Guinea: Selected Issues and Statistical Appendix
Author: International Monetary Fund
Publication Date:
14 Aug 2003
eISBN:
9781451815184
Language:
English
Keywords:
ISCR; CR; microfinance institution; broad money growth; Guinea; inflation equation; inflation variable; inflation risk; consumer price index market basket; inflation trend; Inflation; Demand for money; Monetary aggregates
Download PDF (703.7 KB)

The paper first discusses price trends and the relationship between money growth and inflation. Second, it focuses on the central challenge of improving competitiveness and promoting exports to enhance growth in the economy. Finally, it reviews the microfinance sector. The study also includes the following statistical data: economic and financial indicators, consumer price index, central government revenue and expenditure, monetary survey, structure of interest rates, balance of payments, composition of imports and merchandise exports, nominal and effective exchange rates, and external public debt.

Abstract

The paper first discusses price trends and the relationship between money growth and inflation. Second, it focuses on the central challenge of improving competitiveness and promoting exports to enhance growth in the economy. Finally, it reviews the microfinance sector. The study also includes the following statistical data: economic and financial indicators, consumer price index, central government revenue and expenditure, monetary survey, structure of interest rates, balance of payments, composition of imports and merchandise exports, nominal and effective exchange rates, and external public debt.

II. Inflation and Monetary Pass-Through in Guinea1

A. Introduction

6. This paper develops stylized facts about the inflationary process in Guinea, focusing particularly on the relationship between money growth and inflation. The adequate assessment of inflationary pressures, and the forecasting of inflation trends is particularly important in designing economic policy, notably a proper fiscal policy based on accurate revenue and expenditure forecasts. To this end, we examine the influences of changes in broad and reserve money on CPI inflation using quarterly data for the September 1991-March 2003 period. Limited data availability for the period calls for caution when interpreting the results. The findings nonetheless provide indications on how to interpret changes in monetary aggregates and guide monetary policy.

7. Since independence, price trends in Guinea have been largely distorted by administratively controls, and have not reflected market mechanisms, in particular up to the late 1970s. The partial and progressive liberalization that followed, in the context of significant fiscal imbalances and loose monetary policy, resulted in periods of hyper-inflation. Only in the late 1980s and early 1990s was CPI inflation reduced to single-digit levels. Given the importance of exogenous factors in the determination of price levels in the country, the role of the monetary pass-through in CPI inflation might have been underestimated.

8. To provide evidence on the links between inflation and money growth, we build a bivariate inflation model containing monetary growth and CPI inflation. The study shows that, in the last ten years, a significant long-run relationship may be established between money growth and CPI inflation. Using an error correction model and impulse response analysis, we also find that money has an immediate and lasting impact on inflation. The results support the argument in favor of an active monetary policy in order to maintain inflation at reasonable levels. The consideration of broad money and reserve money variables suggests that monetary policy may act in two related ways: (i) by direct liquidity management, to contain reserve money growth; and (ii) by ensuring that the policy mix does not lead to excessive broad money expansion in the economy.

9. The remainder of this chapter is organized as follows. Section provides a summary of recent developments in CPI inflation. Section C discusses the composition of the Guinean CPI and other data-related issues. Section D presents the theoretical framework of the model. Section E investigates the long-run relationship between money and inflation, while Section F focuses on short-run dynamics. Section G concludes and provides policy recommendations.

B. Developments in Inflation and Money Growth

10. The 12-month rate of change in the CPI fell from 7.2 percent in December 2000 to 1.1 percent in December 2001, and was contained below 3 percent until October 2002. Policy slippages in 2000 had resulted in increased inflationary pressures, as fiscal spending increased dramatically, and as its monetary financing led to a significant monetary overhang. Twelve-month CPI inflation returned to double-digits in the second half of 2000, reaching 10.4 percent, and remained around or above 7 percent until the last quarter of 2001, when fiscal discipline and monetary tightening eventually contributed to curbing inflation.

11. Developments in the last quarter of 2002 and first quarter of 2003 show a resurgence of inflation, rising to 6.1 percent in December 2002 and 10.4 percent in March 2003. With a very substantial increase in credit to the government, and in base and broad money, a significant drop in net foreign assets of the central bank, further increases in inflation, reflecting the loose policy stance, would pose serious risks to the economy, including prolonged output losses, a long-term distortion in the allocation of resources, and an acute impact on the poor segments of the population. These developments illustrate the consequences of “stop-and-go” economic policies.

Guinea: Trends in Inflation, 1962–2003

The examination of past inflationary trends in Guinea reveal the risks associated with inflation in a country that has been exposed to prolonged periods of hyperinflation. While price levels during the first two decades after independence were largely administered, and inflation remained under control, averaging 6.7 percent from 1964 to 1979, it accelerated sharply with the first liberalization attempts, as supply responses were weak and excessive fiscal spending, and its monetary financing, increased inflationary pressures. With the extensive reform program that followed the 1984 coup, inflation first increased dramatically, culminating at an annual average of 65 percent in 1986 when the national currency was devalued by 92 percent, trade was liberalized, and price controls removed (except those on fuel and rice).

uA02fig01

Guinea: Consumer Price Index, 1963-2002

(Annual percent charge)

Citation: IMF Staff Country Reports 2003, 251; 10.5089/9781451815184.002.A002

After 1986, the annual average rate of inflation dropped: sharply (see figure II.2). This positive performance was supported by favorable climatic conditions leading to very significant increases in agricultural output and food production and to low price increases in food produce, by low price rises in basic commodities, including housing and transport, and by exchange rate stability and a decline in imported prices, in particular in the prices of imported rice (see figure II.1). Aside from 1995, when elections and heavy rains pushed prices upward, the declining trend in inflation continued. Annual average CPI inflation was under 2 percent in 1997, and remained under control until 1999, largely due to the relative stability of monetary policy. Broad money growth was limited, bank credit to the government contained, and net foreign assets remained at adequate levels.

Figure II.1.
Figure II.1.

Guinea: Consumer Price Index, Dec. 1992 - March 2003

(12-month year-on-year growth rates, in percent)

Citation: IMF Staff Country Reports 2003, 251; 10.5089/9781451815184.002.A002

Source: Guinean authorities, DNSI.
Figure II.2.
Figure II.2.

Impulse Response Analysis on the Inflation Variables

Citation: IMF Staff Country Reports 2003, 251; 10.5089/9781451815184.002.A002

C. Composition and Structure of the Guinean CPI

12. The Guinean CPI covers 165 items since 1987. The components and their weights in the CPI are summarized in table II.1. Food items account for a large proportion of the basket, which suggests that factors affecting food prices dominate movements in the CPI (see figure II.1). These factors include agro climatic conditions, wages, domestic inputs, and transportation costs, with rainfall playing a key role. In particular, the scarcity of food products between harvests leads frequently to tensions in domestic prices, and to sudden jumps in the price level - illustrated in Figure II.1 in the summer months. Figure II. 1 also suggests that domestic prices are more volatile than imported product prices, and weigh more heavily on CPI movements. In particular, the dramatic reduction in the rate of inflation until December 1993 was mainly driven by the decline in the inflation of non-tradable goods prices in the CPI, while the decline in imported products prices was less marked.

Table II.1.

Guinea: Consumer Price Index Market Basket

(In percent)

article image
Source: DNSI, 2003

13. A number of data weaknesses may be identified at the onset of the study. The index may not reflect socio-economic realities in the capital city. Also, the weights (based on a 1991 survey) and some of the items composing the index are obsolete.2 Further, information is being collected in five markets in different neighborhoods of Conakry. Supermarkets and small retail shops, which account for a growing share of domestic sales, are excluded from the sample. Another limitation of the index is the absence of a breakdown between urban and rural indices, as it covers only the capital city, Conakry.3 In light of those weaknesses, there has been some concern that official figures may not reflect the actual rate of inflation.

14. The statistical relevance and accuracy of the various measures of inflation is an additional issue, developed further in appendix I. In the case of Guinea, 12-month inflation growth rates tend to overweight outlier observations, often related to the price jumps between harvests, as mentioned above. Measures of inflation may therefore be distorted. The use of quarterly data tracks better current developments in inflation, while limiting the presence of noise in the data.

D. Theoretical Framework

15. An inflation equation is derived to measure the impact of the relevant explanatory variables and predict the inflationary outcome of economic policy and exogenous factors. In a small open economy, the price level is commonly postulated to be influenced both by money demand and by imported inflation. The overall price level (CPI) is a weighted average of the price of domestic goods and imported goods. The price of imported goods is determined exogenously in the world market, and valued domestically according to the level of the exchange rate. Given limited data available and the shortness of the time series, the study focuses on domestic determinants of the price level. The price of goods is assumed to be determined domestically by the difference between real money demand and supply, the demand for real money balances being determined by the following four variables: real income; inflationary expectations; expected domestic interest rates; and expected foreign interest rates:

p=md(y,pe,rd,rw)−ms[1]

Equation [1] specifies a long-term relationship in the money-market. An increase in real income, expectations of a reduction in the rate of inflation, an increase in expected domestic interest rates, and a reduction in foreign interest rates will increase money demand, and thereby lead to inflationary pressures.4

16. Two alternative bivariate models are considered: z1t = [Δpt,Δrest] and z2t = [Δpt,ΔM2t] containing monetary growth and inflation. Two monetary variables are used, corresponding to the narrowest (reserve money, rest) and the broadest (broad money, M2) monetary aggregates. Under the hypothesis of a long-term relationship between the variables, we proceed by including all the variables in a single dynamic system, where none of the variables is assumed a priori to be exogenous. We estimate a vector auto regression (VAR) model and apply cointegration tests to verify whether there are any long-run relationships among the variables. The analysis is based on an auto-regression model of the form:

Δxt=μ+∑i=lkΓtΔxt−i+Πxt−i+εt[2]

where xt is the vector of the endogenous variables included in the model (Δp,Δm) the parameters μ and Π are allowed to vary without restrictions; k is the lag length of the model; and εt is a vector of errors with mean zero. The existence of a cointegrating relationship between the variables is tested by analyzing the rank of the matrix Π, using the methodology developed by Johansen (1988, 1995).

E. Long-Run Relationships Between Money and Inflation

17. The empirical analysis is conducted using quarterly data. End-of-period values are used, and all variables are in logarithms. The time series are not seasonally adjusted in order to preserve their time series properties.5 The variables are as follows:

  • Inflation variables: As the Guinean authorities do not compute a non-core inflation index, the non-food index (NONF) is used as an alternative to the composite CPI index (CPI). A stronger relationship between money growth and nonfood inflation is expected a priori in the model, as the composite index is more sensitive to exogenous events, such as geoclimatic conditions, and may overweight food products (see Section C).

  • Monetary variables: Two monetary aggregates are used alternatively, broad money (M2) and reserve money (RES). Quarterly monetary data are available only from September 1991, restricting the sample for the model to September 1991-March2003.

18. Standard augmented Dickey-Fuller (ADF) unit root tests suggest that the variables are integrated of order 1,1(1), i.e., they are nonstationary in levels while the first differences of the variables are stationary (see Table II.3). As a result, and given that the two variables are considered endogenously determined, we proceed to formulate an unrestricted VAR model, with appropriate lags.6 The results are presented in appendix II.

19. The Johansen methodology is used to test the presence of a cointegrating relationship. Two series of results are reported, the trace statistics and the maximum eigenvalue test, both of which provide indication as to whether the hypothesis of cointegration between the variables should be rejected. Economic theory should then be used to place restrictions and identify the long-run equations. Given the paucity of data in the case of Guinea, long run equations, in the form of the money demand equation defined above, may not be fully identified. Variables such as real income, and domestic interest rates are not available for the period under study. Caution should therefore be exercised in the interpretation of the identified long-run relationship between the price level and monetary aggregates. The coefficients may represent only partially the extent of the relationship, and an omitted variable bias may affect the coefficients. The results from the cointegration analysis are presented in Table II.4. The VAR model consists of the appropriate lags on the endogenous variables, with a term trend, restricted to lie within the cointegration space.7 In all four models, the hypothesis of no cointegration between the endogenous variables is rejected, and one-long run relationship is identified between the price level and money at the 5 percent level of confidence.

The four long-run relationships may be written as follows:

LCPI = 0.317LM2 + 0.004Trend

LCPI = 0.477LRES + 0.004Trend

LNONF = 0.257LM2 + 0.004Trend

LNONF = 0.160LRES + 0.007Trend

20. The model performs well in terms of explaining the price level as a function of money, suggesting that increases in the money stock have a substantial long-term effect on inflation.8 The impact of reserve money on the composite CPI is found larger than the impact of broad money, as expected. However, the results on the non-food price index are somehow counterintuitive, with (i) the impact of monetary variables being more subdued,9 (ii) broad money having a larger impact than reserve money on the non food price level. In addition, the coefficients of the money variables are lower than could be expected, suggesting that a doubling in the money supply results in an increase in prices of only about 16 to 48 percent. Two explanations may be given for such low coefficients. First, lack of data and incomplete model specification (as suggested in the above paragraph) may introduce an error-in-variables problem. This would explain why the time trend enters positively in the model, compensating for low coefficients on money variables. Second, the price level may be affected only by large changes in the money supply. The positive coefficient of the trend term would then suggests persistence of inflation over time.

F. Short-Run Inflation Dynamics

21. The dynamic version of the long-run relationship estimated in equation [2] can be specified as an error correction model of the form:

Δxt=μ+∑i=1kΓiΔxt−1+αβ′xt−1+εt[3]

which differs from a simple vector autoregressive model in first differences in the β′xt−i term, called the error correction term. The error-correction model approximates deviation from the equilibrium and estimates the short-run response necessary for the system to return to its equilibrium.

22. We estimate the error-correction model using the full information maximum likelihood. While the system estimates error-correction equation for each of the two variables in the model, we report in Table II.5 only the four error-correction equations corresponding to the inflation equation. In each of the four equations, the coefficient of the error correction term is significant at the 5 percent level, and varies in levels between -0.604 and -0.733, suggesting that about two-third of the excess in the money supply is reabsorbed through inflation in each quarter. However, the suggested impact of lagged money growth is counterintuitive, as the model suggests that acceleration of the rate of money growth results in a reduction in inflation. Finally, coefficients on lagged inflation are significant and positive (at varying lags depending on the variables entered in the model), suggesting that inflation is persistent over time.

23. The dynamic vector error-correction model estimates the dynamic impact of changes in the variables in the model. An impulse response analysis traces the impact of a shock to the i-th variable to all of the other endogenous variables in the model. The effect of a change in the monetary and the inflationary variable on current and future values of the inflation variable is represented in Figure II.2. In response to a shock in money growth, inflation rises significantly over a period of about two years, after which it stabilizes at a higher level. This response is consistent with what was expected, the increase in the stock of broad money or reserve money resulting in a permanent increase in price levels. The impact of a contemporaneous shock in inflation on future values of inflation suggests an immediate jump in price levels, which decreases over time. After between 5 and 6 periods, the impact turns negative, and leads to a reduction in inflation. This may suggest that, experience of high inflation leads to a policy response that eventually reduces the rate of inflation. This interpretation is supported by the experience in Guinea since 1991. It also confirms that there is a significant role for monetary policy in addressing inflation risks.

G. Conclusion

24. The findings in this chapter support a long-term relationship between money and inflation, with money growth passing through inflation. Short-term dynamics reinforce this impact. Impulse response analysis show that a shock in the money stock will have an increasing impact on inflation over two years, and stabilize at a higher level after that time.

25. From a policy perspective, both reserve money and broad money are found to significantly impact inflation over time. This reveals the need for monetary policy to focus on both aggregates. Active liquidity management (presently in the form of issuance of sterilization bills) is needed to meet reserve money targets. To prevent excessive broad money expansion, an adequate macroeconomic policy mix is required, including, notably, a sustainable fiscal policy. The financing of the government deficit through treasury bills, even though sounder than its monetary financing, would lead to an expansion in net domestic assets, and hence in broad money, with a substantial impact on price levels in the economy.

26. The resurgence of high inflation in recent months is very specific, notably as it is counter-seasonal. This suggests that the role of monetary expansion may have been greater in recent trends than over the period examined in this model, and hence reinforces the urgent need for a monetary tightening if inflation is to remain at sustainable levels.

References

  • Blinder, A., “Commentary,” Federal Reserve National Bank of St. Louis Review, 1997 (May/June), pp. 157–60

  • Cecchetti, S., “Measuring Short-Run Inflation for Central Bankers,” Federal Reserve National Bank of St. Louis Review, 1997 (May/June), pp. 143–55

    • Search Google Scholar
    • Export Citation

  • Dickey, D., and Fuller, W. A., “Likelihood Ratio Statistics for Autoregressive Time Series with a Unit Root,” Econometrica, 1981, Vol. 49 (June)

    • Search Google Scholar
    • Export Citation

  • Johansen, S., “Statistical Analysis of Cointegrating Vectors,” Journal of Economic Dynamics and Control, 1988, Vol. 12, pp. 231–54

    • Search Google Scholar
    • Export Citation

  • Johansen, S., Likelihood-Based Inference in Cointegrated Vector Autoregressive Models, 1995, Oxford University Press, Oxford

Appendix I: Measures of Inflation and Early Signals in Guinea

A common way to measure inflation is to use data on the year-on-year rate of change, sometimes complemented by the twelve-month annual average rate of change. Twelve-month growth rates present a summary of the trends over the last twelve months, but reflect only partially the most recent changes in inflation and can therefore be misleading. Inflationary trends in Guinea over the last year are illustrative. The annualized one-month and three-month rates indicated as early as July 2002 the emergence of inflationary pressures, while the twelve-month rates remained at low levels until October 2002. During the same period, monetary policy remained lax and failed to absorb the excess liquidity injected in the system. In order to provide early monetary policy responses, the one-month and three-month growth rates may provide adequate additional indicators.

Table II.2.

Guinea: Developments in the Overall CPI, January 2000–December 2001

(Unless otherwise indicated)

article image
Source: Guinean authorities, and staff calculations

Misleading information may come from the fact that simple-moving averages are very sensitive to shocks and outliers. In Guinea, outliers in inflation series are frequent, and their impact is persistent in the published 12-month growth rates of inflation. To illustrate the argument, we correct the CPI index for outliers, and replace the monthly price change by the average of the previous and next month price changes. The lasting impact of a one-month price change is clear. For example, the 5 percent monthly increase in September 2000 -the largest monthly increase by far in the series- caused the twelve-month rate of inflation to be 4.8 percentage points higher than it would have been without the outlier observation for the next eleven-month period.10 In the same way, a base effect distorts twelve-month inflation time series, when the drop (or increase) in the series is caused by the dropping out of the twelve-month average rate of the abnormally high (or low) rate observed in the twelfth month before. In the Guinea data, this is observed most clearly in the reduction in the twelve-month growth rate from 8.3 percent in August 2001 down to 3.4 percent in September 2001. One of the solutions proposed in the literature11 is to use the three-month growth rates, which take into account more accurately current inflation developments while reducing the amount of noise from the one-month growth rate.

Appendix II: Unit Root Tests and Estimation Results

Table II.3.

Unit Root Tests

article image
Notes:1. The ADF is the augmented Dickey-Fuller test. The null hypothesis is that a series contains a unit root.2. The number of lags was determined automatically based on the Schwartz Information Criteria (SIC).3. The reported statistics are for the tests with a constant term and a linear trend included.

The existence of a cointegrating relationship is tested by analyzing the rank of the matrix Π. If the coefficient matrix Π has reduced rank τ < n, where n is the number of endogenous variables, then there exist k × τ matrices α and β each with rank τ such that Π = αβ′ and β′yt is I(0). β contains the cointegrating vectors, and the elements of α are known as the adjustment parameters in the VEC model, i.e., the coefficients of the error correction terms, or the speed of adjustment to the long run equilibrium.

Table II.4.

Long-Run Economic Relationships between Money and Inflation

article image
Note: In order to perform the cointegration tests, the series are assumed to have an intercept and a trend, restricted to the inside the cointegration space.
Table II.5.

Vector Error Correction Estimates for D(LCPI) and D(LNONF)

article image
Notes: T-statistics are reported in italic; ** indicates significant at 5 percent level, * indicates significant at 10 percent level.

Results for normality are the multivariate extensions of the Jarque-Bera residual normality tests, based on the Cholesky orthogonalization method. Except for one equations, we may not reject the hypothesis of multivariate normality at the 5 percent degree of confidence.

1

This paper was prepared by Rodolphe Blavy.

2

This is the case for some items that are not consumed in Guinea anymore, for which the index is frozen, accounting for more than 3 percent of the CPI. Also, new consumption patterns are not captured, including mobile phones, transports, construction, and food products for which consumption has increased very substantially, e.g., rice, potatoes.

3

The authorities have acknowledged the urgent need to revise the CPI methodology used in Guinea, based on a 1995 survey of consumption patterns, and on the results of the poverty map to be established in 2003. The number of items in the index will be expanded to 317, and the weights for each item will be determined based on a survey of 4416 households, compared to 300 households for the present index. In the medium-term, the index will cover all major urban centers in the country.

4

The independent role of geoclimatic conditions in influencing food prices, the impact of domestic petroleum price fluctuations, and the indirect impact of electoral spending during major national elections may also be substantial.

5

Similar results were obtained with seasonally adjusted series.

6

The appropriate lag length is chosen based on the results of lag exclusion tests, and the various lag length criteria computed in e-views, including the sequential modified LR statistic, the final predictor error, the Akaike information criterion, the Schwarz information criterion, the Hannan-Quinn information criterion.

7

Alternative specifications have been explored, in particular models with a constant, with and without timetrend. The results are not reported here and are broadly similar to those presented. The model proposed here is retained as it provides the strongest cointegrating relationship between the variables. A forthcoming working paper will present detailed results and discussion of the various models.

8

Results on the pair wise Granger test of causality reinforce the findings that money and inflation are to be treated as endogenous variables. However, the Wald statistic is lower for the inflation variable in the money equation, suggesting that the direction of causality is indeed from money to prices.

9

One possible explanation for the impact of monetary variables on the non-food CPI being more subdued is a consequence of the absence in the model of variables accounting for the changes in the nominal exchange rate and for other import price shocks. However, the results obtained after inclusion of the nominal exchange rate in the model provided insignificant coefficients for that variable.

10

A similar phenomenon was identified by M. Kaufman and R. Luzio in the Selected Issues Paper on “Price Stability and the Choice of Inflation Target for Monetary Policy”, Kenya, IMF 2002

Same Series

Other IMF Content

Other Publishers

Cover IMF Staff Country Reports
Guinea: Selected Issues and Statistical Appendix
Author: International Monetary Fund
Volume/Issue:
Volume 2003: Issue 251
Publisher:
International Monetary Fund
ISBN:
9781451815184
ISSN:
1934-7685
Pages:
83
DOI:
https://doi.org/10.5089/9781451815184.002
  • View in gallery

    Guinea: Consumer Price Index, 1963-2002

    (Annual percent charge)

  • View in gallery

    Guinea: Consumer Price Index, Dec. 1992 - March 2003

    (12-month year-on-year growth rates, in percent)

  • View in gallery

    Impulse Response Analysis on the Inflation Variables

Guinea: Consumer Price Index, 1963-2002

(Annual percent charge)

Guinea: Consumer Price Index, Dec. 1992 - March 2003

(12-month year-on-year growth rates, in percent)

Impulse Response Analysis on the Inflation Variables