Labour Supply in a High Unemployment Economy: Regional Dummies
Findings on the regional dummies for males do not support our argument of higher participation of residents from the region of Prishtina. The coefficients on regional dummies for males are individually insignificant (based on the t-test) and also jointly insignificant (X2=9.60, p=0.1428). For females, 4 out of 6 coefficients on regional dummies are significant and positive (based on the t-test), while the 6 of them are jointly significant (X2 =27.90, p=0.001). In two of these cases (Peja and Gjakova that are among the western and southern regions) findings support our expectations of higher participation due to working in predominantly agricultural regions. Given that our sample is representative at the regional level, we have estimated the same regressions with regional employment-to-population ratios substituting for the regional dummies to capture the discouragement effect due to a more depressed labour market. The estimated coefficient was insignificant for both genders. Other coefficients did not change their significance level and hardly changed their size. This is an expected finding given the current regional boundaries encompass small distances and males can easily commute to take advantage of better employment chances or higher wages. While for females, there is some evidence that they are less likely to commute and their labour supply decisions are made within the regional context.
The size of the effect of explanatory variables on the probability of labour supply is shown in Table A3. Panel A shows the (base) probability of labour force participation for an individual with given characteristics. Rural males and females have a higher probability of participation compared to their urban counterparts respectively. Panel B shows the change in the probability of labour supply due to a change in some of the characteristics keeping others the same. An interesting finding is the large effect of education on participation for females in particular. Increasing the level of education from less than upper-secondary education to upper-secondary increases the probability of participation for urban males by 15 percentage points (from 0.60 to 0.85), for urban females by 27 percentage points, for rural males by 5 percentage points and for rural females by 21 percentage points. Note also the large decrease in the estimated probability of participating in the labour force for a 50 years old female (urban and rural females).
To sum up, unlike most of the studies reviewed above we define participation as being in employment or unemployment (which as explain above and in Figure A1, both follow ILO guidelines). This acknowledges the difference between actual (observed) and desired hours of labour supply, which takes into account the demand side constraints on labour supply. We find evidence for the expected concavity of LFP rate with respect to age. Findings also support the view that education affects positively the probability of labour supply and this effect is estimated to be larger for urban residents compared to their rural counterparts. In line with findings elsewhere, our estimate suggests that married females are less likely to participate than non-married females. For rural females, there is weak evidence for the expected negative effect of a household member abroad on the probability of labour supply. The expected negative effect of household non-labour income on labour supply is supported only for urban males.