GLOBAL INTEGRATION OF INDIA'S MONEY MARKET: INTEREST RATE PARITY IN INDIA

WORKING PAPER NO 164





GLOBAL INTEGRATION OF INDIA’S MONEY MARKET:
INTEREST RATE PARITY IN INDIA






Vipul Bhatt
Arvind Virmani









JULY 2005









INDIAN COUNCIL FOR RESEARCH ON INTERNATIONAL ECONOMIC RELATIONS
Core-6A, 4th Floor, India Habitat Centre, Lodi Road, New Delhi-110 003
Website: www.icrier.org

---

GLOBAL INTEGRATION OF INDIA’S MONEY MARKET:
INTEREST RATE PARITY IN INDIA






Vipul Bhatt
Arvind Virmani









JULY 2005


The views expressed in the ICRIER Working Paper Series are those of the author(s) and do not
necessarily reflect those of the Indian Council for Research on International Economic Relations
(ICRIER).

---

CONTENTS
Page No
FOREWORD
I
1
INTRODUCTION
1
2
LITERATURE REVIEW
2
3
MODEL AND ESTIMATION
4
3.1
Estimating Equations
4
3.2
Econometrics
6
3.3
Data Sources
7
4
RESULTS
9
4.1
Stationarity and Co-integration
9
4.2
Covered Interest Parity
10
4.3
Un-covered Interest Parity
11
4.4
Exchange Risk and RBI Intervention
12
5
CONCLUSIONS
13
6
REFERENCES
15
7
APPENDIX: TESTS
17
7.1
Order of integration
17
7.1.1
Sequential ADF Test for unit root
17
7.1.2
Phillips-Perron Test
18
7.2
Test for Co-integration: Johansen’s Methodology
18
7.3
Call Money Assymmetry
19

TABLES
Table 1 : Unit Root Test Results................................................................................................ 9
Table 2: Results of Johansen’s Co-integration Test................................................................ 10

FIGURE
Figure 1: 3-Month Forward Premium (F) and India-US Interest Differential (Idiff)
8

---

Foreword

Financial openness exists when residents of one country are able to trade
assets with residents of another country, i.e. when financial assets are traded goods. A
weak definition of complete financial openness, which one might refer to as financial
integration, can be given as a situation in which the law of one price holds for
financial assets- i.e. domestic and foreign residents trade identical assets at the same
price. A strong definition would add to this the restriction that identically defined
assets e.g. a six-month Treasury bill, issued in different political jurisdictions and
denominated in different currencies are perfect substitutes in all private portfolios.
The degree of financial integration has important macroeconomic implications in
terms of the effectiveness of fiscal and monetary policy in influencing aggregate
demand as well as the scope for promoting investment in an economy.

The paper shows that the short-term (up to 3 month) money markets in India
are getting progressively integrated with those in the USA even though the degree of
integration is far from perfect. Covered interest parity is found to hold for while
uncovered interest parity fails to hold. The difference between the two can be
attributed to the existence of an exchange risk premium over and above the expected
depreciation of the currency. Analysis of RBI interventions in response to foreign
exchange shocks suggests that these may play a role in the deviations from interest
parity. Further work needs to be done however on this as well as on instruments of
other maturity such as 1 month and 6 month (for which consistent data was not
available).

Arvind Virmani
Director & Chief Executive
ICRIER


July 2005


i

---

1
INTRODUCTION
Financial openness exists when residents of one country are able to trade
assets with residents of another country, i.e. when financial assets are traded goods. A
weak definition of complete financial openness, which one might refer to as financial
integration, can be given as a situation in which the law of one price holds for
financial assets- i.e. domestic and foreign residents trade identical assets at the same
price. A strong definition would add to this the restriction that identically defined
assets e.g. a six-month Treasury bill, issued in different political jurisdictions and
denominated in different currencies are perfect substitutes in all private portfolios.
The degree of financial integration has important macroeconomic implications in
terms of the effectiveness of fiscal and monetary policy in influencing aggregate
demand as well as the scope for promoting investment in an economy.
The free and unrestricted flow of capital in and out of countries and the ever-
increasing integration of world capital markets can be attributed to the process of
Globalization. The benefits of such integration are liquidity enhancement on one hand
and risk diversification on the other, both of which are instrumental in making
markets more efficient and also facilitate smooth transfers of funds between lenders
and borrowers. India began a very gradual and selective opening of the domestic
capital markets to foreign residents, including non-resident Indians (NRIs), in the
eighties. The capital market opening picked up pace during the nineties. In this paper
we try and estimate the degree of financial integration between India and the rest of
the World, by focussing on the degree of integration of the Indian money market with
global markets.
Frenkel (1992) in his review of Capital Mobility measurement outlined four
different definitions of perfect capital mobility that are in widespread use, of which
three are of relevance to the current paper. These are real interest parity, uncovered
interest parity and covered interest parity. (i) Real interest parity hypothesis states
that international capital flows equalise real interest rates across countries. (ii)
Uncovered interest parity states that capital flows equalise expected rates of return on
countries’ bonds regardless of exposure to exchange risk. (iii) Covered interest parity
states that capital flows equalise interest rates across countries when contracted in the
same currency. Frenkel (1992) shows that these three definitions are in ascending
order of specificity in the following sense. Only definition (iii) that the covered

1

---

interest differential is zero is an unalloyed criterion for “capital mobility” in the sense
of the degree of financial market integration across national boundaries. Condition
(ii) that the uncovered interest differential is zero requires that (iii) hold and that there
be zero exchange risk premium. Condition (i) that the real interest differential be zero
requires condition (ii) and in addition that expected real depreciation is zero.
2
LITERATURE REVIEW
The uncovered interest parity (UIP) theory states that differences between
interest rates across countries can be explained by expected changes in currencies.
Empirically, the UIP theory is usually rejected assuming rational expectations, and
explanations for this rejection include that expectations are irrational, see Frankel and
Froot (1990) and Mark and Wu (1998), or that time-varying risk premia are present,
see Domowitz and Hakkio (1985) and Nieuwland et al. (1998), respectively. In a
survey of 75 published estimates, Froot and Thaler (1990) report few cases where the
sign of the coefficient on interest rate differentials in exchange rate prediction
equations is consistent with the un-biased-ness hypothesis and not a single case where
it exceeds the theoretical value of unity. This resounding unanimity on the failure of
the predictive power of interest differentials is virtually unique in the empirical
literature in economics.
A third explanation was provided by McCallum (1994a), who observes that
regressing the change in spot exchange rates on the forward premium, one typically
finds a negative regression parameter of -4 to -3 contrary to the expected parameter of
+1. McCallum argues, however, that this finding may be consistent with the UIP
theory, if one introduces policy behavior. Assuming policymakers adjust interest rates
in order to keep exchange rates stable, and that they are interested in smoothing
interest rate movements, McCallum derives a reduced form equation for the spot
exchange rate under rational expectations. In fact, this results in a negative theoretical
relationship between the change in the spot exchange rate and the forward premium
consistent with his empirical findings. Christensen, M. (2000) extend the data set used
by McCallum to include the recent 8 years and find that $/DM, $/£ and $/Yen for the
period 1978.01m to 1999.03m behave amazingly well according to the modified UIP
theory developed by McCallum. However, when he estimates the policy reaction
function, its structural parameters are inconsistent with the UIP relationships
estimated.

2

---

Nevertheless, there appears to be overwhelming empirical evidence against
UIRP, at least at frequencies less than one year (see Hodrick (1987), Engel (1996) and
Froot and Thaler (1990)). Fama (1984) focuses on statistical properties of this
relation. He finds that from the end of August 1973 to the end of 1982, the variance of
the exchange risk premium has been large, exceeding the variance of expected future
spot rates changes of the dollar against each of ten other major currencies (over
monthly intervals). On the other hand Frankel and Froot (1987), among others,
propose an explanation of UIP deviations based on the existence of asymmetries
between currencies. Using survey data to approximate the exchange rates’ behaviour,
they show that agents were expecting a 10% depreciation of the Dollar against the
Mark over 1981-85 whereas the differential in corresponding interest rates was only
around 4%. Given that this empirical evidence has not stopped theorists from relying
on UIRP, it is fortunate that recent evidence is more favourable. Bekaert and Hodrick
(2001) and Baillie and Bollerslev (2000) argue that doubtful statistical inference may
have contributed to the strong rejections of UIRP at higher frequencies. Chinn and
Meredith (2001) marshal evidence that UIRP holds much better at long horizons.
They test this hypothesis using interest rates on longer-maturity bonds for the U.S.,
Germany, Japan and Canada. The results of these long horizon regressions are much
more positive — the coefficients on interest differentials are of the correct sign, and
most are closer to the predicted value of unity than to zero. Ravi Bansal and Magnus
Dahlquist (2000) conclude that the often found negative correlation between the
expected currency depreciation and interest rate differential is, contrary to popular
belief, not a pervasive phenomenon. It is confined to developed economies, and here
only to states where the U.S. interest rate exceeds foreign interest rates
The covered interest parity (CIP) postulates that interest rates denominated in
different currencies are equal once you cover yourself against foreign exchange risk.
Unlike the UIP, there is empirical evidence supporting CIP hypothesis. Empirical
studies such as Frenkel and Levich (1975, 1977, 1981), Frankel (1989), among others,
find that the CIP holds in most cases on the Eurocurrency market (where remunerated
assets have similar default and political risk characteristics) since the collapse of the
Bretton Woods regime in early 1970’s. Lewis(1995) shows that risk premia do not
vary significantly and often switch sign, contrary to what the observed stability of the
countries’ global creditor or debtor status would predict. However she explains that
not only the conditional variance of exchange rate is not significant enough to account

3

---

for risk premia movements, but also that risk premia examined in the short run should
concern capital flows and investors with similar temporal horizons, such as currency
traders, hedge funds and mutual funds managers. Frankel (1991) reports mean
covered interest differentials (CIDs) for the period 1982 to 1987 for a selection of
developed and developing economies using monthly observations of the 3-month
local money market rate against the equivalent Eurodollar rate. Focusing on the East
Asian economies in the sample – Japan, Hong Kong, Malaysia and Singapore – the
null of a zero differential is rejected for the first three economies, though only
marginally in that the CIDs are very low. Chinn and Frankel (1992) found that the
CIDs were small for Japan, Hong Kong and Singapore, but large for Malaysia.
In the Indian context, Varma (1997) has undertaken an analysis of the covered
interest parity. His posits a structural break in the money market in India in
September 1995, with CIP become effective from that point on for the first time in the
Indian money market. The structural break itself is attributed to interplay between the
money market and the foreign exchange market. The period after 1995 is however
witness to several deviations from the CIP. Varma has used rates on Treasury bills,
certificates of deposit and commercial paper and call money rate to analyse the Indian
money market. For the foreign rate he has calculated an implicit euro-rupee rate for
six, three and overnight maturity. Thus he uses a mix of actual and constructed rates
of different maturity. A rigorous test requires use of interest rates on identical
instruments (e.g. maturity, risk) and a consistent forward rate (period of forwards
should be identical to that of instruments). This is perhaps the first time that such a
test is being carried out for India.
3
MODEL AND ESTIMATION
3.1
Estimating Equations
One of the key implications of international financial integration is on the
degree of movement/co-movement of interest rates in countries over time and their
comparison in terms of convergence or having a common trend. The relationship
between two countries’ interest rates is termed as interest rate parity.
The interest rate theory proposes that given perfect capital mobility, perfect
capital market and fixed exchange rates the interest on identical assets (identical in
terms of maturity etc) would be equal across countries. However, in the real world

4

---

with capital controls, flexible exchange rates and imperfect capital markets
divergence between interest rate is frequently observed and persist over long periods.
Given the reality of non-frictionless capital markets and flexible exchange rates the
recent versions of the interest parity theorem attribute this divergence to the
expectation about exchange rate movements. Based on the preference individuals
have for risk there are two versions of this basic relation:

a)
Uncovered Interest Rate Parity- Assume that individuals are risk neutral. With
no capital controls and perfect capital markets the interest differential between
two countries is equal to change in exchange rate:

it – it* = St+1-St .
where
it is domestic interest rate
it*/ is foreign interest rate on similar asset ( identical in all respects except for
yield and currency denomination)
St is the spot exchange rate.
A risk neutral person would replace St+1 by his expectation about future
exchange rate. So we get
it – it* = E(St+1) – St
Any deviation from UIP can be attributed to currency associated risks in the
absence of hedging agreements- namely currency premium and expectation bias.

b) Covered Interest Parity- Assume that individuals are risk averse. Such an
individual would like to cover himself for any unexpected currency fluctuation
during the tenure of the deal. Given the forward contract market, he would
purchase a forward contract and use the exchange rate mentioned in the contract.
Then any difference in interest rate should be equated to forward premium. This
is called CIP:

it – it* = Ft- St
or
it – it* = ft
where Ft is forward rate and ft is forward premium.

5

---

Any deviation from CIP would suggest that the markets are inefficient,
regulations like capital controls exist and costs like sovereign risk, individual
borrowing constraints are not accounted for.
3.2
Econometrics

To test the basic relation of interest rate parity we can think of a linear regression of
the following type:
Equation 1: ∆St = α + β ( it – it*) + εt

For Uncovered Interest Parity we would expect α to be 0 and β to be equal to 1.

For covered interest parity we would use the following regression:
Equation 2: ft = α + β ( it – it*) + εt

and then test for β = 1.
The problem with using Ordinary Least Square as an estimation technique
relates to the issue of non-stationarity of the time series involved in the above
equation. In case of non-stationary times series the estimate of β would be spurious
and biased. However if we can show that the two variables in question are
cointegrated than the OLS estimates are super consistent and would converge to their
true value faster (see). Thus before drawing inferences based on the results of
ordinary least squares it is imperative to check the variables namely F (3-month
forward premium) and IDIFF (3-month TB auction rate differential between India and
U.S). In case the two series are integrated of the same order we can then test for
cointegartion between the two non-stationary variables.
For covered interest parity we need to test for β = 1 where β is the coefficient
of IDIFF. Formally,

Ho: β = 1 - Covered Interest Rate Parity holds
H1: β ≠ 1 – There is no interest rate convergence.

The above test uses a standard t- statistic given by:
t = (β - 1)/σβ ~ tn-2

6

---