Vocational Training Program for the Unemployed Impact Evaluation 2010-2012

The selection of provinces for evaluation began with a list of the 39 provinces which had at least two significantly oversubscribed training courses in 2009. These provinces were first stratified by whether they had an unemployment rate above or below the median of 10 percent in 2009. Ten provinces were then randomly selected from each strata with probability proportional to the percentage of individuals trained in 2009. Three additional provinces (Antalya, Gaziantep, and Diyarbakir) were included in the sample at the request of ISKUR because of their importance in representing varying labor market conditions across Turkey. As a result, 23 provinces were selected for inclusion in the evaluation.

Power calculations gave a target sample size of 5,700 individuals. This target was divided among the 23 provinces in proportion to the number of trainees in these provinces in the previous year. Thus Istanbul accounts for 21.8 percent of the sample, Kocaeli, Ankara and Hatay collectively 28 percent, and the remaining half of the sample is split among the other 19 provinces.

The evaluation team worked with regional ISKUR offices to determine the actual courses from within each province to be included in the evaluation. The key criteria used to decide which courses to include in the evaluation were i) the likelihood of the course being oversubscribed (which ensures the most popular types of training, for which there would be demand for further scale-up, are included); ii) inclusion of a diversity of types of training providers to enable comparison of private and public course provision; and iii) course starting and ending dates. The evaluation includes courses that started between October and December 2010 and finished by May 2011 (75 percent had finished by the end of February 2011). The timing of the evaluation was determined by the fact that it tends to be a time of year when people in Turkey are more likely to seek training through ISKUR.

This resulted in a set of 130 evaluation courses spread throughout Turkey, of which 39 were offered by private providers and the remainder were mainly government-operated. Courses were advertised and potential trainees applied to them following standard procedures. Applications were then screened to ensure they met the eligibility criteria of ISKUR and the course provider. Training providers were then asked to select a list of potential trainees that was at least 2.2 times capacity.

The ISKUR Management Information System (MIS) stratified applicants for each course by gender and whether or not they were less than 25 years old. Within these strata, the MIS randomly allocated trainees at the individual level into one of three groups: a treatment group who were selected for training, a control group who were not, and a waitlisted group who the training provider could select into the training if there were drop-outs. Since training providers are paid on the basis of number actually trained, if individuals assigned to treatment drop out of training, providers look to quickly fill in the empty spots.

The final evaluation sample consisted of 5,902 applicants, of which 3,001 were allocated to treatment and 2,901 to control groups. There were 173 individuals who applied to more than one course.