The Effectiveness of Social Investment Policies: Training and Childcare in OECD Countries

Tuesday, 17 July 2018: 09:15
Oral Presentation
Jae Hyoung PARK, London School of Economics and Political Science, United Kingdom
What is the relationship between expenditure on social investment policies (SIPs) and socio-economic outcomes across OECD countries since mis-1990s? Even though SIPs such as active labour market policies (ALMPs) and work-family policies (WFPs) aim to provide opportunities and reduce barriers for labour market entry, there is a debate on whether these policies are actually biased toward or against vulnerable groups more susceptible to new social risks in terms of the 'Matthew effect' (Cantillon et al 2001; Cantillon 2011; Cantillon and Van Lancker 2013). Furthermore, a growing number of scholars argue that SIPs should be complemented by social protection policies (SPPs) in order to compensate for a perverse effect of SIPs (Esping-Andersen et al 2002; Vandenbroucke and Vleminckx 2011; Leoni 2015). This paper focuses on what has been the impact of public spending on ALMPs and WFPs on the labour market outcomes of low-skilled workers and women in terms of three indicators: equality, job quality, and gender equality. We narrow down the critical elements of SIPs into spending on training and spending on childcare, most representative one of ALMPs and WFPs programmes respectively. Training, a major 'building block' of ALMPs, aim at improving the prospect of finding a job or increasing earning capacity, while childcare as a work-family reconciliation policy linked to increasing mothers' participation rather than to increasing children's human capital formation. When evaluating the effectiveness, we also consider the possibility of interaction between spending on training and spending on unemployment benefits as part of traditional SPPs. A time series cross-section (TSCS) analysis of the OECD countries is conducted for empirical estimation. In a process of model specification, we control for non-stationarity (i.e., highly persistent or strongly dependent time-series) and time-lag problem choosing the correct lag structure.