• Keine Ergebnisse gefunden

Downward nominal wage rigidity during the crisis – aggregate

Im Dokument in the Estonian Private Sector (Seite 147-151)

6. IMPACT OF ECONOMIC CRISIS ON DOWNWARD NOMINAL

6.3. Downward nominal wage rigidity during the crisis – aggregate

In order to get a first impression of the reaction of wages to the recent economic crisis a very simple set of statistics will be used: the share of wage cuts and wage freezes in the total number of wage changes. Before we can look at the results, certain methodical issues have to be clarified. Registry data, although it reflects very accurately the actual change in total wages, does not tell much about the employer’s intentions behind cutting wages. So if the registry data shows a small wage cut, it is difficult to believe that this wage cut is a result of a determined action to cut workers’ wages by exactly this amount (say 0.3%). It is far more plausible that such small changes are the result of some technical issues, for example a worker being absent for half a day for personal reasons.

The same applies also to wage increases. On the other hand, it is very difficult to draw a clear line between ‘probably random’ wage cuts and wage reductions that were the result of employer-side austerity measures.

In the histogram-location approach a 1% or 2% histogram bin is usually used, which also means that wage changes that fall into the range +/–0.5%

around zero (or +/–1% respectively) are considered to be wage freezes and changes that fall to the left of this range are wage cuts.

In Figure 45 wage freezes and wage cuts are depicted for period 2002–2009.

The wage changes data is prepared according to the same methodology that was described in chapter 3.5. The different lines on the figure designate the different ranges for wage changes that will still be considered as a wage freeze. Most of the empirical work done in this thesis uses a 2% bin width. For this range, the share of wage freezes varies between 4.3% (the lowest value (2007)) and 8.1%

(the highest value (2009)). While economic growth left the share of wage freezes more or less constant, during the economic crisis the share of wage freezes almost doubled. Next we can look at the wage cuts using the same range (+/–1%). It is evident that the share of wage cuts shows significantly higher variation than the share of wage freezes. During solid economic growth the share of wage cuts dropped from 26.7% to 15.8% in 2007. However, in 2009 the share of wage cuts increased to as high as 55.1% which is 3.5 times more than in 2007.

148 Figure 45. Wage freezes and wage cuts

Source: Estonian Tax and Customs Board, author’s calculations

What happens with higher wage cuts? Figure 45 also provides information on the dynamics of wage cuts of more than 5%. It is clear that they follow the same pattern. In 2009 47% of observations in wage change distribution were wage cuts of more than 5%, which is 3.8 times more than in 2007.

The first conclusion from the wage change distributions is that the share of wage freezes and of wage cuts increased substantially during the period 2007–

2009. However, the increase in wage cuts was significantly higher than the increase in wage freezes and the extent of the share of wage cuts indicates that this was a very wide-spread practice.

It is difficult to put wage freezes and cuts separately on the same picture with DNWR – it would be easier to interpret only one indicator. The rigidity coefficient used in the histogram-location approach estimated the share of wage cuts that did not happen and were instead converted into wage freezes. The maximum number of wage cuts that can theoretically be converted into wage freezes cannot exceed the number of observations in the zero bin of the distribution. Thus, the maximum possible value of the rigidity coefficient can be obtained by dividing the share of wage freezes by the combined share of wage freezes and wage cuts in the wage changes distribution:

0,0%

= ( ) ( ) (6–1)

where ρmax is the maximum share of wage cuts that were not enforced due to DNWR, f(x) is the probability density function of wage changes, and u and -u are respectively the upper and lower bounds of the range of wage changes that are still considered to be wage freezes.

Unlike the ‘real’ DNWR coefficient, the maximum size of the DNWR can easily be observed from the wage changes distribution. If there are changes in ρmax over the business cycle then this could be an indication of the dynamics of DNWR as well. Table 36 lists wage freezes and wage cuts and the theoretical maximum values of DNWR for the different ranges used for defining wage freezes.

In the sub-chapter 4.5 DNWR was analysed on the Estonian aggregate level with 1% as the histogram bin width. If we concentrate on the columns using +/–

0.5% as the range for defining wage freezes we can see that the size of maximum DNWR declined from its highest value of 17.8% in 2006 to 9.4% in 2009. This is a significant reduction. For the period 2002–2008 the DNWR coefficient estimated with the histogram location approach was 9.3%. Unless all zero wage growth observations in 2009 were the result of wage rigidity (i.e.

redistributed from the left side of the distribution to the zero location), it can be said that DNWR has declined during the crisis. The ratio of actual DNWR to the maximum DNWR coefficient for the period 2002–2008 was around 53%. If the same proportions also prevailed for 2009, then the actual DNWR for 2009 would be 4.9%.

It is also interesting that the indicator of maximum DNWR reaches its peak in 2006, while the peak of the economic boom was achieved only in 2007. It seems that the maximum DNWR indicator has certain leading indicator properties over mean wage growth, which reacts more slowly to economic downturns (see Figure 46).

150

Figure 46. Dynamics of median wage growth and the maximum DNWR indicator Source: Estonian Tax and Customs Board, author’s calculations

Table 36. Incidence of wage freezes, wage cuts and maximum DNWR in Estonia 2002–

2009

Source: Estonian Tax and Customs Board, author’s calculations

From the maximum DNWR coefficients, the main conclusions do not depend on the range that is used in defining wage freezes, as the economic crisis brought about a significant decline in the maximum possible values of DNWR and this applies to all definitions of wage freezes.

To summarise, there is no exact measure of DNWR that could be applied for analysing this phenomenon during severe economic crises. However, when relying on second best alternatives such as looking at the share of wage cuts and

-5%

2002 2003 2004 2005 2006 2007 2008 2009 Median earnings growth Maximum DNWR -/+1%

wage freezes in the wage change distribution and a new indicator, the maximum DNWR indicator, the picture is quite clear. The economic crisis led to a signi-ficant increase in wage cuts and a sizeable reduction in the maximum DNWR indicator. The theoretical maximum of wage cuts that were transformed into zero wage changes was 9.4%, although this would mean that all observations of wage freezes are in fact non-enacted wage cuts. From the previous estimates of DNWR coefficients and the maximum rigidity indicator there is reason to believe that the actual DNWR coefficient for 2009 is significantly smaller. This also means that when comparing 2009 with the period 2002–2008 there was a decline in the DNWR coefficient.

6.4. Downward nominal wage rigidity during the crisis –

Im Dokument in the Estonian Private Sector (Seite 147-151)