Ghana: Selected Issues and Statistical Annex

This Selected Issues paper and Statistical Annex examines the impact of cocoa taxation on cocoa supply in Ghana. The paper describes historical developments in cocoa production. The effects of the taxation of cocoa in Ghana are evaluated and a dynamic model of cocoa supply is estimated and used for simulations. The paper concludes that the most important factors adversely affecting the cocoa sector were government policies. Specifically, in the late 1960s and the 1970s, the effective cocoa duty rates were punitive and the cocoa sector was further hit by policies of overvalued exchange rate.

Abstract

This Selected Issues paper and Statistical Annex examines the impact of cocoa taxation on cocoa supply in Ghana. The paper describes historical developments in cocoa production. The effects of the taxation of cocoa in Ghana are evaluated and a dynamic model of cocoa supply is estimated and used for simulations. The paper concludes that the most important factors adversely affecting the cocoa sector were government policies. Specifically, in the late 1960s and the 1970s, the effective cocoa duty rates were punitive and the cocoa sector was further hit by policies of overvalued exchange rate.

III. The Transmission Mechanism Between Money and Inflation in Ghana 1/

A. Introduction

Inflation has been persistently high in Ghana for a number of years, and its direct impact on Ghana’s long-term growth has been reasonably well documented in the empirical literature. 2/ These studies found that in the case of Ghana, high inflation rates affected capital accumulation and total factor productivity and hence long-term growth potential. 3/

The objective of disinflation remains high on the government’s economic policy agenda, and the government has identified monetary policy as the most appropriate instrument in achieving this objective. 4/ However, the successful implementation of a disinflation policy based on monetary targeting depends crucially on one major assumption. Inflation should be linked to a controllable monetary target, the demand for which is empirically related to a well-defined set of economic variables in a fairly predictable way. This paper therefore looks at the determinants of inflation in Ghana and examines the extent to which it is a monetary phenomenon.

1. Inflation and monetary growth - recent developments

After Ghana launched its Economic Recovery Program (ERP) in 1983, the annual inflation rate fell from 123 percent in 1983 to 10 percent by 1991. Since 1992, however, this downward trend has reversed and inflation has risen sharply; at the end of 1995, the annual inflation rate was 71 percent (Chart III.1). Table III.1 shows actual and target rates of inflation and money growth in Ghana over the period 1990-95; it indicates that the authorities did not generally meet the prescribed targets. 5/

Table III.1.

Ghana: Actual and Programmed Inflation and Money Growth over the Period 1990-95

(In percent)

article image
Source: IMF staff estimates.

In percentage points.

2. Review of empirical models of inflation in Ghana

A number of authors have attempted to study empirically the factors that explain inflation in Ghana. The existing evidence seems to be mixed between monetary and supply-side explanations. Chhibber and Shafik (1990) conclude that inflation in Ghana is largely money driven, but Sowa and Kwakeye (1991) argue that inflation in Ghana is influenced largely by real factors, particularly food production. Kapur et al. (1991) 1/ specified a Walrasian general equilibrium model, which combines monetarist, cost-push, and structural explanations of inflation. Based on this theoretical model, they estimated a reduced-form equation of inflation in Ghana over the period 1984-90. The inflation regressions they estimated yielded high explanatory power and suggest that inflation in Ghana is primarily a function of domestic supply shocks, and domestic (petroleum price) and external (import cost) cost-push factors. More recently, Johnson et al. (1994) estimated models of inflation using various monetary aggregates based on monthly data from January 1990 to December 1993 and concluded that base money has the highest information content for the predictability of inflation in Ghana.

CHART III.1
CHART III.1

GHANA Inflation and Broad Money Growth, 1985–95 1/

(In percent)

Citation: IMF Staff Country Reports 1996, 069; 10.5089/9781451814736.002.A003

Sources: IMF, International Financial Statistics, Ghanaian authorities; and staff estimates.1/ Annual rates of change based on average quarterly data.

This paper builds upon and extends the previous work cited above, notably Kapur et al. (1991), in a number of ways. Firstly, in the spirit of disequilibrium adjustment models, it is argued that the prevalence of high transactions costs and information lags precludes rapid adjustment of actual inflation to its underlying long-run equilibrium level. This suggests that inflation models should incorporate a disequilibrium adjustment mechanism and thus a dynamics process. Secondly, note is taken of the fact that the inflation variable is based on the consumer price index (CPI) and that petroleum prices and imported good prices are by construction components of the CPI basket. The retail prices of petroleum and other imported goods may be taken as good leading indicators of inflation in Ghana, but this does not necessarily imply that they are the true underlying factors that cause inflation in the economy. Thirdly, none of the empirical models of inflation in Ghana so far has addressed the effects of inflationary expectations. Lucas (1976) has argued that econometric models, which do not explicitly model the effects of expectations, are prone to systematic prediction error.

The rest of this paper is organized as follows. Section B presents a model of inflation and the data description. Sections C and D present the econometrics test of the model, without and with inflationary expectations, respectively. Section E summarizes the conclusions. The Appendix contains the detailed econometric results of the inflation models.

B. The General Inflation Model and Data Description

The theoretical inflation model adopted in this paper is summarized in equation (1) below. That is, inflation (π) is a function of money growth (Δm) and movements in the opportunity cost of holding money (Δr). 1/

π=f(Δm,Δr).(1)

A priori, excess money supply growth in the economy should lead to a rise in inflation. Conversely, a rise in interest rates should draw liquidity from the economic system and lead to a reduction in inflation.

On the basis of this theoretical monetarist model, a general econometric model of inflation may be specified using a standard dynamic error-correction model (ECM) parameterization incorporating the relevant error-correction term ECM as follows:

A(L)πt=α0+B(L)α1Δm2t+C(L)α2Δrt+D(L)α2Δrt+D(L)α3Δet+F(L)α3Δet+F(L)α4ECMt1+Q1+Q2+Q3+μt,(2)

where π is the inflation rate based on the consumer price index (CPI), m2 is broad money, r is the domestic interest rate proxied here by the lending rate, e is the nominal exchange rate in terms of U.S. dollars, Q is a seasonal dummy, and μ is an error term. A(L) F(L) are polynomials of the form A(L) = ΣαrLr, in which L is a lag operator such that Lixt = xt-1, α0 is a constant term, and ECM is the error-correction term. All data variables are quarterly time series and expressed in natural logarithm (with the exception of the interest rate). The sample period is 1984 Q1-1995 Q4.

The existence of transactions costs and information lags implies that there are delays in the adjustment of actual inflation to its long-run equilibrium level. Estimation of the dynamic relationship between money and inflation should therefore incorporate a disequilibrium adjustment mechanism. This study is based on an error-correction (ECM) approach. The ECM is derived from the optimizing response of individuals to past disequilibria and can be thought of as a more general intertemporal version of partial adjustment (Nickell (1985)). The ECM can provide an insight into both the short-run and the long-run determinants of inflation.

C. A Dynamic Econometric Model of Inflation in Ghana

1. Stationarity status of data variables and test for cointegration

Recent advances in time series econometric modeling have established that the stationarity status of data variables should be identified before running a regression model. A data variable is said to be stationary if the mean and variance are invariant to time. If the issue of nonstationarity is not addressed, empirical inferences, and the policy prescriptions that are drawn from them, may be invalid.

This section tests for the stationarity of the data set. In identifying the order of integration of each data variable, two unit roots tests are applied. The results of these unit roots tests are presented in Table III.6. From the unit roots tests, inflation can be identified as a stationary 1(0) series, while money (m2), the domestic interest rate, and the exchange rate are all nonstationary I(1) series. Table III.7 reports the seasonal unit roots tests of the data series and confirms that the four data variables are all seasonally stationary.

To test for the existence of a long-run equilibrium relationship between the four data variables, following Johansen and Juselius (1990), a vector autoregression (VAR) with a constant, a trend, and a full set of seasonal dummies was estimated. The results are reported in Table III.8. According to the matrix testing for the number of cointegrating vectors, there is clearly one significant cointegrating vector given by β’ in the eigenmatrix β’ and one marginally significant cointegrating vector represented by β’ 2.

The first cointegrating vector, or estimated equilibrium solution, indicates that in the long run, inflation in Ghana is positively related to money supply and the exchange rate and negatively related to the domestic interest rate. The second marginally significant equilibrium cointegrating vector is however not easy to identify. 1/ The first cointegrating vector therefore provide a basis for defining the disequilibrium adjustment term of our error-correction model (ECM).

2. An error-correction model of inflation

On the basis of the stationarity and cointegration analysis and adopting a general-to-specific modeling procedure, an inflation equation was estimated using the general dynamic error-correction model specified in equation (2). Throughout the modeling process various parameterizations of variables were considered, but the specification adopted and reported in this paper outperformed all other alternative specifications. The final inflation equation was estimated using the recursive ordinary least squares estimator (ROLS).

3. Economic interpretation of the error-correction model of inflation

The estimated inflation model presented in Table III.2 is in a stationary first difference vector autoregressive space. The estimated ECM of inflation is generally consistent with the theoretical model in (1). The ECM of inflation indicates that growth in money supply (m2) has no immediate impact effect on inflation in Ghana. However, according to the regression model, growth in money supply tends to translate into inflation between the first and second quarters, with the maximum effect coming in after the second quarter. The effect of money on inflation dissipates after three quarters. As expected, the sum of the lagged coefficients on money, which is an estimate of the long-run multiplier, sums to unity in the limit. Further econometric analysis through the recursively computed one-step Chow (1960) statistics indicates parameter constancy and model stability (at the 5 percent significance level) of the estimated inflation equation.

Table III. 2.

The Estimated Nonexpectations Inflation Model in Ghana

article image
Source: Staff estimates; * and ** denote significance at the 5 percent and 1 percent levels, respectively.
Inflation Model
A(L)πt=α0+B(L)α1Δm2t+C(L)α2Δrt+D(L)α3Δet+F(L)α3Δet+F(L)α4ECMt1+Q1+Q2+Q3+μt,(3)

where A(L)…..F(L) are polynomials of the form A(L) = ΣαrLr, in which L is a lag operator such that Lixt = xt-1 and α0 a constant term.

(1) π is the inflation rate based on the consumer price index (CPI),

(2) m2 is broad money,

(3) r is the domestic interest rate proxied here by the lending rate,

(4) e is the nominal exchange rate in terms of U.S. dollars,

(5) ECM is the error-correction term,

(6) Q is a seasonal dummy, and n is an error term.

Based on the estimated model, interest rate or exchange rate movements in Ghana do not affect inflation significantly in the short run. Thus the results here are in accordance with the findings in Chhibber and Shafik (1992), which showed that there is no direct relationship between inflation and the official exchange rate in Ghana. They argued that short-run movements in the exchange rate could not explain the high and persistent inflation in Ghana. Nonetheless, our results show a significant relationship between inflation and the official exchange rate in the long run. Thus, it can be argued that in making short-run projections for inflation in Ghana, interest rates and exchange rate developments may be ignored, but in making long-range forecasts, both ought to be considered. From a policy perspective, this suggests that the stabilization of the value of the cedi should help to reduce the rate of inflation in the medium to long run. The estimated inflation model also indicates that seasonal factors do influence inflation in Ghana. Finally, the coefficient of the error-correction term in the inflation equation indicates that on average, actual inflation in Ghana adjusts to its underlying long-run equilibrium by 3.5 percent per quarter. This is quite low and underscores the fact that inflation was generally out of equilibrium in Ghana during the period of our study, 1984-95.

D. Incorporating Expectations in the Inflation Model

The above econometric analysis of inflation in Ghana has not incorporated the effects of inflationary expectations. In many countries, inflationary expectations quickly get embodied in interest rates and wage demands and they also affect spending behavior. From a more theoretical perspective, Lucas (1976) has argued that in situations where agents act rationally, then their behavior will necessarily change systematically with changes in the policy regime. Thus, any inference on economic behavior based on econometric models that do not explicitly model expectations effect will be prone to systematic predictive error. Previous empirical models of inflation in Ghana have not addressed the effects of inflation expectations.

1. Modeling expectations

The issue of modeling expectations in empirical work has generated considerable debate, and each method has its theoretical shortcomings as well as its econometric implications. Traditionally, inflation expectations can be modeled using the adaptive expectations operator. That is, expected inflation is specified as a weighted average of past inflation, and thus,

E(π)t=λ1πt1+λ2πt2....+λnπtn.(4)

A theoretically more appealing expectations assumption is that of rational expectations (RE). Most expectations formation models either assume that complete learning is possible, as in the extreme version of rational expectations, or that no further learning is possible at all. However, here we assume rational expectations but incorporate learning into the RE model. Our learning process of inflation is denoted by L(π) and specified as thus,

L(π)t=[0,t=t0[1+exp(αβ(tt0+1))]1,t0<t<t0+531,tt0+53],(5)

where at the beginning of our sample period (t=t0), it is assumed that people in Ghana do not know much about the inflation expectations generator but that as time increases within our sample, Ghanaians improve on their learning process. Throughout the sample period, people in Ghana are assumed to be improving on their learning specification about inflation (n) and attain rational expectations (perfect foresight) after the ex ante forecast period. Thus our learning-adjusted rational expectations expected inflation proxy becomes E(π)t - f(L(π)t, πt. 1/

We shall embed both expectations proxies (adaptive and learning- adjusted rational expectations variables) as competing variables into our estimated econometric model of inflation for Ghana. We denote the adaptive expectations variable by AE(π) and the learning-adjusted rational expectation inflation variable by LRE(π). The incorporation of the inflationary expectations gave rise to an endogeneity problem in our model. To circumvent this problem, we moved away from recursive ordinary least squares (ROLS) to recursive instrumental variable (RIV) estimation. Augmenting our inflation model by the two specified expectations proxies above, and re- estimating the model using the recursive instrumental variable method yielded the inflation equations that are reported in Tables III.3 and III.4.

2. Interpretation of the estimated expectations models of inflation

In estimating our inflation equations by instrumental variable method, we had to choose instruments for the inflation expectations variables. In theory, the instrument set (It) should have no correlation with the composite disturbance term μt but have a reasonably high degree of correlation with the inflation expectations proxy. 1/ The pursuit of robustness in our inflation estimation produced {Δmt-1, Δmt-2, Δet, ECMt-2|It} as the optimal instrument vector set for the adaptive expectations inflation model (Table III.3) and {Δmt-2, Δmt-3, Δmt-4, Δet, ECMt-2|It} for the learning-adjusted rational expectations inflation model (Table III.4). 2/

Table III.3.

The Estimated Adaptive Expectations Inflation Model in Ghana

article image
Source: Staff estimates; * and ** denote significance at the 5 percent and 1 percent levels, respectively.
General Inflation Model
A(L)πt=α0+B(L)α1Δm2t+C(L)α2Δrt+D(L)α3Δet+E(L)α4ECMt1+α5AE(π)t+Q1+Q2+Q3+μt,(6)

where A(L)…..E(L) are polynomials of the form A(L) = ΣαrLr, in which L is a lag operator such that Lixt = xt-1 and α0 is a constant term.

(1) π is the inflation rate based on the consumer price index (CPI),

(2) m2 is broad money,

(3) r is the domestic interest rate proxied here by the lending rate,

(4) e is the nominal exchange rate in terms of U.S. dollars,

(5) ECM is the error-correction term,

(6) AE(π) is expected inflation proxied by adaptive expectations,

(6) Q is a seasonal dummy, and μ is an error term.

Table III.4.

The Estimated Rational Expectations Inflation Model in Ghana

article image
Source: Staff estimates; * and ** denote significance at the 5 percent and 1 percent levels, respectively.
General Inflation Model
A(L)πt=α0+B(L)α1Δm2t+C(L)α2Δrt+D(L)α3Δet+E(L)α4ECMt1+α5LRE(π)t+Q1+Q2+Q3+μt,(7)

where A(L)…..E(L) are polynomials of the form A(L) = ΣαrLr, in which L is a lag operator such that Lixt = xt-1 and α0 is a constant term.

(1) π is the inflation rate based on the consumer price index (CPI),

(2) m2 is broad money,

(3) r is the domestic interest rate proxied here by the lending rate,

(4) e is the nominal exchange rate in terms of U.S. dollars,

(5) ECM is the error-correction term,

(6) LRE(π) is the learning-adjusted rational expectations inflation proxy,

(7) Q is a seasonal dummy, and μ is an error term.

Both expectations proxies are significant in the estimated inflation equation and improve the explanatory power of the model. However, the incorporation of the expectations variables does not drastically alter the broad conclusions reached with the non-expectations inflation model. According to our expectations models of inflation, money growth does not affect inflation contemporaneously but does so after a lagged period of one to two quarters. As in the non-expectations model of inflation, the maximum effect of excess money growth on inflation comes in after two quarters. However, we note that the learning-adjusted rational expectations variable dominates the adaptive expectations variable in our model in terms of both the relative significance in the model and the explanatory power, namely the resulting R2s. In general, the diagnostics of the estimated augmented learning-adjusted rational expectations inflation model is better than the adaptive expectations ECM of inflation (see Table III.9). Further comparative analysis (by non-nested encompassing tests reported in Table III.10) corroborates the dominance of the learning-adjusted rational expectations inflation model over the adaptive expectations model. This is not surprising given the incorporation of learning into our model in which we assumed that people in Ghana take time to fully understand the true inflation-generation process.

3. Forecast analysis of inflation

The within-sample forecast of inflation based on the expectations- augmented model of inflation is good. Actual and forecast inflation for 1995 with the standard error bounds, based on the dominant learning-adjusted rational expectations model, is reported in Chart III.2. As is evident from this chart, our dominant model projects a lower inflation than the actual throughout 1995, but the difference between the actual and forecast inflation narrows toward the end of the year. The ex ante inflation forecast analyses for 1996 based on the three estimated models are reported in Table III.5 and these indicate that on average, inflation in Ghana is unlikely to fall below 25 percent in 1996.

CHART III.2
CHART III.2

GHANA Inflation: Actual and Forecast for 1995 from the Estimated Error–correction Model 1/

Citation: IMF Staff Country Reports 1996, 069; 10.5089/9781451814736.002.A003

Source: Staff estimates.1/ Annualized rates.
Table III.5.

Summary Statistics of the Ex ante Inflation Forecast for 1996 Based on the Estimated Inflation Equations 1/

(In percent)

article image
Source: Staff estimates.

The analysis is based on the Fund’s monetary projections profile for 1996. The program target for broad money supply growth rate is 5 percent.

Model 1 is the estimated nonexpectations inflation model.

Model 2 is the estimated adaptive expectations model of inflation.

Model 3 is the estimated learning-adjusted rational expectations model of inflation.

The econometric models analyzed above are not definitive. 1/ Moreover, we do not have output variable in the estimated inflation equations that are reported here and this precludes the examination of possible Phillips curve-type effects on inflation in Ghana. 2/ Further econometric analysis was carried out by restricting a trend variable in our estimated cointegrating space. However, this did not yield a meaningful inflation/money long-run solution. This suggests that total factor productivity or output in general may not be a significant determinant of inflation in Ghana. In fact, Bruno and Easterly (1994) find that inflation in the range of 40 percent and above per annum over a prolonged period has strong negative growth effects in Ghana. Thus the causation in any case appears to run from inflation to growth rather than the converse.

E. Conclusions

This paper has developed econometric models of inflation in Ghana, using a dynamic error-correction modeling approach. Both the estimated long-run and short-run dynamic inflation equations have produced coefficient estimates that underline the impact of money on inflation in Ghana.

The analysis in this paper has shown that growth in money supply does not immediately affect inflation in Ghana but gets translated into inflation between the subsequent one to two quarters, with the maximum impact effect coming in after two quarters. Thus, according to our regression analysis, inflation in Ghana seems to be largely a monetary phenomenon, but its effect is subject to variable lags. Second, there exists no direct immediate short-run impact effect of movements in the nominal exchange rate on the rate of inflation in Ghana. However, the models suggest that in the medium to long run the stabilization of the value of the cedi should be an important part of the anti-inflation policy in Ghana. Third, we find seasonal factors important in explaining inflation in Ghana. Finally, the estimated models here indicate that actual inflation has adjusted very slowly to its equilibrium level, and this underlines the fact that inflation in Ghana has generally been out of equilibrium over the years.

Embedding inflationary expectations in the estimated inflation model for Ghana improves its explanatory power. Thus expectations do seem to influence inflation in Ghana and policies directed at dampening inflationary expectations should indeed reduce the rate of inflation. Outside sample forecasts of inflation in Ghana based on the estimated models reported here, and assuming the staff money supply projected profile for 1996, indicate that on average, inflation is unlikely to fall below 25 percent in 1996.

In summary, the estimated models in this paper have clearly identified an empirical relationship between money and inflation, which supports a monetary explanation for the high rate of inflation in Ghana. The implication is that maintaining a tight monetary policy should be the key element in gradually bringing inflation under control.

Table III.6.

Unit Roots Tests Results of the Individual Variables

(1984 Quarter 1-1995 Quarter 4) 1/

article image
Source: IMF, International Financial Statistics; staff estimates.
Table III.7.

Seasonal Unit Roots Tests Results of the Individual Variables

(1984 Quarter 1-1995 Quarter 4) 2/

article image
Source: Staff estimates.
Table III.8.

The Johansen-Juselius VAR Cointegration Results 1/

(1984 Quarter 1-1995 Quarter 4)

article image
article image
Source: Staff estimates.
Table III.9.

Competing Inflation Model Performance

article image
Table III.10.

Nonnested Encompassing Tests for Inflation Model 2 Versus 3

article image
Source: Staff estimates.

References

  • Abbey, J.L.S., Macroeconomic Review and Outlook - Executive Prognosis,Center for Economic and Policy Analysis (Accra, Ghana, 1996).

    • Search Google Scholar
    • Export Citation
  • Baba, Y., D.F. Hendry, and R.M. Starr, The Demand for Ml in the U.S.: 1960- 1988,Review of Economic Studies, Vol. 59 (1987), pp. 2561.

    • Search Google Scholar
    • Export Citation
  • Banerjee, A., J.J. Dolado, J.W. Galbraith, and D.F. Hendry, Cointegration, Error-Correction and the Analysis of Non-Stationarity Data Oxford: Oxford University Press (1993).

    • Search Google Scholar
    • Export Citation
  • Banerjee, A., J.J. Dolado, D.F. Hendry, and G.W. Smith, Exploring Equilibrium Relationships in Econometrics Through Static Models: Some Monte Carlo Evidence,Oxford Bulletin of Economics and Statistics, Vol. 48 [1986], pp. 253377.

    • Search Google Scholar
    • Export Citation
  • Bruno, M. and W. Easterly, Inflation Crises and Long-Run Growth,World Bank Draft Working Paper, (Washington: The World Bank, 1994).

    • Search Google Scholar
    • Export Citation
  • Charemza, W.W. and D.F. Deadman, New Directions in Econometric Practice, University of Leicester.

  • Chhibber, A. and N. Shafik, Exchange Reform, Parallel Markets and Inflation in Africa: The Case of Ghana,World Bank Working Papers, WPS 427 (1990).

    • Search Google Scholar
    • Export Citation
  • Chhibber, A. and N. Shafik, The Inflationary Consequences of Devaluation with Parallel Markets: The case of Ghana,Mimeo, (Washington: The World Bank, 1992).

    • Search Google Scholar
    • Export Citation
  • Dickey, D.A. and W.A. Fuller, Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root,Econometrica, Vol. 50, (1981), pp. 105772.

    • Search Google Scholar
    • Export Citation
  • Engle, R.F. and C.W.J Granger, Cointegration and Error Correction: Representation, Estimation and Testing,Econometrica, Vol. 55 (1986), No. 2, pp. 251276.

    • Search Google Scholar
    • Export Citation
  • Fischer, S., The Role of Macroeconomic Factors in Growth,Journal of Monetary Economics Vol. 32 (1993), pp. 48583.

  • Friedman, M. (1962). “The Interpolation of Time Series by Related Series,Journal of American Statistical Association (December 1962).

    • Search Google Scholar
    • Export Citation
  • Friedman, M., A Theoretical Framework for Monetary Analysis,Journal of Political Economy, Vol. 78 (1970), pp. 193238.

  • Hendry, D., Econometric Modeling with Cointegrated Variables,Oxford Bulletin of Economics and Statistics, Vol. 48 (1986), pp. 201212.

    • Search Google Scholar
    • Export Citation
  • Hendry, D. and J.A. Doornik, PC-GIVE (7.00): An Interactive Econometric Modelling System, Institute of Economics and Statistics, (Oxford, 1994).

    • Search Google Scholar
    • Export Citation
  • Hendry, D., N. Ericsson, and H.A. Tran, Cointegration, Seasonality, Encompassing and the Demand for Money in the U.K.,Federal Reserve Board, (Washington D.C., 1992).

    • Search Google Scholar
    • Export Citation
  • Johansen, S., Statistical Analysis of Cointegrated Vectors,Journal of Economic Dynamics and Control, Vol. 12 (1988), pp. 23154.

    • Search Google Scholar
    • Export Citation
  • Johansen, S. and K. Juselius, Maximum Likelihood Estimation and Inference on Cointegration- with application to Demand for Money,Oxford Bulletin of Economics and Statistics, Vol. 52 (1990), pp. 169210.

    • Search Google Scholar
    • Export Citation
  • Johansen, S. and K. Juselius, Identification of the Long-run and Short-run Structure - An Application to the ISLM Model,Journal of Econometrics, (forthcoming) [1995].

    • Search Google Scholar
    • Export Citation
  • Johnson, O., C. Pazarbasioglu, H. Fosgaard, and P. Hayward, Monetary Reforms to Enhance the Effectiveness of Monetary and Exchange Market Operations,Manuscript, Monetary and Exchange Affairs Department (Washington: International Monetary Fund, 1964).

    • Search Google Scholar
    • Export Citation
  • Kapur, I., M.T. Hadjimichael, P. Hilbers, J. Schiff, and P. Szymczak, Empirical Estimates of Inflation,” Appendix II in Ghana: Adjustment and Growth, 1983-91, IMF Occasional Paper, No. 86 (Washington: International Monetary Fund, 1991.

    • Search Google Scholar
    • Export Citation
  • Lucas, R.E., Econometric Policy Evaluation: A Critique,” in The Phillips Curve and Labor Markets, ed. K. Brunner and A.H. Meltzer, Amsterdam: North Holland Publishing Company, Carnergie Rochester Conference Series on Public Policy, Vol. 1, (1976), pp. 1946.

    • Search Google Scholar
    • Export Citation
  • Maddala, S. [1977]. Econometrics, Singapore: McGraw-Hill.

  • Nickell, S.J., Error Correction, Partial Adjustment and All That: An Expository Note,Oxford Bulletin of Economics and Statistics, Vol. 47 (1985), No. 2, pp. 119129.

    • Search Google Scholar
    • Export Citation
  • Sowa, N.K., “Policy Consistency and Inflation in Ghana,” Unpublished manuscript, Department of Economics, University of Ghana, Legon [1992].

    • Search Google Scholar
    • Export Citation
  • Sowa, N.K. and J.K. Kwakye, “Inflationary Trends and Control in Ghana,” Unpublished manuscript, Department of Economics, University of Ghana, Legon [1991].

    • Search Google Scholar
    • Export Citation

Statistical Annex

Table 1.

Ghana: Gross Domestic Product by Sector, 1990–95

article image
Sources: Statistical Service; and staff estimates.

Revised estimates.

Estimates.

Including restaurants and hotels.

Table 2.

Ghana: Gross Domestic Product by Expenditure Category, 1990-95

article image
Sources: Statistical Service; and staff estimates.

Revised estimates; reflects government accounts on commitment basis as of 1994 onwards.

Estimates.

Table 3.

Ghana: Composition and Growth of Gross Domestic Product by Sector, 1990–95

article image
Sources: Statistical Service; and staff estimates.

Revised estimates.

Estimates.

Table 4.

Ghana: Composition and Growth of Gross Domestic Product by Expenditure Category, 1990-95

article image
Sources: Statistical Service; and staff estimates

Revised estimates; reflects government accounts on commitment basis as of 1994 onwards.

Estimates.

Approximated by total central government capital expenditure, including capital outlays financed by external project aid.

Approximated by net lending to state enterprises by the Government; excluding investment financed by state enterprises’ savings and/or domestic borrowing, which is included in private investment.

Table 5.

Ghana: Cocoa Bean Production, Consumption, Prices, Payments to Farmers, and Export Receipts, 1984/85-1995/96

article image
Source: Cocoa Board.

Includes sales to processing companies; most of the processed products are then exported.

Including bonus payments until 1986/87, but excluding bonus payments thereafter.

Main crop.

Midcrop.

Estimates.

Table 6.

Ghana: Operations of the Cocoa Board (Cocoa Division), 1988/89-1995/96 1/

article image
Source: Cocoa Board.

Crop year ending September 30.

Mainly discount charges on bills drawn to finance the purchases of cocoa, export duty, and operations of the Cocoa Board.

Includes provision for doubtful debts and depreciation. Includes all other Cocoa Board costs in 1992/93.

Includes outlays for produce inspection, research, construction of feeder roads, and subsidies for insecticides and spraying. Includes a provision of 8.5 billion cedis for retrenchment in 1993/94.