Heikensten: Introduction on monetary policy
Allow me as usual to begin by thanking you for your invitation. I think it goes without saying that these hearings are a highly valuable feature of the open discussion of monetary policy that we are all keen to safeguard. It is also my hope that the hearings can help spread knowledge about monetary policy issues to a wider audience, especially since they are now broadcast live on television.
As is customary at the first hearing of the year I intend to begin by looking back on developments last year. As we all know, inflation turned out sharply lower than the Riksbank’s target in 2004. When that happens it is natural that a debate on monetary policy should arise in the public domain. It is hardly an exaggeration to say that the discussions in the past year have been unusually lively. I commented on the debate in detail when I was here in October, but there may be reason to touch upon it again today.
Subsequently I intend to present our current assessment of the inflation outlook and explain the interest rate decision we made yesterday, after which I will round off with some general reflections on Swedish monetary policy and the conditions under which it operates.
A look back at 2004
Developments in 2004
Sweden’s economic performance in the past year was characterised by high growth coupled with markedly low inflation. Economic growth last year was in actual fact considerably higher than normal; in the past 25 years growth has been higher on only six occasions. Even though demand in general grew strongly in 2004 the same cannot be said about employment. We have yet to see a clear turnaround in the labour market.
The Riksbank’s monetary policy decisions in 2002 and 2003, which are relevant to the inflation outcome in 2004, were initially guided by the measure of underlying inflationary pressures that the Bank has used most in recent years, UND1X. In 2003, though, the Bank shifted its focus for a period to the measure UND1X excluding energy prices. Inflation according to these measures averaged 0.9 per cent and 0.8 per cent, respectively, in 2004. In terms of the CPI, inflation was 0.5 per cent. The fact that CPI inflation turned out significantly lower than the other two measures is essentially because the fairly extensive monetary easing that began at the end of 2002 and continued through 2003 and into 2004 has kept down households’ mortgage interest expenditure.
So why was inflation in 2004 so low? In order to get a better picture of one of the causes we need to go back to the turn of the year 2002-2003. At that time inflation was pushed up unusually high after electricity prices rose to record levels. The Riksbank’s view then was that it was a temporary effect; that electricity prices would fall back and that the inflation rate would therefore come down again during the year, which it did. As regards CPI and UND1X inflation, that meant that the twelve-month change – or the change in relation to the corresponding month a year earlier – fell steeply in early 2004. So the drop in inflation at the start of 2004 was partly an expected recoil of the temporary increase a year before. (Figure 1). The Riksbank had thus anticipated in early 2003 that CPI inflation would undershoot the target in 2004.
Inflation was also being depressed by low import prices. Imported producer prices have developed weakly for a couple of years, although they rose again in 2004 partly because hikes in commodity prices pushed up the prices of intermediate goods. As regards consumer prices, inflation on imported goods has turned out unexpectedly low for some time. The surprisingly low import prices could be a result of Swedish importers and producers buying from low-cost countries to a greater extent than before, something that is difficult to capture with the measure that we normally use to gauge developments in world market prices. The low outcome is most likely also attributable to conditions in Sweden, including increased competition in many sectors, not least the retail trade.
This brings me to a third factor that has contributed to the low inflation, namely productivity, which grew stronger than expected by the Riksbank and other forecasters. (Figure 2). This is the other side of the unexpectedly weak labour market conditions, which in turn have dampened domestic price and wage pressures.
Neither are the causes of the high productivity growth fully understood. One possible hypothesis is that the large-scale IT investment in many sectors at the end of the 1990s is now beginning to pay off. Another theory argues that Sweden is a highly open economy and thus strongly influenced by the currently rapid integration of low-cost countries into world trade. As regards the developments in both productivity and imported inflation there is considerable uncertainty over the strength and durability of the present tendencies.
The Riksbank and unemployment
As I said earlier it is natural that a debate on monetary policy should arise when inflation deviates considerably from target. On this occasion the debate has been fuelled further by the rise in unemployment in 2003-2004. When I was here in the autumn I devoted a lot of attention to the issue of what the Riksbank could do about the unemployment situation. The discussion has continued since then.
It is important to remember that the Riksbank’s objective is price stability. We try to fulfil this objective by setting interest rates at a level that enables the general level of demand to grow in line with what the economy is able to sustain in the long run without giving rise to inflation, and that ensures that economic agents have confidence in the Bank to meet the inflation target. So we only have an indirect influence on the general level of demand and employment. And the poorer than expected employment conditions in the past two years have not been due to developments in demand. On the contrary, demand has proved surprisingly strong.
Nonetheless it can of course be maintained that employment could have been somewhat higher if the Riksbank had kept interest rates lower, something that would have been possible given what we now know about inflation. It is difficult to object to this assertion, which of course is why the Riksbank’s critics keep coming back to it. But an assessment of what we could or could not have done must be based on what we knew when the decisions were made, not on what we know now. One way to try to approach this issue is to compare our assessments and actions with other players’ forecasts and opinions on policy at the time in question.
The Inflation Report that we published today shows that the economic and inflation outlook on which the Riksbank based its monetary policy in 2002-2003 was in no way significantly different from that held by other leading forecasters of the Swedish economy during this period. (Figure 3). If anything it seems that the Riksbank was one of the most successful forecasters, or perhaps rather one of the least unsuccessful, although it also has to be remembered that our forecasts, unlike others’, had been based on the assumption of an unchanged repo rate.
Part of the recent monetary policy debate has seen our forecasts compared with those of the National Institute of Economic Research (NIER). It turns out, however, that the NIER’s view of the appropriate repo rate path has not differed more than marginally from the Riksbank’s. (Figure 4). On average the NIER recommended a repo rate that was barely 0.2 percentage points lower. Given the effects that one generally expects from monetary policy, inflation in 2004 would only have been marginally closer to the target if we had followed the NIER’s recommendations. Nor would we have had more than a very slight impact on employment. The point is that the differences in opinion regarding future inflation and the stance of monetary policy have been small in relation to the drop in inflation and the rise in unemployment.
To sum up, we can conclude that the low inflation outcome in 2004 cannot be attributed to a monetary policy that was contrary to the consensus view at the time. Instead, it was essentially a result of developments that practically nobody managed to foresee. Another possible conclusion is that we would have had to conduct a dramatically different monetary policy if inflation was to have turned out much closer to the target in 2004. For example, the Riksbank would probably have had to cut the repo rate rapidly and sharply at the beginning of 2002. Such a stance would most likely have caused something of a stir. Inflation was high at that time and nobody predicted the kind of dramatic decline that would have justified a substantial rate cut. To give you an idea of the sentiment at the time, one of the country’s leading trade union representatives warned us not to take the inflation target lightly.
Has policy been asymmetrical?
One recurrent feature of the recent discussion is claims that the Riksbank is more concerned about high inflation than low inflation. It has been said that inflation since the introduction of the Bank’s target – from 1995 onwards – has turned out somewhat below target on average.
Personally, I can understand fears that central banks could be ”overly zealous” in their fight against inflation, especially if a country has experience of a high-inflation regime, as is the case in Sweden. Such risks have also been one of the arguments in favour of opening the Riksbank to scrutiny and of promoting a policy that is as clear as possible. An open debate and good opportunities to scrutinise the Bank appear to me to be the best antidote against an introverted and exaggerated inflation-fighting perspective.
Having said that, though, I think that there is every reason to object to the recurrent interpretation that the low inflation is an expression of an excessively hawkish stance on the part of the Riksbank. Firstly, for a large part of the period in question monetary policy has been guided by a measure of underlying inflation UND1X, and so policy should first and foremost be evaluated in terms of this measure, in accordance with the Bank’s clarification in 1999. UND1X inflation has averaged 1.9 per cent since 1995, which is undeniably quite close to the target. (Figure 5). The fact that CPI inflation during this period has been lower – 1.4 per cent – is mainly due to the sharp decline seen in the general level of interest rates, partly because a low-inflation environment has been secured. Furthermore, ten years may be too short a period over which to assess policy, at least when a number of those years have involved an adjustment from one regime to another, as in Sweden’s case.
Another way to approach this issue is to study the episodes when inflation has undershot the Riksbank’s target. That has mainly happened on three occasions. On the first of these, 1996-1997, a high level of interest rates in the previous years had without doubt contributed to both a drop in the general level of demand and to low inflation. The period was also characterised by an occasionally heated debate on monetary policy. The reason the Riksbank decided to keep policy tight at the time was that credibility for inflation-targeting had not yet been established.
The two subsequent periods, 1998-1999 and 2004, were completely different. On both these occasions the repo rate had been relatively low the years before at the same time as the general level of demand was growing fast. The main factors generating the low inflation were changes on the supply side of the economy: deregulation, lower import prices and surprisingly rapid productivity growth. The Bank’s policy had hardly been called into question at all when it was conducted in 1996-1997; it was only in retrospect that the criticism came, and this has largely been the pattern in recent times, as I mentioned earlier. If one wants to argue that the Riksbank has systematically conducted excessively tight policy one should reasonably be able to show that other forecasters called the Bank’s policy strongly into question when it was formulated.
So, in my opinion, there are on the whole fairly natural explanations for why inflation on average has been somewhat below target over the past ten-year period. It is not because the Riksbank has considered deviations below target to be less serious than deviations above target; instead, it is partly because we have undergone an adjustment from a high-inflation to a low-inflation regime, which affected policy in large measure during the first of those three episodes, and partly because inflation on different occasions has been depressed by supply factors that neither we nor others have been able to predict.
Allow me to conclude the retrospective part of this speech by underlining that I value an open, lively debate on monetary policy. I have previously pointed out before the Committee that we would be happy to come here more often and that we are open to other forms of discussion than this one. If the Committee’s members have ideas to improve the material that we present, we would be interested in hearing them. The Riksbank is also considering initiating an annual forum for monetary policy debate, where our assessments and policy can be discussed against the background of regular independent evaluations of the forecasts that we – and other participants in the general debate – have produced.
The current assessment
Cyclical upswing continuing
With that I shall now look forward and briefly discuss the assessment of economic and inflation prospects in our latest Inflation Report and the decision we took yesterday regarding the repo rate.
As usual I shall begin with a brief look at the international situation. Economic growth abroad has remained high. There are many indications that international economic activity will continue to strengthen in the period ahead and that resource utilisation will gradually pick up. High productivity growth and weak resource utilisation have kept down cost developments in a number of industrialised countries. Coupled with the increased integration of low-cost countries into world trade this has contributed to low global inflationary pressures. In the period ahead, price pressures are expected to remain modest even though they will increase as economic conditions improve and there is a reduction in spare production capacity. That means that market interest rates are also anticipated to rise from today’s unusually low levels. There are positive signals emanating from the US economy while the euro area, and notably Germany, has seen somewhat weaker growth than expected. (Figure 6.) Overall, growth in Sweden’s export markets is forecast to be largely the same as in the previous Inflation Report.
The Swedish economic recovery began in 2003 and it is predicted to continue during the forecast period. In 2004 robust international demand for Swedish exports contributed to the high economic growth. Manufacturing activity now appears to be entering a slightly slower phase. Indicators and data point to a continuation of firm, healthy growth, albeit at a somewhat lower rate than in early 2004. Growth in domestic demand is assumed to increase while the contribution to aggregate demand from net exports will diminish. (Table 1). In 2005 business investment and private consumption are anticipated to account for the majority of the pick-up in demand. Compared with the December Inflation Report the forecast for GDP growth in 2005 has been revised up marginally to 3.2 per cent, mainly due to higher investment and increased services trade. Towards the end of last year investment in the economy grew more than expected and there is much to suggest that the fast increase will continue in early 2005.
Despite last year’s robust productivity growth the number of employed fell for the second consecutive year. The fact that firms were able to boost output so sharply without hiring new staff was due to a rise in both productivity and average working time. It is reasonable to believe, though, that a continuation of strong GDP growth will result in higher employment during the forecast period. Different indicators from the labour market also suggest that a turnaround is on the way. For example, the number of temporary employees has risen since summer 2004 and the number of new job vacancies reported to Swedish employment offices rose in annual terms in the period November-January after having dropped for almost two years in a row.
There is reason to expect average productivity growth in the economy to decrease somewhat as the economic upswing continues. Nevertheless, the Riksbank’s assessment remains that there has been a lasting increase in the rate of productivity growth, and the Bank’s forecasts for the coming years are therefore relatively high in a longer historical perspective. The forecast for domestic cost pressures this year has been revised down somewhat all the same compared with the assessment in December on the basis of new data for 2004.
Resource utilisation in the Swedish economy was low last year. In particular, the demand for labour is still limited even though some indicators of labour shortages show an upward tendency. Indicators from manufacturing also suggest that capacity utilisation has picked up recently. The Riksbank’s overall assessment is that resource utilisation has risen – albeit from a low level – and will continue to increase as more and more spare resources are put to use.
The inflation outlook
As I mentioned, both domestic and imported inflation were low in the past year. Domestic inflation was expected to be low at this point in time and the rate of price increases for domestically produced goods in recent months has also been broadly in line with the Bank’s previous forecast. In spite of this, the forecasts for domestic inflation this year and next year have been revised down. One reason is that the downward pressure on prices in the retail food trade is anticipated to be greater than previously assumed. How big the effects will be is difficult to say, though, as is their significance for the general inflation rate in the longer term. Over a longer horizon, domestic inflation is expected as before to rise gradually as the cyclical upswing continues and resource utilisation picks up.
Imported inflation has recently turned out lower than we anticipated in the previous Inflation Report. There are several reasons for this, which I touched upon earlier. For instance, increased imports from low-wage countries appear to have led to greater downward price pressure than previously forecast even though these projections were optimistic about price developments for imported goods. The price-dampening factors are also estimated to affect imported inflation in the period ahead, and the forecast has therefore been revised down both for this year and next.
The effects of the factors that have resulted in falling prices of imported goods in the past year are expected to gradually subside, however. It is reasonable to believe that rising global resource utilisation will lead to higher international price pressures. In addition, a gradual pick-up in domestic resource utilisation and mounting cost pressures are forecast to contribute to larger price increases for imported goods and services as well.
All in all, the forecasts for domestic and imported inflation mean that the forecast for UND1X inflation has been revised down compared with the Inflation Report in December. (Figure 7.) Both imported inflation and domestic price pressures are expected to be lower mainly in the near term but also during the latter part of the forecast period. The downward revision for the coming year is not due to any change in the assessment of economic prospects, but rather to a number of specific factors that have resulted in higher competition in the world market and in Sweden. But as the cyclical upswing continues and firms’ costs increase, there is still assumed to be a gradual rise in inflation. In the main scenario, which is based on an unchanged repo rate in the coming two years, both CPI and UND1X inflation is expected to be on target two years ahead. The annual averages for inflation will be low throughout the forecast period, however.
Allow me in this context to briefly say something about the fact that the weak price developments in the recent period have caused the word ”deflation” to reappear in the media, often in the form of an eye-catching headline. There is always reason to keep the risk of deflation in mind, since falls in the general price level are sometimes associated with considerable problems in the economy. It is important, however, to ascertain whether the subdued price developments are a result of weak demand or due to positive changes on the supply side of the economy. At the moment there is no doubt that the latter is the dominant factor. If a pair of jeans becomes cheaper because the dollar has fallen, or the price of plasma TVs has been halved because their production has increased in China, or the price of food goes down because the competition in the retail food trade has toughened, it is something that we essentially should be happy about. But should there be a sharp fall in demand then the situation would be completely different of course.
So what risks do we see that could cause inflation to deviate from the path in the main scenario? There are a number of risks in the inflation assessment that have been in focus for some time, e.g. the twin deficits in the US economy, the developments in the Swedish krona, and productivity growth. In addition, new risks have emerged in connection to specific factors, such as the developments in food prices, and these are creating great uncertainty over how inflation will be affected in the period ahead – not least in the short run.
Compared with the previous Inflation Report the main scenario on this occasion has taken account of the downside risks to inflation to a greater extent; e.g. increased international, as well as domestic, competition is now forecast to have a somewhat bigger impact on Swedish prices than previously assumed. There are arguments, however, that this could nevertheless entail an underestimation of inflation as much as an overestimation. That was shown, for instance, by the latest price data released last Friday, where the inflation outcome for February was actually a little higher than the forecast in the Inflation Report that we published today. Individual monthly figures for inflation are difficult to forecast, however, and next month the forecast could just as easily overshoot the outcome.
In the main scenario productivity is also assumed to follow the normal cyclical pattern, but just like the effects of the international competition there are arguments to suggest that this pattern may imply an underestimation or overestimation of the future trend. The overall assessment of the different risk factors is that the risk outlook is balanced, i.e. that the risks of inflation turning out higher than in the main scenario are as large as the risks of inflation turning out lower. Thus, the balance of risks does not change our assessment. Inflation in the coming two years will be low, coming into line with the target at the end of the forecast period.
The point of departure for our discussion about the future path of the repo rate is the assessment of inflation prospects in the coming years that I just presented. Allow me in this context to point out that the Riksbank does not conduct monetary policy according to a mechanical rule, even though that sometimes seems to be the perception. There are also other aspects than inflation one to two years ahead that must be taken into account when setting the repo rate.
Last year we encountered a similar situation to that today, when the Bank's forecasts suggested that inflation would undershoot the target for almost the entire two-year period ahead, even though it was estimated to be on target in two years’ time. One reason we decided not to cut rates then – a reason that is still valid today – was that we saw that demand was growing strongly already. So, monetary policy appeared to have had the desired effect. This development has been reinforced now that business investment seems to have begun to pick up in earnest and that private consumption is also stronger. In other words, economic activity is already firm and GDP growth is above what is judged to be sustainable in the long term. Given that, a rate cut now would probably mean in practice that we would have to tighten policy more quickly in the future. So it is a question of trying to avoid excessive fluctuations in the real economy and interest rates.
Another, partly related aspect that we also have reason to consider when making decisions about the repo rate is that the low inflation is due in large measure to markedly weak price developments for certain products as a result of increased competition, e.g. in the retail food trade and textile industry. It is probably not always prudent to fully counter this through highly loose policy that would drive up domestic demand. Should that price outlook change, we may find that excessively high price pressures have been established, which may then prove difficult to break. It remains to be seen how long the effects on inflation of the price cuts for certain products will last. This argument, too, can essentially be said to involve an ambition to avoid excessive swings in interest rates and the real economy.
Yet another factor that has been discussed for some time by the Executive Board is the rapid growth in household debt and the increases in house prices. A number of commentators have begun to highlight risks in this area. We have previously concluded that the developments are not a cause for great concern. Even though that largely remains our assessment, it has gradually become clearer that the developments in credit and house prices are one argument against looser monetary policy. A rate cut followed by a faster hike could bring about problems through their effects on household indebtedness and consumption.
The inflation assessment I presented earlier indicates that inflation will be in line with the target at the end of the two-year period on which we normally focus. At the same time the issues that I have just discussed give cause for a certain amount of caution in interest rate policy in line with the Bank’s clarification of monetary policy in 1999. All in all, the arguments I have put forward advocate leaving the repo rate unchanged, which was also the decision we took yesterday.
As usual there is reason to underline that even though this is our assessment today conditions can of course change in the period ahead. A lot can happen: the imbalances in the United States could result in financial unease that sharply curtails the international economic upswing, or the credit and debt developments in Sweden may prove considerably more worrisome than foreseen today. We may also find that we have overestimated or underestimated the impact on inflation of changes in competitive pressures. So, claims in the media that the Riksbank has closed the door on a particular course of action in the future are misleading as usual.
Allow me to conclude with a few general reflections on the conditions under which monetary policy operates and regarding the expectations one can have on it. Even though developments since the shift in policy regime in the early 1990s have been remarkably good this has, as we now know, neither stopped inflation from occasionally deviating sharply from target nor prevented output and employment from fluctuating somewhat.
What is important to realise is that there are always going to be fast changes in the economy and inflation that cannot be foreseen nor fully offset in the short run. No central bank can therefore have full control over inflation and economic activity over time. What monetary policy should normally be able to achieve is to ensure that inflation over a number of years is fairly well in line with the inflation target. This should also improve the chances of ensuring stable developments in demand in the economy. That in turn should help smooth the fluctuations in the cyclical component of unemployment, even though the relationship between demand and unemployment does not always have to be especially strong, as demonstrated in particular by recent developments.
It is just as important to realise that monetary policy, and for that matter stabilisation policy in general, cannot be used to permanently raise growth and reduce unemployment. If the aim is to achieve a permanently higher employment rate and lower unemployment, it is essentially a question instead of trying to influence the structural component of unemployment, i.e. the component that does not depend on fluctuations in demand for goods and services in the economy. And that requires completely different measures than those of monetary policy. Examples include how to improve the business climate and how to make the labour market as efficient as possible.
It is vital to discuss and evaluate monetary policy, but it is also important that the discussion of economic policy and developments on the whole be based on a realistic view of what monetary policy can accomplish. If so, the discussion is more likely to help us in our efforts to conduct monetary policy in the best way possible.