From: The wage returns to on-the-job training: evidence from matched employer-employee data
Panel A: Papers using worker level data - developed countries | ||||||
---|---|---|---|---|---|---|
Name of study | Data, country and time period | Dependent variable | Training | Other independent variables | Effects training on wages | Controls endogeneity training |
Bassinini, Booth, Brunello, De Paola and Leuven (2005) | Worker Level Data for Europe (1995-2001) | log hourly wage | sum over the sample period (6 years) of training events | age, gender, marital status, schooling, year, country and industry. | Between 3.7% and 21.6% depending on the country. | No |
Albert, Serrano, Hernanz (2010) | Worker Level Data for Europe (1995-2001) | log hourly wage | dummy: having participated in training between January of the previous year and the date of the interview | gender, educational attainment, potential labour market experience, firm size, industry affiliation, working time, occupation, seniority | Positive for OLS but not statistically significant for FE. | Yes |
Fixed effects | ||||||
BudrÃa and Pereira (2004) | Worker Level Data for Portugal (1998-2000) | log hourly wage | dummy: having ever participated in training | age, experience, schooling (and interactions between these variables and training), part time, tenure, sector, firm size. | 12% for men and 37% for women | Yes |
Excluded instruments (selection model): having a second activity and having lived abroad. | ||||||
Dearden, Reed and Reenen (2006) | Worker Level Data for U.K. (1992,1997) | log hourly wage | dummies: having participated in training in the previews 4, 3, 2, and 1 quarters | age, gender, occupation, dummy for no qualifications, firm size, industry. | 0.15% (production sector only) | No |
Leuven and Oosterbeek (2002) | Worker Level Data for Holland (2001) | log hourly wage | dummy: having participated in training in the previous 12 months | age, gender, schooling, firm size. | Not statistically different from zero | Yes |
Randomization: control group composed by people that were planning to engage in a training activity by did not because of some random event. | ||||||
Leuven and Oosterbeek (2004) | Worker Level Data for Holland (1999) | log hourly wage | dummy: having participated in training in the previous 12 months | age, schooling, tenure, firm size. | Not statistically different from zero (for 40-year-old workers) | Yes |
They use the RD data design method. They explore the discontinuity introduced by a new tax law that allows tax deduction for firms’ expenditures on training for workers with more than 40 years. So the decision of training workers around age 40 suffers and will be influenced by an exogenous effect (the law). | ||||||
Lillard and Tan (1986) | Worker Level Data for U.S.A. (1983) | log annual wage | dummies: having participated in training (formal and informal) in the current job | experience, schooling, tenure, union member, dummy for non-white, tenure, region, long run state unemployment rate, cyclical sensitivity of state unemployment. | 22% for formal training | No |
Sousounis (2009) | Worker Level Data for U.K. (1998-2005) | log weekly wage | dummy: having participated in training in the previous 12 months | age, gender, marital status, dummy for having children with less than 12 years in the household, race, schooling, dummies for having changed job, private sector, part time managerial position, supervisor, firm size, region and time. | Negative (-3% for OLS) but not statistically significant for FE. | Yes |
 |  |  |  |  |  | Fixed effects |
Panel B:Papers using worker level data - developing countries | ||||||
Name of study | Country and time period | Dependent variable | Training | Other independent variables | Effects training on wages | Controls endogeneity training |
Chung (2000) | Worker Level Data for Malaysia (1976, 1988) | log hourly wage | dummy: having ever participated in training | age, marital status, nationality, schooling, dummies for employers and unpaid family workers. | 20%-30% (for women) | Yes |
Excluded instruments (selection model): having a bank account, level of education in 1976, and parents occupational status. | ||||||
Frazer (2006) | Worker Level Data for Ghana (1991-1999) | log hourly wage | dummy: having participated in an apprenticeship | gender, potential experience, schooling. | Not statistically different from zero for the whole sample but 17% for self-employed. | No |
Johanson and Wanga (2008) | Worker Level Data for Tanzania (2006) | log hourly wage | dummy: having ever participated in training (per type: on-the-job, informal apprenticeship, vocational, college/advanced) | experience, gender, occupation, schooling, rural dummy and region. | 38% for on-the-job training, 27% for formal apprenticeship, 47% for vocational training and 77% for college certificated training. | No |
Kahyarara and Teal (2008) | Worker Level Data for Tanzania (1997-2000) | log monthly wage | dummies: current and past on-the-job training and going on a short training course in the previews six months | gender, potential experience, occupation, schooling, tenure, dummy for capital city, firm fixed effects. | 22% for current on-the-job training, not statistically different from zero for past on-the-job training and 17% for short training courses. | No |
Monk, Sandefur and Teal (2008) | Worker Level Data for Ghana (1984, 2000) | log monthly wage | dummy: having participated in an apprenticeship | gender, potential experience, schooling, log hours worked per week, IQ score, interaction between apprenticeship and schooling, city. | 50% for people with no formal education. The return declines as education rises. | Yes |
Members of the household that also made an apprenticeship, dummy for household access to credit, and a dummy for having internal piped water in the house as a wealth indicator. | ||||||
Rosholm, Nielsen, Debalen (2007) | Matched employer-employee data for Kenya and Zambia (1995) | log monthly wage | dummy: having participated in training in the previous 12 months | age, ethnicity, experience, gender, occupation, schooling, tenure, union participation and familiar relations within the owners of the firm, ownership, industry, location, size, financial situation, skill demand, turnover, unionization, training annual expenses. | 2.3% for Kenya and not statistically different from zero for Zambia. | Yes |
Matching Estimators Method (Local Linear Matching) | ||||||
Almeida, Faria | Matched employer-employee data for Malaysia (2002) and Thailand (2004) | log hourly wage | dummy: having received formal on the job training since having joined the firm | educational attainment, gender, age, tenure, experience, marital status, occupation, union participation, computer, bank account, internet transaction, training at a previous employer, size, foreign capital, exports, education of the work force, education of the manager, new production technologies, industry, region. | 7.7% for Malysia and 4.5% for Thailand. | Yes |
 |  |  |  |  |  | Matching Estimators Method (Local Linear Matching) |
Panel C: Papers using firm level data | ||||||
Name of study | Data, country and time period | Dependent variable | Training | Other independent variables | Effects training on wages | Controls endogeneity training |
Almeida and Carneiro (2008) | Firm Level Data for Portugal (1995-1999) | log value added per employee | average number of hours of training per employee | log employees, log capital stock, share occupation group, share low educated workers, share males workforce, cubic polynomial on average wage workforce, year dummies, region and sector | 24% for firms providing training. | Yes |
First differences, GMM: past level of training as a instrument for current training. | ||||||
Barrett and O’Connel (1999) | Firm Level Data for Ireland (1993, 1995) | productivity (out-put divided by total employment) growth | average training days per worker | investment, change on employment, sector, size, innovation, restructuring, management quality, dummies for labor incentives strategies. | Increasing one day of training per worker increases productivity growth by 0.03%. | Yes |
Dependent variable is productivity growth | ||||||
Tan and Lopez-Acevedo (2003) | Firm Level Data for Mexico (1992, 1999) | log monthly wage | dummy: firm offered training in the previous 12 months | average years of schooling of the workforce, percentage of women, occupation, ownership, exports, size, industry and region | Training returns increased from 5% to 7% from 1992 to 1999. | Yes |
 |  |  |  |  |  | Excluded instruments (selection model): years in operation, R&D, computerization, unionization. |
Panel D: Papers using industry level data | ||||||
Name of study | Country and time period | Dependent variable | Training | Other independent variables | Effects training on wages | Controls endogeneity training |
Dearden, Reed and Reenen (2000) | Industry Level Data for U.K. (1984-1996) | log hourly wage | industry aggregated incidence for training in the previous 4 weeks | log capital per worker, log hours per worker, log of R&D over sales, region, time and tenure dummies, proportion of: men, age groups, occupation, qualified workers, small firms. | Raising training incidence by 5% increases wages and productivity by 1.6% and 4% respectively. | Yes |
 |  |  |  |  |  | Panel data: Within groups estimator |