First a word of thanks for the invitation to speak here today.
The question I shall be trying to illuminate is how the choice of a stabilisation policy regime is likely to affect the conditions for achieving economic growth that is satisfactory and sustainable. To this end I shall be using three approaches. First I shall present a historical review aimed at investigating the nature of the relationship between growth and inflation. This I shall do using data from the twentieth century. At the same time I shall comment on what macroeconomic research has to tell us about the growth-inflation relationship. Then I shall reverse the question and consider how economic growth influences monetary policy in an inflation-targeting regime. In conclusion I shall present some reflections on how the new stabilisation regime we set up in Sweden in the 1990s - with low inflation and stable macroeconomic conditions - has affected the prospects for economic growth.
Inflation and growth in a longer perspective
In the period since the inflation-targeting regime was introduced in Sweden in 1993, annual CPI inflation has averaged approximately 1.8 per cent. Annual GDP growth in this period has averaged approximately 2.5 per cent. This is markedly different from the picture in the late 1980s. It is also something of a contrast with developments elsewhere, as can be seen from Fig. 1. In recent years Sweden has had both higher growth and lower inflation than both the euro area and Germany, whereas in the second half of the 1980s the opposite was the case. In this comparison we must of course bear in mind that a part of the good performance recently has had to do with capacity utilisation being rather low for so long after the crisis in the early 1990s.
Still, what we have experienced is appreciably lower inflation and higher growth. Such periods have occurred earlier, however, as is clear from Fig. 2. In the twentieth century up to World War One (1900-14), for instance, annual inflation averaged about 1 per cent and growth almost 3.5 per cent. The diagram also shows that besides the combination of low inflation and high growth, there have been periods with the opposite. But from this one cannot definitely conclude that there is a stable causal connection - that low inflation causes higher growth. The relationship between growth and inflation is plotted in Fig. 3; each point represents the intersection of inflation and growth rates for Sweden over five-year periods. A correlation does exist but is not particularly strong.
While it is possible that low inflation does lead to higher growth, it is equally conceivable that the relationship between growth and inflation is due to them both being affected by some other variable that has not been taken into account. One example could be oil price increases, which push inflation up at the same time as through various mechanisms they depress growth. Perhaps we should not be too surprised that the relationship between inflation and growth is not all that strong. According to rational expectations theory, it is not so much actual inflation as its unexpected development that is liable to affect the behaviour of households and firms.
There are studies which do suggest that low inflation may be beneficial for growth, particularly when the comparison is between rates of inflation above 10 per cent and levels close to our current target. Some of these studies have been done, for example, under the auspices of the International Monetary Fund and the World Bank. However, many of these studies suffer from what statisticians often refer to as the identification problem. Inflation has usually fluctuated most in periods when it has been high, while periods of low inflation have tended to be associated with a more stable rate. The specific importance of inflation's average rate is then difficult to determine because it is hard to distinguish this from effects of inflation's fluctuations. It is easy to see that large fluctuations in inflation may be harmful for real economic growth; they naturally result in considerable uncertainty among households and firms, with the risk of negative effects not least on the propensity to invest. As shown in Fig. 4, growth in Sweden has in fact varied most in decades when inflation fluctuated markedly.
Stabilising inflation at a low level may thus be one way of reducing the risk of fluctuations in its rate and thereby in the real economy too. It follows that it should also contribute to growth being somewhat higher in the longer run. A more complex question is the exact level at which inflation should be stabilised; much has been written about that issue but I shall not be going into it here. Personally I am fond of a definition that is usually attributed to Alan Greenspan: inflation should be so low that it does not appreciably affect the behaviour of households and firms. That, I believe, amounts to something fairly close to the figure of 2 per cent we now use in Sweden. It is worth noting that in a number of other countries the inflation target is also in the vicinity of this level.
The role of growth in the inflation targeting regime
Although a clear causal connection between inflation and growth is hard to demonstrate, in an inflation targeting regime economic growth is a central consideration for monetary policy. I shall now move on and briefly describe our view of growth when forecasting inflation and say something about the practical problems in this context.
The traditional Keynesian view is that variations in demand cause the actual development of production (measured, for example, in terms of GDP) to fluctuate around a long-term trend (Fig. 5). When GDP is above the trend there is normally demand pressure that is often inflationary. Countering such inflationary impulses is the Riksbank's task. Similarly, we normally act if actual growth is on the way to being below its long-term potential rate. Thus, our efforts to fulfil the inflation target can also help to smooth the fluctuations in economic activity. Trend GDP is usually assumed to be exogenous, that is, given in advance and not something that monetary policy can influence. What we can influence, in the first place by adjusting the repo rate, is the general fluctuations in demand and thereby in inflationary pressure.
Monetary policy's impact on demand is commonly considered to be lagged. The Riksbank therefore has to forecast inflation a number of years ahead and use the forecast as a guide to repo rate decisions. If a forecast based on the assumption of an unchanged repo rate points to inflation in the forecast period being above the targeted rate, there are normally grounds for raising the repo rate. Similarly, there are normally reasons for lowering the repo rate if the forecast points to inflation being below the target.
Growth has a central position in the assessment of inflation. Having arrived at a picture of future demand, this is compared with the growth that is considered to be feasible without leading to resource utilisation being so high that inflation is liable to take off. Concepts such as potential growth and total resource utilisation are a problem because they cannot be observed directly and have to be estimated instead with various econometric and statistical methods. Fig. 6 illustrates the complex nature of reality - different ways of measuring potential growth evidently give rather different results. Moreover, potential growth is liable to vary over time but it may not be possible to verify such variations until growth and inflation outcomes are available. For these and other reasons, the Riksbank uses a combination of many different methods. In the case of potential growth in the coming years, for example, we also make relatively detailed assessments of labour supply, together with various productivity forecasts that, for instance, take the cyclical pattern into account.
As the estimation of the economy's production capacity has changed over time, so of course have the assessments of resource utilisation at a particular time. This is evident from Fig. 7. A good example is the assessments for 1995-96. In November 1995 we calculated that the economy's unutilised resources - estimated in terms of the output gap - would be fully utilised by the end of 1996, after which the output gap was calculated to be positive. In last October's Inflation Report, however, the assessment was that the output gap in 1995-96 had been markedly negative at around 5-6 per cent. A part of the difference has to do with methodological developments in the measurement of the output gap. The results may also have been affected by statistical revisions that can go quite far back. Moreover, the appraisal of unutilised resources may change over time as new information becomes available. So even our present opinion about the size of the output gap will presumably be modified in the future as new methods are developed and new statistics are taken into account.
For clarity's sake I should underscore that neither is the Riksbank's assessment of resource utilisation based solely on econometric calculations. Here we likewise use additional information of other kinds about shortages and so on, for instance from business tendency surveys. An unusually good example of this happens to be just the winter and spring of 1996, when we accorded little weight in practice to the calculated indicator of resource utilisation and paid more attention to business tendency data and information about the situation in the labour market.
Recent years also provide many good examples of the difficulties in gauging the rate of potential growth or the level of resource utilisation. It is only some years ago that a major topic of debate was the likely consequences of the so-called new economy. Productivity growth was surprisingly strong for a couple of years, not just in the United States but to some extent also in Sweden. And although growth was high, inflation remained appreciably lower than expected year after year. Against this background it could therefore be argued that potential growth had become somewhat higher. Some of the reforms in the early 1990s, involving taxes, social security and deregulations, pointed in the same direction. If that were the case, there could perhaps also be room for lower interest rates.
But these are only some of the problems involved in assessments and decision-making. Fundamental changes in supply can also have effects on demand. Better conditions for growth strengthen expectations of profits and that in turn pushes share prices up. Besides favouring investment, this may fuel consumption by share-owning households. Rising productivity may send a message to households that there is more room for wage increases and thereby also favour consumption. Such a course of events could just as well lead to higher interest rates as lower. The difficulty in assessing the strength of the various mechanisms and still more their dynamics over time is obviously considerable.
The higher inflation outcomes in the past year raise some similar questions, though now with the opposite sign because inflation has been higher than expected even though aggregate demand has been lower. In a number of Inflation Reports in the past year this has prompted us to discuss whether our assessments of how much the economy can deliver without increased inflation have been too optimistic.
Before turning to my third and last issue, I should like to draw two conclusions about this set of problems:
- In practice, the only course open to a central bank is presumably to collect new information about resource utilisation and potential growth and gradually incorporate it in assessments. That exposes one to criticism for being too optimistic as well as too pessimistic. It is only in retrospect that one knows the right answer. The Riksbank's attitude to the new economy was, as I said at the time, "tentative but not dismissive". Our forecasts were constructed accordingly. We were somewhat more optimistic than before about potential growth and resource utilisation but the adjustment amounted to just some tenths.
- The other conclusion concerns the accuracy of inflation targeting. We presumably have to accept that variations occur in inflation. Achieving a rate of exactly 2 per cent year after year is not feasible in practice, though it should be possible as an average in the longer run. It would not in fact even be desirable to aim for exactly 2 per cent in every situation. That would result in unnecessarily large fluctuations in interest rates and economic growth. So given the gaps in our and other central banks' knowledge - as witness, for example, our difficulties in measuring resource utilisation - there are good reasons for being sceptical about fine tuning the economy through monetary policy. We don't know enough to do that.
Growth effects of an inflation targeting regime
Finally, I want to say something about what the changeover to a low inflation environment, with stable rules for economic policy, may have meant for growth. My remarks are based, at least to some extent, on personal experience of what has happened in Sweden in the past ten to fifteen years. There is support in research for some but not all of my observations. Here, then, are four areas where I see clear changes from the 1970s and '80s in directions that favour growth.
Productivity
An important explanation for the acute economic crisis in the early 1990s was the weak productivity in the 1970s and '80s. In this respect there seems to have been a change for the better, though it is still too early to tell how permanent it will be, see Fig. 8. The change is presumably attributable in part to the clearer framework which the inflation targeting regime has created for business operations. The short-cuts to profits that were available in the 1970s and '80s, when problems with competitiveness were repeatedly resolved with devaluations, do not exist today. Both the motives for and the acceptance of continuous rationalisation and heightened efficiency have been strengthened.
The composition of investment
In the 1980s Sweden invested a great deal in housing and other real estate. The picture in the past decade has been different, with investment in machinery playing a leading role (Fig. 9). The sectoral pattern has also changed, with considerably more investment in advanced technology. There are, of course, many explanations for these differences, such as the structure of the tax system and the introduction of new technologies. But it seems to me that the more stable macroeconomic environment has played a major part. Many investment decisions in the 1980s, not least in real estate, were based on false premises. Moreover, the high level of costs in the wake of inflation hit industrial investment. The investment propensity was probably weakened by the difficulties in predicting profitability and general economic trends. The academic literature provides support for these notions. When inflation is high and volatile, price mechanisms function less efficiently. It becomes difficult to distinguish profitable investments from those that are less viable.
Wage formation
Here, too, there have been appreciable improvements. Five significant changes in this field in the 1990s have been identified by Carling et al and at least two of them are connected with the monetary policy regime:
- One is that the lower inflation improves the chances that low nominal wage increases will yield higher real wages. The existence of such a real wage trend is confirmed by Fig. 10; the nominal rate of wage increases in the 1990s was lower than in the 1970s and '80s, yet the real wage has developed comparatively strongly. The picture in the 1970s and '80s was different. When fiscal policy failed to secure low inflation and we had a series of devaluations, wage formation more or less collapsed. Wages and prices chased each other in an upward spiral.
- The other is that instead of acting via price increases and impaired competitiveness, high wage increases have a direct impact in the form of lower output, investment and employment. Nowadays the labour market organisations have a specific inflation target to start from; it is admittedly over- or under-shot at times but it does serve as a clear and useable benchmark.
Structural policy
The new environment, with a precise target for inflation, has also improved the conditions for generating changes that strengthen competition and heighten efficiency in the Swedish economy. The chief argument in favour of such measures is, of course, that they help to promote prosperity. That was also our argument in the Finance Ministry in the late 1980s, for instance when we were working on the deregulation of agriculture and the housing sector. But it was not always an argument that made an impression on strong producer interests. Today there is the additional argument that increased competition can help to keep inflationary pressure down and that in turn - at least for a time - can pave the way for somewhat lower interest rates and thereby higher growth.
Concluding remarks
The realignment of stabilisation policy has thus, in my opinion, been important for improving conditions for economic growth in Sweden. In several respects the new stabilisation policy environment is more conducive to growth than its predecessor. The role of monetary policy in this context is to safeguard the value of money, a goal the Riksbank has expressed operationally as 2 per cent inflation. That in turn lays a good foundation for growth and employment. It must, however, be born in mind that in addition to low inflation, high economic growth presupposes a sound educational system, a suitable climate for enterprise and investment, efficient tax and social security systems and so on. The demands on economic policy are further accentuated by foreign competition in a world that is changing rapidly, with high ambitions for increased prosperity.
Questions of this kind deserve considerably more attention from media as well as in discussions among economists. For our well-being in the longer run they are far more important than whether the repo rate moves up or down a quarter of a percentage point.
Notes:
1) Note, moreover, that the figure shows moving five-year means; the relationship in terms of annual change figures looks much weaker.
2) See e.g. Fischer, S. Modern Central Banking", i Capie, F., m.fl. (eds.), The future of central banking: The tercentenary symposium of the Bank of England, Cambridge University Press, 1994.
3) A linear trend or an HP filter was often used earlier to measure potential growth, while more recent estimations have been based on the UC method; see Apel, M. och P. Jansson "System Estimates of Potential Output and the NAIRU", Empirical Economics 24(3), 1999, 373-88.
4) Heikensten, L. (2000), "Monetary policy and the new economy", speech to the Parliamentary Finance Committee, February 24.
5) Data for the United Kingdom support this thesis: Beaudry et al. (2001), Monetary instability, the predictability of prices and the allocation of investment: an empirical investigation using U.K. panel data, American Economic Review 91, June.
6) Carling et al. (1998), Nya förutsättningar för lönebildning (New conditions for wage formation), Ekonomisk Debatt 26.