RESONATE
Worker flows, labor market outcomes, and agent-based modeling
Project financed by UEFISCDI, under programme Postdoctoral research
grant agreement PN-III-P1-1.1-PD-2019-1180
project hosted by Star-UBB Institute in the Business Informatics Research Center
project hosted by Star-UBB Institute in the Business Informatics Research Center
This project is designed around the study of labor market dynamics characterized by the full set of flows of workers between firms in large, realistic, labor markets such as entire domestic economies, taking into consideration both theoretical and empirical frameworks. We are interested in investigating labor market related topics, that can answer a set of questions of high-interest from scientific and socio-economic points of view. For this purpose we will focus on the detailed modeling of the flows of workers between firms, on constructing and characterizing the network of all those job flows, and on isolating the impact of the pattern of those flows on various worker and firm labor market outcomes. In this project, we consider an alternative for labor market investigation based on developments in agent-based modeling (ABM) and the advent of high-performance computing. The main innovation resides in the synergy between the state-of-the-art modeling tools used in microeconometrics with regard to large linked-employer-employee-data (LEED) and the Agent-Based Modeling and associated simulations techniques.
Research team
|
|
Scientific mentorProf.dr. Gheorghe Cosmin Silaghi |
Research objectives
The overall goal of this project is to respond questions concerned with the (evolution of) worker or firms labor market output with the help of agent-based modeling, by constructing and simulating worker flows, based on calibration by means of massive national or large regional LEED datasets. In particular, we consider the following specific project objectives:
To construct, for detailed study, a full-fledged, synthetic network of worker-firm labor market flows typical of national (or large-regional) economies, by means of massive, computationally-intensive, simulation of agent-based models – directly calibrated on the most relevant, very detailed, labor market empirical features from existing linked-employer-employee data.
To apply our fully-fledged network of labor flows and associated ABM framework to study one very researched theme in empirical labor studies nowadays, namely whether moving to firms which already has employed former co-worker of yours is increasing your salary and/or is offering a greater job protection, compared to moving to firms where you do not know anyone from previous workplaces, hence you could not have been refereed for those jobs.
To reveal the power (or the weaknesses) of agent-based modeling by contrasting the obtained results with those that would come out by applying standard econometric tools, for instance those used in structural labor econometrics.
Results
Articles under evaluation
The results of the research activity carried out within the project are embodied in two articles:
1. Cheregi, V.I., Is labor matching effectiveness dependent on education level, age or gender?
-submitted to the journal “Baltic Journal of Economics”, Taylor & Francis
2. Cheregi, V.I., Silaghi, G.C., Agent-based modeling for the study of labor market dynamics
-submitted to the journal „Simulation Modeling Practice and Theory”, Elsevier
Conference presentations
The papers were presented in three scientific conferences, as follows:
A. The International Conference: European Economics and Finance Society (EEFS), Twenty Second Edition, 14th-17th June 2024, organized by the National and Kapodistrian University of Athens, Greece.
-article presented: "Is labor matching effectiveness dependent on education level, age or gender?" (author: Valentina Ioana Cheregi)
B. The Conference: Cluj Economics and Business Conference (CEBC 2023), Third Edition, 7th-9th December 2023, organized by Babeș-Bolyai University, Faculty of Economics and Business Administration, Cluj-Napoca, Romania.
-article presented: “Is labor matching effectiveness dependent on education level, age or gender? A Beveridge curve approach” (author: Valentina Ioana Cheregi)
C. The International Conference: Globalization and Higher Education in Economics and Business Administration (GEBA 2023), Edition XV, October 19-21, 2023, organized by the Alexandru Ioan Cuza University in Iași, Faculty of Economics and Business Management.
-article presented: “Agent-based modeling for the study of labor market dynamics” (authors: Valentina Ioana Cheregi, Gheorghe Cosmin Silaghi)
Scientific Reports:
Summary of results:
The postdoctoral project entitled "Worker flows, labor market outcomes, and agent-based modeling" (acronym RESONATE) was designed around the study of labor market dynamics with special focus on the flows of workers between firms. We used an alternative to the conventional estimation methods used in economics for the investigation of labor market related topics. The approach we follow uses the agent-based methodology (ABM) to explore the detailed structure of the labor market, by specifically accounting for the network structure that characterizes the interactions between firms and workers in the labor market framework.
The main objective was to respond questions concerned with the labor market output with the help of agent-based modeling. Specifically, our main focus was the use and applicability of this methodological approach in studying labor market related topics.
The scientific approach was guided by several essential steps in exploring the synergy between the theoretical framework of the labor market and the applied framework of agent-based modeling.
We will briefly describe in the following the main stages completed throughout the project:
2020 (September-December) and 2021 (January)
The first activities carried out were meant to analyze the related literature, in order to identify the essential elements required to study the dynamics of the labor market using the agent-based methodology (ABM). To this end, we reviewed a number of over 80 scientific articles to explore the utility of such a methodology.
The literature review stage confirmed the importance and the necessity of agent-based modeling in studying labor market dynamics. The advantages it possesses as compared to the standard approach, allow the study of some phenomena at a disaggregated level, by relying on a bottom-up perspective, heterogeneity and realistic description of the reality.
2023 (January-December)
The preliminary research conducted strengthened our conviction regarding the importance of using an agent-based methodology for capturing with more accuracy the particularities of the labor market. However, we were not able to identify in the related literature a unitary and integrated vision regarding the implementation of such a model in practice. Thus, in order to fill the gap in the literature, we channeled our effort in two directions:
1. We considered imperative to examine how the use of agent-based methodology (ABM) and the study of labor market dynamics combine. To achieve this objective, we conducted a detailed analysis following a double perspective:
a. First we followed a methodological perspective specific to the field of computer science to determine the essential elements (ingredients) required for building an agent-based model.
b. Second, we tried to synthesize and analyze to what extent the existing studies in the literature managed to incorporate these elements.
This double-perspective, distinguishes our approach from existing works in that, starting from the advantages that the ABM methodology offers, it tries to serve as a guide for the implementation of ABM in labor market studies. The paper we wrote in this stage ("Agent-based modeling for the study of labor market dynamics") develops a step-by-step procedure and details the main ingredients to consider when implementing an agent-based model for studying labor market characteristics.
2. Simultaneously, in order to test the capability of the agent-based methodology in the study of the labor market, we conducted a macroeconomic study that explores the dynamics of the labor market in a regional perspective. The objective was to perform a comparative analysis based on the estimation methods traditionally used in economics.
We sought the econometric validation of the well-known negative relationship between unemployment and the vacancy rate, known as the Beveridge Curve (BC). The article we wrote as a result of this analysis ("Is labor matching effectiveness dependent on education level, age or gender?") aimed to explore the role of the workforce structure on the efficiency of the labor matching process. Data were collected for a number of aggregate variables for the Central and Eastern European countries between 2005 and 2023. Our choice is motivated by the fact that the existing literature, for this particular sample of countries, do not usually take into account the structure of the labor force (in terms of age, gender or education) in explaining the dynamics between unemployment and vacancies.
The results of the study confirm the validity of the negatively sloped Beveridge curve and point to a significant impact of education, gender and age on labor market outcomes. In particular, the association between employees and companies is facilitated by a greater share of the workforce that has higher education, while the opposite is true for the share of women and younger people in the workforce, both coefficients estimated in regression indicating a greater probability of deteriorating the matching process.
2024 (January-July)
In the final part of 2023 and in 2024 we focused also on results dissemination. We have participated and presented our papers in three scientific conferences. The comments and suggestions we received were subsequently incorporated in our work.