Gerhard Krug

(When) Is job-finding via personal contacts a meaningful concept for social network analysis? A comment to

Abstract Chua (2011) argues that in a meritocratic context, institutions restrict the usefulness of social networks in exerting influence on job seekers’ earnings. Regressing job-finding via personal contacts on earnings, he finds negative effects of influence via personal contacts, especially for the well-educated and individuals working in the state sector. In this comment, I argue that these results are ambiguous because (1) the analysis does not sufficiently distinguish between job ‘search’ methods and job ‘finding’ methods, (2) job-finding method indicates information flow rather than a personal contacts influence, and (3) it remains unclear whether Chuas analysis reflects the effect of network usage in job search per se or the effect of self-selection into network usage by individuals with low earning potential.

Do Lower Caseloads Improve the Performance of Public Employment Services? New Evidence from German Employment Offices

The caseworker-to-clients ratio is an important, but understudied, policy parameter that affects both the quality and cost of public employment services that help job seekers find employment. We exploit a large-scale pilot by Germanys employment agency that hired 490 additional caseworkers in 14 of its 779 offices. We find that lowering caseloads caused a decrease in the rate and duration of local unemployment as well as a higher re-employment rate. Disentangling the mechanisms that contributed to this improvement, we find that offices with lowered caseloads increased monitoring and imposed more sanctions but also intensified search efforts and registered additional vacancies.

Private and Public Placement Services for Hard-to-Place Unemployed

The authors analyze a randomized field experiment in two German labor market agencies that provide public and private provision of intensive job placement services. The findings, based on analysis of administrative agency data over 18 months in 2009–2010, show that assignment to public employment services reduced accumulated days in unemployment by one to two months, compared to an assignment to a private provider. The effects, however, were short-lived. Moreover, two-thirds of the effect is attributable to labor force withdrawals. Finally, several important differences in the modes of service provision are only partially attributable to inherent aspects of in-house production and contracting out.

Augmenting propensity score equations to avoid misspecification bias – Evidence from a Monte Carlo simulation

Propensity score matching is axa0semi-parametric method of balancing covariates that estimates the causal effect of axa0treatment, intervention, or action on axa0specific outcome. Propensity scores are typically estimated using parametric models for binary outcomes, such as logistic regression. Therefore, model specification may still play an important role, even if the causal effect is estimated nonparametrically in the matched sample. Methodological research indicates that incorrect specification of the propensity score equation can lead to biased estimates. Augmenting the propensity score equation with terms that represent potential nonlinearity and nonadditivity, as proposed by Dehejia and Wahba and more recently by Imbens and Rubin, represents axa0means of avoiding such bias. Here, we conduct axa0Monte Carlo simulation and show that the misspecification bias is rather small in many situations. However, when the propensity score equation and/or the outcome equation are characterized by strong nonlinearity and nonadditivity, the misspecification bias can be severe. Augmentation is shown to reduce this bias, often substantially. The Dehejia-Wahba (2002) algorithm performs better than the Imbens-Rubin algorithm, especially when the outcome equation is strongly nonlinear and nonadditive.ZusammenfassungPropensity Score Matching ist eine semi-parametrische Methode zur Drittvariablenkontrolle bei der Schätzung kausaler Effekt eines Treatments, einer Intervention oder einer Handlung auf eine bestimmte Zielvariable. Propensity-Scores werden typischerweise unter Verwendung parametrischer Modelle für binäre Ergebnisse geschätzt, etwa der logistischen Regression. Daher stellt sich trotzdem die Frage der korrekten Modellspezifikation, selbst wenn der kausale Effekt in der gematchten Stichprobe nichtparametrisch geschätzt wird. Studien zeigen, dass eine falsche Spezifikation der Propensity-Score-Gleichung zu verzerrten Schätzungen führen kann. Um solche Verzerrungen zu vermeiden, haben Dehejia und Wahba und kürzlich Imbens und Rubin Algorithmen zur Anreicherung der Propensity-Score-Gleichung mit Termen vorgeschlagen, welche eine potenzielle Nichtlinearität und Nichtadditivität in der Modellspezifikation abbilden sollen. In der vorliegenden Arbeit wird eine Monte-Carlo-Simulation durchgeführt und es zeigt sich, dass die Verzerrung aufgrund von Fehlspezifikation in vielen Situationen eher klein ist. Wenn jedoch die Propensity-Score-Gleichung und/oder die Outcome-Gleichung durch starke Nichtlinearität und Nichtadditivität gekennzeichnet sind, kann die Fehlspezifizierungs-Vorspannung schwerwiegend sein. Anreicherungsalgorithmen reduzieren solche Verzerrungen oft erheblich. Der Dehejia-Wahba Algorithmus scheint hierzu besser geeignet als der Algorithmus von Imbens-Rubin (2015), insbesondere dann, wenn auch die Ergebnisgleichung stark nichtlinear und nichtadditiv ist.

Do Lower Caseloads Improve the Performance of Public Employment Services? New Evidence from German Employment Offices: Do lower caseloads improve public employment services?

The caseworker‐to‐clients ratio is an important, but understudied, policy parameter that affects both the quality and cost of public employment services that help job seekers find employment. We exploit a large‐scale pilot by Germanys employment agency, which hired 490 additional caseworkers in 14 of its 779 offices. We find that lowering caseloads caused a decrease in the rate and duration of local unemployment as well as a higher re‐employment rate. Disentangling the mechanisms that contributed to this improvement, we find that offices with lowered caseloads increased monitoring and imposed more sanctions but also intensified search efforts and registered additional vacancies.