Bäckström: Monetary policy and unemployment
First a word of thanks for the invitation to this evening's discussion. This is the third year in succession that your Association's agenda has featured monetary policy. My first introductory talk, in 1995, was about the importance of low inflation and the background to and implications of focusing on price stability. Last year I took up matters to do with monetary policy's construction and effects, as well as the considerations behind the monetary policy decisions in recent years. This evening I shall be elaborating on those discussions and then deal with monetary policy in the context of Sweden's high unemployment.
Economic developments in Sweden
Following the deep recession in the Swedish economy in the early 1990s, we are now in the fourth year of economic expansion. Since 1994 production's annual growth rate has averaged 2.8 per cent, which is considerably above the trend to which we have been accustomed in recent decades.
In 1994 and 1995 the GDP growth rate averaged as much as 3.5 per cent. This increase was concentrated to the internationally oriented sector and industrial activity registrered the fastest upswing in a long time. This led to capacity problems despite relatively high investment growth. The strong expansion was therefore not sustainable.
Moreover, the credibility of economic policy was impaired by weak central government finance and an unfavourable history of inflation. There was a lack of confidence that the expansionary trend could continue without generating higher inflation. Inflation expectations moved up--more and more people began to count on rising inflation. All this was manifested in, for example, rising long-term interest rates, a weak exchange rate, rising producer prices and accelerating wage increases.
The Riksbank therefore tightened interest rate policy, beginning in August 1994. The restrictive turn was criticised in some quarters because inflation had not yet started to accelerate. But our intention was precisely that. If monetary policy measures are taken in time, the inflationary process can be checked before it has taken off. Once inflation has got going, more pronounced monetary policy measures are needed for a return to low inflation.
Vigorous efforts have also been made over these years to tackle the central government financial deficit. With its six-monthly control stations, the convergence programme enables people to keep an eye on the political system's determination to complete the consolidation of the budget and fulfil the goal of balance in 1998.
Government representatives have also expressed an ambition to achieve a surplus after that. With the improvement in central government finance, confidence in economic policy grew appreciably and inflation expectations fell back notably in the autumn of 1995 and during 1996. This in turn is a fundamental prerequisite for lastingly stable growth. During 1996 the Riksbank was in a position to reduce the repo rate by degrees to the lowest level for more than thirty years.
Towards the end of 1995 the growth of production slackened and the increase during 1996 probably stopped at between 1 and 1.5 per cent. However, the slowdown seems to have been relatively gentle. The demand situation appears to be good, particularly when one considers that the growth of private sector production was held back by marked destocking in 1996. Stock adjustments are not uncommon after some years of rising output and they do not seem to have had negative effects on private consumption. In the second half of 1996 private consumption picked up and export demand went on rising. As the destocking comes to an end, domestic and international demand should therefore be able to generate an accelerating growth of production this year and in 1988.
Considering the economy's unutilised resources, I do not expect inflationary capacity restrictions to arise in the coming years. If inflation expectations also remain low and confidence in economic policy in general is stable, this, too, will contribute to a subdued development of future inflation.
Since the autumn there have been some months in which the level of consumer prices was lower than a year earlier. In a price analysis, however, it is important to distinguish transitory effects on inflation from the underlying inflationary process. In January this year the level of consumer prices was 0.1 per cent lower than a year earlier. Strictly speaking, however, what Sweden is experiencing is not deflation. The falling price level has to do with transitory effects, above all the Riksbank's marked reduction of the repo rate during 1996. Excluding house mortgage interest costs, the rate of inflation in January 1997 was about 1.5 per cent.
The outlook for inflation points to room for a fall in bond rates. Moreover, underlying economic factors suggest that, after the depreciation in recent months, the krona is undervalued and should be capable of appreciating.
The present monetary stance appears to be well balanced. After the major changes in the past year, for example in the repo rate, time is needed to assess their effects. Another matter to consider is the recent months' developments in financial markets. They, too, motivate a cautious line in monetary policy.
I note that the Swedish economy seems to have made good progress in adjusting to a low inflation regime, with stable, sustained growth. What gives cause for concern, however, is that the recovery of employment appears to be much more sluggish and that unemployment is persistently high.
High cost of unemployment
Diagram 1: Total and registered unemployment in Sweden 1965-96
As shown in Diagram 1, registered unemployment in Sweden rose sharply in the early 1990s; from a level between 2 and 3 per cent, it shot up to around 8 per cent. This was accompanied by increased participation in labour market programmes. Total unemployment accordingly reached around 13 per cent, an increase of almost 10 percentage points or up to half a million people. This is the highest level since the 1930s. It is primarily among youth that unemployment is high, as shown in Diagram 2. Sweden seems to have landed in much the same situation as many other countries in Europe.
Diagram 2: Age-related unemployment
High unemployment is associated with grave social problems and large economic costs.
Unemployment's long-run economic costs in terms of GDP can be said to be of approximately the same magnitude as the unemployment rate. Sweden's increase in unemployment can in this way be estimated to cost around 10 per cent of GDP a year, which is equivalent to SEK 150 billion. Considering the sensitivity of general government finance to variations in unemployment and production, the cost to the public sector, in the form of lower tax revenue and increased expenditure on unemployment support, for example, may actually be as high as 9 per cent of GDP or SEK 135 billion.
Conversely, an appreciable reduction of unemployment implies decreased social tensions as well as substantial economic gains. Calculated in the same way, halving registered unemployment from 8 to 4 per cent, for example, would confer an estimated long-term gain in production of 4 per cent of GDP and strengthen government finance by as much as over SEK 50 billion a year.
Simple calculations of this type are obviously uncertain but they do serve to illustrate the sort of economic benefits that may be reaped from an appreciable fall in unemployment.
What can the Riksbank and monetary policy do?
There was a time when many people believed that the problem of very high unemployment could be tackled with a generally expansionary economic policy. Lower interest rates, devaluations or fiscal stimuli were seen as instruments for boosting demand, raising production, improving employment and thereby lowering unemployment. Even today, advocates of similar views are not uncommon. As the consolidation of the budget places constraints on fiscal policy, attention concentrates on monetary policy. Representatives of this line argue that the problem of unemployment could be solved if only the Riksbank had a wider objective than combating inflation. Some say that the Riksbank ought therefore also to have explicit goals for growth, employment and unemployment, goals that should be as important as keeping inflation under control.
You will hardly be surprised to hear that I do not share such opinions. The current situation in the labour market is more complicated than that.
An expansionary economic policy, aimed at boosting demand above the economy's long-term production potential, would actually be liable to worsen the labour market situation in the longer run. Inflation would move up and the economy would function less well. Neither is it the case that higher inflation would generate permanently higher employment and lastingly lower unemployment. Let me give you an example.
Suppose inflation is 2 per cent initially and monetary policy is given an expansionary turn that disregards the sustainable level of output. If wage agreements have been concluded for a considerable period, an upward shift in inflation to, say, 5 per cent could in the best case contribute to a short-run increase in employment because firms' real wage costs are reduced. In time, however, 5 per cent inflation will be perceived as normal and wage demands will be pitched accordingly. With higher nominal wage increases, real wage costs will rise again and unemployment will tend to return to its initial level. A renewed attempt to stimulate employment and reduce unemployment would then entail a further acceleration of inflation from 5 to, say, 8 per cent and the procedure would be repeated. Ultimately, all that is achieved is higher average inflation and an economy than functions less efficiently; unemployment stays at its initial level.
The level of unemployment to which an economy returns when inflation expectations have been adjusted to actual inflation is usually referred to as equilibrium unemployment. It is immune to monetary policy and determined primarily by structural factors such as demographic changes, technology and the institutional conditions in the labour market. What I have just described is the textbook model for the long-term Phillips curve.
The message is that an economy has a path for long-term growth that is determined by underlying, structural factors. Under certain circumstances, fiscal and monetary policy are capable of influencing fluctuations in activity around this trend, though one should not expect to be able to fine-tune demand. A basic idea behind a monetary policy focused on price stability is that the economy is capable of sustained growth along this long-term path. If this path accommodates a rapid increase in production, growth will be high; but if potential output rises slowly, actual growth will be weak even if monetary policy were to be very expansionary and possibly affect the situation temporarily.
A very important conclusion from this is that permanently low unemployment cannot be achieved by accepting permanently high inflation. In other words, sustained employment cannot be generated by implementing an expansionary monetary policy and disregarding unemployment's equilibrium level or the potential level of output. The narrower the gap between actual production and the capacity ceiling and the lower unemployment happens to be in relation to its equilibrium level, the greater will be the risk of higher inflation.
From this also follows a need to know whether the equilibrium level of unemployment has risen. If that is the case, measures should be taken specifically to lower the equilibrium level, a task that is outside the scope of monetary policy.
Where is unemployment's equilibrium level?
What is unemployment's current equilibrium level in Sweden? As always when one goes from a theoretical analysis to practical economics, everything becomes much more complex. Unemployment's equilibrium level is not an observable quantity. One has to make do with estimates and their inevitable uncertainty means that interpretations have to be cautious.
The available estimates suggest that equilibrium unemployment in Sweden moved up in the early 1990s and, in terms of registered unemployment, is now somewhere in the interval between 4.5 and 7.5 per cent. The OECD also counts on a marked increase in the 1990s; in the latest report on Sweden the current level is put at between 6 and 7 per cent.
There is not, in fact, complete agreement that equilibrium unemployment has moved up sharply. But what suggests that it has, even though we cannot be sure of its exact level, is that otherwise the high unemployment should have acted as a strong restraint on wage formation. The actual development of wages does not indicate this.
It therefore has to be concluded that unemployment cannot be lowered appreciably by monetary policy alone. Most economic observers also consider that even with decent growth in the coming years, unemployment will remain high. A lasting return just to something resembling the unemployment figures we have been accustomed to in Sweden therefore requires a proper discussion of which measures are needed to bring equilibrium unemployment down.
The picture in the USA, Western Europe and Sweden
It is not just in Sweden that unemployment is high. This is something which is leaving its mark on virtually the whole of the EU area. Moreover, the level of European unemployment has been high during most of the 1980s. In the United States, on the other hand, unemployment has fluctuated around 6 per cent for the past quarter-century without any upward trend.
The direct explanation for the difference between the EU area and the United States is the appreciably faster growth of employment over there. In the past quarter-century and more, employment in the United States has risen over 60 per cent as against only 15 per cent in the EU area. The number of new jobs certainly grew by some per cent in Europe when activity was recovering but almost as many jobs disappeared in the downturns. In the United States, on the other hand, over the business cycle there have been net increases.
Economists differ as to the underlying causes. For some economists, equilibrium unemployment has moved up on account of underlying structural factors. One hypothesis, put forward by, among others, the American economist Paul Krugman, is that certain categories of labour (low educated, for example) are having increasing difficulty in competing in the labour market. This is mainly bound up with a technology that replaces manual labour with machines but presumably it also has to do with increased competition from countries where wage levels are considerably lower. It is often pointed out that the effects of this are countered by a wider wage spread that in turn has stimulated an expansion of the private services sector. Employment has therefore risen rapidly in the United States. In Europe, on the other hand, the result has been increased unemployment; the labour market and wage formation here have not permitted the same development as in the United States.
A similar effect, leading to a stronger employment trend, may have come from a lower average real wage rise in the United States compared with Europe. The steep real wage increase in the EU area has favoured capital-intensive sectors and stimulated the use of capital-intensive technology. At the same time, low-productivity enterprises have shrunk or closed down. This has shown up in increased productivity and weakly rising employment. The opposite has occurred in the United States: sluggish real wages have limited the need to rationalise labour and invest in capital-intensive equipment and this has resulted in rapidly rising employment.
Diagram 3a: Real wage and employment, USA and EU, 1970-97
As shown in Diagram 3a, since 1970 the average real wage in the United States has risen by a total of 15 per cent but by as much as 60 per cent in the EU area. The change in employment, as I mentioned earlier, has been the opposite.
Diagram 3b: Real wage and employment, Netherlands and Sweden, 1970-96
Europe, however, does have some positive examples. One of the few countries here that has managed to turn the employment curve upwards from the mid 1980s is the Netherlands, as shown in Diagram 3b. In the course of a decade unemployment has been virtually halved, from 12 to just over 6 per cent. One factor behind this is a very moderate real wage rise, so that employment has risen on this account. The labour market has probably become more adaptable. Another phenomenon in the Netherlands is an increase in part-time employment; this probably reflects a change in labour demand but also a shift in supply. The case of the Netherlands shows that the employment trend can be turned.
This indicates, in my opinion, that in the current discussion in Sweden about a new model for wage formation, the prototype ought not to be Europe as a whole. If the average real wage rise in Sweden were to match the rate observed in Europe in the past quarter-century, it might make a reduction of unemployment more difficult. The model should rather be the Netherlands.
The pattern in Sweden resembles that for the EU area. In the period as a whole employment has hardly changed. As in Europe in general, real wages have risen relatively strongly, though the trend has been held back by a series of devaluations that pushed inflation up, presumably with a negative effect on production.
Diagram 4: Cumulative change in employment in Sweden 1960-96 in the private and public sectors and in total
The great challenge Sweden faces in the future development of employment is evident from Diagram 4, which shows that for a long time there has been no net increase in private sector jobs. This is the trend that must be broken. While there has certainly been some increase in employment in boom years, in the subsequent downturns virtually as many jobs have disappeared. The overall increase in employment has come from an expansion of public sector jobs; one million jobs were created in this sector between 1960 and 1990. Now that, for various reasons, the political system has been obliged to arrest the public sector's expansion, employment has also fallen.
Diagram 5: GDP in the USA, EU, Netherlands and Sweden, 1970-96
It is important, I think, to underscore that, notwithstanding the marked differences in employment, GDP growth does not differ substantially between the United States, the EU area, the Netherlands and Sweden. Therefore it is not high demand generated by a more expansionary monetary and fiscal policy that can account for the creation of so many new jobs in the United States or, in the past decade, in the Netherlands. This is an indication that structural factors lie behind the development in Western Europe as a whole. An important part seems to be played, not least, by wage formation.
The weak growth in Sweden is also evident from Diagram 5. This is another interesting observation in the present context. When potential output rises slowly for structural reasons, growth will be weak even if economic policy is expansionary. In spite of devaluations and fiscal expansion with a view to boosting demand, growth has remained lower than in many other industrialised countries. This is an important lesson to bear in mind when discussing monetary policy's functions and tasks.
Other economists have been less interested in why employment has been slack and unemployment has risen and more concerned with unemployment's tendency to remain at a high level. The observation that unemployment, having risen, tends to stick at a high level or falls back to normal levels only gradually is usually described in terms of hysteresis and persistence. An explanation for the occurrence of persistence tendencies is provided by the insider-outsider model. Briefly, an increase in unemployment does not lead to wage restraint because those who have lost the jobs–the outsiders–exercise no influence on wage setting. Instead of a restrained wage rise, with lower labour costs and a possibility of taking on additional labour, the wage increases remain high and the outsiders remain out of work. Strictly speaking, this does not constitute an increase in equilibrium unemployment but it still amounts to unemployment that is more or less permanent.
In Sweden, wage increases tended to accelerate in the 1995 round of settlements even though unemployment was high. Statements by representatives of labour market organisations are indicative of a propensity to give real wage increases priority over increased employment. Allow me to cite (from Finanstidningen, 14th October 1996):
"There is no direct link between our coming wage claims and the number who are unemployed". This is an illustration of the theories about insiders and outsiders in the labour market.
The costs of hiring and firing labour are one source of insiders' influence over wage formation, which can affect the possibility of bringing unemployment down. During the latest economic upswing in Sweden the number of hours worked rose by the equivalent of 130,000 new jobs but the increase in the number of persons in employment stopped at 85,000. The direct explanation is that average working time lengthened, which naturally may have reflected a rising propensity to go from part-time to full-time work and to do overtime in order to ease a strained personal economy. But it is also conceivable that employers have become less inclined to hire additional labour and more prone to make greater use of the existing labour force to cope with rising demand. An indication of this could be employers' willingness to introduce working hours related to the activity in the economy.
The differences between the explanatory models should not be exaggerated. They have several features in common and this makes them difficult to separate clearly in practice. Most of the evidence suggests that the high equilibrium unemployment in Western Europe as well as in Sweden has several causes. As the structures and institutions in the Swedish labour market resemble those in the rest of Western Europe in many respects, it is reasonable to suppose that the same forces have been at work in our country but that their impact and importance were concealed for a long time by an expansionary economic policy that was ultimately unsustainable.
High and more or less permanent unemployment, or high equilibrium unemployment, is a sign that the labour market is functioning poorly. These problems are rooted in turn in wage developments because it is price formation which is the central market mechanism.
Conclusion
The best way in which monetary policy can contribute to an economy that functions better economy with employment and growth is by keeping inflation low and stable.
The economy can then expand along its long-term growth trend. If this path, which is determined by structural factors, permits a rapid increase in output, growth will then be high. If potential output rises slowly, on the other hand, economic growth will be weak even if monetary policy were to be very expansionary.
Most things suggest that traditional demand policy is not capable of lowering unemployment appreciably. Most economic observers also expect that even with decent growth, unemployment will remain high in the coming years.
Whatever the causes of the high unemployment, an important key to turning this tendency lies in the functioning of the labour market. I have not considered conceivable specific measures, such as stimulating private sector employment, contributing to efficient wage formation and encouraging individuals to look for and accept jobs and maintain and develop their competence. I believe that, as Assar Lindbeck pointed out recently, the problem of unemployment in Sweden and, indeed, in the whole of Western Europe has to do with a great many factors connected with how the labour market functions.
Although the most effective measures cannot be identified for certain in advance, the central concern is the workings of the labour market and thereby also the labour market organisations. This applies to employer and employee representatives alike. The climate in the labour market will be of significance for the possibility of bringing the high unemployment down.
One issue is whether a mood of mutual understanding can be created so that employment has priority in wage setting. An important lesson seems to be that Sweden ought not to opt for a standard in wage formation that matches the average level in Western Europe; such a relatively high level could impede a reduction of unemployment. The model should rather be the Netherlands.
Another issue is whether a mood of mutual understanding can be created about the reform and construction of labour market rules and other measures that can help to reduce unemployment.
My intention this evening has not been to present a deeper analysis of the labour market problems, nor to answer questions about what needs to be done. But perhaps I have contributed by raising some of the questions that need to be answered and by defining the function of monetary policy and what it can do. If we are to achieve a lasting return just to something resembling the unemployment figures we have been accustomed to in Sweden, there will have to be a proper discussion of which measures are needed to bring unemployment down.
