Bergström: Is the Riksbank to blame for unemployment?
Thank you for the invitation to come to Brunnsvik to speak about the Riksbank and how we work.
Over the past ten years, Sweden’s macroeconomic performance has been relatively robust, both in comparison with other OECD countries and with the economic performance of the 1970s and 1980s. But there are problem areas, and I intend today to discuss perhaps the most serious of these – the employment situation.
It's natural in the employment debate that questions are put to the Riksbank, since our conduct of monetary policy affects short-term economic developments. This makes it tempting to criticise the Riksbank and to expect us to be able to solve the employment problems through a looser, more inflation-tolerant monetary policy. However, the experiences of many decades of research and of actual economic developments are unambiguous: inflation is no solution to unemployment problems.
To elucidate this it's important to differentiate between two issues: on the one hand the long-term relationship between a stable, low inflation rate – which is the Riksbank’s statutory objective – and employment. On the other hand the impact of monetary policy on economic activity and employment in the short term. The latter issue has been discussed more and more in the public domain recently, where it has been claimed that excessively tight monetary policy has resulted in the loss of up to 75,000 jobs, as if a simple relationship existed between interest rate decisions and jobs.
Allow me to begin with a retrospective approach to address the issue of monetary policy's effects on employment in the long run.
Employment and inflation in the long term – a retrospective view
During the first decades of the post-war era, one of the fundamental theses of Keynes' legacy was that there was a direct trade-off between unemployment and inflation, implying that decision-makers could secure lower unemployment by allowing inflation to rise. Inflation was moderate and full employment was achieved.
Owing to this line of reasoning, the fight against inflation was completely overshadowed by demand management. In addition, the power of monetary policy was constrained by the exchange-rate cooperation at that time – Bretton Woods – which had a fixed exchange rate as its target. When I started my economics education in the early 1960s, monetary policy revolved around quantitative controls of the credit market. Interest in stabilisation policy theory was instead tied to fiscal policy. The economics curriculum focused on the impact of fiscal policy on demand developments. How big are the multiplier effects of public expenditures, tax changes and expenditures financed in different ways? The supply side, i.e. the issue of the long-term sustainable growth rate under full capacity utilisation, was almost entirely overlooked. Even growth theory was sometimes based on demand analysis rather than on the growth of the factors of production and on technological progress.
During this period, full employment was the top priority of Swedish economic policy, as is evident in every fiscal plan from the 1950s and 1960s. The idea was that full employment involved a small rate of frictional unemployment of around 2.5 per cent, even though the level was never stated explicitly in policy documents. That level was assumed to correspond to people that at the time of each measurement were in the process of changing jobs or in the process of leaving the education system to enter the labour market or similar. According to the prevalent view, higher unemployment could always be explained by insufficient demand.
It was generally accepted among professional economists and the public that the government could fine-tune the economy with the aid of fiscal policy - expenditures, tax receipts and tax structures. It was a question of bridging the cyclical slumps. These ideas prevailed on both sides of the Atlantic, even though there was a school of sceptics both in Europe and the United States.
This policy was highly successful for a long time. For the greater part of the post-war period, unemployment in Sweden remained clearly below 4 per cent. On an annual basis it did not exceed 4 per cent at any time between 1950 and 1990.
Why didn't the successes last? In order to properly understand what happened, we must consider the gradual shift that occurred in expectations and in economic policy during the post-war period. The thinking at the beginning of that period was still marked by the interwar period's mass unemployment but also by the price stability of that time. Prices fell during some periods and rose during others. On the whole, the price level was stable during the 1930s, which was also the Riksbank’s explicit target at the time. Inflation expectations during the first post-war period were marked by these experiences, even to such an extent that the government prior to the post-war period announced in advance that the price level would fall as the productivity gains from the transition to peacetime production materialised.
As the general price level actually rose, but never dropped, the public’s expectations of inflation were influenced accordingly. Expectations of rising prices worked their way slowly into people’s minds. During the period 1956-64, the general level of prices increased on average by 3.4 per cent a year; during the period 1965-74 by 5.6 per cent and from the mid-1970s to the mid-1990s by an average of 8.1 per cent a year. Between 1973 and 1982, Sweden devalued the krona six times after the Bretton Woods system, with its fixed exchange rates, had been abandoned. By the end of the 1980s, following successive increases in the inflation rate and repeated devaluations, inflation expectations had been anchored at a high level. It became increasingly obvious that the trade-off between inflation and unemployment that had previously been taken for granted did not apply, that each new attempt to raise employment by expanding demand only resulted eventually in increasingly higher inflation (see Chart 1).
It is easy to understand why. When inflation rises, it affects the wage demands in the next round of wage negotiations. Wage-earners are interested in the actual purchasing power of their wage and structure their wage demands accordingly. Inflation thus increases without any effect on employment. The conclusion is that in the long run there is no trade-off between inflation and unemployment.
But even though inflation does not have any long-term impact on employment, it does affect economic growth and welfare. High inflation disrupts the mechanism by which goods’ relative values are signalled via prices, and distorts the reasons to save. This problem is perhaps especially significant for investment when the return is expected to be yielded over a long period in the future. In such cases, uncertainty over price developments can have a curbing effect. Furthermore, there is an increase in uncertainty about future developments in wages and costs, making it more difficult for agents in an economy to make economic decisions. Lenders may decide to set a higher rate of interest to compensate for any rise in inflation, and firms may decide not to invest in new machinery or to hire employees.
The experiences during the 1970s and 1980s contributed to the formation of the new monetary policy regime, which targets low, stable inflation. The inflation target has been set at 2 per cent. That level allows some leeway to zero inflation, which, among other things, reduces the risk of deflation and enables adjustments in relative wages without normally requiring nominal wage decreases for different groups in the labour market.
On the whole, as I mentioned, it has also worked well. Inflation has fallen at the same time as economic growth has been relatively high compared with the immediately preceding decades (see Chart 2). Unemployment has gradually declined since the crisis in the early 1990s, even though it is considerably higher than in the 1960s.
After the introduction of the inflation target, we thereby had maybe the clearest evidence that there is no trade-off between inflation and unemployment since both fell at the same time during the 1990s. The absence of a long-term relationship between inflation and unemployment has also been confirmed by extensive studies in a large number of OECD countries.
Monetary policy and employment in the short term
Having now spoken about the Riksbank's inability to influence employment in the long run, the question remains of how monetary policy can affect employment in the short term. Monetary policy can have a short-term impact on economic activity and therefore at best reduce the fluctuations in the cyclically determined component of unemployment.
Monetary policy is estimated to affect inflation 1-2 years ahead. Consequently, monetary policy decisions are based on a forecast of future inflation. It is difficult, however, to keep inflation perfectly in check. It may deviate from the inflation target, due, for instance, to unexpected events or political decisions. Likewise there may sometimes be reason, in the event of a sharp deviation, to return inflation slowly to the target so as to avoid substantial real economic costs. But like other forecasters, the Riksbank can of course also make mistakes in its forecasting. For all these reasons inflation may be above target during some periods, and be below target during others. In order to take account of and clearly reflect this reality, the inflation target is set within a tolerance band of ± 1 percentage point. In exceptional cases, this tolerance band can also be exceeded during short periods. Fortunately it seems that the Riksbank's forecasting ability has at least not been poorer than most other forecasters', possibly even somewhat better.
As you might guess the Riksbank doesn't receive any congratulatory telegrams for creating thousands of new jobs when we overshoot the inflation target by a few tenths of a percentage point, but when we undershoot to the same extent some economists are quick to put a price on the miss in the shape of lost jobs. We welcome all criticism that is underpinned by solid analysis, since it enables us to enhance our forecasting ability. However, I think it’s important to clarify a number of points regarding the criticism in which a number of commentators have claimed that the Riksbank’s actions have caused the loss of 20,000 to 75,000 jobs.
The first is that we have an entirely symmetric approach to deviations from the target. It is just as bad to undershoot the target as to overshoot it. Sometimes I get the impression that at least some of the criticism is based on the idea that the Riksbank has some kind of "hidden agenda" that is leading us to deliberately try to keep inflation below our set target, or at least that we think it’s less problematic if inflation is below target than above it. I can assure you that we have no such “hidden agenda”. The Riksbank always aims to meet the inflation target.
The second is that there is no such simple relationship between interest rate decisions and unemployment, as the criticism suggests. Were the repo rate able to steer developments in the economy so directly, inflation targeting would be much easier, but this is not the case. The economy is exposed to new shocks all the time, and it is difficult to discern their significance and how they can be countered by monetary policy. What the Riksbank can do, in a complex world, is to influence resource utilisation and economic activity so that inflationary tendencies are kept fairly well in check, at the same time as a host of other factors affect employment in parallel and interact with the monetary policy effects. In this work we try to avoid unnecessary fluctuations in output and employment, but we are not able to fine-tune the economy exactly.
The fact that the Riksbank's power to influence employment is both short-term and relatively limited is not inconsistent with our ability to keep inflation in check. The latter is normally achieved by affecting inflation expectations. As long as our monetary policy is credible, it is enough to make small interest rate adjustments to keep inflation close to target. In this way, the Riksbank sends out a signal that it is keeping a close eye on inflationary tendencies and everyone knows that stronger measures await if inflation expectations take off in either direction.
So, there is no simple relationship between monetary policy and employment. Nevertheless, for the sake of argument, let's go along with our critics and assume that such a relationship exists. In that case, estimates merely show that even if the repo rate had been a full percentage point too high for a whole year, the effect on today’s employment situation would at most have been a few thousand jobs, and not the tens of thousands claimed by the Riksbank's critics. In light of this, claims that the Riksbank is to blame for the loss of tens of thousands of jobs are rather questionable. Neither have our critics, despite requests, presented any convincing estimates to support this view. I think it’s a poor show on the part of government officials, trade union economists and trade union leaders to accuse the Riksbank of bringing unemployment upon tens of thousands of people.
An interesting addition in this regard is that the most recent periods when inflation has been below target, at the end of the 1990s and the beginning of this decade, have been marked by unexpectedly high productivity and unexpectedly low imported inflation (which few forecasters had anticipated). So it's not the case that the low inflation is a result of the Riksbank actively tightening monetary policy and dampening demand. On the contrary, these have been situations in which interest rates have been low. In spite of that, inflation has been weak owing to positive supply shocks. The impact of these shocks on employment differs from the effect produced by excessively tight monetary policy. Too low inflation due to high productivity growth and cheap intermediate goods stimulates growth, demand and employment. Consequently, the price of missing the inflation target has likely been lower in terms of growth and employment than if the cause had been a sharp monetary tightening that had caused demand to drop and inflation to fall.
The fact that employment depends on a number of other factors than monetary policy decisions and cyclical variations is illustrated, among other things, by the miscalculations of the National Labour Market Board just prior to 2004. In November last year the Board estimated that Swedish economic growth would be 1.4 per cent in 2003 and 1.9 per cent this year, while unemployment was forecast to be 5.1 per cent this year. Moreover, the Board believed that the repo rate should be cut from 2.75 to 2.25-2.5 per cent. These forecasts were roughly in line with those of the Riksbank at that time, with the exception that we, as usual, based our forecasts on the assumption that the repo rate would be kept unchanged. A year later we can say that economic growth has exceeded both the National Labour Market Board’s and our expectations at the same time as the monetary stance has become looser. Nevertheless, labour market growth has been weaker. This shows both that forecasters can be surprised by unexpected developments in the economy and that unemployment at present is not chiefly due to insufficient demand for goods and services owing to low growth and tight monetary policy.
Structural policy measures needed for a better functioning labour market
So, the weak employment growth is, in my opinion, not primarily a result of low demand for goods and services in the economy – the economy is growing by between 3 and 4 per cent a year – but instead has to do with the rapid structural change that puts increasingly high demands on the functioning of the labour market. I now intend to touch upon these structural problems.
Firstly, the employment problem is even bigger than that reflected in open unemployment data. At the beginning of the 1990s the supply of labour dropped at the same time as employment fell, which meant that the decline in employment was not fully mirrored in unemployment figures. It is a common pattern for both the size of the labour force and employment to be correlated with economic activity. When activity weakens, it leads not only to a fall in employment but also to a decrease in the number of people who say that they want to work. This is natural when people know that it’s difficult to get a job. But despite historically high economic growth, the employment rate has not recovered since its steep decline in connection with the crisis in the early 1990s. A large proportion of the working age population is still outside the labour force, which implies that there in practice is considerable hidden unemployment in addition to those openly unemployed. Furthermore, the fact that open unemployment has not fallen more during the 1990s’ economic boom indicates that the labour market does not function satisfactorily.
Another sign that the unemployment problems are structural in nature is the big differences between regions and between occupational categories. In spite of a single monetary policy for the entire Swedish economy, unemployment figures in some regions are double those in others, while the proportion of people receiving benefit from their unemployment fund is twice as high in some occupational categories as in others (see Charts 3 and 4). These differences seem to remain over long periods, indicating low mobility in the jobs market. Another possible indication of such shortcomings is that the same vacancy ratio today is combined with somewhat higher total unemployment than 15-20 years ago.
At the same time there is a large, growing need for mobility in the labour market. A comparison of different sectors’ employment growth during periods of expansion and contraction shows that many sectors suffer from job losses in both upturns and downturns, while others see jobs created both in upturns and downturns. The old cyclical pattern, where jobs are shed during a slump and are then created once the economy turns around, no longer appears to hold true. Jobs that are lost in a slowdown are being replaced, especially in manufacturing, to some degree by increased productivity and information technology. Growth is leading to new jobs but in different areas than before. This means that more people have to change workplace or sector, which in turn requires considerable adaptability. More jobs have to be created in, for example, the services sector, perhaps in completely new firms.
Consequently, high growth does not necessarily imply increased job security in the traditional sense. Instead, high growth may reflect rapid change and high productivity growth, and that more jobs are disappearing in some parts of the economy at the same time as new jobs need to be created in other areas. This pattern has been observed in the United States in the debate on jobless growth.
This trend is reinforced by firms’ increased possibilities to outsource parts of their production to other countries, a phenomenon that has recently been getting more and more attention in the economic debate. Several studies from the United States, where outsourcing is considered especially extensive, show that the direct effect on employment is fairly small. However, there is an indirect effect since the pressure to change increases, i.e. the need for staff decreases in some sectors but rises in others.
In such a time of rapid structural change, the cost of low labour-market mobility increases, while it becomes more important that new firms and jobs are created. More people will find themselves in a transition between companies and between sectors, and sometimes between residential locations. If the incentives are low for workers to quickly change jobs and move to a new sector, at the same time as the incentives for firms to employ labour from a different sector are reduced by high initial wages, costs of sick leave and job security, there is a risk that more people will get stuck in the transition. This would entail a rise in frictional unemployment. In addition, the risk of long-term unemployment rises if more people become discouraged and stop looking actively for work.
The Swedish labour market has previously functioned efficiently in these respects. Think of the closing down of almost the entire textile industry and of the shutdown of the Swedish shipyards at the same time as the paper and pulp industry and the iron and steel industry were rationalised considerably during the 1970s. Not least the trade union movement, more specifically the Swedish Trade Union Confederation, helped facilitate structural change. Read the official reports from 1941, 1951 and 1961 for the Confederation's conferences in those years. The findings of these reports are something we need to take in today. The trade union movement realised that structural change was the prerequisite for increasing real wages and for generally higher prosperity.
Conclusions
I believe that it's fully possible with the right methods to achieve both an employment rate that is sharply higher and an unemployment rate that is considerably lower than today's level. It is important, however, that the fundamental cause of the unemployment is discussed seriously and that the calls for change are directed at those who actually have the power to carry out the required changes.
All factual criticism of the Riksbank's forecasts is good, but there's a risk of concentrating the debate on our interest rate decisions and of the Riksbank's deviations from the inflation target, the effects of which are transitory and even themselves out over time, overshadowing other, much more important factors that really can affect employment in the long run. We risk forgetting the much more powerful tools that other decision-makers, outside the Riksbank, have at their disposal to influence employment. What is relevant is not why unemployment is 5.8 per cent instead of 5.0 per cent but rather why it is not 4 per cent or even lower on average over an economic cycle. Equally important is why the employment rate has continued to drop in Sweden since the 1980s.
In my opinion, measures must be taken to enhance the functioning of the labour market. The demands are increasing in line with the global trends that we’ve seen in recent years in the shape of outsourcing and technological progress.
I have never believed that the shift in regime to a low-inflation policy would make it impossible to return to full employment according to the old norm even in the long run. I would not have taken the job I have now if I wasn't convinced that it is possible to combine low unemployment with low inflation, as in the first decades after the Second World War. I understood and criticised the fact that the transition from high inflation would be costly in terms of temporary job losses and a decline in income. But that cost is now behind us and the conditions are in place for a lower level of unemployment in the long run. However, this requires a willingness to discuss the real causes of the weak employment growth in Sweden.
