A Decade of Economic Reforms - Review by RBI
[Source: RBI Report on Currency and Finance 2001-2002 dated March 31, 2003]

Module: 5 - External Sector - Capital Account, External Debt and
Exchange Rate: Approach, Development & Issues

Foreign Exchange Reserves: Approach, Developments and Issues

International Reserves and Optimality

he policy of accretion to reserves to meet the transactions and precautionary needs generally involves both financial and economic cost to the country. In the absence of any unique definition of the opportunity cost, one comes across several interpretations in the literature, the applicability of which depends on the particular economic context. When an economy is both foreign exchange and saving constrained, the opportunity cost could be the rate of return on domestic investment (assuming that the foreign exchange used to finance unsatiated investment demand would have fetched the return on domestic investment). When an economy is only foreign exchange constrained, the opportunity cost could be in terms of foregone consumption. When an economy is not foreign exchange constrained ( i.e., all productive forex demand are met before the reserves are built up), the opportunity cost would depend on the difference between the cost of borrowing and the return on reserve assets. If high cost borrowings are used to build reserves on which modest returns are obtained because of the emphasis on safety and liquidity of reserve assets, such reserve build-up policies may prove more costly. Another form of cost - often known as the quasi-fiscal cost - arises from the higher rate of return on domestic assets vis-à-vis foreign assets. Since sterilised intervention gives rise to an offsetting change in the domestic asset holding of a central bank when its foreign assets increase, the overall profits arising from the total asset portfolio of the central bank’s balance sheet may decline. As a result, the profits transferred to the Government -which represents a major source of non-tax revenue for the Government - may decline. For countries operating with large fiscal imbalances, such a decline in non-tax revenue could prove more costly, forcing a higher mobilisation of tax revenue or a cut in expenditure that may affect growth and development.

The development of money demand theory along the lines of optimal inventory control or buffer stock model provides a natural theoretical benchmark for the study of optimal reserve accumulation and has spurred a large body of empirical literature in the late 1960s and early 1970s (Grubel, 1971; Kelly, 1970). Heller (1966) used a cost-benefit approach and stressed that the issue is one of choosing an optimal level of reserves at which the central bank minimises the total expected cost i.e., the sum of the adjustment cost that is incurred when reserves reach some lower bound and the opportunity cost. Heller’s estimation indicated that optimal reserves are directly proportional to the variability of exports and are inversely related to the opportunity cost variable (defined as foregone earnings or economic welfare). Frankel and Jovanovic (1981) and Landell-Mills (1989) also found a negative relationship between the demand for reserves and its opportunity cost. Flood and Marion (2001), however, did not find any relationship between reserves and opportunity cost. The sign of the propensity to import has also turned out to be ambiguous in the literature. Several studies have found an inverse relationship, (Heller, 1966; Kelly, 1970; Heller and Kahn, 1978; Landell-Mills, 1989) while some others found a positive relationship between import propensity and reserves (Frankel, 1974; Edwards, 1984). While the inverse relationship between reserves and the propensity to import has been interpreted as the extent of adjustment to an external shock, the direct relationship is an indicator of the degree of openness. Eaton and Gersovitz (1980) formally introduced the concept of transactions demand for reserves. In addition to the ‘conventional’ variables, Ben-Bassat and Gottlieb (1992) introduced risk of default or sovereign risk into the assessment of precautionary demand for reserves. Lane and Burke (2001) adopted a broad approach to identify the potential determinants of reserves and found:

1. trade openness to be the most important variable along with financial deepening (M2/GDP);

2. smaller and more volatile industrial countries hold larger reserves than their larger, less volatile counterparts; and

3. more indebted developing countries had smaller reserve ratios. The traditional determinants identified in the empirical literature on foreign exchange reserve holdings are presented in Table below.

Analytically and empirically, reserve holding appears to be an under-researched area. Clear standard methodologies have yet to evolve on this subject. However, relatively high demand for international reserves by countries in Far East and relatively low demand by some other countries has attracted attention. Aizenman and Marion (2002) recently showed that for 125 developing countries, reserve-holding over 1980-1996 period could be predicted by the size of international transactions, their volatility, the exchange rate arrangement and political considerations. Sovereign risk and fiscal liabilities led to relatively large precautionary demand for reserves.

Aizenman and Marion (2002) showed that there has been a structural break in the equation for the demand for reserves in several Asian countries that were affected by crisis in 1997. Using a standard estimating equation, they found that over the period 1960-96, their model over-predicted the reserve holdings of countries such as China, Taiwan, Hong Kong, South Korea and Singapore implying thereby that their reserves were low in relation to what was estimated as desirable in the model. Out of sample forecasts of desirable levels generated for 1997-99, however, under predicted the reserve holding (i.e., actual reserves holdings exceeded the predicted desirable level). Such a result has to be assessed in the context of the overall change in policy stance of the emerging market economies in the aftermath of the Asian crisis that viewed high reserves as an appropriate policy of "self-insurance".

In India, the determinants of reserve holdings appear to have changed considerably and evolved over the recent years with the gradual opening up of the capital account, the Asian crisis (with contagion emerging as a significant determinant), the imposition of sanctions following the nuclear explosion in Pokhran in May 1998 and the subsequent rating downgrade on foreign currency borrowings. These new factors, by their very qualitative nature, are difficult to quantify. Taking recourse to exceptional external borrowings in the form of RIBs (1998-99) and IMDs (2000-01) could be interpreted as the manifestation of the precautionary demand for reserves by the authorities; but for RIBs and IMDs, accretion to reserves during these two years would have been negative.

The substantial growth in reserves since 2001-02 has generated a debate regarding the cost of holding reserves. While the cost of reserves is secondary to properly meeting the overall objective behind holding reserves, it is important to note that in India, in the last few years, almost the whole addition to reserves has been made without increasing the overall level of external debt, which has hovered around US $ 100 billion during the previous five years. The increase in reserves largely reflects higher remittances, quicker repatriation of export proceeds and non-debt flows (RBI, 2003). Even after taking into account foreign currency denominated NRI flows (where interest rates are linked to LIBOR), the financial cost of additional reserve accretion in India in the recent period is low (RBI, 2002).

While the reserve build up policies of the emerging market economies like India reflect the importance of appropriate reaction to the vastly altered conditions prevailing at both national and global levels, high reserve policies also entail several other implications, particularly for monetary management and in terms of the quasi-fiscal costs, both of which pose a different type of challenge to the policy makers.

Cross-country experiences of surges in capital inflows indicate that in the context of the limited capacity of the economy to absorb capital flows in the form of higher productive investment and the resultant implications for the exchange rate, monetary authorities often intervene in the foreign exchange market to absorb the surplus in the market and thereby avoid nominal appreciation of the exchange rate. A non-sterilised (or partially sterilised) intervention can, however, cause a sharp rise in the monetary base and hence higher inflation. Real appreciation resulting from higher inflation could erode external competitiveness; lower interest rates could also fuel lending and consumption boom, that can potentially lead to a sharp deterioration in the current account balance and culminate in a possible currency crisis. When the conflict between the policy objectives of checking nominal appreciation and limiting inflation emerges, central banks attempt to counter inflation through sterilised intervention, which by nature appears money supply neutral.

Among the instruments available for sterilisation, recourse to open market operations (OMOs) is particularly effective if inflows are temporary and there exists a near perfect elastic demand for domestic government securities. However, if the demand for government securities is not perfectly elastic in view of the limited absorptive capacity of market participants and the underdeveloped nature of the financial markets, OMOs can cause domestic interest rates to rise, nullifying the impact of sterilisation as higher interest rates could attract larger capital inflows. Sterilisation typically involves exchanging high-yielding domestic assets for low-yielding foreign assets and, therefore has quasi-fiscal costs. The degree of effectiveness of sterilised intervention would therefore depend on: (i) the sensitivity of domestic interest rates to OMOs and (ii) the degree to which foreign capital flows respond to such interest rate variations. The combined effect, which is captured through the estimated "offset coefficient", indicates that the value of the coefficient may range between 0 and (-) 1, with values close to zero indicating "effectiveness of sterilisation" and values close to (-) 1 indicating "ineffective sterilisation". Over the period April 1993 to March 1997, the offset coefficient for India turned out to be (-) 0.3, suggesting that sterilisation of the surges in capital flows experienced during the 1990s was effective (Pattanaik, 1997). Sterilisation induced increase in interest rates, however at times, may be more than offset by the change in the stance of monetary policy. When capital flows are persistent and sterilisation proves ineffective, cross-country studies show that countries often attempt a few options available to them. These include: fiscal adjustment, easing of restrictions on capital outflows, accelerated trade liberalisation, lower interest rates on foreign currency deposits, prepayment of costly debt, adoption of new sterilisation techniques such as foreign exchange swaps, switching of government deposits from the banking system to the central bank, imposition of taxes on domestic assets purchased by foreigners, fixing of ceilings on foreign borrowings by domestic residents, and recourse to indirect capital controls, such as variable unremunerated reserve requirements on certain categories of foreign borrowing. In the Indian case, capital flows have had a softening effect on interest rates and positive effect on broad money growth.

Capital inflows have served the twin purpose of meeting India’s domestic investment-saving gap and the need for reserve accretion consistent with the standard indicators of reserve adequacy. The latter objective has become more prominent recently in tandem with the pattern seen in many Asian countries in the aftermath of the Asian crisis.

The surge in capital flows often poses a challenge to the conduct of monetary and exchange rate policy. As a result of large accretion to reserves resulting from surges in capital flows in the face of low domestic absorption, there has been a substantial increase in net foreign exchange assets (NFA) of the Reserve Bank. In the period since 1992-93, the NFAs have grown at a higher rate than net domestic assets (NDAs) (excepting 1995-96). In fact, NDAs held by the Reserve Bank had to be brought down by sterilisation through OMOs (particularly since 1996-97) to check the expansion in the monetary base. The extent of sterilisation, however, varied from year-to-year, depending on the requirements to modulate liquidity conditions. With sustained surge in capital flows since 1999-2000, the NDAs of the Reserve Bank even recorded a decline in absolute terms on account of significant OMOs (net sales) .

There is hardly any formal analysis available for emerging market economies (including India) to evaluate the monetary impact of large capital flows econometrically. One way to assess the impact is to simulate the emerging monetary condition in counterfactual scenarios where reserve accretion is placed at lower levels. For this purpose, an interest rate reaction function has been estimated that relates domestic interest rate to its possible determinants like inflation, output and capital flows along with a conventional money demand equation in terms of income and interest rate. Controlling for the impact on interest rates from the accretion to reserves, the counterfactual path generated through a simulation exercise under lower reserves accumulation points towards a perceptible monetary impact of reserves in the 1990s in terms of higher broad money demand.

In the period of surge in capital flows and significant accretion to foreign exchange reserves with the monetary authority, the literature points to the emergence of newer channels of monetary expansion, especially, if there is a disassociation between the growth of broad money and reserve money. In India, over the 1980s and 1990s, notwithstanding a fall in reserve money growth, broad money grew almost at the same level in the two decades, which may be symptomatic of such a disassociation. The stylised simulation model in Box VII.7 is an attempt at exploring the likely mechanism that characterises the underlying process of monetary expansion. The exercise suggests that had there been a lower reserve accumulation with the Reserve Bank, the interest rate would perhaps have been higher, thus, possibly signifying that the liquidity effect of reserve accumulation would have outweighed the sterilisation effect in the Indian context.

In the 1990s, composition of reserve money expansion has shifted from domestic assets to foreign assets reflective of the surge in capital flows, accumulation of reserves and sterilisation operations of the Reserve Bank. Sterilisation operations have so far been successful in arresting reserve money growth to levels where the inflationary potential of capital flows has been kept in control. However, future concerns on monetary management may arise, among others, from the following factors:

• need to ensure sufficient stock of Government securities at the hands of the Reserve Bank for sterilisation operations of the required magnitude;

• the absorption capacity of the financial system for Government securities once the credit demand picks up; and

• the upward drift in money multiplier arising partly from fall in currency-deposit ratio and CRR has sustained higher growth of broad money in the face of declining reserve money growth. With financial liberalisation and market-determined interest rates, there is evidence that innovations to money multiplier emanate from interest rates and other macroeconomic variables (such as growth) apart from the conventional proximate determinants (Jha and Rath, 2001). To the extent that these newer determinants of money multiplier are not entirely under the control of the monetary authority, it poses a challenge for the Reserve Bank to manage the emerging monetary conditions