Retaining through Training, Even for Older Workers Matteo Picchio - PowerPoint PPT Presentation

Retaining through Training, Even for Older Workers Retaining through Training, Even for Older Workers Matteo Picchio CentER, ReflecT, Tilburg University Sherppa, Ghent University IZA Jan C. van Ours CentER, ReflecT, Tilburg University University of Melbourne IZA, CEPR, CESifo Amsterdam 26 October 2011 M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Introduction Introduction On-the-job training of the workforce is important At macro level , the accumulation of human capital is the main engine of growth (Lucas, 1993) At micro level , on-the-job training is key factor for a sustained competitive advantage (Wright et al., 1995) In the recent crisis, temporary shorter working hours arrangements, often in combination with training, have been used as instruments ⇒ Many countries reported measures to provide training to existing workers at risk of job loss (OECD, 2010). M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Introduction Aim We study the effect of firm-provided training on the probability of being at work in the Netherlands. We also focus on the effect for older workers , as the labour market position of older workers is cause of concern in the NL (as in many European countries): demographic trends are causing an ageing of the workforce; older workers’ job separation rates are often a one-way street out of the labour force. M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Related Literature Related Literature: Adult Training The question whether adult training affects labour market performances and productivity of workers has been the core of substantial research. Training programmes are found to have a modest or no effect on unemployment exit rates (Gerfin and Lechner, 2002; Andr´ en and Andr´ en, 2006; Cr´ epon et al., 2007; Lechner et al., 2008; Lalive et al., 2008; Sianesi, 2008) and on wages (Heckman et al., 1999). Positive effects on future employment stability (Cr´ epon et al., 2007; Lechner et al., 2008) and job satisfaction (Jones et al., 2009) M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Related Literature Related Literature: On-the-Job Training Another branch of the literature focused on on-the-job training. It increases employment prospects (Gritz, 1993; Bonnal et al. 1997); wages (Bartel, 1995); firm productivity (Bartel, 1994; Barrett and O’Connell, 2001; Conti, 2005; Dearden et al., 2006) M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Related Literature Related Literature: in the NL Some studies have tried to infer the impact of different kind of training programmes on future employment prospects: Ridder (1986): especially women and minorities benefit from training programmes; Pavlopoulos et al. (2009): training increases the chances for upward wage mobility. M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Data and Sample Data Data are from the 1994–2001 waves of the European Community Household Panel (ECHP). The longitudinal ECHP for the NL comprise a number of individual records that range from 12,000 to 13,000 per year, for a total of 100,716 records. M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Data and Sample Sample The 1994 wave is lost due to lack of information in training. We remove individuals younger than 26 . We remove the self-employed . We remove observations with missing values . We remove individuals that are not in the sample for at least 3 consecutive waves . We end up with an unbalanced panel of 7,257 individuals , for a total of 33,348 records in the period 1996–2001. M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Data and Sample M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Data and Sample M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Data and Sample M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Econometric Model Econometric Model We model the impact of the firm-provided training indicator on the employment probability The training indicator is a potentially endogenous human capital variable: self-selection due to unobserved heterogeneity (ability, intelligence, motivations, labour force attachment); feedback effects from shock in the employment status to future training participation M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Econometric Model Econometric Model The interrelated dynamics between employment status and training participation are specified using a panel data dynamic bivariate unobserved effects probit model a jit = c ji + η jit is individual heterogeneity c ji is the permanent component capturing time-invariant characteristics (ability, intelligence) η jit is the transitory component (motivations, labour force attachment) M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Econometric Model Correlated RE and Initial Conditions The presence of unobserved heterogeneity generates 2 problems: get rid of the fixed effects c 1 i and c 2 i ⇒ Mundlak (1978) 1 version of Chamberlain’s (1984) approach initial conditions due to the dynamic and the correlation 2 between the unobserved heterogeneity and initial observations of the outcome variables ⇒ Wooldridge’s (2005) approach M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Econometric Model Unobserved Heterogeneity Distributions In order to avoid too strict parametric assumptions on the distribution of the permanent and transitory components of unobserved heterogeneity, we assume that v i ≡ ( v 1 i , v 2 i ) is a random draw from a discrete distribution 1 with a priori unknown number of support points η it ≡ ( η 1 it , η 2 it ) is a random draw from a discrete distribution 2 with a priori unknown number of support points. The number of support points is chosen by minimizing the AIC. M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Econometric Model Likelihood Function As v i and η it are not observed and are assumed to be random terms from bivariate distributions, we integrate them out when the model is estimated by ML. The contribution to the likelihood function of individual i is given by where Φ denotes the cdf of the standard normal distribution. M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Econometric Model Predicting Probabilities and APEs M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Econometric Model Identification We exploit different identification sources. The sequencing : training during the year before the interview 1 date can be assumed to be predetermined w.r.t employment status. However, firms might be more likely to train those workers that they seek to retain. To the extent that some of the time-varying variables are 2 strictly exogenous, the imposed stability of the structural parameters over time jointly with their time-variation will provide a multiplicity of instruments (Bhargava, 1991). Some household characteristics are included in the employment 3 equation but excluded from the training equation. Multiple observations per individual exploited to identify the 4 unobserved heterogeneity distribution. M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Estimation Results Univariate Models M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Estimation Results Benchmark Model M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers

Retaining through Training, Even for Older Workers Estimation Results Retaining Older Workers M. Picchio & J.C. van Ours Tilburg University Retaining through Training, Even for Older Workers