Leadership Experiences, Labor Market Entry, and Early Career Trajectories

Notes: Estimates based on Equation 1 using the threshold as an instrument for being elected into the student union council. All models include list fixed effects. The model does not include any additional covariates and corresponds to the first column in Table 3 below. The sample includes the first five candidates on each side of the threshold. Sample restrictions: Column 4 drops three observations with missing data on first year of studies. Column 5 drops 323 observations with missing data on employment in the year before (cohorts 1982 and 1985 and three observations not residing in Sweden). Column 6 drops 246 observations without lagged information on student union representatives within that university. The last three columns have slightly fewer observations due to missing values on the outcome variables (from truncation of the sample window). Standard errors are clustered at the individual level and robust to heteroskedasticity. Significance:

* p < 0.10,

** p < 0.05,

*** p < 0.01.

Notes: Estimates are from instrumental variables models (see Equation 1) using the threshold as an instrument for being elected to the student union council. Covariates are the ones presented in Table 2 and indicators for missing values of the last three of these. “Bandwith 5 + 5” indicates closest five on each side. Sample restrictions: All columns drop 64 observations with missing data on employment in the year after election (1982 cohort and ten not residing in country). Column 5 drops an additional 690 observations ranked four to five slots away from the threshold. Column 6 drops 1,584 observations ranked two to five slots away from the threshold. Standard errors (within parentheses) are clustered for repeated observations at the individual level and robust to heteroskedasticity. Significance:

↵* p < 0.10,

↵** p < 0.05,

↵*** p < 0.01.

Notes: Estimates are from instrumental variables models (see Equation 1) using the threshold as an instrument for being elected to the student union council. All models include list fixed effects. The sample consists of the five candidates closest to the threshold on each side. Sample restrictions: All columns drop 64 observations with missing data on employment in the year after election (1982 cohort and ten not residing in country). Samples in Columns 4 and 5 use 2,103 observations in total since an additional three observations are dropped due to missing data on first year of study. Samples in Columns 6 and 7 use 1,839 observations in total since observations with missing data on employment in the year before are dropped (1982 and 1985 cohorts and three not residing in country). Note that eight observations are missing due to not residing in country in the year after, which explains the discrepancy between here and Column 5 of Table 2. Standard errors (within parentheses) are clustered for repeated observations at the individual level and robust to heteroskedasticity. Significance:

↵* p < 0.10,

↵** p < 0.05,

↵*** p < 0.01.

Notes: Estimates are from instrumental variables models (see Equation 1) using the threshold as an instrument for being elected to the student union council. All models include list fixed effects. The sample consists of the five candidates closest to the threshold on each side. Sample restrictions: Credit points are only available from 1993, which reduces the number of observations in Panel A. The graduation analysis of Panel B excludes students with prior degrees. This reduces the sample size by 244 observations from the main sample of 2,170. The reason for using this restriction is that we lack information about new degrees in cases where students take multiple degrees at the same level (for example, dual bachelor degrees). Standard errors (within parentheses) are clustered for repeated observations at the individual level and robust to heteroskedasticity. Significance:

↵* p < 0.10,

** p < 0.05,

*** p < 0.01.

Notes: Estimates are from instrumental variables models (see Equation 1) using the threshold as an instrument for being elected to the student union council. All models include list fixed effects. The sample consists of the five candidates closest to the threshold. Covariates are the ones presented in Table 2 (sex, immigrant status, age, years since start of studies, employment status the year before the election, and previous experience of council membership) and indicators for missing values of the last three of these. Note that these “covariates” include a dummy for past seats. Sample restrictions: Column 1 drops 64 observations with missing data on employment in the year after election (1982 cohort and ten not residing in country). Column 2 drops 128 observations with missing data on elected representatives in the year after. Column 3 drops 64 observations with missing data on employment the year after election (1982 cohort and ten not residing in country) and an additional 730 observations with missing data on election outcomes the year before. Columns 4 and 5 drop 31 observations not residing in Sweden three years after election, and Column 5 also drops 644 observations with missing values on election outcomes in the year after. Standard errors (within parentheses) are clustered for repeated observations at the individual level and robust to heteroskedasticity. Significance:

↵* p < 0.10,

↵** p < 0.05,

↵*** p < 0.01.

Notes: The outcomes are measured in the year following the election. The dependent variable is calculated using elected candidates during the previous five years, excluding those who also ran during the relevant (election) year. Estimates are regression coefficients from instrumental variables models (see Equation 1) using the threshold as an instrument for being elected to the student union council. All models include list fixed effects. The sample consists of the five candidates closest to the threshold. Sample restrictions: All columns drop 64 observations with missing data on employment the year after election (1982 cohort and ten not residing in country). Standard errors (within parentheses) are clustered for repeated observations at the individual level and robust to heteroskedasticity. Significance:

↵* p < 0.10,

** p < 0.05,

*** p < 0.01.