This Selected Issues paper attempts to uncover the long-term determinants of the demand for foreign exchange reserves in Tunisia. It assesses the adequacy of current and projected reserves holdings in light of the country’s policy choices. The paper describes recent trends in foreign exchange reserves in Tunisia. Econometric evidence on the determinants of the demand for foreign reserves in Tunisia is presented. The results are used to forecast the desired level of reserves given Tunisia’s medium-term macroeconomic framework and to draw policy implications.

Abstract

This Selected Issues paper attempts to uncover the long-term determinants of the demand for foreign exchange reserves in Tunisia. It assesses the adequacy of current and projected reserves holdings in light of the country’s policy choices. The paper describes recent trends in foreign exchange reserves in Tunisia. Econometric evidence on the determinants of the demand for foreign reserves in Tunisia is presented. The results are used to forecast the desired level of reserves given Tunisia’s medium-term macroeconomic framework and to draw policy implications.

I. Assessing Reserves Adequacy in Tunisia 1

A. Summary

1. Tunisia foreign exchange reserves increased over the last decade. They rose from 1.6 months of imports and 62 percent of short term external debt in 1992 to about 3 months of imports and 100 percent of short term external debt in 2003.

2. The demand for foreign exchange reserves is well explained by key fundamentals. Foreign exchange reserves are positively correlated with economic size and current account vulnerability, and negatively correlated with exchange rate flexibility and the opportunity costs of holding reserves.

3. The estimated long run relationship between foreign reserves holdings and their determinants yields a demand for reserves that is lower than currently forecasted in Tunisia’s medium term macroeconomic framework. The model estimates that 2.5 months of reserves corresponding to 84 percent of short term debt coverage would be appropriate given balance of payments forecasts. Tunisia’s medium term scenario builds reserves equivalent to 3 months of imports and 105 percent of short term external debt coverage.

4. The planned gradual move to a floating exchange rate regime could further reduce the demand for reserves. However, increased exchange rate volatility in the transition to floating regime and possible new balance of payments risks emanating from capital account liberalization could require a reserve cushion to allow for some central bank intervention. As increased exchange rate flexibility fosters market liquidity, which helps reduce market volatility, the demand for larger reserves should diminish in view of the opportunity costs of holding reserves. Moderate reserve coverage would also avoid providing incentives for inappropriate debt management practices. Sound liability management by both the public and private sectors reduces the need for foreign exchange reserves.

B. Introduction

5. Foreign exchange reserves adequacy is a key component of good macroeconomic management. Foreign exchange reserves can be used to smooth random and temporary balance of payments shocks to maintain an exchange rate parity, avoid the macroeconomic costs of adjustment to temporary shocks, and smooth adjustment to the macroeconomic impact of some permanent shocks. Foreign reserves can also be used to smooth exchange rate volatility in illiquid foreign exchange markets. Foreign exchange reserves holdings, however, imply an opportunity cost usually defined as the difference between the highest possible return forgone from an alternative investment and the yield on foreign reserves. For these reasons, an analysis of the adequacy of reserves is important for sound macroeconomic management.

6. Econometric evidence in emerging market countries (IMF, 2003) shows that the demand for international reserves can be reasonably explained and forecasted using key fundamentals. In a panel setting, the econometric evidence showed that reserves holdings depend positively on measures of an economy’s size and external vulnerability, and negatively on exchange rate flexibility and opportunity costs of holding reserves.

7. The goal of this paper is to apply IMF (2003) to uncover the long run determinants of the demand for foreign exchange reserves in Tunisia, and to assess the adequacy of current and projected reserves holdings in light of the country’s policy choices. The Tunisian authorities have decided to gradually liberalize the capital account of the balance of payments to accompany the country’s increased integration (trade and financial) into the world economy. To minimize the risks of increased international integration2 and to maintain monetary policy independence in an open capital account environment, the Central Bank of Tunisia (BCT) is gradually moving from a real effective exchange rate targeting framework to a floating exchange rate regime. To this end, the BCT is implementing a new monetary framework that defines price stability as its primary objective. The framework adopts broad money as the reference intermediate target of monetary policy to achieve an inflation target and base money as the operating target. 3 While the planned move to a flexible exchange rate regime should reduce the need for holding large reserves, foreign exchange reserves could still be needed for precautionary reasons and to prevent excessive short term exchange rate volatility.

8. The paper is organized as follows. Section C describes recent trends in foreign exchange reserves in Tunisia. Section C also discusses measures of liquidity in the foreign exchange market since a more liquid foreign exchange market reduces the need for central bank intervention to smooth volatility. Section D presents econometric evidence on the determinants of the demand for foreign reserves in Tunisia. The results are used to forecast the desired level of reserves given Tunisia’s medium term macroeconomic framework and to draw policy implications.

C. Foreign Exchange Reserves and Markets in Tunisia

Trends in Foreign Exchange Reserves

9. Foreign exchange reserves in Tunisia increased over the last decade (Charts 1, 2, and 3). They rose from 1.6 months of imports of goods and services in 1990 to about 3 months in 2003 and represent US$ 3 billion. Over the same period, the ratio of reserves to short term external debt increased from 52 percent to about 100 percent.4 These ratios are somewhat lower than in many other emerging market countries where reserves accumulation in recent years appears to have been higher than justified by fundamentals (IMF, 2003). The short-term debt reserves coverage in Tunisia appears more than adequate judging by the 100 percent benchmark coverage, which empirical evidence (Bussière and Mulder, (1999) has determined is adequate to avert capital account crises in emerging market countries. Bussière and Mulder 1999) note, however, that the presence of a current account deficit or an overvalued exchange rate would require more reserves. In this regard, the 3 months of reserves coverage that has often been used as a benchmark cushion for current account related vulnerabilities remains a rule of thumb. The IMF (2003) empirical framework analyzing the determinants of the demand for foreign reserves would therefore be useful to shed further light on reserves adequacy in Tunisia.

Chart 1.
Chart 1.

Reserves in Months of Imports

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A001

Source: IMF (2003) and staff estimates.
Chart 2.
Chart 2.

Reserves to Short Term Debt

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A001

Source: IMF (2003) and staff estimates
Chart 3.
Chart 3.

Reserves to Broad Money

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A001

Source: IMF (2003) and staff estimates.

Foreign Exchange Market

10. Tunisia liberalized its foreign exchange market in 1994 when an interbank market was created. The BCT gradually reduced its market presence giving financial institutions more role in managing foreign exchange flows. Transactions among financial institutions, thus, represented 80 percent of dinar/foreign exchange transactions in 2003 and 100 percent of transactions amongst foreign currencies (Table 1). Forward and swap markets have also been created, although turnover in these markets is low relative to the spot market.

Table 1

Dinar/Foreign Currency Transactions

(In millions of US dollars)

article image
Source: Tunisian authorities.

In percent of total dinar/foreign currency transactions

11. Total turnover in the foreign exchange market has however stagnated in recent years. This could be due to several factors that inhibit market activity. A foreign exchange surrender requirement, which the BCT reduced from 50 percent to 30 percent in 2003, still exists. Banks are required to close their foreign exchange positions if a loss greater than 3 percent is incurred on the position even though they are allowed to hold positions up to 20 percent of their capital. Banks’ role in managing their foreign exchange holdings is limited by an obligation to deposit end-of-day foreign exchange balances at the central bank, which the BCT plans to eliminate. The BCT posts daily bid/ask spread dinar quotes, thereby reducing exchange rate uncertainty in the market. The market bid ask spread has remained relatively stable at 25 basis points, and daily exchange rate variations average below 0.5 percent reflecting limited market uncertainty (Charts 4 and 5). No noticeable improvement in market liquidity measured by market turnover and the ratio of exchange rate volatility to turnover (Charts 6 and 7) can be observed in recent years.5 The planned increase in exchange rate flexibility and elimination of BCT bid/ask quotes should help develop market activity provided banks are given more latitude to take open positions.

Chart 4:
Chart 4:

Bid/Ask Spread (basis points)

(Dinar/US Dollar Exchange Rate)

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A001

Chart 5:
Chart 5:

Absolute Value of Daily % Changes in Dinar/Dollar Exchange Rate

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A001

Chart 6:
Chart 6:

Turnover (value traded)

(in millions of dinars)

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A001

Source: Tunisian authorities.
Chart 7:
Chart 7:

Conventional Liquidity Ratio

(Daily Absolute Exchange Rate Changes over Value Traded)

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A001

12. In a context of increased exchange rate flexibility, the BCT might need to intervene in the foreign exchange market only to limit market volatility. Market volatility is currently low in part because of market practices discussed above. Volatility could increase in the future as financial institutions are given a larger role in determining the exchange rate. As this occurs, the current flexible real effective exchange rate targeting framework could be used to define an exchange rate band, which would guide short term intervention policy when the BCT ceases to post daily bid ask spreads of the dinar. The BCT would however need to gradually move away from this policy to avoid market perception of an exchange rate guarantee (Duttagupta et al, 2004). Ultimately, foreign exchange intervention may need to be limited to announced annual foreign exchange reserves accumulation targets that the long run determinants of the demand for foreign exchange reserves discussed below would suggest.

D. Econometric Evidence on the Determinants of Foreign Exchange Reserves

13. Following on IMF (2003), we estimated bivariate regressions between real reserves and its determinants (Table 2). IMF (2003) measured economic size by population and real GDP per capita; current account vulnerability by openness to trade and exports volatility6; capital account vulnerability by financial openness and potential for capital flight by residents; exchange rate flexibility by the actual volatility of the exchange rate; and the opportunity cost of holding reserves by the difference between domestic and foreign interest rates. The bivariate regressions indicate that foreign reserves and most of its identified determinants have the expected correlation signs. Foreign reserves vary positively with measures of economic size and current and capital account vulnerability, and negatively with exchange rate flexibility and the opportunity cost of holding reserves. In the multi-variable regression, best results are obtained when economic size is captured by the logarithm of real GDP per capita and current account vulnerability by openness to trade (imports plus exports relative to GDP) and the current account deficit relative to GDP.

Table 2.

Determinants of Foreign Exchange Reserves (OLS)

article image
Source: Tunisian authorities.

14. The data series, however, have unit roots implying that a regression of foreign reserves on its determinants could be spurious unless the variables are cointegrated. The best-fit multiple regression (Table 2, Equation 1) has good residuals’ properties and the OLS Engle Granger procedure for cointegration indicates that its residuals are stationary (Table 3). This suggests that the variables are cointegrated and the estimated equation is a long run relationship. The Johansen VAR procedure (Table 4) also indicates that the variables are cointegrated confirming that equation (1) is not spurious. Both λmax and λtrace statistics, however, suggest that there are two cointegrating vectors. 7 The existence of another cointegrating relationship, and the potential misspecification of foreign reserves explanatory variables’ equations in the VAR setting may be a reason why some of the variables do not have expected signs in the long-run foreign reserves equation estimated by the Johansen procedure. In fact, real GDP per capita and the current account balance to GDP are not significant in the VAR estimated long run relationship.

Table 3.

Augmented Dickey Fuller Tests for Unit Roots

article image
Rrmg = real reserves minus gold; lrgdpc=log of real GDP per capita; opent=ratio imports and exports to GDP; volexr = standard deviation of monthly dinar/dollar exchange rate; ratedif=nominal short term interest rate differential between the United States and Tunisia; cagdp=current account deficit to GDP Resid = residuals of Equation 1 in Table 2** and * mean significance at the 1 percent and 5 percent levels respectively.Samples are from 1987 to 2002
Table 4

Cointegration Analysis

(Johansen methodology)

article image

** and * denote significance at the 1 percent and 5 percent levels Sample is 1983-2001. Variable are as defined in Table 3

15. Equation 1 is, therefore, our preferred long run relationship. Coefficients in equation 1 have the expected signs and produce good in sample forecasts of the level of reserves (Chart 8).

(Equation1)172.21+24.086(5.77)*LRGDPC(5.34)1.535*VOLEXR(1.78)+18.097*OPENT(2.46)0.347*RATEDIF(1.99)+34.59*CAGDP(2.00)
Chart 8:
Chart 8:

Fitted vs Actual and Projected Reserves

(in US dollars)

Citation: IMF Staff Country Reports 2004, 360; 10.5089/9781451837841.002.A001

Source: Staff estimates.

16. The projected reserves in Tunisia’s current macroeconomic framework are somewhat higher than forecasted using the estimated long run equation 1. The projected foreign reserves build up would hold reserves at 3 months of imports and 105 percent of short term debt in the medium term. Equation 1 suggests that 2.5 months of reserves coverage (closer to the 10 year historical average of 2.7 months), and 85 percent of short term debt coverage may be appropriate. The fitted values are based on the average exchange rate volatility of the last 3 years and an interest rate differential of about 200 basis points. As exchange rate flexibility increases, the need for holding reserves should further diminish. Increased exchange rate volatility in a more flexible exchange rate environment and possible new balance of payments risks emanating from capital account liberalization could, however, require a reserve cushion to allow for central bank intervention. As increased exchange rate flexibility fosters market liquidity, which reduces market volatility, the demand for larger reserves should diminish in view of the opportunity costs of holding reserves. Moderate coverage of short term external liabilities may also encourage better private and public sectors debt management. As noted in IMF, 2004, sound liability management by both the public and private sectors should play a major role in containing exposures and rollover risks, and help reduce the need for foreign exchange reserves.

References

  • 2003, Aizenman, J., and Marion, N.The High Demand for International Reserves in the Far East: What is going on?Journal of International Economics, 17, 370400.

    • Search Google Scholar
    • Export Citation
  • 1999, Bussière, M. and Mulder, C.External Vulnerability in Emerging Market Economies: How High Liquidity can Offset Weak Fundamentals and the effects of Contagion”.

    • Search Google Scholar
    • Export Citation
  • 1975, Claassen, E-M.Demand for International Reserves and the Optimum Mix and Speed of Adjustment Policies”, American Economic Review Vol.65. No 3. 44653

    • Search Google Scholar
    • Export Citation
  • 2004, Duttagupta, R., Fernandez G., Karacadag, C.From Fixed to Float: Operational Aspects of Moving Toward Exchange Rate Flexibility

    • Search Google Scholar
    • Export Citation
  • 2001, Flood, R. and Marion, N.Holding International Reserves in an Era of High Capital Mobility”, Brooking Trade Forum.

  • 2003, International Monetary FundAre Foreign Reserves in Asia Too High”. Chapter II by Edison, H. in World Economic Outlook, September.

    • Search Google Scholar
    • Export Citation
  • 2002, Laurens, B. and Sarr, A.Liberalization of the Capital Account in Tunisia—Progress Achieved and Prospects for Full Convertibility”. IMF Country Report 02/120, Chapter II.

    • Search Google Scholar
    • Export Citation
  • 2003, Laurens, B. and Sarr, A.Tunisia—Strengthening the Monetary Framework and InstrumentsIMF Country Report 03/246, Supplement.

    • Search Google Scholar
    • Export Citation
  • 2002, Lybek, T. and Sarr, A..Measuring Liquidity in Financial Markets”. International Monetary Fund Working Paper, WP/02/232.

1

Prepared by Abdourahmane Sarr.

2

The potential risks are the loss of external competitiveness as trade integration increases and the provision of an exchange rate guarantee to investors as financial integration intensifies.

3

See Laurens and Sarr (IMF Country Report 03/259, 8/21/03; and IMF Country Report 02/120, 6.17/02) for a description of the monetary policy framework and planned sequencing of capital account liberalization). These papers also discuss the structural policies that support a flexible exchange rate regime, namely deep foreign exchange and money markets, and a healthy and well supervised financial sector.

4

Short term debt includes current amortization and non-residents’ deposits and excludes suppliers’ credit. Deposits of residents in foreign currency and local currency convertible into foreign currency are negligible (less than 0.3 percent of M3), and there is no foreign currency public debt to residents. There is, therefore, no need for augmented reserve coverage indicators that would include these liabilities.

5

See Sarr and Lybek (2002) for a discussion of liquidity measures.

6

We replaced exports volatility by the current account deficit to GDP as a measure of current account vulnerability in our single country empirical context and annual exports data.

7

Weak exogeneity was rejected, implying that a test of cointegration using a single equation error correction model of foreign reserves would not be appropriate.

Tunisia: Selected Issues
Author: International Monetary Fund