A. Country Characteristics

The Gambia is the smallest country in continental Africa in terms of area. It has a population of about 1.6 million, which has been growing at an annual rate of 2.3 percent. With a per capita gross national income of $320 in 2007, it is one of the poorest countries in the world⁵—with a score of 0.502, The Gambia ranks 155 out of 177 countries (and 94 out of 108 developing countries) in the 2007/2008 UN Human Development Index (see United Nations Development Programme, 2007). Nearly 60 percent of the population live in poverty (see Republic of The Gambia, 2006). About 80 percent of the population depends on agriculture, which is dominated by groundnuts. However, because its productivity is low, the agricultural sector contributes only about one-fifth of real GDP. The service sector, which accounts for over 50 percent of real GDP, has been the major contributor to economic growth.

B. Macroeconomic Performance

The performance of The Gambia’s economy has been uneven. In 1988–2007, real GDP growth averaged 4.5 percent and inflation 6.6 percent a year, although both varied significantly from year to year (see Figures 1 and 2). Exogenous, political, and policy-induced shocks are the main reasons for uneven performance. Since 1984 The Gambia has been in seven different IMF programs, the most recent signed in February 2007 for a three-year Poverty Reduction and Growth Facility arrangement of SDR 14 million.⁶

View Full Size

The Gambia: Real GDP Growth, 1988-2007

(In percent per year)

Citation: IMF Working Papers 2009, 192; 10.5089/9781451873399.001.A001

Source: IMF staff estimates

• Download Figure
• Download figure as PowerPoint slide

View Full Size

The Gambia: Average Inflation, 1988-2007

(In percent per year)

Citation: IMF Working Papers 2009, 192; 10.5089/9781451873399.001.A001

Source: Computed from the consumer price index data from the authorities.

• Download Figure
• Download figure as PowerPoint slide

The Gambia experienced a severe economic crisis in the mid-1980s primarily due to the cumulative effects of two decades of unsustainable macroeconomic policies, an overvalued exchange rate, price controls, and overexpansion of the public sector, exacerbated by droughts and terms of trade deterioration. To restore stability the authorities implemented a comprehensive adjustment program—the Economic Recovery Program (ERP)—in June 1985; it was followed by the Program for Sustained Development in 1990. Through 1993 these reforms led to significant improvements in economic performance despite continued poor rainfall and low world groundnut prices (see World Bank, 2003). However, between 1993 and 1996 the economy was set back by a series of adverse shocks: reinforced border and transit controls in Senegal, the suspension of convertibility of CFA francs outside the CFA zone, the 50 percent devaluation of the CFA franc in 1994, and a fall-off in foreign aid and tourism after the 1994 military coup.

The country enjoyed a period of growth in 1998–2001 when political stability was established after elections in 1997 and 2001, and the authorities renewed their commitment to improving economic policies. Though very bad weather led to a 3 percent decline of real GDP in 2002, growth in real GDP rebounded in 2003; but fiscal slippages, an accommodating monetary policy, and international reserves that had been falling since 2001 fueled an abrupt depreciation of the dalasi (55 percent in nominal effective terms) and a surge in inflation (from less than 1 percent in 2000 to 21 percent in August 2003). In response, in 2004 the authorities began to tighten their monetary and fiscal policies. As a result, inflation fell to just 0.4 percent year-on-year in December 2006. Output grew at an annual average of 6.4 percent for 2003–06. In December 2007, The Gambia became the 23rd country to reach the completion point under the Enhanced Heavily Indebted Poor Countries (HIPC) Initiative.⁷ Although it is expected to receive the equivalent of $514 million in nominal debt relief from the HIPC and Multilateral Debt Relief initiatives, the country is still at high risk of debt distress.

The Gambia is an open economy. Since January 1986, it has adhered to a managed floating exchange rate regime. Until that point the dalasi was pegged to the pound sterling at a rate of D5 = £1. On January 20, 1986, an interbank market for foreign exchange was introduced; since then the exchange rate has been determined by market forces. Effective June 30, 2002, the IMF reclassified The Gambia’s exchange rate arrangement from independently floating to a managed float with no predetermined path. Its exchange system is free of restrictions on the making of payments and transfers for current international transactions. After relative stability in 1988–99, the exchange rate depreciated rapidly in 2000–03 due to lax fiscal and monetary policies; however, since 2004 it has been appreciating in response to prudent macroeconomic policies. Figure 3 presents the exchange rate movements in The Gambia during 1988–2007.

View Full Size

The Gambia: Exchange Rate Developments, 1988-2007

(Dalasi/U.S. dollar)

Citation: IMF Working Papers 2009, 192; 10.5089/9781451873399.001.A001

Source: IMF, International Financial Statistics.

• Download Figure
• Download figure as PowerPoint slide

C. Financial Sector Development⁸

The Gambian financial system is small. As of 2006, apart from the CBG the financial system consists of 8 commercial banks (including one Islamic bank),⁹ the post office savings bank, 11 insurance companies, 68 savings and credit associations, 38 foreign exchange bureaus, and several money transfer agents. The commercial banks overwhelmingly dominate the financial sector, accounting for 97 percent of total sector assets.

Commercial banking is highly concentrated: the largest two banks (Standard Chartered Bank and Trust Bank Limited) collect 71 percent of deposits, extend 58 percent of loans, and hold 66 percent of assets. Nearly 15 percent of bank deposits are denominated in foreign currency (see IMF, 2007b, Table 29). All the commercial banks are privately owned. Foreign participation is high: 100 percent for two banks and between 70 percent and 90 percent for three others.

Despite the dominance of the banking system, financial intermediation is low; the loans-to-deposits ratio is only about 40 percent. This is due not to any interest rate controls (interest rates have been deregulated since 1985) but to a variety of other factors—high lending rates, the attractiveness of government securities to banks because of low risk and high yields, and a weak legal environment (in particular, lack of an effective commercial court to enforce contracts). These factors also partly explain why real lending rates and interest spreads at commercial banks have been high (see Figure 4).

View Full Size

The Gambia: Real Lending Rates and Interest Rate Spreads, 1988-2007

(Percent per year)

Citation: IMF Working Papers 2009, 192; 10.5089/9781451873399.001.A001

Source: Gambian authorities.

• Download Figure
• Download figure as PowerPoint slide

The financial sector has been gradually deepening. The ratio of M2/GDP has more than doubled, from 25 percent in 1986 to 52 percent in 2007 (see Figure 5). Growth has been particularly steep since 2000 primarily due to strong capital inflows; a crack-down on informal exchange markets that were diverting funds from formal banking channels; financial innovation (e.g., allowing residents to have foreign currency deposits); and increasing banking sector activity (in terms of both numbers of banks and branches). Among 44 sub-Saharan African countries, The Gambia ranked sixth in terms of M2/GDP in 2007.¹⁰

View Full Size

The Gambia: Measures of Financial Deepening, 1988-2007

(Percent)

Citation: IMF Working Papers 2009, 192; 10.5089/9781451873399.001.A001

Source: International Monetary Fund, International Financial Statistics.

• Download Figure
• Download figure as PowerPoint slide

The Gambia also has a government debt securities market. The CBG first began issuing treasury bills in 1986 for liquidity management purposes. However, since the mid-1990s, the treasury bills have also been issued to meet the financing requirements of the government. Nearly 50 percent of outstanding treasury bills are held by the commercial banks, although in the past public corporations held most of them.¹¹ Individuals are also allowed to buy them. Treasury bills are reasonably liquid; they can be rediscounted by the primary dealers or the CBG.¹²

Banks depend on the CBG for rediscounting treasury bills to meet their urgent liquidity needs. Until 2004 the CBG rediscount rate was a 3 percent premium over the yield on treasury bills. However, since its inception in July 2004, the Monetary Policy Committee (MPC) has been announcing the rediscount rate (independent of current yields on the treasury bills) after its bimonthly meetings to signal to market participants the direction of monetary policy. Many banks use the rate to set their prime lending rates.

Private debt securities are not common. There is no stock market, but there is a small and growing foreign exchange market in which commercial banks and foreign exchange bureaus are the main players. Until 2003 there was an active parallel market but transactions in it are now few. Interbank activity on overnight borrowing and foreign exchange transactions is negligible, although it has been growing slowly in recent years.

Financial innovation is taking place, but slowly. In 2002 banks began to accept foreign-currency deposits. Since 2007 they have also been providing automated teller services. There has been no restriction on foreigners buying government debt. In fact The Gambia opened up its financial sector to investors in the late 1980s as part of its ERP to promote development of a more competitive banking system and more efficient financial markets. There was some foreigner interest in government securities in 2003–04 when yields reached 31 percent, but outside interest waned as yields fell to 10–15 percent.

A. General Framework

There are many theories about demand for money; they tend to emphasize transactions, speculative, precautionary, or utility motives that money satisfies. Though implicitly they address a broad range of hypotheses, interestingly, almost all of them share common elements (variables). In general, they elicit a relationship between the quantity of money demanded and a few important economic variables linking money to the real sector of the economy (see Judd and Scadding, 1982, p. 993). What separates these theories is that although they consider similar variables to explain the demand for money, they often differ in the role they assign to each. Consequently, the one clear fact that emerges from the literature is that the empirical work is motivated by a blend of theories.

The empirical work generally begins with the following specification of the functional relationship for long-term demand for money:

where the demand for real balances M/P is a function of the chosen scale variable (S), representing economic activity, and the opportunity cost of holding money (OC). M stands for the monetary aggregate in nominal terms and P for the price.

The opportunity cost of holding money consists of both its own rate and alternative returns on money. The own rate of money represents the return money yields when it is held in the form of current deposits used mainly for transactions. Failure to include the own rate of money often leads to breakdown of the estimated money demand function, especially during periods of financial innovation (see Ericsson, 1998). Alternative returns represent yields on instruments that are close substitutes for money that are foregone by holding money. They include returns on domestic financial and real assets; in an open economy, they may also include yields on foreign assets. Based on the currency substitution literature, yields on foreign assets are generally represented by foreign interest rates and expectations for depreciation of the domestic currency.¹⁵

The choices for domestic alternatives to money (e.g., government securities, commercial paper, corporate bonds, certificates of deposits, and return on equities) will vary from country to country depending on the nature and depth of the financial markets. The expected rate of inflation is typically used to represent the yield on real assets. Where there is hyper-inflation, where interest rates are controlled by the government, and where the financial sector is not well developed, only the expected rate of inflation is used because it better represents the alternative return on money. However, in countries that are experiencing moderate inflation where alternative financial assets are available, both expected rate of inflation and the yield on one or more alternative financial assets are included in the money demand function. The rationale is that yields on alternative financial assets may not fully incorporate the expected inflation rate.

Money demand functions are usually estimated in log linear form with monetary aggregates and scale variables entering in logarithms, but other variables are specified either in logarithms or in levels.¹⁶ Coefficients of variables specified in logarithms directly indicate elasticities; those in levels, semi-elasticities.

B. Model Specification for Long-Run Demand for Real M2 in The Gambia

A. Unit Root Tests

The augmented Dickey-Fuller (ADF) test was applied to examine whether the variables to be used in the estimation have unit roots. The test is conducted by including both constant variables and a trend variable. Seasonal dummies are also used because the time series are not seasonally adjusted. The test began with 12 lagged difference terms in order to whiten the residuals; but the Akaike information criterion (AIC) is finally applied to choose the maximum number of lags. The results indicate that the variables LRBM, LRGDPA, DDR, TBRA, CPI2000, INFA, DEPR, XRD$, NEER, LIBOR3M, and USTBR3M are nonstationary. All variables are integrated to the order 1, that is I(1), confirmed by the fact that the first differences of all these variables are I(0). Table 1 presents the result of unit root tests. Appendix IV shows the graphs of the variables used.

Table 1.

Unit Root Test Results

Note: ** indicates the rejection of unit roots at 1 percent significance level.

Table 1.

Unit Root Test Results

	Test Statistic	Lag		Test Statistic	Lag
LRBM	-1.439	1	dLRBM	-16.540	0 **
LRGDPA	-2.053	1	dLRGDPA	-14.450	0 **
DDR	-2.155	0	dDDR	-14.840	0 **
TBRA	-3.314	10	dTBRA	-9.685	0 **
CPI2000	-0.420	2	dCPI2000	-12.570	0 **
INFA	-2.675	0	dINFA	-13.560	0 **
XRD$	-2.140	8	dXRD$	-11.470	0 **
DEPR	-12.110	0 **	dDEPR	-23.740	0 **
NEER	-1.419	4	dNEER	-10.980	0 **
LIBOR3M	-2.983	7	dLIBOR3M	-10.040	0 **
USTBR3M	-3.384	8	dUSTBR3M	-9.562	0 **

Note: ** indicates the rejection of unit roots at 1 percent significance level.

View Table

Table 1.

Unit Root Test Results

	Test Statistic	Lag		Test Statistic	Lag
LRBM	-1.439	1	dLRBM	-16.540	0 **
LRGDPA	-2.053	1	dLRGDPA	-14.450	0 **
DDR	-2.155	0	dDDR	-14.840	0 **
TBRA	-3.314	10	dTBRA	-9.685	0 **
CPI2000	-0.420	2	dCPI2000	-12.570	0 **
INFA	-2.675	0	dINFA	-13.560	0 **
XRD$	-2.140	8	dXRD$	-11.470	0 **
DEPR	-12.110	0 **	dDEPR	-23.740	0 **
NEER	-1.419	4	dNEER	-10.980	0 **
LIBOR3M	-2.983	7	dLIBOR3M	-10.040	0 **
USTBR3M	-3.384	8	dUSTBR3M	-9.562	0 **

Note: ** indicates the rejection of unit roots at 1 percent significance level.

B. Cointegration Tests

Since all the variables are I(1), the Johansen-Juselius multivariate cointegration technique (see Johansen, 1988, and Johansen and Juselius, 1990) is applied to examine the existence of a cointegration relationship between LRBM and its determinants as shown in equation 3 above (see Table 2). The model included a trend, constant, seasonal dummies, 12 lags, and a impulse dummy of D200301 to account for many financial innovation measures introduced in January 2003 (see Appendix I). The AIC is applied to choose the maximum lag length.

Table 2.

Cointegration and Weak-Exogeneity Tests

Note: The system includes 2 lags for each variable, a constant, trend, seasonal dummies, and the dummy variable D2003m1. The estimation period is 1988:1-2007:6 (234 observations).

^1/r stands for the number of ranks.

^2/** indicates the significance at 1 percent level.

^3/** and * indicate that the null hypothesis of weak-exogeneity is rejected at 1 percent and 5 percent significance level, respectively. The probability of getting any number exceeding the χ² value shown above is less than the figure presented within the squared brackets.

Table 2.

Cointegration and Weak-Exogeneity Tests

Cointegration test
Eigenvalues						0.1760	0.1186	0.0621	0.0272	0.0226
Null hypotheses 1/						r = 0	r <= 1	r <= 2	r <= 3	r <= 4
λ trace 2/						101.6500 **	56.3600	26.8100	11.8100	5.3500
	Adjusted for degrees of freedom					97.3000 **	53.9500	25.6700	11.3000	5.1200
	P-values					0.0090	0.2580	0.7560	0.8530	0.5870
λ max 2/						45.2800 **	29.5500	15.0000	6.4600	5.3500
	Adjusted for degrees of freedom					43.3500 **	28.2900	14.3600	6.1800	5.1200
	P-values					0.0090	0.1390	0.6980	0.9360	0.5880
						Standardized Eigenvectors β’
						LRBM	LRGDPA	DDR	TBRA	INFA	Trend
						1.000	-1.264	-0.402	0.223	-0.227	-0.022
						-0.598	1.000	-0.071	0.008	0.010	-0.001
						-2.537	59.665	1.000	0.161	-0.006	-0.165
						-202.620	890.550	2.975	1.000	-1.010	-0.912
						-4.380	-540.050	-8.227	5.938	1.000	1.930
					Standardized Adjustment Coefficients α
						LRBM	LRGDPA	DDR	TBRA	INFA
					ΔLRBM	-0.0066	0.0219	-0.0001	-0.0644	0.0298
					ΔLRGDPA	0.0008	-0.0099	-0.0009	-0.0201	0.0142
					ΔDDR	0.0134	1.4580	-0.0290	0.0002	-0.0004
					ΔTBRA	-0.2828	0.1355	-0.0415	-0.0010	-0.0009
					ΔINFA	0.2167	-6.2224	-0.1075	0.0008	-0.0040
Weak-exogeneity test 3/
			Variables			ΔLRBM	ΔLRGDPA	ΔDDR	ΔTBRA	ΔINFA
				α1 = 0		χ2(1) = 6.4371 [0.0112]*
				α2 = 0		χ2(1) = 0.97973 [0.3223]
				α3 = 0		χ2(1) = 0.10742 [0.7431]
				α4 = 0		χ2(1) = 15.659 [0.0001]**
				α5 = 0		χ2(1) = 1.2776 [0.2583]
		α2 = 0; α3 = 0; α5 = 0				χ2(3) = 2.4117 [0.4915]

Note: The system includes 2 lags for each variable, a constant, trend, seasonal dummies, and the dummy variable D2003m1. The estimation period is 1988:1-2007:6 (234 observations).

^1/r stands for the number of ranks.

^2/** indicates the significance at 1 percent level.

^3/** and * indicate that the null hypothesis of weak-exogeneity is rejected at 1 percent and 5 percent significance level, respectively. The probability of getting any number exceeding the χ² value shown above is less than the figure presented within the squared brackets.

View Table

Table 2.

Cointegration and Weak-Exogeneity Tests

Cointegration test
Eigenvalues						0.1760	0.1186	0.0621	0.0272	0.0226
Null hypotheses 1/						r = 0	r <= 1	r <= 2	r <= 3	r <= 4
λ trace 2/						101.6500 **	56.3600	26.8100	11.8100	5.3500
	Adjusted for degrees of freedom					97.3000 **	53.9500	25.6700	11.3000	5.1200
	P-values					0.0090	0.2580	0.7560	0.8530	0.5870
λ max 2/						45.2800 **	29.5500	15.0000	6.4600	5.3500
	Adjusted for degrees of freedom					43.3500 **	28.2900	14.3600	6.1800	5.1200
	P-values					0.0090	0.1390	0.6980	0.9360	0.5880
						Standardized Eigenvectors β’
						LRBM	LRGDPA	DDR	TBRA	INFA	Trend
						1.000	-1.264	-0.402	0.223	-0.227	-0.022
						-0.598	1.000	-0.071	0.008	0.010	-0.001
						-2.537	59.665	1.000	0.161	-0.006	-0.165
						-202.620	890.550	2.975	1.000	-1.010	-0.912
						-4.380	-540.050	-8.227	5.938	1.000	1.930
					Standardized Adjustment Coefficients α
						LRBM	LRGDPA	DDR	TBRA	INFA
					ΔLRBM	-0.0066	0.0219	-0.0001	-0.0644	0.0298
					ΔLRGDPA	0.0008	-0.0099	-0.0009	-0.0201	0.0142
					ΔDDR	0.0134	1.4580	-0.0290	0.0002	-0.0004
					ΔTBRA	-0.2828	0.1355	-0.0415	-0.0010	-0.0009
					ΔINFA	0.2167	-6.2224	-0.1075	0.0008	-0.0040
Weak-exogeneity test 3/
			Variables			ΔLRBM	ΔLRGDPA	ΔDDR	ΔTBRA	ΔINFA
				α1 = 0		χ2(1) = 6.4371 [0.0112]*
				α2 = 0		χ2(1) = 0.97973 [0.3223]
				α3 = 0		χ2(1) = 0.10742 [0.7431]
				α4 = 0		χ2(1) = 15.659 [0.0001]**
				α5 = 0		χ2(1) = 1.2776 [0.2583]
		α2 = 0; α3 = 0; α5 = 0				χ2(3) = 2.4117 [0.4915]

Note: The system includes 2 lags for each variable, a constant, trend, seasonal dummies, and the dummy variable D2003m1. The estimation period is 1988:1-2007:6 (234 observations).

^1/r stands for the number of ranks.

^2/** indicates the significance at 1 percent level.

^3/** and * indicate that the null hypothesis of weak-exogeneity is rejected at 1 percent and 5 percent significance level, respectively. The probability of getting any number exceeding the χ² value shown above is less than the figure presented within the squared brackets.

Of various formulations tested, only the one involving variables real M2, real GDP, interest rate on short-term deposits, yield on treasury bills, and expected inflation produced a cointegration relationship.¹⁹

Both trace and maximal eigenvalue tests rejected zero in favor of at least one cointegration vector. ²⁰ The results are significant at the 1 percent level, even when the critical values are adjusted for degrees of freedom, as suggested by Osterwald-Lenum (1992). Based on the signs and magnitudes of the estimated coefficients, the unique cointegration vector can be interpreted as the long-run demand for real M2. The standard errors presented within parentheses indicate that all coefficients except LRGDPA are significant.

Long-run income elasticity is 1.26, which is slightly higher than the quantity theory suggests for broad money. Long-run elasticity typically exceeds one for many countries where the financial systems are underdeveloped and monetization is faster than output growth.²¹ The velocity in these cases shows a declining trend. In the case of The Gambia, velocity has been declining, from about 400 percent through the mid-1990s to about 300 percent in the late 1990s to about 200 percent since 2003.²² Model misspecification, especially omitting important opportunity cost variables, also leads to high income elasticity. The omitted opportunity cost variables could be highly correlated with money; and in such situations, the scale variable usually picks up the effects of these omitted variables.

Coefficients of both the own rate of money (DDR) and the alternative return (TBRA) carry expected signs, indicating that as the interest rate on demand deposits increases, so does demand for real M2; and as yields on treasury bills increase, demand for real M2 decreases.

Expected inflation turns out to be positively related to the demand for real M2. There may be two reasons for this:

• First, the financial sector is shallow and dominated by the banking sector (although there were only 41 bank branches plus nearly 70 savings and credit association in the entire country as of end-2007), and there are not many alternatives for money. For instance, equity market capitalization and the outstanding value of debt securities amount to more than 200 percent of GDP in industrial countries (and more than 300 percent in Japan and the United States) and about 126 percent for emerging markets as a group in 2006 (see IMF, 2007c); but, in The Gambia, the only alternative for money is treasury bills and the amount outstanding was valued at only 27 percent of GDP in 2007. Inflationary expectations also diverge among the population (depending on whether urban or rural; rich or poor; and the goods they buy); consequently for transactions and precautionary reasons agents may demand more money in real terms (in terms of percent increase) than the expected rise in inflation.²³

• Second, data deficiency in CPI may also contribute for expected inflation not having a negative relationship with demand for money. The old CPI, which had 1974 weights and measured prices in the low-income area of Greater Banjul only, was not revised until 2007.

Interestingly, the foreign-influence variables seem not to be important in explaining demand for real M2 for The Gambia. When foreign interest rates (three-month LIBOR or yields on three-month U.S. treasury bills) or expected depreciation variables (depreciation of the dalasi against the U.S. dollar or measured using nominal effective exchange rates) were introduced, either the model failed to produce a cointegration relationship or produced multiple cointegrating vectors with coefficients having implausible values or carrying the wrong signs. Considering how open The Gambia’s economy is, inflation may already reflect the effect of depreciation. Also, since this is a poor country, only a handful of domestic players will be able to exploit the arbitrage opportunities of investing in foreign securities.

The diagnostic tests (see Appendices V and VI) indicate that except for DDR and TBRA, the equations are free of error autocorrelation and heteroscedasticity. However, the normality tests are rejected for all the variables because of a handful of large outliers for each. The one-step residuals shown in Appendix VII confirm that in most cases these residuals are within ±2 standard errors around 0 for a given equation. Some of these outliers are associated with (1) shocks affecting output (LRGDPA) during 1995 and 2002; (2) drastic changes in the own rate of money (DDR) at some points in 1988, 1992, 1999, 2003, and 2005; and (3) a dramatic increase in yields on TBRA in 2002–04 after the authorities decided to quell inflationary pressures with a tight monetary policy. The chow tests shown in Appendix VIII magnify these outliers and show breaks in 1995, 2002, and 2005.

The cointegration results need to be interpreted with caution. Attempts to account for the outliers by introducing dummy variables changed the coefficients of variables significantly or yielded no cointegration relationship. An attempt to redo the cointegration tests by introducing monthly data on tourist arrivals as a proxy for real GDP did not yield a cointegration relationship either.²⁴ Before drawing a firm conclusion about the stability of the model, we analyzed the behavior of the alpha coefficients.

Adjustments to restore long-run equilibrium in response to disequilibrium are indicated by the alpha coefficients. The first term in the alpha vector represents the direction and speed at which DLRBM responds to deviation from the equilibrium relationship. Similarly, the second term shows how fast and in what direction DLRGDPA adjusts to bring LRBM back to equilibrium. The alpha coefficient for LRBM is –0.00659; it represents the estimated feedback coefficient for the money equation. The negative coefficient implies that lagged excess holding of money induces a smaller amount during the current period. The negative sign for the coefficient is important because it indicates that the model is valid; any disruptions in the long-run relationship will be reduced in future periods.

The numerical value of –0.00659 for the alpha coefficient for the LRBM equation implies that adjustment toward equilibrium takes place by about 0.7 percent in the first month after disequilibrium. The adjustment is apparently slow in comparison to other countries (see Ericsson and Sharma, 1996) because there are so few alternatives for money in The Gambia. The alpha coefficients also show that most of the adjustments take place in the equations for LRBM, DDR, TBRA, and INFA. The standard errors indicate that the coefficients for LRBM and TBRA are significant.

The weak exogeneity tests also indicate that the adjustments are primarily carried through LRBM and TBRA. Other variables—LRGDPA, DDR, INFA, and the trend—are weakly exogenous; therefore, changes in money are not affecting these variables. It intuitively makes sense why TBRA is not weakly exogenous. As part of meeting the operational targets on reserve money, the authorities adjust the quantity of treasury bills supplied to the market. They also in a way signal to the market what will be the upper bound on yields from auctions by announcing the rediscount rate after each MPC meeting is concluded. Data show a correlation coefficient of 0.97 between yields on treasury bills and rediscount rate.

Although weak exogeneity was not rejected for LRGDPA, its coefficient moved around significantly while doing the similar tests for other variables. We also found earlier that the cointegration tests yielded insignificant coefficient for LRGDPA. Moreover, the parameter constancy tests indicate structural breaks and any attempts to correct for these breaks yielded no good alternative cointegration formulation. All these observations point to a conclusion that the long-run relationship for LRBM is not stable in The Gambia. Consequently, a short-run model is not carried out.

Although the study did not find a stable long-run relationship for demand for broad money in real terms, the analysis provides some valuable information on the effect of interest rates and inflation on demand for money. First, by the size and sign of the coefficients, interest rate on short-term deposits and yield on treasury bills give an indication to what extent demand for money can be influenced by changes in interest rates. Second, inflation appears to incorporate exchange rate developments. Therefore, monetary aggregates and explanatory variables have information content on the status of the economy. As these variables are important for understanding the transmission mechanism of monetary policy, improvements in data quality will aid in better conduct of monetary policy in The Gambia.²⁵