Earnings of Swedish Young Men: Griliches’ (1976) study on the wage premium for formal schooling in America is replicated on the Swedish Malmö Longitudinal Study, an individual data base with unusually good ability measures collected at young age. The model is tried for two years, when the men were 25 and 30 years old, respectively. The wage premium is shown to be slightly less in the Swedish material. While Griliches found the ability bias to be between ten and twenty per cent, it is found to vary a lot between the alternative ability variables and between the two years tested in the Swedish material. Grade point average as an ability variable gives a large bias, approximately thirty percent in both years, while IQ from the age of twenty gives the highest bias, thirty-five percent. The schooling coefficient increases in a few cases when other background variables are used as proxies for, or together with, ability. Ability by way of intelligence quotas is not statistically significant before the men are thirty years old. A 2SLS is used towards the end to correct for the potential endogeneity of the schooling variable. A Hausman test does not reject the equality between the 2SLS estimator and the OLS estimator in either case.
Life Cycle Earnings and Wage Premiums: In this paper I use the Malmö Longitudinal Study and construct actual age-earnings profiles for the 1928 cohort of Malmö men. The emerging age-earnings profiles are rather similar to the approximated profiles by Mincer (1974). Real earnings for the highest educated men are found to have decreased significantly during the late 70s and early 80s. The same development is found for the wage premiums over the life cycle. A similar pattern has been found in other Swedish studies, but then from cross-sectional data. A significant difference is found between the wage premiums for vocational and lower secondary school, even though they involve the same number of years in education. Finally, ability is introduced in the model, and a proportionally much larger ability bias is found for the lower levels of education. This is explained with the larger spread of ability in the lower educational groups.
Comparing Annual and Lifetime Earnings: Distributions and Wage Premiums for a cohort of Swedish Men: In this paper I use the Malmö Longitudinal Data set and calculate actual lifetime earnings for the 1928 cohort of Malmö men. I compare distributions of, and the educational wage premiums for, annual and lifetime earnings. The distribution of lifetime earnings is found to be less than the average annual distribution, but not for each separate annual estimate. When comparing annual and lifetime wage premiums, the significance for the lower levels of education tend to disappear. The annual wage premiums seem to be overestimated compared to the lifetime premiums, except for academic studies which are underestimated annually.
Proxying ability by family background when estimating returns to schooling is not a good idea – it will only make matters worse: (written together with Erik Mellander) A regression model is considered where earnings are explained by schooling and ability. It is assumed that schooling is measured with error and that there are no data on ability. Regressing earnings on observed schooling then yields an estimate of the return to schooling that is subject to positive omitted variable bias (OVB) and negative measurement error bias (MEB). The effects on the OVB and the MEB from using family background variables as proxies for ability are investigated theoretically and empirically. The theoretical analysis demonstrates that the impact on the OVB is uncertain, while the MEB invariably increases in magnitude. The empirical analysis shows that the MEB strongly dominates the OVB, making the total bias in the estimated return negative. The total bias becomes more negative as more family background variables are added, driving the estimated return even farther away from the true value.
Parental Income, Lifetime Income and Mortality: (written together with Mårten Palme) We study the relation between parental economic resources and mortality among elderly. We use a data set on a cohort of individuals born in 1928 in the county of Malmö in southern Sweden, which contains exceptionally detailed measures of parental household income from 1937. The very rich information on individual earnings throughout their entire life cycle allows us to construct a measure of lifetime earnings. Date and cause of death are obtained from national registers. Using Cox proportional hazard models we find an inverse relationship between parental income and mortality, also when controlling for individual lifetime income and when studying those with high education separately. A competing risk analysis shows the relation between parental income and mortality to apply to cancer as the cause of death while own income later in life seems more important for circulatory diseases as the cause of death.
The Malmö Study is one of the longest longitudinal individual databases existing. In this chapter its history and substance is presented in more detail than in the other articles in the thesis. The main sources on historical background are Fägerlind (1975), Tuijnman (1989), and Furu (2000). I discuss my choice of variables for the econometric chapters from the wealth of information available in the Malmö data set. Some composite variables are presented in Section C. They were created to test Eliasson’s (1998) so called platform hypothesis of cumulative learning at school and on the job. Even though some variables (for instance recurrent education) appear to contribute significantly to earnings, the material collected so far is not up to capturing a possible interactive (cumulative) effect. For the time being we regard this work as an ongoing effort, but nevertheless present data created as an illustration of future uses of the material, for deeper studies on the determination of earnings and job careers. This work has been done in co-operation with Gunnar Eliasson.
Stockholm: KTH , 2005. , 10 p.
Business and economics, returns to education, ability bias, earnings distributions, lifetime earnings