is it worth to study two majors? the case of poland dominik buttler education and work: (un-) equal...
DESCRIPTION
Context of the analysis Enrollment Ratio, tertiary level, Poland 1990/911995/962000/012005/062010/112011/122012/132013/14 gross12,922,340,748,953,853,151,849,2 nett9,817,230,638,040,840,640,238,6 Source: Central Statistical Office of PolandTRANSCRIPT
Is it Worth to Study Two Majors? The Case of Poland
Dominik ButtlerEducation and Work: (Un-) equal TransitionsSofia, 24-25 September 2015
Schedule
• Context of the analysis and research question • Literature• Empirical strategy • Dataset • Results• Conclusions and further questions
Context of the analysis
Enrollment Ratio, tertiary level, Poland
1990/91 1995/96 2000/01 2005/06 2010/11 2011/12 2012/13 2013/14
gross 12,9 22,3 40,7 48,9 53,8 53,1 51,8 49,2
nett 9,8 17,2 30,6 38,0 40,8 40,6 40,2 38,6
Source: Central Statistical Office of Poland
Context of the analysis
Enrollment Ratio, tertiary level, Poland
1990/91 1995/96 2000/01 2005/06 2010/11 2011/12 2012/13 2013/14
gross 12,9 22,3 40,7 48,9 53,8 53,1 51,8 49,2
nett 9,8 17,2 30,6 38,0 40,8 40,6 40,2 38,6
90' 91' 92' 93' 94' 95' 96' 97' 98' 99' 00' 01' 02' 03' 04' 05' 06' 07' 08' 09' 10' 11' 12' 13'0
50100150200250300350
Higher education institutions in Poland, 1990-2014
private public
Source: Central Statistical Office of Poland
Context of the analysis
Share of students by field of HE in selected countries, 2013
country education humanities socialbusiness, law sum
Poland 12,0 7,0 35,7 54,7Bulgaria 6,8 5,0 40,2 52,0Switzerland 9,5 6,6 34,9 51,0France 2,5 8,8 38,3 49,6Germany 9,1 10,7 28,9 48,7Romania 2,0 6,9 39,6 48,5Czech Republic 9,3 7,0 31,6 47,9Sweden 12,4 8,8 26,7 47,9UK 8,0 8,8 26,1 42,9
Source: eurostat
Context of the analysisSa
lary
HE
degr
ee/s
alar
y ot
hers
Shar
e of
wor
kers
with
HE
degr
ee
Source: Gajderowicz, Grotkowska, Wincenciak, 2012
Salary and employment of workers with HE degree
Introduction of fees for studying a second major
• Second major (two majors) vs. double major• justification referring both to economic efficiency and social justice • full time students at public universities affected
– 10,1 % students declared studying more than one major (8,5 % at public universities)
• possible negative impact on the financial stability of unpopular departments
• relatively strong protests of students• new law declared to be unconstitutional
Is it worth to study two majors, after all?
Literature
• human capital and/or signaling theory
• Hemelt (2010): a wage premium of 3,2% (slightly higher for women) among graduates in US (National Survey of College Graduates)
• Del Rossi and Hersch (2008): a wage premium up to 2,3% among graduates in US (National Survey of College Graduates), high premium of arts, humanities or social science students taking business as a second major
• Zafar (2012): Double Major: One for me, One for the Parents (sample of Northwestern University sophomores)
Empirical strategy
• The empirical analysis with the use of the unique dataset from the survey Study of Human Capital in Poland
• Descriptive statistics on what majors are combined
• Regression analysis, wage determinants in the sample of tertiary education graduates
• Propensity score techniques, estimation of the average treatment effect on the treated– Nearest neighbor matching– Kernel matching
Empirical strategyRegression analysis
Y – net monthly wageD – whether someone graduated from two majors (binary variable)X – set of independent variables:
Empirical strategyRegression analysis
Y – net monthly wageD – whether someone graduated from two majors (binary variable)X – set of independent variables:
job characteristics variables• professional experience• tenure• employment type• managerial position
Empirical strategyRegression analysis
Y – net monthly wageD – whether someone graduated from two majors (binary variable)X – set of independent variables:
job characteristics variables• professional experience• tenure• employment type• managerial position
two major decision variables• education of parents• whether someone had to work after high school• whether someone graduated from vocational school
Empirical strategyRegression analysis
Y – net monthly wageD – whether someone graduated from two majors (binary variable)X – set of independent variables:
job characteristics variables• professional experience• tenure• employment type• managerial position
two major decision variables• education of parents• whether someone had to work after high school• whether someone graduated from vocational school
socio-demographic status• educational level (PhD, master, postgraduate)• gender• marital status• city size
Empirical strategyEstimation of the average treatment effect on the treated
Empirical strategyEstimation of the average treatment effect on the treated
Empirical strategyEstimation of the average treatment effect on the treated
Y – net monthly wageD – whether someone graduated from two majors (binary variable)PSM – based on (binary) variables which could possibly influence both the wage and the decision to study two majors• gender• whether at least one parent had HE degree• whether someone had to work after high school• whether someone graduated from vocational school at the secondary level
The Dataset
• Study of Human Capital in Poland
• Population/society survey (study of the population of working age)– people of working age (18-59 W, 18-64 M)– detailed information on employment status, job characteristics, curriculum,
qualifications and competences, training needs – five waves (2010-2014), around 17 000 cases each (not a panel)
• only one wave (2013) used in the analysis due to the selection of variables– individuals with HE degree– full time employees or self-employed
ResultsTwo majors and field of study
Share of students who graduated from more than one major (in %) by filed of studyartistic 11,76social sciences 7,59humanities 7,52natural sciences 5,66education 5,48law 5,11economics 4,13biology 4,05formal sciences 4,04engineering 3,20medicine 2,75agriculture 1,53total 3,20N=15205
ResultsTwo majors and field of study
Most popular combinations of study fields (as a % of two majors graduates)education & education 10,27
humanities & humanities 8,83
humanities & education 6,37
education & social sciences 6,16
economics & social sciences 5,75
economics & economics 4,11
N=487
Share of students who graduated from more than one major (in %) by filed of studyartistic 11,76social sciences 7,59humanities 7,52natural sciences 5,66education 5,48law 5,11economics 4,13biology 4,05formal sciences 4,04engineering 3,20medicine 2,75agriculture 1,53total 3,20N=15205
Regression analysisDescriptive statistics
variable mean sd
lnwage 7,859 0,473
phd 0,016
master 0,761
two majors 0,043
postgraduate 0,107
tenure 9,788 8,613
experience 13,467 9,913
self employed 0,125
married 0,692
N=1547
Regression analysisDescriptive statistics
variable mean sd variable mean
lnwage 7,859 0,473 female 0,612
phd 0,016 village 0,261
master 0,761 city less 100 0,359
two majors 0,043 city more 100 0,346
postgraduate 0,107 warsaw 0,034
tenure 9,788 8,613 parents high 0,184
experience 13,467 9,913 vocational 0,319
self employed 0,125 job at school 0,293
married 0,692
N=1547
Determinants of monthly earnings (ln)OLS regression coefficients
model 1phd 0.601***master 0.087***two majors 0.117**postgraduate 0.149***tenuretenure2experienceexperience2self employedparents_hivocationaljob_schoolr2_a 0.145N 1547
variables not reported: married, female, female*married, city size, cons*** p<0.01; ** p<0,05; * p<0,1
Determinants of monthly earnings (ln)OLS regression coefficients
model 1 model 2phd 0.601*** 0.515***master 0.087*** 0.065**two majors 0.117** 0.104**postgraduate 0.149*** 0.098***tenure 0.018***tenure2 -0.000***experience 0.014***experience2 -0.000self employed 0.166***parents_hivocationaljob_schoolr2_a 0.145 0.226N 1547 1547
variables not reported: married, female, female*married, city size, cons*** p<0.01; ** p<0,05; * p<0,1
Determinants of monthly earnings (ln)OLS regression coefficients
model 1 model 2 model 3phd 0.601*** 0.515*** 0.491***master 0.087*** 0.065** 0.048*two majors 0.117** 0.104** 0.099*postgraduate 0.149*** 0.098*** 0.109***tenure 0.018*** 0.017***tenure2 -0.000*** -0.000***experience 0.014*** 0.015***experience2 -0.000 -0.000self employed 0.166*** 0.164***parents_hi 0.072***vocational -0.044*job_school -0.089***r2_a 0.145 0.226 0.238N 1547 1547 1547
variables not reported: married, female, female*married, city size, cons*** p<0.01; ** p<0,05; * p<0,1
Determinants of monthly earnings (ln)OLS regression coefficients
model 1 model 2 model 3 model 4phd 0.601*** 0.515*** 0.491*** 0.488***master 0.087*** 0.065** 0.048* 0.048* two majors 0.117** 0.104** 0.099* 0.059 postgraduate 0.149*** 0.098*** 0.109*** 0.107***tenure 0.018*** 0.017*** 0.017***tenure2 -0.000*** -0.000*** -0.000***experience 0.014*** 0.015*** 0.015***experience2 -0.000 -0.000 -0.000 self employed 0.166*** 0.164*** 0.163***parents_hi 0.072*** 0.073***vocational -0.044* -0.045* job_school -0.089*** -0.087***r2_a 0.145 0.226 0.238 0.237 N 1547 1547 1547 1547
variables not reported: married, female, female*married, city size, cons*** p<0.01; ** p<0,05; * p<0,1
Estimation of treatment effectsdeterminants of studying two majors, logistic regression coefficients
coeff.female 0,148parents_hi 0,248*vocational -0,272**job_school 0,204*cons -1,859***ps_r2 0.0235N 1547
*** p<0.01; ** p<0,05; * p<0,1
Estimation of treatment effectsaverage treatment effect on the treated, ATT
Matching treated controls difference ATTS.E. (corr) T-stat (corr)
treatment matched %
NN 3092 1909,51 1183,04 0,619 371,37 3,18 100
Kernel 3092 2808,19 283,80 0,101 206,65 1,37 100
Estimation of treatment effectsPSM, NN matching , balance test
mean %reductvariable treated control bias t p>t
vocational school U 0,194 0,324 -2,24 0,025
M 0,194 0,194 100 0 1
Estimation of treatment effectsPSM, NN matching , balance test
mean %reductvariable treated control bias t p>t
vocational school U 0,194 0,324 -2,24 0,025
M 0,194 0,194 100 0 1
parents he U 0,283 0,179 2,16 0,031
M 0,283 0,283 100 0 1
job at school U 0,373 0,289 1,48 0,14
M 0,373 0,373 100 0 1
female U 0,701 0,608 1,53 0,125
M 0,701 0,701 100 0 1
PSM, kaliper matching , balance test
mean %reductvariable treated control bias t p>t
vocational school U 0,194 0,324 -2,24 0,025
M 0,194 0,205 91,2 -0,16 0,869
parents he U 0,283 0,179 2,16 0,031
M 0,283 0,263 80,5 0,26 0,793
job at school U 0,373 0,289 1,48 0,14
M 0,373 0,364 89,6 0,10 0,918
female U 0,701 0,608 1,53 0,125
M 0,701 0,749 48,5 -0,62 0,536
Conclusions and further research
• Some evidence of a positive return to graduating from two majors– results not very robust– lack of useful variables, especially in PSM procedure
• Other possible effects worth investigation, e.g. impact on competences, unemployment, job match, quality of life
• Determinants of studying two majors– What characteristics share students of two majors?– Studying two majors as a safe studying strategy?
Thank you for your attention!