Question 1. Why worry about the monetary transmission mechanism (MTM) in low-income countries (LICs), in particular?

The environment for monetary policy in many LICs has undergone a fundamental transformation in recent decades, with much lower inflation, fewer signs of fiscal dominance, and an end black market exchange rate premia. And now, as we describe in Berg and others (forthcoming), many central banks in low-income countries in Sub-Saharan Africa are modernizing their monetary policy frameworks.

At the same time, their economies are profoundly different from those of emerging market and advanced countries, with among other features much smaller financial sectors, a high incidence of supply shocks, and a large share of food in consumption. Monetary policy frameworks have also been very different, with greater reliance on monetary aggregates to conduct policy.

Some variant of inflation targeting is the obvious benchmark for those countries with an independent currency. But how can a central bank emphasize an inflation objective as its nominal anchor if it does not know how monetary policy influences the macroeconomy?

Question 2. What is the evidence that transmission in LICs is radically different?

Mishra and others (2012) argue that transmission in LICs is “weak and unreliable.” They note that the correlation between short and long rates is generally much weaker in LICs than in emerging markets or advanced countries, and that vector-auto-regression (VAR) based evidence in developing countries tends to find insignificant effects of “monetary policy shocks” on inflation and output.

In ongoing work, we are investigating the question posed in Mishra and others (2012) as to whether this evidence reflects the techniques or the “facts on the ground.” Early results suggest that there are indeed reasons to worry about the application of some standard empirical techniques in conditions typical of LICs. In particular, data scarcity and regime changes imply that periods of consistent policy regime with adequate data are very short, rarely longer than say 10 years. Simulations suggest that 10 years is generally too short to get significant coefficients in a standard small VAR quarterly data, even if data are measured correctly and monetary policy is correctly identified. And indeed the regimes in the region have not been stable (Berg and others, forthcoming).

It is worth remembering that even in the United States, where data series are long and structural change less important, it took many years and dozens of papers before anomalies such as the “liquidity puzzle” (that a monetary policy tightening seemed to lower interest rates) and the “price puzzle” (that a monetary tightening seemed to increase subsequent inflation) were solved in VARs.

Question 3. What else can be done to identify the MTM?

As Summers (1991) and Romer and Romer (1989) argue, views on the real effects of monetary policy in the United States have been more influenced by the narrative arguments of Friedman and Schwartz (1963) and by reference to the real effects of the “Volcker disinflation,” than by formal statistical analysis. Thus, much of our research has focused on country-specific analyses rooted in detailed consideration of particular episodes.

In Berg and others (2013), we examine a significant tightening of monetary policy that took place in October 2011 in four members of the East African Community (EAC): Kenya, Uganda, and Tanzania, as well as the somewhat different experience of a fourth EAC country, Rwanda. The events we studied took place in the context of sizeable commodity price shocks, one peaking in 2008 and the second in 2010–11, as well as the global financial crisis. Policymakers generally did not tighten in the face of the first shock, a response validated by the collapse of these prices and of external demand in 2009. The second boom was more persistent and, perhaps inspired by the earlier episode, policy remained accommodative in most of the countries in question.

Throughout 2011, concerns increased about the adequacy of the policy stance. However, it was unclear when a tightening might come or how strong it would be. Thus, when it came it was at least partly unexpected—“unusual” in the language of Friedman and Schwartz (1963). We can thus ask, what did this large monetary policy tightening shock do?

Question 4. What happened, then, when policy was tightened?

We find clear evidence of the monetary policy transmission mechanism, especially in Kenya and Uganda: after a large policy-induced rise in the short-term interest rate, lending and other interest rates rise, the exchange rate tends to appreciate, output growth tends to fall, and inflation declines (Figure 1).

Kenya and Uganda: Selected Macro-Variables Jan. 2010–Dec. 2013

Citation: IMF Research Bulletin 2014, 002; 10.5089/9781498320795.026.A003

Source: IMF Working Paper 13/197.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Kenya and Uganda: Selected Macro-Variables Jan. 2010–Dec. 2013

Citation: IMF Research Bulletin 2014, 002; 10.5089/9781498320795.026.A003

Source: IMF Working Paper 13/197.

• Download Figure
• Download figure as PowerPoint slide

Kenya and Uganda: Selected Macro-Variables Jan. 2010–Dec. 2013

Citation: IMF Research Bulletin 2014, 002; 10.5089/9781498320795.026.A003

Source: IMF Working Paper 13/197.

• Download Figure
• Download figure as PowerPoint slide

Question 5. But what about the special features of LIC economies?

There can be no question that the unusual features of LIC economies can make a large difference to the MTM. The trick, however, lies in figuring out what matters when.

For example, in our narrative case studies in Berg and others (2013), there were several “dogs that didn’t bark,” i.e., features that did not manifest themselves as critical but which had received substantial attention from policymakers and analysts. For example, the depth of the financial sector did not distinguish the two countries with the clearest transmission (Kenya and Uganda) from the two others. In terms of the size of the financial sector, Kenya is the outlier, with a much larger financial sector than the other three (Figure 2). Yet the MTM looks quite similar in Kenya and Uganda and somewhat different in the other two countries.

Kenya, Rwanda, Tanzania, and Uganda: Size of the Financial Sector

Citation: IMF Research Bulletin 2014, 002; 10.5089/9781498320795.026.A003

Source: IMF Working Paper 13/197.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Kenya, Rwanda, Tanzania, and Uganda: Size of the Financial Sector

Citation: IMF Research Bulletin 2014, 002; 10.5089/9781498320795.026.A003

Source: IMF Working Paper 13/197.

• Download Figure
• Download figure as PowerPoint slide

Kenya, Rwanda, Tanzania, and Uganda: Size of the Financial Sector

Citation: IMF Research Bulletin 2014, 002; 10.5089/9781498320795.026.A003

Source: IMF Working Paper 13/197.

• Download Figure
• Download figure as PowerPoint slide

Question 6. What did seem to influence the MTM?

The feature that seems to most clearly distinguish the four countries we examined was the nature of the policy regime itself. Uganda had just officially announced an “Inflation Targeting-Lite” regime in July 2011, after attempting “flexible money targeting” from late 2009. Kenya, meanwhile, also announced in September 2011 operational reforms designed to emphasize the role of the central bank policy interest rate in the conduct of policy. Meanwhile Tanzania and Rwanda during this period placed primary emphasis on reserve money growth as an operational target with broad money aggregates as intermediate targets, while also announcing a central bank policy rate. Rwanda added substantial exchange rate intervention to the mix, such that the nominal exchange rate exhibited little volatility around a steady rate of depreciation.

We argue in the paper that these differences in regime may have shaped the MTM. In Tanzania, for example, the observed increase in the policy rate (and tightening through quantity instruments) did not translate into higher lending rates, but other aspects of the MTM (credit growth, exchange rate, inflation, possibly output) seemed to respond.

Question 7. Does this mean that the monetary policy problem is very similar in LICs and Emerging Markets (EMs)?

Not so fast. It surely remains the case that, while the main elements of the MTM seem to be there in recognizable form in some important cases, LIC central banks face a number of unusual, albeit not unprecedented, challenges.

Other work in the IMF’s Research Department has focused on the role of food price shocks (Andrle and others, 2013) and of money targeting (Andrle and others, 2013b) in a forward-looking policy framework. The bank-dependent financial systems shaped the nature of and response to the global financial crisis in LICs, which manifested in sudden stops of capital, increases in the country risk premium, adverse movements in the terms of trade, and a flight to safety domestically (Baldini and others 2012). One important implication is that various credit and money aggregates moved in complex and superficially contradictory ways, complicating the task of inferring the stance of monetary policy from these aggregates. Another is that monetary policy may have fairly limited scope to buffer the real effects of such shocks.

More work remains to be done on the role of limited financial markets and the banking system. Research under-way jointly with the Bank of Uganda involves collecting a large set of loan-level observations to understand some aspects of the MTM that cannot be readily observed with the aggregate data, such as the role of credit rationing.

Finally, the topic of exchange rate management is an important one in LICs, perhaps even more so than in more developed countries. Central banks have been using sterilized intervention along with more standard instruments, and Benes and others (2013) and Ostry and others (2012) are beginning to come to grips with the question of when and how to do so effectively. A related and under-researched topic is the strength of monetary transmission and the role of monetary policy in pegged regimes with limited capital mobility (See for example Blotevogel, 2013).

Despite these important uncertainties, we believe that our results should be encouraging for those central banks that are attempting to move toward more forward-looking policy frameworks. It also suggests that models that embed the standard MTM, albeit carefully modified and augmented to capture critical country-specific features, can be useful.