International Capital Mobility in Emerging Markets: New Evidence

Text Preview:
Review of International Economics, 9(4), 626640, 2001

International Capital Mobility in Emerging Markets:
New Evidence from Daily Data

Michael Kumhof*



Abstract
This paper analyzes daily covered interbank interest differentials for three emerging markets before and
after the 1997/98 financial crises, and compares them with those of four developed economies. It examines
descriptive statistics of covered differentials and the long-run equilibrium (cointegrating) relationship
between their interest rate and forward discount components. Mean differentials and their volatility were
moderate before crises, but increased dramatically during crises. The main reasons are temporarily effective
capital controls, large bank default risk premia, and capital market imperfections. The evidence for a
cointegrating vector consistent with covered interest parity is strong, implying that, despite large short-term
deviations, covered interest parity does hold as an equilibrium relationship.


1. Introduction
Assumptions about countries' degree of integration into international capital markets
are key building blocks of macroeconomic models of open economies. The standard
assumption of "perfect capital mobility" states that domestic agents can freely borrow
and lend in international capital markets at the world real interest rate. It is difficult
to devise unambiguous tests of this hypothesis. Montiel (1993) contains an overview
of standard approaches used in the empirical literature, with specific reference to
emerging markets. These include using the magnitude of capital flows as an indicator,
testing for lack of effectiveness of sterilization, FeldsteinHorioka savinginvestment
correlations, Euler equation approaches, and interest arbitrage conditions including
covered, uncovered and real interest parities. As pointed out by both Montiel (1993)
and Frankel (1992, 1993), all but covered interest parity tests cannot be interpreted
unambiguously as tests of a country's integration into international capital markets.
  Covered interest parity states that, under full integration, capital flows should
equalize the returns on any two assets that differ only in their country of issue and
currency of denomination, while being identical in terms of maturity, liquidity, and
default risk. This requires that their interest differential be equal to the forward dis-
count between the two currencies. As shown by Obstfeld (1995), for eurocurrencies
the covered interest differential equals the onshoreoffshore interest differential, an
unambiguous measure of international capital market integration. Obstfeld (1995) and
Frankel (1993) show that for developed economies covered interest parity holds almost
perfectly, but for emerging markets this has long been hard to test for lack of good
data.


* Kumhof: Stanford University, Stanford, CA 94305, USA. In 2001/2: on leave at the IMF, Washington, DC
20431, USA. Tel: 202-623 9020; Fax: 202-623 4740; E-mail: kumhof@stanford.edu. The author thanks the IMF
for financial support. Helpful comments were received from two anonymous referees, Menzie Chinn, Sajjid
Chinoy, Michael Horvath, Subir Lall, and seminar participants at Columbia and Stanford. Giovanni Facchini
provided excellent research assistance. Help with obtaining data is gratefully acknowledged from Halim
Alamsyah (Bank Indonesia), the Bank of Thailand, Alejandro Werner (Banco de Mexico), and Maria de las
Nieves Rebollar (Banamex).


 Blackwell Publishers Ltd 2001, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA
                                CAPITAL MOBILITY IN EMERGING MARKETS                          627

   This paper uses daily data on emerging-market interest rates and forward exchange
rates which have recently become available. Econometric tests taking account of the
nonstationarity of interest rates and forward discounts are used. The covered interest
parity hypothesis is tested only for the country's banking sector.
   The main difficulty in testing for covered interest parity in emerging markets has
been that liquid forward foreign exchange markets with publicly quoted prices did not
exist until very recently. For larger emerging markets, forward markets were created
in the 1990s, but only monthly data were typically available and used in studies such
as that of de Brouwer (1997). An arbitrage condition like covered interest parity
should, however, hold continuously, making data of higher frequency useful. Daily data
are now available on Bloomberg from the mid-1990s for some large emerging markets
including Indonesia, Mexico, and Thailand. They are used in this study, and compared
with data for four European economies from the same source.
   The existing literature often tests for covered interest parity by regressing the
forward discount on the nominal interest differential. For an example, see Chinn and
Frankel (1994).This assumes that nominal interest differentials are stationary, but espe-
cially between emerging market currencies and the US dollar they are often highly
nonstationary. In that case ordinary least squares leads to spurious results if the inter-
est differential and the forward discount are not cointegrated, and conventional rejec-
tion criteria lead to faulty inference even if they are. This paper therefore analyzes only
descriptive statistics of covered differentials and the long-run equilibrium (cointegrat-
ing) relationship between their interest rate and forward discount components. It is
found that mean differentials and their volatility were moderate before financial
markets crises, but increased dramatically during crises. The main reasons are tem-
porarily effective capital controls, large bank default risk premia, and a variety of
capital market imperfections. The evidence for a cointegrating vector consistent with
covered interest parity is strong, implying that, despite large short-term deviations,
covered interest parity does hold as an equilibrium relationship.
   Interest parity tests with specific sets of interest rates are commonly interpreted
as evidence on the integration of the whole country into international capital
markets. This is likely to be an overinterpretation, as the capital market integration
of different sectors may differ greatly. Chinn and Dooley (1997) examine the firm
sector, through tests of interest parities for lending interest rates. They find that
"capital is not completely mobile . . . because the market in bank lending is not
well integrated with either domestic or international capital markets." This paper
examines only the domestic banking sector, by using the domestic interbank rate for
parity tests.


2. Emerging Financial Markets
This section provides some country-specific information, mainly on Indonesia and
Thailand, which will be useful to understand and interpret the empirical results. For a
comprehensive analysis of many aspects of the Asian crisis, the reader is referred to
Rivera-Batiz (2000). The focus here is more narrowly on the institutional features char-
acterizing financial markets in these countries. The discussion is based on conversa-
tions with International Monetary Fund officials and on Lall (1997) and International
Monetary Fund (1997).
   Forward foreign exchange and money markets in Thailand and Indonesia were
highly segmented before and especially immediately after the Asian crisis, while
Mexico's markets were far more liquid and far less segmented. The following remarks

                                                                      Blackwell Publishers Ltd 2001
628      Michael Kumhof

describe common features of Thai and Indonesian financial markets.1 They are fol-
lowed in two subsections by country-specific information.
   The forward foreign exchange markets in Thailand and Indonesia did not always
quote prices on a fully hedged basis; i.e., based on domesticforeign interest differen-
tials. This was true especially after the Asian crisis, and reflects segmentation between
money and foreign exchange markets. Quotes were instead provided on a speculative
basis, based on depreciation expected by currency traders.
   Money markets themselves were very highly segmented into two separate markets,
one between strong (creditworthy) banks and another between the remaining weaker
banks. Strong banks were able to rely on retail deposits and, before the crisis, on inter-
national borrowing. These banks commanded low money market borrowing rates and
were mostly net lenders to the central bank. Weak banks had access to a separate
money market with higher interest rates. These banks were net borrowers from the
central bank, which therefore effectively intermediated funds from strong to weak
banks.
   Market participants state that the forward discount accurately reflected expected
depreciation, while interest rates did not either because of heavy market intervention
by the central bank or because of significant additional counterparty risk premia.


Thailand
The Thai baht had been under speculative pressure for several months before its final
collapse in July 1997. International Monetary Fund (1997) reports that currency attacks
took place in July 1996, January 1997, and May 1997.
  The Bank of Thailand's defense of its currency in 1997 can be broken into two
phases. Before 15 May it intervened heavily in the forward market, effectively com-
mitting a major portion of its reserves. On 15 May it allowed interest rates to rise and
switched to imposing an array of capital controls on local banks. These measures pre-
vented speculative pressure from hitting the spot market, and created a liquidity
squeeze for holders of short baht positions. This led to very high offshore baht bor-
rowing rates and a very strong offshore baht exchange rate for a short period. Because
the large price differences between offshore and onshore markets provided a strong
incentive to circumvent the controls, speculative pressure continued. On 2 July the
Bank of Thailand gave up its currency defense. Capital controls remained in place,
but had become ineffective by late July. This led to additional controls which became
difficult to enforce. Many, but not all, controls were removed in early 1998.
  Thai financial institutions were known to have large currency mismatches, and very
poor loan books especially because of exposure to a weak property sector. They there-
fore had to pay large risk premia which increased steeply following the crisis.


Indonesia
In Indonesia the money/capital markets consisted of two main segments, the interbank
money market and the market in short-dated central bank intervention bonds, Serti-
fikat Bank Indonesia (SBI). Foreign banks and investors had full access to the SBI
market, but hardly participated in the interbank market, which involved much higher
counterparty credit risk.
   Indonesian banks and corporates had been optimistic about the rupiah for the first
half of 1997, and took long rupiah positions via substantial foreign borrowing, swaps,
and options. They continued to do so for several days following the Thai devaluation.

 Blackwell Publishers Ltd 2001
                                  CAPITAL MOBILITY IN EMERGING MARKETS                         629

This fact, and the poor quality of their loan book, meant that following the specula-
tive attack on the rupiah they had to pay very large risk premia. This explains why the
interbank rate, which is calculated as the average of the borrowing rates paid by some
of the largest banks, exceeded the SBI rate by a large margin. As will be seen, it was
these risk premia which mostly explain deviations from covered interest parity in the
case of Indonesia.


3. Data and Their Time Series Properties
Tables 1 and 2 contain information on data sample periods and the chosen sample
break points for the seven economies studied. In all cases the domestic interest rate is
an onshore domestic one-month interbank borrowing rate, the foreign interest rate
is the one-month LIBOR, and the forward discount is calculated from the offshore
forward foreign exchange rate. Figures 14 show the time-series graphs. With one
exception all data are from Bloomberg. The Mexican forward discount data are from
Banamex, a commercial bank, which was able to furnish a longer series than
Bloomberg.
   Because of the qualitative changes in emerging markets' time series during times of
crises, the paper separately examines the full sample, the pre-crisis and the post-crisis
series for each country.2 The full sample in all cases has at least 465 observations. For
developed economies, for purposes of comparison, the sample periods were chosen to
be approximately equal to those available for the three emerging economies. One addi-
tional data series from September 1992 through April 1994 (400 observations) for


Table 1. Data for Emerging Markets

                                  Indonesia              Thailand                 Mexico

Beginning of sample         7 October 1996         2 March 1997           1 April 1996
End of sample               28 January 1999        28 January 1999        28 January 1999
No. of observations         543                    465                    681
No. of missing values       13                     10                     5
Not available before        Forward rate           Forward rate           I'bank rate (TIIE)
Sample break date           11/14 July 1997        14/15 May 1997         7/10 August 1998
Event at break              Jump in I'bank rate    Jump in I'bank and     Jump in I'bank
                              (spot rate jumped       forward rate           and spot rate
                              1 week later)           (spot rate jumped
                                                      45 days later)
Observations before break   181                    61                     573
Observations after break    362                    404                    108



Table 2. Data for Developed Markets

                            France            Germany          Ireland                Portugal

Beginning of sample     1 April 1996      3 April 1996      3 April 1996           3 April 1996
End of sample           28 January        27 January        30 December            28 January
                          1999              1999              1998                   1999
No. of observations     693               692               675                    670


                                                                       Blackwell Publishers Ltd 2001
Download Link:
Share Link: Forum Link:

More on Business

  • Picture: Gr 10 Maths Literacy - Paper 1

    Gr 10 Maths Literacy – Paper 1

    File Size: 249.96 KB, Pages: 10, Views: 1,340,811 views

    GRADE 10 EXEMPLAR EXAMINATION NOVEMBER 2006 MATHEMATICS LITERACY PAPER 1 Minimum time: 1 hours Maximum time: 2 hours 75 marks PLEASE READ THE FOLLOWING CAREFULLY 1. This paper consists of: 5 questions an answer sheet with grid paper for question 4 (d). 2. Answer all …
  • Picture: Improving cash flow using credit management - CIMA

    Improving cash flow using credit management – CIMA

    File Size: 374.27 KB, Pages: 28, Views: 1,270,087 views

    Improving cash flow using credit management The outline case sponsored by Improving cash flow using credit management sponsored by Albany Software focuses on developing award-winning software to transform financial processes and is the market leader in electronic payment solutions. Albany makes financial transfers and subsequent …
  • Picture: MATERIAL SAFETY DATA SHEET - compressorparts.com

    MATERIAL SAFETY DATA SHEET – compressorparts.com

    File Size: 311.77 KB, Pages: 8, Views: 521,290 views

    Product Name: MOBIL RARUS SHC 1025 Revision Date: 16Feb2007 Page 1 of 8 _____________________________________________________________________________________________________________________ _ MATERIAL SAFETY DATA SHEET SECTION 1 PRODUCT AND COMPANY IDENTIFICATION PRODUCT Product Name: MOBIL RARUS SHC 1025 Product Description: Synthetic Base Stocks and Additives Product Code: 606343-00, 973764 Intended Use: …
  • Picture: Onan RV Generator Troubleshooting Guide - Flight Systems, Inc.

    Onan RV Generator Troubleshooting Guide – Flight Systems, Inc.

    File Size: 2,729.64 KB, Pages: 18, Views: 37,558 views

    RV GENERATOR TROUBLESHOOTING GUIDE A Service of FLIGHT SYSTEMS 207 Hempt Road Mechanicsburg, PA 17050 Ph: 717 590 7330 Fax: 717 590 7327 www.flightsystems.com CONTENTS RV Generator Troubleshooting Overview 1 RV Generator FAQs 5 Onan RV Generator Charts: Applications Chart 8 Test Information for Troubleshooting …
  • Picture: Interpretive Guide for the Achievement Levels Report (2003

    Interpretive Guide for the Achievement Levels Report (2003

    File Size: 624.72 KB, Pages: 19, Views: 38,998 views

    Interpretive Guide for the Achievement Levels Report (2003 Revision) ITBS/ITED Testing Program The purpose of this Interpretive Guide is to provide information to individuals who will use the Achievement Levels Report for monitoring the achievement of student grade groups, both at the building level and …

Leave a Reply

Your email address will not be published. Required fields are marked *