Developing analytical tools to nowcast and forecast the macroeconomic, distributional, and epidemiological effects of the crisis and related public health vs. economic policy responses in Luxembourg
Project MODVid: Modeling the macroeconomic and distributional effects of Covid-19 and restarting scenarios
How does the project address the COVID-19 crisis?
MODVid involves 6 PI’s and about 20 partners from Unilu, STATEC and LISER. It aims to inform public decisions during the crisis by providing estimates of the macroeconomic, distributional, and epidemiological effects of the crisis and of restarting scenarios in Luxembourg.
What are the main points addressed?
Building upon the previous RECOVid project, the new research project is comprised of four complementary and interdependent work packages (WP). They are:
- WP1 – Short-run health and macroeconomic effects
- WP2 – Short- and medium-term distributional effects
- WP3 – Occupational sorting after the crisis
- WP4 – YMAA in the post-Covid wealth crisis
Research Outputs ( to-date)
Policy Brief: How bad will the Covid-19 second wave be for Luxembourg’s economy?
MODVid is coordinated by Frédéric Docquier (LISER) and includes about 20 researchers from these Luxembourg institutions:
- Luxembourg Institute of Socio-Economic Research (LISER)
- University of Luxembourg
- Statistics and Economic Studies of the Grand Duchy of Luxembourg (STATEC)
Aim of the research
MODVid is a set of four complementary and interdependent work packages. Two of them aim to inform ”burning” public decisions to be made during the crisis. These WP’s address priorities identified in the WP0, WP6 and WP7 of the Covid-19 Task Force set up by Research Luxembourg. Our philosophy here is to develop a set of evolving and interconnected modeling tools that aim to deliver fast results first, and then increasingly precise results and predictions as the set of available data on socioeconomic variables and leading indicators increases. In addition, Covid-19 will induce uncertain medium-term effects on firms and households. Future policy actions will be needed to limit the persistent effects on welfare and wealth inequality. Hence, two other WP’s aim to help decision makers avoid a rise in inequality and large welfare losses in the aftermath of the crisis.
Work Package details:
1. Short-run health and macroeconomic effects (WP-I)
WP-I analyzes jointly the public health and economic effects of the Covid-19 crisis week after week throughout the year 2020. We develop a model that can be used to produce nowcasts of the costs of the crisis, and simulation results that can inform public decisions about the relevant modalities and timing/extent of a potential restarting strategy. We treat economic and epidemiological trends as interdependent: Covid cases affect the number of workers available to firms, and employment shocks generate on-the-job interactions which affect the infection curve. WP-I sheds light on the scale of these interactions, on the effectiveness of lockdown and deconfinement plans, and on the role played by cross-border workers.
The economic block of the model consists of an extended input-output (I/O) model, which accounts for both demand-side and supply-side mechanisms. We enrich the standard I/O model with binding supply-side constraints of various sorts. Firstly, the disruption of the global value chain can prevent firms to produce the quantity demanded on the market. Secondly, some workers are infected by Covid-19 and cannot supply labor. In addition, school closures imply that many workers are forced to take parental leave. And last but not least, containment measures reduce the permitted levels of employment and activity in lockdown industries. The epidemiological block decomposes the sector specific labor force available at each period into sixteen groups involving 4 regions of residence (within the Greater Region) and four infection status (susceptibles, symptomatic infected, asymptomatic infected, recovered). The dynamics of these stocks is governed by contamination rates on the job and in the place of residence. The model is calibrated using pre-crisis data and a wide range of post-lockdown leading indicators from STATEC, ADEM, IGSS, LCSB, LIH, Chamber of Commerce, etc.
WP Leaders: Frédéric Docquier (LISER-CB) and Tom Haas (STATEC).
Partners: Chiara Peroni (STATEC), Joël Machado (LISER-LM), Michal Burzynski (LISER-LM), Michel Beine (Unilu-Econ), Pierre Picard (UniluEcon); Françoise Kemp (Unilu-LCSB), Stefano Magni (Unilu-LCSB).
2. Short- and medium-term distributional effects (WP-II)
In the first part of the WP-II, macroeconomic scenarios from WP-I will be included into a microsimulation model developed and calibrated in LISER. This model produces an estimate of the whole distribution of disposable incomes in the population, along with estimation of the total tax revenues and benefit spending. We use it to forecast the distributive impacts of Covid-19. For example, one can simulate what would happen if a fraction p of workers from sector s will lose their jobs and/or that a fraction q of workers from sector s experience a wage cut of x percent. We assess the impact on inequality or poverty indicators, as well as the budgetary impact of reduced taxes and increased benefits. The simulation can be augmented by discretionary policy responses: governments can modify tax-benefit policy parameters in response to the crisis, and these can be combined with the earnings shock to assess distributive and budgetary impacts. In reverse, assuming government have a target poverty rate in mind, the model could help determine how fiscal parameters need to be modified to avoid exceeding that target. The accuracy of the simulation crucially depends on the accuracy of the earning and employment shocks, which are simulated into the model. In the first stage, we use proxies from WP-I for labor market effects of the lockdown on the distribution of earnings in different sectors and by occupation.
In the second part of WP-II, we will refine the microsimulation analysis and develop a mechanism to deliver real-time snapshots of the income distribution. The key methodological challenge is the lack of up-to-date data. We overcome this gap by developing a “nowcasting” methodology based on microsimulation a la O’Donoghue et al. 0This consists of using the most recent data on employment, wage/price levels to calibrate a simulation model of household incomes, taxes and benefits. Existing methods are relatively crude, applying price inflation factors, changing proportionally the employment rate in specific industries, and using a fiscal block to explain the policy consequences. In Part 2, we develop a more sophisticated approach that allows us to better explain the heterogeneity of changes in the population by utilizing a dynamic income generation model (IGM) to nowcast income data using leading indicators on personal and household attributes. The IGM relies on a system of hierarchically structured, multiple equation models for detailed income sources, whereas taxes and benefits are calculated using the EUROMOD microsimulation model.
We will extend the methodology recently developed in LISER to simulate counterfactual real-time income distributions and to identify the impact of Covid-19. We plan to do this on a monthly basis over the course of the crisis to support planning in government and monitor the cumulative impact of the wide variety of changes on vulnerable groups. This requires different data sources: e.g., (i) calibration data to index income growth (monthly national statistics – Statec); (ii) calibration data to align the labour market (monthly registry on the labour force – Statec); (iii) Microdata on which to estimate and simulate (EU-SILC).
Leader: Philippe Van Kerm (Unilu/LISER) for Part I and Denisa Sologon (LISER-LIC) for Part II.
Partners: In Part I: Nizamul Islam, Vincent Vergnat, Nathalie Lorentz, Kristell Leduc (LISER LIC). In Part II: I. Kyzyma, J. Linden (LISER-LIC); C. O’Donoghue, J. McHale (NUI); Philippe Van Kerm (UL-LISER LIC).
3. Occupational sorting after the crisis (WP-III)
Turning our attention to labor market issues, important questions are: What share of workers fromvulnerable occupations is capable of switching job types? How does this occupational mobility depend on sectoral-specific job survival rate? What are the expected consequences for within and across sector wage inequality? How many workers will be forced to join unemployment? The intensity of lockdown, and its economic consequences for firms and workers are strongly dependent on the character of work within occupations and sectors. Employees in different niches of labor markets are heterogeneously affected by the imposed restrictions. Teleworking and remote work organization allow to maintain business-as-usual performance in specific occupations. On the contrary, many occupations that involve personal interactions and face-to-face contacts are totally frozen
In this work package we investigate the medium-term impact of Covid-19 lockdown on workers’ occupational allocation, wage inequality and unemployment in Luxembourg and across EU member states. Our goal is to quantify the labor market implications of partial annihilation of selected sectors of economic activity. By simulating scenarios of different degree of severity, we compute changes in workers’ assignment to jobs, unemployment measures, and final production across sectors of regional and national economies. Our study is targeted at identifying those segments of labor markets that are most vulnerable to lockdown, to highlight the consequences of insufficient efforts to preserve jobs inducing firms’ mass bankruptcies, layoffs, and unemployment. The methodology implemented in this project includes building and calibrating a theoretical model of labor markets with multiple occupations and sectors of economic activity. Workers, equipped with multidimensional, heterogeneous skills, decide about their optimal sorting across occupations. Firms operating within sectors set the structure of occupational employment to fulfil tasks. The equilibrium allocation of workers to tasks determines wage distributions, within- and between occupation inequality, and the endogenous rate of unemployment. Within this approach, we quantify the consequences of lockdown for labor market assignment, as well as simulate various
scenarios of post-lockdown economic environment for wage distributions, inequality, occupational mobility of workers, and sectoral production. Our methodology allows us to design tailor-made policies targeted at supporting the most vulnerable workers.
Leader: Michal Burzynski (LISER-LM).
4. YMAA in the post-Covid wealth crisis (WP-IV)
This project helps assess the risks of a contagion of the Covid-19 health crisis to households’ welfare: one can expect the spread of a major socioeconomic crisis through (at least) four processes: a. mass unemployment and loss of income, b. a housing market crisis, c. family disruptions, and d. health issues (physical and mental disorders). This study focuses on a specific potentially vulnerable population, young-mid-aged adults (YMAA) – typically in the first two decades of their involvement in the labour market. High debt to income ratios and dependence of households’ welfare on upward job mobility, family stability, and health condition define the specific vulnerability of this demographic group. We focus mostly on process b., based on income, as well as wealth and debt balance, because of its specific demographic importance. With the exception of the 2008 financial crisis, the last generation of YMAA experienced a context of relative stability, easier access to credit, and rapid growth of the housing index (with different magnitudes across countries), favouring home ownership. This context favoured early decisions to buy, since the later one’s home aquisition, the smaller the surface one’s household could expect to obtain1. In the disruptive context of the post-COVID19 era, indebted YMAAs face the risks of asset bubble explosion and its consequences (cf Case-Shiller Index). This increases their socioeconomic vulnerability. Is this middle generation to be sacrificed? Particular within the middle class?
In work package (i), we define and describe the population at risk of wealth losses based on their debt versus assets position, and reassess the concept of vulnerability through over-indebtedness risks. The work package (ii) analyzes the regimes of accumulation of birth cohorts spanning from the silent generation born in the late 1930 to the Millennials (today’s YMAAs). This preCOVID-19 regime will be longitudinally modelled based on income and wealth data (generally through panel studies) within different social strata (based on education, parents’ background, etc.) in a set of contrasted developed countries, using panels in Germany (Gsoep), the U.S. (Psid), and Korea (Klips) spanning 1990-2018, as well as in Luxembourg (through the EUHFCS (2011-2014-2017)). Novel age-period-cohort modelling techniques (Bar Haim et al 2019) developed in Luxembourg (APCD, APCTLAG, APCGO) with UPenn colleagues will provide the rules of the ‘routine scenario’ of wealth accumulation before the disruption triggered by COVID-19. Work package (iii) contrasts the earlier regime (the counterfactual ‘routine scenario’ of wp (ii)) with simulated ‘disruptive scenarios’ modelling stresses of different intensities on four dimensions: negative shocks on wealth, health, family disruptions, mass unemployment, etc. This comparison of scenarios will help measure impact, understand and forecast consequences, and suggest best social policies for a resilient society.
Leader: Louis Chauvel (Unilu).
Partners: Jason Settels (Unilu), Francisco Ceron (Unilu), Herbert Smith and Hyunjoon Park (UPenn).
1- See Chauvel, L. and A. Hartung (2016). Malaise in the Western Middle Classes. UNESCO, World Social Science Report Challenging Inequalities. Chauvel L., A. Hartung, E. Bar-Haim, P. Van Kerm Philippe (2019). Income and Wealth Above the Median”. Research on Economic Inequality, 2019, vol. 27, pp. 89-104.
RESEARCH LUXEMBOURG TASK FORCE
Didier Goossens: email@example.com
LUXEMBOURG INSTITUTE OF SOCIO-ECONOMIC RESEARCH
Carole Wiscour-Conter: firstname.lastname@example.org