Book
Share
Chapter

2 Inflation Targeting in Uganda

Author(s):
Andrew Berg, and Rafael Portillo
Published Date:
April 2018
Share
  • ShareShare
Show Summary Details
Author(s)
Martin Brownbridge and Louis Kasekende 

1 Introduction

The Bank of Uganda (BOU) introduced its inflation targeting (IT) monetary policy framework in July 2011, replacing a monetary targeting (MT) framework which had been in operation for two decades. This chapter reviews Uganda’s experience with IT, examines the pertinent lessons that have been learned so far, and discusses the technical and institutional challenges of implementing the framework.

The chapter is organized as follows. Section 2 discusses the reasons why the BOU replaced its MT framework with IT. It discusses what the BOU regarded as the essential prerequisites for the introduction of IT and also outlines the main features of the IT framework and the differences with MT. Section 3 examines how the introduction of IT has required fundamental changes in relationship between fiscal policy and monetary policy and the challenges that these changes have entailed. Section 4 discusses the formation of monetary policy in the IT framework, including the challenges of forecasting inflation and estimating the output gap. Section 5 discusses the implementation of monetary policy, in particular the evolving methods the BOU has used to achieve its operating target of aligning interbank interest rates with the policy interest rate. Exchange rate policy and how this has impacted on monetary policy is discussed in section 6. The BOU’s communications strategy is examined in section 7. Section 8 evaluates the outcomes in terms of the performance of the IT framework, including assessing the evidence for the strength of the monetary transmission mechanism and the extent to which the BOU’s primary policy target has been achieved. Section 9 offers some conclusions and lessons from Uganda’s experience with IT.

2 Motives for the Introduction, and Salient Features, of it in Uganda

The BOU’s primary motive for introducing an IT framework was the recognition that the monetary targeting framework, which had been used by the BOU since the early 1990s, was becoming obsolete. In particular, its efficacy was being undermined because the development of the financial sector and its increasing integration with the regional and global economy was making both the money demand function and the money multiplier more unstable and unpredictable. Between 1999/2000 and 2010/11, the velocity of circulation of domestic currency broad money (M2) declined, from 9.4 to 5.5, and the rate of decline was erratic (see Table 2.1). The annual change in the velocity in this period varied between positive 1 per cent and negative 14 per cent, which made setting intermediate monetary targets on the basis of a forecast of velocity difficult and, therefore increased the probability that errors in forecasts of demand for money would lead to sub-optimal monetary targets. The broad money multiplier was also quite volatile in the last three years of this period, when it rose from 2.2 in 2007/08 to 2.6 in 2010/11 (Table 2.1). An unstable money multiplier complicates the task of achieving broad money targets through control of base (reserve) money. Because an IT framework uses a policy interest rate rather than the monetary base as the operating target, autonomous shifts in the velocity of money or the money multiplier do not automatically alter the stance of monetary policy.

Table 2.1.Velocity of Broad Money (M2) and M2 money multiplier, 1999/2000 to 2010/11
1999/20002000/20012001/20022002/20032003/20042004/20052005/20062006/20072007/20082008/20092009/20102010/2011
M2 velocity change in velocity9.49.1

-2%
8.3

-10%
7.7

-7%
7.7

0%
7.8

1%
7.6

-3%
7.5

-2%
7.1

-6%
6.8

-4%
6.4

-5%
5.5

-14%
M2 money multiplier2.22.22.12.42.22.22.32.22.22.42.52.6
Source: BOU.
Source: BOU.

The IT framework offered two further advantages over MT. First, it provided a mechanism for clearly signalling the stance of monetary policy to the public, through the public announcement of a policy interest rate. Second, it offered the BOU much greater scope for short-term fine tuning of monetary policy, through adjustments to the policy interest rate, in response to macroeconomic shocks.

Before deciding to replace its MT framework with IT, the BOU considered whether Uganda was ready to implement the latter. At around the turn of the millennium, when several emerging market economies were adopting IT frame works, it was believed that the successful adoption of IT was dependent upon a number of demanding institutional and technical preconditions being met; for example deep and efficient financial markets to facilitate the transmission of monetary policy and sophisticated technical capacities in the central bank to forecast inflation and simulate the impact of policy changes on target variables (e.g. IMF, 2004; Masson et al., 1997). Subsequent research, much of it conducted by the IMF Research Department, suggested that the actual preconditions for a successful introduction of IT could be more narrowly defined. Freedman and Otker-Robe (2010) identified three essential preconditions: i) the central bank’s primary policy objective must be the control of inflation; ii) there must be no fiscal dominance; and iii) the central bank must have instrument independence. These are essentially institutional preconditions, which the BOU believed could be met in Uganda. Controlling inflation, with a publicly announced target for core or underlying inflation, had been the BOU’s primary objective of monetary policy since the 1990s. The 1995 Constitution of Uganda conferred operational independence on the central bank.1 The issue of fiscal dominance was not regarded as a serious threat to the success of an IT framework, given the Government’s good track record of fiscal discipline since the early 1990s, but this issue is complex, and is examined in more detail in Section 3. Freedman and Otker-Robe also argued that the adoption of IT stimulates the central bank to improve its technical capacities and implement reforms to develop financial markets.

Table 2.2.Key Features of the Monetary Targeting and IT Frameworks in Uganda
MT frameworkIT framework
Primary policy objectiveInflationInflation
Secondary policy objectiveOutput
InstrumentsPrimary securities auctionsSecondary market operations
Operating targetReserve moneyShort-term interest rate
Intermediate targetBroad moneyInflation forecast
Frequency of adjustments to policy stanceUsually annuallyInitially monthly, now bi-monthly
CommunicationsMinimalIntegral

Table 2.2 delineates the main features of the IT framework in Uganda and compares them with the MT framework which it replaced. The primary policy objective, controlling core inflation, was unchanged with the introduction of IT, as was the numerical target: 5 per cent for annual core inflation, although with the introduction of IT the BOU explicitly stated that its objective is to achieve the 5 per cent target on average over the medium term.2 With the introduction of IT, the BOU added a secondary objective, stabilizing real output as close as possible to estimated potential output. Under the MT framework, real output had been viewed as being mainly exogenous to demand management policies, and instead determined on the supply side of the economy. Developments in the economy during the 2000s, not least the impact of the global financial crisis, suggested that the demand side shocks could also affect real output, although supply-side shocks were clearly still an important source of output volatility.

The operating target of monetary policy in the IT framework is a short-term money market interest rate: the seven-day interbank interest rate. In the MT framework, reserve money was the operating target. The MT framework utilized an explicit intermediate target, broad money. An intermediate target does not play such a critical role in the IT framework, because it is much less rules-based than monetary targeting, although the inflation forecast can be viewed as an intermediate target in the IT framework. A key difference between the MT and IT frameworks is that while in the former, monetary policy is determined by current conditions, notably an estimate of current demand for a monetary aggregate, the latter is forward-looking in that monetary policy is determined by a forecast of future inflation, based on the premise that there are lags between a change in the policy interest rate and its full impact on inflation.

Communications with the public play an integral role in the IT framework, because of the importance of the central bank being able to influence inflationary expectations and convince the public of the credibility of monetary policy. As such, the BOU radically overhauled its communications policy to support IT, with the centrepiece being the press briefing given by the Governor after each Monetary Policy Committee meeting, at which the interest rate decision is announced and explained in a Monetary Policy Statement.

3 The Separation of Fiscal and Monetary Policy and the Issue of Fiscal Dominance

The condition of no fiscal dominance was interpreted by the BOU as entailing two conditions. First, that the central bank is not forced to finance the government’s domestic borrowing requirement, and second, that government borrowing is sustainable, so that the market will not take the view that eventually the government debt will have to be monetized (à la Sargent and Wallace, 1981).

The second condition is clearly applicable to Uganda. Although public debt has risen since 2009, it is still relatively low and far below the thresholds at which it would be considered to be unsustainable, as shown in Table 2.3. As such, the likelihood that a government default would force the BOU to refinance public debt looks remote.

Table 2.3.Gross Nominal Public Debt, Per cent of GDP: 2009–16
YearExternalDomesticTotal
200910.36.116.5
2010136.719.7
2011167.923.9
201213.58.922.4
201315.41025.4
201415.811.527.3
201518.112.530.7
201620.713.634.3
Source: BOU.
Source: BOU.

However, the fact that public debt is sustainable is not sufficient by itself to prevent central bank financing of the government’s domestic borrowing requirement: there needs to be both political commitment and institutional safeguards to prevent this from happening. Unfortunately, the institutional arrangements for fiscal and monetary policy under the MT framework, while adequate for that framework, could not meet the requirements of an IT framework.

Under the operational arrangement in place during the MT period, fiscal and monetary policy was intertwined. The government was subject to ceilings on its net domestic financing (under the IMF-supported programmes), but it did not explicitly issue securities to fund its domestic financing requirement, nor aim to avoid any borrowing from the central bank. Instead, the BOU issued government securities, through primary auctions, to mop up the liquidity needed to meet its reserve money targets. This meant that the distribution of the government’s domestic financing requirement, between the central bank and market participants, was determined passively as the outcome of the interaction of the needs of the reserve money programme and the magnitude of domestic financing. It also meant that it was impossible to distinguish where fiscal policy, in terms of the government’s financing needs, ended and where monetary policy began. As a consequence, monetary and fiscal policy both lacked transparency.

This was not a satisfactory arrangement for an IT framework. Two changes were needed to support the implementation of IT: to clearly separate the domestic financing of the budget from monetary policy and to avoid government borrowing from the central bank. The first reform has been accomplished. Since the start of the 2012/13 fiscal year, the primary issues of government securities (which take place in three out of every four weeks) are only used for mobilizing finance for the budget. The Ministry of Finance, Planning and Economic Development (MFPED) determines the size of these issues based on the domestic financing requirements of the budget, and the annual amount of net issues of securities is announced in the budget.3 Monetary policy is now conducted on the secondary market, through conducting repurchase or reverse repurchase operations and, more occasionally, through secondary market sales of the BOU’s own holdings of government securities.4 As a consequence, it is now possible for the market to distinguish clearly between fiscal and monetary policy operations. Separating monetary and fiscal policy operations in this manner has been a crucial reform in the implementation of the IT framework.

The issue of government borrowing from the BOU is more complex. Up until the time of the global financial crisis, the government had, in most fiscal years, saved money with the BOU (net financing from the BOU was negative). This was a passive outcome of the needs of the macroeconomic programme. Between 1995/96 and 2008/09, the government saved money with the BOU at an annual average rate of 1.1 per cent of GDP, while overall net domestic financing (which includes financing from the BOU) averaged negative 0.3 per cent of GDP. However, since the global financial crisis, the government’s domestic borrowing requirement has risen—between 2009/10 and 2015/16, net domestic financing averaged 2.0 per cent of GDP—and it is no longer axiomatic that it will make savings with the BOU.

Since the introduction of IT, the actual record of government financing from the BOU has been mixed, as shown in Table 2.4. For the purposes of monetary policy, the most pertinent measure of net financing for the budget is the change in the net government position with the BOU, excluding the project accounts and other dedicated accounts such as oil revenue accounts, which hold funds raised from specific sources, such as donor project funds, and are earmarked for specific expenditures.5 The BOU financed the government budget in three of the first five financial years of the implementation of the IT framework, and in two of these financial years the financing was more than 10 per cent of base money at the start of the year.

Table 2.4.Government Net Financing from the Bank of Uganda: Shs Billions, 2011/12–2015/16
2011/20122012/20132013/20142014/20152015/2016
Billions of Shillings323-2-293149466
Per cent of Money Base11.2-0.1-8.8410.3
The computation of net financing excludes transactions to and from dedicated accounts, such as the oil revenue and donor project accounts as well as transactions related to monetary policy operations. Net financing as a per cent of base money is calculated using base money at the start of the relevant financial year.Source: BOU.
The computation of net financing excludes transactions to and from dedicated accounts, such as the oil revenue and donor project accounts as well as transactions related to monetary policy operations. Net financing as a per cent of base money is calculated using base money at the start of the relevant financial year.Source: BOU.

Government borrowing from the central bank is potentially problematic for monetary policy for two reasons. The first is that it entails the creation of money. Under the IT framework the BOU has to use its secondary market instruments to implement its monetary policy, but if government borrowing is large and sustained, this leads to the creation of ‘structural liquidity’ which may require a different set of sterilization instruments and may be costly for the central bank, as discussed in Section 5. Second, central bank financing of the budget can undermine the BOU’s credibility to fight inflation.

Although the MFPED is committed in principle to avoid borrowing from the BOU, there are no practical institutional mechanisms which can prevent it. There are constraints on government borrowing from the central bank in both the BOU Act and the Public Financial Management Act, but neither act precisely defines government borrowing and there are differences of opinion within government as to whether borrowing should be defined in terms of a stock or a flow of resources through time; it is the latter which is most relevant for macroeconomic management. Government has introduced a Single Treasury Account System (STA) for its accounts in the central bank which might eventually provide a mechanism for curbing borrowing from the central bank. Ideally a floor should be placed on the balance in the STA and the MFPED should then manage its cash flows to ensure that this floor is not breached. Parliament, through the Parliamentary Committee on the Economy and Budget, should also play a more assertive role in monitoring the government borrowing from the BOU and holding it to account for deviations from budget plans.

4 Formulating Monetary Policy

The policy interest rate—the Central Bank Rate (CBR)—is set on a bi-monthly basis by the BOU’s Monetary Policy Committee (MPC).6 The BOU sets the rate based on two key considerations: a 12-month forecast of core inflation, plus an assessment of the risks to the forecast, and an estimate of the output gap. The inflation forecast takes priority because controlling core inflation is the BOU’s primary target. The technical challenges of formulating monetary policy are threefold. The first is making accurate inflation forecasts, the second is to estimate the output gap, and the third is to determine, if outturns are forecast to deviate from targets, how large a change in the policy interest rate is needed to achieve the targeted outturns.

Medium-term forecasts of inflation require a good understanding of the determinants of inflation, the capacity to project changes, over the medium term, in the exogenous drivers of inflation, and a robust forecasting model. When the IT framework was first introduced, an inflation forecast derived from a VAR model was used by the MPC. However, although a VAR model, which projects past trends forward, is useful for short-term forecasts, it is much less so for medium-term forecasts. This is partly because inflation is quite volatile in Uganda, with quite large cyclical swings from troughs to peaks over periods of from seven to twenty months. Consequently, current levels of inflation are a poor guide to future inflation beyond a few months’ time. A large part of the volatility in core inflation is caused by the impact on domestic prices of volatility in the exchange rate (discussed in Section 6). Econometric estimates of the long-run pass-through of the exchange rate to inflation in Uganda are around 0.5 (Bwire, Anjugo, and Opolot, 2013; Apaa Okello and Brownbridge, 2013). For the purposes of forecast ing inflation this presents a problem: one of the major variables driving inflation is very volatile and is itself very difficult to forecast.

In 2014 the Research Department of the BOU replaced the VAR model with a semi-structural quarterly projection model (QPM). The QPM, which is a dynamic stochastic general equilibrium model, incorporates forecasts of determinants of inflation such as the exchange rate, the fiscal deficit, and the output gap. It is embossed in a forecasting and policy analysis system (FPAS) that includes near term forecasting, data management, and other elements to shape the discussion, formulation, and communication of the policy stance.7 However, the effectiveness of this approach faces two challenges. The first is to make robust forecasts of variables such as the exchange rate. The second is to calibrate the model to accurately reflect the causal factors in the inflation process, including the relative magnitude and timing of these factors in driving inflation.

Table 2.5 provides details of the BOU’s core inflation forecasts and the out turns. In most cases the forecasts were 12-month forecasts, but occasionally the forecast period was slightly less or more than 12 months. The forecasts were included in the Monetary Policy Statements issued immediately after the MPC meetings. The outturn data are taken from the 2005/06 base year CPI series, because this was the series in use at the time when the forecasts were made. However, the 2005/06 base year series was discontinued in November 2015 and replaced with a 2009/10 base year series, hence the outturn data from December 2015 onwards are based on the latter series. Two points can be made about the core inflation forecasts. First, it has been difficult to forecast the turning points in the inflation cycle; for example, the BOU had not expected the upward swing of the inflation cycle in 2015–16 to peak until mid-2016, yet the actual peak occurred at the end of 2015. Second, the forecasts have underestimated the volatility of inflation, especially with respect to the downward swings of the cycle, when core inflation turned out to be lower than forecast (e.g. in October and November 2014). On average, the published forecasts shown in Table 2.5 were higher than the outturns by 1.4 percentage points.8

Table 2.5.Core Inflation Forecasts and Outturns
Date at which forecast was madeForecast (per cent)Outturn (per cent)Date to which forecast applies
Oct 137–82.4Oct 14
Nov 136.5–7.52.3Nov 14
Jan 146.5–7.52.7Late 2014
Mar 145.5–6.53.7Mar 15
Apr 146–74.6Apr 15
May 143–74.8May 14
Jun 145–64.9Jun 15
Aug 145.5–6.55.5Aug 15
Oct 1456.3Oct 15
Dec 1457.5Dec 15
Feb 154–66.7Feb 16
Apr 157–96.9Jun 16
Jun 158–106.9Jun 16
Jul 158–105.7Jul 16
Source: Forecasts are from the BOU’s Monetary Policy Statements; outturn data are from UBOS.
Source: Forecasts are from the BOU’s Monetary Policy Statements; outturn data are from UBOS.

Incorporating the output gap into the setting of the CBR is very problematic for several reasons. The first is the difficulty of making robust estimates of potential output. The BOU has estimated potential output by fitting a trend, using a Hodrick-Prescott filter, through the quarterly real GDP data. However, this method does not capture potential structural breaks in the time series data. Furthermore, because of the recent rebasing of GDP estimates by the Uganda Bureau of Statistics (UBOS), a fully consistent time series for GDP only extends back to 2008/09, which obviously undermines the efficacy of estimates of potential output derived from an H-P filter. In addition, there are often substantial revisions made to the quarterly estimates in subsequent quarters. Consequently, neither the estimates of actual quarterly output, nor the estimates of potential output, are very reliable. In December 2010, for example, six months before the BOU introduced the IT framework, the BOU had concluded that, although aggregate demand was increasing, the economy was still operating with a negative output gap, based on the annual GDP estimates for 2009/10, whereas subsequent revisions to the national accounts data showed that real GDP in 2010/11 was substantially above potential. The BOU also prepares its own composite indicator of economic activity on a monthly basis. This can provide indications on whether growth in real activity is accelerating or slowing down, but by itself it provides no information on whether there is a positive or a negative output gap.

In setting monetary policy, the BOU aims to follow the Taylor Principle,9 whereby the real interest rate is raised whenever inflation is forecast to rise above the policy target and vice versa.10 The risks to the inflation outlook are also taken into account. The factors driving forecast inflation also affect the policy response. The BOU reacts less aggressively to supply-side shocks to prices which are expected to be reversed; in such cases the BOU’s main concern is to prevent spill overs from the supply shocks which might prove more persistent; for example increases in school fees arising from food price shocks. When large changes in the CBR are warranted, as with the sharp rise in the inflation forecast between February and June 2015 shown in Table 2.5, the BOU spreads the required interest rate change out over several months to avoid too much disruption to financial markets.

5 Implementing Monetary Policy

The key task of implementing monetary policy is to align a short-term risk-free interest rate with the CBR so that it sets a benchmark for other interest rates in the economy. Interbank lending in Uganda mostly entails overnight and seven-day loans. In Uganda the seven-day interbank rate serves as the operating target for monetary policy.

Implementation of monetary policy takes place through regular interventions in the money market, through an offer to the commercial banks for either a repo (which removes liquidity) or a reverse repo. The repos/reverse repos are trans acted at the CBR, with the BOU accepting all offers from the banks which are consistent with the CBR (i.e. the BOU fixes the price of liquidity and allows the market to determine the quantity). In effect, if there is too much liquidity in the market, the BOU offers to pay the banks the CBR on their surplus reserves.

Until May 2012, the BOU had auctioned a fixed quantity of repos or reverse repos, allowing the market to determine the interest rate, although a ceiling was imposed on the rate for the reverse repo and a floor on that for the repo, with the ceilings and floors linked to the CBR. The switch to the current modalities has brought two advantages. First, it has ensured that the repo/reverse repo rate matches the CBR at every issue and second, it obviates the need for the BOU to make precise liquidity forecasts before issuing a repo or reverse repo. All that the BOU needs to know to prevent the seven-day interbank rate from deviating from the CBR is whether the banking system in aggregate will have surplus or deficit liquidity.

5.1 Challenges for the Implementation of Monetary Policy

Monetary policy implementation has faced a challenge from structural liquidity. Structural liquidity refers to long-term liquidity, rather than simply temporary fluctuations of liquidity. It is also an approximation of how much liquidity the BOU needs to remove from the banking system to align interbank interest rates with the CBR. The sources of structural liquidity are factors which lead to the creation of reserve money in excess of its demand (by the banks and the public), which are mainly the accumulation of foreign exchange reserves by the BOU and government borrowing from the BOU.

Figure 2.1 provides estimates of structural liquidity, in the 21 months from January 2015 to the beginning of October 2016. These estimates are derived by adding together the deviation of actual excess bank reserves over the average level of excess reserves for the whole period,11 plus the volume of liquidity which has been mopped up or injected by the BOU through the issuance of repos or reverse repos and the net sales of the BOU’s recapitalization securities on the secondary market. As can be seen in Figure 2.1, structural liquidity was always positive in this period, but it rose very sharply in the final quarter of the 2015/16 financial year, mainly because of the government borrowing from the BOU, shown in Table 2.4. A stagnation of demand for reserve money during 2016 has also contributed to the build-up of structural liquidity.

Figure 2.1.Structural Liquidity and Issuance of Monetary Policy Instruments to Mop it up, Shilling Billions: Jan 2015–October 2016

Where the bars depicting net repos show a negative number, as at 2 Feb 2015, the BOU had injected liquidity with reverse repos.

Source: BOU.

The problems emanating from structural liquidity are twofold. First, when the volume of structural liquidity is very large, it exceeds the amount of recapitalization securities which the BOU holds to mop up liquidity for longer than very short periods. As a consequence, the BOU has to use short-term repos with maturities of seven days or less to mop up liquidity which will remain in the banking system for much longer than seven days. Consequently, large volumes of repos mature each week, injecting liquidity back into the system, which can weaken the BOU’s control over liquidity conditions. Second, mopping up large volumes of liquidity is expensive for the BOU, as it must pay the interest costs of the securities it issues, or forgo the interest income.

6 Exchange Rate Policy and Monetary Policy

Since the 1990s Uganda has maintained a floating exchange rate, primarily motivated by the conviction that exchange rate flexibility helps to stabilize the real economy in the face of external shocks. However, the nominal exchange rate has often been very volatile in Uganda, as can be seen in Figure 2.2, driven by both current account and capital account shocks. The latter include fluctuations in short-term flows of portfolio capital which are invested in the domestic money and securities markets. The BOU is not indifferent to exchange rate volatility, for several reasons. Volatility causes disruption to economic agents needing to trans act in foreign exchange. There have also been extended periods when the real effective exchange rate has been overvalued, thereby undermining external competitiveness and long-run economic growth.12 Given that the current account of the balance of payments has widened in recent years, from 6.8 per cent of GDP in 2008/09 to 9.7 per cent of GDP in 2015/16, the BOU has been concerned to avoid a deterioration of competitiveness. Nominal exchange rate movements are also a major determinant of inflation, as discussed in Section 4.

Figure 2.2.Nominal and Real Effective Exchange Rate Indices (2009/10 = 100), Monthly Averages, January 2010–August 2016

Source: BOU.

The volatility of the exchange rate poses dilemmas for the central bank. It is too important for the economy to simply ignore. On the other hand, including the exchange rate among the targets for monetary policy (i.e. including the exchange rate among the variables in the reaction function for the policy interest rate) would compromise the BOU’s ability to deliver on its target for inflation and possibly also generate more volatility in output and inflation rates (Eichengreen, 2002). Instead the BOU has attempted to manage the volatility of the exchange rate, when necessary, through sterilized intervention, defined as intervention which leaves the target interest rate (the seven-day interbank rate) unchanged. The intervention takes place through BOU sales or purchases of foreign exchange to the interbank foreign exchange market (IFEM) while sterilization is implemented using the money market interventions described in Section 5. Sterilized intervention is effective in influencing the exchange rate in Uganda because capital mobility is far from perfect, given the small size of domestic financial markets. BOU interventions can be large relative to daily flows in the interbank foreign exchange market. In effect, the BOU has two instruments to address two targets, as discussed by Ostry et al. (2012) and in Chapter 13.

Sterilized intervention is motivated by two objectives. The first is to dampen short-term exchange rate volatility; i.e. sharp daily movements in the exchange rate, especially when the BOU believes that such movements are not driven solely by economic fundamentals. These interventions are usually quite successful in stemming rapid appreciation or depreciation but there have been infrequent occasions, when the exchange rate has been under very strong pressure to depreciate rapidly, when the BOU has not fully sterilized foreign exchange sales, thereby tightening liquidity; this was because the foreign exchange sales alone were not sufficient to stem the depreciating pressures. Obviously, in such circumstances, achieving the operating target (the seven-day bank rate) is temporarily subordinated to the need to restore stability to the exchange rate.

The second motive for intervention is to avert sustained real appreciation beyond what is estimated to be the long-term equilibrium real effective exchange rate. This is more problematic because, if the appreciation pressure is strong, it often requires repeated interventions, the sterilization of which contributes to the accumulation of structural liquidity and hence creates potential problems for the implementation of monetary policy, as discussed in Section 5. The BOU intervened in 2013 to stem real appreciation, but the build-up of structural liquidity constrained the scope for doing this.

The BOU has also purchased foreign exchange from the IFEM for purposes of accumulating foreign reserves. Given that it sells far more foreign exchange to the government (for external debt servicing, government imports, etc.) than it purchases from the government (e.g. from donor budget support inflows), if the BOU did not purchase foreign exchange from the IFEM, its foreign reserves would be steadily depleted. These purchases are not intended to influence the exchange rate or send any signal to the market about the BOU’s desires with respect to the exchange rate. Hence the BOU has carried out these purchases on a regular, predictable basis, buying a small amount each working day through an auction process, with the exact amount purchased each day dependent on the offers received from banks. These purchases are also sterilized, directly through the BOU’s money market interventions and indirectly by government debt issuance.13

7 Communications

The BOU radically revamped its communications strategy to support the introduction of the IT framework. The main objectives of the communications strategy are to enhance the signal of the policy stance and to influence inflationary expectations. In particular, it aims to make the public aware of the policy interest rate decision and the reasons which underlie it, especially the core inflation forecast.

The key component of the communications strategy is the press briefing which follows the MPC and at which the Governor reads out a monetary policy statement (MPS) and answers questions from journalists. The MPS is uploaded to the BOU’s website. The BOU also posts a monthly monetary policy report on the website, which includes material from the reports presented to the MPC. Finally, the BOU uses speeches by the Governor and Deputy Governor to address monetary policy issues and to explain how the IT framework works; these are also posted on the website. In drafting the MPS and the monetary policy reports, the BOU aims to be transparent and to avoid painting an unwarrantedly optimistic picture of future prospects. For example, the inflation forecast presented to the MPC is included in the MPS.

The communications strategy has been successful in strengthening the coverage of the BOU’s monetary policy in both the local and foreign media. All of the serious local newspapers and electronic media carry reports of the press briefing, often with quotes from the Governor. The local TV and radio stations also cover the press briefing and sometimes supplement this with interviews with BOU officials. Specialist international media, such as Reuters and Bloomberg, and more occasionally MSBC Africa, also carry reports on the interest rate decision. Some of the large international banks which operate in Africa, such as Standard Bank, provide to their clients regular market analysis on Uganda which makes reference to the MPS and other communications from the BOU. Hence it is possible to draw two conclusions which are positive for monetary policy. First, there is much greater awareness among the general public in Uganda of what the BOU is doing, because of the widespread media coverage. Nevertheless, the quality of macroeconomic analysis, even in the serious local media, is not very high. Second, among economic and financial sector specialists in the media and private sector, there is a greater understanding of the reasoning behind the BOU’s monetary policy actions; for example, where the BOU believes the main threats to inflation reside.

8 How Effective has Monetary Policy been in the IT Framework?

In this section we review the evidence on the effectiveness of the IT framework in Uganda in terms of the capacity of monetary policy to achieve its targets. This is essentially a discussion about the monetary policy transmission mechanism: i.e., do changes in the policy interest rate bring about the desired changes in macroeconomic variables, particularly inflation? The monetary policy transmission mechanism entails two stages. The first is an interest rate transmission, whereby a change in the policy interest rate affects market interest rates in the economy. The second stage entails changes in market interest rates affecting private sector behaviour and hence macroeconomic variables. We start by examining the interest rate transmission mechanism.

8.1 How Strong is the Interest Rate Transmission Mechanism?

The link between the CBR and the seven-day interbank rate is the first stage in the interest rate transmission mechanism. Between April 2012 and August 2016, the monthly average seven-day interbank rate was close to the CBR, deviating in absolute terms on average by only 52 basis points over this period.14 The months in which there were larger deviations of the average seven-day interbank rate from the CBR were mostly those in which the exchange rate was under strong pressure such that the BOU restricted liquidity to the banking system because of fears that this might fuel further depreciation, as was the case for some months in 2015.

The second stage in the interest rate transmission mechanism involves changes in the seven-day interbank rate affecting longer-term interest rates, notably time deposit rates and bank lending rates. Average time deposit rates, which are heavily influenced by a few wholesale depositors, have been slightly more volatile than seven-day interbank rates, but they have tracked the CBR quite closely (Figure 2.3). Between April 2012 and August 2016, the average absolute deviation between the average monthly time deposit rate and the CBR was 115 basis points.

Figure 2.3.CBR, Seven-Day Interbank Rate, 364-Day TB Rate, Average 7–12 Month Time Deposit Rate and Average Lending Rate; July 2011–August 2016

Source: BOU.

The BOU’s influence over bank lending rates is less than over deposit rates. Estimates by staff of the Research Department indicate that the bank lending rate responds to a 100 basis point change in the CBR by slightly less than 50 basis points (Sande and Apaa Okello, 2013). Abuka et al. (2015) find that a 100 basis points rise in the seven-day interbank rate is associated with a rise of between 33 and 49 basis points in the bank lending rate. There has also been an asymmetry in the speed of adjustment of the lending rate to the CBR. As can be seen in Figure 2.3, when the CBR was raised sharply in 2011, the bank lending rate also increased over roughly the same period, but when the CBR was reduced in 2012, the reduction in bank lending rates was drawn out over a much longer period of time.

8.2 Do Changes in Market Rates Affect Macroeconomic Variables?

Changes in market interest rates can affect macroeconomic variables through various channels—volumes of bank credit to the private sector, savings decisions by consumers, exchange rates, and private sector expectations of inflation and other variables. In practice, it is difficult to isolate and estimate the impact of these different channels. However, a few studies have been completed which examine the empirical evidence for the transmission mechanism since the introduction of the IT framework, which we briefly review below.15

Opolot (2013) uses balance sheet data from commercial banks to estimate the response of credit aggregates to changes in Treasury Bill interest rates, as a proxy for the stance of monetary policy, during the period 2000–2012 which straddles both the MT and IT frameworks in Uganda. He finds that a rise in interest rates reduces bank lending, although the impact is quite weak. He also finds that high levels of liquidity in banks dampen the bank lending channel. Abuka et al. (2015) use microeconomic data from the Credit Reference Bureau database to investigate how changes in bank lending rates affected volumes of loans disbursed by banks over the period from 2010 to 2014. They find that a rise in the banking lending rate reduces the probability that a bank will grant a loan to a borrower and the volume of credit granted to borrowers. They also find that the degree to which lending rates affect credit aggregates in individual banks depends on the strength of those banks’ capital and liquidity positions; higher capital ratios weaken the bank lending channel but higher liquidity ratios strengthen it. They find that the quantitative impact of a given change in lending rates in Uganda is only about half that estimated for advanced economies. Hence, there is evidence of a functioning bank lending channel in Uganda, albeit one which is weaker than that in advanced economies. This finding is consistent with theories of credit market imperfections which suggest that imperfect information about the quality of borrowers causes banks to ration credit rather than allowing the interest rate to clear the credit market.

Berg et al. (Chapter 5) employ a narrative or event study approach to evaluate the efficacy of the monetary policy transmission mechanism in East African countries, focusing on the monetary tightening triggered by the steep rise in inflation in these countries in 2011. They found evidence of an effective transmission mechanism in Uganda, which they attributed to the clarity and transparency of the monetary policy regime. In the episode that they studied, a policy-induced rise in interest rates led to an appreciation of the exchange rate, a fall in inflation, and a decline in output growth.

8.3 Inflation Outcomes under the IT Framework

Figure 2.4 depicts annual core inflation from July 2006 to September 2016. Unfortunately, there is no single consumer price series which spans this period. Instead there is a 2005/06 base year series which began in July 2006 and was discontinued in November 2015, and a second series, with a 2009/10 base year, which began in July 2011. IT was introduced when inflation was climbing very rapidly, as a result of a combination of rising global and regional food prices, steep nominal exchange rate depreciation (which had begun in early 2010, as shown in Figure 2.2), and very strong credit growth. After introducing IT in July 2011, the BOU raised the CBR by 10 percentage points to 23 per cent by October 2011 (Figure 2.3), with the results noted by Berg et al. in Chapter 5. Core inflation peaked in October–November and then fell back sharply, so that by September 2012 it was back to around 5 per cent.

Figure 2.4.Annual Core Inflation Rates: July 2006–October 2016 (two series)

Source: UBOS.

Table 2.6 compares core inflation outcomes under both the MT and IT frame works. We have excluded the period of high inflation from February 2011 to June 2012 from the analysis, because this period spanned both frameworks. Furthermore, given the lags in the monetary transmission mechanism, an inflation outcome which occurred during the first few months of the operation of the IT framework cannot fairly be attributed to that framework, and certainly not to it alone. Two notes of caution are warranted in comparing inflation outcomes under the two frameworks. First, we have had to use different CPI series for each framework, the 2005/06 series for the MT framework and the 2009/10 series for the IT framework, because the latter series does not cover any of the period in which the MT framework was operated by the BOU and the former covers only part of the period in which the BOU has been implementing the IT framework. Second, inflation outcomes are not purely the result of monetary policy but also reflect exogenous shocks which are not identical to both the periods of the MT and IT frameworks.

Table 2.6 shows that, since inflation was brought under control in mid-2012 (a period of 52 months), core inflation has averaged 5.1 per cent, very close to the BOU’s policy target of 5 per cent over the medium term. Core inflation rates were higher under the MT framework, averaging 8.2 per cent in the 55 months from July 2006 to January 2011, although part of the higher average inflation can be attributed to the global food price shocks and the impact of the global financial crisis on the exchange rate in 2008 and 2009. Core inflation has also been less volatile, with a lower standard deviation, since the IT framework was introduced.

Table 2.6.Average Annual Core Inflation and Standard Deviations of Core Inflation Rates under Monetary Targeting and Inflation Targeting Frameworks
Monetary TargetingInflation Targeting
July 2006–January 2011July 2012–October 2016
Annual Average core inflation (per cent)8.25.1
Standard deviation (percentage points)3.11.8
Source: BOU and UBOS.
Source: BOU and UBOS.

9 Conclusions

After more than five years of implementing IT in Uganda, what lessons can be learned from this experience? In this conclusion we ask three questions. How feasible is IT in a low-income economy? What are the basic prerequisites for the adoption of IT? What are the most important elements for the success of IT?

There are undoubtedly challenges to implementing an IT framework in a low-income economy; these include shallow financial markets, which impede the monetary policy transmission mechanism, volatile exchange rates, supply price shocks which make inflation more volatile and difficult to forecast, and a lack of reliable and timely data, especially on the real sector of the economy. Nevertheless, an IT framework can offer a central bank traction over the medium-term evolution of inflation, which is at least as effective, if not much more so, than the alternative of a MT framework. The evidence reviewed in the previous section suggests that the use of a policy interest rate as the operating target of monetary policy drives a functioning monetary policy transmission mechanism, through both the interest rate transmission from the policy rate to market interest rates and the subsequent impact on bank lending and nominal exchange rates, together with the public announcement of the policy interest rate as a signal of the monetary policy stance.

Uganda’s experience bears out the findings of Freedman and Otker-Robe (2010), that the essential preconditions for the adoption of IT are that the central bank must have operational independence to set monetary policy and that controlling core inflation must be the unambiguous primary target of monetary policy. Each of these factors has been important, both in shaping the BOU’s monetary policy and in establishing its credibility as a central bank with a strong commitment to control inflation, a commitment which has been noted in some of the commentaries by market participants, such as international banks. The absence of fiscal dominance is also important, but this is difficult to define precisely in practice. Ideally it should preclude borrowing by the government from the central bank except in small quantities and on a purely temporary basis, with effective institution mechanisms to enforce this. The latter are lacking in Uganda and BOU financing of the budget has not been negligible or purely temporary, as discussed in Section 3. This highlights the importance of a fourth precondition for the adoption of IT; the central bank must have sufficient instruments at its disposal, usually marketable government securities, to conduct monetary policy through open market operations. If government does borrow from the central bank, this implies that the latter will require a larger volume of instruments, ceteris paribus, with which to manage the liquidity created by government borrowing.

What are the elements which have contributed to the successful adoption of IT in Uganda? We would argue that a key factor is that the BOU has constructed the IT framework around a set of basic principles which have guided the development of operational procedures and the setting of policy. As a result monetary policy has clear, coherent, and transparent objectives which have contributed to better understanding of it by market participants and the public, and which in turn has an important influence on the effectiveness of monetary policy. These principles are discussed below.

First, the primary determinant of monetary policy is the inflation forecast, together with an assessment of risks to the forecasts. As such, the BOU has implemented monetary policy on a forward-looking basis, putting faith in its inflation forecasts, which is necessary given the lags in the transmission mechanism. Although it has proved difficult to make accurate inflation forecasts, given the volatility of inflation, the forecasts nevertheless provide the monetary policy committee with clear guidelines for setting the policy interest rate, based on the Taylor principle. The BOU has been prepared to raise the policy interest rates when inflation is forecast to rise, even if, at the time of the decision, actual inflation, as opposed to forecast inflation, was still close to target. This was probably important in curbing the rise in inflation in the second half of 2015, driven by a steep depreciation of the exchange rate. The BOU has also been prepared to implement quite aggressive changes in the monetary policy stance when inflation is forecast to rise, which is necessary because the monetary policy transmission mechanism is weaker than in advanced economies and hence larger changes in the policy instruments are needed to bring about a given change in the target variables.

Second, the BOU has disentangled monetary policy from fiscal policy, by clearly separating monetary operations carried out on the secondary market from the issuance of securities to finance of budget which are conducted in the primary auctions. This is important for two reasons. It has ensured that the BOU has full control over its monetary policy operations, and it has enhanced the transparency of monetary policy.

Third, the BOU has focussed the implementation of monetary policy, through the regular interventions in the money market, on the objection of aligning the seven-day interbank rate with the policy interest rate (i.e. achieving the operating target of monetary policy) rather than any other objective. It has not pursued multiple operating targets, such as trying to target both interest rates and monetary aggregates at the same time. Since introducing the IT framework, the BOU has been flexible in experimenting with different approaches to determine the most effective methods for achieving this objective, such as different strategies for issuing repos, but the operating target has not changed.

Finally, the BOU has prioritized its communications strategy, putting emphasis on explaining to the public the reasons for monetary policy decisions and making public its medium-term inflation forecasts. Communications is viewed as an essential complement to monetary policy decisions and actions, because of the important role which public understanding of monetary policy plays in enhancing its effectiveness in an IT framework.

References

    AbukaC.AlindaR. K.MinoiuC.PeydroJ. L. andPresbiteroA. F. (2015). Monetary Policy in a Developing Country: Loan Applications and Real Effects. Working Paper 15/270. Washington, DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation

    Apaa OkelloJ. andBrownbridgeM. (2013). Exchange rate pass through and its implications for monetary policy in Uganda. Working Paper 10/2013. Bank of Uganda.

    • Search Google Scholar
    • Export Citation

    BwireT.AnjugoF. L. andOpolotJ. (2013). A Re-estimation of exchange rate pass-through to domestic prices in Uganda: Evidence from a Structural Vector Auto-Regression (SVAR). Working Paper 05/2013. Bank of Uganda.

    • Search Google Scholar
    • Export Citation

    DavoodiH. R.DixitS. andPinterG. (2013). Monetary Transmission Mechanism in the East African Community: An Empirical Investigation. Working Paper 13/39. Washington, DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation

    EichengreenB. (2002). Can emerging markets float? Should they inflation target?Working Paper 36. Central Bank of Brazil.

    ElbadawiI.KaltaniL. andSotoR. (2012). Aid, Real Exchange Rate Misalignment and Economic Growth in Sub-Saharan Africa. World Development40 (4) 681700.

    • Search Google Scholar
    • Export Citation

    FreedmanC. andOtker-RobeI. (2010). Important Elements for Inflation Targeting for Emerging Economies. Working Paper 10/113. Washington, DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation

    GoncalvesC. E. (2015). Taylor visits Africa. Working Paper 15/258. Washington, DC: International Monetary Fund.

    International Monetary Fund. (2004). Monetary Policy Implementation at Different Stages of Market Development. Washington, DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation

    JohnsonS.OstryJ. D. andSubramanianA. (2007). The Prospects for Sustained Growth in Africa: Benchmarking the Constraints. Working Paper 52/07. Washington, DC: Inter national Monetary Fund.

    • Search Google Scholar
    • Export Citation

    MassonP. R.SavastanoM. A. andSharmaS. (1997). The Scope for Inflation Targeting in Developing Countries. Working Paper 97/130. Washington, DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation

    MontielP. (2013). The monetary transmission mechanism in Uganda. International Growth Centre.

    MugumeA. (2011). Monetary Transmission Mechanisms in Uganda. Working Paper 01/2011Bank of Uganda.

    Mutebile-TumusiimeE. (2014 June 27). Remarks by the Governor at the Uganda Bankers’ Association Informal Dinner Kampala.

    OpolotJ. (2013). Bank Lending Channel of the Monetary Policy Transmission Mechanism in Uganda: Evidence from Panel Data Analysis. Working Paper 01/2013. Bank of Uganda.

    • Search Google Scholar
    • Export Citation

    OstryJ.GhoshA. R. andChamonM. (2012). Two Targets, Two Instruments: Monetary and Exchange Rate Policies in Emerging Market Economies. Staff Discussion Note 12/01. Washington, DC: International Monetary Fund.

    • Search Google Scholar
    • Export Citation

    SandeD. andApaa-OkelloJ. (2013). Interest Rate Pass Through in Uganda. Working Paper 02/2013. Bank of Uganda.

    SargentT. J. andWallaceN. (1981). Some Unpleasant Monetarist Arithmetic. Federal Reserve Bank of Minneapolis Quarterly Reviews5 (3) 117.

    • Search Google Scholar
    • Export Citation

    TaylorJ. B. (1993December). Discretion versus policy rules in practice. Carnegie-Rochester Conference Series on Public Policy39195214. Elsevier.

    • Search Google Scholar
    • Export Citation
1Article 162 of the 1995 Constitution states that: ‘In performing its functions the Bank of Uganda shall conform to the Constitution but shall not be subject to the direction or control of any person or authority’.
2Core inflation excludes the prices of food crops and utilities, fuel, and energy which together comprise 17.6 per cent of the overall (headline) consumer price basket. The BOU targets core inflation, rather than headline inflation, because the prices of food crops and energy, fuel, and utilities are subject to supply-side shocks to a greater extent than other components of the consumer price index, and are thereby much less influenced by monetary policy, and also because they (especially food crop prices) are more volatile. The standard deviation of food crop inflation was more than double that of core inflation between 2006 and 2015.
3Under the monetary targeting framework, in which primary securities issues were used only for monetary policy, the BOU alone determined the amount issued.
4These were issued to recapitalize the BOU, beginning in May 2013.
5This is the definition of temporary advances from the central bank to the government which is used to compute the indicative target on this variable in the Policy Support Instrument programme which Uganda has with the IMF. The indicative target was introduced in 2015.
6Initially the CBR was set on a monthly basis, at the start of every month, but the BOU switched to a bi-monthly basis, in the middle of the relevant month, at the start of the 2014/15 fiscal year. The change was motivation by the fact that the inflation volatility which characterized 2011 and 2012 had been reduced and, as such, inflation forecasts did not vary much from month to month, hence a monthly change in the CBR was not necessary to ensure that the monetary policy stance was aligned with medium-term forecasts; see Mutebile-Tumusiime (2014).
7The model was developed with technical assistance from the IMF’s Research Department.
8Where the forecast was a range, we have used the mid-point of the range to calculate the deviation of outturn from forecast.
10Goncalves (2015) estimated an interest rate reaction function for three East African central banks in a small econometric model to assess whether they followed the Taylor principle. He found that interest rates in Uganda were consistent with the Taylor principle.
11Banks always hold some reserves in excess of their minimum statutory level of required reserves, as a buffer against shocks. Excess bank reserves averaged Shs 119 billion over the period January 2015 to October 2016, which is 10 per cent of the banking system’s required reserves.
12Many cross-country empirical studies have identified a robust correlation between the level of the real exchange rate and economic growth in developing countries over the long term, with a more depreciated real exchange rate supporting higher long-term growth and/or accelerations in the growth rate (e.g. Elbadawi, Kaltani, and Soto, 2012; Johnson, Ostry, and Subramanian, 2007).
13If the government fully funds its domestic borrowing requirement by issuing securities to the market, it must accumulate a surplus in domestic currency with the central bank to offset its net foreign currency purchases from the central bank. Hence in principle, the foreign currency which is purchased by the BOU, over and above that which is used to accumulate foreign reserves, and which is sold to government, should not need to be sterilized. However, there may be timing differences between the purchase of foreign currency by the BOU and government debt issuance which necessitate temporary sterilization by the BOU.
14We have excluded the first nine months of the IT period (July 2011–March 2012) because in this period the BOU intentionally aimed to keep the seven-day interbank rate close to the top of the band around the CBR, which at that time was four percentage points above the CBR, because inflation was very high. This was done by issuing sufficient repos to push the weighted average repo rate close to the top of the band around the CBR. After March 2012, the BOU adjusted its intervention policy so that the weighted average repo rate was closer to the CBR itself.
15Empirical studies which cover the period before the introduction of the IT framework in Uganda include Mugume (2011), Montiel (2013), and Davoodi et al. (2013). The first two of these studies finds a weak monetary policy transmission mechanism, while that of Davoodi et al. finds evidence of a more substantial channel.

    Other Resources Citing This Publication