Abstract
This Selected Issues paper analyzes the underlying sources of growth in Uganda, suggesting that the contribution to growth from total factor productivity has been minor, while the high population growth poses a significant challenge to sustain a rapid improvement in living standards. The paper takes a closer look at the monetary transmission mechanisms in Uganda, aimed at assessing the appropriate choice of intermediate target and mix of liquidity sterilization instruments. It also focuses on the recent financial sector reforms undertaken by the government.
II. The Choice of Intermediate Target and Mix of Sterilization Instruments20
A. Introduction
25. Uganda has succeeded in stabilizing inflation after years of very high inflation in the late 1980s to early 1990s. A key factor in bringing inflation under control has been the implementation of tight financial policies along with prudent macroeconomic management (IMF 1996). Despite its success in achieving price stability, the Bank of Uganda (BOU) conducts monetary policy in a difficult setting characterized by a high propensity to transact in cash, and shallow financial markets. This is further complicated by the large foreign exchange inflows in the form of grants and loans (in excess of 10 percent of GDP) to finance the county’s large fiscal deficit and poverty reduction strategy, which injects vast amounts of liquidity into a thinly monetized economy. Against this background, monetary conditions have been characterized by high and volatile interest rates, reflecting pressures due to sterilization operations and difficulties in liquidity management.
26. In conducting monetary policy to achieve its ultimate price objective, a central bank may use either a monetary quantity or an interest rate as its intermediate target.21 Since 2002, the BOU has been targeting the monetary base to achieve its medium-term goal of maintaining underlying (nonfood) inflation below 5 percent. However, interest rates have remained high and volatile, albeit declining recently, questioning the appropriateness of a monetary target and the feasibility of using an interest rate as an operating target. For an operational target to succeed in achieving a price objective there should be an identifiable transmission mechanism from the target to the price objective so that policymakers can determine the level they should seek.22 Therefore, the second section of the chapter attempts to identify the transmission mechanism of monetary aggregates and interest rates on inflation. As it turns out, monetary aggregates, particularly broad money and to a lesser extent, reserve money, seem to have a significant impact on prices in the sample period, whereas interest rates and NDA do not, suggesting that the BOU should continue to operate within a monetary targeting framework at present.
27. The government’s preeminent monetary policy challenge is to absorb the vast amounts of liquidity being injected into the economy as it draws down its external grants and loans so as to meet its price objective.23 The BOU’s sterilization operations rest on two pillars: it can either absorb the liquidity by selling foreign exchange, or, (on behalf of the government) issue treasury bills and bonds.24 For a given monetary target, when excess liquidity is mopped up through increased sales of foreign exchange, the sales carry the risk that an exchange rate appreciation (or a slower depreciation) may adversely affect competitiveness and hence, the export sector.25 At the same time, the use of domestic debt sales for sterilization purposes may (i) put undue upward pressure on interest rates, crowding out credit to the private sector; (ii) erode the BOU’s profitability as it absorbs the burden of the spread between the interest rate it pays on open market paper and that which it receives from foreign reserve holdings and/or other investments; and/or (iii) put pressure on the government budget, as it ultimately bears the costs of BOU’s sterilization operations. The budgetary costs of additional domestic debt issuance are clear, but the optimal mix of sterilization instruments in the short run will crucially depend on the price elasticities of the treasury bill and foreign exchange markets, on the one hand, and on the differential impact of exchange rate and interest rate changes on the real economy, on the other. However, it is important to bear in mind that under a floating exchange rate regime the amount of foreign exchange sales should be reasonably closely related to the amounts of net donor flows over time to facilitate the requisite current account adjustment and avoid a significant real exchange rate misalignment, which needs to be closely monitored to ensure that the mix chosen does not eventually cause the real exchange rate to deviate far from its equilibrium.26 The price elasticities of the treasury bill and foreign exchange markets are expected to be high because of the shallowness of markets, therefore, the third section of this chapter will provide a preliminary examination of the impact of interest rate and exchanges rate changes on the business cycle.
Section II attempts to identify the transmission of monetary aggregates and interest rates on prices using a recursive Vector Autoregression (VAR) approach encompassing the key sources of inflation in low-income countries (LICs). The third section estimates the marginal impact of interest rate and exchange rate changes on the manufacturing production index and exports, given the lack of high frequency GDP, which would help gauge the weight to attach to the mix of tools for liquidity management. Given the BOU’s preannounced objective of maintaining core inflation below 5 percent and its intent to progress toward a formal inflation targeting (IT) regime, the fourth section provides a short exploration of the scope for adopting a full-fledged IT framework. Section five presents conclusions.
B. Monetary Transmission Mechanism
Inflationary dynamics are evaluated using a six-variable recursive VAR approach.27 The structure is a stripped-down version of McCarthy (2000), who employs a production chain model of price determination. The underlying model of the distribution chain translates into a recursive structure of the variance-covariance matrix, which enables the identification of shocks stemming from external and policy developments and their effects on consumer inflation. The structural shocks are recovered from the VAR residuals using the Cholesky decomposition of the variance-covariance matrix.28 The VAR comprises six variables in the following order: international oil prices, coffee prices, output gap, exchange rate, monetary aggregates or interest rates, and consumer prices.
28. The system encompasses the four key sources of inflation identified by Loungani and Swagel (2001) in their analysis of the sources of inflation in developing countries. First, as discussed by Agenor and Montiel (1999), inflation in developing countries is often linked to underlying fiscal imbalances. Such imbalances can lead to an increase in inflation by causing excessive money creation, as in Sargant and Wallace (1981), or by triggering a balance of payments crisis and resulting in an exchange rate depreciation, as in Liviatan and Pierman (1986). Another source of inflation may be due macroeconomic overheating—that is, an excessive expansion of aggregate demand over potential output supply—as examined by Coe and McDermott (1997) for 13 Asian economies, estimated by the influence of an activity variable such as the output gap. A third source of inflation, examined by Ball and Mankiw (1995) is supply-side ‘cost shocks’—movements in the prices of particular goods, such as oil and food prices, that lead to persistent changes in the aggregate price level. Finally, as discussed by Chopra (1985), inflation may have a substantial inertial component arising from the sluggish adjustment of inflationary expectations or the existence of staggered wage contracts (see Calvo, 1983; and Christiano, Eichenbaum, and Evans, 2001). The specification encompasses the more traditional Phillips curve explanations of price changes.29
Data
29. Monthly headline and core consumer price inflation (CPI) data are available for 1990 on from the Uganda Bureau of Statistics, with substantial quality improvements introduced in 1997.30 Monthly UK Brent oil prices and New York Uganda coffee prices published by the IMF’s World Economic Outlook (WEO) are used for international supply shocks. Monthly exchange rates, monetary aggregates, and interest rates are from the monetary survey and treasury bill auctions released by the BOU. Demand conditions are identified by using deviations of the industrial production index from its potential using a HP filter.
30. An analysis of the time-series properties of the variables reveals that the variables are integrated of order one or I(1), except for industrial production and exports, which are trend stationary. Therefore, the I(1) series are differenced for estimation purposes to avoid the spurious and inconsistent regression problem (Hendry 1 995).31 Results of Granger causality tests are somewhat inconclusive, showing little evidence of Granger causality in either direction, but lend support for a transmission of external and domestic shocks to consumer prices, particularly exchange rate and broad money to core inflation.
Results
31. Inflation is driven by its own innovations, as well as by monetary and exchange rate shocks.32 Variance decompositions, which break down the variation in the endogenous variables down to the component shocks in the VAR, are reported below over 12 months. Generally, variances of all variables are largely explained by their own innovations at all time horizons. Exchange rate changes and money growth each explain about 10–15 percent of the variation in core inflation. The same holds true for headline inflation, although less significantly so. Broader monetary aggregates (M2 and M3) seem to explain a larger share of the variation of core inflation than reserve money, while interest rates and NDA have an insignificant effect. Impulse response functions trace out the effects of orthogonal shocks to international supply conditions, exchange rates, and monetary aggregates on prices through the dynamic structure of the VAR (see below). Again, own innovations have a significant effect on core and headline inflation, while exchange rate shocks, which have the second largest effect, persist for about three to six periods followed by monetary shocks. Impulse responses indicate that demand shocks may have a negative influence on inflation, but the variance decompositions suggest a discounting of their importance. A striking feature is the very low persistence of own innovations and other shocks on consumer prices.
32. The exchange rate pass-through to headline inflation is greater than to core inflation.33 The cumulative impulse response to an exchange rate shock of one standard deviation (or 2.4 percent) gives an exchange rate pass-through elasticity (percent change in price level t periods after the shock over an initial percent change in the exchange rate) of 0.24 for headline inflation and 0.17 for core inflation after six months. The less than one-to-one exchange rate pass-through may be explained by the degree of openness and strategic behavior in the form of pricing-to-market under imperfect competition.
33. Overall, monetary aggregates have an identifiable transmission mechanism on consumer prices while interest rates do not. For example, a 1 percentage point increase in M2 leads to a 0.2 percent rise in core inflation in three months. This supports the continuation of BOU’s monetary targeting framework as a means of controlling inflation. Therefore, the BOU’s problem has become to determine how much (if any) of the large aid-induced domestic liquidity injections need to be neutralized, and how this is to be achieved. The presence of a relatively stable money demand function in Uganda (see Hensridge, 1999; and Nachega, 2001) provides a useful framework to determine the level of monetary growth that is appropriate to achieve a given inflation target, therefore, the next section addresses the yet uninvestigated instrument mix to sterilize excess liquidity.34
Variance Decomposition of Core Inflation
Citation: IMF Staff Country Reports 2005, 172; 10.5089/9781451838749.002.A002
Accumulated Response to Cholesky One S.D. Innovations
Citation: IMF Staff Country Reports 2005, 172; 10.5089/9781451838749.002.A002
C. The Mix of Instruments for Sterilization
34. The mix of tools for sterilizing liquidity could affect the real economy in the short term. As mentioned earlier, the BOU’s sterilization operations rest on two pillars: net treasury bill issuance and foreign exchange sales; therefore, this section provides a preliminary investigation of the optimal mix of sterilization instruments in the short term based on the differential impact of exchange rate and interest rate changes on the industrial production index and exports, given the lack of high-frequency GDP data.
35. The time-series properties of exports and industrial production provide a simple methodology to estimate the marginal impact of interest rate and exchange rate changes. The de-trended series of exports and industrial production are explained by differenced series of interest rates, exchange rates, and other potential determinants in a simple regression framework.35 The presence of widespread credit rationing in low-income countries, endogeneity problems, and dynamic effects are also considered.
36. The impact of interest rate and exchange rate changes on industrial production is significant. Real interest rate and exchange rate changes have a statistically significant impact on industrial production, with a higher interest rate negatively affecting growth as expected, but a real exchange rate appreciation supporting a manufacturing expansion. The results are robust, controlling for the impact of domestic credit, dynamic effects, and potential endogeneity problems. The counter-intuitive positive impact of an appreciation is probably explained by Uganda’s heavy reliance on intermediate imports by the manufacturing sector, although the absence of data on intermediate imports precludes a formal test. Moreover, the impact of interest rates is over and above its potential impact by affecting the availability of credit, which is also statistically significant.
37. Export performance is negatively affected by currency appreciation, but credit availability is also important.36 Short-term export fluctuations are significantly hurt by a real exchange rate appreciation as expected, while interest rates do not show a statistically significant effect, but have a negative sign. However, credit availability is a significant determinant of export growth in the short term, with an elasticity similar to that of real exchange rate changes. This result cautions against a narrow focus on real exchange rate dynamics in explaining export performance.
38. A simple and preliminary exercise to determine the optimal mix suggests the BOU should mop up a larger share of excess liquidity through foreign exchange sales rather than net treasury bill issuance. Assigning equal weights to the export sector and industrial production sector, an arbitrary assumption suggests that less than 50 percent of excess liquidity should be sterilized by net treasury-bill issuance, mainly driven by the compensatory effects of an exchange rate appreciation on industrial production. Obviously, the mix would be skewed even further toward greater foreign exchange sales if we assumed a close relationship between interest rates and credit growth, which could not be precisely estimated for Uganda. The significant impact of both interest rate and exchange rate changes on private sector activity does not support a corner solution of only relying on one instrument for sterilization purposes.
Dependent Variable: De-trended Real Industrial Production Index
Monthly data from June 1993-June 2004)
Dependent Variable: De-trended Real Industrial Production Index
Monthly data from June 1993-June 2004)
| Baseline | With Domestic Credit to the Private Sector |
With Lag Dependent Variable |
All Regressors in Lags |
|
|---|---|---|---|---|
| Constant | 0.01024 | 0.04476 | 0.01388 | 0.02991 |
| (t-statistic) | 0.43930 | 2.54354 | 0.82998 | 1.80445 |
| p-value | 0.66120 | 0.01220 | 0.40820 | 0.07360 |
| ∆ real private credit | 0.67194 | 0.28049 | 0.34527 | |
| (t-statistic) | 8.28309 | 3.21361 | 4.04206 | |
| p-value | 0.00000 | 0.00170 | 0.00010 | |
| ∆ real exchange rate | 0.88228 | 0.65984 | 0.25454 | 0.23341 |
| (t-statistic) | 8.24634 | 7.15062 | 2.72050 | 2.58497 |
| p-value | 0.00000 | 0.00000 | 0.00750 | 0.01090 |
| ∆ real interest rate | -0.0099 | -0.00406 | -0.00117 | -0.00255 |
| (t-statistic) | -2.50268 | -2.74835 | -1.8790 | -2.1781 |
| p-value | 0.00616 | 0.00690 | 0.0450 | 0.0215 |
| De-trended Real industrial production index (-1) | 0.56878 | 0.54836 | ||
| (t-statistic) | 7.57747 | 7.47957 | ||
| p-value | 0.00000 | 0.00000 | ||
| R-squared | 0.51138 | 0.67126 | 0.77599 | 0.77722 |
| Adjusted R-squared | 0.49965 | 0.66337 | 0.76688 | 0.76809 |
| Durbin-Watson stat | 0.65709 | 1.04380 | 2.22851 | 2.11098 |
| F-statistic | 43.60687 | 85.07838 | 85.21653 | 85.12329 |
| Prob(F-statistic) | 0.00000 | 0.00000 | 0.00000 | 0.00000 |
Dependent Variable: De-trended Real Industrial Production Index
Monthly data from June 1993-June 2004)
| Baseline | With Domestic Credit to the Private Sector |
With Lag Dependent Variable |
All Regressors in Lags |
|
|---|---|---|---|---|
| Constant | 0.01024 | 0.04476 | 0.01388 | 0.02991 |
| (t-statistic) | 0.43930 | 2.54354 | 0.82998 | 1.80445 |
| p-value | 0.66120 | 0.01220 | 0.40820 | 0.07360 |
| ∆ real private credit | 0.67194 | 0.28049 | 0.34527 | |
| (t-statistic) | 8.28309 | 3.21361 | 4.04206 | |
| p-value | 0.00000 | 0.00170 | 0.00010 | |
| ∆ real exchange rate | 0.88228 | 0.65984 | 0.25454 | 0.23341 |
| (t-statistic) | 8.24634 | 7.15062 | 2.72050 | 2.58497 |
| p-value | 0.00000 | 0.00000 | 0.00750 | 0.01090 |
| ∆ real interest rate | -0.0099 | -0.00406 | -0.00117 | -0.00255 |
| (t-statistic) | -2.50268 | -2.74835 | -1.8790 | -2.1781 |
| p-value | 0.00616 | 0.00690 | 0.0450 | 0.0215 |
| De-trended Real industrial production index (-1) | 0.56878 | 0.54836 | ||
| (t-statistic) | 7.57747 | 7.47957 | ||
| p-value | 0.00000 | 0.00000 | ||
| R-squared | 0.51138 | 0.67126 | 0.77599 | 0.77722 |
| Adjusted R-squared | 0.49965 | 0.66337 | 0.76688 | 0.76809 |
| Durbin-Watson stat | 0.65709 | 1.04380 | 2.22851 | 2.11098 |
| F-statistic | 43.60687 | 85.07838 | 85.21653 | 85.12329 |
| Prob(F-statistic) | 0.00000 | 0.00000 | 0.00000 | 0.00000 |
Dependent Variable: De-trended Real Exports
(monthly data fro June 1993-June 2004)
Dependent Variable: De-trended Real Exports
(monthly data fro June 1993-June 2004)
| Baseline | With Domestic Credit to the Private Sector |
With Lag Dependent Variable |
All Regressors in Lags |
|
|---|---|---|---|---|
| Constant | 0.0374 | 0.0349 | 0.04985 | 0.0347 |
| (t-statistic) | 2.3890 | 3.3905 | 3.3094 | 3.3485 |
| p-value | 0.0008 | 0.000 | 0.000 | 0.000 |
| ∆ real private credit | 0.45890 | 0.3944 | 0.31945 | |
| (t-statistic) | 2.8459 | 2.9569 | 2.90634 | |
| p-value | 0.00000 | 0.00170 | 0.00010 | |
| ∆ real exchange rate | -0.9435 | -0.5043 | -0.4509 | -0.3938 |
| (t-statistic) | 2.349 | 2.2348 | 2.032 | 1.8837 |
| p-value | 0.0038 | 0.0049 | 0.010 | 0.039 |
| ∆ real interest rate | -0.0054 | -0.0045 | -0.0040 | -0.00347 |
| (t-statistic) | -0.8943 | -0.7459 | -0.5484 | -0.4589 |
| p-value | 0.3159 | 0.4480 | 0.5896 | 0.7548 |
| De-trended Real | ||||
| Exports (-1) | 0.3905 | 0.4580 | ||
| (t-statistic) | 4.4385 | 3.9054 | ||
| p-value | 0.00000 | 0.00000 | ||
| R-squared | 0.37138 | 0.4339 | 0.4659 | 0.4223 |
| Adjusted R-squared | 0.35403 | 0.41564 | 0.4365 | 0.39359 |
| Durbin-Watson stat | 1.4589 | 1.3935 | 1.9230 | 2.5490 |
| F-statistic | 22.8539 | 26.4395 | 31.5360 | 29.4959 |
| Prob(F-statistic) | 0.00000 | 0.00000 | 0.00000 | 0.00000 |
Dependent Variable: De-trended Real Exports
(monthly data fro June 1993-June 2004)
| Baseline | With Domestic Credit to the Private Sector |
With Lag Dependent Variable |
All Regressors in Lags |
|
|---|---|---|---|---|
| Constant | 0.0374 | 0.0349 | 0.04985 | 0.0347 |
| (t-statistic) | 2.3890 | 3.3905 | 3.3094 | 3.3485 |
| p-value | 0.0008 | 0.000 | 0.000 | 0.000 |
| ∆ real private credit | 0.45890 | 0.3944 | 0.31945 | |
| (t-statistic) | 2.8459 | 2.9569 | 2.90634 | |
| p-value | 0.00000 | 0.00170 | 0.00010 | |
| ∆ real exchange rate | -0.9435 | -0.5043 | -0.4509 | -0.3938 |
| (t-statistic) | 2.349 | 2.2348 | 2.032 | 1.8837 |
| p-value | 0.0038 | 0.0049 | 0.010 | 0.039 |
| ∆ real interest rate | -0.0054 | -0.0045 | -0.0040 | -0.00347 |
| (t-statistic) | -0.8943 | -0.7459 | -0.5484 | -0.4589 |
| p-value | 0.3159 | 0.4480 | 0.5896 | 0.7548 |
| De-trended Real | ||||
| Exports (-1) | 0.3905 | 0.4580 | ||
| (t-statistic) | 4.4385 | 3.9054 | ||
| p-value | 0.00000 | 0.00000 | ||
| R-squared | 0.37138 | 0.4339 | 0.4659 | 0.4223 |
| Adjusted R-squared | 0.35403 | 0.41564 | 0.4365 | 0.39359 |
| Durbin-Watson stat | 1.4589 | 1.3935 | 1.9230 | 2.5490 |
| F-statistic | 22.8539 | 26.4395 | 31.5360 | 29.4959 |
| Prob(F-statistic) | 0.00000 | 0.00000 | 0.00000 | 0.00000 |
D. The Scope for Inflation Targeting
The preconditions for inflation targeting in developing countries proposed in the literature are not fully met, although Uganda is quite advanced compared with most other LICs and could plausibly inch toward such a framework in the medium term. The main prerequisites for adopting an IT framework in developing countries, as identified by Masson and others, (1997), are as follows: a degree of monetary policy independence, in particular, freedom from fiscal dominance and no commitment to a particular level of or path of the nominal exchange rate (or any other nominal anchor, such as wages); agreement on a well-defined framework containing an explicit target for future inflation and a commitment to target that rate as an overriding objective; the presence of a model for forecasting future inflation; and an operating procedure for adjusting monetary instruments in case the forecast inflation differs from its target. Uganda has largely overcome concerns of fiscal dominance and “fear of floating, ” although donor dependence is a concern. It also has a well-articulated inflation target of maintaining underlying inflation below 5 percent in the medium term, which has many of the characteristics of a suitable target (see Svensson, 2000), although more research is needed whether it is an ‘optimal’ target. However, one of the major shortcomings is the presence of potentially conflicting objectives in the Central Bank Act, 1993, including those of supporting socioeconomic development and lack of formal central bank independence with no binding limits on government recourse to central bank borrowing, although the BOU enjoys a significant degree of independence over the use if its instruments. Models for predicting inflation are present but are not fully specified or as precise as highlighted in the paper. The greatest weakness is probably the absence of a procedure, or rule, for adjusting monetary instruments systematically in case the forecast inflation (or any other objective) differs from its target in the presence of numerous shocks. Overall, Uganda would need to make some progress on all these fronts before contemplating a switch to an IT framework.
E. Conclusions
The BOU should continue to operate within a monetary targeting framework at present, while a preliminary investigation of the optimal mix of sterilization instruments in the short term suggests a fair mix of both net treasury bill issuance and foreign exchange sales, unless there is an much greater weight on the export sector than on industrial (or manufacturing) production. The BOU could potentially move toward an IT framework in the long term, but would need to make progress in a number of areas and conduct further research, including on the monetary transmission mechanism as the economy becomes more sophisticated. A very preliminary, simplistic exercise to determine the optimal mix suggests that more than 50 percent of excess liquidity should be sterilized through foreign exchange sales, but the tentative conclusion requires further research. It is important to bear in mind that the optimal mix of sterilization instruments is a short-run consideration, and it is necessary to closely monitor the level of real exchange rate misalignment to facilitate the requisite current account adjustment that is related to the large donor inflows. Overall, there is a need for more research into operationalizing quantitative frameworks that would maximize the objective function of the central bank in countries like Uganda, which are subject to large donor flows and numerous exogenous shocks, given the depth of the financial system and macroeconomic linkages.
References
Adam, C. and others, 2004, “Exchange Rate Policy and the Management of Official and Private Capital Flows in Africa,” IMF Working Paper 04/216 (Washington: International Monetary Fund).
Adam, C., 2001, “Uganda: Exchange Rate Management, Monetary Policy, and Aid,” unpublished BOU and EPRC, Kampala.
Agenor, P. and Montiel, 1999. “Development Macroeconomic,” 2nd Edition Princeton.
Ball, L. and Mankiw, G., 1995, “Relative Price Changes as Aggregate Supply Shocks,” Quarterly Journal of Economics, pp.161-194.
Calvo, G. A., 1983, “Staggered prices in a utility maximizing framework,” Journal of Monetary Economics, 12, 383-398.
Chopra, A., 1985, “The Speed of Adjustment of the Inflation Rate in Developing Countries: A Study of Inertia,” IMF Staff Papers, Vol. 32, No 4.
Christiano, L., Eichenbaum, M. & Evans, C., 2001, “Nominal Rigidities and the Dynamic Effects of a Shock to Monetary Policy,” Working Paper 0107, Federal Reserve Bank of Cleveland
Coe, D. and McDermott, C., 1997, “Does the Gap Model Work in Asia?” IMF Staff Papers, Vol. 44, No 1.
Gali, J. and Gertler, M., 1999, “Inflation Dynamics: A Structural Econometric Analysis,” Journal of Monetary Economics, Vol. 44, pp. 195-222.
Hendry, D., 1995, “Dynamic Econometrics,” Oxford University Press.
Henstridge, N., 1999, “De-monetization, Inflation, and Coffee: the Demand for Money in Uganda,” Journal of African Economies, Vol. 8.
Sharer R., De Zoysa, H., and McDonald, C., 1996, “Uganda: Adjustment with Growth,” Occasional Paper No. 121 (Washington: International Monetary Fund).
Liviatan, N. and Piterman, S., 1986, “Acceleration of Inflation and Balance of Payments Crises, 1973‒84,” in The Israeli Economy, ed. By Yoram Ben-Porath, Harvard.
Loungani, P. and Swagel, P., 2001, “Sources of Inflation in Developing Countries,” IMF Working Paper 01/198 (Washington: International Monetary Fund).
Masson, P. and others, (1997), “The Scope for Inflation Targeting in Developing Countries,” IMF Working Paper 97/130 (Washington: International Monetary Fund).
McCarthy, J., 2000, “Pass-Through of Exchange Rates and Import Prices to Domestic Inflation in Some Industrialized Countries” (unpublished), Federal Reserve Bank of New York.
Nachega, J., 2001, “Financial Liberalization, Money Demand, and Inflation in Uganda,” IMF Working Paper 01/118 (Washington: International Monetary Fund).
Sargent and Wallance, 1981, “Some Unpleasant Monetarist Arithmetic,” Federal Reserve Bank of Minneapolis, Quarterly Review, Vol. 5, pp. 1-17.
Svensson, L. (2000), “Open-Economy Inflation Targeting,” Journal of International Economics, Vol. 50.
Prepared by Shanaka J. Peiris
Alternatively, it could peg its exchange rate to another country’s currency, but this option is not pursued here since Uganda’s floating exchange rate regime is deemed to have served it well and a regime switch is not under consideration.
Moreover, the Central Bank would require a liquid market in the target instrument that is not dominated by its own operations, which is not the case in Uganda given the limited activity of the interbank and other securities markets.
External grants and loans, which are credited to the BOU’s foreign correspondent bank accounts, have no domestic liquidity impact until they are drawn down and spent domestically by the government. A lower fiscal deficit due to higher taxes or lower expenditures due to less external financing would reduce the amount of liquidity injected.
In addition, the BOU also uses repos for fine-tuning ‘temporary’ liquidity variations, although there have been a greater reliance repos for sterilizing structural liquidity as of late.
Sales of foreign exchange could be used as a tool of sterilization only to a point until the foreign exchange reserve target becomes binding, which is not presently the case in Uganda.
This is consistent with the strategy suggested by Adam and others (2004).
A structural VAR model was not attempted given the purpose of the paper and the lack of a rich array of reliable high-frequency macroeconomic data (e.g., productivity).
The Cholesky decomposition imposes the correct number of restrictions for just identification and imposes a recursive structure on the system; so that the most endogenous variable is ordered last, i.e., it is affected by all contemporaneous ‘structural’ shocks. The results of the VAR thus could be highly susceptible to the ordering chosen.
The system does not, however, encompass New Keynesian Phillips curve specifications, which are based on firm microfoundations, and is a weakness in the approach taken. See Gali and Gertler (1999) for such a framework.
Food and foodstuff have a 27.4 percent weight on the CPI basket, based on the living standard survey of 1997/98.
However, first differencing may lead to the loss of information on the long-run relationships between the variables, which is a weakness in the approach.
The VAR is estimated from June 1993 to June 2004 on a monthly basis in log differences, with the lag length determined by Akaike information criteria.
This may be explained by the greater share of nontradable goods in core inflation.
An estimation of a money demand function with more recent data also suggested a money demand function with an income elasticity of unity and the presence of a clearly identifiable single structural break.
Exchange rate and interest rate volatility are also often quoted as potential determinants of economic performance in the short term, but the inclusion of ‘volatility’ indicators such as standard deviations and forecasts of Generalized Autoregressive Conditional Heteroskedastic (GARCH, 1,1) models did not show significant effects, and thus are excluded from the empirical framework. In this sense, the results for Uganda differ from Adam (2001).
The contrasting determinants of industrial production and exports are compatible given the very small share of manufacturing in total exports.
