis pasar yang efisien. Oleh karena itu, validitas LOOP sangat penting untuk pasar keuangan. Ada beberapa aliran utama dalam literatur keuangan yang menggunakan konsep LOOP. Salah satunya mempelajari integrasi keuangan dengan menguji validitas LOOP di pasar modal. Contoh terbaru adalah Yeyati, Schmukler dan Van Horen (2009), yang menggunakan premi lintas pasar untuk menilai integrasi keuangan. Aliran lain yang menggunakan LOOP secara ekstensif berfokus pada penemuan harga. Logika LOOP digunakan dengan cara berikut: karena harga aset yang sama berubah di pasar yang terpisah, kedua pasar menyesuaikan untuk kembali ke LOOP. Namun, dalam aliran literatur ini, fokusnya bukan pada validitas LOOP sebagai sebuah konsep. f)

Ringkasan Karya teoretis dan empiris pada LOOP/PPP sangat luas dan multidimensi. Namun,

tampaknya tidak ada konsensus khusus tentang topik tersebut. Temuan dalam literatur ekonomi telah sepenuhnya membatalkan versi absolut LOOP, dan meskipun inovasi metodologis (seperti model TAR) menguji versi relatif hukum dengan relatif berhasil, hasil akhirnya masih memungkinkan untuk kisaran tertentu di mana harga barang yang sama dapat berbeda. Itu

membatalkan bentuk asli undang-undang. Pendekatan dan fokus makalah ini berbeda dari literatur sebelum-sebelumnya dengan cara berikut. − Pertama dengan menggunakan aset keuangan yang fokus pada LOOP. − Kedua, sebelum model kompleks seperti VECM atau TAR digunakan, bentuk dasar LOOP harus diuji menggunakan OLS. Jika model VECM atau TAR digunakan, secara otomatis diasumsikan bahwa LOOP tidak berlaku karena model ini hanya menguji konvergensinya. − Ketiga, pengujian mengenai apakah krisis keuangan berpengaruh pada LOOP.

3. a.

Analisis Jangka Pendek Deskripsi Data − Menggunakana sampel 54 perusahaan Kanada yang melakukan perdagangan di Toronto Stock Exchange (TSX) dan New York Stock Exchange (NYSE) − Daftar saham yang cross-listed di TSX dan NYSE diambil dari situs Toronto Stock Exchange − Kumpulan data adalah periode dua tahun dari 2 Januari 2008 hingga 31 Desember 2009; itu mencakup periode Krisis Keuangan 2007-2009, yang memungkinkan memeriksa pengaruhnya terhadap LOOP. Untuk uji cross-sectional LOOP, saya memilih 45 hari perdagangan yang dipilih secara acak dari sampel data. Dari Tabel 1 dapat dilihat bahwa meskipun sebagian besar saham berkonsentrasi di sektor

Bahan Dasar, sampel cukup terdiversifikasi dan mencakup berbagai sektor lainnya. Hal ini mengurangi potensi masalah spesifik sektor. Faktor penting lainnya adalah sampel tidak memiliki masalah likuiditas perdagangan, karena hanya 5% saham dalam sampel yang memiliki rata-rata volume perdagangan harian yang relatif rendah b) Estimasi Hasil Estimasi dilakukan dalam dua langkah. Pertama, versi hukum satu harga berikut ini diperkirakan untuk setiap hari secara individual menggunakan metode OLS dengan kesalahan standar Putih yang kuat: Inpj- Inp*j = βjin Ej+uj

(3)

Di mana pj adalah harga penutupan setiap perusahaan di New York Stock Exchange, p *j adalah harga penutupan di Bursa Efek Toronto, dan E j adalah nilai tukar antara dolar AS dan Kanada pada waktu penutupan pasar . βj estimasi dilakukan dengan menggunakan metode kuadrat terkecil tertimbang. Karena setiap individu βj memiliki varian tersendiri σ2bj, bobot yang digunakan sama dengan 1/ σ2bj untuk memberikan signifikansi yang lebih tinggi pada perkiraan dari hari-hari yang kurang volatil. Gambar 1 menunjukkan distribusi perkiraan harian βj dan distribusi batas atas dan bawah interval kepercayaan 95 persen. Agar LOOP bertahan, garis LOOP harus berada di antara batasbatas interval kepercayaan. Hal ini terjadi pada sebagian besar hasil, namun perkiraan, dan terlebih lagi intervalnya, menunjukkan volatilitas yang tinggi pada paruh pertama tahun 2008 yakni awal dari Krisis Keuangan. Terbukti bahwa saat Krisis Keuangan mereda menjelang akhir tahun 2008, volatilitas perkiraan turun secara substansial. Bahkan dalam kasus di mana ada beberapa penyimpangan, perkiraan yang dihasilkan masih sangat dekat dengan satu dengan interval kepercayaan yang sempit. Untuk menguji pengaruh volatilitas dalam estimasi harian, saya membagi sampel menjadi tiga subsampel –1 Januari – 31 Juni 2008; 1 Januari – 31 Desember 2008 dan 1 Januari – 31 Desember 2009. Saya memperkirakan hasil keseluruhan untuk setiap periode waktu menggunakan kuadrat terkecil yang tidak tertimbang dan regresi gabungan yang menyesuaikan volatilitas harian. Tabel 2 merangkum hasil regresi ini. Dalam setiap regresi, LOOP berlaku untuk semua subsampel serta sampel penuh karena interval kepercayaan dipusatkan di sekitar 1 dalam setiap kasus. Namun, cukup jelas bahwa menyesuaikan volatilitas harian meningkatkan kualitas hasil secara signifikan. Semua perkiraan regresi gabungan jauh lebih dekat ke 1, kesalahan standar secara signifikan lebih rendah, dan interval kepercayaan jauh lebih sempit. LS tidak berbobot '08 Jan - '08 Juni

2008

2009

1.092

1,001

(.053)

(.008)

1.14 (.09) 95% CI N

.94 - 1.34 14

Sampel lengkap

1,05 (.03) 0,98 - 1,20 25

.98 - 1.02 20

0,99 - 1,11 45

Kumpulan OLS '08 Jan - '08 Juni

2008

2009

1.01

1.021

1,001

(.028)

(.008)

(.003)

Sampel lengkap

1,005 (.004) 95% CI

0,95 - 1,07 1,004 - 1,03

N

14

25

0,994 - 1,009 20

0,998 - 1,012 45

Tabel 2 Selanjutnya, hasil untuk subsampel tahun 2009 mengkonfirmasi ketika Krisis Keuangan mereda pada tahun 2009, begitu pula volatilitas interval kepercayaan, yang mengarah ke dukungan kuat dan mendukung validitas hukum satu harga di pasar.

4.

Analisis Jangka Panjang

a) Deskripsi Data Dan Tes Root Unit Untuk tujuan analisis jangka panjang, saya menggunakan seluruh kumpulan data dari 2 Januari 2008 hingga 31 Desember 2009. Untuk menguji pengaruh Krisis Keuangan terhadap deviasi dari LOOP I memecah data menjadi dua bagian. Set pertama mencakup tahun 2008 dan set kedua mencakup tahun 2009. Saya kemudian melakukan estimasi pada setiap subset secara terpisah dan estimasi dapat dibandingkan untuk memeriksa apakah ada efek krisis pada penyimpangan dari LOOP. Berbeda dengan data cross-sectional yang digunakan untuk analisis jangka pendek LOOP, kumpulan ini menyajikan kumpulan data deret waktu untuk setiap saham dan nilai tukar. Oleh karena itu, uji akar unit harus dilakukan pada setiap deret waktu untuk menentukan stasioneritas. Saya menguji akar unit setiap seri harga saham yang diperdagangkan di NYSE, setiap seri harga saham yang diperdagangkan di TSX dan seri nilai tukar. Hasilnya konsisten dengan studi Eun dan Sabherwal (2003) bahwa baik deret harga saham maupun nilai tukar $US/C$ tidak stasioner, dan semua deret adalah stasioner first-difference; dengan demikian, semua data ini terintegrasi dengan orde 1, yaitu (1) b) Model

Dengan adanya akar unit dalam deret waktu, teknik konvensional adalah menguji adanya hubungan kointegrasi antara deret tersebut dan kemudian menggunakan model VECM untuk menguji stasioneritas residu. Jika residualnya stasioner, maka disimpulkan bahwa LOOP relatif berlaku dan kecepatan konvergensi menarik. Pendekatan itu, bagaimanapun, mengandaikan bahwa LOOP tidak berlaku dalam bentuk aslinya. Menggunakan fakta bahwa setiap seri stasioner dalam perbedaan pertama, saya menggunakan model berikut untuk menguji validitas LOOP. Membiarkan dan

mewakili harga saham j pada penutupan hari perdagangan t di NYSE,

mewakili harga saham yang sama j pada penutupan hari perdagangan t di TSX.

Selanjutnya, mari

mewakili nilai tukar $/C$ pada hari t pada saat penutupan kedua pasar.

Maka persamaan LOOP adalah (4) Mengambil log dan bergerak

ke sisi kiri untuk menghindari masalah kausalitas

Granger karena perubahan harga di NYSE dapat mempengaruhi perubahan harga di TSX dan sebaliknya, persamaan (4) menjadi (5) Karena adanya akar unit di setiap seri harga, perbedaan pertama harus diambil. Dengan menambahkan konstanta dan kesalahan white noise, model regresi berikut diperkirakan untuk setiap saham secara terpisah: (6) Di mana Dan

c)

,

adalah gangguan dengan mean 0 dan varians

.

Metodologi Melihat persamaan 6, dapat dilihat bahwa ia menguji bentuk asli dari LOOP relatif.

Hipotesis bersama dari

akan memberikan jawaban tentang keabsahan hukum.

NS fitur penting dalam model adalah penghilangan semua jenis biaya, berbeda dengan model TAR yang dominan dalam literatur ekonomi. Pembenaran untuk penghilangan ini bergantung pada dua asumsi.

− Asumsi pertama adalah bahwa perusahaan yang terlibat dalam perdagangan di pasar ekuitas di berbagai negara memiliki akses ke mata uang negara yang bersangkutan tanpa terlibat dalam perdagangan valuta asing. − Asumsi kedua adalah bahwa pemain utama yang terlibat dalam perdagangan dan memanfaatkan peluang arbitrase adalah lembaga keuangan besar seperti bank investasi dan dana lindung nilai. Bersama-sama, kedua asumsi ini memungkinkan satu pembenaran untuk teknik estimasi OLS, karena tanpa adanya biaya transaksi, non-linier yang terkait dengan biaya tersebut juga hilang. Pembenaran lain untuk OLS adalah bahwa seri perbedaan harga, untuk setiap saham yang sesuai j dan seri perbedaan nilai tukar,

, adalah stasioner , juga stasioner. Itu

menegaskan bahwa masalah akar unit yang ada dalam analisis level telah diselesaikan. Oleh karena itu, OLS dapat digunakan untuk menguji spesifikasi asli dari LOOP relatif, dan hanya jika tidak memiliki teknik yang berbeda seperti VECM atau TAR yang dapat digunakan Setelah perkiraan untuk

diperoleh untuk setiap saham, hipotesis bersama dari

diuji. Kelemahan dari uji hipotesis, bagaimanapun, adalah bahwa nonpenolakan hipotesis tidak membuktikan hukum untuk dipegang, tetapi hanya bahwa tidak ada cukup bukti untuk menolaknya pada tingkat signifikansi yang tinggi itu. Dalam hal itu, interval kepercayaan 95% dan 99% dibangun untuk melihat apakah 1 termasuk dalam interval tersebut. Berdasarkan interval kepercayaan ini, kesimpulan yang lebih kuat dapat dibuat apakah LOOP relatif berlaku Langkah terakhir dalam analisis adalah agregasi. LOOP mungkin atau mungkin tidak berlaku untuk saham tertentu, tetapi itu may masih bertahan pada tingkat agregat. Oleh karena itu, dua portofolio dibangun, di mana satu portofolio terdiri dari satu saham dari setiap perdagangan saham di NYSE dan yang lainnya terdiri dari satu saham dari setiap perdagangan saham di TSX. Dengan melakukan itu, tes semua dikumpulkan

bersama-sama dapat

memberikan tes PPP untuk pilihan portofolio. Agar PPP bertahan, selisih persentase perubahan nilai portofolio harus sama dengan persentase perubahan nilai tukar. Metode kuadrat terkecil tertimbang digunakan untuk pengujian ini, karena masing-masing dan bobot yang digunakan adalah sama dengan

.

memiliki varian tersendiri

Teknik ini diterapkan pada tiga periode waktu yang berbeda: 1) 2008-2009, 2) 2009-2010 dan 3) 2008-2010. Periode pertama merupakan masa Krisis Keuangan, oleh karena itu, perbandingan antara

memberikan wawasan tentang pengaruh Krisis Keuangan

pada LOOP relatif pada tingkat individu, sedangkan perbandingan antara (perkiraan dari regresi WLS) memberikan efek pada LOOP relatif pada tingkat agregat. Periode ketiga digunakan untuk menguji validitas LOOP relatif selama seluruh periode waktu. d) Koreksi Korelasi Serial Untuk setiap saham, tes LM Breusch-Godfrey dilakukan untuk menguji korelasi serial. Kemudian, kriteria informasi Bayesian (BIC) telah digunakan untuk menentukan jumlah lag untuk OLS regression dengan kesalahan standar Newey-West. Prosedur ini diperlukan karena dengan adanya korelasi serial, kesalahan standar OLS tidak benar, yang berpotensi dapat menyebabkan interval kepercayaan dan kesimpulan hipotesis yang salah. Kesalahan standar yang benar juga diperlukan untuk regresi WLS yang benar karena varians digunakan sebagai bobot e)

Hasil Tabel A1, A2 dan A3 memberikan perkiraan OLS untuk

OLS dan kesalahan standar Newey-West, dan

interval kepercayaan 95%,

untuk setiap regresi untuk periode waktu yang

sesuai. Dapat dilihat bahwa LOOP relatif yang dimiliki hanya 15 dari 54 saham selama periode 2008 sampai 2010 dianggap. Ketika data dipecah menjadi dua sub-periode, hasilnya meningkat − LOOP berlaku untuk 28 saham pada periode 2008-2009 − LOOP berlaku untuk 24 saham pada periode 2009-2010 Hasil 2008-2009 disorot oleh kesalahan standar yang lebih tinggi dari perkiraan yang memperbesar interval kepercayaan. Namun, hasil ini tidak dapat meyakinkan tentang validitas LOOP karena menunjukkan berlaku untuk beberapa saham dan tidak berlaku untuk yang lain. Tabel berikutnya memberikan hasil agregat untuk Waktu periode

β

yang diperoleh dari regresi WLS.

95% kepercayaan diri 99% kepercayaan diri

0.8478221 2008-2009

.8213454 - .8742988

.8125435 - .8831007

.8953701 – 9.22015

.890946 - .9264392

(.131945) 2009-2010

0.9086926

(.0066421) 0.8756673 2008-2010

.8596788 - .8916559

.8543635 - .8969711

(.0079678) Tabel 3 – Hasil WLS Dapat dilihat bahwa paritas daya beli relatif gagal untuk setiap perioode ketika portofolio agregat dipertimbangkan. Namun, pengaruh krisis keuangan tersebut dapat ditelaah dengan membandingkan perkiraan periode 2008-2009 dan 2009-2010. Jelas bahwa krisis keuangan menciptakan penyimpangan yang lebih besar dari loop, karena NS

perkiraan untuk tahun 2008-

2009 jauh lebih rendah dari perkiraan tahun 2009-2010. Selanjutnya, kesalahan standar untuk periode waktu 2008-2009 adalah dua kali kesalahan standar untuk periode 2009-2010, menjelaskan mengapa perkiraan untuk periode 2008-2010 lebih rendah dari perkiraan untuk 2009-2010, karena peristiwa tahun 2008 memiliki pengaruh yang signifikan. Hasil ini konsisten dengan kesimpulan dari literatur bahwa volatilitas yang lebih tinggi menciptakan penyimpangan yang lebih tinggi dari hukum. Hasil menarik lainnya yang dapat diamati adalah bahwa meskipun LOOP relatif gagal untuk menahan,

nilainya selalu di bawah satu. Artinya, perubahan nilai tukar lebih besar

daripada perubahan harga relatif. Apa yang bisa menjelaskan perilaku ini? Satu penjelasan yang mungkin dapat diberikan oleh model overshooting nilai tukar Dornbusch (1976). Wawasan model ini adalah bahwa nilai tukar melampaui nilai ekuilibrium jangka panjangnya karena kekakuan harga dalam jangka pendek. Ketika selama krisis keuangan, bank sentral mengambil langkah-langkah dan menurunkan suku bunga, karena ekspektasi inflasi, perubahan nilai tukar dapat menjadi lebih tidak stabil, melampaui nilai jangka panjang dan oleh karena itu berkontribusi pada

memperkirakan berada di bawah satu.

Meskipun hasil ini memberikan wawasan tentang efek krisis keuangan, mereka juga menciptakan kontradiksi teoretis. Berikut adalah beberapa alasan untuk dieksplorasi yang dapat menjelaskan situasi ini: 1) Mungkin, ada semacam hubungan keseimbangan dinamis yang tidak ditangkap oleh OLS. 2) Pengaruh krisis belum sepenuhnya hilang sehingga ekspektasi inflasi belum mereda, sehingga estimasi belum sepenuhnya kembali ke 1, meskipun sempat membaik menjadi 0,9 pada 2009-2010.

3) Salah satu asumsi tentang biaya transaksi tidak berlaku, dan dalam hal ini model TAR harus digunakan. Namun, itu tidak memecahkan teka-teki, karena model TAR hanya memberikan konvergensi jangka panjang dan tidak membuktikan LOOP dalam bentuk aslinya.

5.

Kesimpulan Dalam makalah ini, saya menggunakan saham perusahaan Kanada yang terdaftar secara

silang untuk menguji validitas teori LOOP. Tujuan utama dari analisis ini adalah untuk mencoba dan menetapkan kasus dasar di mana, sebelum pengujian, orang akan mengharapkan hukum untuk dipegang. Tujuan lainnya adalah untuk menguji versi LOOP absolut dan relatif dalam bentuk aslinya yang sebenarnya. Analisis cross-sectional menunjukkan bahwa LOOP berlaku untuk sebagian besar hari dalam sampel, dan bahwa sebagian besar volatilitas dalam perkiraan jatuh pada awal 2008 – awal Krisis Keuangan. Setelah menyesuaikan ini dengan memecah sampel dari 45 hari yang dipilih secara acak menjadi tiga subsampel dan memperkirakan ulang menggunakan regresi gabungan, saya menemukan dukungan kuat yang dimiliki LOOP. Selanjutnya, hukum satu harga berlaku sempurna untuk sampel tahun 2009. Mengingat hasil ini teori memprediksi bahwa LOOP relatif akan berlaku juga. Namun, analisis jangka panjang menunjukkan bahwa LOOP relatif mungkin berlaku atau mungkin tidak berlaku untuk saham tertentu. Selanjutnya, ketika pilihan portofolio dipertimbangkan (versi PPP), hukum secara mengejutkan gagal dan perkiraan lebih rendah dari 1. Selain itu, hasil menunjukkan bahwa Krisis Keuangan 2007-Tahun 2009 berdampak negatif terhadap penyimpangan hukum, dengan meningkatkannya secara signifikan. ARTIKEL “WHY NOT DIVERSIFY INTERNATIONALLY RATHER THAN DOMESTICALLY? ➢ Motivasi utama dalam memegang portofolio saham yang terdiversifikasi adalah untuk mengurangi risiko. ➢ Risiko portofolio dalam hal variabilitas pengembalian akan lebih kecil daripada risiko bagianbagiannya yang terpisah. ➢ Jelas sekali, semakin besar jumlah sekuritas dalam portofolio, semakin kecil kemungkinan kerugian portofolio sebagai akibat dari kemalangan satu perusahaan. ➢ Total risiko portofolio akan bergantung tidak hanya pada jumlah sekuritas yang dimasukkan dalam portofolio, tetapi juga pada risiko masing-masing sekuritas individu dan sejauh mana risiko ini independen satu sama lain.

Ex: portofolio 10 sekuritas elektronik kemungkinan besar mendapat manfaat lebih sedikit dari diversifikasi daripada portofolio yang terdiri dari saham yang dipilih dari 10 industri yang berbeda. ➢ portofolio yang terdiversifikasi secara internasional cenderung membawa risiko yang jauh lebih kecil daripada portofolio domestik yang khas. A. DIVERSIFIKASI DOMESTIK Berikut contoh diversifikasi domestic berdasarkan artikel (contoh gambar kurva yg diambil 5 negara) Prosedur yang digunakan: ➔ Portofolio yang berisi peningkatan jumlah saham yang berbeda dihasilkan (untuk: mengurangi ketergantungan pada sampel tunggal) termasuk beberapa dengan ukuran yang sama, dan rata-rata ukuran risiko kemudian digunakan.

Gambar menunjukkan pengaruh diversifikasi terhadap pengurangan risiko. ➔ Sumbu vertikal mengukur portofolio risiko relatif terhadap keamanan tipikal dari negara. ➔ Sumbu horizontal memberikan jumlah surat berharga yang termasuk dalam portofolio. ➔ Ketika diversifikasi meningkat, risiko portofolio menurun di semua negara tetapi tidak secara proporsional. Sangat cepat pengurangan marjinal dalam variabilitas penambahan keamanan ekstra dalam portofolio menjadi lebih kecil. Terbukti pada Wall Street, pasar yang besar, memiliki lebih banyak peluang untuk diversifikasi. Sebagian besar perusahaan Amerika secara terbuka menawarkan saham biasa sementara sebagian besar bahkan

perusahaan Eropa terbesar adalah masih milik swasta. Karena investor Eropa akibatnya menemukan bahwa pasar domestik mereka kekurangan keragaman peluang investasi yang dinikmati oleh Amerika, diversifikasi internasional adalah relatif lebih menarik bagi mereka.

B. DIVERSIFIKASI INTERNASIONAL ➔ pengurangan risiko yang lebih besar dapat dicapai dengan mendiversifikasi portofolio secara internasional. ➔ Beberapa penulis: bahwa pergerakan harga saham di berbagai negara hampir tidak berhubungan: Perubahan harga di Bursa Paris tampak independen dari fluktuasi harga saham di bursa London, dan seterusnya. Ketika sekuritas dari satu negara (katakanlah A.S.) berkinerja lebih buruk dari yang diharapkan, pasar lain kemungkinan akan lebih baik, sehingga mengimbangi kerugian. Cukup dengan berinvestasi di saham yang berbeda negara, risikonya berkurang drastis. Efektifitas Diversifikasi Internasional dalam mengurangi Risiko ➔ Risiko investasi dalam sekuritas asing diasumsikan hanya disebabkan oleh variabilitas harga dan bukan fluktuasi nilai tukar, yang akan dipertimbangkan secara eksplisit di bagian berikutnya. Hubungan empiris antara risiko portofolio dan jumlah kepemilikan untuk portofolio internasional yang dibangun dengan peluang yang sama untuk memegang sekuritas di setiap negara diberikan pada Gambar 9. Kurva diversifikasi untuk pasar AS direproduksi dalam gambar ini untuk tujuan perbandingan (perbandingan yang sama tentu saja bisa dibuat untuk investor dari negara manapun).

➔ Keuntungan dari diversifikasi internasional cukup besar. Dalam hal variabilitas pengembalian, portofolio saham AS yang terdiversifikasi dengan baik secara internasional (dengan jumlah kepemilikan).

➔ manfaat diversifikasi internasional bagi investor Jerman atau Swiss akan lebih besar dibandingkan dengan portofolio domestik yang sangat terdiversifikasi. ➔ Meningkatkan ukuran portofolio domestik di atas 20 saham tampaknya hanya mencapai pengurangan risiko yang relatif kecil, pengurangan substansial masih dapat dicapai untuk portofolio internasional dengan ukuran yang sama. Metode Membangun Portofolio Internasional secara Independen Dari sudut pandang praktis, bagaimanapun, sangat tidak mungkin bahwa seorang individu atau bahkan reksa dana akan memiliki 500 atau 1000 saham yang berbeda dari semua negara. Beberapa aturan seleksi praktis diperlukan untuk memperoleh diversifikasi yang baik dengan ukuran portofolio yang wajar. 1. Memastikan diversifikasi geografis yang baik dengan memilih saham lintas negara, seperti dalam portofolio yang dilaporkan pada Gambar 9. 2. Cara yang lebih konvensional adalah memilih saham lintas industri. Saham dari semua negara juga dapat diklasifikasikan berdasarkan industri; dalam hal ini pemilihan oleh industri secara otomatis akan memberikan beberapa diversifikasi antar negara. 3. Metode seleksi ketiga akan menjadi kombinasi dari dua lainnya: Pengurangan risiko internasional akan dicapai dengan secara sadar memilih saham di kedua negara dan industri. Sekali lagi, portofolio internasional telah dibangun secara independen mengikuti ketiga metode ini Hasilnya tampak pada Gambar berikut.

➔ diversifikasi antar industri lebih rendah daripada diversifikasi antar negara. Kecuali untuk portofolio yang sangat besar (di mana kedua metode akan memilih saham yang sama), risiko portofolio terdiversifikasi di seluruh negara lebih rendah daripada risiko satu (internasional) diversifikasi lintas industri. ➔ Seperti yang diharapkan, prosedur gabungan dengan diversifikasi industri dan geografis memberikan hasil yang sedikit lebih baik. C. Risiko Nilai Tukar ➔ Keuntungan dari investasi internasional yang disajikan di atas mungkin dikurangi oleh banyak faktor kelembagaan, politik dan psikologis. ➔ Fluktuasi harga dengan demikian hanya mewakili sebagian dari risiko total investasi asing, karena investor mungkin juga khawatir dengan kemungkinan pengenaan kontrol nilai tukar dan pembatasan modal pada kepemilikan asing.

➔ Yang lebih penting adalah adanya risiko nilai tukar, terutama di hari-hari ini ketidakstabilan moneter internasional. ➔ Salah satu cara untuk menghilangkan risiko nilai tukar dari investasi portofolio internasional adalah dengan melakukan lindung nilai atas kepemilikan asing. Dalam banyak kasus, risiko nilai tukar dapat dihilangkan dengan membeli kontrak pertukaran forward. ➔ Jika, di sisi lain, investor sekuritas asing tidak melindungi dirinya dari fluktuasi nilai tukar, dia sebenarnya berspekulasi pada mata uang. ➔ Risiko pertukaran memiliki imbalannya dan spekulasi semacam itu mungkin cukup menguntungkan. ➔ Investor juga dapat melakukan lindung nilai atas investasi awalnya dengan membeli kontrak pertukaran berjangka, menghapus sebagian besar pertukarannya risiko dengan demikian. ➔ Untuk menunjukkan pengaruh risiko nilai tukar, portofolio internasional telah dibangun, dengan pengembaliannya pada setiap titik waktu dihitung dalam dolar dengan asumsi tidak ada perlindungan terhadap mempertaruhkan pertukaran. ➔ Pengaruh diversifikasi (jumlah saham) terhadap risiko portofolio tampak pada Gambar 1. Seperti yang dapat diharapkan, risiko portofolio yang tidak dilindungi terhadap risiko nilai tukar lebih besar daripada portofolio tertutup. Namun, total risikonya masih jauh lebih kecil daripada portofolio domestik yang sebanding. Jelas, misalnya, pemegang saham asing sangat diuntungkan dari devaluasi dolar. Portofolio Internasional yang tidak terungkap tentu merupakan lindung nilai yang baik terhadap devaluasi dolar. KESIMPULAN Kendala terakhir untuk pengembangan skala penuh dari dana multinasional baik dalam hal investasinya dan pelanggannya adalah ketakutan bahwa beberapa negara mungkin secara tak terduga memberlakukan kontrol pertukaran, membekukan modal yang diinvestasikan. Satu-satunya cara untuk mengurangi dampak dari potensi ancaman terhadap likuiditas dana akan berinvestasi di suatu negara secara proporsional dengan bagian langganan penduduknya. Ringkasnya, manfaat dari diversifikasi internasional begitu besar sehingga mereka harus dengan cepat menghidupkan kembali perkembangan reksa dana internasional yang sukses di AS—di bawah kepemimpinan, mungkin, kelompok-kelompok Wall Street yang paling dihormati, daripada beberapa petualang dengan kejujuran yang meragukan.

American Finance Association Wiley International Portfolio Choice and Corporation Finance: A Synthesis Author(s): Michael Adler dan Bernard Dumas Sumber: The Journal of Finance, Vol. 38, No. 3 (Jun., 1983), hlm. 925-984 Diterbitkan oleh: Wiley untuk American Finance Association

URL Stabil: http://www.jstor.org/stable/2328091 Diakses: 06-10-2015 14:36 UTC

Penggunaan Anda atas arsip JSTOR menunjukkan penerimaan Anda terhadap Syarat & Ketentuan Penggunaan, tersedia di http://www.jstor.org/page/ info/tentang/kebijakan/terms.jsp JSTOR adalah layanan nirlaba yang membantu para sarjana, peneliti, dan siswa menemukan, menggunakan, dan membangun berbagai konten dalam arsip digital tepercaya. Kami menggunakan teknologi informasi dan alat untuk meningkatkan produktivitas dan memfasilitasi bentuk beasiswa baru. Untuk informasi lebih lanjut tentang JSTOR, silakan hubungi [email protected]. American Finance Association dan Wiley bekerja sama dengan JSTOR untuk mendigitalkan, melestarikan, dan memperluas akses ke The Journal of Finance.

http://www.jstor.org

Konten ini diunduh dari 130.237.165.40 pada Sel, 06 Okt 2015 14:36:38 UTC Semua penggunaan tunduk pada Syarat dan Ketentuan JSTOR

THE JOURNAL OF FINANCE * VOL. XXXVIII, TIDAK. 3 * JUNI 1983

Pilihan Portofolio Internasional dan Keuangan Perusahaan: Sebuah Sintesis MICHAEL ADLER dan BERNARD DUMAS* STRUKTUR teori keuangan internasional sebagian besar mencerminkan teori keuangan domestik. Mulai dari teori mikro pilihan portofolio individu yang diperoleh, melalui agregasi dan kliring pasar, hubungan penetapan harga ekuilibrium dan pengorbanan risiko-pengembalian. Ini memberikan tujuan untuk memaksimalkan nilai perusahaan dari mana aturan keputusan dapat dihitung. Urutan analitis ini sama apakah ada satu atau lebih pasar modal. Untuk membedakan antara pengaturan domestik dan internasional, dibutuhkan konsep ekonomi kebangsaan. Pendekatan alternatif untuk ekonomi internasional pada dasarnya berbeda dalam konsepsi mereka tentang apa itu bangsa. Teori Ricardian mengidentifikasi negara berdasarkan teknologi dan preferensi konsumsi mereka. Teori perdagangan internasional Heckscher-Ohlin mendefinisikan negara sebagai zona di mana faktor fisik produksi dibatasi.' Dalam ekonomi moneter, unit ekonomi individu yang memegang mata uang yang sama dalam portofolio mereka sebagai alat pembayaran diakui sebagai milik negara yang sama, dan mata uang dibedakan satu sama lain oleh fakta bahwa mata uang tersebut dikeluarkan oleh bank sentral yang berbeda.2 Dalam portofolio teori, dua jalan sejauh ini telah dieksplorasi dalam upaya untuk menangkap dimensi internasional. Sebagian besar literatur terbaru yang berasal dari Solnik [179] telah dikhususkan untuk model di mana negara didefinisikan sebagai zona unit daya beli umum atau, lebih tepatnya, sebagai himpunan bagian dari investor *

Penulisnya adalah Profesor Bisnis, Universitas Columbia dan Profesor, CESA (HEC, ISA, CFC), masing-masing. Artikel ini dimulai saat Dumas menjadi Profesor Tamu di Columbia dan diselesaikan saat dia menjadi Profesor Tamu di Berkeley. Kami berterima kasih kepada Andr6 Saurel, yang membantu menyediakan sebagian dari basis data, dan kepada JeanFrancois Dreyfus, yang berkontribusi dalam pemrosesan data. Kami menerima komentar kritis yang berharga dari Profesor Bradford Cornell, Jeffrey Frankel, Bruce Lehmann, David Modest, Patrice Poncet, Richard Roll, Piet Sercu, Bruno Solnik, Ren6 Stulz dan, khususnya, Michael Brennan. Kesalahan mungkin tetap ada terlepas dari upaya mereka: itu adalah tanggung jawab kami. Kami mohon maaf kepada penulis yang karyanya tidak dikutip. Ini bukan indikasi kualitas pekerjaan mereka tetapi hanya ketidaktahuan kami atau terbatasnya cakupan survei ini. 1 Generalisasi teori Heckscher-Ohlin untuk ekonomi dengan produksi tidak pasti dan perdagangan sekuritas dapat dianggap sebagai milik bidang keuangan internasional tetapi tidak akan ditinjau di sini. Lihat Helpman dan Razin [85, 86, 87], Baron and Forsythe [20], dan Dumas [48, 49].

2 Pendekatan ini erat dengan teori portofolio dan berasal dari analisis neraca pembayaran, terutama karena yang terakhir mengambil orientasi moneter yang sangat kuat di tahun enam puluhan di bawah pengaruh Mundell [145], dan sejak munculnya "keseimbangan portofolio". " pendekatan: Branson [28]. Hubungan dengan teori portofolio, bagaimanapun, tidak lengkap karena yang terakhir hanya hari ini dalam proses memperkenalkan kepemilikan uang ke dalam portofolio investor individu (untuk upaya awal, lihat Roll [159], Kouri [108], dan barubaru ini. Fama dan Farber [58], Hodrick [90], Poncet [151], dan Dumas [49]). Perangkat yang paling umum digunakan adalah injeksi saldo uang riil sebagai argumen terpisah dari fungsi utilitas.

925 Konten ini diunduh dari 130.237.165.40 pada Sel, 06 Okt 2015 14:36:38 UTC Semua penggunaan tunduk pada Syarat dan Ketentuan JSTOR

926 The Journal of Finance yang menggunakan indeks harga yang sama dalam mengempiskan pengembalian moneter yang diantisipasi mereka. Kelompok investor nasional digambarkan dengan penyimpangan dari Purchasing Power Parity (PPP) yang menyebabkan mereka mengevaluasi pengembalian dari sekuritas yang sama secara berbeda. Seperti yang akan dirinci di bawah, penyimpangan ini mungkin timbul baik dari perbedaan selera konsumsi atau dari perbedaan harga berbagai komoditas yang dapat diakses oleh investor. Heterogenitas dalam evaluasi pengembalian individu ini memainkan malapetaka dengan hasil Pemisahan, Agregasi, dan Penetapan Harga Aset standar dari Teori Portofolio. Sebagian besar makalah ini akan dikhususkan untuk tinjauan dan perluasan beberapa upaya yang dilakukan untuk memulihkan hasil ini.3 Menyelesaikan masalah pilihan portofolio ketika pengembalian riil investor berbeda adalah langkah pertama yang diperlukan menuju teori keuangan internasional yang sesungguhnya. Dalam praktiknya, negara-negara selanjutnya dapat dipisahkan oleh manifestasi kedaulatan seperti pajak dan kontrol perbatasan yang membatasi transaksi keuangan swasta antar negara. Teori ekonomi keuangan tidak dengan mudah menangani ketidaksempurnaan seperti itu yang cenderung mengelompokkan pasar modal internasional. Oleh karena itu, kami menurunkan masalah yang terkait dengan segmentasi ke satu bagian menjelang akhir. Ini membuat kita sebagian besar bebas untuk mengadopsi pasar modal dunia terpadu sebagai paradigma untuk sebagian besar lainnya. Survei kemudian dapat disusun berdasarkan turunan dari model penetapan harga aset internasional varians rata-rata (IAPM) di mana beberapa negara ditampilkan sebagai wilayah yang penduduknya memiliki indeks daya beli yang berbeda. Prinsip pengorganisasian ini mencerminkan keadaan seni di lapangan, memungkinkan kita untuk mengidentifikasi kontribusi dan kesalahpahaman dalam literatur yang berkembang, dan mengarah ke rencana berikut. Urutan pertama bisnis adalah untuk menentukan apakah, dalam istilah empiris, negara dapat secara bermakna dibedakan oleh penyimpangan PPP. Oleh karena itu, setelah meninjau secara singkat kondisi keberadaan indeks harga, Bagian I mensurvei bukti empiris tentang perilaku stokastik tingkat

inflasi, nilai tukar, dan deviasi PPP. Bukti ini dengan kuat menunjukkan bahwa deviasi PPP signifikan dalam hal ukuran, bahwa mereka bertahan untuk periode yang panjang tetapi bervariasi, dan sangat acak. Oleh karena itu masuk akal untuk menganggap bahwa investor yang tinggal di negara yang berbeda memiliki tolok ukur yang berbeda untuk mengukur pengembalian riil dan risikonya. Akibatnya, orang akan mengharapkan komposisi portofolio mereka juga berbeda. Sebelum mengembangkan teori mean-variance, mengingat bahwa fungsi utilitas tidak kuadrat secara universal, lebih lanjut berguna untuk mengetahui apakah tingkat pengembalian internasional pada dasarnya terdistribusi normal. Bagian II pertama-tama membahas bukti samar mengenai pertanyaan ini. Kesimpulan sementaranya adalah bahwa normalitas bukanlah asumsi yang tidak dapat dipertahankan. Bagian ini kemudian melanjutkan ke masalah yang lebih luas tentang struktur korelasi pengembalian nominal dan riil. I Perhatikan bahwa pendekatan PPP-deviasi terhadap keuangan internasional pada prinsipnya tidak mengandaikan adanya beberapa mata uang. Karena hubungan PPP

hampir selalu dinyatakan dalam bentuk nilai tukar yang menghubungkan dua tingkat harga yang dinyatakan dalam dua mata uang yang berbeda, doktrin Pembelian "PPPI telah dikaitkan dengan teori keseimbangan nilai tukar mengambang. Bahkan, penyimpangan PPP sangat mungkin terjadi. , dan biasanya akan terjadi di bawah nilai tukar tetap, atau memang di dunia dengan mata uang yang unik; sebaliknya PPP bisa dibayangkan berlaku di manamana, menyebabkan kita mengenali satu "negara" di dunia, di hadapan banyak mata uang yang terkait secara acak nilai tukar berfluktuasi. Secara empiris, bagaimanapun, kita akan menemukan di bawah sebagian besar penyimpangan PPP terkait dengan pergerakan nilai tukar selama periode nilai tukar mengambang.

Konten ini diunduh dari 130.237.165.40 pada Sel, 06 Okt 2015 14:36:38 UTC All use tunduk pada Syarat dan Ketentuan JSTOR

Keuangan Internasional 927 Bagian III beralih ke masalah pilihan portofolio yang optimal di pasar modal dunia terpadu tanpa pajak o r biaya transaksi tetapi preferensi konsumsi investor secara nasional heterogen. Setiap investor, terlepas dari kebangsaannya, memiliki akses gratis ke menu aset yang sama: semua saham, asing dan domestik, dan satu obligasi jangka pendek bebas default per negara. Semua obligasi berisiko secara nyata. Investor diasumsikan untuk memaksimalkan aditif waktu, utilitas yang diharapkan von NeumannMorgenstern dari fungsi konsumsi seumur hidup.4 Memperkenalkan nilai tukar acak untuk menerjemahkan pengembalian masa depan dalam satu mata uang ke mata uang lain menimbulkan masalah teknis yang sudah dikenal. Pada satu tingkat mungkin sepele: perubahan unit pengukuran yang tidak menyentuh realitas ekonomi yang mendasarinya.5 Meskipun demikian, translasi mata uang menghasilkan produk dari variabel acak yang distribusi probabilitasnya sulit diperoleh. Secara khusus,

produk dari dua variabel normal tidak normal. Oleh karena itu, kami mengadopsi dalam Bagian III metodologi waktu kontinu Merton [138, 139, 140] yang sebagian besar menawarkan jalan keluar dari kesulitan tersebut. Teknik ini pada dasarnya membenarkan paradigma mean-variance dengan alasan pendekatan Samuelson [167] dan, dengan menggunakan proses Ito, secara efektif mengubah produk variabel acak menjadi penjumlahan.6 Bagian IV dan V fokus pada penetapan harga internasional aset. Teori Penetapan Harga Aset Modal yang melibatkan multiplisitas unit pengukuran, dengan masalah terjemahan yang menyertainya, telah disediakan oleh Fischer [62], Grauer, Litzenberger, dan Stehle [GLS] [79], Friend, Landskroner, dan Losq [FLL] [70 ] dan Hodrick [90]. CAPM ini adalah pernyataan ulang dalam hal pengembalian nominal CAPM tradisional Hitam [23] yang secara implisit berurusan dengan pengembalian nyata (yaitu, kempes). GLS mengasumsikan pasar yang lengkap dan menggunakan formulasi state-of-the-world yang menghindari masalah yang berhubungan dengan penerjemahan, karena operasi ini kemudian dapat dilakukan pada basis state-by-state. Fischer dan FLL menggunakan metodologi waktu-berkelanjutan; mereka tidak merujuk secara eksplisit ke beberapa mata uang tetapi cara mereka menerjemahkan tingkat pengembalian riil ke dalam nominal dan sebaliknya dapat juga digunakan untuk menerjemahkan dolar ke franc. Hodrick mendalilkan keberadaan satu barang yang diperdagangkan di seluruh dunia dengan harga yang sama dan memungkinkan pengembalian yang tidak stasioner. CAPM ini hanya berlaku jika investor menggunakan indeks harga yang sama dalam I fungsi utilitasTime additive von Neumann-Morgenstern memiliki kelemahan utama. Kelengkungan fungsi utilitas sesaat (U(*) dalam lampiran) secara bersamaan memainkan peran parameter penghindaran risiko dan elastisitas substitusi antara konsumsi titik waktu yang berbeda. Untuk upaya menguraikan dua aspek preferensi ini, lihat Selden [171].

5 Tidak peduli pengaturan ekonomi apa yang dipilih, keputusan unit ekonomi rasional (investor, perusahaan, dll.) harus invarian ketika seseorang mengubah unit pengukuran atau mata uang akuntansi, di mana pengembalian dinyatakan. Ini adalah kriteria rasionalitas minimum (yang mungkin tidak dapat diterima oleh bendahara perusahaan yang peduli dengan tampilan laporan akuntansi akhir tahun mereka). Untuk memverifikasi bahwa kriteria ini dipenuhi oleh metode keputusan yang diberikan, seseorang harus menerjemahkan dari satu mata uang ke mata uang lain menggunakan nilai tukar acak dan memeriksa bahwa keputusan yang diperoleh tidak berubah. Jika demikian, prinsip ketidakrelevanan mata uang pengukuran ditegakkan dan semuanya baik-baik saja dengan teori yang ada. 6 Metode ini membatasi setiap saat distribusi bersyarat dari tingkat pengembalian sesaat yang harus normal. Tetapi tingkat pengembalian pada interval waktu yang terbatas tidak begitu dibatasi karena parameter dari proses Ito sendiri mungkin variabel.

Konten ini diunduh dari 130.237.165.40 pada Sel, 06 Okt 2015 14:36:38 UTC Semua penggunaan tunduk pada Syarat dan Ketentuan JSTOR

928 The Journal of Finance

mengempiskan pengembalian,7 asumsi yang tidak realistis di tingkat internasional. Namun demikian, bahkan dalam pengaturan sederhana ini, seseorang dapat memperoleh ekspresi untuk nilai tukar forward8 sebagai fungsi dari distribusi nilai tukar spot masa depan yang sesuai dengan jatuh tempo kontrak forward. Sebuah kesimpulan kuat muncul: nilai tukar forward umumnya tidak lagi menjadi prediktor yang tidak bias dari kurs spot masa depan. Ini berbeda dari nilai yang diharapkan dari tempat dengan dua premi. Salah satunya adalah akibat dari risk aversion para investorspekulan. Premi lainnya akan ada di bawah netralitas risiko dan itu muncul dari adanya inflasi acak. Kami akan mendapatkan hasil yang sama, tetapi dalam pengaturan yang lebih umum di mana PPP dilanggar. Ketika PPP tidak berlaku, heterogenitas dalam perilaku pilihan portofolio membatasi agregasi tuntutan individu ke dalam CAPM. Bagian IV menyelamatkan CAPM standar sejauh mungkin, dengan membatasi cakupannya pada penetapan harga beberapa aset relatif terhadap yang lain. Jumlah aset lain ini sama dengan jumlah negara dan, untuk tujuan praktis, aset tersebut diidentifikasi dengan Treasury Bills lokal. Bagian V kemudian membahas penetapan harga aset yang tersisa ini. Bagian VI melihat lebih dekat masalah kesejahteraan yang mungkin terkait dengan keacakan nilai tukar. Ada banyak perdebatan tentang kapan dan apakah risiko pertukaran adalah "nominal" atau "nyata" dan apakah itu penting atau tidak. Isunya rumit: tergantung pada faktor-faktor seperti bagaimana uang dimasukkan ke dalam setiap perekonomian dan mengapa uang itu disimpan; tentang kelengkapan pasar modal; dan tentang bagaimana pemerintah meningkatkan pendapatan dan menggunakan hasil penciptaan uang. Kemajuan dapat dicapai dengan mengabaikan ketidaksempurnaan pasar modal. Pada akhirnya, bagaimanapun, ini harus dihadapi, mungkin lebih di internasional daripada di pengaturan domestik murni. Oleh karena itu, Bagian VII beralih ke literatur tentang segmentasi. Beberapa makalah di bidang ini bertujuan untuk menghitungkeseimbangan kondisi. Lainnya membahas keuntungan kesejahteraan dari menjembatani hambatan investasi dan kemungkinan keputusan perusahaan yang optimal. Secara empiris, tingkat keparahan ketidaksempurnaan pasar yang cenderung menghasilkan segmentasi dan luasnya segmentasi itu sendiri belum dapat diukur. Menyelesaikan masalah ini tetap menjadi tantangan utama untuk penelitian masa depan. Bagian VIII, akhirnya, beralih ke pertanyaan tentang kebijakan perusahaan. Ini berfokus terutama pada penghindaran risiko valuta asing, yaitu, kebijakan lindung nilai. Dalam pasar modal internasional yang lengkap, sempurna, dan terpadu, lindung nilai perusahaan tidak akan relevan. Bagian ini kemudian mengeksplorasi kesulitan dalam mengukur eksposur dan keputusan lindung nilai dalam keadaan di mana hal itu mungkin penting.

Pendahuluan ini berutang panjangnya pada perlunya menetapkan batasbatas survei ini. Permintaan maaf ditujukan kepada penulis yang telah menyumbangkan wawasan penting tentang topik yang terkait erat dengan topik yang dibahas di sini. Secara khusus, kita akan membahas sangat sedikit masalah dalam ekonomi makro. Uji Empiris Saldo Portofolio 7 Dalam salah satu bagian artikel mereka, GLS memang mengizinkan penyimpangan PPP tetapi investor mereka kemudian harus memiliki penghindaran risiko yang sama. Artikel GLS berkontribusi besar pada fokus pada peran penyimpangan PPP dan penguraian risiko nilai tukar menjadi komponen yang murni "nominal" dan murni "nyata". Pembahasan lebih lengkap akan diberikan pada Bagian VI.

'Nilai tukar forward adalah nilai tukar yang ditentukan dalam kontrak untuk mengirimkan mata uang asing pada titik waktu di masa depan. Lihat Bagian V.

Konten ini diunduh dari 130.237.165.40 pada Sel, 06 Okt 2015 14:36:38 UTC Semua penggunaan tunduk pada Syarat dan Ketentuan JSTOR

International Finance 929

pendekatan arus modal yang diprakarsai oleh Branson [28] telah menyoroti validitas teori portofolio dalam keuangan internasional tetapi kami tidak akan memiliki ruang untuk meninjaunya. Literatur makroekonomi yang luas tentang netralitas uang yang terkait erat dengan pembahasan Bagian VI, hanya akan disebutkan secara sepintas. Literatur tentang perdagangan dihilangkan sama sekali; lebih lanjut, bahkan untuk topik-topik yang secara langsung berada dalam cakupan survei ini (misalnya, literatur empiris tentang paritas daya beli dan kurs forward sebagai prediktor kurs spot berikutnya), ketika banyak makalah telah diterbitkan, hanya beberapa makalah yang representatif yang akan dikutip . I. Paritas Daya Beli dalam Keuangan Internasional PPP pada dasarnya memiliki dua kegunaan dalam teori pasar modal danperusahaan internasional

keuangan. Salah satunya adalah sebagai antara,

ukuran kesamaan, atau perbedaan peluang konsumsi di negara yang berbeda. Yang lainnya adalah sebagai kemungkinan pengaruh pada arus kas yang dihasilkan oleh aktivitas produksi atau perdagangan perusahaan. Perhatian kita pada bagian ini adalah pada aspek yang pertama sedangkan aspek yang terakhir akan dibahas pada Bagian VIII. Penyimpangan dari PPP, yang menurut survei kami sebagai aturan, berfungsi untuk membedakan satu negara dari yang lain. Ketika terjadi penyimpangan PPP, investor di berbagai daerah akan mengukur pengembalian riil mereka secara berbeda dan keinginan untuk memiliki portofolio yang umumnya berbeda. Untuk memperbaiki gagasan, pertama-tama mari kita bedakan PPP dari

paritas harga komoditas (CPP), juga dikenal sebagai hukum satu harga. CPP adalah kondisi arbitrase instan yang berlaku antara harga barang yang diperdagangkan identik di dua lokasi tanpa adanya hambatan perdagangan. Ini juga dapat berlaku antara barang-barang yang tidak diperdagangkan yang merupakan substitusi dekat (atau dapat diubah menjadi) barangbarang yang diperdagangkan.9 Sebaliknya, PPP adalah hubungan antara tingkat harga rata-rata tertimbang, bukan harga komoditas individual. Dalam praktiknya, tingkat harga diukur dengan indeks yang dihitung relatif terhadap beberapa periode dasar; menulis paritas tingkat harga pada dua saat yang berbeda dan mengambil rasio mengarah ke apa yang disebut "PPP Relatif." Jika tingkat harga secara sewenang-wenang ditetapkan menjadi sama dalam beberapa tahun dasar, setiap perubahan harga satu barang relatif terhadap yang lain, akan cukup untuk menciptakan deviasi PPP di tahun-tahun berikutnya. Mari kita membahas secara singkat kondisi yang diperlukan agar PPP dapat bertahan dengan tepat. Pertama, indeks harga konsumen yang dihitung oleh lembaga statistik nasional dunia harus merupakan representasi yang valid dari kemungkinan konsumsi dan preferensi warganya. Lampiran merinci asumsi yang diperlukan. Jika konsumen suatu negara adalah pemaksimal utilitas (dengan utilitas tambahan waktu), mereka akan memiliki indeks harga komposisi invarian di mana bagian anggaran mengungkapkan selera mereka jika preferensi mereka juga homotetik. Mengingat preferensi homotetik seperti itu, indeks biaya hidup yang dipublikasikan memperkirakan indeks yang tepat dan dapat dibandingkan secara wajar di seluruh I

negara.10CPP dapat diuji secara langsung dengan membandingkan harga absolut setelah diterjemahkan ke dalam mata uang umum. Tes kedua adalah apakah harga relatif dua barang sama di mana-mana. 10 Kita tidak mengetahui adanya pengujian langsung terhadap homotetisitas preferensi. Jika asumsi homotheticity dilanggar, dua indeks harga variabel (satu mencerminkan anggaran "rata-rata" dan anggaran "marginal" lainnya

Konten ini diunduh dari 130.237.165.40 pada Sel, 06 Okt 2015 14:36:38 UTC Semua penggunaan tunduk pada Persyaratan JSTOR dan Kondisi

930 Jurnal Keuangan Agar PPP dapat bertahan dengan tepat, kondisi yang cukup mencakup preferensi homotetis, seperti di atas; CPP sehubungan dengan setiap barang yang termasuk dalam indeks; dan, di samping itu, selera yang sama untuk menjamin bahwa komposisi indeks negara yang berbeda akan Kondisi ini tidak perlu, terlihat dari perdebatan tahun 1920-an bahwa PPP juga dapat muncul meskipun ada perbedaan selera dan adanya barang yang tidak diperdagangkan, asalkan terdapat cukup substitusi antara barang yang dikonsumsi dan antara barang yang diperdagangkan dan yang tidak diperdagangkan. barang dalam

produksi untuk menghasilkan korelasi yang tinggi antara harga komoditas individu. Detailnya tidak perlu kita perhatikan di sini. Sebagaimana dicatat, pentingnya deviasi PPP tions dalam keuangan internasional berasal dari cara di mana investor menghitung pengembalian riil dari keamanan yang diberikan. Pertimbangkan seorang investor AS yang memegang sekuritas Jepang dengan pengembalian nominal tertentu (mungkin acak). Jika pengembalian nominal (atau moneter) ini diukur dalam yen,1" ia pertama-tama akan menerjemahkannya ke dalam dolar dan kemudian mengempiskannya menggunakan Indeks Harga Konsumen AS atau, lebih baik lagi, indeksnya sendiri, yang dinyatakan dalam dolar.12 Ini menghasilkan AS daya beli pendapatan asing ini. Tetapi, pertimbangkan seorang investor Jepang yang memegang sekuritas yang sama dan mengharapkan pengembalian nominal nominal (yen atau dolar) yang sama. Pengembalian riilnya diperoleh dengan mengempiskan pengembalian nominal yen oleh indeks harga miliknya, atau indeks harga Jepang. diukur dalam yen. Jika PPP berlaku tepat, yaitu, jika dua indeks harga persis sejalan dengan nilai tukar, kedua investor akan melihat pengembalian riil secara identik. Gagasan mereka tentang pengembalian riil dari sekuritas yang sama akan berbeda sejauh bahwa indeks harga mereka, dinyatakan atau diterjemahkan ke dalam mata uang yang sama, berbeda, yaitu, sejauh mana PPP dilanggar.Namun, ini tidak berarti bahwa penyimpangan itu sendiri masuk ke dalam kalkulus investor atau, oleh karena itu, menonjolkan dengan cara apapun sebagai sumber risiko yang terpisah. Masing-masing menghitung pengembalian riilnya menggunakan indeksnya sendiri, terlepas dari perbandingan apa pun dengan indeks orang asing. Untuk distribusi pengembalian nominal sekuritas tertentu, keberadaan PPP tidak menghilangkan atau mengurangi risiko investasi asing secara absolut atau relatif terhadap investasi dalam negeri.13 saham) diperlukan untuk memodelkan pilihan portofolio mean-variance. Lihat bagian terakhir dari Breeden [30]. Lihat juga Stulz [191].

11

Mulai dari pengembalian yang diukur dalam Yen tidak penting. Hasil akhirnya, hasil nyata, tidak tergantung pada fakta itu. Jika kami telah menetapkan bahwa keamanan, atau pengembaliannya, dalam mata uang Yen, itu akan menjadi indikasi untuk struktur pembayaran keamanan, yaitu bahwa pengembalian dalam Yen ditetapkan sebelumnya. Spesifikasi seperti itu akan berlaku untuk deposito di bank Jepang atau sekuritas pendapatan tetap Yen. Untuk jenis sekuritas lainnya, seperti saham, tidak peduli apa mata uang pembayaran dividen acak, tidak ada denominasi apriori. Tetapi, berdasarkan struktur hasil probabilistik dari sekuritas, seseorang dapat menetapkan denominasi implisitnya. Ini adalah objek perhitungan eksposur; lihat di bawah, di Bagian IV dan VIII. 12 Jelas indeks biaya hidup yang diberikan oleh lembaga statistik nasional memiliki dimensi mata uang (terlepas dari fakta bahwa mereka biasanya dikutip sebagai angka tanpa dimensi yang dihitung relatif terhadap tahun dasar di mana indeks ditetapkan pada 100). Oleh karena itu, mereka dapat diterjemahkan dari satu mata uang ke mata uang lainnya. Dengan demikian, seseorang dapat memperoleh indeks biaya hidup rumah tangga Prancis yang

dinyatakan dalam dolar AS. 3 Meskipun demikian, perbedaan harga antara wilayah perdagangan pada umumnya akan mempengaruhi kegiatan perdagangan dan produksi perusahaan dan oleh karena itu pengembalian mereka serta kesejahteraan umum (lihat Bagian VI dan VIII). Tapi ini adalah masalah yang terpisah. Kami sedang memeriksa pilihan portofolio sekarang dengan pengembalian sekuritas yang diberikan. Dalam konteks ini, deviasi PPP muncul hanya ketika membandingkan pilihan portofolio investor yang berbeda, bukan dalam memeriksa risiko yang ditanggung oleh salah satu dari mereka.

Konten ini diunduh dari 130.237.165.40 pada Sel, 06 Okt 2015 14:36:38 UTC Semua penggunaan tunduk pada Syarat dan Ketentuan JSTOR

International Finance 931

0

cr_

~~~~~~COLU)

z U

I)

O

a_

tnnt(v/ U /Foc

I .I

n .St I . III . . . ._

1973 1974 1975 1976 1977 1978 1979 Catatan: Observasi bulanan Kurs Dollor/F.Fr.spot

(An

St) dan

Forward (tn Ft ) pada RasioAS/Perancis

Biaya Hidup

Indeks

Ctn(COLUS/COLF) (skala setara dengan

nilai tukar spot pada bulan awal )] Juni 1973-Juli 1979. Sumber: JA Frenkel (68 )

Gambar 1. Deviasi PPP AS-Perancis, 1973-1979

Bukti empiris mengenai deviasi PPP penting terutama untuk mengungkapkan bahwa pengembalian riil investor memang berbeda dan oleh karena itu komposisi portofolio mereka harus berbeda Gambar 1 adalah tipikal dari catatan empiris: pola serupa ditampilkan oleh semua mata uang utama lainnya. Implikasi yang jelas adalah PPP dilanggar, selama periode terakhir, penyimpangan PPP besar dan kumulatif di merasakan bahwa penyimpangan dalam arah tertentu telah berlangsung lama tetapi bervariasi. Jelas, PPP adalah hipotesis yang dipertanyakan untuk jangka pendek; apakah lebih baik dalam jangka panjang? Dalam hal Gambar 1, apakah fakta bahwa nilai tukar tampaknya berputar secara tidak teratur di sekitar rasio indeks harga berarti bahwa penyimpangan PPP pada akhirnya membalikkan diri? Mari kita periksa dulu bukti mengenai dua penyebab utama penyimpangan PPP: "4 penyimpangan CPP dan perbedaan dalam keranjang konsumsi nasional. CPP biasanya berlaku dalam margin biaya transaksi yang sempit untuk barang-barang homogen yang diperdagangkan di pasar lelang terorganisir seperti bursa komoditas Emas dan logam lain yang mudah diarbit adalah contoh yang baik. Karena Cassel mengusulkan hipotesis PPP, bagaimanapun, penulis lain termasuk sebagian besar '4 Kami tidak akan mensurvei secara mendalam literasi yang banyak tentang PPP. Ini telah ditinjau oleh Balassa [18] dan Officer [ 148] dan dalam Journal of International Economics edisi Mei 1978.

Konten ini diunduh dari 130.237.165.40 pada Sel, 06 Okt 2015 14:36:38 UTC Semua penggunaan tunduk pada Syarat dan Ketentuan JSTOR

932 The Journal of Finance baru-baru ini Isard [ 96], Kravis dan Lipsey [112], dan Richardson [156] telah menemukan bukti penyimpangan yang persisten dan besar dari CPP (hingga 5% selama periode tarif tetap dan hingga 20% di bawah floatin tarif g). Penyelidikan empiris yang mendalam telah dilakukan oleh Katseli-Papaefstratiou [101] dan Crouhy-Veyrac et al. [38]. Sampel gabungan mencakup sebagian besar kelas barang yang diperdagangkan; penyimpangan sebagian besar tidak dapat dijelaskan. Keberadaan mereka telah dikaitkan dengan ketidaksempurnaan yang mencegah arbitrase seperti pajak, tarif, dan biaya transaksi; informasi tak simetrik; atau monopoli dan penetapan harga yang diskriminatif. Sebagian mereka mungkin ilusi. Barang-barang dalam klasifikasi perdagangan komoditas yang sempit sekalipun tidak homogen. Selain itu, mungkin ada kesalahan pengukuran harga, terutama pada kontrak jangka panjang. Namun, secara keseluruhan, pelanggaran CPP tampaknya menjadi aturan daripada pengecualian. Sejauh menyangkut PPP, Kravis dan Lipsey [112] mengutip Kravis et al.

[111] telah menghasilkan perbandingan tingkat harga absolut pada tahun 1970. Tabel 1 mereka direproduksi di sini sebagai Tabel I kami. Perbandingan dilakukan dalam dua cara: sekali menggunakan bobot nasional lokal, dan sekali menggunakan secara seragam untuk semua negara set bobot yang sama (yaitu bobot AS) untuk berbagai kelompok komoditas. Pemeriksaan sederhana menunjukkan bahwa penyimpangan lebih lebar ketika menggunakan bobot yang berbeda daripada saat menggunakan bobot yang sama. Oleh karena itu, perbedaan selera konsumsi nasional memang berkontribusi terhadap penyimpangan PPP. Akan tetapi, ada pandangan yang sering dikemukakan, bahwa sementara PPP bisa gagal kapan saja, ini adalah hipotesis jangka panjang yang bagus. Keyakinan ini tampaknya didasarkan pada gagasan bahwa deviasi rata-rata, di mana rata-rata diambil dalam jangka waktu yang lama, tampaknya cenderung nol untuk sebagian besar negara. Gailliot [72] menghitung persentase penyimpangan indeks harga grosir dari paritas daya beli dengan dolar AS, dari rata-rata 1900-4 ke rata-rata 1963-7; mereka adalah, untuk Kanada .04, Prancis -.01, Jerman .04, Italia -.11, Jepang .26, Swiss .14, dan Inggris Raya .11 atau .02, tergantung pada indeks yang dia gunakan. Aliber dan Stickney [16] menghitung deviasi PPP tahunan dan menemukan bahwa deviasi ratarata menurun Tabel I

Tingkat Harga Relatif pada tahun 1970 (AS = 100) Bobot Sendiri Bobot AS Tingkat Harga Tingkat Harga yang Diperdagangkan Tidak Diperdagangkan Tidak Diperdagangkan PDB Negara Barang PDB Barang Barang Kenya 34.9 58.5 22.3 74.2 96.7 48.9 India 24.0 46.8 10.3 47,6 74.4 17,3 Columbia 31,8 57,6 21,0 62.0 91.6 28.6 Hongaria 44.7 62.7 30.7 66.0 95.0 33.0 Italia 66.2 84.0 50.4 82.9 107.0 55.5 Jepang 61.1 82.5 46.4 76.4 95.3 55.0 Inggris Raya 65.8 87.0 55.9 FR 79,8 99,2 60,9 95,4 114,5 73,5 Prancis 73,8 83,4 63,4 89,1 105,9 70,4 Amerika Serikat 100,0 100,0 100,0 100,0 100,0 100,0 Sumber: IB Kravis, Z. Kenessey, A. Heston, dan R. Summers [111]. Konten ini diunduh dari 130.237.165.40 pada Sel, 06 Okt 2015 14:36:38 UTC Semua penggunaan tunduk pada Syarat dan Ketentuan JSTOR

International Finance 933

karena jumlah pengamatan tahunan meningkat. Hasil seperti itu diharapkan jika penyimpangan PPP mengikuti pola seperti yang tercermin pada Gambar 1. Pertanyaannya adalah apa implikasinya untuk pemodelan tingkat harga komoditas dan nilai tukar secara absolut, dan relatif satu sama lain. Haruskah kita memilih proses stokastik untuk kuantitas ini yang menggabungkan pembalikan terhadap PPP? Jawabannya tentu saja harus berasal dari model keseimbangan umum, tetapi pada tahap ini, kita hanya dapat mendiskusikan elemen-elemen dari formulasi yang masuk akal. Bukti empiris yang tersedia tidak menunjukkan kecenderungan apapun terhadap pembalikan (atau korelasi serial non-nol). Gailliot [72] mengklaim berdasarkan pengamatan biasa bahwa deviasi positif rata-rata 5 tahun dalam satu dekade diikuti oleh deviasi negatif pada dekade berikutnya, antara tahun 1903 dan 1967. Bukti seperti itu tampaknya tidak kuat. Roll [161] memberikan tes pertama keberadaan korelasi serial dalam deviasi PPP. Menggunakan data bulanan IMF untuk dua puluh tiga negara antara tahun 1957 dan 1976, Roll tidak menemukan bukti korelasi serial dalam pengumpulan sampelnya. Namun, beberapa ukuran korelasi serial, baik positif maupun negatif, muncul di beberapa negara, terutama Iran, Argentina, dan Meksiko. Hasil masing-masing negara tidak dianalisis secara rinci. Tes lain oleh Rogalski dan Vinso [158] mencapai kesimpulan yang sama. Adler dan Lehmann [10], menggunakan regresi lag terdistribusi, menguji hipotesis bahwa deviasi PPP mengikuti martingale (dan karena itu tidak menunjukkan korelasi serial). Berbagai sampel data bulanan dan tahunan (baik untuk periode tarif "tetap" dan fleksibel) dan berbagai durasi kelambatan (hingga sepuluh tahun) semuanya menghasilkan hasil yang serupa: sebagai aturan, model martingale tidak ditolak.15 Hasil empiris ini menunjukkan bahwa PPP dilanggar secara instan dan dapat diperkirakan akan dilanggar untuk setiap horizon peramalan. They therefore suggest the heterogeneity of national consumption tastes as a foundation for international finance. The economic interpretation of the empirical success of the martingale process is not straightforward, however. A theoretically sound model of PPP deviations would incorporate both the action of costly commodities trading and that of almost costless information-efficient foreign exchange trading. Costly commodi ties trading would cause the law of one price (CPP) to be violated instantaneously (ex post) and would allow the exchange rate to fluctuate within a band (akin to the gold points of the previous century) on either side of its PPP level. On the other hand, costless information-efficient foreign exchange and bond trading causes the current spot exchange rate to be the best predictor of the future exchange rate adjusted for interest rates (see Section V) and further, since interest rates anticipate inflation, it also causes the current real exchange rate to be the best predictor of the future exchange rate adjusted for the anticipated inflation difference between the two countries. This last statement is what may be termed the ex ante PPP hypothesis and it leads to the martingale model.16 But it is not clear how the commodities and the foreign exchange markets

15 Ie, the fact that a currency is below its PPP value is no indication that it will subsequently appreciate absolutely or even relative to price levels. Nevertheless, a number of commercial foreign exchange forecasting services sell forecasts based on PPP deviations. 16 As was stressed by one referee, ex post PPP, and not just ex ante PPP, would be needed to remove the heterogeneity of perceived real returns across investors.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

934 The Journal of Finance interact. If the martingale model held strictly, an absolute deviation taking place today would never tend to be corrected; this is surely an undesirable feature of this model. For the present, all that can be safely said is that the actual behavior of PPP deviations cannot be statistically distinguished from a martingale. More specifi cally, under floating rates, PPP deviations behave approximately like the ex change rates themselves17'18 (see Genberg [73] and our Table II) and the latter more or less follow a martingale (Poole [152]). II. The Empirical Structure of International Returns Since much of the research in international finance involves either applications of statistical regression techniques or extensions of meanvariance portfolio theory, we should briefly review the little that is known about the probabilistic structure of international returns. Two separate aspects are covered in this section: one is the question of the probability distribution which best describes these returns; and the second is the matter of estimating empirically the scope for risk reduction through international diversification and, more generally, the validity of market regression models as descriptors of the risk structure of international asset prices. Let us take each in turn. A. The Probability Distribution of International Returns The first issue is whether the returns on international investment are normally distributed and, if they are not, what probability distribution best describes them. Unfortunately, no one yet has investigated the distributions of real returns. Farber, Roll, and Solnik [61] examined the distribution of R = (St+,/St) [(1 +

where S is the spot exchange rate and rk and rn are the foreign and domestic nominal interest rates: R represents the nominal excess rate of return (over the domestic riskfree rate) on speculation in the money rk)/(l + rn)] - 1

market.19 For monthly data between 1964 and 1975, they discovered that the distribution of R departed from normality but less, apparently, than the distribution of the exchange rate change itself: the comparison was not made in great detail. It is the distribution of exchange rate changes which has received the most attention. Besides Farber, Roll, and Solnik [61], investigations include those of Westerfield [202], Dooley and Shafer [42], Papadia [149] and, most recently, McFarland, Pettit, and Sung [135]. In all cases, severe departures from normality were discovered, generally more so for pre-1973 data than for the floating-rate period. Following the research of Granger and Morgenstern [77] and Fama and 17 Ie, price levels fluctuate little in comparison to exchange rates. To some extent, this may be due to a diversification effect between commodities prices. See Cornell [35]. 18 This indicates that PPP deviations and the portfolio issues arising from them are properly raised within the context of a multi-currency world. PPP deviations also occur domestically (ie, across one country) but they are probably less important and less random.

19 Under interest rate parity (cf. Section V), R can be rewritten R

=

S,+1Ft - 1 where Ft is the forward rate: R is then also the nominal return to forward speculation for a contract size equal to one unit of domestic currency. If the domestic interest rate properly anticipates domestic inflation, it is also an estimate of the real rent from a foreign investment position (in excess of the domestic real interest rate). This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 935 Table II

Sources of Deviations from PPP: 1971 February-1979 December. Correlations of eight exchange rates (against the US dollar) with their respective PPP levels. 107 observations The correlations are measured after taking first differences on both sides. Germany .996 Belgium .995

Canada .923 France .998 Japan .938 Netherlands .994 United Kingdom .961 Switzerland .983 Data Source: OECD main economic indicators, various issues.

Roll [59], and the observation that the empirical exchange-rate distributions had fatter tails and greater kurtosis than the normal, further tests suggest that these distributions may be stable-paretian with characteristic exponent between 1.3 and 1.8. The currently prevailing hypothesis, then, is that the distributions of the exchange rates belong to the stable, infinite-variance class.20 Alternative views are, however, by no means excluded. One view is that the observed exchange rate variations are drawn from nonstationary normal distri butions; ie, that the exchange rates themselves follow a stochastic process with variable parameters. In order to account for the fat tails in exchange rate variations, this process should have a tendency to diverge or oscillate more widely than does a regular Brownian motion; ie, the variance would have to grow faster than linearly with time. One other view is that exchange rates, as a result of government intervention, or for some other reason, undergo discrete jumps. This could be modeled by means of combined Poisson and diffusion process, provided the mean frequency of these jumps is approximately constant. Kouri [107] calculated portfolio choices and some elements of the equilibrium (the forward premium) under the Poisson assumption. Both alternative views lack a good empirical analysis so far. 20As Granger [78] points out, it is, of course, difficult actually to prove that any sequence of observed random variables in fact has a stable, non-normal distribution. Strong indication but not proof is provided by tests of the stability of the characteristic exponent as increasing numbers of (log) price or exchange-rate changes are added together, both chronologically and in randomized

order. Westerfield tested only the chronological sequences and found the characteristic exponents to be reasonably stable for sums of between one and ten weekly observations: characteristic exponents rose with the length of sequence but reached a maximum of 1.69. McFarland, Pettit, and Sung employ daily data between 1/75 and 7/79. Tests of chronological and randomized sequences both showed a tendency for characteristic exponents to rise slowly

but to remain below 1.7 for sequences of up to 15. However, when the chronological sequences were lengthened to 25, at the cost of reduced sample size, the characteristic exponents of six out of eight exchange rates reached 2.00, the critical value for normality. The authors express little confidence in this comforting result. This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

936 The Journal of Finance The major question raised is whether these various distribution assumptions make mean-variance theory totally inapplicable. With regard to the infinite variance hypothesis, one may note that, even if returns distributions are stable, they may be approximately normal within a finite range; if, further, utility functions are appropriately bounded, the expected utility integral computed over such a normal approximation may be approximately mean-variance. There is a clear need, at this point, for better theories of approximation. The nonstationarity hypothesis would create only minor complications at the theoretical level; the state variables which shift the parameters of the stochastic process must merely be made explicit. Meanvariance theory remains valid locally. The real difficulty is empirical; it would arise when attempting to identify the state variables and estimating the functional link between the parameters to the process and the state variables. Of course, the wider the class of permissible model formulations, the more hazardous the statistical task.21 Similar comments may be made regarding the possibility of adding on some Poisson jumps. B. Correlations among International Returns and Diversification If national financial markets are not perfectly (positively) correlated, investors should be able to reduce their portfolio-variance-risk without sacrificing expected return by international diversification. This simple insight gave rise to a series of papers, including Grubel [80]; Levy and Sarnat [122], who computed also internationally-efficient combinations of stock market indices subject to a short selling constraint; and Solnik [181] who demonstrated that the additional vari ance reduction could be obtained with a relatively small number of securities. Table III illustrates, with data for nine stock markets during the 1971-9 period, the kind of correlation patterns which underlie these earlier studies. Panel A presents correlations between pairs of nominal returns which are not translated into US dollars or adjusted for US inflation: these seem quite similar to those presented in Lessard [117] for the period 1959-73. Translation and deflation do not exchange the coefficients very much: the correlations among real returns in US terms appear in Panel B. The thing to notice about both panels is that the correlations are fairly small. The same pattern holds true for returns measured in other currencies and deflated by other indices. One other method used to display the potential for risk reduction consists

in regressions of individual stock returns or of national market indices on a world market index. This is the route taken by early writers such as Agmon [12], Solnik [180], and Lessard [118]. In Lessard's experiments, the residuals of regressions of the stock's national index or of its local industry index on the world market index proxy were introduced as additional orthogonal factors into the market model test. In the actual event the fits produced by Agmon's single index, Lessard's multiple factor, and Solnik's "national" and "international" factor models were relatively weak. Generally, at least 40 percent of the variation was left unex plained. This result is not in itself surprising, given the low correlations between 21 This is especially true when the results of the statistical estimation are to be fed into an optimization program. See the end of Section III.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 937 Table III

Correlations of Monthly Price Relatives For Nine Major Stock Market Indices: 1/71-12/79 Panel A: Nominal Monthly Returns in Original CAN FRA JAP NET UK SWI USA EV DEV.

Currency

ST. GER BEL

GER 1.00 0.005 0.041 BEL 0.48 1.00 0.008 0.038 CAN 0.24 0.45 1.00 0.011 0.051 FRA 0.44 0.60 0.34 1.00 0.009 0.070 JAP 0.35 0.29 0.26 0.13 1.00 0.013 0.050 NET 0.59 0.66 0.52 0.50 0.37 1.00 0.005 0.047 UK -.03 0.06 -.01 -0.6 0.04 -.04 1.00 0.008 0.820 SWI 0.52 0.64 0.46 0.42 0.37 0.63 -.14 1.00 0.004 0.050 USA 0.28 0.49 0.69 0.34 0.34 0.52 -.005 0.53 1.00 0.005 0.046 Panel B: Real Monthly Returns in $US, Deflated by USCPI ST. GER BEL CAN FRA JAP NET UK SWI USA EV DEV. GER 1.00 0.011 0.090 BEL 0.51 1.00 0.012 0.085 CAN 0.25 0.40 1.00 0.004 0.053 FRA 0.41 0.59 0.40 1.00 0.018 0.144 JAP 0.47 0.40 0.28 0.26 1.00 0.011 0.059 NET 0.54 0.65 0.47 0.51 0.39 1.00 0.010 0.104 UK 0.11 0.09 0.66 -.02 0.06 0.12 1.00 0.071 0.764 SWI 0.56 0.70 0.46 0.55 0.48 0.60 -.07 1.00 0.007 0.058 USA 0.34 0.48 0.69 0.38 0.33 0.53 0.02 0.49 1.00 -.002 0.046 Sources: (a) Capital International, Perspectives, various issues; (b) OECD Main Economic Indi cators.

the indices from various exchanges. It is consistent also with the earlier results of Blume [25] and King [103] for the US market. The size of the firmspecific residual variances in any of these regressions, domestic or international, certainly seem to suggest that considerable scope exists for risk

reduction from diversifi cation. However, one should be careful about taking this conclusion much further. The main limitation of this kind of analysis arises from the choice of the index or indices against which the regression is run. In domestic finance, regressing a security's return on any efficient-portfolio return produces residuals whose var iance is easily interpreted as "diversifiable risk," since an investor choosing that portfolio would require no compensation for bearing that risk. Further, in a market which is at equilibrium and in which practically all investors do diversify, the market portfolio is one portfolio known to be efficient (Roll [160]). The normative implication is then fairly simple: diversifiable risk can be measured against the market portfolio. But, when investors' purchasing power units differ by nationality, as we allow below, they will in principle differ in their concept of what an efficient portfolio is (except if they all have logarithmic utilities: see Section III); and there will be no implication that at market equilibrium the market portfolio should be efficient in any sense or, a fortiori, that its nominal rate of return measured in any currency should serve as a benchmark for valuation. Any segmentation will compound this effect. As a result, there can be This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

938 The Journal of Finance no simple way to infer the extent to which the gains from variance-risk reduction can be reaped by different nations' citizens merely from analyzing the structure of securities' markets data alone. The data themselves may be consistent with any number of models of portfolio selection and market equilibrium, with or without segmentation. The availability of risk reduction will depend upon which one is true. This point should put an end to all further attempts to base measures of such quantities as "risk reduction" or "diversifiable risk" on sample estimates of the means and variances of and correlations among market or industry indices. The covariance matrix also says nothing about the presence or absence of segmentation.22 Suppose, however, that we have an international CAPM where the market portfolio nominal rate of return is the proper benchmark, as is the case, for instance, in Solnik [179]. National or industry factors in this setting are theoret ically irrelevant.23 One may then regress individual stock returns on a single international market index. To an approximation, the residual risk may be diversified away.24 One may, of course, introduce additional orthogonal factors such as the residuals from regressions of national indices on the world index or from regressions of industry indices on the world index. The expanded specifi cation cannot produce any information regarding the relative benefits of inter national as opposed to interindustry or purely

domestic diversification. This observation underscores the confusion surrounding the question of whether interindustry diversification within countries or international diversification across industries, or some combination of both, can be replicated by international diversification across national indices. This question, which no one has examined systematically, is of considerable practical interest. In short, the main use of single or multiple-factor market models is probably, much as in Sharpe's [175] diagonal model, the reduction of data requirements for the computation of optimal portfolios. The variability of the world index combined with other factors, or alternatively, the variability of foreign market indices combined with orthogonal factors representing local industry groupings and the world market both leave large fractions of individual assets' variabilities unexplained. The potential for international diversification to reduce risk seems unquestionable. Beyond simplifying data requirements,25; however, the available 22 See Adler and Dumas [3] and Section VII below. Two more criticisms may be leveled against these statistical descriptions. One is potentially devastating: rates of return may not be stationary so that it may not be legitimate to apply correlation or regression analysis to them. A second one may be less important empirically: most of the published work refers to US current dollar returns when in actually deflated returns should have been employed. 23 This is not a critique of Solnik's [180] empirical procedure. Quite the opposite: he was precisely trying to verify that national factors (orthogonal to the world portfolio) received a zero price. 24 See, however, Friend and Losq [71] who argued that a simple extension of the CAPM such as Agmon's [12] in which all investors worldwide hold only the world market portfolio, will tend to overstate the gains from international diversification. 25 These statistics do have descriptive, if not normative, value, however, and they may lead to some more fundamental analysis: one could seek the sources of the correlations between financial markets and the origins and differences between means and variances of return. Eg, is there a connection between these and patterns of trade and specialization?

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 939 market-model results offer neither guidelines for the construction of optimally diversified international portfolios nor the basis for evaluating the benefits or value of such diversification. These matters are addressed next. AKU AKU AKU. Portfolio Choice In view of the fairly low correlation between national financial markets observed in the previous section, it is possible to reap important gains from international portfolio diversification. In the present section we seek the optimal portfolios of worldwide investments which maximize these gains.

Consider a world of L + 1 countries and currencies. Without loss of generality, we measure nominal returns in terms of the L + 1st currency. Nominal rates of return given in another currency can easily be translated by multiplying one plus the foreign-currency rate of return by the ratio of the end-of-period to the beginning-of-period exchange rate. There are N nominally risky securities, whose nominal price dynamics in terms of the measurement currency are given by stationary Ito processes (Brownian motions):26

dYilYi

=

,ui dt + oi dzi i = 1

...

N(1

where Yi is the market value of security i in terms of currency L + 1; ,ui is the instantaneous expected nominal rate of return on security i; oi is the instantaneous standard deviation of the nominal rate of return on security i; and zi is a standard Wiener process and dzi is the associated white noise. We also define Q as the N x N matrix of instantaneous covariances fik of the nominal rates of return on the various securities. Finally, there is one (the N + lst) security which is nominally riskless: an interest earning bank deposit or short-term bond denominated in the measurement currency. The instantaneous nominal rate of interest paid on this deposit is denoted r. In some applications, it will be useful to distinguish two subsets among the nominally risky securities: the last L securities may be taken to be the nominal bank deposits denominated in the non-measurement currencies, while the first n(n = N - L) would be stock securities paying a random dividend.27 If we 26

See the critique below.

27Formally Equation (1) only allows for income in the form of capital gains. A simple change of notation can incorporate dividends. Let a (constant return-to-scale) dividend dx be paid in an interval of time dt with dx/Y = Md dt + od dZd

and let the price actually behave as: dY/Y

=

(M - Ad) dt + o- dz - od dzd (1')

then the total rate of return on the stock is given by Equation (1) in the text. Stock price behavior

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC

All use subject to JSTOR Terms and Conditions

940 The Journal of Finance partition accordingly the covariance matrix Q, its southeast block then contains the covariances of exchange rates. There are L + 1 national investor types, each with homothetic utility func tions.28 The price index P' of an investor of type 1, expressed in the measurement currency, follows a stationary process:29

dP11P1 = 7rl dt + a' dzl I=1 *- L + 1 (2) where xri and ai, standard deviation of the instanta

are the expected value and neous rate of inflation as seen by investor 1. We call wl the N x N vector of covariances a7 of the N risky securities returns with investor l's rate of inflation. The superscript 1 will be dropped whenever we consider one (generic) investor in isolation. While they are not strictly inconsistent with the empirical evidence presented in Sections I and II above, one may nevertheless raise strong objections to the a priori specification of the functional forms ((1) and (2)) for the dynamics of stock and commodities prices when they should actually be endogenous. Lucas [129] can be interpreted as having shown that the stationary Brownian motion for asset prices in equation (1) is not consistent with the positive risk aversion of (time-additive von Neumann-Morgensterm utility endowed) investors. It would be consistent only with risk neutral investor behavior where ,u would be the rate of time discount (applicable uniformly to all securities). Under risk aversion, asset prices multiplied by discounted marginal utilities must follow martingales and so discounted asset prices by themselves generally do not, in contradiction with equation (1). For further details, see Lucas [129]. Rosenberg and Ohlson [164] also pointed out that the portfolio choices (9) to be derived from the equations of motion (1) are unreasonable. In a domestic setting at least, these asset demands would imply that the prices of assets relative to each other would all be functions of one and the same random factor (the weighted average of risk tolerances a) and so would all be perfectly correlated. These internal flaws can be corrected by introducing nonstationary Ito proc esses where the parameters iu and o- would be functions of a vector of state variables. The resulting portfolio choices would contain one more (hedge) fund per state variable, in addition to the two funds which appear in (9) below. See Merton [140] or Breeden [30], Stulz [191], and Hodrick [90]. The functions ,u (.) and ar (.) would in turn have to be endogenized at equilibrium. Procedures for so doing have been provided by Cox, Ingersoll, and Ross [37]. They amount essentially to interpreting the CAPM as a functional equation in these unknown factors, rather than as an algebraic equation giving the expected return on a security. This final step, so far, has only been performed for economies populated with identical consumer-

investors. (1') is still a geometric Brownian motion because (a dz - od dzd)/(o72 + ao - 2powrd)'12

is a Wiener process, for so long as u and Jd are constants (p stands here for the correlation between dz and dzd.) 28

See footnote 10 for references to the more general case.

29

See the critique below. This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 941 For this reason, we are not prepared to propose a complete generalequilibrium model of international capital markets. The heterogeneity in consumption tastes which characterizes international finance would require as yet unknown proce dures for computing equilibrium prices. State variables which will ultimately have to be introduced into the model could easily be handled in the portfolio computation below, in the manner of Merton [140] or Breeden [30], but this added complication would not speak to the specific feature of the field which is the heterogeneity in the purchasing power of money. In order to focus on that aspect and to illustrate its implications most vividly, we restrict ourselves to stationary Brownian motions; ie, to constant,u's and 's in Equation (1). The equation of motion (2) can equally be criticized on the ground that commodities prices should be endogenized. In addition, it might have been preferable to start the analysis with the price dynamics of individual commodities rather than with that of a price index. Were individual prices to follow stationary Brownian motions, price indices generally would not, as expenditure shares fluctuate with relative prices (even if preferences are homothetic30). Only when expenditure shares are constant (Cobb-Douglas utility function) can both indi vidual prices and indices simultaneously be Brownian. But we have no reason to favor one assumption over the other. Note that we leave unspecified the reason for the PPP deviations (dP'/P' - dP1/P1); they may equally arise from differences in tastes (as in Stulz [191]) or from CPP deviations. If, as suggested by the empirical evidence of Section I, ex ante PPP holds and PPP deviations follow a martingale, the parameters of the stochastic process in (2) can be restricted accordingly. Assuming homothetic direct utility functions, the material of the Appendix implies that we may express an investor's objective function as:31

rT

Max E V(C, P, s) ds (3)

where C is the nominal rate of consumption expenditures, P is the price level index, and V(.) is a function homogeneous of degree zero in C and P =

expressing the instantaneous rate of indirect utility. Calling w- Iwi, the (N + 1) x 1 vector whose components sum to 1 and which indicates the investor's portfolio choice among the available investment opportunities, his wealth dynamics are:32 dW = [YNJ1 wi(Ai -

r) + r]W dt - C dt -

ZN=1 wii

dzi (4)

where W is nominal wealth. Denoting by J(W, P, t) the maximum value of (3) subject to (4), the Bellman principle states that this function must be stationary 30 Homotheticity only guarantees that indices do not fluctuate with wealth, as relative prices are kept constant. If, in addition, utilities were not homothetic, recall from Breeden [30] and Stulz [191] that two indices (one based on average and the other on marginal expenditure shares) would be needed. A similar observation had been made by Adler and Dumas [6] in the more restrictive context of utilities assuming a quadratic form. 31 Some bequest function could be added without modifying the results we wish to obtain. Income from sources other than security returns is ruled out. 32 See Merton [138]. The last portfolio variable WN+1 has been eliminated on the basis of ziN+ w1 = 1.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

942 The Journal of Finance or that its total expected rate of increase must be identically zero:

O Max[V(C, P, t) + Jt + Jw[ffYj1 wi(i - r) + r]W -C

C,w

+ JPP7r + X2JW,W EN1 ,k=l WiWkUi,kW + X Jp,pcP2 p

+ JWP

EN=1 wiai,,r WP] (5)

The homogeneity of degree 0 of the function V(C, P, t) implies that J(W, P, t) and C(W, P, t) which satisfy (5) must be homogeneous of degree zero in W and P. By Euler's theorem: jp _-( W/P) Jw

and therefore:33 JP,W -(l/P)Jw - (W/P)Jw,w

JP,P -( W/P)Jw,p + (W/P2)Jw -2(W/P2)JW + (W/P)2JW,W

Substituting into (5)

O Max[V(C, P, t) + Jt c,W

+

-

Jwt[Z'=L wi(Mi r) + r

-

7r + = 7r-

C}

+ XJw,w{g=1 S^=1 WiW^ - 2 wia, + 4}W2] W2]

The derivatives with respect to the decision variables C and wi are set equal to zero: Vc = Jw (6) 0 = Jw(Ai - r - Ci,7r) + Jw,w(X^=l Wkai,k - a?,)W (7)

Defining: a = -Jw/Jw, w W as the investor's risk tolerance,34 we can rewrite

(7) in the form of a required nominal yield on security i: Ai

=r+

(1--4, +- =1 Wkfi,k, (8)

an equation we shall have occasion to refer to again. Solving for the optimal portfolio directly in vector notation, we get: w

=

a(1

-F-1(- -

r1))

+

(1 - a)(1 _ lQ-1@) (9)

where 1 is an N x 1 vector

of ones and 1 ' its transpose; u is the vector of 3

This procedure has been used by Fischer [62] and Losq [127].

4 This was not the only possible definition of risk tolerance. An alternative definition (Breeden [30]) is: -Vc/VccC. The two are not equivalent, since the wealth elasticity of consumption is not generally equal to 1.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC

All use subject to JSTOR Terms and Conditions

International Finance 943 nominal expected returns, pi; Q is the N x N matrix of instantaneous covariances ai,;k of the nominal rates of return on the various securities; X is the N x 1 vector of covariances ai1, of the N risky securities returns with the investor's rate of inflation. As has been pointed out by Kouri [107] and Losq [127], the optimal portfolio is the combination with weights a and 1 - a of two component portfolios which we now interpret. As is well known,35 the logarithmic utility function ((p)tln(C/ P)) implies a = 1. The first component portfolio (with coefficient a) is therefore the portfolio of a logarithmic investor. The formula indicates that its composition is independent of the behavior of commodity prices. This is not a new result:36 ln(C/P) = ln C - ln P and therefore commodities prices separate out in the objective function and have no influence on the decisions. As a result, the logarithmic component is the same for all investors, irrespective of nationality; a logarithmic investor would be nationless. Geometrically, this implies that the Markowitz efficient frontiers of all the investors have one point in common (where they are all tangent necessarily). Naturally, the composition is independ ent of the choice of the measurement currency.37 The second component portfolio (with weight 1 - a) is the portfolio of an investor with zero risk tolerance (a = 0). It is therefore, for any given investor, his global minimum variance portfolio in real terms. The formula based on nominal rates of return bears out this interpretation: Q-lc is the vector of regression coefficients of the investor's rate of inflation on the various securities' returns. This portfolio is thus the one whose nominal rate of return is the most highly correlated with the investor's rate of inflation (in measurement currency); or, in other words, it is the best possible hedge against inflation. By Ito's lemma, the random part of real returns is nothing but the random part of nominal returns minus the random part of the rate of inflation. The regression just mentioned minimizes the variance of this difference which is the variance of the real portfolio return. Since this hedge portfolio involves the rate of inflation of commodities' prices, it is investor specific. Its nominal returns composition is independent of expected (gL) since the formula involves only covariances; this is appropriate, since this portfolio minimizes the variance without regard for profitability. The composition is also independent of the choice of measurement currency. We summarize these results in the following formulae (describing investor l's portfolio) and a theorem: Wi a 'Wl,og + (1 - al)Wh (10)

wlog= _Q-j(gu rl)) (11) ,,,h = ( 1@ = 11 ... L + (

3

1

(12)

Merton [139].

36

Hakansson [82].

3 The incredulous reader may check in Sercu [174], Appendix A, where calculations are performed explicitly, using translation of returns, to verify that fact.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

944 The Journal of Finance THEOREM.38 (Optimal portfolio strategy for the individual investor). Every investor in the world holds a combination of: -

the universal logarithmic portfolio with weight a.

- his

personalized hedge portfolio which constitutes the best protection against inflation as he perceives it, with weight 1 - a.

We have computed the logarithmic portfolio39 as well as the hedge portfolios of US and French investors as they would have been over the years 1971-9, based on ex post monthly rates of return. They are displayed in Table IV. The logarithmic portfolio exhibits large positive and negative weights, negative weights implying borrowing or short selling. This is the component portfolio which takes advantage of expected rate of return differentials. There are exact opposite entries in Canadian and US dollar deposits, because, during this period, US interest rates were above Canadian ones without offsetting exchange rate changes.40 There is a similar pseudo-arbitrage between the Deutsche Mark on the one hand and the Belgian Franc and the Guilder on the other. When comparing stocks and bank deposits, there is a clear tendency for bank deposit entries to be negative in those currencies where the stock entry is positive and vice-versa, although the numbers are by no means exactly opposite. The reason is that exchange rate variations tend not to offset, and to be wider than, stock price variations; hence there is an incentive to hedge stock purchases against currency risk by means of local borrowing. The hedge portfolios are even more striking. Although we have only shown the US and French hedge portfolios, the pattern is the same for all nationalities.4" An investor's hedge portfolio is almost entirely made up of a nominal bank deposit (or Treasury Bill or short-term bond) denominated in his home currency. The reason is that exchange rate fluctuations are much wider than

price level (CPI) fluctuations, as we observed earlier in Section II. Also, contrary to a 38

It is known since Black [23] that mean-variance investors who care only about real returns and who are confronted with nationality distinctions, only need (any) two efficient funds. The specific choice of the two funds made here is specially telling in the context of purchasing power differences across investors. 3 All our statistics and portfolios are conditional. Nominal interest rates presumably contain some ex ante information on ensuing inflation rates, exchange rates and stock returns. We have therefore regressed all rates of return on the nine comcomitant inte rest rates of the various currencies and computed all statistics and portfolios from the residuals of the regression. This procedure amounts to treating nominal interest rates as pseudo state variables. As a result, the logarithmic portfolio composition is a (linear) function of the interest rates prevailing and it fluctuates from month to month. We show in Table IV the average portfolio over the entire decade. One referee objected to

this statistical treatment on the grounds that it introduces errors in the variables. In practice, the procedure makes little difference: nominal interest rates simply "explain" very little of ensuing variations. It was introduced only in order to ensure that portfolio choices would be exactly invariant with respect to the choice of measurement currency. 4 The same empirical result is obtained by Braga de Macedo [26] over the period October 1973 to April 1978. It arises, of course, from the strong correlation between the two currencies. But this multicollinearity itself renders unreliable the two figures -21.81 and 22.01. This and other critical statistical problems will be discussed below. 41 The weights to be placed on their home-currency deposit by the ten national investor types are: for German investors, 1.009; for the Belgians, 1.029; for the Canadians, 0.983; for the French, 0.988; for the Japanese, 1.007; for the Dutch, 0.973; for the British, 1.030; for the Swiss, 1.021; and for US nationals, 0.983.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

Ihternational Finance 945 popular belief, but full in line with the observations of Lintner [125] and Fama and Schwert [60], stocks are not good hedges against inflation, for various reasons. As a consequence, investors who are very risk averse prefer to bear fully their home inflation risk rather than to bear exchange rate uncertainty or stock price uncertainty. It is not clear whether the result would be modified if real estate or commodities such as rare metals were included in the array of possible investments. The log-portfolio calculations, however, should not be taken at face value. They suggest improbably that investors individually and in the aggregate should short some securities and hold more than 100% of the available supply of others. Beyond the possibility that the calculated weights violate typical short selling constraints such a result, if sustained, would imply that international capital markets are not in equilibrium (a point on which the evidence will be reviewed in sections IV and V).

Moreover the estimates are plagued by major statistical problems which undermine their significance.42 No statistical theory, to our knowledge, gives the sample distribution of the estimated wlog from equation (11). We are, therefore, unable to build confidence intervals for the optimal log-portfolio composition in Table IV. Simulation experiments using a single risky asset suggest that these confidence intervals are much wider than the hypothesized [0, 1] range. Had we run them with several risky assets, multicollinearity would have compounded the problem. Presumably, it would be possible to develop more efficient estimates of i&logthan the one obtained simply by premultiplying the estimated average return vector by the inverse of the estimated covariance matrix. Bawa et al. [22] have worked on a Bayesian theory aimed in that direction but few applications have been made. The hedge portfolio composition is, nonetheless, very clearcut and it is doubtful that, as long as one uses CPI's as indicators of inflation, any statistical problem could raise doubts about it. In fact, the small variability of CPI's relative to securities' returns and exchange rates in the countries we have considered provides a rationale for the early work of Solnik [179], generalized

recently by Sercu [174],43 where it is assumed that each investor ignores his homecurrency inflation (or assumes it to be null) and therefore considers rates of

return expressed in home currency units as being real returns. Quite evidently this is a case of deviation from PPP44 since different people regard the same securities' returns differently. One consequence of the assumption is that the home currency bank deposit or Treasury Bill is seen as riskless in real terms by the national investors (and only by them). It is not difficult to verify on the basis of equation 42 This is quite apart from the issue, which we leave aside, of the possible nonstationarity of the rate-of-return distribution. 43 Solnik assumed independence of exchange rates and nominal local-currency stock returns, and he further made some (internally inconsistent) assumptions regarding exchange rate behavior. Sercu corrected these deficiencies. One referee pointed out that Solnik had produced an appendix to his work where independence was no longer assumed, but he had then reached no simple statement of portfolio strategy. 44 One controversy arose from Solnik's assumption No. 7 which seemed to imply the absence of trade between countries! This assumption is actually unnecessary, for so long as there are several goods and different consumption tastes, PPP deviations could occur even if trade were unhampered.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

946 The Journal of Finance Table IV

Universal Logarithmic Portfolio and Investor Hedge Portfolios for US and French Investors, Computed from Nominal Rates of Return February 1971 to December 1979

Logarithmic Hedge Portfolios Portfolio US French (weights sum to 1) Investor Investor Stocks Germany -6.18 0.021 0.025 Belgium 6.15 0.000 -0.032 Canada 4.68 0.000 -0.028 France -1.59 0.000 0.011 Japan 3.01 0.005 0.002 Netherlands 1.34 -0.011 0.014 United Kingdom 0.01 0.000 0.000 Switzerland 0.90 0.001 -0.022 United States -6.75 -0.020 0.020 Bank Deposits Deutsche Mark 11.57 -0.029 -0.047 Belgian Franc -9.22 -.003 0.059 Canadian Dollar -21.81 0.034 0.046 French Franc 3.02 0.004 0.988 Japanese Yen -0.68 0.034 -0.007 Guilder -2.79 0.016 -0.009 British Pound -4.10 -0.024 -0.005 Swiss Franc 1.43 -0.017 0.017 US Dollar 22.01 0.983 -0.032 Data Source: Morgan Guaranty, World Capit al Markets for one-month deposit rates; Capital International, Perspectives for stock price indexes and dividend rates; OECD, Main Economic Indicators for US and French CPI's. Warning: Monthly dividends are taken to be the last 12-month dividend divided by twelve.

(12) that in this case the hedge portfolio reduces to the home deposit.

Consider the portfolio problem of an investor of country 1; since he assumes that his rate of inflation measured in currency l is zero (or nonrandom), the same rate of inflation translated into the measurement currency L + 1 and to be introduced into the covariance vector wi, reduces to the rate of change of the (L + 1/l)th exchange rate; and since the translated rate of return on the currency l treasury Bill to be introduced into the covariance matrix Q is also equal, in its random part, to the same exchange rate change, the regression of inflation on securities which underlies formula (12) will produce a unit coefficient on the currency-i Bill and zeros on all other securities. Hence:45 46 Solnik [179], in one of his separation theorems, had further split up the logarithmic portfolio into two: one whose return was independent of exchange rate changes and one which was fully dependent upon them (the latter being made up of bank deposits only). Sercu [174] did the same but only in order to show that Solnik could be generalized. While this procedure is always open (see also our Section VIII) even in the general case, there is really no point to it since the logarithmic portfolio in its entirety is held by all investors. If we split it into two parts, the two subportfolios will be held in the same proportion by all.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 947 COROLLARY. (Solnik; Sercu)

When home inflation (measured in home currency and using home consump tion weights) is zero (or nonrandom), every investor in the world holds a combination of: -

the universal logarithmic portfolio with weight a.

his home currency Treasury Bill or bank deposit, with 1 - a. Statements such as these generated considerable interest among macroecono mists (Kouri and de Macedo [110], Dornbufsch [43], Krugman [113]) aiming to explain spot exchange rates and, more specifically, to find links between the current account and the exchange rate.46 Portfolio balance theorists such as Branson [28], Girton and Henderson [75, 76], and Kouri [109] worked with postulated asset demand functions and assumed that investor's exhibited "home habitat" preference: they demand their home-currency denominated asset rela tively more than foreigners do. A country's current account surplus which places more wealth in the hands of home investors consequently raises the demand for home currency assets and causes a rise in the value of the home currency. The corollary above provides the microeconomic foundation for this reasoning: if a shift in wealth in favor of the home country leaves the world logarithmic demand more or less unchanged, it will usually raise the demand for the home Treasury bill. Braga de Macedo [26] and Krugman [113] suggested that the reasoning will be correct only if the home risk tolerance a < 1 (ie, home investors are holding the home bill rather than borrowing at that rate).47 Strong further assumptions are required, however, to make this an exact equilibrium argument. These are detailed in the next section. One is tempted to accept Krugman's conjecture that the magnitude of the effect is

small. Actually, in much of this body of literature, the optimal portfolio strategy theorem has been specialized in another way which is often less favorable to the argument. Imagine that a country's output prices are fairly stable (or nonrandom) when expressed in the local currency. A given investor consumes in certain proportions goods produced in his home country which have a stable price and goods imported from various foreign countries whose translated prices fluctuate like exchange rates. He composes his purchasing power index accordingly. When, following formula (12), this index is regressed on the various securities' returns, including the translated returns on foreign Treasury Bills, which vary like exchange rates, the weights of the hedge portfolio obviously reconstruct the consumption mix. When the investor spends 10% of his consumption budget on goods imported from France, 10% of his hedge portfolio is devoted to French franc bank deposits as a hedge against the translated price of French imports.48 COROLLARY. (Kouri and de Macedo)

When the various countries' local-currency output prices are nonrandom, the weights falling on foreign currency Treasury Bills or bank deposits in an inves 46

We are grateful to Jeffrey Frankel for bringing this important insight to our attention. 47 A frequently accepted value is a = X (risk aversion equal to 2). 48 Kouri and de Macedo actually have a more general model where output prices are allowed to be random. But the theorem to be given, which is our responsibility, captures their contribution most strikingly.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

948 The Journal of Finance

tor's hedge portfolio replicate his consumption mix according to origin of the goods.49 The contrast between the two corollaries should serve to highlight the depend ence of the hedge portfolio composition on the choice of the commodities price index: the CPI versus one which incorporates explicitly the prices of imports. The reason for this difference is to be found in the odd behavior of national CPI's: they do not seem to reflect immediately the variations in import prices arising from exchange rate changes. Why they do not is an open issue presumably linked to the behavior of importing firms and to the theory of commercial contracting. IV. Partial Pricing: n Assets Priced Relative to L + 1 Other Assets

In the tradition of the Capital Asset Pricing Model (CAPM) of Sharpe [176], Lintner [124], and Mossin [143], equilibrium in the capital market is character ized by a relationship between the required yields on the various assets. The demand is assumed to originate from investors who hold optimal portfolios given by Equation (10) and the supply is assumed to be fixed. One asks the question: what return must this security bring relative to another security so that investors are willing to hold both in the proportion in which they are available? In the international context, the heterogeneous perceptions of real returns, due to PPP deviations, will prevent us from answering this question for every security; we shall have to take as given the expected rates of return of as many securities as there are countries (L + 1) and price the other securities (n = (N + 1) (L + 1)) relative to these. The nominal yield required on the various securities by an individual investor in order for him to be willing to hold a given portfolio w is given to us by Equation (8) which we reproduce here for convenience, emphasizing with a superscript the terms which depend on the identity 1 of the investor:

zi =

r + (1 -

1/a')a,i

+ (1/a') kL wkik; i = 1

...

N (8 repeated) As we saw, this

equation is equivalent to the formulation of portfolio demand (9)

yields.

It

may

be rewritten in the following =

for

given

form:50

r+

a', + (1/a') Zk=1 Wk(Ui,k - aN,j; i = 1 (13)

In this formula, the last term contains the covariance of the nominal return on security i with the nominal return on security k minus the covariance between the nominal return on security i with the rate of inflation; this difference is, in effect, the covariance between the nominal return on security i with the real return on security k; hence the summation is the covariance between the nominal return on security i and the real return on the investor's portfolio. 4 Here is a comment by Braga de Macedo [26]: "That the effect of an increase in the relative demand for country 1 goods increases the relative demand for country 1 currency is similar to the condition for stablity in a flow view of the foreign exchange market, whereby the demand for currency is derived from the supply and demand for exports and imports." Some tests on the external demand for US dollars were run by Dumas and Poncet [50]. 60 Recall

0 that EkI' wk = 1 and that Ui,N+l = for all i since the last security is nominally riskless. This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 949

The intuitive meaning of (13) is therefore as follows. A security must bring a nominal return in excess of the nominal rate of interest, which is made up of two premia. The last one is a risk premium proportional to the covariance of the security's nominal rate with the investor's real portfolio return. A covariance with a portfolio is the usual measure in the CAPM literature of the risk contributed by a security to a portfolio (its marginal risk). When investors are concerned with their purchasing powers, they relate the required nominal yield on each asset to the real returns on their benchmark portfolio, much as one would expect. The first premium in (13) is not a risk premium as it would exist even if the investor exhibited zero risk aversion (1/l = 0). It reflects the fact that investors predicate their portfolio choices on real returns. The expected real return on a security depends on the expected value of the nominal return, the expected value of inflation, and the covariance between the nominal return and inflation. Deflation involving a product, it generates a covariance when computing expected values. We may perhaps call this the inflation premium. Formulae (8) or (13), although valid for every investor, are unfortunately not usable directly to obtain, in an empirically meaningful fashion, the required yield on the various securities. This is because we cannot observe the individual portfolio holdings, wl. The only portfolio which is directly observable by reading the prices in the newspaper is the aggregate one, given by the relative market capitalizations of all the securities on the market: the market portfolio5' wvm with W= W Wk/1l W'

where the summation is taken over all the investors and Wl is investor l's nominal wealth. We must therefore transform (8) into an equation valid at the aggregate market level. The operation of aggregation is typically performed by multiplying (8) by al and taking an average over all investors, where the weights are their relative wealths. But in the present case, with PPP deviations and with investor specific measures of inflation, the result will be disappointing:

E (1 - al)Wlr m =i

r+

(1 - 1/am) El (1 - a') Wl + (1/am) Lk= Wkmik i= 1 *..N (14)

where

am

=

(El Wla')/(El W')

The disappointment is that the second term of (14) is now unobservable. Indeed, it contains the covariances of security i with the various investors' rate of inflation, weighted by their wealth and by one minus their risk tolerance. It is evidently out of the question to measure each individual's risk tolerance.52 But 51 Roll [160] objects that, in a world where the prices of most assets (eg, real estate or human capital) are observed infrequently, if ever, the market portfolio itself is not observable. Worse yet, no proxy measure of this portfolio will work adequately in a test of the CAPM. 52 This stumbling block is intrinsic to the international setting of heterogeneous investors' rate of inflation. No model known to us is capable of collapsing the wealth and risk-tolerance weighted

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

950 The Journal of Finance

note that the problem may not be as large as it seems. The summation in the second term can be performed in two steps: once over all the individuals of the same nation who use the same deflator and therefore have the same aj,,; and then once again over the several nations, using the national wealth weighted average risk tolerances in lieu of the individual ones. We therefore have only a sum of L + 1 terms, one for each nation. The result is a CAPM containing L + 1 terms of covariance with inflation in addition to the intercept and the covariance with the market. This "multi-beta" CAPM may be tested directly for so long as the number of data points (ie, securities) is sufficiently large (N + 1 > L + 3). The hypotheses to be tested are that the intercept is equal to the nominal measurement-currency interest rate, that the regression coefficients on all the covariance terms sum to one and that the coefficient on the covariance with the market is positive. There exists, however, another procedure involving a prior analytic transfor mation of (14), which leads to a useful economic result. When the model is exact, this second approach is strictly equivalent to the previous one. It starts with the observation that the main difficulty arises from the L + 1 national weights (summing to one): (1 - a') W/(1 -a)WI

which are not observable. In order to compute them, one may reverse the problem partially; ie take the expected yields on L nominally risky securities (eg, the last L ones) as given, assume that they conform exactly to the model and use them to solve for the unknown weights. The weights so obtained can then be substituted back into (14) to compute the required yields on the other

n(n = N - L) securities. The effect of this procedure is to make the covariances between rates of return and inflation independent of investors or, equivalently, to set the covariances between rates of return and PPP deviations, which reflect the differences among investors, equal to zero.

To achieve the requisite transformation of (14), we introduce 'Yj: this is the vector of regression coefficients from a regression of the returns of security i(i < n), on the last L securities, so specified as to render the residuals independent of PPP deviations. For the given covariance matrix, the definition of y' emerges from:53 k=n+l _Yi,krk,ir

O=

Ci,or Ek=n+l /i,k0k,7r n,

+L = L+T - +L _ L+L.

n

average rate of inflation into one observable number. The same problem arises, for instance, in the consumption CAPM of Breeden [30] and Stulz [191] when PPP is violated. The technique leading to (15) and (16) is an extension of Sercu [174] to the case with random domestic inflation rates. Pricing stocks relative to bonds requires partitioning and inverting the covariance matrix. The partition of the inverse corresponding to stocks can then be identified as the inverse of a matrix of the residuals from regressions of each stock on the set of bonds, so structured as to render these regression residuals themselves independent of PPP deviations. The Py emerge naturally as the coefficients of these regressions and represent the weights of a portfolio of bonds which immunizes stock i from the PPP deviations. A stock combined with its associated hedge portfolio is a hedged stock. Equation (15) defines the -yi vector as the solution to setting the covariance between the hedged stock and PPP deviations equal to zero. This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 951 The suggested procedure then leads to the following CAPM: k-n+ 'Yi,k) +

(1 -

-

7/of)(4, Ek=n+1 'Yi,kOk,ir)

+

-

i- =n+1 Yi,kAk=

(l/am)[JN=, wNa(ij,1 Ek'=n+l oYi,k',k'j)]; i = 1,

... ,

r(1 - E

n; Vl; (16)

where it will be noted that the right-hand side has the same value (by virtue of (15)) no matter which national rate of inflation (1 = 1, * * L + 1) is used. The economic interpretation of the above is clearest if one visualizes the ye's as side investments in the last L securities which would accompany negatively each unit investment in security i, (icn). The purpose of these auxiliary investments is revealed by equation (15) which is a condition on the net nominal return from security i and its associated bundle of securities held short. The left hand side is the covariance of this net return with investor l's purchasing power index and the right-hand side is the covariance with investor L + l's index. Taking the difference between the left and the righthand sides, equation (15) says that the side investments associated with

security i are chosen in such a way that the net return is linearly independent of the purchasing power deviation between investor 1 and investor L + 1 (dP'/P' dPL+1/PL+%) and generally independent of all the L basic PPP deviations which may arise between L + 1 national investor groups. The side investments therefore constitute a hedge of security i against PPP deviations. It is natural to choose the L, non-measurement currency bonds as the hedging vehicles. The result is a CAPM which prices stocks relative to the L bonds, that is, which provides the expected returns on stocks only when all L + 1 nominal interest rates are given. Consider, then, the following zero-investment bet: -

invest 1 measurement currency unit in security i,

1

-

borrow (short sell) ' n,kunits in securities k = n + 1 to n + L, borrow - E n+1 ik units riskfree in the measurement currency. The net expected return (in measurement currency) of this bet is the left-hand side of (16) minus the first term on the right-hand side. According to (16), this net expected return is linearly related to the covariance with inflation and the covariance with the market, exactly as in the nominal CAPM of, eg, Friend, Landskroner and Losq [70]. Hence: THEOREM. The net nominal required yield on a security hedged against PPP

deviations is given by the traditional nominal Capital Asset Pricing Model.54 As noted, equation (15) specifies the y's as coefficients of a regression of the first n securities on the last L ones, such that the residuals are independent of PPP deviations. We use the L nonmeasurement-currency bonds as the refer ence securities. These securities are nominally risky only because exchange rates are random. In practice, the regression can be estimated by instrumental variables techniques, [100, p. 278], with the PPP deviations playing the role of 6 The reader should carefully avoid a misunderstanding regarding the words "hedged against PPP deviations." The theorem means that we, as financial economists, know how to price a security by the standard CAPM once we have associated to it a combination of securities which constitutes a hedge against PPP deviations. There is no implication that PPP deviations are some kind of separate

risk against which any one investor would want to hedge. See the remarks at the end of Section I and in Section VI below.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

952 The Journal of Finance the instrumental variables. This is largely equivalent to a 2 SLS procedure. In the first stage, one regresses the exchange rates on the PPP deviations: in

the second stage, one regresses the stock returns on the fitted values from the first stage. The coefficients in the second stage provide estimates of the yj vector; and by construction the residuals are independent of the PPP deviations. In the Solnik-Sercu special case (see Section III) of zero local currency inflation as seen by local investors, PPP deviations are collinear with exchange rates. Hence, there is no distinction between the instrumental variables and the regressors, and the regression reduces to an ordinary least squares regression of the first n securities (stocks) on exchange rates. In that case, the regression coefficients y may be interpreted as the sensitivities of the stocks with respect to exchange rates, or as their "exposures" to exchange risks. We shall return to this notion in our Section VIII on corporate policy. The theorem may thus be specialized: COROLLARY No. 1 (Solnik-Sercu). When local inflation rates are zero, the net nominal required yield on a security hedged against exchange risks by means of multi-currency borrowing and lending, is given by the traditional nominal55 Capital Asset Pricing Model. There is actually an alternative way to derive and to state this corollary. We saw in Section III that, in this special case, the hedge portfolios are entirely made up to the investors' respective home Treasury bills. Stock securities, that is, receive a zero weight in the hedge portfolios and are therefore held by investors only as part of their logarithmic portfolio. Hence: COROLLARY No. 2. When local inflation rates are zero, the world market portfolio of stocks and the stock part of the logarithmic portfolio are proportional to each other. If, in addition, Treasury bills are in zero net supply, we have mwilog=

wr

i =1,

...

,n

Substituting (11) into this relationship and partitioning out the elements corresponding to stocks would directly produce the CAPM alluded to in Corollary No. 1. The y coefficients would appear as one partitions the inverse covariance matrix. One word of warning is in order: this corollary does not say that the market portfolio as a whole is efficient for anybody. By way of illustration, we have computed the coefficients y for the stock market returns included in our sample (described in Section II). This was done by regressing each market index (translated into dollars) on all exchange rates using the instrumental variables technique described above. Because exchange rates are so closely correlated with PPP deviations (cf. Table II), however, we may regard these coefficients as approximate exposures to exchange risk (ie, OLS regressions on exchange rates). The numbers therefore are of great descrip tive interest. They are displayed in Table V. It appears, from the diagonal elements of the array, that most European stock markets and Canada are

6 In that case, it is immaterial whether one uses a nominal or real CAPM as the second term on the right-hand side of (16) (the covariance with inflation) is now equal to zero.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 953 in terms

Note:

United

-0.045.

United

Japan

the France Canada

of

See

Belgium

Sources

Switzerland

Germany

of

francs

regression the parentheses .

Stocks

The

against

States Netherland s

and Kingdom

data:

"Exposure"

See

PPP

the

text.

instead to

'y

of Table

coefficients

the

or

Actually

French US

-0.057

0.078

0.276

US

the

(0.084)

0.005

2.608

IV.

(0.079)

1.540

-0.033

0.130

(1.652)

deviations.

0.064 (0.069) (0.048) (0.098) the

franc

(0.057) dollars,

dollar

array

so

Germany

deposit

(0.097)

They

(0.095)

the column US 0.168

(the

0.319

-0.021

1.514

0.758

0.081

are

(0.113)

(2.223)

-0.115

(0.106)

-0.038

-0.114

would

is

(0.132)

(0.077)

completed

(0.092)

(0.128)

Belgium

dollar

(0.131)

(0.064)

also

have

measurem ent

column

0.803 0.151

globally

1.931

combinations

0.112 0.189 0.058 -1.765

(0.347) (0.373)

(0.428)

(7.289)

(0.432)

-0.311 -0.205

currency)

(0.303)

(0.253)

needed

disappeared,

(0.420)

Canada

national

(showing

(0.212)

approximately

but

the

be

may

the

Table

invariant

to

to

every the

0.074

-1.673

0.047

0.071

very

0.146

1.771

0.071

Deposits

(0.059)

(0.055)

(1.162)

0.118

hedge V

0.188

France

obtained

exchange

of

(0.048)

other

by

(0.068)

same

row

(0.069)

choice

of

(0.067)

(ie,

rates

(0.034)

one

US

(0.040) the number

-0.109

-0.057

-1.632

-0.136

exposures the Exchange

numbers as -0.250

(3.644)

(0.173) -0.140 (0.214)

would the (0.184)

(0.216)

(0.151) reference

(0.210)

0.439

-0.218

-0.264 Japan (0.106)

dollar

(0.126)

US

row

have

Rates)

been complementation

dollar

2.481

currency. to

invested

0.209

1.576

0.268

0.084

0.273

the

(0.076)

(0.067)

0.099

0.160

(0.082)

(1.600)

(0.094)

0.135

in

ie,

(0.092)

(0.055)

(0.095)

(0.047)

rates

if

one.

Netherlands

of

the

complements same. to

one)

Eg,

return

(3.78)

everything 0.007

national 0.407

0.167 -9.220

0.331 0.207

0.033

Standard

(0.180)

(0.157)

(0.224)

(0.194)

0.138

(0.222)

-0.048

United

exposure

the

(0.218)

on

had

been

(0.110)

stock

would

have

(0.131)

stocks to

of

deviations

Kingdom

are

US

come

0.269

-0.344

-0.244

-0.559

-0.483

2.621

0.086

-0.404

(0.158)

(0.181)

(0.147)

(0.183)

(3.087)

0.068

(0.128) out

(0.177)

(0.107)

the

measured dollar

(0.090)

Switzerland

securities

shown is

in of This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

954 The Journal of Finance overexposed56 to their respective currency. The United Kingdom, however, pro vides results which are difficult to rationalize. Japanese and Swiss stock are remarkably well diversified as far as their vulnerability to exchange rate changes are concerned. Finally, United States stocks are mostly exposed to the US/ Canada exchange rate but very little exposed to the overall posture of the US currency. We cannot offer a theory which explains these results, but it is conceivable that an international extension of Fama [57] would provide one. Specialized versions of the International Asset Pricing Model (15) have been submitted to empirical tests (Solnik [180] and Agmon [12]). The tests have been inconclusive both from a statistical standpoint and also in view of Roll's [160] general critique of such tests: the world market portfolio is an elusive entity which is probably badly proxied by any available index. In Solnik [180] the IAPM, which puts a price on the systematic risk measured against the world market portfolio, was tested against the alternative hypothesis that national factors (orthogonal to the world market portfolio) also receive a price. The absence of statistical significance was due to the large specific risks of individual securities and to the relatively small share of the variance of returns explained by national factors. For details on empirical tests, see Solnik [182] and the discussion by Dumas [45]. V. The International Structure of Interest Rates and the Forward Exchange Market In a world of L + 1 nations, we have so far succeeded in pricing all assets except L + 1 of them, and we take these assets to be the L + 1 local currency bank deposits or Treasury bills. As far as these assets are concerned, we have no choice but to use Asset Pricing Model (14) which, as we have already observed, is not directly testable with the usual data. Letting exchange rates appear explicitly, we have ri + Oi = rL+l + (1

-

1/am)(El(l - a1)W's4,,)/(1 -al)Wl + (1/atm)

EN

1 WkmSi,k; i = 1, ... , L (17)

where ri is the nominal interest rate on the currency i bank deposit; 6i is the expected value of the instantaneous rate of change of the exchange rate of currency i against the measurement currency L + 1; rL+1 is the measurement currency interest rate, so far denoted simply r; s, is the covariance of exchange rate i with national investor l's rate of inflation; and Si,k

is the covariance of exchange rate i with the translated return on asset k, including for k = n + 1 to N, the covariances with the exchange rates themselves.57

56 One hundred percent exposure would be the case where the array contained ones in the diagonal and zeros everywhere else. Such was implicitly the assumption in the original Solnik [179] model where local-currency rates of return were assumed independent of exchange rates. A one-dollar investment in German stocks would then simply be hedged by borrowing one dollar's worth of

Deutsche Marks. 57

si, is another notation for a' +1, and Si another notation for Un+1,k

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 955

Forward exchange contracts are redundant in our model.58 Were they to exist then, in the absence of impediments to arbitrage, the forward rate would be set by interest rate parity (IRP). Calling fi the percentage difference between the forward and spot rates (premium if positive, discount if negative), IRP implies -

fi= rL+j ri (18)

ie, the forward premium equals the interest rate differential. As a result, at equilibrium we have the so-called "Fisher open" or "uncovered IRP" relation ship:59 fi

=

-

-

i (1 - 1/am)[(EX (1 al)WWs4/X)/l (1 a')Wl] -

(1/a ) zk=l Wk Si,k (19)

or fi= i- [(E1 (1 - al)Ws4,)/Xl (1 - al)W'] -

(1/am) &N+1 Wkm[Si,k- (El (1 a')W's4,Ii_)/1El (1 al)Wl] (20) Equations (19) and (20) imply that

the forward rate is biased predictor of the ensuing spot. Repeating here a comment made previously in various contexts, the spread appearing in (20) between the forward rate and the expected spot is made up of two premia. One would exist even in the absence of risk

would

nonrandom

aversion (1/ am = 0) but disappear under measurement-currency inflation and Purchasing Power Parity (ie nonrandom inflation for all when all rates are expressed in the measurement currency). The mathematical and economic rea sons for its existence were given following Equation (13). Under risk neutrality, the expected value of the spot rate deflated must be equal to the forward rate also deflated.60 The covariance between the spot rate and the deflator (deflators 58 Selling francs forward is equivalent to selling borrowed francs spot and investing the proceeds in dollars. Hence the forward rate must be equal to the simultaneous spot rate corrected for the interest rate difference. Otherwise some arbitragers could reap instantaneous riskless profits (aside from default risk). We do not review here the evidence on the Interest Rate Parity relationships. See Frenkel and Levich [69], Herring and Marston [89] and, for a review, Kohlhagen [106] Section II. See also footnote 68 below.

69In the

Solnik-Sercu special case of no local inflation as seen by local investors, we have: S> = Si,n+t(= rn+i,n+l)

ie, translated nonmeasurement currency inflation rates behave like exchange rates.

60

ie:

where FE(1/I) = E(S/I);

F = forward exchange rate (in measurement currency units pe r unit of foreign currency); S = future spot rate quoted the same way; I = price index expressed in measurement currency units; and E = expected value operator Reasoning in real terms, as we do here, produces the inflation premium and also disposes of the so called Siegel

[177] paradox. Siegel argued that it is impossible simultaneously that F = E(S) and, after changing measurement currency, that 1/F = E(1/S) since by Jensen's inequality, E(1/S) > 1/ E(S). The paradox was quickly dismissed as a trivial mathematical inconvenience without economic or empirical significance. Actually, in our formulation, there is no paradox. Note that in the above equation, the price index I has a currency dimension. If we switch currencies around, as Siegel did, This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

956 The Journal of Finance

in the absence of PPP) is the source of this premium.61 The second premium in (20) is a plain risk premium and it is linked to the covariance of the exchange rate with "the" real return on the world market portfolio. This real return is computed using an average worldwide rate of inflation, where the weights in the average are, as in the first premium, the national wealths times one minus national risk tolerances. Again, the unobservability of the risk tolerance is the reason for the nontestability of formulae (17), (19), and (20). The world market portfolio composition is given as usual by the relative capitalizations of all the assets. It contains, therefore, all the assets which are not globally in zero net supply; Frankel [64] baptized these "outside assets." When all assets are "inside" (all in zero net supply) and returns are stationary, there is no risk premium; ie the formula reduces to what it would be if l/lam = 0.

This equilibrium model of the forward rate is useful in at least two respects. First, it provides a focus for a short review of the empirical literature concerned with the "efficiency" of the exchange markets. Second, it serves to reveal the shortcomings of the socalled "Modern Theory" which features in the conven tional account of the forward markets. There are no direct tests of Equation (20) and it is unlikely that there will be until the problem of estimating risk tolerance under diverse consumption pref erences is solved. Roll and Solnik [163] tested a very special version in which the second term of (20) was omitted and equal weights were used in the third. They were unable to establish conclusively that risk premia exist. Frankel [65, 66] must be credited for testing the hypothesis that the risk premium may fluctuate with the supplies of "outside assets" and specifically with the supply of govern ment debts (cumulated government deficits). But he was unable to produce evidence of a significant link with these quantities. What appears most frequently in the literature is time-series analyses of the nominal difference between the (logarithms of the) forward and subsequent spot rates, uncorrected for inflation. This difference reflects the nominal returns to forward speculation (sell forward, buy at the future spot) or, equivalently, the forecast error if one takes the forward rate as a predictor of the spot. The literature asks two related questions: is the expected forecast error zero and are forecast errors serially uncorrelated? Taken together, the hypothesis is that the we must also translate the price index (I -> IS). Then we have:

(1/F)E(S/I)

=

E[(1/S)/(I/S)]

which is equivalent to the first equation. Jadi tidak ada kontradiksi. To counter one referee's misunderstanding, note that we are not arguing here that the inflation premium is unimportant; quite the opposite; we say that it takes care of the Siegel paradox. 61

Solnik [179] overlooked this premium because he was missing the term 1 xa in his version of Equation (11).

As a result, his comments regarding the weighting of the various assets in the overall premium are incorrect. This is unfortunate: he had identified the weights with the net investment position of each country. Under his restrictive set of assumptions (recall from footnote 43 that he assumed not only absence of local inflation but also independence of exchange rates and local currency stock returns; so that: SZ, = Si,n+1 and Si,k = Si,n+k), his interpretation holds for the weights appearing within the second premium of (23) only.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 957

forecast error follows a random Walk.62 This provides an indirect test of whether the premia in (20) are zero. Evidence against the random walk hypothesis, which was at first the prevailing one, is mounting.63 Levich [120] discovered biases which were of opposite sign depending on the direction of change of the spot rate. These he attributed not to risk premia, but to transaction costs which penalize speculation and keep the forward below the expected spot when the latter is rising and vice-versa. Cornell [34], Geweke and Feige [74], Hansen and Hodrick [83], and Cumby and Obstfeld [39] all found in floating rate data, both post-World War I and in the recent past, instances where the unconditional mean bias was significantly different from zero, and whether it was zero or not, other cases where forecast errors were serially correlated. These results do not, of course, imply anything about the efficiency of the forward exchange markets although some of the authors above motivated their tests by appeals to the efficient markets hypothesis.64 They are, however, consistent with the existence of biases or premia, perhaps of the type described in (20), which fluctuate widely and in a serially correlated fashion. Indeed, Stockman [187] was able to find evidence of variable premia: when he split his sample into two subperiods, the premia which were not significantly different from zero in each subperiod were nevertheless significantly different from each other, for some currencies. There is need for a general equilibrium theory which would identify the exogenous determinants of the premia. Equation (20) further enables us to confront Tsiang's [196] theory of the forward exchange market, subsequently baptized the Modern Theory (MT) by Stoll [189]. The MT apparently still enjoys currency in the thinking in central banks. In its later manifestations, the MT was used to account for deviations from IRP, to justify the forward rate as a predictor of the future spot, and as an underpinning for official intervention, often righteously termed "counter specu lation." In the MT, forward transactions are contracted between two distinct and separate classes of traders. Speculators take open forward positions and link the forward rate to the expected future spot rate. Arbitragers demand forward contracts when the forward rate deviates from IRP and link the forward rate to 62 Absence of serial correlation does not imply a random walk. But, since the hypothesis is usually couched in the form of a regression model, the assumption of stationarity of the residual term is needed anyway for statistical purposes. The random walk model is really the one being tested. 63 See Kohlhagen [104] and Dufey and Giddy's [44] "submartingale" model. The literatures on the efficiency of the foreign exchange market prior to and until 1977 has been exhaustively surveyed by Kohlhagen [106]. There is no need to reproduce this work here. Also discussed there are some crucial macroeconomic questions which are related to the matter of efficiency but are too remote from our topic of international portfolio choice to be discussed here; these are: the impact of trade flows on the forward rate due to "hedging pressure" (Levin [121], Dooley [40]); the impact of official intervention (Kohlhagen [105]); and whether speculative activity can be destabilizing ("bandwagon"

effects) and, if so, what measurement method would allow us to identify the periods where it is.

64

By itself,

market efficiency does not imply that the forward rate is equal to the expected spot. This used to be a frequent misconception. In the presence of risk aversion, or nonconstant required real returns, there is also no implication that market efficiency leads to serially uncorrelated returns. See Lucas [129]. But in practice, risk aversions are sufficiently low as to produce very low serial correlations. Witness the autocorrelations computed by Fama [56] on New York Stock Exchange data.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

958 The Journal of Finance interest rates. Hedgers protecting previously-established exposures cannot be modeled in the MT. While their motives are akin to those of arbitragers, their calculations are like those of speculators and they are therefore lumped with the latter. The key flaws in this specification are the assumed specialization of traders and the assumption that interest rates in the MT are purely exogenous, set by the whims of central bankers independently of expectations.65 When IRP prevails, the equilibrium forward rate in the MT is therefore also exogenous. At IRP, the arbitrage demand is infinitely elastic. Any shift in speculators' expectations of the future spot, due say to government counterspeculation, leaves both the arbitrage schedule and the IRP forward rate unchanged. The arbitrage volume adjusts automatically to meet the speculative demand leaving the false impression that governments can successfully induce capital flows by forward intervention. Due to the exogeneity of interest rates, speculators' and arbitragers' demand curves for forward contracts are, implausibly, perfectly independent. In contrast, there are no separate classes of traders in the theory leading to Equation (20). Investors are identifiable by their price deflators, not by their transactions motives. Their portfolio problem has been solved globally: their demand for forward exchange66 is an indistinguishable component of their vector asset-demand function. Their motives may include arbitrage, speculation, and hedging when the latter is identified with the need to diversify. Any linear decomposition of forward transactions according to purpose is, however, essen tially arbitrary.67 The probability distribution of the exchange rate is simultane ously a determinant of both speculation and diversification. Because IRP holds, the arbitrage demand is potentially infinite but no arbitrage flows, accommodat ing or otherwise, will actually occur. More importantly, interest rates are endog enous. Equation (17) makes clear that, given expectations regarding future spot rates, interest rates in various currencies cannot be set independently: conversely, interest rate differentials reflect anticipations of future spot rates. The forward rate in (20) is therefore jointly at IRP and equal to the certainty equivalent of the future spot rate. In the MT, the equilibrium forward rate generally falls between the exogenous IRP level and the expected future spot rate.68 To forecast the future spot, all one 65 There are others, quite aside from the difficulty with hedgers. First, speculators will not bet on the difference between the forward and expected spot rates unless they are risk neutral or, if they are risk averse, unless forward contracts are the only available risky asset. Second, in creating the link between the forward market and capital flows, the MT confusingly identified the spot transactions associated with covered interest arbitrage (borrow one currency, sell it spot, invest in a second) with "hot," interest-sensitive, short-term capital movements. In reality, these would not be covered by buying forward the borrowed currency and therefore, unlike arbitrage, will tend to be independent of the forward rate and of deviations from IRP. 6 This is a figure of speech. As noted in footnote 58, forward contracts are redundant instruments when borrowing and lending is allowed in all currencies. Hence we never introduced (and could not have because of the perfect substitutability) a separate demand for forward exchange. In the present context the expression refers to the equivalent demands for bank deposits or loans. The situation will be different below when arbitrage becomes risky.

6 Adler and Dumas [5 and 6] showed that the additive separation of the speculation and hedging purposes is possible with a quadratic utility function which generates a linear marginal utility function.

'

To account for IRP deviations at equilibrium was one of the MT's design objectives. Nowadays, This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 959

does is to compare the concurrent actual and IRP forward rates. This demon strably false proposition is the mechanical result of the MT's identification of forward equilibrium at the intersection of the speculator's inelastic demand curve (with its intercept at the expected spot) and the arbitragers' inelastic supply curve for forward contracts (with an intercept at IRP). The majority of the authors cited associated the finite elasticities of the two curves to traders' aversion to a notion of default risk which rises with market volume. To refute the MT's prediction, Adler and Dumas [5, 6] and Kouri [107] modeled the portfolio demand for forward exchange in the presence of an exogenous default risks on both forward conctracts and banking transactions, which made interest arbitrage risky. The results generalize Equation (20). Interest rates and the forward rate are endogenized: both are functions of spot rate expectations.69 IRP may be violated due to the default risks. The demand for forward contracts cannot be decomposed additively among the motives: rather the arbitrage motive acts multiplicatively on the others. And above all, the forward rate need not be bracketed by the IRP rate and the expected spot. Following the demise of the MT, the theory leading to Equation (20) has taken its place in the tool kit of macroeconomists. It produces as a special case the model, mentioned in Section III, which was used by Krugman [113] to suggest a link between exchange rates and the current account.70 This link is directly apparent in Equation (17). Assume 1 - ai > 0 and ai = a L+l and take some

wealth from country L + 1 and transfer it to country i. This modified situation may be the result of a history of large current account surpluses or smaller current account deficits for country i at the expense of country L + 1.71 The (comparative-static) transfer induces a drop of the interest differential (ri - rL+1) in favor of currency i or a drop of its expected rate of appreciation (06).72 The the daily press reveals that IRP holds to within very narrow transaction cost tolerances in the Eurocurrency interbank dealer market even during turbulent periods. Nevertheless, most, of the papers in the area presume that IRP is violated. Many reasons have been given: transactions costs (Branson [27], Frenkel and Levich [69]; institutional constraints (Einzig [52], Sohmen [178], Canterbery [32]; interest rates functionally related to the volume of arbitrage (Prachowny [153], Frenkel [67] and Kenen [102]; default risk in arbitrage transactions (Stoll [189], Grubel [81]) or political risk (Aliber [14], Dooley and Isard [41]). Deviations from IRP are frequently observed in comparisons of domestic money market or local government T-bill rates. Internal markets may therefore be segmented while the offshore markets are not. IRP violations depend on where one looks. 69 In one section of Kouri [107], interest rates are taken to be exogenous but then restrictions correctly follow on the expected behavior of spot rates. The fact that interest rates, as much as forward rates, reflect exchange rate anticipations have been noted before these formal models were constructed. See Branson, Katz, and Willett [29], Pippenger [150], and th e literature reviewed in Kohlhagen [106], Section llc. 70

Via a wealth effect rather than an arbitrage effect.

7' Larger wealth may also be the result of capital gains on country i's preferential holding of home currency assets. These gains would, however, induce a current account deficit. 72 This conclusion requires a world where local inflation rates are negligible so that sfr,Z = 0 and Si = Si,n+i is the variance of the exchange value of currency

i and is therefore positive. Further the market portfolio of stocks is the stock component of the logarithmic portfolio (wm/am = Wk,10g; k = 1 ... n) which must be held constant. And, the market portfolio of bonds is zero if these are in zero net supply. Simple examination of Equation (17) then produces the stated result. Note, however, that the conclusion may be mitigated or strengthened by changes in the logarithmic portfolio induced by the lower value of r, + 0i - rL+l. This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

960 The Journal of Finance

dynamics of wealth accumulation and exchange rates are yet to be worked out in keeping

with this

observation.73

VI. Some Welfare Questions Associated with Exchange Risk The function of capital markets is the allocation of the bearing of risks. Among these, the risk arising from holding nominal assets is of special interest in international finance. To define terms, recall the development of Section I. A resident of a country holding a foreign security with a given probability distri bution of foreign-currency rate of return would compute his real return by first translating into home currency and then deflating by means of the home price index. Consequently, the currency risk arising from holding a nominal asset denominated in foreign currency (ie paying a fixed amount of foreign money) is linked to the randomness of the exchange rate times the home purchasing power index. Exchange risk, which may be identified with the randomness of the exchange rate, is never borne alone but only in conjunction with home purchasing power uncertainty. The currency risk associated with holding a nominal security denominated in home currency is simply the randomness of the home purchasing power index, if any. The question is under what conditions these currency risks affect welfare and how they are allocated across individuals. We first consider a world in which money is only a unit of account but is not held in the investors' portfolios and is issued by no one. Exchange rates are then arbitrary (random) numbers, exogenously given, translating one measurement unit into another. It should be clear that the multiplicity of benchmarks for value by itself has no impact on welfare. If there is a welfare issue, it arises from the presence in the financial markets of nominal securities, ie, securities whose payoff is linked to the fluctuations in the purchasing power of one monetary unit of value. The issue, then, is in what circumstances would such securities be willingly held by investors at equilibrium; for, if they were not held, the random ness of exchange rates and price levels, holding constant the physical payoffs (outputs) on the other securities, could not possibly have an impact on welfare. When the financial market allows individuals to trade risks and insurance in every conceivable dimension of their choice, the allocation of risk bearing is Pareto optimal. In that case, a strong assertion of financial theory is that all consumption risks are mutualized. Consider a random event which is to cause an individual to lose something to the benefit of another individual (a zero-sum risk). Then it would be optimal for these two risk-averse persons to precontract: the first person would buy insurance from the second one and pay him a fixed premium and, if the risk materializes, its effects would be cancelled by invoking the insurance policy. All personal risks would disappear in this manner and in the end individual consumption (assumed to be the only source of utility) would only be a function of aggregate consumption. Individual investors, that is, would only bear the impact of social (aggregative) risks. This function,

relating individ 7 This current account argument may seem reminiscent of the "hedging pressure" theory wherein trade flows induced trading firms to hedge in the forward market, thereby affecting the equilibrium rate (Levin [121], Dooley [40]). But really it is not, as we are referring now to the accumulated current account and to the stock of wealth resulting from it rather than to trade flows.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 961 ual and aggregate consumptions, is called an optimal sharing rule.74 If there are several goods, each person's consumption of each good is a function of the aggregate consumption of all goods. If the investor's utility functions are of the von Neumann-Morgenstern type, the sharing rules are nonrandom functions. Consider now a pure exchange economy (aggregate consumption of each good equals its aggregate exogenous output) endowed with a Pareto optimal capital market and where the various moneys are only units of account.75 In this economy,76 the only securities which would be held are those which achieve the optimal sharing rule, ie, which serve to allocate the risks of aggregate output of the various goods. But, the various currency risks are not aggregate risk; they arise only in connection with the holding of nominal securities whose payoffs are linked to no underlying physical output.77 For every borrower there is a lender. The gain of the one is the loss of the other. And, at equilibrium, there will be neither nominal borrower nor nominal lender; nominal securities will not be held.78 This is an important conclusion and a fairly robust one provided one clarifies the phrase "nominal securities will not be held." If a nominal security is issued by a corporation and the proceeds are used to repurchase stock, it will, in the present setting, be bought by the stockholders of that corporation (Modi gliani-Miller [141]). This is not to be seen as an exception to the above statement. If a government issues nominal bonds to purchase claims on future output, the bonds will be purchased and the claims on output sold (possibly short) by the taxpayers of that government (Wallace [200]).79 This is not an exception to the above statement either. In both cases, stockholders or taxpayers will earn gains or losses on their dividends or on their tax bill which exactly offset their losses or gains on the holding of nominal bonds. If we switch, however, to an exchange economy in which the capital market is not Pareto optimal because of restrictions in the array of tradable securities, then some personal risks will not be hedgable. In that case, capital market participants will look for proxies, ie, securities which are correlated with the risks they wish to hedge. In general they will make up a portfolio of all the available securities and compose it in such a way as to achieve the best possible correlation. To the extent that nominal securities are available, they will be used for these purposes and will generally be held.80 Increased currency risk could then reduce or improve welfare. 74 See Rubinstein [165]. The concept is identical to the one of contract curves in an Edgeworth box formulation where the edges of the box would represent consumption in the various states of nature. 75 We must add the following purely technical assumption: there is no nominal security whose payoff in terms of one commodity is perfectly correlated with the aggregate consumption of that commodity. This is only meant to avoid indeterminancy.

76This is the kind of economy considered by Grauer, Litzenberger, and Stehle [79]. We do not introduce, however, their assumption of identical consumption tastes for all individuals, which leads to PPP holding exactly (ie, ex post: see Section I). 77

See infra the case of corporate or government bonds.

78 Unless they are also real in the eyes of some investors, as in the Solnik-Sercu special case. 79 The government budget deficit (on current account) bites into consumable output and as such constitutes a nonhedgable aggregative risk. The reasoning in the text keeps the deficit constant. 8 In the model of Sections III, IV, and V the market was generally not Pareto optimal and nominal assets were indeed held at equilibrium. In the SolnikSercu special case the market became Pareto optimal but nominal assets also became real and were held for that reason.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

962 The Journal of Finance

Therefore, it appears that currency risks matter or do not matter depending on conditions in the capital market. This was the conclusion reached by Grauer, Litzenberger, and Stehle [79] but in the context of an economy where consumers had identical tastes for the various commodities and PPP prevailed. It should be clear that their conclusion owes nothing to this assumption. What is crucial, however, to the reasoning is the assumption that, as one varies currency risks, one leaves unchanged the probability-distributions of the consum able outputs which are the real aggregate risks to be borne by investors. This assumption may not be tenable in some settings. For instance, in the Solnik Sercu special case, where local currency inflation in each country is zero but tastes differ, Grauer, Litzenberger, and Stehle make the valid point that, if CPP prevailed for each good, then exchange rates could not fluctuate and currency risks would not exist unless the relative prices between goods and their outputs fluctuated; and this in turn implies that one could not vary currency risks without varying output risks. In that case, currency risks may be said to matter, but only because they vary in step with output risks.81 The juxtaposition of the Grauer et al. model where PPP was assumed and currency risks did not matter, with the Solnik model where PPP did not prevail and currency risks mattered, led some astray. It is often stated in the literature (Aliber [15, p. 106], Jaffee [98], Cornell [36]) that, with PPP, exchange rates are not linked with relative commodities prices or that their variations are offset by price levels, so that exchange risk becomes a purely nominal uncertainty which matters to no one. The result is to identify the exchange risk which matters (sometimes called "real exchange risk") with deviations from PPP, and the exchange risk which does not matter ("nominal exchange risk") with fluctuations in the PPP level of exchange rates (ie, the ratio of price indices). This seems wrong for the above reasons and also because the risk of PPP deviations is not a separate risk which is to be borne by anyone: the foreign deflator does not enter any domestic investor's risk calculus. PPP deviations, as distinct from variations in the purchasing power at home of domestic and foreign securities, will therefore not affect any investor's financial decisions. At some point, one must stop thinking in terms of price levels and exchange rates reflecting only exogenous random changes in measurement units, and come to grips with the fact that moneys are held by households and issued by Central Banks. But, at that stage, one must also realize that the question of the relevance of exchange risk becomes illformulated since exchange rates and price levels are endogenous. The issue then becomes that of the welfare impact or non-neutrality of monetary policies in a multi-currency world. It is a very complex one, for which few statements remain valid outside a particular context or model formu lation. The vast macroeconomic literature which addressed this issue recognized at least four channels of influence:82 (a) The rate of monetary expansion affects the nominal rate of interest which

81 Two other instances of links between consumable output and nominal quantities will be encountered below and in Section VIII: monetary policy may have a real effect and the outputs and sales flows of firms may be affected by eg, a change in the selling price abroad of the finished product, relative to the production cost at home. 82 In addition, government's fiscal policy directly influences the amount of output available for consumption as the budget deficit (on current account) is a drain on real resources. The deficit may

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 963

influences real money balances, consumption, and the demand for securities (savings). If output is kept constant, a rise in the nominal rate of interest increases the cost of the liquidity services of money and lowers welfare (Bailey [17], Barro [21]). If output is allowed to vary, the conclusion depends on the setting chosen (Mundell [146], Fischer [63]); (b) Monetary policy alters the relative stocks of available assets. As their prices adjust to maintain market equilibrium, the discrepancies which appear between the market prices of physical assets and their replacement costs affect aggregate demand and have therefore real effects (Tobin [194]). This argument is independent of the liquidity services of money. It deals only with the compo sition of the government's versus the private portfolios. Some irrelevance prop ositions are being developed (Wallace [200], Chamley and Polemarchakis [33]) which tend to invalidate it; (c) Random monetary policies induce equivocation in price signals (Lucas [128], Sargent and Wallace [170], Santomero and Seater [169], Weiss [201]), leading to a short-term inflation-employment trade-off; (d) Exchange rates affect export competitiveness and employment (Laursen and Metzler [115]). Capital market theory is not, so far, capable of incorporating these effects into a generalequilibrium framework (see Lucas [130, 131] and Helpman and Razin [88]). Attempts at introducing moneys into portfolios have been made in the partial equilibrium context of the capital asset pricing model: Kouri [108], Stockman [187], Fama and Farber [58], Landskroner and Liviatan [114], Staple ton and Subrahmanyam [184], Poncet [151], and Dumas [49]. Some limited welfare issues are discussed in Fama and Farber [58] and Dumas [49]. In both of these, money is a separate argument in investors' utility functions and is held because it yields liquidity services. Fama and Farber [58] make the point that $1 of money and a nominal (short-term) bond paying $1 carry the same currency risk. Hence a "separation" exists between the decision to hold money versus nominal bonds and the decision as to the composition of the remainder of the portfolio. People may decide how much they want to hold of nominal assets in their overall portfolio and then divide this amount between money and bonds depending on the amount of liquidity services they wish to use (the cost is as usual the foregone nominal rate of interest). The implication is that the presence of money does not increase or reduce the amount of currency risks people have to bear, as compared to the situation where the government would only issue nominal bonds to finance its purchases. The point is quite general; but, in order to illustrate, consider the simplest case analyzed in Dumas [49] where financial markets are Pareto optimal and money is issued to make transfer payments (or reduce taxes). The random benefits to transfer recipients are exactly equal to the purchasing power losses of money holders. Pre-contracting between the two overlapping groups of households can occur: transfer recipients may borrow in a state-

contingent way from money holders and, in exchange, remit the transfers to them later.83 This possibility makes it unnecessary in this case for anyone to be financed by money creation, thereby creating a link between consumable output on the one hand and prices and exchange rates on the other. 83 If the amount of the transfers are fixed nominally, the market need not be a complete one: nominal bonds of maturities can be used for the purpose.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

964 The Journal of Finance

bear the purchasing-power risk of currencies. While in full agreement with Fama and Farber on the matter of currency risk, Dumas [49] points out that they overlooked another risk of a monetary origin: the risk of fluctuations in the nominal rate of interest.84 As has been recognized by macroeconomists (see the monetary channel of influence (a) above), the opportunity cost of holding money balances, measured by the nominal interest rate, is a determinant of aggregate welfare (along with aggregate consumption) and, of course, the next period's nominal interest rate is also a separate argument of the optimal sharing rules. Because nominal interest-rate randomness is an aggregate risk, it cannot be pre-contracted away and each investor must bear a share of it. While the above discussion has de-emphasized PPP deviations as a measure of exchange risk (especially those arising from differnces in investors' consump tion tastes), the matter may be different when CPP is violated and individual commodity prices are misaligned across the world. CPP deviations are sympto matic of barriers to free trade. It is then not clear how investors trading contingent claims in a supposedly integrated world capital market will take receipt or make delivery of the physical payoffs resulting from their bets. The notion that the world-wide aggregate amount of each good constitutes a pool of freely allocable resources must be called into question. The concepts of, aggregate consumption and of sharing rules may well lose their meaning. To take the issue further requires a model capable of accounting for the CPP deviations. Anticipating the results of such a model, it will undoubtedly remain improper loosely to identify currency risks (or the part of them that matters) with CPP (or PPP) deviations. There is a presumption that the wider the CPP deviations, the larger will be the amount of welfare foregone as a result of insufficient or inefficient trading. But one can only speculate as to how widely-fluctuating, random CPP deviations will be linked empirically to variability in exchange rates or to their product with domestic price levels. Whether increased exchange or currency risks reduce welfare is an open question. VII. Segmentation Segmentations of the international commodities markets which produce CPP deviations may disturb the worldwide allocation of risk: because the goods market is partly segmented, so is the capital market.85 Independently, however, capital markets may be separated along national lines owing either to investors' inhibi tions or official restrictions. Investors may be inhibited by a lack of information, the fear of expropriation or, more generally, discriminatory taxation.86 Official restrictions may include exchange and border controls which restrict foreigners' 8 Technically, the reason they overlooked it is that they worked with an indirect utility function of current consumption, current balances, and end-of-period wealth. This is incorrect: when money is held, the indirect utility cannot be a function of future wealth alone; it also depends on the future nominal interest rate.

85 The CPP deviations need not be random for this to happen. When investors encounter difficulty in collecting payoffs in consumable form, they may be deterred from investing in the first place. 86 The randomness of exchange rates is not to be conceived as a source of segmentation since by proper hedging, repatriated payoffs may be rendered independent of exchange rates.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 965

access to local capital markets, reduce their freedom to repatriate capital and dividends, and limit the fraction of a local firm's equity that they may own. These manifestations of sovereignty serve in part to define nations as distinct segments of the international capital market. As a phenomenon, segmentation is not unique to the international arena. It has received attention also at the domestic level, from Rubinstein [166] and, more comprehensively, from Lintner [126]. Capital market segments in these papers generally consist of groups of investors and of groups of securities which each investor class is allowed to trade. The multiplicity of possible groups of investors and securities led the authors to define gradations ranging from com plete segmentation to partial segmentation with overlap.87 The main objective and contribution of these models was to show that a sufficient amount of diversity among investors in different segments would lead to firms having optimal, value maximizing, interior capital structure decisions. The cost is that separation properties generally break down: portfolio separation for individuals and the independence of capital budgeting and financing decisions for firms. Lintner's encyclopedic treatment attempts parenthetically to establish conditions in which separation is restored. Neither paper questions the existence or uniqueness of equilibrium in segmented markets or whether the risk allocation will be Pareto optimal. Both assume that value maximization will be unanimously supported irrespective of the type of segmentation being postulated. Whether this last assumption can be maintained, at least to an approximation, is a question that continues to bedevil this strand of the literature. Despite an absence of empirical justification, it is possible that segmentation can safely be ignored at the domestic level. Most authors do. Internationally it is harder to avoid if only because, from time to time and place to place, governments try to insulate their capital and goods markets from the rest of the world. To the extent that segmentation exists in practice, international corpo rations may be able to play an important role by recognizing the causes of the segmentation and by planning transactions to enable their stockholders to reap the welfare gains from integration.88 When stock markets are segmented, for example, home-country firms can purchase shares in foreign firms and provide an indirect diversification as well as a rate-ofreturn arbitrage service. Adler [1] and Adler and Dumas [3] calculated the valuemaximizing foreign acquisition and the resulting home and foreign market equilibria when both markets sepa rately are described by a CAPM. They calculated that the value maximizing foreign acquisition would not be welfare maximizing from the viewpoint of home investors. This point was pursued by Lee and Sachdeva [116]. They pointed out that what produced the result was the implicit assumption that home firms had monopoly power in the home capital market. When these firms behave at home as pure competitors, as in Ekern and Wilson [53], home welfare is maximized

87 Rubinstein further remarked that segmentation might not be an essential concept. Incomplete diversification of domestic portfolios can result from other causes such as nontraded assets, default risks in borrowing, taxes, and transaction costs. Internationally, however, such factors may operate to separate financial markets from each other. We do not therefore attempt any further distinction between segmentation and the imperfections that may cause it.

8

The welfare gains from integration have been evaluated by Subrahmanyam [193].

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

966 The Journal of Finance

while the welfare of investors in the host country is generally minimized. Their work contributed an essential insight. When conditions in the domestic market are those leading to the Modigliani and Miller theorem, there is no optimal foreign acquisition decision for any single, individual home-firm. What exists is an optimum for the total amounts of foreign shares to be acquired by all the firms in the home-market segment. The allocation of this total acquisition among home firms is irrelevant for investors holding a diversified home market portfolio. Stapleton and Subrahmanyam [133] calculated numerically optimal decisions and capital market equilibrium in a variety of stock market settings. Adler and Dumas [4] established the principle that for each kind of segmentation there is a corresponding domestic valuemaximizing decision for unrestricted home firms: when the bond market is segmented, there is an optimal foreign versus home borrowing decision; and there is an optimal forward contracting decision when stockholders' access to the forward exchange market is restricted. As before, these decisions will not be firm-specific if the domestic market is free of imper fections and bankruptcy costs. The acquisitions of foreign shares that firms make for the purpose of providing their owners with international diversification should be distinguished clearly from private foreign direct investment (PFDI), although they might accidentally be so classified in the balance of payments statistics. For one thing, PFDI involves the purchases of control, a consideration excluded from the acquisition decisions discussed above. Segmentation among the financial markets does not seem particularly important in this connection.89 Most explanations of MNC's behavior and of synergy, however, appeal to imperfections in, and segmentations of, not the financial markets but of the markets for products, factors, and technology. To paraphrase Kindleberger, direct investment falls more within the province of industrial organization and monopoly theories than financial theory.90 Nevertheless, the fact that MNC's foreign operations can be viewed as (more 89 Ragazzi [154] offers an intriguing theory based on segmentation within countries which are financially underdeveloped. In such countries, he hypothesizes two capital markets, one for the trading of very large, controlling blocks of shares and one for the trading of minority holdings. Expected returns for a given level of risk would be higher in the market for control. Ragazzi suggests that MNC's might finance themselves in the minority market, repackage the funds, and enter the

oligopolistic market for controlling interests, thus reaping the difference in rates of return. 90 full review of the determinants of PFDI is beyond the scope of this survey. Segmentation of product markets underlies Adler and Stevens's [10] analysis of MNC's exporting and investment de cisions. Early accounts of the spread of international investment, summarized in Hufbauer and Adler [95] and elsewhere, emphasized the migration of labor-intensive industries to low wage countries: such investment bridges a segmentation in the factor market. By the same token, however, capital intensive industries should concentrate in, and preferably export capital intensive goods from, countries like the US where the cost of capital may be relatively low. This proposition has proved empirically questionable. The failure of the factor proportions account led to the analysis of PFDI as a

channel for the transfer of technology: Hufbauer [92, 93, 94] and Vernon [199]. The product life cycle account follows from the view that the market for technology is segmented and monopolistic. This notion also underlies the influential proposal by Johnson [99] and Magee [132] that MNC's be modeled as monopolistic producers, not of goods, but of information. Their efforts to appropriate the externalities tend to segment the information market between private information held by the innovating firms and public information. Other explanations of PFDI resort to economies of scale in the production of goods and information or in the ability of large firms to negotiate concessions from host governments. Few feature financial market segmentations as key contributing factors.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 967

or less diversified) portfolios of controlling shares raised the hypothesis that they may offer a (partial) diversification service. Can one replicate true international diversification by purchasing a portfolio of MNC stocks? Agmon and Lessard [13] regressed the returns of 217 US MNC's on the US index and an interna tional factor. In second pass regressions, they found the coefficient of the world factor to be correlated with a sales measure of MNC's international involvement and, therefore, suggested that perhaps one can. However, using a sample of 40 European and 23 US firms between 1966 and 1974, Jacquillat and Solnik [97] concluded that one cannot. Basically, a multiple regression of MNC returns on various national indices showed that these firms correlate highly with their respective national indices and very little with foreign stock markets. These negative results were confirmed by Brewer [31] and Senschak and Beedles [173]. The issue is not easy to settle empirically, as noted in Adler [2]. If investors can diversify costlessly into the shares of foreign firms with equal access to the same projects as US MNC's, geographical diversification of projects ceases to be a service that MNC's can valuably perform on behalf of their stockholders. It may be impossible to detect diversification benefits if MNC's operate in markets where individuals also can trade. If any "foreign investment" effect is to be observed then, based on the theory of PFDI, it is likely to be the result of MNC's monopoly advantages abroad. Errunza and Senbet [54] independently followed this reasoning. They found that a measure of MNC's monopoly returns, ie, market value minus replacement cost, was significantly correlated with a sales measure of MNC's foreign involvement. The clear implication is that it is hard empirically to unravel the effects of monopoly from potential diversification benefits. While segmentation within and among capital markets may not be central to explanations of PFDI, the possibility that it exists is important and perhaps crucial in connection with the analysis of corporate financial decisions. Unfor tunately, there is as yet no definitive empirical method for determining whether and to what extent the international capital market is segmented. There appear to be four conceivable avenues of investigation, not all of which have been followed in the literature. The first is misguided. Section II reviewed several studies of the correlations between national stock markets. A theme sometimes encountered in these papers is that low correlations indicate segmentation on the grounds that integrated national markets would tend to fluctuate together. This inference, however, is incorrect. There are national random factors (politics, etc.) which affect selec tively the production activities of any one country. They are reflected in stock returns but this is no evidence of segmentation. Further, output mixes vary considerably among countries partly as a consequence of the specialization induced by international trade. Random shocks may affect selectively specific industrial sectors. They may, therefore, have a relatively heavy impact on those stock markets where these sectors are large but not in others. Small correlations among national stock market indices are generally consistent with perfect capital market integration.

Prospectively, a better approach for detecting segmentation is to analyze the correlations among national consumption rates. As was pointed out in Section This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

968 The Journal of Finance

VI, consumption risks (as opposed to production risks) are mutualized in an integrated and Pareto optimal capital market. Consequently, small random (unanticipated) fluctuations in national consumption rates should in such a market be perfectly correlated with the aggregate random consumption rate. There are some difficulties in defining aggregate consumption risk when differ ences in tastes lead to PPP deviations. These are not impossible to overcome.91 The problem of CPP violations may, however, be more severe for reasons already given at the end of the previous section. The third possible method is, like the first, based on an analysis of securities' prices. Rather than trying to infer segmentation directly from the correlation structure, however, the idea is to derive competing capital asset pricing models, with and without segmentation, and to confront them with data to see which one fits best.92 Stehle [195] attempted to test the hypothesis that the US market is completely isolated against the null that it is completely integrated with the rest of the world. To avoid the problems associated with PPP deviations, he assumed a world of logarithmic investors (ai = 1 and, therefore, atm = 1 in Equation (17)). With complete segmentation, the proper market portfolio for US investors would be represented by the US index whereas under complete integration a world index would be appropriate. Unfortunately, his empirical evidence did not significantly discriminate between the two competing models.93 The last approach, which also relies on CAPM concepts, was initiated by Black [24]. A variation and extension was recently proposed by Stulz [190]. Rather than postulate the extremes of either complete segmentation or none at all, these papers employ a continuous parameter of segmentation in the form of a propor tional tax. In Black, the tax is on an investor's net holdings (longs minus shorts) of risky foreign asets while in Stulz the tax on both long and short positions is positive. Borrowing at home and abroad is riskless and untaxed in both models. This apparently minor difference in specification, nevertheless, produces two quite different CAPMs with a common feature: the world portfolio will not be efficient for any investor in either one. The segmentation test in each case would essentially consist of fitting the derived CAPM to stock price data and either estimating the value of the implied tax rate or detecting its effects via security market line analysis. So far, this approach remains in the realm of theory. 91 Breeden [30] is credited for pointing out that the mutualization of consumption risks implies the perfect correlation of consumption rates. Admittedly, (Stulz [191]) consumption mixes vary across the world, leading to PPP deviation, and destroying the perfect correlation. But examining correlations of consumption rates of individual goods would eliminate the problem. Such a route was not open to the authors who studied the stock market returns: examining industry rates of return would have resolved the specialized issue but would not have dealt with the issue of country-specific

production risks (which are not supposed to be mutualized). 92 An alternative method, in the same spirit, would use the Arbitrage Pricing Theory (Roll and Ross [162] and test whether random factors which are common to stocks of different countries receive the same price in the different national stock markets (an APT in the absence of PPP would be needed for this purpose). This suggestion was made by one referee. We are thankful to him.

93 Solnik [182] following the Roll [160] methodology argues in favor of comparing actual and optimal portfolios (the latter computed on the basis of actual returns) rather than comparing the actual return statistics to their theoretical values (19). He claims that optimal aggregate portfolios under segmentation and under integration would not be sufficiently different (because of the low correlation of returns across countries) to permit a clear-cut conclusion.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 969

VIII . International Corporate Financial Decisions: Hedging Policy The discussion of the previous section clearly implies that the analysis of international corporate financial decisions rapidly becomes problematic once capital market imperfections and segmentations enter the picture. As these are often hard to ignore in practice, the literature in this sub-field is correspondingly slim. We shall therefore focus on the subject which has received the most attention, hedging policy, after the briefest of reviews of other decision problems. A few papers address longterm decisions. Mehra [137] confirms that all the Modigliani and Miller (M & M) propositions continue to hold in a perfectly integrated, two country capital market with random exchange rates, identical investors but no (or identical) taxes and no inflaton. The capital budgeting criterion in such a world is independent of the choice of measurement currency, the nationality of the investing firm, and of the financing decision. The same conclusion regarding investment and financial decisions will also emerge, how ever, from any of the (quasi-) complete-market, tax-free international asset pricing models such as Solnik's [179], GLS's [79] or, for that matter, our own in Section IV above. Adler and Dumas [4] and Senbet [172] introduce different tax rates at home and abroad with the result that there emerges a value-maximizing foreign borrowing decision, the investment and financing decisions become interdependent and planning becomes a programming problem. Adding other sources of segmentation does not change this general picture but complicates it considerably. Exact solutions depend very much on the specific provisions of the assumed tax regime and on the specific constraints imposed by the assumed imperfections. Lessard [119] offers a pragmatic compromise which treats the decisions separately. One would like to know how good an approximation his procedure is. Exchange rate variations can affect firms along several dimensions: through their impact on short and long-term monetary assets and liabilities and on physical assets; and via their effect on the volume of sales and the associated production plans. Exchange risk is but one of many environmental risks with which firms contend. Isolating its effects is a matter of decomposing the varia bility of some measure of the firm's results among the various risk sources. Once the extent of the firm's vulnerability to foreign currency risk is determined, it may readily be modified using financial hedging instruments such as forward contracts, swaps (borrowing one currency and lending the proceeds in another), and a variety of insurance schemes. Two questions then arise. Should a firm hedge at all? If there are circumstances in which it should, then by how much? While these issues are by no means completely settled, we may review the progress to date and the problems that remain. In a perfectly integrated world capital market like that of Section IV, where the M & M propositions all hold, Baron [19] and Dumas [46] show that corporate hedging, like the choice of debt versus equity, is irrelevant regardless of the firm's exposure. In such a world,

firms need not do what investors can do equally well for themselves. This irrelevance proposition is independent of the existence of risk and inflation premia in the relationship between the forward and expected future spot rates. The equilibrium forward rate in Equation (23) is set precisely This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

970 The Journal of Finance

at the level where no value is gained or lost by hedging. When a firm hedges, there is a change in its risk posture which is precisely offset by a value-preserving change in its expected return. What the proposition depends on, in other words, is the dual assumption of symmetric information and the (quasi-) completeness of the world's financial markets. Integration of the foreign exchange and stock markets guarantees that the same (linear) valuation functional is used in both. This allows the value additivity principle to be applied: the value of a firm with a forward contract is equal to the value of the same firm without a contract plus the value of a contract. In the absence of transactions costs, forward contracting involves no exchange of money between the parties: the forward rate adjusts until the value of the contract at the time it is entered is exactly zero. The values of the hedged and unhedged firms are therefore equal on the hedging-decision date. We should emphasize that the market conditions which guarantee that this variant of the M & M theorem will hold are the only reasons for hedging to be irrelevant. Hedging is also claimed to be unnecessary by followers of Aliber [15] on the generally faulty ground that the net gain from it (equal to the forward rate minus the ensuing spot) is sometimes positive and sometimes negative and tends to zero over long periods. Aliber's own argument was empirical and more restrictive: he observed that the average deviation of the forward from the future spot rate tended towards zero as the number of observations included in the average increased.94 He proceeded to deduce that long-term nominal foreign currency assets are not exposed in the long run, and therefore that these assets, at least, need not be hedged. The argument is flawed for several reasons. The most important is that it ignores risks whose expected values are zero. It is wrong to base on reasoning dealing with averages the theoretical argument that risk avoidance is irrelevant.95 There is no reason to suppose that stockholders or corporate managers are risk-neutral and care only about long-run expected values. If hedging is useless, it can only be for the reasons given in the previous paragraph. As is the case with most financial decisions that do not matter in M & M's theory, practitioners do devote time and resources to hedging exchange risk. Even those who ignore balance sheet translation exposures substitute some other target. What accounts for this activity? One reason may be that the market is not as integrated as the theory requires. Segmentation and other imperfections, including bankruptcy costs in addition to the ones already mentioned, may be part of the explanation; but their impact is hard to measure. Information may not be symmetrically distributed. Firms do not publish their exposures by currency so treasurers must be better informed in this regard than investors. Perhaps there is room for managers to make hedging decisions for stockholders.96 94 Note that the long term-average observation would only show that the expected value of the forward rate is equal to the expected value of the spot, not that the forward rate is equal to the conditional expected value of the spot. 9 Also, when risk is considered, the exposure of long-term nominal foreign-currency assets is found to have little to do with deviations of the forward from the future spot rate: we discuss exposure below.

9

Asymmetric information raises thorny problems. If managers attempt to enter forward trans actions on behalf of shareholders, what should they assume regarding the forward contracting

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 971 Organization and accounting theories may provide additional clues. Managers may not willingly further the interests of stockholders. Many of them in fact object to value maximization partly on the generally questionable grounds that it is unfair to evaluate their performances relative to market prices over which they have no control. Instead, they prefer or may be required to serve some accounting objective such as minimizing the exchange losses reported in financial statements. This last is reflected in Rodriguez's [157] survey and would account for the popularity of Aliber's averaging arguments. There is for the present no comprehensive theoretical framework for dealing convincingly with this issue. We can offer some preliminary notions as a step in that direction. For a point of departure, let us postulate that "market-value stabilization" is the objective.97 A firm's exposure to a given foreign exchange rate can then be defined as the sensitivity of the domestic real market value of the firm's equity, as of a given future date, to the concomitant random variations in the future domestic purchasing power of the foreign currency on that date. The measure of sensitivity is an amount of foreign currency: it is the amount deposited in the bank which would render the firm as vulnerable to foreign currency risk at the target date, as does its commercial activity. A perfect hedge consists of the forward transaction in foreign currency for the said maturity required to render the random variations in the future real domestic market price of the stock independent of the randomness of the future purchasing power of the foreign currency. As one performs this sensitivity measure vis 'a vis all currencies simultaneously,98 the result is an equivalent portfolio of foreign (and also domes tic) currencies which the firm is implicitly holding on account of its commercial activity. An optimal hedge is equal to the simple difference between the implicit pre-existing portfolio and some desired portfolio which meets the postulated objective. To state a theoretical definition of exposure is to reveal also its practical limitations. Economic theory has not yet progressed to the point where it is possible in the case of a firm, security, or commodity accurately to associate a specific future market price and its probability with each possible level of the future exchange rate. Much of the managerial literature (Lietaer [123], Makin [133], Adler and Dumas [7] seeks to avoid the problems involved with market value objectives by relying on accounting numbers.99 These suffer from at least behavior of the stockholders themselves? The answer is especially difficult to provide when stock holders' nationalities differ. Asymmetric information is not the only rationale for corporate hedging: one referee pointed out that hedging entails fixed (information and transactions) costs which would become prohibitive if stockholders hedged on their own account on a day-to-day basis. 9 Perhaps, this objective would be derived from the desire to minimize the probability of default (default being the circumstance where the value of equity is zero).

98 If the postulated objective function can be put in the mean-variance form (as, for instance, when the market value of the equity is normally distributed), the exposures to the various currencies are coefficients of a multiple regres sion (across states of nature) of the market value of the equity on the purchasing powers of all foreign currencies (Dumas [46]). The same procedure applies, of course, to any asset or security. Approximate stock market exposures were displayed in Table V.

99

Accountants themselves somewhat arbitrarily classify balance sheet items into exposed and non exposed. Exposure is then simply the net amount of exposed assets minus liabilities in today's balance

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC

All use subject to JSTOR Terms and Conditions

972 The Journal of Finance one deficiency:100 they do not incorporate the delayed effects of an exchange rate change on the firm's ensuing cash flows.101 To remedy this limitation and to accommodate managerial concerns, Adler and Dumas [7] proposed that treasurers be concerned with the mean and variance of the consolidated net worth or the consolidated cash balance102 (in the company's functional currency), measured at some cut-off date. This date would be chosen posterior to the target date at which the exchange rate changes being analyzed are to take place.103 Either nominal objective may be justified as a signal of default risk. A firm is equally bankrupt when the real or nominal value of its equity reaches zero; and cash balances can be compared to their deficiency levels, irrespective of the measure ment unit. Equipped with such a mean-variance objective, the firm could proceed as before. Levels of the cut-off-date net worth or cash balance would be associated, perhaps by simulation, with levels of the target-date exchange rates after taking account of the responses the firm might plan to make in each state. Exposures to each currency will be represented by the coefficients in a regression (across states) of the target variable on the set of exchange rates. The exposures measured in this way or on the basis of the earlier market-value reasoning will be global in the sense that they encompass the sensitivity of the firm as a whole to exchange rates. They can, however, be decomposed among components which represent the different kinds of influence that exchange rates can have. There are at least five categories of these: (1) the impact on short-term nominal net assets with maturity equal to the target date; (2) the impact on longer term nominal net assets with maturities falling beyond the target date; (3) the impact on the salvage value of existing physical assets and on the replacement cost or purchase price of physical assets to be replaced or acquired; (4) the impact on sales prices and unit costs; and (5) the indirect impact via sales prices on the volume of sales and consequently on the planned volume of production and other physical activities. Accounting measures of exposure have imperfectly dealt with only the first two of these effects. The fifth has so far been completely ignored. There is some confusion regarding the rest. sheet: the time-dated forward-looking nature of exposure is lost. Under FASB 8 exposure was equal to foreign accounts receivable plus cash minus long-term debt, accounts payable, and short-term debt; it was generally negative. Under FASB 52, all items are considered exposed; exposure is then equal to net foreign assets and it is necessarily positive. '0 One other deficiency is that accounting numbers are nominal in nature and no price index seems inherently satisfactory to deflate them as has been made clear by the endless debate on inflation accounting. For a questionable attempt at solving the firm's numeraire problem, see Eaker [51] . 101 These would have been included in a market-value based measure of exposure. 102 This approximation, too, is ad hoc. It suffers from the same potential problems of non-optimality and inconsistency as any firm-specific quantity. It has, however, the major virtue of being able to bring all the potential effects of the exchange rate into the picture. There is little to choose between book net worth and the cash balance on this count. The former accommodates long-term debt while the latter focuses on treasurers' professed main concern, cash, and is independent of any accounting rule. 103 The cut-off date only determines the precision of the measurement whereas the target date is an essential parameter of exposure measures.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 973

Adler and Dumas [8] considered channel (2); ie, nominal long-term, foreign currency bonds. The market values of such bonds may be exposed more or less than 100% in the sense that their exposures, and therefore the amount of shorter term forward contracts required for a perfect hedge, may be larger or smaller than their face or redemption values. Whether they are over- or under-exposed depends on the degree of serial dependence in anticipated exchange rates (and not on the difference between the forward and ensuing spot exchange rates, as in Aliber [15]). Aliber and Stickney [16] linked channels (3) and (4) to the existence of deviations from commodity price parity (CPP) between countries. Their specific contention was that physical assets are not exposed because the average deviation from purchasing power parity (relative to a base period) empirically tends to zero over the long haul. In Section (2), this evidence was seen to be consistent with the hypothesis that PPP deviations follow a martingale, hardly a situation of no risk. Their argument is also highly questionable for additional reasons. As far as channel (3) is concerned, it is clear that exchange rates may influence the reference currency market value of a physical asset located in any country once exchange rates and goods prices are correlated. Physical assets regardless of their location are indeed generally exposed to exchange risk. It is equally clear, however, that the issue of comparing the exposures of (identical) physical assets located in different countries does not arise when exposure is properly defined. Hence comparative prices (as in CPP) have no role to play. Traditional accounting trailed this red herring: the original cost book value of physical assets located at home does not fluctuate and is by definition not exposed. The CPP misconception presumably arose from an implicit comparison with home assets. As for channel (4), it is clear that exchange rate variations may erode or improve the firm's competitiveness abroad. But note that a firm generally purchases goods and services in one country, transforms them, and may sell them in some other country. Except in the case of a pure shipping firm, the goods bought in one place and sold in another are never the same. Even if CPP held exactly, there remains the possibility that the prices of the output good and the input goods will have different exposures. Profitability and net cash flows would then be affected by exchange rates. It is small solace to know that on the average over the long run goods prices may be internationally linked. The firm should not be concerned with averages when it is planning its risk-bearing strategy. IX. Conclusion The best conclusion of a survey paper is undoubtedly one filled with directions for research prompted by the shortcomings of existing theory. The shortcomings have probably been apparent to the reader, but it may be useful to recapitulate. Deviations from parity of individual commodities prices are a phenomenon about which existing microeconomics has very little to say. A model of interna tional goods markets must be constructed before we can say anything serious regarding the financial side. We have mentioned that segmentation of the goods market can induce segmentation of capital markets. The manner in which the This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

974 The Journal of Finance foreign exchange market reacts to CPP or PPP deviations (see Section I) can and will be properly analyzed only when the physical events (shipments, produc tion, etc.) which are being anticipated by this market are themselves made explicit. Macroeconomists would also benefit from such models, since fluctuations in employment may be linked to the

international competitiveness of a country's products. Disregarding possible inconsistencies, we dealt with portfolio choices and asset pricing in a unified worldwide capital market with PPP deviations. As in much of finance theory so far, the functional form of the dynamics of securities prices was postulated ab inito, leaving only the cross-sectional relationship between risk and return parameters to be determined in market equilibrium. Following Lucas [129] and Cox, Ingersoll, and Ross [37], the stochastic process of asset prices should instead be fully endogenized by dynamic methods borrowed from func tional analysis and by imposing an assumption of rational expectations.104 Such a project would come at an auspicious time since balance-of-payments theorists, who have so far postulated ad hoc asset demands, have lately become interested in utility maximization (see Obstfeld [167]). The introduction of money balances in portfolios which we discussed briefly in Section VI would tie together the concerns of financial micro- and macroeconomists, which are becoming remark ably convergent. We are on the threshold of a true stochastic theory of the balance of payments. One difficulty, however, looms large. In a complete financial market with available instruments for all maturities and investors holding rational expectations, there is no need for portfolio revisions; prices adjust but not portfolios. One way to account for international capital flows under rational expectations is if there remain unhedgable or unexpected risks for at least some maturities. Other avenues for introducing portfolio revisions may prove fruitful. Delicate analytical choices will have to be made. On the empirical side, tests and measures of the degree and specific sources of segmentation of international capital markets are becoming essential. It is almost impossible to progress without having some knowledge of the true meaning of national borders in finance. Initial tests should be based not on an analysis of securities prices, but on the principle that in integrated capital markets all consumption risks are mutualized. It is to be feared, however, that data on consumption behavior are not sufficiently reliable to be submitted to stochastic analysis. In particular, it may prove difficult to distinguish in a relatively stable fashion the expected and unexpected variations of consumption rates or to trace the effects of specific imperfections. As far as policy implications are concerned, corporate financial behavior still awaits a proper paradigm which will have to be provided by general financial theory. In the meanwhile, some mean-variance portfolio choice will serve as a framework for exchange risk hedging decisions. The thrust should be towards practicality and towards a good understanding of the dynamics of the problem: should one hedge stocks or flows and how to long-term and short-term hedging instruments interact? 104 In order for the theory to have content, the state variables upon which prices and decisions are contingent should be fully specified. Stulz [191] stopped short of this goal. In addition, the difficult problems associated with aggregating across investors' diverse preferences will have to be solved.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 975 APPENDIX Conditions for the Existence of Price Indices

As in standard portfolio theory, we assume that investors maximize a time additive von Neumann-Morgenstern expected-utility of lifetime consumption function. But we introduce here several commodities so that there is some difficulty in defining what is meant by the consumption rate. The basic objective function of the investor must now be formulated as a function of the several consumption rates achieved for the various commodities: rT

Max E U(c(s); s) ds (1) f

where E(.) = the expected-value operator conditional on the information available at time t; and

c(s)

=

Icg(s); g

=

1, * G} is the vector of consumption rates for the G goods at time s

Optimization of the consumption mix at each point in time leads to an equivalent the indirect function V(.): objective in terms of utility rT

Max EV (C(s); P(s); s) ds (2) where C(s) is the rate of nominal consumption expressed in some arbitrary monetary unit per unit of time; and = P(s) IPg(s); g = 1, * * * G} is the vector of prices for the G goods at time s, expressed in the same monetary unit, and

V(C; P; s) Max U(c; s) stcP = C, c 2 0 (3) Assuming that the function U(.) is sufficiently well behaved for the function V(.) to exist, be unique and - for the purposes of later derivations - to be continuous and twice differentiable, the function V(.) satisfies a relationship known as Roy's identity (see Varian [198], page 93): VP=-c Vc (4)

where V,p is the 1 x G vector of partial derivatives of V with respect to the price of goods; c is here the 1 x G vector of optimal consumption rates; and Vc is the partial derivative of V with respect to the consumption budget C. The formulation (2) proves somewhat cumbersome when it comes to portfolio choices because all prices of consumption goods appear separately in the objective function. A more compact, albeit special, formulation is possible when the function U(.) is such that there exists an invariant price index, ie, a compression This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

976 The Journal of Finance

of the price vector P into a single scalar leading to the same decisions and valid at all levels of the consumption budget C. Precisely: DEFINITION. There exists an invariant price index when the indirect utility function V(.) either satisfies the property

V(C; P; s) C(V(1; P; s) (5) or can be transformed by some monotonic transformation into a function which satisfies (5). The condition to be satisfied by the direct utility function U(.) in order for there to exist an invariant price index was pointed out by Samuelson and Swamy [168]: LEMMA. A necessary and sufficient condition for the indirect utility function V( * ) to satisfy property (5) is that the direct utility function U( * ) be homogeneous of degree one with respect to the consumption rates c.

Proof: The proof is given in Varian ([198], pages 14 and 42) apropos cost and production functions exhibiting constant returns to scale. A homothetic function is defined as the composition of a monotonic transform with a function homogeneous of degree one. THEOREM (Samuelson and Swamy). A necessary and sufficient condition for there to exist an invariant price index is that the dire ct utility function U(.) be homothetic with respect to the vector of consumption rates c.

Proof: It is obvious from the definition (3) of V(.) that applying a monotonic transformation to U(.) applies the same transformation to V(.) and vice versa. When property (5) is satisfied, V(1; P; s) is evidently one over the invariant price index (or, it is the purchasing power index). It can be computed in practice only if one knows explicitly the function V(.). However, one may expect that for small percentage changes in prices, the local log-linear approximation of the function V(.) may be revealed by the consumer's budget allocation. Indeed, when (5) holds, one may write Roy's identity (4) as CVp(1; P; s) -cV(1; P; s) (6) so that, for each commodity g, [Pg/V(1; P; s)]Vpg(1; P; s) cgPg/C PROPOSITION. The elasticities of the index function with respect to prices are revealed

by budget shares.

This result is not affected if a monotonic transformation is applied to the function V(.), as the derivative of the monotonic transformation would only appear as a factor on both sides of Equation (6). In practice it means that percentage variations of the price index may be computed as an average of the percentage variations of individual commodity prices weighted by budget shares. This is, approximately, the way in which national statistical institutes across the world compute cost-of-living indices, a procedure which is valid under homothetic preferences only. This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 977 References 1. M. Adler. "The Cost of Capital and Valuation of a Two-Country Firm." Journal of Finance 29 (March 1974), 119-33. 2. --. "Investor Recognition of Corporate International Diversification: Comment." Journal of Finance 36 (March 1981), 187-91. 3. and B. Dumas. "Optimal International Acquisitions." Journal of Finance 30 (March 1975), 1-20. 4. . "The Long-Term Financial Decisions of the Multinational Corporation." In EJ Elton and MJ Gruber (eds.), International Capital Markets. Amsterdam: North Holland, 1975. 5. . "Portfolio Choice and the Demand for Forward Exchange." The American Economic Review 66 (May 1976), 332-40. 6. . "Default Risk and the Demand for Forward Exchange." In H. Levy and M. Sarnat (eds.). Financial Decision Making under Uncertainty. New York: Academic Press, 1977. 7. . "Foreign Exchange Risk Management." In B. Antl (ed.). Currency Risk and the Corporation. London: Euromoney Publications, 1980: 145-158. 8. . "The Exposure of Long-Term Foreign Currency Bonds." Journal of Financial and Quantitative Analysis 15 (November 1980), 973-95. 9. . "Should Exposure Management Depend on Translation Accounting Methods?" Euro money (June 1981), pp. 132-8. 10. M. Adler and B. Lehmann. "Deviations from PPP in the Long Run: A Random Walk?" Working paper, Columbia University, August 1982. 11. M. Adler and GVG Stevens. "The Trade Effects of Direct Investment." Journal of Finance 29 (May 1974), 655-76. 12. T. Agmon. "The Relations Among Equity Markets in the United States, United Kingdom, Germany and Japan." Journal of Finance 27 (September 1972), 839-56. 13. - and DR Lessard. "Investor Recognition of Corporate International Diversification." Journal of Finance 32 (September 1977), 1049-56. 14. RZ Aliber. "The Interest Rate Parity Theorem: A Reinterpretation." Journal of Political Economy 81 (November/December 1973), 1451-9. 15. . Exchange Risk and Corporate International Finance. London: Macmillan, 1978. 16. and CP Stickney. "Accounting Measures of Foreign Exchange Exposure: The Long and Short of It." Accounting Review 50 (January 1975), 44-57. 17. MJ Bailey. "The Welfare Cost of Inflationary Finance." Journal of Political Economy 64 (April 1956),93110. 18. B. Balassa. "The Purchasing-Power Parity Doctrine: A Re-appraisal." Journal of Political Economy 72 (December 1964), 584-96. 19. DP Baron. "Flexible Exchange Rates, Forward Markets and the Level of Trade." American Economic Review 66 (June 1976), 253-66. 20. - and R. Forsythe. "Models of the Firm and International Trade under Uncertainty." American Economic Review 69 (September 1979), 565-74. 21. R. Barro. "Inflationary Finance and the Welfare Cost of Inflation. Journal of Political Economy 80 (September 1972), 978-1001. 22. VS Bawa, SJ Brown, and RW Klein. Estimation Risk and Optimal Portfolio Choice. Amsterdam and New York: North Holland Publishing Co., 1979. 23. F. Black. "Capital Market Equilibrium with Restricted Borrowing." Journal of Business 45 (July 1972),444-54.

24. . "International Capital Market Equilibrium with Investment Barriers." Journal of Financial Economics 1 (December 1974), 337-52. 25. ME Blume. "On the Assessment of Risk." Journal of Finance 26 (March 1971), 1-10. 26. J. Braga de Macedo. "Portfolio Diversification across Currencies." Working Paper, Princeton University, November 1980. 27. WH Branson. "The Minimum Covered Interst Differential Needed for International Arbitrage Activity." Journal of Political Economy 77 (November/December 1969), 1028-35. 28. WH Branson. "Asset Markets and Relative Prices in Exchange Rate Determination." IIES Seminar paper No. 66, Stoc kholm, December 1976.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

978 The Journal of Finance 29. , SI Katz, and TD Willett. "Exchange Rate Systems, Interest Rates, and Capital Flows." Princeton Essays in International Finance, No. 78, January 1970. 30. DT Breeden. "An Intertemporal Asset Pricing Model with Stochastic Consumption and Investment Opportunities." Journal of Financial Economics 7 (September 1979), 265-96. 31. HL Brewer. "Investor Benefits from Corporate International Diversification." Journal of Financial and Quantitative Analysis 16 (March 1981), 113-26. 32. ER Canterbery. "Foreign Exchange, Capital Flows and Monetary Policy." Princeton Studies in International Finance, Department of Economics, Princeton University, 1965. 33. C. Chamley and HM Polemarchakis. "Asset Markets, General Equilibrium and the Neutrality of Money." Working paper, Cowles Foundation, September 1981. 34. B. Cornell. "Spot Rates, Forward Rates and Exchange Market Efficiency." Journal of Financial Economics 5 (August 1977), 55-66. 35. . "Relative Price Changes and Deviations from Purchasing Power Parity." Journal of Banking and Finance 3 (September 1979), 263-79. 36. . "Inflation, Relative Prices and Exchange Risk." Financial Management 9 (Autumn 1980), 30-34. 37. JC Cox, JE Ingersoll, Jr., and SA Ross. "Theory of the Term Structure of Interest Rates." Econometrica, forthcoming. 38. L. Crouhy-Veyrac, M. Crouhy, and Jacques Melitz. "More About the Law of One Price." European Economic Review 18 (July 1982), 325-44. 39. RE Cumby and M. Obstfeld. "A Note of Exchange Rate Expectations and Nominal Interest Differentials: A Test of the Fisher Hypothesis." Journal of Finance 36 (June 1981), 697-704. 40. MP Dooley. "A Model of Arbitrage and Short-term Capital Flows." International Finance Discussion Papers, No. 40 (January 1974), Division of International Finance, Board of Governors of the Federal Reserve System. 41. and P. Isard. "Capital Controls, Political Risks, and Deviations from Interest-Rate Parity." Journal of Political Economy 88 (April 1980), 370-84. 42. and JR Shafer. "Analysis of Short-Run Exchange Rate Behavior." Board of Governors of the Federal Reserve System, International Finance Discussion Papers, No. 76, February 1976. 43. R. Dornbusch. "Exchange Risk and the Macroeconomics of Exchange Rate Determination." In R. Hawkins et al., editors. The Internationalization of Financial Markets and National Economic Policy. Greenwich, CT: JAI Press, 1982. 44. G. Dufey and I. Giddy. "The Random Behavior of Flexible Exchange Rates: Implications for Forecasting." Journal of International Business Studies (Spring 1975), 1-32. 45. B. Dumas. "Discussion." Journal of Finance (May 1977), 512-15. 46. . "The Theory of the Trading Firm Revisited." Journal of Finance 33 (June 1978), 1019- 29. 47. . "The Theorems of International Trade under Generalized Uncertainty." Journal of International

Economics 10 (November 1980), 481-98. 48. . "Trade Theorems with Less than Perfectly Correlated Disturbances." CERESSEC Working Paper, 1980. 49. . "Trading Rules and Equilibrium in a Monetary Economy." In M. Sarnat (ed.). Capital Markets and Inflation. Cambridge, MA: Ballinger, 1982. 50. and P. Poncet. "La Demande de Dollars des Personnes ne resident pas aux Etats-Unis." Working paper, 1980, Archives de I'ISEA, forthcoming. 51. MR Eaker. "The Numeraire Problem and Foreign Exchange Risk." Journal of Finance 36 (May 1981), 419-26. 52. PA Einzig. A Dynamic Theory of Forward Exchange. New York: Macmillan, 1961. 53. S. Ekern and R. Wilson. "On the theory of the Firm in an Economy with Incomplete Markets." Bell Journal of Economics and Management Science 5 (Spring 1974), 171-80. 54. VR Errunza and LW Senbet. "The Effects of International Operations on the Market Value of the Firm: Theory and Evidence." Journal of Finance 36 (May 1981), 40118. 55. EF Fama. "The Behavior of Stock Market Prices." Journal of Business 38 (January 1965), 34-105. 56. . Foundations of Finance. New York: Basic Books, 1976.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 979 57. . "Stock Returns, Real Activity, Inflation and Money." American Economic Review 71 (September 1981), 545-65. 58. and A. Farber. "Money, Bonds and Foreign Exchange." American Economic Review 69 (September 1979), 639-49. 59. and R. Roll. "Some Properties of Symmetric Stable Distributions." Journal of the American Statistical Association 63 (September 1968), 817-36. 60. EF Fama and GW Schwert. "Asset Returns and Inflation." Journal of Financial Economics 5 (November 1977), 115-46. 61. A. Farber, R. Roll and B. Solnik. "An Empirical Study of Risk Under Fixed and Flexible Exchange Rates." In K. Brunner and AH Meltzer (eds.). Stabilization of the Domestic and International Economy. Amsterdam: North Holland, 1977: 235-75. 62. S. Fischer. "The Demand for Index Bonds." Journal of Political Economy 83 (June 1975), 529- 34. 63. . "Capital Accumulation on the Transition Path in a Monetary Optimizing Economy." Econometrica 47 (November 1979), 1433-9. 64. J. Frankel. "The Diversifiability of Exchange Risk." Journal of International Economics 9 (August 1979), 379-94. 65. J. Frankel. "A Test of Perfect Substitutability in the Foreign Exchange Market." Southern Economic Journal 48 (October 1982), 406-16. 66. J. Frankel. "In Search of the Exchange-Risk Premium: A Six-Currency Test Assuming Mean Variance Optimization." Journal of International Money and Finance (December 1982), forthcoming. 67. JA Frenkel. "Elasticities and the Interest Parity Theory." Journal of Political Economy 81 (May/June 1973), 741-7. 68. . "Flexible Exchange Rates, Prices and the Role of News: Lessons from the 1970's." Journal of Political Economy 89 (August 1981), 665-705. 69. and RM Levich. "Covered Interest Arbitrage: Unexploited Profits?" Journal of Political Economy (April 1975).

70. I. Friend, Y. Landskroner, and E. Losq. "The Demand for Risky Assets under Uncertain Inflation." Journal of Finance 31 (December 1976), 1287-98. 71. I. Friend and E. Losq. "Advantages and Limitations of International Portfolio Diversification." In M. Sarnat and GP Szego (eds.). International Finance and Trade. Cambridge, MA: Ballinger, 1979: 3-16. 72. HJ Gailliot. "Purchasing Power Parity as an Explanation of Long Term Changes in Exchange Rates." Journal of Money, Credit and Banking 2 (August 1970), 348-57. 73. H. Genberg. "Purchasing Power Parity under Fixed and Flexible Exchange Rates." Journal of International Economics 9 (May 1978), 247-76. 74. J. Geweke and E. Feige. "Some Joint Tests of the Efficiency of the Markets for Forward Foreign Exchange." Review of Economics and Statistics 61 (August 1979), 334-41. 75. L. Girton and D. Henderson. "Central Bank Operations in Foreign and Domestic Assets under Fixed and Flexible Exchange Rates." In P. Clark, D. Logue, and R. Sweeney (eds.). The Effects of Exchange Rate Adjustments. Washington: US Government Printing Office, 1976: 151-79. 76. L. Girton and D. Henderson. "Financial Capital Movements and Central Bank Behavior in a TwoCountry, Short-run Portfolio Balance Model." Journal of Monetary Economics 2 (Jan uary 1976), 33-61. 77. CWJ Granger and 0. Morgenstern. Predictability of Stock Market Prices. Lexington, MA: HeathLexington, 1970. 78. CWJ Granger. "A Survey of Empirical Studies on Capital Markets." In EJ Elton and MJ Gruber (eds.). International Capital Markets. Amsterdam: North Holland, 1975: 3-36. 79. FLA Grauer, RH Litzenberger, and R. Stehle. "Sharing Rules and Equilibrium in an International Capital Market under Uncertainty." Journal of Financial Economics 3 (June 1976), 23356. 80. HG Grubel. "Internationally Diversified Portfolios: Welfare Gains and Capital Flows." American Economic Review 58 (December 1968), 1299-1314. 81. . Forward Exchange, Speculation and the International Flow of Capital. Stanford: Stanford

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

980 The Journal of Finance University Press, 1966. 82. NH Hakansson. "On the Relevance of Price-Level Accounting." Journal of Accounting Research 7 (Spring 1969), 22-31. 83. LP Hansen and RJ Hodrick. "Foward Exchange Rates as Optimal Predictors of Future Spot Rates: An Econometric Analysis." Journal of Political Economy 88 (October 1980), 828-53. 84. D. Heckerman. "On the Effects of Exchange Risks." Journal of International Economics 3 (November 1973), 379-88. 85. E. Helpman and A. Razin. "Uncertainty and International Trade in the Presence of Stock Markets." Review of Economics Studies 45 (June 1978), 239-50. 86. . "Welfare Aspects of International Trade in Goods and Securities." Quarterly Journal of Economics 92 (August 1978), 489-508. 87. . A Theory of International Trade under Uncertainty. New York: Academic Press, 1980. 88. . "Comparative Dynamics of Monetary Policy in a Floating Exchange Rate Regime." Working paper, Tel-Aviv University, 1981. 89. RJ Herring and RC Marston. "The Forward Market and Interest Rates in the Eurocurrency and National Money Markets." In Carl H. Stein, John H. Makin, and Dennis E. Logue (eds.). Eurocurrencies and the International Monetary System. Washington: American Enterprise Institute, 1976.

90. R. Hodrick. "International Asset Pricing with Time-varying Risk Premia." Journal of Interna tional Economics 11 (November 1981), 573-7. 91. D. Hsu, RB Miller, and DW Wichern. "On the Stable Paretian Behavior of Stock Market Prices." Journal of the American Statistical Association 69 (March 1974), 108-13. 92. GC Hufbauer. Synthetic Materials and the Theory of International Trade. Cambridge, MA: Harvard University Press, 1966. 93. . "The Impact of National Characteristics and Technology on the Commodity Composi tion of Trade in Manufactured Goods." In R. Vernon (ed.). The Technology Factor in International Trade. New York: Columbia University Press, 1970: 145-232. 94. . "The Multinational Corporation and Direct Investment." In PB Kenen (ed.). Inter national Trade and Finance: Frontiers for Research. New York: Cambridge University Press, 1975: 253-320. 95. G. Hufbauer, and M. Adler. Overseas Manufacturing Investment and the Balance of Payments. Washington, DC: Government Printing Office, 1968. 96. P. Isard. "How Far Can We Push the Law of One Price?" American Economic Review 67 (December 1977), 942-8. 97. B. Jacquillat and B. Solnik. "Multinationals are Poor Tools for Diversification." Journal of Portfolio Management (Winter 1978). 98. DM Jaffee. Comment on RZ Aliber: "The Firm under Pegged and Floating Exchange Rates." Scandinvaian Journal of Economics 78:2 (1976), 323-6. 99. HG Johnson. "The Efficiency and Welfare Implications of the International Corporation." In Kindleberger (ed.). The International Corporation. Cambridge, MA: MIT Press, 19: 35-36. 100. J. Johnston. Econometric Methods. New York: McGraw Hill, 1972. 101. L. Katseli-Papaefstratiou. "The Reemergence of the Purchasing Power Parity Doctrine in 1970's." Special Paper in International Economics No. 13 (December 1979), Princeton. 102. PB Kenen. "Trade, Speculation and the Forward Exchange Rates." In RE Baldwin et al. (eds.). Trade, Growth and the Balance of Payments. Chicago: Rand MacNally, 1965. 103. BF King. "Market and Industry Factors in Stock Price Behavior." Journal of Business 39 (January 1966), 139-90. 104. SW Kohlhagen. "The Forward Rate as an Unbiased Predictor of the Future Spot Rate." University of California, Berkeley, 1976. 105. . "'Rational' and 'Endogenous' Exchange Rate Expectations and Speculative Capital Flows in Germany." Weltwirtschaftsliches Archiv 113 (December 1977), 624-44. 106. . The Behavior of Foreign Exchange Markets-A Critical Survey of the Literature. NYU Monograph Series, 1973. 107. P. Kouri. "The Determinants of the Forward Premium." IIES Seminar Paper 62, University of Stockholm, August 1976. 108. . "International Investment and Interest Rate Linkages under Flexible Exchange Rates."

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 981 In R. Aliber (ed.). The Political Economy of Monetary Reform. London: Macmillan & Co., 1977. 109. . "The Exchange Rate and the Balance of Payments in the Short Run and in the Long Run: A Monetary Approach." Scandinavian Journal of Economics 78:2 (1976), 280-304. 110. P. Kouri and J. Braga de Macedo. "Exchange Rates and the International Adjustment Process. Brookings Papers (1:1978), 111-150. 111. IB Kravis, Z. Kenessey, AW Heston, and R. Summers. A System of International Comparisons of Gross Product and Purchasing Power. Baltimore: Johns Hopkins University Press, 1975. 112. IB Kravis and RE Lipsey. "Price Behavior in the Light of Balance of Payments Theory." Journal of International Economics 8 (May 1978), 193-246.

113. P. Krugman. "Consumption Preferences, Assets Demands, and Distribution Effects in Inter national Financial Markets." Working Paper No. 651, National Bureau of Economic Research, March 1981. 114. Y. Landskroner and M. Liviatan. "Risk Premia and the Sources of Inflation." Journal of Money, Credit and Banking 13 (May 1981), 205-14. 115. S. Laursen and L. Metzler. "Flexible Exchange Rates and the Theory of Employment." Review of Economics and Statistics 32 (November 1950), 281-99. 116. WY Lee and KS Sachdeva. "The Role of the Multinational Firm in the Integration of S egmented Capital Markets." Journal of Finance 32 (May 1977), 479-92. 117. DR Lessard. "World, National and Industry Factors in Equity Returns. Journal of Finance 29 (May 1974), 379-91. 118. . "World, Country and Industry Relationships in Equity Returns: Implications for Risk Reduction through International Diversification." Financial Analysts Journal (January/Feb ruary 1976), 2-8. 119. . "Evaluating Foreign Projects: An Adjusted Present Value Approach." In DR Lessard (ed.). International Financial Management. Boston, MA: Warren, Gorham & Lamont, 1979: 577-92. 120. RM Levich. "Tests of Forecasting Models and Market Efficiency in the International Money Market." In JA Frenkel and HG Johnson (eds.). The Economics of Exchange Rates. Reading, MA: Addison-Wesley, 1978. 121. JH Levin. Forward Exchange and Internal-External Equilibrium. Ann Arbor: Michigan International Business Series, No. 12, 1970. 122. H. Levy and M. Sarnat. "International Diversification of Investment Portfolios." American Economic Review 60 (September 1970), 668-75. 123. BA Lietaer, Financial Management of Foreign Exchange: An Operational Technique to Reduce Risk. Cambridge, MA: MIT Press, 1971. 124. J. Lintner. "The Valuation of Risk Assets and the Selection of Risky Investments in Stock Portfolios and Capital Budgets." Review of Economics and Statistics 47 (February 1965), 13- 37. 125. . "Inflation and Security Returns." Journal of Finance 30 (May 1975), 259-80. 126. . "Bankruptcy Risk, Market Segmentation and Optimal Capital Structures." In I. Friend and J. Bicksler (eds.). Risk and Return in Finance, Vol. 11. Cambridge, MA: Ballinger Publishing Co., 1977: 1-128. 127. E. Losq. "Commodity Price Uncertainty and Capital Market Equilibrium." McGill University, mimeo, 1977. 128. RE Lucas. "Some International Evidence on Output-Inflation Trade-offs." American Economic Review 63 (June 1973), 326-34. 129. . "Asset Prices in an Exchange Economy." Econometrica 46 (November 1978), 1429-46. 130. . "Equilibrium in a Pure Currency Economy." Economic Inquiry 18 (April 1980), 203-20. 131. . "Interest Rates and Currency Prices in a Two-country World." Working paper, Univer sity of Chicago, July 1981. 132. S. Magee. "Multinational Corporation, the Industry Technology Cycle and the Transfer of Technology to Developing Countries." Journal of World Trade Law, 1978. 133. JH Makin. "Portfolio Theory and the Problem of Foreign Exchange Risk." Journal of Finance 33 (May 1978), 517-34. 134. J. Mayshar. "Transaction Costs and the Pricing of Assets." Journal of Finance 36 (June 1981),

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

982 The Journal of Finance

583-98. 135. JW McFarland, RR Pettit, and SK Sung. "The Distribution of Foreign Exchange Price Changes: Trading Day Effects and Risk Measurement." Journal of Finance 37 (June 1982), 693-716. 136. RI McKinnon. "The Exchange Rate and Macroeconomic Policy: Changing Postwar Percep tions." Journal of Economic Literature 29 (June 1981), 531-57. 137. R. Mehra. "On the Financing and Investment Decisions of Multinational Firms in the Presence of Exchange Risk." Journal of Financial and Quantitative Analysis 13 (June 1978), 227-44. 138. RC Merton. "Lifetime Portfolio Selection under Uncertainty: The Continuous-Time Case." Review of Economics and Statistics 51 (August 1969), 247-57. 139. . "Optimum Consumption and Portfolio Rules in a Continuous-Time Model." Journal of Economic Theory 3 (December 1971), 373-413. 140. . "An Intertemporal Capital Asset Pricing Model." Econometrica 41 (September 1973), 867-87. 141. E. Modigliani and MH Miller." "The Cost of Capital, Corporation Finance and the Theory of Investment." American Economic Review 48 (April 1958), 261-97. 142. Moon Ho Lee. Purchasing Power Party. New York: Marcel Dekker, Inc., 1976. 143. J. Mossin. "Equilibrium in a Capital Asset Market." Econometrica 34 (October 1966), 768-83. 144. . The Economic Efficiency of Financial Markets. Lexington, MA: DC Health, 1977. 145. RA Mundell. International Economics. New York: Macmillan, 1968. 146. . Monetary Theory: Inflation, Interest and Growth in the World Economy. New York: Goodyear Publishing Company, 1971. 147. M. Obstfeld. "Macroeconomic Policy and World Welfare ander Flexible Exchange Rates." Working paper, Columbia University, March 1980. 148. LH Officer. "The Purchasing Power Theory of Exchange Rates: A Review Article." IMF Staff Papers 23 (March 1976), 1-60. 149. F. Papadia. "Forward Exchange Rates as Predictors of Future Spot Rates and the Efficiency of the Foreign Exchange Market." Journal of Banking and Finance 5 (June 1981), 217-40. 150. JE Pippenger. "Spot Rates, Forward Rates, and Interest Rate Differentials." Journal of Money, Credit and Banking 4 (May 1972), 375-83. 151. P. Poncet. "Optimum Consumption and Portfolio Rules with Money as an Asset." Working paper, ESSEC, Department of Finance, November 1979. 152. W. Poole. "Speculative Prices as Random Walks: An Analysis of Ten Time Series of Flexible Exchange Rates." Southern Economic Journal 83 (1966). 153. MF Prachowny. "A Note on Interest Parity and the Supply of Arbitrage Funds." Journal of Political Economy 78 (May/June 1970), 540-5. 154. G. Ragazzi. "Theories of the Determinants of Direct Foreign Investment." LM.F. Staff Papers 20 (July 1973), 471-98. 155. SF Richard and M. Sundaresan. "A Continuous Time Equilibrium Model of Forward Prices and Future Prices in a Multigood Economy." Journal of Financial Economics 9 (December 1981), 347-72. 156. JD Richardson. "Some Empirical Evidence on Commodity Arbitrage and the Law of One Price." Journal of International Economics 8 (May 1978), 341-52. 157. RM Rodriguez. Foreign-Exchange Management in US Multinationals. Lexington, MA: DC Heath, 1980. 158. R. Rogalski and JD Vinso. "Price Level Variations as Predictors of Flexible Exchange Rates." Journal of International Business Studies (Spring/Summer 1977), 71-81. 159. R. Roll. "Interest Rates on Monetary Assets and Commodity Price Index Changes." Journal of Finance 27 (May 1972), 251-77. 160. . "A Critique of the Asset Pricing Theory's Test; Part I: On Past and Potential Testability of the Theory." Journal of Financial Economics 4 (March 1977), 129-76.

161. . "Violations of Purchasing Power Parity and their Implications for Efficient International Commodity Markets." In M. Sarnat and GP Szego (eds.). International Finance and Trade. Cambridge, MA: Ballinger Publishing Co., 1979. 162. and S. Ross. "An Empirical Investigation of the Arbitrage Pricing Theory. Journal of Finance 35 (December 1980), 1073-1103.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

International Finance 983 163. R. Roll and B. Solnik. "A Pure Foreign Exchange Asset Pricing Model." Journal of International Economics 7 (May 1977), 161-80. 164. B. Rosenberg, and J. Ohlson. "The Stationary Distribution of Returns and Portfolio Separation in Capital Markets: A Fundamental Contradiction." Journal of Financial and Quantitative Analysiss 11 (September 1976), 393-402. 165. M. Rubinstein. "An Aggregation Theorem for Securities Markets." Journal of Financial Eco nomics 1 (September 1974), 225-44. 166. . "Corporate Financial Policy in Segmented Securities Markets." Journal of Financial and Quantitative Analysis 8 (December 1973), 749-62. 167. PA Samuelson. "The Fundamental Approximation Theorem of Portfolio Analysis in Terms of Means, Variances and Higher Moments." Review of Economic Studies 37 (October 1970), 537-42. 168. - and S. Swamy. "Invariant Economic Index Numbers and Canonical Duality: Survey and Synthesis." American Economic Review 64 (September 1974), 566-93. 169. A. Santomero and J. Seater. "The Inflation-Unemployment Trade-off: A Critique of the Literature." Journal of Economic Literature 16 (January 1978), 499-544. 170. T. Sargent and N. Wallace. "Rational Expectations, the Optimal Monetary Instrument and the Optimal Policy Rule." Journal of Policitcal Economy 83 (April 1975), 241-54. 171. L. Selden. "An OCE Analysis of the Effect of Uncertainty on Saving under Risk Preference Independence." Review of Economic Studies 46 (January 1979), 73-82. 172. LW Senbet. "International Capital Market Equilibrium and the Multinational Firm Financing and Investment Policies." Journal of Financial and Quantitative Analysis 14 (September 1979), 455-80. 173. AJ Senschak and WL Beedles. "Is Indirect International Diversification Desirable?" Journal of Portfolio Management 6 (Winter 1980), 49-57. 174. P. Sercu. "A Generalization of the International Asset Pricing Model." Revue de l'Association Frangaise de Finance 1 (June 1980), 91-135. 175. WF Sharpe. "A Simplified Model for Portfolio Analysis." Management Science 9 (January 1963), 277-93. 176. . "Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk." Journal of Finance 19 (September 1964), 425-42. 177. JJ Siegel. "Risk, Interest and Forward Exchange." Quarterly Journal of Econoics 86 (May 1972), 303-9. 178. E. Sohmen. Flexible Exchange Rates. Chicago: University of Chicago Press, 1969. 179. BH Solnik. "An Equilibrium Model of the International Capital Market." Journal of Economic Theory 8 (August 1974), 500-24. 180. . "The International Pricing of Risk: An Empirical Investigation of the World Capital Market Structure." Journal of Finance 29 (May 1974), 48-54. 181. -. "Why Not Diversify Internationally Rather than Domestically?" Financial Analysts Journal 30 (July/August 1974), 48-54. 182. . "Testing International Asset Pricing: Some Pessimistic Views." Journal of Finance 32 (May 1977), 503-11.

183. RC Stapleton and MG Subrahmanyam. "Market Imperfections, Capital Asset Equilibrium and Corporation Finance. Journal of Finance 32 (May 1977), 307-21. 184. . "Uncertain Inflation, Exchange Rates and Bond Yields." Journal of Banking and Finance 5 (March 1981), 93-107. 185. R. Stehle. "An Empirical Test of the Alternate Hypotheses of National and International Pricing and Risky Assets." Journal of Finance 32 (May 1977), 493-502. 186. GVG Stevens. "Two Problems in Portfolio Analysis: Conditional and Multiplicative Random Variables." Journal of Financial and Quantitative Analysis 6 (December 1971), 1235-50. 187. AC Stockman. "Risk, Information and Forward Exchange Rates." In JA Frankel and HF Johnson (eds.). The Economics of Exchange Rates. Reading, MA: Addison-Wesley, 1978. 188. . "A Theory of Exchange Rate Determination." Journal of Political Economy 88 (August 1980), 673-98. 189. HR Stoll. "Causes of Deviations from Interest Rate Parity." Journal of Money, Credit and Banking 4 (February 1972), 113-17.

This content downlo aded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

984 The Journal of Finance 190. RM Stulz. "On the Effects of Barriers to International Investment." Journal of Finance 36 (September 1981), 923-34. 191. . "A Model of International Asset Pricing." Journal of Financial Economics 9 (December 1981), 383-406. 192. . "The Forward Exchange Rate and Macroeconomics." Journal of International Economics 12 (May 1982), 285-99. 193. MG Subrahmanyam. "On the Optimality of International Capital Market Integration." Journal of Financial Economics 2 (August 1975), 3-28. 194. J. Tobin. "A General Equilibrium Approach to Monetary Theory." Journal of Money, Credit and Banking 1 (February 1969), 15-29. 195. and WR Brainard. "Asset Markets and the Cost of Capital." In B. Balassa and R. Nelson (eds.). Economic Progress, Private Values and Public Policy: Essays in Honor of William Fellner. Amsterdam: NorthHolland Publishing Co., 1977. 196. SC Tsiang. "The Theory of Forward Exchange and the Effects of Government Intervention in the Forward Exchange Market." LM.F. Staff Papers 7 (April 1959), 75-106. 197. RB Upson. "Random Walk and Forward Exchange Rates: A Spectral Analysis." Journal of Financial and Quantitative Analysis 7 (September 1972), 1897-1906. 198. HR Varian. Microeconomic Analysis. New York: Norton, 1978. 199. R. Vernon. "International Investment and International Trade in the Product Cycle." Quarterly Journal of Economics 80 (May 1966), 190-207. 200. N. Wallace. "A Modigliani-Miller Theorem for Open-Market Operations." American Economic Review 71 (June 1981), 267-74. 201. L. Weiss. "The Role for Active Monetary Policy in a Rational Expectations Model." Journal of Political Economy 88 (April 1980), 221-33. 202. JM Westerfield. "An Examination of Foreign Exchange Risk under Fixed and Floating Rate Regimes." Journal of International Economics 7 (May 1977), 181-200.

This content downloaded from 130.237.165.40 on Tue, 06 Oct 2015 14:36:38 UTC All use subject to JSTOR Terms and Conditions

Sebuah Model Equilibrium Model dari International Capital Market ➢ Model intertemporal pasar modal internasional dikembangkan dalam kerangka Model Penetapan Harga Aset Modal Sharpe-Lintner-Mossin. ➢ Dimensi fundamental dari pasar internasional ini adalah adanya risiko nilai tukar dan mekanisme yang memberikan perlindungan kepada investor yang tidak mau menanggung risiko semacam itu. ➢ Sebuah teorema reksa dana diturunkan yang menyatakan bahwa semua investor akan acuh tak acuh antara memilih portofolio dari aset asli atau dari tiga dana: a. portofolio saham yang dilindung nilai terhadap risiko nilai tukar (pasar portofolio) b. portofolio obligasi, spekulatif dalam dimensi risiko nilai tukar c. aset bebas risiko negara mereka sendiri ➢ Komposisi dana ini tidak tergantung pada preferensi atau kebangsaan investor (kecuali untuk aset bebas risiko). Premi risiko keamanan di atas tingkat bebas risiko nasionalnya ditampilkan proporsional dengan risiko sistematis internasionalnya. ➢ Koefisien proporsionalitas adalah premi risiko pasar kerja di seluruh dunia tingkat obligasi. ➢ Perbedaan antara suku bunga 2 negara ditunjukkan sama dengan perubahan paritas yang diharapkan antara kedua negara ditambah istilah tergantung pada kovarians ris nilai tukar. ➢ Nilai tukar forward adalah perkiraan bias dari nilai tukar masa depan. THE INTERNATIONAL ASSET PRICING MODEL ➢ Model Penetapan Harga Aset Modal yang dikembangkan oleh Sharpe, Lintner, dan Mossin adalah dasar untuk teori baru pasar modal. ➢ Model yang diterapkan pada pasar nasional tunggal menunjukkan bahwa semua investor dapat membuat keputusan investasi mereka dengan memilih portofolio mereka dari dua dana: portofolio pasar dan aset bebas risiko. ➢ Model yang memprediksi bahwa pengembalian ekstra yang diharapkan dari memegang aset sebanding dengan kovarians pengembaliannya dengan portfoho pasar ('"beta"). ➢ Dengan asumsi yang tidak terlalu restriktif, R. Merton, telah mengembangkan model ekuilibrium intertemporal pasar modal. Sifat intertemporal model ini memungkinkan untuk menangkap efek yang tidak akan pernah muncul dalam model statis. ➢ Model Penetapan Harga Aset Modal antarwaktu memiliki peran penting keterbatasan karena hanya mempertimbangkan investasi nasional. Oleh karena itu teori ini hanya dapat berlaku di pasar modal dunia yang sepenuhnya tersegmentasi. Tidak benar bahwa itu dapat dengan mudah diperluas hanya dengan memasukkan asing peluang investasi dalam portofolio pasar; model ini tidak akan mewakili keseimbangan umum dan kehilangan semua substansi ekonomi dan menarik. Faktanya, sangat sedikit upaya yang dilakukan untuk menggunakan pendekatan semacam ini untuk mengembangkan model ekuilibrium internasional pasar modal. Di antara berbagai kerumitan tugas semacam itu adalah ketiadaan dari aset bebas risiko universal (dan suku bunga yang berbeda) dan adanya risiko nilai tukar yang mengubah karakteristik investasi yang sama dari negara yang berbeda.

➢ Kemungkinan perubahan paritas menyiratkan bahwa aset yang sama mungkin menghasilkan pengembalian yang berbeda (dan karenanya ekspektasi pengembalian yang berbeda) untuk warga negara dari negara yang berbeda. Dalam kerangka mean-variance set peluang investasi yang dihadapi investor dari berbagai negara akan bervariasi, bahkan di pasar modal internasional yang sempurna dan terintegrasi.

1. International Capital Market Structure Sejumlah asumsi umum harus dibuat tentang struktur pasar modal; sebagian besar asumsi standar kesempurnaan pasar A-1 Pasar modal selalu dalam ekuilibrium (yaitu, tidak ada perdagangan dengan harga tidak seimbang). A-2 Pasar modal sempurna, tanpa biaya transaksi, pajak, atau kontrol modal. Investor adalah penerima harga. A-3 Aset dapat dijual pendek. A-4 Di setiap negara terdapat pasar (obligasi) untuk pinjaman dan pinjaman pada tingkat yang sama (namun tingkat ini tidak harus sama di semua negara). A-5 Perdagangan aset dan mata uang berlangsung terus menerus dalam waktu. Ini menyiratkan dunia nilai tukar yang fleksibel. A-6 Investor memiliki ekspektasi yang homogen tentang nilai tukar variasi dan distribusi pengembalian dalam hal asset mata uang. A-7 Tidak ada batasan pada arus modal internasional. A-8 Konsumsi investor terbatas di negara asalnya. ➢ Asumsi A-1 hingga A-4 adalah asumsi standar pasar sempurna. ➢ Model pasar internasional seperti itu tidak dapat dibangun tanpa biaya, dan asumsi yang meragukan dari ekspektasi homogen (A-6) juga harus dibuat di sini. Seperti yang dijelaskan Merton, A-5 mengikuti langsung dari A-2. ➢ Jika tidak ada biaya untuk bertransaksi dan aset dapat dipertukarkan dalam skala berapa pun, maka investor akan lebih memilih untuk dapat merevisi portofolio mereka setiap saat (baik mereka benar-benar melakukannya atau tidak). ➢ Dua poin terakhir merangkum semua asumsi internasional perilaku dan struktur pasar. A-7 adalah pusat model ini. Sejauh pola konsumsi investor yang bersangkutan A-8 menyiratkan ketat pemisahan nasional, tetapi tidak melarang pembelian barang asing dengan mata uang lokal. Asumsi ini akan dibahas lebih rinci pada halaman 505. 2. Tingkat Dinamika Pengembalian dan Struktur Nilai Tukar

Diasumsikan bahwa ekspektasi secara riil adalah homogen di semua investor dan bahwa harga dinamika. Dijelaskan dengan rumus. 3. Struktur Preferensi dan Dinamis Persamaan Anggaran Dalam model keseimbangan antarwaktu, umumnya tidak mungkin untuk asumsikan, pada saat yang sama, keacakan dalam pengembalian investasi, konsumsi harga dan pendapatan. Kami akan membuat asumsi penyederhanaan bahwa pendapatan upah diketahui secara pasti 4. Persamaan Optimalitas: Permintaan Aset TEOREMA PEMISAHAN 1. Semua investor akan acuh tak acuh antara memilih portofolio dari aset asli atau dari n + 1 dana, di mana pilihan yang mungkin untuk dana tersebut adalah: ➢ portofolio pasar saham dunia (lindung nilai terhadap risiko nilai tukar) ➢ n obligasi masing-masing negara. Proporsi dana berisiko yang diinvestasikan pada aset i adalah:

N.B.: Seperti yang diharapkan, teorema dua dana standar berlaku untuk semua investor dari satu negara; mereka dapat memilih dari dua dana: - aset bebas risiko domestik mereka - portofolio semua aset berisiko (termasuk obligasi asing) yang komposisinya tergantung pada kebangsaan investor (bukan pada preferensinya). Dana ini akan bervariasi untuk setiap negara. 5. Lebih Lanjut Tentang Risiko Pertukaran TEOREMA PEMISAHAN 2. Semua investor akan acuh tak acuh antara memilih portofolio dari kumpulan aset asli atau dari 3 dana, di mana pilihan yang mungkin untuk dana tersebut adalah: -portofolio semua saham yang dilindung nilai terhadap risiko nilai tukar (portofolio pasar dunia) -portofolio obligasi, spekulatif dalam dimensi risiko nilai tukar -aset bebas risiko negara mereka. Proporsi dana yang diinvestasikan dalam setiap aset ditentukan oleh karakteristik pasar 6. Hubungan Hasil Keseimbangan Antara Aset Kita sekarang menurunkan pasar ekuilibrium kondisi kliring untuk model. Karena fungsi permintaan untuk saham dan obligasi dapat dipisahkan, kondisi kliring dapat diselesaikan secara independen.

Hubungan ekuilibrium antara pengembalian yang diharapkan pada aset nasional individu dan pengembalian yang diharapkan di pasar internasional dapat diturunkan dari kondisi ini. a. Stocks Perbedaan yang paling jelas antara hubungan (16) dan perkawinan Hubungan Model Penetapan Harga Aset adalah: (1) risiko sistematis adalah risiko sistematis internasional, yang melibatkan: kovarians return saham dengan portofolio pasar dunia. (2) Ri dan Rm, secara umum akan berbeda. Asumsi bahwa suatu negara dapat mengontrol tingkat bunganya secara sewenang-wenang sekarang akan dijatuhkan. b. Bonds portofolio nol-beta (tanpa risiko pasar) tidak selalu memiliki pengembalian yang diharapkan sama dengan tingkat bebas risiko negara investor. Hal ini berlaku untuk setiap portofolio obligasi asing.

KESIMPULAN • • •

• •

• •

•

Model keseimbangan antarwaktu dari pasar modal internasional memperhitungkan risiko nilai tukar dan keberadaan suku bunga yang berbeda di seluruh dunia. Beberapa "teorema reksa dana" dapat diturunkan dengan implikasi penting 'untuk kebijakan investasi. Yang terpenting dari teorema ini menyatakan bahwa semua investor akan acuh tak acuh antara memilih portofolio dari aset asli atau dari tiga dana, yaitu: a. portofolio saham yang dilindung nilai terhadap risiko nilai tukar (pasar portofolio) b. portofolio obligasi, spekulatif dalam dimensi risiko nilai tukar c. aset bebas risiko negara mereka sendiri. 2 portofolio pertama tidak tergantung pada preferensi investor atau kewarganegaraan. Hubungan penetapan harga risiko untuk saham telah diturunkan yang menyatakan bahwa: “premi risiko keamanan apa pun atas tingkat bebas risiko nasionalnya sebanding dengan risiko sistematis internasionalnya” Koefisien proporsionalitas adalah premi risiko pasar dunia atas tingkat obligasi dunia. Seperangkat hubungan penetapan harga risiko lainnya menyatakan bahwa perbedaan antara tingkat bunga 2 negara sama dengan perubahan paritas yang diharapkan antara kedua negara ditambah istilah tergantung pada kovarians risiko nilai tukar. Ini menyiratkan bahwa nilai tukar forward adalah perkiraan bias dari nilai tukar masa depan. Perlunya reksa dana yang terdiversifikasi secara internasional. Faktanya, satu dana yang diinvestasikan di semua saham biasa (dengan bobot nilai pasar) akan memenuhi kebutuhan investor

dari negara mana pun. Namun perlu diingat bahwa hasil ini bergantung pada asumsi terbatas tentang ekspektasi homogen, pola konsumsi, kesempurnaan pasar modal.