A rich life cycle model of labor supply in Finland
Antti J. Tanskanen · 2025
Read A rich life cycle model of labor supply in Finland by Antti J. Tanskanen (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-00660.txt.
Introduction
Antti Tanskanen builds a rich life cycle model of work and consumption in Finland. Such models follow a population of simulated individuals across their lifetimes, choosing jobs and spending under real rules. Finland's wide variety of employment states and life situations demands a large state space, so each agent's circumstances can be tracked faithfully through career, family, and benefit spells.
The model reproduces key statistics of the Finnish labor market. It matches employment and unemployment rates by age, the spread of effective marginal and participation tax rates people actually face, and the share of part-time work. As a policy test, Tanskanen runs the Orpo government's program through the model, tracing its effects on employment and public finances. The simulator turns reform debates into measurable forecasts.
Catalog record
| CREATOR | Antti J. Tanskanen |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 70 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2511-00660
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A rich life cycle model of labor supply in Finland Antti J. Tanskanen Abstract A life cycle model of consumption and labor supply describes employment decisions of a collection of individuals during their lifetime. We develop a life cycle model describing a heterogeneous population operating in Finland under a wide variety of employment states and life situations. A rich life cycle model requires a large state space representing the possible states of simulated agents. The results demonstrate that the model reproduces a number of statistics of the Finnish employment market such as the age structures of employment rate and unemployment rate, distributions of observed effective marginal tax rates and participating tax rates, and proportion of part time work. As an application of analysis of a reform, we analyze how the program of Orpo government influences employment and public finances in Finland. 1 Introduction A life cycle model of consumption and labor supply (abbreviated LCM in the following) describes how people make decisions on whether to work or not, and how much to work. If saving is included in an LCM, it can be used to describe how people smoothen consumption over their life-times (Modigliani and Brumberg 1954). Stochastic LCMs have been applied to assess the influence of a social security reform on employment and unemployment rates (Hakola and Määttänen 2007, Määttänen 2014, Tanskanen ja Kotamäki 2021), retirement (Jiménez-Martín and Sánchez Martín 2007, Blau 2008, Rust 1989), and on the internal rate of return on pension contributions (Tanskanen 2020b). An LCM is often used to study optimal savings behavior during an individual’s life cycle (Modigliani and Brumberg 1954) and consumption patterns (Gourinchas and Parker 2002). LCMs have demonstrated more successes than failures in these contexts (Browning and Crossley 2001). 1.1 Analyzing social security reform A social security reform often aims either to improve social security of certain groups of individuals or to revise the benefits to encourage employment. In both cases, estimating how the reform affects the financial sustainability of a social security scheme is crucial for the viability of the reform. Due to complexity of social security schemes, the influence of a social security reform on employment and financial sustainability is quite hard to analyze, especially when behavior of people is taken into account. If a reform is not financially sustainable, there is little reason to adopt it. A stochastic LCM of labor supply simulates how rational individuals make employment decisions, when the future is uncertain. An individual makes her employment decisions by maximizing the expected utility in an LCM. The cornerstone of this modeling approach is the utility function, which balances the value of free time against the benefits of consumption. Consumption, facilitated by wages, social security benefits, and savings, must be weighed against the opportunity cost of lost free time due to work. Striking this balance determines the labor supply. An LCM aims to provide insight into how individuals respond to policy reforms, such as changes in social security or taxation schemes. 1.2 Population, employment decisions and state transitions Work incentives vary across different household types. A single household with children, a couple with children, a couple without children, and a single household without children all exhibit different optimal behaviors. A comprehensive LCM should be able to find the optimal behavior for each of these groups at various stages of life, taking also into account that in the future, the single agent may be part of a couple. Transitions between employment states are influenced by a mix of factors. Some transitions, such as changing employment status, are deliberate decisions made by the individual. Others, like moving to disability pension, occur with exogenous probabilities. Certain transitions, like quitting a job, may result from either an individual’s decision or external circumstances. Agents decide their next course of action, such as remaining employed or transitioning to unemployment, taking all these possibilities into account. 1.3 Finding optimal behavior Dynamic programming is a well-known method for solving an LCM (see, e.g., Heer and Maussner 2009, Rust 1989, Määttänen 2014) giving a good approximation to the optimal behavior. The approximation converges toward the globally optimal solution in the limit of infinitely tight grid (Puterman 1990). However, applicability of dynamic programming is constrained by its large demand for resources and finite available computation time (Sutton and Barto 2018). Dynamic programming finds optimal behavior best in a relatively small state space with less than 10 discrete or discretized state variables. Modeling a modern social security scheme in, e.g., a Nordic country, requires a large number of state variables to describe the social security benefits and wages in detail. In an LCM describing Finnish social security (Tanskanen 2020a), the number of state variables was 13. To find an optimal behavior in this model, dynamic programming would need to solve the model on a grid with \penalty\ 10^{21} states, which is computationally unfeasible. Solving an LCM in a large state space requires the use of approximative algorithms (Rust 1997). Recent advances in machine learning algorithms enable solving a rich LCM with a large state space approximately. Reinforced learning (abbr. RL) algorithms are alternatives for dynamic programming that can find an approximately optimal solution even for an LCM with a huge state space. Although theoretical knowledge of convergence properties of RL algorithms is not fully developed, the performance of RL algorithms, such as ACKTR (Wu et al. 2017), seems to be good in practice (Tanskanen 2022). This is supported by theoretical results on the convergence (Wang et al. 2019). When the state space is huge, the algorithm must be able to generalize from the observed states to the unobserved ones. This makes deep reinforced learning methods based on neural networks well suited for this task. 1.4 Comparison with other methods The most common way to estimate the impact of a reform on employment is to use an appropriate elasticity describing the response of employment on the examined change (Cahuc et al. 2014). This is a solid, experiment-based way of estimation. For example, if a reform changes the maximum duration of unemployment benefit, the elasticity of the unemployment duration to the potential duration of unemployment benefit is the appropriate elasticity for the analysis of the response. Ministry of Finance’s (2024) estimates of the impact of reforms are mostly made using appropriate elasticities. Each employment supply elasticity describes a single pathway influencing a particular group of people. For example, elasticity of unemployment duration with respect to potential duration of unemployment benefit will only measure how the length of unemployment spells will respond to a change in the maximum duration of unemployment benefit. To estimate the response of unemployment to a change in the unemployment benefit level, we need a different elasticity. This raises important questions: How should various responses be aggregated? Are the pathways correlated? Are the same individuals affected via multiple pathways? Can the responses be summed without double-counting? These are not issues with an LCM. Microsimulation is used to estimate how demographic and policy changes influence individual outcomes, and it enables a better understanding of the effects of chosen policies. Microsimulation is often based on register data on, e.g., wages and social security benefits. A microsimulation model provides a static analysis of how a reform influences net income across