A comparison of labour market responses to the global downturn.

Holland, Dawn ; Kirby, Simon ; Whitworth, Rachel 等

The global economic downturn has led to a crisis in labour markets, with an estimated 15.2 million job losses across the OECD economies, equivalent to a rise in the OECD unemployment rate from 5.5 to 8.9 per cent. Initially, the rise in unemployment appeared lower than expected. In Holland, Kirby and Whitworth (2009) we demonstrated a simple rule of thumb between output growth and the unemployment rate in the OECD as a whole, based on Okun's approach. Using a dataset that spans the period 1988-2008, regression analysis suggests that on average a 1 per cent decline in output is associated with a rise of 0.6 percentage points in the unemployment rate across the OECD economies. Between the first quarter of 2008 and the first quarter of 2009, output in the OECD economies declined by 4.8 per cent. The unemployment rate rose by 1.9 points over this period, as compared to 2.9 per cent given by the rule of thumb. However, the labour market tends to lag production. While most of the major economies started to grow again in the second or third quarters of 2009, OECD unemployment continued to rise into the final quarter of the year, with a cumulative increase in the OECD unemployment rate of 3.4 percentage points--even higher than that suggested by our rule of thumb.

The peak in the OECD unemployment rate may already be behind us, as unemployment rates have stabilised in Canada, Germany, Hungary and Finland and even started to decline in Japan. This note delves more deeply into the response of OECD labour markets to the global economic downturn. While at the aggregate OECD level labour input has declined more or less in line with expectations given the decline in output, we have seen stark differences in the labour market responses of individual economies. We will highlight some of these key differences and identify those countries where labour market prospects are expected to continue to deteriorate this year.

Figure 1 illustrates the cumulative output loss between the first quarter of 2008 and the first quarter of 2009 against the rise in the unemployment rate to the end of 2009 in a selection of OECD economies. Countries that lie below the OECD regression line, such as Germany, Japan and Italy, have exhibited a relatively small rise in the unemployment rate given their output loss, whereas those that lie above the regression line estimated by Holland, Kirby and Whitworth (2009) have exhibited a greater rise in unemployment than expected. Notably, Spain, Ireland and the US have all experienced what might be considered excessive increases in unemployment given their output declines.

The unemployment rate measures the proportion of the labour force that is not currently in work, and a rise in the unemployment rate can either indicate job losses, or it can reflect new entrants into the labour force who have yet to find employment. The two have very different implications for the productive capacity of the economy. While the unemployment rate is a useful indicator of wage pressures in the economy, a more direct measure of labour input can give deeper insight into the implications for production. Total labour input is given by total employment multiplied by the average hours worked per employed person. Any adjustment in working time has important implications for output but will not be reflected in a change in the unemployment rate.

[FIGURE 1 OMITTED]

Figure 2 plots the loss of output against the decline in labour input since the start of the downturn in a selection of OECD countries. Those countries that lie above the regression line have seen a greater decline in labour input than expected, based on the simple rule of thumb, while those below the regression line have seen a smaller decline in labour input than expected. Spain, the US and Ireland all lie above the regression line as they do in figure 1. However, it is interesting to note that Sweden also lies above the regression line rather than below it, the opposite of what is seen in figure 1. Another interesting development is shown by the positions of Germany and Japan, which lie on the regression line rather than well below the line as shown in figure 1, suggesting that labour input has declined by more than the shift in the unemployment rate suggests.

In a downturn, firms bring in lower levels of revenue, and if they cannot find a way to reduce costs many will go bankrupt. There are three routes through which firms can reduce their labour costs: reducing employment levels, reducing average hours worked per employee, and reducing average wages. All three have no doubt been at work to varying degrees in all the major economies. Cultural, regulatory and bargaining differences will determine the individual country responses to some degree. In figure 3 we decompose the change in labour input illustrated in figure 2 into the change in employment and the change in average hours worked. Average hours have declined by more than employment in the US, Germany, Sweden, Japan, the UK and Australia, whereas the decline in labour input in Spain and Finland is entirely down to a reduction of employment. Italy, France and Australia have experienced very little in the way of labour input loss. This is particularly surprising in the case of Italy, where output declined by more than 6 per cent between the first quarter of 2008 and the second quarter of 2009.

[FIGURE 2 OMITTED]

Another interesting distinction across countries is the extent to which declines in employment reflect job losses by employees or job losses by the self-employed. Those made redundant may become self-employed, reducing the impact on total labour input and potential output. At the same time, in a banking crisis we may find many small firms and family businesses going bankrupt, leading to a reduction in self-employment. In countries where employment protection is strong, we may see a smaller decline in employee jobs, but sharper declines in self-employment, which tends to have little in the way of social protection. Where labour market regulation is relatively weak, we may be more likely to see employee job losses. Figure 4 illustrates the change in self-employment and the change in employees between the first quarter of 2008 and the third quarter of 2009. Self-employment has increased in the US, Finland, Canada and most notably France since the onset of the downturn, partially offsetting the decline in employment. In Ireland, the self-employed and employees have suffered about equally, whereas the self-employed in Spain, Sweden, Japan and Italy have experienced relatively higher rates of job losses.

[FIGURE 3 OMITTED]

[FIGURE 4 OMITTED]

Employment levels may also be maintained if employees are willing to accept wage cuts as an alternative to layoffs. Figure 5 illustrates average annual real wage growth in the selected group of countries between 2000 and 2006, compared to the real wage growth observed since the beginning of 2008 on an annualised basis. Real wage growth has been weaker than average in France and the UK, which helps to explain why these countries have not suffered as sharp a decline in labour input as countries like the US and Spain, where real wage growth has soared due to the unexpected drop in inflation. Real wages have declined sharply in Italy, which can explain much of the resilience of employment in Italy. Ireland has also seen a sharp slowdown in real wage growth, although this appears to have done little to sustain employment, while the recent rise in real wage growth in Canada and the Netherlands has not pushed up unemployment as much as might be expected. Nonetheless, there is clearly a relationship between recent real wage growth and the change in labour input, with a correlation of 52 per cent.

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

There are clearly stark differences in individual country labour market responses to the global downturn. In order to determine whether employment has declined by more or less than expected, given country-specific developments in output, average hours worked and real wages, we run a series of simulations using the National Institute global econometric model, NiGEM. (1)

The response of the labour market to a decline in output will depend to some extent on the type of shock driving the decline. For our diagnostic shocks, we apply a rise in the investment risk premium, which NIESR has argued has been a driving force behind the global financial crisis (see for example Barrell, 2009). Figure 6 illustrates the expected decline in employment in response to a 1 per cent decline in output in the first year. (2)

Based on the estimated set of NiGEM equations, which reflect historical dynamic developments in each country, we would expect employment to decline by slightly more in the UK and the US, and by less in Italy and Sweden, but in all cases employment is expected to decline by 0.1-0.6 per cent in response to the 1 per cent decline in output.

While average hours worked per employee show some cyclical response to the shock, the NiGEM model is structured so that the bulk of labour input adjustment is effected through employment. In order to assess the impact of reduced working time, we repeat the diagnostic shock, applying a 1 per cent decline in average hours worked in each economy. Figure 7 illustrates the offset in terms of employment expected in response to a 1 per cent reduction in average hours worked. Spain and Sweden show slightly stronger employment responses to working time than Germany and the US. In all cases, there is some short-run decline in total labour input associated with a decline in average hours worked.

Our final diagnostic simulation looks at the impact of real wage developments on employment. We repeat the initial investment risk premium shock, applying an additional shock to real wages in each country. Figure 8 illustrates the offset in employment expected from an additional 1 per cent decline in real wages. Employment in the US, UK and Germany is more responsive to wages than in the other economies, while employment in Finland is relatively insensitive to real wage developments.

As a final step, we calibrate an expected decline in employment in each country, based on the simulation studies and actual changes in output, hours and real wages. The estimates are calculated as follows:

[DELTA][E.sup.e] = [[alpha].sub.1][DELTA]GDP + [[alpha].sub.2][DELTA]H + [[alpha].sub.2][DELTA]RW

where [DELTA][E.sup.2] is the expected change in employment between 2008Q1 and 2009Q3, [DELTA]GDP is the actual change in output between 2008Q1 and 2009Q1, [DELTA]H is the actual change in average hours between 2008Q1 and 2009Q3, [DELTA]RW is the difference between average real wage growth 2008Q1-2009Q3 and average real wage growth 2000-2006 (as illustrated in figure 5), and [[alpha].sub.1], [[alpha].sub.2], and [[alpha].sub.3] are the parameters illustrated in figures 6, 7 and 8, respectively.

[FIGURE 7 OMITTED]

[FIGURE 8 OMITTED]

[FIGURE 9 OMITTED]

Figure 9 plots our estimates against the actual change in employment. Those that lie above the line have seen a smaller fall/larger increase in employment than expected, given actual developments in output, average hours worked and real wages, while those that lie below the line have seen excessive declines in employment. Employment has declined significantly more than expected in Ireland in particular, but also in Sweden and Spain. Employment developments in the US, Finland, Canada, France, Italy and Australia are more or less in line with expectations, while employment in the UK was slightly worse than expected and employment in Germany has held up better than expected.

Figure 9 supports the initial conclusion from our simple rule of thumb. There has clearly been a bias towards underperforming in OECD employment, with seven of twelve countries experiencing sharper employment declines than expected and only three countries performing better than expected. The unemployment rate in Germany, Japan, Finland, Sweden, Italy and the Netherlands has risen by less than expected, given their output declines. However, in the case of Germany and Japan, this masks a decline in average working hours, so that the decline in total labour input is more closely in line with expectations. Sweden's labour input decline even appears excessive, given the decline in output. Italy has seen very little decline in either the unemployment rate or total labour input, and this is partly a reflection of real wage developments, with sharp declines since the onset of the crisis. A sharp rise in real wages in the US and Spain goes some way towards explaining the excessive rise in unemployment observed in these countries.

In general, developments in output, average hours worked and real wages can explain much of the country-specific labour market responses observed since the onset of the crisis. Developments in Ireland, Spain and Sweden clearly reflect additional factors, which may include a structural shift out of the construction sector in Ireland and Spain. The UK labour market has performed slightly worse than expected, and has underperformed relative to the other G7 economies. In Germany, employment levels have been maintained despite the sharp loss of output. Given the relatively strong real wage growth, this must be putting firm profits under pressure, and there is a strong possibility of a delayed labour market response, with job losses materialising this year.

REFERENCES

Barrell, R. (2009), 'Long-term scarring from the financial crisis', National Institute Economic Review, 210, October, pp. 36-8.

Holland, D., Kirby, S. and Whitworth, R. (2009), 'Labour markets in recession: an international comparison', National Institute Economic Review, 209, pp. 35-41.

NOTES

(1) For further detail on the structure of the NiGEM model, see the discussion on pp. 17-21 and on http://nimodel.niesr.ac.uk.

(2) For the purposes of this note, we assume a linear response to the size of shock applied. While NiGEM is not strictly a linear model, the degree of non-linearity is small and would have little impact on the results.

Dawn Holland, Simon Kirby and Rachel Whitworth *

* National Institute of Economic and Social Research. E-mail: [email protected], [email protected] or [email protected]