Skip to main content
Large-scale Assessments in Education

An IERI – International Educational Research Institute Journal

Table 1 Returns to skills: international evidence.

From: Skills, earnings, and employment: exploring causality in the estimation of returns to skills

 

Pooled

Australia

Austria

Belgium

Canada

Chile

Cyprus

Czech R.

Denmark

Estonia

Finland

Baseline (OLS)

.200***

(.003)

.205***

(.016)

.182***

(.012)

.151***

(.009)

.204***

(.009)

.367***

(.033)

.141***

(.020)

.122***

(.018)

.135***

(.009)

.177***

(.013)

.140***

(.012)

Measurement error—literacy (2SLS)

.221***

(.003)

[61223]

.216***

(.014)

[2824]

.218***

(.013)

[2096]

.169***

(.011)

[3141]

.231***

(.009)

[7334]

.378***

(.029)

[649]

.173***

(.023)

[847]

.159***

(.022)

[581]

.156***

(.009)

[5153]

.168***

(.015)

[3096]

.154***

(.013)

[2812]

Measurement error—second PV (2SLS)

.232***

(.003)

[65811]

.250***

(.015)

[2264]

.211***

(.012)

[1859]

.184***

(.011)

[2687]

.242***

(.009)

[5054]

.402***

(.030)

[2238]

.182***

(.024)

[1142]

.149***

(.019)

[1377]

.162***

(.010)

[3346]

.209***

(.015)

[2873]

.164***

(.014)

[2062]

Reverse causality—years school (2SLS)

.517***

(.007)

[7569]

.492***

(.039)

[197]

.471***

(.036)

[179]

.322***

(.022)

[324]

.489***

(.023)

[553]

.673***

(.044)

[342]

.752***

(.083)

[93]

.313***

(.040)

[149]

.370***

(.023)

[382]

.450***

(.032)

[369]

.477***

(.036)

[170]

Reverse causality—parents (2SLS)

.461***

(.012)

[1862]

.516***

(.083)

[48]

.384***

(.045)

[69]

.315***

(.041)

[79]

.338***

(.032)

[199]

.635***

(.085)

[47]

.502***

(.117)

[38]

.255***

(.056)

[41]

.286***

(.040)

[94]

.487***

(.064)

[95]

.364***

(.080)

[24]

Omitted variables—parents (OLS)

.182***

(.003)

.189***

(.017)

.165***

(.012)

.138***

(.010)

.194***

(.009)

.324***

(.034)

.129***

(.021)

.112***

(.019)

.127***

(.010)

.155***

(.015)

.137***

(.012)

Omitted variables—health (OLS)

.194***

(.003)

.204***

(.016)

.173***

(.011)

.149***

(.009)

.204***

(.009)

.335***

(.035)

.132***

(.020)

.115***

(.017)

.132***

(.009)

.164***

(.014)

.135***

(.012)

Observations

45064

1433

1115

1221

7197

903

938

1066

1879

1763

1480

 

France

Germany

Greece

Indonesia

Ireland

Israel

Italy

Japan

Korea

Lithuania

Netherl.

Baseline (OLS)

.175***

(.009)

.239***

(.015)

.101***

(.019)

.245***

(.036)

.230***

(.023)

.273***

(.020)

.135***

(.018)

.181***

(.014)

.210***

(.019)

.172***

(.016)

.180***

(.011)

Measurement error—literacy (2SLS)

.179***

(.010)

[4326]

.268***

(.015)

[2976]

.106***

(.024)

[601]

.331***

(.049)

[538]

.259***

(.023)

[1583]

.321***

(.021)

[1551]

.154***

(.020)

[1515]

.162***

(.016)

[2778]

.235***

(.017)

[4204]

.169***

(.019)

[1982]

.217***

(.014)

[2252]

Measurement error—second PV (2SLS)

.200***

(.010)

[4940]

.274***

(.015)

[2249]

.137***

(.024)

[763]

.306***

(.043)

[661]

.280***

(.021)

[1152]

.320***

(.020)

[2233]

.145***

(.019)

[1535]

.221***

(.017)

[1862]

.249***

(.018)

[2211]

.198***

(.018)

[1590]

.215***

(.014)

[2082]

Reverse causality—years school (2SLS)

.361***

(.019)

[524]

.456***

(.030)

[361]

.455***

(.068)

[62]

.747***

(.080)

[190]

.512***

(.042)

[172]

.572***

(.042)

[190]

.423***

(.045)

[220]

.478***

(.037)

[266]

.563***

(.040)

[332]

.600***

(.054)

[156]

.456***

(.039)

[178]

Reverse causality—parents (2SLS)

.354***

(.037)

[129]

.370***

(.049)

[86]

.448***

(.146)

[13]

1.381***

(.394)

[12]

.509***

(.074)

[64]

.562***

(.075)

[68]

.466***

(.093)

[51]

.717***

(.131)

[32]

.540***

(.068)

[82]

.401***

(.062)

[67]

.567***

(.096)

[30]

Omitted variables—parents (OLS)

.157***

(.010)

.228***

(.017)

.091***

(.019)

.210***

(.034)

.214***

(.025)

.251***

(.021)

.114***

(.020)

.166***

(.015)

.187***

(.019)

.150***

(.017)

.173***

(.012)

Omitted variables—health (OLS)

.172***

(.009)

.234***

(.015)

.099***

(.019)

.231***

(.036)

.230***

(.023)

.264***

(.019)

.135***

(.018)

.180***

(.014)

.208***

(.019)

.158***

(.016)

.176***

(.011)

Observations

1715

1296

623

806

1033

908

1019

1319

1441

1260

1012

 

New Zealand

Norway

Poland

Singapore

Slovak R.

Slovenia

Spain

Sweden

Turkey

U.K.

U.S.

Baseline (OLS)

.191***

(.013)

.132***

(.008)

.190***

(.018)

.466***

(.015)

.188***

(.021)

.186***

(.012)

.225***

(.017)

.130***

(.009)

.208***

(.035)

.220***

(.015)

.254***

(.028)

Measurement error—literacy (2SLS)

.206***

(.013)

[2848]

.131***

(.008)

[4754]

.250***

(.021)

[1483]

.444***

(.014)

[8089]

.184***

(.023)

[1782]

.222***

(.013)

[3120]

.261***

(.020)

[2265]

.139***

(.010)

[2646]

.273***

(.036)

[509]

.262***

(.016)

[1978]

.278***

(.022)

[2947]

Measurement error—second PV (2SLS)

.221***

(.012)

[3184]

.152***

(.009)

[2990]

.221***

(.020)

[1587]

.501***

(.015)

[7153]

.225***

(.023)

[1709]

.224***

(.013)

[2484]

.259***

(.018)

[2350]

.143***

(.009)

[2489]

.246***

(.032)

[703]

.247***

(.015)

[2484]

.274***

(.023)

[3389]

Reverse causality—years school (2SLS)

.460***

(.036)

[217]

.330***

(.026)

[257]

.642***

(.055)

[161]

.723***

(.025)

[1055]

.742***

(.071)

[142]

.506***

(.030)

[405]

.653***

(.040)

[322]

.230***

(.023)

[272]

.804***

(.096)

[87]

.606***

(.058)

[117]

.487***

(.033)

[473]

Reverse causality—parents (2SLS)

.363***

(.063)

[43]

.278***

(.036)

[82]

.740***

(.133)

[31]

.747***

(.062)

[133]

.654***

(.091)

[66]

.461***

(.043)

[135]

.676***

(.117)

[34]

.201***

(.035)

[55]

.649***

(.149)

[34]

.408***

(.050)

[120]

.443***

(.052)

[131]

Omitted variables—parents (OLS)

.182***

(.014)

.120***

(.009)

.168***

(.018)

.445***

(.016)

.148***

(.020)

.157***

(.014)

.206***

(.019)

.127***

(.009)

.193***

(.036)

.188***

(.019)

.213***

(.030)

Omitted variables—health (OLS)

.186***

(.013)

.128***

(.008)

.183***

(.018)

.456***

(.015)

.182***

(.021)

.178***

(.013)

.224***

(.018)

.127***

(.008)

.208***

(.035)

.214***

(.014)

.236***

(.027)

Observations

1204

1520

817

1507

1193

1306

1192

1316

674

1787

1121

  1. Dependent variable: log gross hourly wage. Each cell reports the coefficient on numeracy skills from a separate regression. Regressions weighted by sampling weights. Model names indicate bias addressed by the specification; “OLS” refers to ordinary least squares estimations, “2SLS” refers to two-stage least squares estimations. Least squares estimates (except for pooled specification) take into account all 10 plausible values of numeracy skills. Sample: full-time employees aged 35–54 years. All regressions control for a quadratic polynomial in actual work experience and gender. Baseline standard model controlling for work experience and gender. Measurement error—literacy numeracy skills are instrumented by literacy skills. Measurement error—second PV numeracy skills are instrumented by second plausible value of numeracy skills. Reverse causality—years school numeracy skills are instrumented by years of schooling. Reverse causality—parents numeracy skills are instrumented by parental education (1 neither parent attained upper secondary education, 2 at least one parent attained upper secondary education, 3 at least one parent attained tertiary education). Omitted variables—parents adds control for parental education. Omitted variables—health adds control for respondent’s health (1 poor, 2 fair, 3 good, 4 very good, 5 excellent). Observations refer to baseline specification. Numeracy skills standardized to SD 1 within country. Pooled specification includes country fixed effects and gives same weight to each country. Robust standard errors (jacknifed in the OLS models due to the complex sampling design) in parentheses. F-statistics of excluded instrument in the first stage of 2SLS models in brackets. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01
Back to article page