Identifying monetary policy rules in South Africa with inflation expectations and unemployment

1 Introduction

South Africa formally adopted an inflation targeting monetary policy regime in February 2000, with the announcement of a 3–6 per cent target for 2002 onwards. Following the development of inflation targeting as a monetary framework, and its implementation by New Zealand’s and Chile’s central banks in 1990, numerous countries chose similar policy frameworks to replace discretionary policies that had aimed at stabilizing a measure such as gross domestic product (GDP), unemployment, or an exchange rate (Svensson 2010).
Inflation targeting includes more than the choice of policy target. It is widely accepted that a full inflation targeting regime includes five elements: (1) the public announcement of medium-term numerical targets for inflation; (2) an institutional commitment to price stability as the primary goal of monetary policy, to which other goals are subordinated; (3) an information-inclusive strategy in which many variables, and not just monetary aggregates or the exchange rate, are used for deciding the setting of policy instruments; (4) increased transparency of the monetary policy strategy through communication with the public and the markets about the plans, objectives, and decisions of the monetary authorities; and (5) increased accountability of the central bank for attaining its inflation objectives (Mishkin 2000). While adoption of inflation targeting always involves a regime with an announced inflation target, individual countries’ frameworks combine some or all of the elements to different degrees. South Africa’s framework includes all five elements and since inception has evolved to strengthen several, especially transparency.

This study examines the operation of monetary policy using quarterly data from 1994 through 2015 and the sub-period covering 2002–15, when monetary policy was formally defined by inflation targeting. The principal research question we address is the extent to which the monetary policy instrument, the South African Reserve Bank’s (SARB) repo rate, has systematically responded to deviations of an inflation measure from its target range and its responsiveness to other variables such as the output gap and unemployment gap. We model the SARB’s monetary policy reaction function as a Taylor rule, with measures related to inflation and the output gap as arguments.

Estimates for many countries have found that a Taylor rule does describe the monetary policy reaction function well, but that result would not be expected for a country that has a simple and rigid regime that targets only currently observed inflation and adjusts policy to correct only for current inflation deviations. In such a model, the coefficient on the output gap or other variables would be expected to be not significantly different from zero. Our estimates shed light on whether the SARB has focused on current inflation alone.

The issue is important for South Africa’s choice of macroeconomic strategy. At various times since the introduction of inflation targeting, political and technical debates have included prominent criticisms of inflation targeting or the interest rate policies resulting from it. For example, pressure for SARB to abandon interest rate policies pursued under inflation targeting has been voiced by the trade union confederation, COSATU, which argued in 2007 and subsequently ‘that the narrow focus on inflation targeting is not an appropriate monetary policy approach’ (COSATU 2007, 2011). Similarly, some independent experts have argued for flexibility in the SARB’s operation of inflation targeting to make interest rate setting responsive to GDP growth and unemployment, as well as inflation (Business News 2016).

Although critics argue against a narrow South African inflation targeting regime defined in terms of current inflation, the SARB, like other central banks, defines its approach as flexible inflation targeting (Marcus 2015). Explicitly, its flexibility is marked by targeting a band of inflation rates rather than a specific level, but it also has discretion over the speed with which it aims to achieve corrections to the target band (policy rate smoothing) and has the ability to take into account the sources of inflation shocks when determining the policy interest rate (van der Merwe 2004). Since SARB’s explicit mode is to target expected inflation, the Monetary Policy Committee (MPC) could be expected to form its own expectations on the basis of maximum available information and take into account numerous indicators of broad economic conditions, including measures of the output gap and labour market conditions.

With unemployment in South Africa remaining high at around 25 per cent, or 33 per cent on a broad measure including discouraged work seekers, the extent to which its monetary policy has actually followed a Taylor rule responding partly to measures of the output gap or unemployment while maintaining its focus on the inflation target is of particular importance. The theoretical foundations of Taylor rule policies relate to medium-term, countercyclical measures, while—as indicated by long-term unemployment persistently being around 66 per cent of the total—structural unemployment is a high proportion of South Africa’s total.

In this paper, we aim to investigate whether a Taylor rule accurately describes the SARB’s policy rate decisions during the official inflation targeting period, which formally commenced in February 2000, by estimating the SARB’s monetary policy reaction function as a modified Taylor rule, framed in terms of measures of expected inflation and labour market slack.

Our results suggest that the SARB’s monetary policy reaction function is not distinguishable from a stochastic modified Taylor rule; its rate setting is a function of both expected inflation and labour market slack. The finding is similar to that of Ellyne and Veller (2011), who find that a Taylor rule equation fits the data more strongly for the period after the formal adoption of inflation targeting than before. Our study strengthens that finding by addressing two sources of potential mismeasurement. Instead of using current inflation as a proxy for expected inflation, we use direct measures of inflation expectations; and instead of using a Hodrick–Prescott (HP) measure of the GDP output gap (Hodrick and Prescott 1997), we use measures of labour market slack.