by Jan van Ours
Posted previously at VoxEU, 27 February 2015
The Great Recession has been characterised by an unprecedented decline in GDP, and unemployment rates remain above pre-Great Recession levels in many countries. This column argues that economic growth is a ‘one size fits all’ solution for the problem of unemployment, because the unemployment rates of different kinds of workers are strongly correlated within countries. That said, economic growth affects above all the position of young workers, and so benefits mostly those who need it the most.
Economic growth in OECD countries has been fluctuating for several decades, but it was rarely negative. The Great Recession is characterised by an unprecedented decline of GDP. In almost every OECD country there was negative real GDP growth of at least a couple of percentage points. The big negative economic shock led to a sharp increase of unemployment in many though not all countries. Now, more than five years after the negative growth shock, unemployment is still high, and in few countries has unemployment returned to its pre-Great Recession level. Above all, economic growth is needed. As I argue in more detail below, economic growth has an attractive ‘one size fits all’ character (see Van Ours 2015 for details). To illustrate my line of reasoning I use annual data from 20 OECD countries over the period 1970–2013.
Unemployment and growth since 1970
Figure 1 provides annual information about unemployment rates and economic growth over the period 1970–2013 for three country blocks – Eurozone countries, non-Eurozone European countries, and non-European countries – and the US. There is a strong correlation between the fluctuations in the unemployment rates, but there are also clear differences. In the US, unemployment rates fluctuate a lot, but over the whole of the calendar time interval there is no clear trend. For the other countries the average unemployment rates are much lower in the early 1970s than later on.
Figure 1. Unemployment rates and GDP growth, 1970–2013
a) Unemployment rates (%)
b) Annual GDP growth (%)
Notes: Eurozone countries: Austria, Belgium, Finland, France, Germany, Ireland, Italy, the Netherlands, Portugal, and Spain. Non-Eurozone European countries: Denmark, Norway, Sweden, Switzerland, and UK. Non-European countries: Australia, Canada, Japan, and New Zealand.
The Eurozone countries on average faced the biggest increase in unemployment. Whereas in the early 1970s the average unemployment rate in the Eurozone countries was substantially lower than in the US, not much different from the non-European countries, and only slightly above the non-Eurozone European countries, there was a strong increase in unemployment rates in the course of the 1970s and early 1980s. From the early 1980s onwards the average unemployment rate in the Eurozone countries is the highest. The immediate impact of the Great Recession in terms of unemployment rates is most severe in the US, but whereas in the US unemployment rates declined after the Great Recession, they kept on growing in the Eurozone countries. The lower part of Figure 1 shows that the fluctuations in GDP growth are very similar. The drop in GDP growth over the Great Recession is unprecedented over the period of analysis. The US experienced negative GDP growth in the early 1980s, but not nearly as severe as in the Great Recession. Over recent years the Eurozone is the outlier, since there is again negative GDP growth, which is not the case in the other country blocks.
A tale of two recessions
Figure 2. Unemployment rates in 1985 and 2013
To illustrate that in terms of unemployment the Great Recession was not unique, Figure 2 compares the unemployment rates in the 20 countries in 1985 and 2013, a couple of years after the 1980s recession and the Great Recession respectively. Clearly, despite the difference of 28 years, the unemployment rates in the two years are highly correlated across the countries. Spain has the highest unemployment rate in both years whereas Austria, Japan, Norway, and Switzerland have the lowest unemployment rates in both years. Portugal is somewhat of an outlier as the unemployment rate in 1985 was average while in 2013 it was higher than in the other countries except for Spain.
Table 1. A tale of two recessions: Splitting up the change in employment rate into a change in the participation rate and a change in the unemployment rate (percentage points)
Note: This decomposition is based on data from Australia, Canada, Finland, Germany, Ireland, Italy, Japan, the Netherlands, Norway, Portugal, Spain, Sweden, and the US. The time periods are 1979–1983 and 2007–2011.
If employment goes down this may cause unemployment to rise and/or non-participation to increase. Table 1 provides an overview of the decomposition in the change in employment rates (employment as a percentage of the working age population) over the 1980s recession and over the Great Recession (including the post-Great Recession years). In terms of change in employment rates the Great Recession has affected men more than women. This was also the case in the 1980s recession. For young men (15–24 years) employment rates went down by 5.3 percentage points in the 1980s recession and 7.4 percentage points in the Great Recession. For young women these numbers are 2.9 and 4.4 percentage points. The unemployment rates for young workers went up simultaneously. For prime-age workers (25–54 years) there are clear differences between men and women. Whereas for women employment rates were hardly affected in the Great Recession and even increased in the 1980s recession, employment rates went down and unemployment rates went up for men. For older workers (55–64 years) there is a clear difference between the 1980s recession and the Great Recession. For older women the 1980s recession did not have a big impact, but over the Great Recession both participation and employment went up substantially. For older men the main effect of the 1980s recession was a substantial drop of 4 percentage points in the employment rate, which was somewhat evenly distributed between an increase in unemployment and a decrease in participation. The Great Recession hardly affected employment of older men, while unemployment went up due to an increase in labour force participation.
Figure 3. Unemployment rates by age and gender, 1976–2013
Note: This graph is based on data from 12 countries: Australia, Canada, Finland, Germany, Italy, Japan, the Netherlands, Norway, Portugal, Spain, Sweden, and the US.
Differential effect by gender, age, and education
Figure 3 shows the development of unemployment rates by age and gender. The unemployment rates of prime-age men and women and old men and women are not very different. All of these unemployment rates fluctuate and show an increase during the Great Recession. The unemployment rates of young men and young women show big fluctuations and the Great Recession is no exception.
Figure 4. Unemployment rates of men by age and educational attainment, 2011
a) Unemployment rates by age
b) Unemployment rates by educational attainment
Figure 4 provides information about cross-country differences in unemployment rates of prime-age men and young men in the year 2011, shortly after the Great Recession. Clearly, these are strongly correlated. In every country the unemployment rate of young men is substantially higher than the unemployment rate of prime-age men. The unemployment rate is highest in Spain, Ireland, and Portugal, and this is the case for both prime-age men as well as young men. The bottom graph of Figure 4 provides information about male unemployment rates in the year 2011 according to educational attainment, distinguishing between low-educated and high-educated workers. Clearly, the unemployment rate among low-educated workers is substantially higher than among high-educated workers. Nevertheless, as with age, if unemployment is high among low-educated workers it is also high among high-educated workers.
Quantifying the effect of economic growth
Okun (1963) proposed a relationship between unemployment and economic growth that is referred to as ‘Okun’s law’. Using US data Okun (1963) related quarterly observations from 1947:Q2–1960:Q4 on changes in the unemployment rate (u) in percentage points to percentage changes in real GNP (y), finding: Δu = 0.30 – 0.30 Δy. This implies that the unemployment rate will increase by 0.3 of a percentage point if real GNP does not grow, and will stay constant if the quarterly growth rate of real GNP is 1%. Okun’s study was not intended to explain the evolution of unemployment through economic growth but to indicate how much potential output is lost when the unemployment rate is above 4% – according to Okun, “a reasonable target under existing labour market conditions” that would be in line with price stability. The relationship between changes in the unemployment rate and economic growth has been studied often – sometimes with unemployment as the dependent variable and growth as the explanatory variable, sometimes the other way around. The latter type of study relates real output to a number of production factors including labour, capital, and technology. Perman and Stephan (2013) is a meta-analysis of 269 estimates of Okun’s relationship as presented in 28 studies. About 60% of all estimates have real output as the dependent variable, three-quarters use country-level data, and slightly more than half of the studies use a static model. Many studies find that the nature of the relationship has changed over time or find that the relationship is different in expansions than during recessions. Therefore, some studies refer to the relationship as ‘Okun’s rule of thumb’. Okun specified an empirical relationship from which it is not clear which way causality runs. From a labour economist’s point of view, growth affecting unemployment makes sense.
My regression results (see Van Ours 2015) provide parameter estimates of Okun’s relationship with GDP growth and lagged GDP growth as regressors, and a distinction is made by age and gender. The estimates of the constant decrease with age, and are larger for men than for women. The negative effects of GDP growth also decrease with age, and they are larger for men than for women. In all estimates the ratio of the constant and the sum of the two parameters for GDP growth is about 2.4, indicating that an annual GDP growth of 2.4% is needed to keep unemployment rates constant. The main difference according to age and gender is that the unemployment rates of young workers are affected the most by cyclical fluctuations while old workers are affected the least. Also, unemployment rates of men are affected more than unemployment rates of women.
The Great Recession was characterised by a drop in GDP that was unprecedented in recent history. However, the Great Recession caused an increase in unemployment that was not unprecedented. The much milder recession of the early 1980s caused a similar increase in unemployment rates. Within countries there is a strong correlation in unemployment rates. If unemployment rates are relatively high in a given country, they are relatively high for various types of workers along dimensions such as gender, age, and educational attainment. The overall picture of unemployment after the Great Recession is mixed. Some countries still have relatively low unemployment rates while others have very high unemployment rates. For the latter countries there is no easy way out. In terms of growth the Great Recession is history. However, the consequences of the Great Recession are still present. First and foremost, to reduce unemployment rates there is a great need for economic growth. Economic growth affects above all the position of young workers. Economic growth causes youth unemployment rates to go down quickly and youth employment rates go up fast. Therefore, economic growth benefits mostly those who need it the most.
Okun, A M (1963), “Potential GNP: Its measurement and significance”, Cowles Foundation Paper 190.
Perman, R and G Stephan (2013), “Okun’s law – A meta analysis”, SIRE Discussion Paper 2013-59, University of Strathclyde.
Van Ours, J C (2015), “The Great Recession was not so Great”, CEPR Discussion Paper 10376 (forthcoming in Labour Economics).