Labour Market Flexibility and Regional Unemployment Rate Dynamics: Spain (1980-1995)

As noted in the previous section, the empirical model consists of three estimated
equations: labour demand, labour supply, wage setting, and the definition of the unem-
ployment rate. The structure of our labour market system is in the spirit of the models
presented in Karanassou and Snower (1998), and Henry, Karanassou and Snower (2000).

Tables 5 and 6 present the estimated models for the high and low unemployment
groups of regions, respectively. Fixed effects estimation implies that within a specific
group, differences in labour market behaviour across regions is captured solely through
fixed effects: only differing constants in the estimated equations (but identical coefficients
for the exogenous variables and the endogenous regressors).³² The Schwarz model selection
criterion prefers this fixed-effect model over heterogeneous models containing individual
region time series regressions.³³

In the labour demand equation, employment depends negatively on the real wage,
and positively on both the level and growth rate of the capital stock. The oil price and
indirect taxes (as a ratio to GDP) have a negative impact on labour demand. The lagged
employment terms capture the employment adjustment process. All the explanatory
variables are highly significant with the exception of the tax rate that is significant at the
20% (14%) size of the test for the high (low) unemployment group of regions.

In the wage setting equation, real wage depends negatively on unemployment and im-
port prices,³⁴ and positively on productivity and benefits. The lag of real wage captures
the adjustment process due to wage staggering. Except for benefits in the low unemploy-
ment group of regions and unemployment in both panels, which are significant at the 15%
size of the test, all other variables are statistically significant at conventional levels.

Finally, in the estimated labour supply equation, the size of the labour force depends
positively on working age population and negatively on the real wage.³⁵ The statistical
significance of past labour force is associated with the labour force adjustment process.

³²We do not show the region-specific coefficients, which are the constants or fixed effects of the model.
Results are available upon request.

³³ Specifically, we select between each of the pooled equations presented in Tables 5 and 6 and the
corresponding individual regressions by using the Schwarz Information Criterion (SIC). We compute the
model selection criteria as follows:

SICpooled = MLL-0.5kpooledlog(NT),

j

SICindividual = MLLi - N [0.5ki log (T)], j = 11,6
i=1

where M LL_pool_ed , MLLi denote the maximum log likelihoods of the pooled model and the ith region
time series regression, respectively; kp_ool_ed , ki are the number of parameters estimated in the fixed effects
model and the individual region time series regression, respectively; N is the number of regions and T is
the time dimension of the sample size. The model that maximises SIC is preferred. (Results are available
upon request.)

³⁴ Real import prices proxy the competitiveness of the economy.

³⁵ The negative impact of the real wage indicates that the income effect dominates.

14

<<   11   12   13   14   15   16   17   >>

More intriguing information