Aggregate Demand and Long-Run Unemployment

Abstract

WHAT DETERMINES the unemployment rate? In answering this question, mainstream economics draws a sharp distinction between the short run and the long run. According to the conventional view, short-run movements in unemployment are strongly influenced by monetary policy and other determinants of aggregate demand. In the long run, however, unemployment returns to a natural rate or NAIRU (the nonaccelerating-inflation rate of unemployment), which is determined by labor market frictions. The NAIRU can change over time for microeconomic reasons, such as changes in labor market institutions. But the conventional wisdom holds that the NAIRU is unaffected by aggregate demand, and thus that demand does not influence long-run unemployment trends. This paper argues that this conventional view is wrong. Monetary policy and other determinants of aggregate demand have strong effects on long-run as well as short-run movements in unemployment. And this is not just a theoretical point. Over the last twenty years the behavior of demand accounts for much of the differences across countries in the evolution of unemployment. Aggregate demand has long-run effects on unemployment because of what Olivier Blanchard and Lawrence Summers have called hysteresis.(1) At a given point in time, there exists a NAIRU: pushing unemployment below a certain level causes inflation to rise. But as demand pushes unemployment away from the current NAIRU, this causes the NAIRU itself to change over time. A number of authors (including myself)(2) have presented empirical evidence in favor of hysteresis, but many students of unemployment remain unpersuaded. The broad goal of this paper is to bolster the case for hysteresis in industrial countries in the 1980s and 1990s. This paper also documents two specific aspects of hysteresis that are important in recent history. The first concerns the role of monetary policy in determining whether hysteresis arises--whether a cyclical rise in unemployment causes a rise in the NAIRU. Here I primarily examine the early 1980s, when most member countries of the Organization for Economic Cooperation and Development (OECD) experienced recessions arising from disinflationary monetary policy. In some countries, such as the United States, the rise in unemployment was transitory; in others, including many European countries, the NAIRU rose and unemployment has remained high ever since. I argue that the reactions of policymakers to the early-1980s recessions largely explain these differences. In countries where unemployment rose only temporarily, it did so because of strongly countercyclical policy: after tight policy produced a recession to disinflate the economy, policy shifted toward expansion, reducing unemployment. In countries where unemployment rose permanently, it did so because policy remained tight in the face of the 1980s recessions. This analysis implies that passive macroeconomic policy during a recession has a high cost: it can lead to permanently higher unemployment. And there do not appear to be substantial benefits from such inaction. I find that countries that combated recessions with expansionary policy still obtained lower inflation from the initial rise in unemployment. Countries that kept policy tight did not gain an additional reduction in inflation. Most previous discussions of hysteresis focus on its role in explaining increases in unemployment. The second part of my analysis asks whether hysteresis works in reverse--whether demand expansions can produce permanent decreases in unemployment. Here I examine the period from 1985 through 1997, when several countries emerged as success stories in reducing unemployment. In particular, I identify four countries where unemployment has fallen substantially since 1985--Ireland, the Netherlands, Portugal, and the United Kingdom--and contrast them with countries where unemployment has stayed high. I find that reverse hysteresis helps explain the success stories: their decreases in unemployment were caused largely by demand expansions. …