Lesetja Kganyago: Issues confronting monetary policymakers in South Africa
Address by Mr Lesetja Kganyago, Governor of the South African Reserve Bank, at the Thomson Reuters Economist of the Year Awards JSE Limited, Sandton,11 August 2015.
The views expressed in this speech are those of the speaker and not the view of the BIS.
Good morning, ladies and gentlemen, and thank you for the opportunity to deliver the keynote address at this prestigious awards ceremony.
This competition is about forecasting a number of economic variables. Predicting the future is difficult at the best of times, and with the economic environment becoming increasingly turbulent, forecasting is becoming that much more difficult. It is also a challenging time for policymakers; in the face of heightened uncertainty, our decisions become more data-dependent than usual. This makes it more difficult for analysts to forecast policy moves. We interact with market participants and other interested parties on a regular basis to help them understand our thinking on various issues.In the recent past, a number of questions have repeatedly come up, and in my address to you today, I will focus on three of these interrelated questions. First, how should monetary policy respond to changes in the exchange rate, particularly when driven by terms-of-trade shocks? Second, given the importance of US policy normalisation in our communication, does this mean that the Bank follows global interest rates in making policy decisions? And finally, how does the Bank think about inflation expectations?
Responding to exchange-rate shocks
Since the beginning of 2014, when the Bank started the tightening cycle, the rand has depreciated by about 18 per cent against the US dollar and by about 6 per cent on a trade-weighted basis. For some time, we have been indicating that the rand exchange rate is one of the main upside risks to the inflation outlook. However, this does not mean that we target the rand in any way: the rand is but one of a number of determinants of inflation acting simultaneously, sometimes in different directions and with varying intensities. Monetary policy actions do not attempt to influence the exchange rate but rather focus on the overall inflation outlook. A complicating issue recently has been the fact that the traditional pass-through relationship from the exchange rate appears to be lower than during previous episodes of rand weakness. Time will tell whether or not this is a cyclical phenomenon which may be reversed at some stage, or whether it is of a more structural nature.
A number of factors have impacted on the rand, and I will highlight three broad categories. First, some domestic developments have impacted negatively on growth and risk perceptions. These include the protracted strikes in the mining and manufacturing sectors last year as well as the deteriorating growth prospects, aggravated by electricity load-shedding. These factors have contributed to the widening of the current account of the balance of payments.
Second is the prospect of US policy normalisation and its implications for the pattern of global capital flows. Expectations about the timing of the first move have kept changing as conflicting data regarding the strength of the US economic recovery emerged. Good news for the US economy proved to be bad news for emerging- market currencies, and vice versa. That the Fed will tighten is not a debate. The question is one of timing. One would expect that the markets would have discounted this to some extent, but uncertainty still remains regarding the starting date as well as the speed and extent of the cycle. The continuing changes to the so-called 'Fed dots', which reflect the expectations of interest-rate changes of individual members of the Federal Open Market Committee (or the FOMC), indicate just how uncertain this process is.
The third broad category to have affected the rand is the ongoing precipitous decline in commodity prices. Commodity prices have been on a persistent downward trend since 2011, reflecting a deterioration in South Africa's terms of trade and contributing to a persistent current-account deficit. This deterioration was reversed briefly with the decline in the international oil price in the second half of 2014. However, in the past two to three months, we have seen a further sharp decline in commodity prices in international markets. For example, since the beginning of May, platinum prices have declined by about 17 per cent, iron-ore prices by 5 per cent (and about 20 per cent since January), gold prices by 8 per cent, and copper prices by 15 per cent. Clearly, these developments, if sustained, pose a downside risk to South Africa's growth outlook, apart from their impact on the current account and the exchange rate. Offsetting this to some extent is the fact that international oil prices have also moderated significantly over the period, having declined from the recent highs of around US$65 per barrel in May to the current levels of around US$50 per barrel.
I should point out, however, that South Africa is not alone in this respect. While a number of idiosyncratic shocks have impacted on the rand, the currencies of other commodity producers have also been affected, in some cases more so than the rand. Since the beginning of the year, for example, during which time the rand has depreciated by about 9 per cent against the dollar, the rand remains more or less unchanged against the Australian dollar and the Russian ruble, but has appreciated by about 8 per cent against the New Zealand dollar, by about 17 per cent against the Brazilian real, by about 2 per cent against the Chilean peso, and by about
3 per cent against the Canadian dollar (although there was generally a good deal of volatility between these two periods). Monetary policy reactions have differed; over this period, monetary policy has been tightened in Brazil, unchanged in Chile, and loosened in Australia, Canada, and New Zealand.
The question is how monetary policy should react to different categories of shocks. Textbooks distinguish between portfolio shocks and terms-of-trade shocks (apart from standard domestic-demand shocks). The challenges posed to the rand by the US policy uncertainty clearly fall into the category of portfolio shocks. Here, literature tells us that policymakers should respond to these shocks by tightening monetary policy. In an inflation-targeting framework, the tightening would depend on the extent to which the weaker exchange rate was expected to impact on inflation rather than acting to protect the exchange rate per se. The role of inflation expectations (and monetary policy credibility) would be central to this.
Reaction to a terms-of-trade shock is less straightforward, as it would also depend on whether it was a decline in an export price (for example the gold or platinum price), which would in itself have less impact on domestic prices, or whether it was an increase in import prices (for example oil prices), which do impact more on domestic inflation. In general, a terms-of-trade deterioration driven by a decline in export prices implies a decline in domestic incomes which can be disinflationary. A fall in export prices may be accompanied by currency depreciation which moderates the impact of the decline in export prices on demand and inflation. A policy response would need to try and 'see through' the first-round effects of the change in the exchange rate. However, again, much depends on the extent to which inflation expectations are well anchored.
Where countries have inflation under control, a response to depreciation and negative terms-of-trade shocks can be more supportive to economic activity. Other commodity producers - like Australia, Canada, Chile, and New Zealand - have low levels of inflation and minimal upside risks to inflation. However, the Australian response could be assisted to a significant degree by the fact that the Prices and Incomes Accord between government and the trade unions, implemented during 1983 and 1996, had broken the back of inflation, and that inflation expectations were well anchored at that time. This is policy capital that Australia has and it could stand them in good stead this time around too. By contrast, the risks to South Africa's inflation are on the upside, inflation is uncomfortably close to the upper end of the target range, and inflation expectations are at elevated levels.
Textbooks may be clear, but reality is a bit more complicated. It is very difficult to disentangle the impact of the various types of shocks when they are happening simultaneously, as is currently the case in South Africa; it is also difficult to know whether these shocks are once-off or likely to persist for some time. It is easy enough to 'look through' a once-off shock, but when the shock is of a protracted or continuous nature, it is a lot more difficult to manage. Furthermore, it is not straightforward assessing the extent to which inflation expectations are anchored, an issue that I will return to in a while.
Global interest rates and domestic monetary policy
A related question is: does the fact that we are concerned about the impact of Fed normalisation mean that we simply follow global interest rates in conducting monetary policy? The simple answer is: no. For a start, it is difficult to talk about a global interest rate when interest rates in the US and the UK are expected to increase in the near future but those in the eurozone and Japan are expected to remain close to the zero bound for a much longer period. With respect to the US rate, the correlation between US and South African long bond yields is much stronger than at the short end. Our decisions on short-term interest-rate changes are focused primarily on the domestic inflation outlook which can include the extent to which expected changes in US rates are assessed to impact on the overall inflation outlook through the exchange-rate channel.
In this respect, there are conflicting views in the market about the way in which we should react to a change in the US interest rate: pre-emptively or reactively? On the one hand, we are told that by acting reactively and delaying the adjustment, we will have to act more aggressively later, ultimately leading to a much higher rate. Implicit in this argument is that inflation would be higher than if we had acted pre-emptively. On the other hand, critics of our recent rate increase argue that by moving pre- emptively we will be starting from a higher repo-rate level when the Fed action begins, which will affect short-term growth.
To some extent, we have tried to chart a middle road between these two views. Our primary mandate is to maintain price stability in the interest of balanced and sustainable economic growth. This implies assessing the impact of our policies on inflation, output, and employment under current conditions. For this reason, our monetary policy tightening cycle has been moderate, and this approach remains our base case - although it is, of course, highly data-dependent. We have not observed a knee-jerk reaction to every adverse move in inflation, particularly among pressures driven by supply-side shocks. But at the same time, completely ignoring the risk of second-round effects could entrench inflation expectations at higher levels and could require stronger action later, with consequences for output.
Monetary policy and inflation expectations
The issue of inflation expectations has cropped up a number of times this morning. Inflation expectations are potentially an important determinant of actual inflation outcomes: to the extent that price-setters set prices on a forward-looking basis, they will have some notion in their minds as to what future inflation is likely to be. Similarly, wage negotiations generally revolve around expected inflation outcomes. In fact, an objective of an inflation-targeting framework is to anchor longer-term inflation expectations within the target range, and this will not only help to maintain inflation within the target, but will also make expectations resilient to temporary shocks to the inflation outlook. This requires monetary policy credibility, which means not only transparency and clear communication, but also demonstrated commitment to act appropriately to keep inflation under control.
All this seems relatively straightforward. However, there are a number of complicated issues and questions which make the interpretation of measured or implied inflation expectations extremely difficult, particularly for policy purposes. First, in reality, it does not necessarily follow that inflation expectations are always correct or that they will be self-fulfilling. Related to this: what do we do when inflation expectations diverge from our own forecasts, which already have inflation expectations incorporated into them? This relates to a second issue: whose expectations should we give most weight to, particularly when there is a divergence between the expectations of different categories of respondents? Do we simply look at those of price-setters?
Third, it is sometimes the case that inflation expectations are formed not on a forward-looking basis but rather on the basis of backward-looking or adaptive expectations. Under such circumstances, published expectations may not be useful predictors of future inflation. It also comes as no surprise that when inflation outcomes have been relatively stable, inflation expectations are also stable. Under such conditions, unless we know how expectations are formed, we cannot be certain if stable expectations are indicative of well-anchored forward-looking expectations or if they are simply telling us what has happened in the past.
Fourth: what is the appropriate time horizon for assessing expectations? As Adam Posen has pointed out, anchored expectations do not necessarily mean unchanging expectations, and short-term inflation expectations should be expected to vary over the business cycles and in the face of shocks to the economy. For this reason, more attention should be given to longer-term expectations. If monetary policy does have credibility, longer-term expectations should not be impacted by shocks that affect short-term expectations.
The Bank uses a number of sources for trying to assess expectations. The broadest is the inflation expectations survey conducted by the Bureau for Economic Research, or BER, on a quarterly basis. It surveys analysts, business people, and trade unionists, asking what their inflation expectations are for the current year and for the next two years. Since mid-2011, respondents have also been asked to give their expectations of inflation in five years' time. Household expectations are also polled, but only for the current year. In addition, there are the surveys that poll financial market analysts on a monthly basis, including the Reuters econometer survey and the Bloomberg survey. The break-even inflation rates, derived on a high-frequency basis from the inflation-linked bond market, give us some indication of the views of bond market participants.
We generally assume that the expectations of the economic analysts are the most forward-looking of the three groups surveyed by the BER. Economic analysts tend to rely on forecasting models similar to our own, and it is not surprising that their forecasts are often quite similar to those of the Bank. But although their forecasts are useful for us to benchmark against, analysts are generally not price-setters in the economy. The same would apply to the break-even inflation rates, which often diverge quite significantly from those of the analysts. (I should point out that we often get confusing or conflicting information when distinguishing between the expectations implicit in the 5-, 10-, and 20-year bonds.)
Over the past few years, average inflation expectations, as shown in the BER surveys, have been relatively stable but deteriorated slightly in the latest survey. Average expectations have been close to or, at times, slightly above the upper end of the target range, although there is generally a variation between the different groups. The expectations of both business and labour are often above those of the analysts, although over time there has been some convergence between the three groupings. Research by the Bank suggests that the expectations of business and labour respondents tend to be more adaptive. So trying to make sense of the different expectations is often quite difficult, and they cannot always be taken at face value.
Furthermore, for reasons mentioned earlier, the Bank focuses less on near-term expectations than on longer-term expectations. It should not come as a surprise if shocks to the economy (for example oil price shocks) cause near-term expectations to change. But focusing on the longer term can also be a challenge as the signals are not always clear. For example, in the most recent BER survey, short-term expectations deteriorated; over the two-to-three-year time horizon inflation expectations improved, on average; but over a five-year horizon they deteriorated. A two-year horizon is consistent with the lag between a change in the policy rate and its full impact on inflation (our models show an 18-month distributed lag for the full effects of an interest-rate change to be felt). Since 2012, the average five-year inflation expectation measured each quarter has varied in the narrow range of 5,8 per cent and 6,2 per cent, with the latest reading at 6,0 per cent following two consecutive months at 5,8 per cent. This is concerning as it is uncomfortably close to or at the upper end of the inflation target range. At this stage, we still do not have a clear view of how inflation expectations are formed, particularly over longer-term horizons. We would like to think that inflation expectations are well anchored because of the credibility of the Bank, but we also recognise that these could be a reflection of past inflation. The persistence introduced by the wage-bargaining process and relatively low levels of competition in parts of the product markets means that inflation expectations matter and can be self-reinforcing. We will therefore continue to monitor the various measures, mindful of these caveats, and focus on the longer-term trends.
Final thoughts
In conclusion, I have tried to convey a sense of how the Bank thinks about some of the issues confronting monetary policymakers. Unfortunately, models and textbook prescriptions generally work more smoothly than what we have to deal with in the real world. In reality, we deal with imperfect information and uncertainty regarding inflation expectations formation, apart from having to cope with an increasingly uncertain future. The global economy is entering unchartered waters, with the prospect of normalisation in the US and a slowdown in China impacting on commodity prices. The truth is that South Africa, along with other emerging-market economies, is likely to face an increasingly turbulent time ahead. While theoretical models can help to guide us, we cannot react in a mechanical way and ultimately have to rely, to a certain degree, on subjective judgment.
I wish all the finalists the best of luck and heartiest congratulations to the winner.
Thank you.