Rosenberg: Monetary policy and the labour market
Introduction
Thank you for the invitation to come here!
After the economic crisis and the changeover in economic policy at the beginning of the 1990s, the Swedish economy has functioned much better than anyone could have imagined. GDP growth has been higher than in the 1970s and 1980s, inflation has been low and more stable and real wages have increased much more quickly than during both of the previous decades. However, developments in the labour market have not been as favourable. Unemployment rose tangibly in connection with the crisis in the 1990s and has since then remained at a higher average level than before.
Inflation has on average been slightly lower than the Riksbank’s inflation target since it was introduced in 1985. During the years 2004-2005 inflation was noticeably below the target. It may therefore be relevant to claim that the high unemployment is largely due to incorrectly balanced monetary policy. There are also debaters who have implied that an economic policy that aims at low inflation can in itself contribute to unemployment being higher than necessary. I intend to devote my speech today to explaining my views on this. I also intend to briefly comment on how I viewed the current situation in the labour market in connection with the monetary policy meeting in May and what significance this had for my stance at the meeting. However, I will not give any new monetary policy signals.
But first I shall begin by providing some historical perspective on this issue by giving an account of the background to the economic policy change of regime at the beginning of the 1990s.
From fine-tuning stabilisation policy to "stable game rules”
Unemployment is currently high in comparison with the levels that prevailed during the 1970s and 1980s. Open unemployment at that time varied between 1.5 and 3.5 per cent. In the mid-1990s open unemployment went up to just over 8 per cent, while during recent years it has varied between 4 and 6 per cent. However, the factors that contributed to holding down unemployment during earlier decades at the same time contributed to the economic crisis at the beginning of the 1990s and also to the need for an economic policy change of regime.
From the 1950s up to the mid-1970s the authorities in Sweden attempted to conduct what is usually known as fine-tuning stabilisation policy. This involved trying to parry fluctuations in economic activity with an active fiscal policy. This policy was relatively successful up to around the mid-1960s. With the target of full employment in their line of vision, politicians were then tempted to conduct an economic policy that was on average too expansionary. It also contributed to a growing central government debt. The economy was fuelled during economic upturns, which led to overheating and inflation. The Riksbank’s task was to try to defend the fixed exchange rate. The social partners were instead to contribute to low inflation by ensuring that wage increases were not too high.
Soaring wage costs and poorer international competitiveness meant that the system broke down in the mid-1970s. To safeguard the objective of full employment, the krona was devalued by a total of around 40 per cent on five different occasions between 1976 and 1982. Each devaluation held unemployment down temporarily, but instead contributed to kindle the inflation fire until it was time for the next devaluation. Unemployment was also held down in that the number of employees in the public sector increased substantially. The number of persons employed in the public sector increased from 15 to 30 per cent between 1960 and 1980. In my opinion, unemployment was held down during this period using methods that were not sustainable in the long term. I consider this to be the first important aspect to bear in mind when discussing why unemployment is now much higher than it was in the 1970s and 1980s.
Towards the end of the 1980s the situation became untenable. Everyone was then counting on the krona being devalued if the rate of increase in prices and wages rose too high. The driving forces for holding down prices and wages had thus disappeared. The speculations against the krona finally led to the Riksbank, after using large interest rate increases to defend the krona, was forced to let the krona float in autumn 1992. We suffered a crisis and unemployment soared. An insight was gradually reached that a new economic policy approach was necessary. Earlier experiences and even research indicated that a healthy economic policy should be based on ”stable game rules” in the form of orderliness in public finances and low, stable inflation. These two principles became the mainstays of the new economic policy framework produced at the beginning of the 1990s. Fiscal policy was subjected to a requirement of long-term sustainability. Monetary policy was given the task of safeguarding price stability.
Monetary policy aims at price stability
One may sometimes gain the impression in the general debate that the economic policy change of regime in the early 1990s meant that monetary policy was given responsibility for stabilisation policy in a broader sense. Those who argue that this is the case think that the Riksbank should give greater consideration to developments in the labour market. I therefore think it is important to point out that the objective of monetary policy is low and stable inflation. This was made clear in the new Sveriges Riksbank Act that came into force in 1999.
The act states that the objective is to maintain price stability but this is not specified in greater detail. The Riksbank has therefore specified a target, whereby inflation should be 2 per cent. To emphasise that it is not possible to continuously achieve the target exactly, the Riksbank has also specified a tolerance interval around the target of plus/minus 1 percentage point. The Riksbank’s inflation target is well in line with the targets chosen by other central banks.
The fact that we have a target for inflation does not mean that we disregard developments in employment and unemployment. The Riksbank conducts what is usually called flexible inflation targeting. This means that we take real economy developments into consideration when making monetary policy decisions. If inflation for some reason deviates from the target, we must decide how quickly inflation should be brought back on target. This depends on how, among other things, unemployment is affected. This is also in line with the preliminary work to the Sveriges Riksbank Act. It is stated there that the Riksbank shall support the targets for general economic policy with a view to achieving sustainable growth and employment without prejudice to the objective of price stability.
There is no simple correlation between inflation and unemployment
Unemployment varies over the economic cycle. Normally, unemployment is below its so-called long-term equilibrium level when economic activity is good. A high level of resource utilisation in the labour market normally also means that the pressure from wage increases is high. This in turn has the short-term effect of pushing up inflation. In economic downturns, however, unemployment is high and resource utilisation is low. This usually contributes to lower wage and inflationary pressures in the short term. The Riksbank therefore also closely follows resource utilisation in the labour market.
In the short term, a central bank can to some extent lower unemployment by increasing demand and inflation with expansionary monetary policy (in the Riksbank's case, by cutting the repo rate). This leads to companies increasing production and taking on more employees. Using the same reasoning, unemployment would therefore rise if inflation was pushed down by tightening monetary policy. It is this relationship that is usually described with the aid of what is known as the short-term Phillips Curve.
Some of the criticism aimed at monetary policy in recent years appears to be based on the assumption that there is a clear-cut long-term relationship between inflation and unemployment. However, it is not possible to buy lower unemployment in the longer term by allowing inflation to rise. Sooner or later wage-earners will require compensation for the higher inflation by raising their wage demands. As prices and wages rise, the stimulating effect of monetary policy will wane. Demand will fall again and employment will decline. In the long term, unemployment will return to its long-term equilibrium level. A systematic, expansionary monetary policy would only have the consequence that inflation and inflation expectations would soar. This correlation is usually illustrated with the aid of the so-called long-term Phillips curve. There is currently a broad consensus on this, both in central bank circles and the academic world. In my opinion, it is therefore not possible to claim that an economic policy that aims for low and stable inflation would in itself lead to higher unemployment.
Another part of the criticism is based on the assumption that it is not possible to buy lower unemployment by pushing up inflation, although there is instead apparently an assumption that if only the Riksbank would ensure that inflation remains at 2 per cent, then unemployment would stabilise around its long-term equilibrium level. In practice, however, there is no such simple relationship between inflation and unemployment. Firstly, unemployment is affected by a number of factors other than inflation, most of which have nothing to do with monetary policy, such as productivity changes, changes in regulatory systems and various shocks that regularly impact on the economy. Similarly, inflation is affected not only by developments in the labour market, but by many other factors. Empirical studies have shown that unemployment “explains” a relatively small percentage of total inflation. This means that inflation developments are largely “explained” by other factors than unemployment. Secondly, it has proved very difficult to assess what is a long-term equilibrium level for unemployment and moreover that this also varies over time. One conclusion of this is that even if the Riksbank were to always succeed in holding inflation at the target, this would be no guarantee that unemployment would be low and stable. Unemployment is instead determined in the long run by how well the labour market functions, for instance, with regard to matching unemployed persons and job vacancies.
My main message here is that there is no clear-cut and stable relationship between inflation and unemployment, either in the long term or the short term. Simplified relationships between inflation and unemployment, such as the short-term and long-term Phillips Curves, may function as pedagogical tools. However, this type of simplified relationship is not practicable in determining how monetary policy should be conducted. Nor can such simple relationships be used to analyse to what degree monetary policy has contributed to higher or lower unemployment.
Monetary policy is based on forecasts
Let me go on to the next issue I intend to take up, namely the conditions under which monetary policy is conducted. This is an important aspect of a discussion of how far monetary policy can be considered to have contributed to higher unemployment.
As monetary policy has an impact on demand and inflation with a time lag, the Riksbank must base its interest rate decisions on a forecast of future economic developments. All forecasts are uncertain. One therefore cannot expect that the Riksbank will always be able to exactly predict what will happen in the economy. It is also difficult to quickly bring inflation back on target once a deviation has occurred. This may also be unsuitable for real economy reasons. It means that it is not reasonable to evaluate monetary policy in the basis of the opinion that inflation must always be on target. One should rather assume that inflation will rarely be exactly on target. In addition, the Riksbank deliberately allows inflation to deviate from target during shorter periods. This has been the case, for instance, when there have been temporary changes in inflation that we have not considered to warrant being counteracted by monetary policy. For example, we assumed that inflation would be below target at the beginning of 2004 due to energy prices having temporarily pushed up price in early 2003. However, inflation turned out to be even lower in 2004 than we had expected. But what I would like to say here is that one must expect inflation to deviate from target sometimes and that there are good reasons for this. I therefore consider that it is wrong to claim that monetary policy contributes to higher unemployment just because inflation is below target during a particular period.
Another thing one can expect from the Riksbank is that we should make the best forecasts we possibly can. It is also important that our forecasts are evaluated. Last year our monetary policy was evaluated by two external professors, Giavazzi and Mishkin. They considered that the Riksbank’s forecasts for inflation compare favourably with those of other forecasters. However, for both 2004 and 2005 both we and other analysts overestimated inflation. This leads me on to some other questions I intend to take up. Why did we overestimate inflation in our forecasts for those years? To what extent is the fact that inflation has undershot the target due to incorrectly formulated monetary policy, and what effects could this have on unemployment?
Why has inflation undershot the target?
Between 1995 and 2006 CPI inflation was on average 1.3 per cent. Measured in terms of UND1X, which is what we normally base our monetary policy on, inflation amounted to an average of 1.7 per cent during the same period.(1) One reason why CPI inflation has been lower than UND1X inflation is that the repo rate has been regularly cut during this period. Regardless of what measure is used, the legal objective of price stability has been achieved. However, inflation has on average undershot the target the Riksbank itself has set.
During 2004-2005 inflation deviated substantially from the Riksbank’s target. Some debaters have, as I mentioned at the beginning, claimed that this has to a large degree contributed to the high unemployment in recent years. Of course, it is true that incorrectly formulated monetary policy could lead to unnecessarily high unemployment. But to be able to assess monetary policy in this regard, it is important to first find out why inflation has undershot the target. It is also important to examine what consequences these factors may have had for unemployment, in addition to the effects related to the formulation of monetary policy.
The analyses made of why inflation has been so low in recent years indicate that it is primarily unexpectedly high and lasting growth in productivity that is the explanation and not contractionary monetary policy.(2) One factor that supports this conclusion is that inflation has been low despite demand and growth in the economy being relatively high. But the higher labour productivity has meant that companies have been able to increase their production without needing to raise their prices so much. It would also appear that it has contributed to companies being able to increase production without needing to recruit new staff to a great extent. This may be an explanation to what is referred to in the debate as “jobless growth”. The high productivity growth has also contributed to the Riksbank being able to cut the interest rate to a historically low level.
My conclusion is therefore that we have first and foremost overestimated inflation because we have underestimated the strength of productivity growth. Giavazzi and Mishkin observed in their evaluation of monetary policy that this has also applied to most other forecasters. We have therefore not overestimated inflation because we have failed to take into account developments in the labour market. Nor is it because our monetary policy decisions have been unreasonable. This was also noted by Giavazzi and Mishkin in their evaluation.
Nevertheless, there is of course reason for the Riksbank to be self-critical. We must become better at understanding the driving forces behind productivity growth. In 2005 we started a project at the Riksbank that is aimed at increasing our understanding of the strong growth in productivity in Sweden in recent years. More specifically, we are studying the relationship between IT investments, organisational changes, internal training and productivity. At the beginning of May this year we organised a seminar where economists from a number of organisations presented their views on developments in productivity. In December we will also be organising a workshop where external researchers will contribute valuable points of view. Hopefully this will make us wiser with regard to developments in productivity growth. However, at the same time, one must have realistic expectations of what can be achieved here; it will probably always be difficult to assess future productivity growth.
What significance would lower interest rates in recent years have had for unemployment?
There are thus natural explanations as to why inflation has been so low in recent years. With the picture we ourselves and most other forecasters had of inflation prospects during 2002-2003, it would probably have been difficult to motivate a much lower interest rate than we chose. With hindsight, it can of course be noted that the repo rate could have been cut more quickly. We have pointed this out ourselves in our own assessments.(3) The question is what this could have meant for unemployment. The Riksbank’s calculations imply, for example, that an interest rate a half a percentage point lower for six months would have led to lower unemployment corresponding to two or three tenths of a percentage point. However, I would like to emphasise that this type of calculation is very uncertain.
Given this, I believe that monetary policy cannot be said to have contributed significantly to the high unemployment. What other possible explanations are there?
What can explain the rise in unemployment?
As I mentioned initially, unemployment was held down in the 1970s and 1980s with an economic policy that was not sustainable in the long term. I believe that this could have concealed structural problems in the labour market that only became visible when the economic policy regime was changed in the early 1990s.
If one studies unemployment, particularly total unemployment, over a long period of time, one can see that it has shown a trend increase ever since the end of the 1960s. This also applies if one excludes the years after the change in economic policy regime. Open unemployment did not fall more than to around 4 per cent at the peak of the most recent economic upturn. This also indicates that there are structural problems in the labour market that have also increased over time. There are, for instance, signs that the matching of unemployed and vacancies has deteriorated after the 1990s crisis. Increased globalisation and stiffer international competition, and the ensuing structural transformation, may also have played a role. The pressure for change in the labour market has been strong and adjustments have not been made as smoothly as might have been desirable.
It is evident that the high unemployment that arose in connection with the crisis has to some extent been persistent. Research shows that the long-term unemployment level may be affected by major changes in actual unemployment. There is also research that show that generous unemployment insurances may contribute to the persistency of unemployment. Especially after major structural crises have occurred, such as the one in Sweden at the beginning of the 1990s.(4)
I shall not develop this further. The functioning of the labour market lies outside the Riksbank’s field of responsibility and there are others who are better qualified to discuss the problems in this area than I am. My main message is that there are many indications of considerable structural problems in the Swedish labour market which may explain the high level of unemployment. These problems cannot be solved by monetary policy. Monetary policy’s effects on unemployment are minor in relation to those due to structural problems.
The current situation in the labour market
Let me before I finish briefly comment on how I viewed the current situation in the labour market in connection with the monetary policy meeting in May.
A large number of central wage agreements are being renegotiated this year. There are several factors governing how high wage costs can be expected to be in future. One important aspect is to what extent the social partners take into account the labour market situation in their wage bargaining rounds. Another is how well the labour market functions with regard to matching the demand for and supply of labour.
The wage bargaining rounds have so far resulted in wage increases that appear to be slightly above what we had estimated in the February Monetary Policy Report. We pointed this out at the monetary policy meeting on 3 May. The labour market situation has strengthened and it is therefore reasonable to assume in future that there will be slightly higher wage increases than we have been used to in recent years.
The information that we had access to at our monetary policy meeting in May indicated that resource utilisation in the labour capacity was slightly more strained than we anticipated in the February report. There were signs that the labour shortage in the business sector has continued to increase. However, it is not exactly easy to assess the level of resource utilisation in the labour market. There are many indications of fairly large labour reserves in total, although this may not be reflected in the recently agreed wage increases.
Employment appeared to be developing in line with the forecast we made in February. However, the labour supply was showing weaker development than we had expected. Altogether, this means that unemployment has been slightly lower than we assessed in our forecasts in February and the labour market situation is therefore slightly more strained.
Our assessment in May was that on the whole, the repo rate would need to be raised gradually and by more during the forecast period than was indicated by the forecast we made in February. I did not consider the changes in the economic outlook to be so great that they required an immediate adjustment in the repo rate. The reason for this was that inflation is low and there are still factors holding it back. What we must do is to balance the pace of our interest rate increases so that inflation does not accelerate more quickly than expected, but so that growth in production and employment are not slowed down too soon. My proposal was therefore to keep the repo rate unchanged in May.
What significance the new information will have in more concrete terms for monetary policy in the future is something we will return to at our next monetary policy meeting in June.
Conclusion
Let me conclude by briefly summarising what I have said. I have mainly focussed on the debate on monetary policy and unemployment. I have also briefly described the current labour market situation, as I assessed it at the previous monetary policy meeting.
I think it is good and necessary that a debate is held on the significance of monetary policy for unemployment. But I consider it essential that this debate is conducted on the basis of reasonable expectations of what monetary policy can and should achieve. The fact that inflation has undershot the target in recent years is mainly due to unexpectedly high productivity growth. The high productivity has been beneficial as it has contributed to good growth. However, it may at the same time explain why employment has shown relatively weak development for some years.
It is not because of monetary policy that unemployment now is much higher than it was in the decades immediately prior to the 1990s crisis. There are many indications that the explanation for this is structural problems in the Swedish labour market. Problems that have increased over time and were long concealed by the unsustainably expansionary economic policy conducted earlier.
With regard to the more current developments in the labour market, unemployment looks to be slightly lower and wage increases to be slightly higher than we assumed in the previous Monetary Policy Report in February. We will present a new assessment of economic activity and inflation prospects in our next Monetary Policy Report in June. This will also show our views on the labour market situation and what consequences this will have for monetary policy.
Thank you for your attention!
Notes:
1)When one measures inflation with the aid of UND1X, one adjusts CPI for the direct effects of changes in indirect taxes and changes in mortgage interest costs.
2)See Giavazzi, F. and F.S. Mishkin (2006), "An evaluation of Swedish monetary policy between 1995 and 2005”, Reports from the Riksdag 2006/07:RFR 1, Committee on Finance. See also Sveriges Riksbank (2006), “Material for assessing monetary policy 2003-2005”, box in Inflation Report no. 1 and Sveriges Riksbank (2005), “Material for assessing monetary policy 2002-2004”, box in Inflation Report no. 1.
3)See the references in note 2.
4)See, for example, Ljungqvist, L and T J Sargent (2006), ”How Sweden’s unemployment became more like Europe's” in Freeman R, B Swedenborg and R Topel (ed), ”Reforming the welfare state – American perspectives on the Swedish model”, SNS Förlag.
