Browsing by Author "Kaila, Martti"

In Finland, children start school during the calendar year they turn seven years old. This creates a discontinuous jump in school starting age. Children born after New Year are on average one year older at the beginning of first grade than the children born before New Year. There exists a rich literature which uses school entry rules like this to study the effects of relative school starting age on various outcomes. This thesis investigates how the relative age difference affects educational outcomes in Finland. The primary data source of this thesis is the joint application register of the Finnish National Board of Education. The sample includes individuals who have applied to upper secondary school during the year they have graduated from comprehensive school between years 1991-2007. Furthermore, the individuals are linked to the Finnish Longitudinal Employer-Employee database. My main outcome variables are the grade point average (GPA) at the end of comprehensive school, and the probability of admittance and graduation from general upper secondary school. I utilize a regression discontinuity design to study whether the discontinuous jump in the school starting age at turn of the year affects these educational outcomes. I estimate the effect in window of ±30 days around new year. I find that the school starting age law generates a significant jump in the school starting age at the turn of the year, which in turn affects educational outcomes. According to my results, those who are born just after new year have on average a 0.15 grade points higher GPA and are significantly more likely to be admitted to and graduate from general upper secondary school. School starting rule is not completely binding. To take this into account, I use a fuzzy regression discontinuity design and show that the estimates on compliers are higher. The findings may be taken as a causal effect of relative school starting age. To support this, I show that the density of assignment variable and various background variables evolve continuously in the vicinity of New Year. Finally, I study heterogeneity in the results and find that the effect is significantly stronger for women than men. My results indicate that the magnitude of relative school starting age effect is similar to what has been found in other countries. This contradicts previous studies on Finland, which document smaller effects. There are three mechanisms which may explain the difference. Firstly, the deviation may arise from the optimal school starting age. Secondly, the gaps may be caused by peer effects and, lastly, relatively older children may perform better since they take the exams at an older age. I cannot distinguish between the different channels, and hence my results should be taken as a combined effect of all mechanisms.