Springe zum Inhalt

Veranstaltungsreihe: IAB-Colloquium

Wissenschaftlerinnen und Wissenschaftler die Ergebnisse ihrer Forschungsarbeit vor und diskutieren diese mit Expertinnen und Experten aus dem IAB. Auch Interessierte aus Arbeitsverwaltung, Politik und Praxis sind willkommen.

We examine the properties of a recommender system we developed at the Public Employment Service (PES) in France, prior to its implementation in the field. The algorithm uses past matches and a very large set of covariates to produce, for each job seeker, a ranking of the available offers and score each pair jobseeker-offer. Using a calibration step that takes advantage of the observation of application sequences, it gives a predicted "matching probability" for each pair.  After a theoretical discussion about the possible strategies to design a recommender system, we compare this new machine learning (ML) algorithm with another matching tool, mimicking the one currently used at the PES, based on a score measuring the "closeness" between the jobseeker's search criteria or preferences and the characteristics of the offer. We quantify the trade-off between the matching probability and the later "preference score" when switching from one system to the other. Next, we examine the issue of congestion.  We show that, on the one hand the ML algorithm based on past matches tends to increase congestion and on the other hand that this strongly reduces its performance. Finally, we show that the use of optimal transport to derive recommendations from the matching probability matrix significantly alleviates this problem. The main lesson at this stage is that an algorithm ignoring preferences and competition in the labor market would have very limited performances but that tweaking the algorithm to fit these dimensions substantially improves its properties, at least "in the lab".

Social science research demonstrates that dispersal policies and restrictions on the freedom of residence have inhibited refugees’ socio-economic integration, presumably because such policies prevent refugees from moving to places where they can employ their skills most fruitfully. However, studies of refugees’ actual residential choices provide little evidence that good economic prospects attract refugees, and some even suggest that refugees often move to deprived cities with frail labor markets. The combination of negative effects of residence restrictions and emerging evidence of disadvantaging secondary migration forms what we call the ‘refugee mobility puzzle’. In this study, we aim at unpacking this puzzle by analyzing the inner-German migration patterns of recent refugees. Specifically, we ask: What attracts refugees to deprived areas, and can their seemingly unfortunate residential choices be understood as moves to opportunity and increased prospects of labor market integration after all? Empirically, we draw on the IAB-BAMF-SOEP Survey of Refugees and track the location of more than 2,000 refugee respondents who were exogenously allocated a place of residence and subsequently became free to move. Based on linear-probability discrete choice models across all German counties and postcodes, we confirm that refugees tend to move to areas with high unemployment. We show that major attractors like housing availability, co-ethnic networks, and service-oriented labor markets are clustered in areas with high unemployment. Taken together, our results complicate recent critiques of dispersal policies and restrictions. On the one hand, our findings show that seemingly disadvantaging relocations into high unemployment areas can conceal potentially improved economic perspectives in relevant labor markets. On the other hand, refugees’ search for affordable housing may turn into an unintended lock-in factor in the mid- and long-run.

This paper studies the interplay between how much workers value workplace flexibility, whether they have such amenities, and how the presence of amenities affects their wages. To overcome the challenge of eliciting quantitative measures of willingness to pay (WTP) at the individual level, we propose the use of dynamic choice experiments, a method which we call the Bayesian Adaptive Choice Experiment (BACE). We implement this method to collect data on the joint distribution of wages, work arrangements, and WTP for different forms of flexibility. We then introduce and estimate a model in which workers may face different prices for job amenities depending on their productivity, extending the Rosen (1986) model of compensating differentials. The model captures key patterns in the data, including (i) the relationship between wages and having amenities, (ii) inequality in workplace amenities across the earnings distribution even when workers value these amenities similarly, and (iii) the tradeoffs across different forms of flexibility. We use the estimates to explore the welfare consequences of workers facing different amenity prices.

Social disparities in track choices are a well-known mechanism for the intergenerational reproduction of inequality. School guidance may help reducing such disparities by narrowing information gaps and by reducing the family influence on students’ decision making. We investigate the potential equalizing role of guidance programs by analysing an intervention carried out in Italy, where students are tracked at age 14 and teacher recommendations are non-binding. The intervention took place in 2018 in the city of Turin and involved 40% of all eighth-grade students, shortly before their transition from comprehensive to tracked education. The students attended four two-hour sessions designed to provide them with information about the educational system and related job market opportunities, and to raise their awareness of their aptitudes and inclinations. We expected the programme to be of particular benefit to low socio-economic status (SES) and migrant students and thus to reduce social gaps in track choices. We adopted a mixed-method research design: with quantitative analyses based on a combination of propensity-score-matching and differences-in-differences techniques, we compared the outcomes of comparable students from the 2017 and 2018 cohorts who were or were not exposed to the intervention in order to assess its impact on inequality; additionally, we use qualitative non-participatory observation to unveil the actual content and implementation of the program and the behaviour of the key actors. We find that while the program contributed to reducing indecision, probably by compelling students to reflect more carefully about their decisions during this crucial transition, it did not have any major effect on social inequalities. Results from the qualitative analysis help us shed light on the mechanisms at play behind this lack of effect. In particular, the heavy emphasis placed on current achievement records, dropout risks, and (short-term) labour-market outcomes may counteract the equalizing potential of the program by pushing low-SES and migrant students towards vocational tracks.

We analyze the effects of large-scale local public infrastructure investments on economic development, exploiting the infrastructure shock following when Brazil was awarded the 2014 FIFA World Cup. We place particular emphasis on effect heterogeneity with respect to the type, location, temporal evolution, and costs and benefits of the investments. Using novel data on monthly night light luminosity at the municipal district level as a proxy for economic activity, we apply Difference-in-Differences and event studies for estimation. Overall, we find strongly positive impacts both in the short and longer run. However, a closer examination reveals that effects are larger and longer-lasting for transport infrastructure as opposed to sports infrastructure, and they are more pronounced in smaller areas. Importantly, we quantify significant negative spatial spillovers. Factoring them in, we still find positive net benefits of transport infrastructure investments two years after the tournament.

While many countries are discussing substantial increases in the minimum wage, policy makers lack a comprehensive analysis of the macroeconomic and distributional consequences of raising the minimum wage. This paper investigates how employment, output and worker welfare respond to increases in the minimum wage beyond observable levels -- both in the short- and long run. To that end, I incorporate endogenous job search effort, differences in employment levels, and a progressive tax-transfer system into a search-matching model with worker and firm heterogeneity. I estimate my model using German administrative and survey data. The model replicates the muted employment response, as well as the reallocation effects in terms of productivity and employment levels documented by reduced form research on the German introduction of a federal minimum wage in 2015.  Simulating the model, I find that long-run employment increases slightly until the minimum wage is equal to 60% of the full-time median wage (Kaitz index) as higher search effort offsets lower vacancy posting. In addition, raising the minimum wage reallocates workers towards full-time jobs and high-productivity firms. Total hours worked and output peak at Kaitz indices of 73% and 79%. However, policy makers face an important inter-temporal trade-off as large minimum wage hikes lead to substantial job destruction, unemployment and recessions in the short-run. Finally, I show that raising the minimum wage largely benefits men. For women, who often rely on low-hours jobs, the disutility from working longer hours outweighs the utility of higher incomes.

Das Arbeitsmarktservice (AMS) hat ein Arbeitsmarktchancen Assistenz-System (AMAS) entwickeln lassen, um den Berater und Beraterinnen eine bundesweit einheitlich erstellte und datenbasierte Information zu den Arbeitsmarktchancen ihrer Kunden und Kundinnen zur Verfügung stellen zu können. Diese zusätzliche Informationsquelle sollte Berater ud Beraterinnen dabei unterstützen, für die Kunden und Kundinnen die passende Betreuung und die effizienteste Fördermaßnahme zu finden. Insbesondere sollte damit die early Intervention verbessert werden.   
Im Vortrag wird über die Herausforderungen, die zur Entwicklung des Arbeitsmarktchancen-Assistenzsystem geführt haben, informiert. Nach einer kurzen Beschreibung des Systems wird auf die folgenden Punkte eingegangen:

  • Daten:  Quelle der Daten (direkt aus dem operativen System, aufbereitet in einem DWH oder andere) und welche werden genutzt; zeitlicher Umfang historischer Daten gegenüber dem Prognosezeitraum, Umfang der Abbildung des Arbeitsmarktes,
  • Methoden: Erläuterung der Logik der Zellenbildung (theoretisches Modell), Auswahlprozess der kategorisierenden Merkmale 
  • Einbettung des Assistenzsystems in die operative Software 
  • Ethische Fragestellungen und wie wurden diese beantwortet, insbesondere die Rolle der AMS-Vermittler und Vermittlerinnen

Status quo:
Das Arbeitsmarktchancen Assistenz-System (AMAS) wurde von Synthesis Forschung 2016 entwickelt und 2019 in einem Probebetrieb (Workshops sowie Schulungen zu AMAS) getestet. Die Datenschutzbehörde hat dem AMS den Einsatz von AMAS mit Bescheid vom 19. August 2020 untersagt (amtswegige Prüfung) sowie eine aufschiebende Wirkung ausgeschlossen, sodass alle Daten und Anwendungen gelöscht werden mussten. Das Bundesverwaltungsgericht hat diesen Bescheid vollumfänglich aufgehoben. Die oberste Instanz, der Verwaltungsgerichtshof hat bisher noch keine Entscheidung getroffen.

Rund die Hälfte der Paare mit Kindern unter 15 Jahren in Österreich lebt nach dem Frau Teilzeit/Mann Vollzeit Modell. In diesem Vortrag gehen wir der Frage nach, wie sich Erwerbsmodelle von Paaren mit Kindern in der COVID-19 Pandemie veränderten und wie sich dies auf die Aufteilung von Kinderbetreuungs- und Hausarbeitszeit auswirkte. Wir bilden Trends über den Verlauf der Pandemie hinweg ab, fokussieren jedoch auf den ersten Lockdown im Frühjahr 2020, wo sich die stärksten Veränderungen zeigten. Als Datenbasis verwenden wir den Mikrozensus (Labour Force Survey) sowie die Daten des Austrian Corona Panel Projects. Zentrale Ergebnisse sind ein (temporärer) Rückgang des Frau Teilzeit/Mann Vollzeit Modells zugunsten einerseits egalitärerer Modelle, andererseits des männlichen Ernährermodells (Frau nicht erwerbstätig/Mann Vollzeit). Dies führte in einem Teil der Familien zu einer veränderten Arbeitsteilung: Väter in Teilzeit und im Homeoffice weiteten ihre Kinderbetreuungs- und Hausarbeitszeit aus. Diese Befunde werden vor dem Hintergrund von Geschlechterrollentheorien und ressourcenbasierten Ansätzen diskutiert.

The presentation offers an overview about the new data service of the Research Data Centre of the Federal Office for Migration and Refugees (BAMF-FDZ). The BAMF-FDZ gives access to register and survey data for migration and integration research. The presentation will introduce the available and future data sets and discuss advantages and limitations. In addition, the application procedure is also explained.

Aufgrund des Ausbruchs der Corona-Krise ist die Gründungstätigkeit in Deutschland zurückgegangen. Gründerinnen und Gründer machten sich aber häufiger selbstständig, um eine sich bietende Geschäftsgelegenheit wahrzunehmen. Ihr Anteil stieg 2020 auf 80 % an, womit die Anzahl an Chancengründungen trotz der rückläufigen Gründungstätigkeit relativ stabil blieb. Die Corona-Krise hat branchenbedingt insbesondere selbstständige Frauen stark belastet. Die Zahl der Gründerinnen blieb 2020 aber nur leicht unter dem Vorjahresniveau. Gründungsinteressierte Frauen scheinen sich schneller auf die neuen Krisenbedingungen eingestellt und letztlich ihre Gründungspläne häufiger doch realisiert zu haben als Männer. So haben Gründerinnen häufiger als Gründer Geschäftsmodellanpassungen vorgenommen. Die meisten (typischen) Gründungshemmnisse wurden 2020 seltener wahrgenommen als üblich. Dies dürfte mit der Corona-Krise zusammenhängen, deren Herausforderungen alles überlagerte und viele Gründungsinteressierte von vornherein abschreckte. Es dürften also eher Menschen gegründet haben, die bereits konkrete Vorstellungen hatten und entsprechend seltener Hemmnisse wahrnahmen. Dennoch gehen Gründerinnen und Gründer mit dem Gründungsstandort Deutschland härter ins Gericht als in den vergangenen Jahren. Coronabedingt verschobene Gründungen dürften 2021 die Gründungstätigkeit stützen.