The main results

In this thesis we have analyzed German labor market for the period from 1970:1 to 2001:2. The primary aim is to disentangle structural shocks as main driving forces behind the rise in German unemployment rate and their propagation mechanism.

A small macroeconomic model serves as the theoretical basis which is in line with the approach of Layard et al. (2005). The model contains an aggregated demand function, a production function, a price setting relation, a wage setting relation, a labor supply function and a definition equation of unemployment. In accordance to the insider-outsider model, the wage setting rule states that nominal wages are chosen one period in advance and are set to make expected employment to be a weighted combination of lagged labor supply and employment. Full hysteresis corresponds to the extreme case where exclusively lagged employment (insiders) is considered in the wage bargaining process. These relations are influenced by exogenous variables, capturing the effects of various structural shocks. Institutional rigidities strengthen the power of insiders and thus exacerbate the inertia in the wage bargaining framework. Such labor market institutions have set the conditions to make the effects of adverse shocks very persistent and produce a very long-lasting rise in the unemployment rate.

SVAR analysis with long run restrictions originated from Blanchard and Quah (1989) is carried out. As compared with previous SVAR analysis of labor markets, novelties of this empirical work are the assumption of full-hysteresis in

the unemployment rate, which is supported by the presence of a unit root in the unemployment series according to Perron tests, and the identification of price shocks as one further structural shock.

Using long-run identifying restrictions achieved from the theoretical model, four structural shocks (price, productivity, aggregate demand and labor supply shocks) are recovered. With the help of impulse response analysis and forecast error variance decompositions, the contributions of various shocks to unemployment evolution in Germany are evaluated and the part of institutional rigidities is captured by hysteresis mechanism.

Empirical results show that no single factor for itself has caused the rise in unemployment. The persistently high unemployment is instead the result of a combination of various shocks as well as hysteresis mechanism.

As regards the structural shocks under investigation, demand shocks are shown to be dominant in accounting for unemployment evolution even in the long run.

Deficient aggregate demand and thus labor demand is no doubt an important reason for the miserable unemployment development in Germany. Although price shocks do not influence unemployment in the long run, they lead to a rise in the unemployment rate in the short/medium run. Since the impact of price shocks is significant and long-lasting, they can to some degree explain the unemployment persistence in Germany. Just like many theoretical and empirical literature about the effect of productivity shocks on the unemployment rate, this thesis does not provide a clear-cut picture concerning productivity shocks either. However, productivity shocks seem to influence unemployment in the short/medium run.

Finally, labor supply shocks are shown to have an effect on the unemployment rate.

In addition to various structural shocks, sluggish propagation mechanisms make transitory shocks have quite long lasting effects. There is rather high degree of inertia in German labor market as supported by the empirical result that the unemployment series seems to follow an I(1) process. Labor market institutions, as represented by the insider-outsider model, are important factors in explaining the hysteresis mechanism.

It can be concluded from the empirical work that it might be too simple blaming solely insufficient effective demand or labor market rigidities for persistently high unemployment in Germany. The sources of unemployment

appear to be a mixture of different factors from demand as well supply side. From the impulse response analysis, shocks of price, productivity, aggregate demand and labor supply are all important determinants of unemployment. Due to the introduction of hysteresis mechanism, all these shocks exert long-lasting effects on the rate of unemployment. Especially aggregate demand shocks influence unemployment even in the long run. Forecast error variance decomposition shows that price and aggregate shocks account for most part of the variation of unemployment, whereas productivity shocks have relatively small influence on the forecast error variance of unemployment and the influence of labor supply shocks is negligible.

Based on these results, we can interpret the unemployment development over the last decades as follows: price shocks like the two oil price crises in the 1970s together with a large productivity slowdown also in the 1970s certainly contributed to the initial rise of the unemployment rate during that period. In the early 1980s, when Monetarism and Supply-Side Economics were dominated, Bundesbank decided to disinflate German economy to overcome the stagflation in 1970s. Bundesbank thus raised the short-term interest rate, which resulted in lower output and hence higher unemployment. This led to lower inflation according to the Phillips curve. With inflation falling and unemployment rising, the central bank lowered the interest rate. However, as compared with the Fed in U.S., Bundesbank chose to disinflate very gradually. Bundesbank thus maintained tight conditions over a long period of time, with the consequence of equilibrium unemployment rate following actual unemployment rate (hysteresis mechanism).

When the central bank finally lowered its interest rate, this did not have any significant effects on the unemployment rate, because the equilibrium rate has risen as well by then. That means, the disinflationary process initiated by Bundesbank lasted too long such that hysteresis effects could arise. Moreover, adverse aggregate demand shocks from tight macroeconomic policy in the post unification era might have played a dominant role in explaining high unemployment in the 1990s.

The analysis casts some doubt on the popular blame accounting for high and persistent unemployment in Germany entirely by rigid wages and labor market institutions. This work illustrated that macroeconomic shocks can very well explain rising unemployment over the sample period. In fact, macroeconomic

shocks are able to explain large fluctuations and upward jumps in unemployment quite well. Such sudden changes, however, are hard to reconcile with a pure micro-based explanation since most of the labor market institutions were already in place even before unemployment started to rise. More generally, macroeconomic distortions are more likely to prevail when unemployment persists for many years and unemployment spells are long, which is exactly the case in Germany.

To recap, the results of the analysis are satisfactory in that the ‘theory-guided’

view on data yields reasonable results regarding impulse response functions and forecast error variance decompositions. Empirical results obtained are generally plausible and hence the advantages of this approach are confirmed in comparison with other standard techniques adopted to explain unemployment development.

However, the SVAR approach has also its limitations despite its rapid development in the empirical research. SVAR models are usually small, with some even important factors neglected. Interpreting residuals in such low-dimensional models as ‘structural’ disturbances is always perilous. Identifying assumptions may be quite controversial when the model is large. Over- parameterization of the reduced form model also affects the precision of the estimates.