For labour market reforms, the findings signal different insights with regard to the role of
economic crisis. Definitely, a bad cyclical situation is not helpful to prepare the ground for
deregulating labour markets, on the contrary: Reform cycles start in periods with significantly
above normal growth, strong internal demand and a high level of consumer and business
confidence. This is an interesting insight: Times of mental depression are obviously no times to
embark on dismantling labour market protection or cutting unemployment benefits. Interestingly
the level, not the change in the unemployment rate, is significantly higher prior to reforms. This
could be cautiously interpreted as a positive impact of structural and not cyclical unemployment
on the likelihood of reforms.
The testing for financial market reforms does not hint towards significant characteristics on the
outset of financial market reforms apart from a higher business confidence and private housing
investment. The findings for tax and product market reforms are more in line with the crisis
hypothesis since here a number of significantly worse business cycle indicators indicate a
relatively poor situation in the two years preceding reforms. The typical situation in the years
before a tax reform is undertaken is thus one of a combination of a relatively comfortable
budgetary situation with a poor growth performance.
The situation prior to reforms thus reveals some insights on the fiscal preconditions for reforms.
Clearly, a favourable budgetary situation simplifies tax and product market reforms. For a better
understanding of the fiscal repercussions of reforms it is necessary to follow fiscal and economic
indicators over the reform cycle which is the next analytical step.
3.3 Developments over the reform cycle
The second, distinct step of the descriptive scrutiny targets at detecting significant differences in
the economic and fiscal setting over the reform cycle. After the first step was meant to show to
which extent the starting position of reforms is different to years of institutional stability, the
attention shifts now towards the evolution over the reform process.
Focusing the analysis on years around major reforms drastically limits the number of included
observations so that the same test as in the first step requiring a large number of observations
cannot be applied. Instead, an analysis of variance is executed comparing means before, during
and after a reform event. Before and after a reform year two years are taken account of. Years
145