Modelling Incomplete Rationality in Retirement Decisions

Info: 

O czym jest projekt?

Większość modeli makroekonomicznych reform systemów emerytalnych zakłada, że ludzie są dalekowzroczni i postępują racjonalnie. W tych modelach, prawidłowa odpowiedź na długowieczność oraz niektóre reformy systemu emerytalnego polega na zwiększeniu prywatnych oszczędności dobrowolnych oraz podaży pracy. Tymczasem dane ze świata rzeczywistego sugerują, że taka reakcja, jeśli w ogóle występuje, jest stosunkowo słaba. Prawdopodobnie jest to przejawem tego, że oszczędności w cyklu życia oraz podaż pracy mogą być pod silnym wpływem czynników krótkofalowych. Zaprojektujemy modele z niepełną racjonalnością, które będą w stanie przynajmniej częściowo replikować tę własność rzeczywistości. W takim modelu będzie można przeanalizować ex ante, czy i jakie instrumenty zachęcające do oszczędzania na emeryturę mogą być skuteczne i podnosić dobrobyt.

Dlaczego projekt jest interesujący?

Od dekad reformy systemów emerytalnych w wielu krajach europejskich, zwłaszcza w Polsce, wprowadza się bez oceny ex ante ich prawdopodobnych skutków – tak pod względem efektów makroekonomicznych jak i dobrobytowych. Nasze wyniki będą w stanie wesprzeć decyzje dotyczące kształtu tzw. III filara emerytalnego, zanim zostaną podjęte jakiekolwiek decyzje.

 

Budget: 

Projekt realizowany wspólnie z Instytutem Badań Strukturalnych w ramach polsko-niemieckiej współpracy badawczej zaczynamy badać systemy emerytalne przy założeniu ograniczonej racjonalności podejmowania decyzji. Naszymi partnerami w tym projekcie są Patrick Puhani oraz Konrad Schaeffer z Leibniz Universität Hannover.

Źródło finansowania | Financing : Narodowe Centrum Nauki, BEETHOVEN 1

Projekt realizowany | Timeline : 02/2016 – 02/2019

Budżet łączny | Total budget: 446 813 zł

  • wynagrodzenia dla podstawowych wykonawców | compensation to researchers: 86 400zł
  • wynagrodzenia dla asystentów badawczych | compensation to research assistants: 140 400 zł
  • komputery i oprogramowanie (m.in. licencje Fortran) | hardware and software (incl. Fortran): 35 000 zł
  • konferencje i inne wyjazdy | conference travels: 102 815 zł
  • materiały + promocja | consumables + visibility: 8 213 zł
  • koszty pośrednie  | overheads: 68 635 zł
Purpose: 

W pierwszej kolejności przeprowadzimy analizę empiryczną zależności pomiędzy cechami psychologicznymi i decyzjami o inwestycjach finansowych. Analiza zostanie przeprowadzona na danych jednostkowych, pochodzących z Niemiec. Dotąd literatura skupiała się na wiedzy finansowej, ale pomijała czynniki behawioralne. Po drugie, przetestujemy empirycznie założenie, które leży u podstaw większości modeli makroekonomicznych: wybór pomiędzy konsumpcją i oszczędnościami wynika z takich samych zależności w cyklu życia. Bazując na dotychczasowych dokonaniach literatury przetestujemy założenie o przechodniości preferencji z wnioskami dla decyzji o oszczędzaniu w cyklu życia. Po trzecie, wykorzystamy wnioski z obu analiz empirycznych, by skonstruować model OLG z prywatnymi oszczędnościami dobrowolnymi, wiedzą finansową oraz ograniczoną racjonalnością. W tym modelu przeprowadzimy eksperymenty polegające na wprowadzeniu instrumentów wspierających oszczędzanie na emeryturę. Przeanalizujemy efekty makroekonomiczne oraz dobrobytowe takich instrumentów.

The objectives of this project are twofold. First, we want to carefully evaluate empirically the role of incomplete rationality in decisions about pension savings and retirement timing. Second, we want to employ this experience in modelling private voluntary pension savings and test whether (and if so – which) policy instruments can effectively be used to foster them. We thus posit two working hypotheses. The first one is implicit: individuals are not fully rational about their choices concerning pension savings and retirement saving. This hypothesis will be addressed empirically. The second one is explicit: there exists a welfare enhancing policy reform that can improve the alignment of individual decisions with a social optimum. The second hypothesis will be addressed by the means of an overlapping generations model. The scope of imperfect rationality will be calibrated to the reference values provided in the empirical part of the project.

  • Aging and inequality

    This data summarizes the evolution of consumption and wealth inequality over the forthcoming decades of longevity. In a defined contribution system, with extending life span on retirement, pension benefits are bound to decline (at least, if the retirement age is not raised). These declining pension benefits will encourage agents to increase voluntary savings in other to smooth consumption over lifetime. This is likely to affect wealth and consumption inequality, despite unchanged institutional arrangement and stable productivity heterogeneity within cohorts.

Systemy emerytalne

1st meeting

At the working seminar of the Beethoven project (4-7 November) Jan presented our overlapping generations model. It will be used to check welfare effects of policies intended as incentives for increased retirement saving. Present were our co-authors from Leibniz University of Hanover: Patrick Puhani and Konrad Schäfer.

2nd meeting

The meeting was held in Hannover. We discussed the potential directions for contributing to the empirical and theoretical literature in our two joint papers.

The theoretical paper rests upon developing a model which adequately captures the response of private voluntary savings to the incentives embedded in the pension system. Current models have a very strong reaction because fully rational agents internalize completely the consequences of changes in the pension system. Their reaction is strong on both the intertemporal choice (savings) and intratemporal choice (labor supply). Meanwhile, empirical evidence is rather weak: people do respond to incentives, but to a much lesser extent. For example, Myck and Lachowska in a recent study using the evidence from Polish pension system reform show that the actual crowding out effect is there, but is rather small. We seek features of the consumer choice that will make model choices closer to what is observed in the data.

The empirical paper intends to address the question of early labor market quits in most advanced countries. Indeed, some earlier research shows that the disincentives embedded in the pension system may contribute to early labor market exits, even in the US. In a widely commented study, French shows that removing some of the kinks in the institutional design of the pension system are likely to extend the careers by as much as two years. However, studies such as this one leave no room for household optimization. We intend to develop the literature further in this direction.

3rd  meeting

The meeting was held in Warsaw and involved the whole Warsaw team and Patrick Puhani. We focused particularly on the theoretical paper and on the features we want the theoretical model to replicate. The model now features a variety of incompletely rational agents, but we still want to exploit further types of behavioral limitations. In the setup with incomplete rationality, we implement instruments fostering private voluntary savings. Given the divers nature of the government instruments, it is imperative to make them comparable in terms of the size of the government intervention. We currently work on the methods that allow us to make them comparable.

4th  meeting

The meeting was held in Hannover and involved the whole Hannover team and Joanna Tyrowicz. We worked the most on visualization of the project results, some very cool movies are to appear on this website shortly. The key component to understand the financial behavior of agents with incomplete rationality concerns the zero-borrowing constraint. In particular, one wants to understand the consequences of the following situation: expecting higher revenues in the future (e.g. because of the government-subsidized old-age savings instruments) the agents may want to borrow against this future income. It is conventional not to allow the agents in the model to have negative net assets, which is very close to the real world. But the consequences of this convention for welfare as well as macroeconomic effects are not very obvious.

5th meeting

The meeting was held in Trier, where Patrick Puhani and Joanna Tyrowicz worked on developing further the theoretical paper. The preliminary results form a model with a variety of incompletely rational agents and three types of voluntary pension savings instruments demonstrate that indeed policy options are not very obvious. Specifically, there are strong interactions between the presence of different types of agents in the economy and the features of the instruments intended at fostering savings. At this point, with those interesting results, we draft the first take on the main policy prescriptions.

6th meeting

The meeting was held in Prague, as the team members were passing by Prague and we could use this opportunity to work closely on the project development. Most of the attention was devoted to the microeconometric evidence on joint (and early) retirement decisions, which serves the purpose of designing our simulation tool for the joint microeconomic paper.  Our model now extends the current state-of-the-art on several dimensions and we are working on ingenious ways to show how our results would look without those multiple extensions (one at the time). 

7th  meeting

The meeting was held in Frankfurt, where we discussed our project outcomes with Alexander Ludwig. A super useful series of meetings for our theoretical paper. We discussed the features of the theoretical setup which are particularly useful for the demonstration purposes. We also chatted a bit about the sensitivity analyses and additional tests that would be of value to convince the elaborate on our findings. In particular, we talked on whether the model should have idiosyncratic shocks and what types of incomplete rationality are of particular interest in this strand of literature. Now, we sit down to implement all these comments in writing.

8th meeting

This was the longest of our meetings, two weeks of day by day hard work! We have concluded the work on our macroeconomic model and derived the results. We still have to run some sensitivity analyses, but the first draft of the text should be available soon. We also gave a lot of work and thought to our microeconomic paper: we studied the features of our SMM model and performed a number of policy experiments. Not all works as it should in this model, but with the insightful comments from WEAI presentation in San Francisco and with the pressure of getting ready for RCEA conference in Waterloo, were super motivated to fix the remaining issues. Finally, we also discussed a lot over the policy evaluation microeconometric paper using exogenous variation in retirement eligibility among women in Germany from the turn of century. We are really hopeful about this paper!

9th meeting

The meeting in London (where Patrick Puhani is spending his sabbatical) was incredibly constructive. We discussed thoroughly the macroeconomic paper, receiving insightful comments from UCL colleagues. We also finalized the policy design in our study. The paper is almost finished and we hope to be able to submit it to further conferences and workshops for 2020 events. For now, we prepare for the presentation at  Spanish Public Choice Workshop. The second part of the meeting was devoted to the microeconometric study. 

Opublikowane | Published

  • Excessive discounting, longevity expectations, and retirement saving: An online survey | Journal of Behavioral and Experimental Economics

    I report results of a major online experiment focused on two behavioural mechanisms that might affect long-term saving: impatience (excessive discounting) and distorted beliefs about own longevity. I observe the longevity expectations to be generally reasonable, both in terms of their mean values and their determinants, although the estimates show large variance and, on balance, slight pessimism. In line with previous studies, I find excessive discounting and (self-reported) insufficient retirement saving to be prevalent. Both expectations and excessive discounting affect retirement saving in the natural direction. However, there is no link between these two determinants: neither overly optimistic nor pessimistic predictions concerning own longevity seem to be systematically linked to excessive discounting. Thus, these two behavioural effects neither strengthen nor cancel each other out.

  • Pension reforms and couples’ labor supply decisions | Labour Economics

    To determine how wives’ and husbands’ retirement options affect their spouses’ (and their own) labor supply decisions, we exploit (early) retirement cutoffs by way of a regression discontinuity design. Several German pension reforms since the early 1990s have gradually raised women’s retirement age from 60 to 65, but also increased ages for several early retirement pathways affecting both sexes. We use German Socio-Economic Panel data for a sample of couples aged 50 to 69 whose retirement eligibility occurred (i) prior to the reforms, (ii) during the transition years, and (iii) after the major set of reforms. We find that, prior to the reforms, when several retirement options were available to both husbands and wives, both react almost symmetrically to their spouse reaching an early retirement age, that is both husband and wife decrease their labor supply by about 5 percentage points when the spouse reaches age 60). This speaks in favor of leisure complementarities. However, after the set of reforms, when retiring early was much more difficult, we find no more significant labor supply reaction to the spouse reaching a retirement age, whereas reaching one’s own retirement age still triggers a significant reaction in labor supply. Our results may explain some of the diverse findings in the literature on asymmetric reactions between husbands and wives to their spouse reaching a retirement age: such reactions may in large parts depend on how flexibly workers are able to retire.

    Joanna
    Tyrowicz
  • Fiscal incentives to pension savings - are they efficient? | Journal of Pension Economics and Finance

    Financing consumption of the elderly in the face of the projected increase in life expectancy is a key challenge for economic policy. Moreover, standard structural models with fully rational agents suggest that about 50-60 percent of old-age consumption is financed with voluntary savings, even in the presence of a fairly generous public pension system. This is clearly inconsistent with either the data, or the alarming simulations of old-age poverty in the years to come. Old-age saving (OAS) schemes are widely used policy instruments to address this challenge, but structural evaluations of such instruments remain rare. We develop a framework with incompletely rational agents: lacking financial literacy and experiencing commitment difficulties. We study a broad selection of OAS schemes and find that they raise welfare of financially illiterate agents and to a lesser extent improve welfare of agents with a high degree of time inconsistency. They also reduce the incidence of poverty at old age. Unfortunately, these instruments are fiscally costly, induce considerable crowd-out and direct fiscal transfers mostly to those agents, who need it the least.

    Joanna
    Tyrowicz
    Krzysztof
    Makarski
    Artur
    Rutkowski
  • Evaluating an old-age voluntary saving scheme under incomplete rationality | Gospodarka Narodowa

    We provide ex ante welfare, fiscal and general macroeconomic evaluation of the voluntary old-age saving scheme recently introduced in Poland (Pracownicze Plany Kapitałowe, Employees’ Capital Plans). ECPs provide tax redemptions as well as lump-sum transfers with the objective to foster old-age savings. Reduction in capital income tax revenues and a rise in expenditure needs to be compensated through adjustment in other taxes. We employ an overlapping generations model (OLG) to gauge the plausible magnitude of the macroeconomic and welfare effects and provide insights in terms of microfoundations of these adjustments. Our OLG model features voluntary participation and innovates relative to the literature by introducing agents with hand-to-mouth preferences. We find relatively high crowding out of private savings. In our preferred specification roughly 0.08 to 0.09 PLN of each 1 PLN allocated to ECPs are actually new savings, the rest being displaced from unincentivized private voluntary savings. The plausible values of the effective capital growth range between 0.03 and 0.42 of 1 PLN in ECPs. ECPs reduce welfare of the fully rational agents, unless they offer a sufficiently large annuity. ECPs provide consumption smoothing and interest income to HTM agents.

    Attached to this publication are data covering the sensitivity analyses for the alternative calibrations of the share of incompletely rational agents in the economy. The file comprises data and program files (stata *.do files) which delivers the replication of tables in figures in the paper for the alternative shares of hand-to-mouth (HTM) agents in population. 

    Artur
    Rutkowski
  • A regression discontinuity evaluation of reducing early retirement eligibility in Poland | International Journal of Manpower

    The reform introduced in Poland in 2009 substantially and abruptly reduced the number of workers eligible for early retirement. This paper evaluates the causal effects of this reform on labor force participation and exit to retirement. We use rich rotating panel from the Polish Labor Force Survey and exploit the discontinuity imposed by this reform. We find a statistically significant, but economically small discontinuity at the timing of the reform. The placebo test shows no similar effects in earlier or later quarters, but in a vast majority of specifications the discontinuity is not larger for the treated individuals, i.e. those whose occupation lost eligibility. We interpret these results as follows: the changes in the eligibility criteria were not instrumental in fostering the participation rates among the affected cohort, i.e. the immediate contribution to increased labor force participation of these cohorts is not economically large.

    Joanna
    Tyrowicz
    Oliwia
    Komada
    Paweł
    Strzelecki
  • Inequality in an OLG economy with intra-cohort heterogeneity and an obligatory pension system | Journal of Economic Inequality

    While the inequalities of endowments are widely recognized as areas of policy intervention, the dispersion in preferences may also imply inequalities of outcomes. In this paper, we analyze the inequalities in an OLG model with obligatory pension systems. We model both policy relevant pension systems (a defined benefit system — DB — and a transition from a DB to a defined contribution system, DC). We introduce within cohort heterogeneity of endowments (individual productivities) and heterogeneity of preferences (preference for leisure and time preference). We introduce two policy instruments, which are widely used: a contribution cap and a minimum pension. In theory these instruments affect both the incentives to work and the incentives to save for the retirement with different strength and via different channels, but the actual effect attributable to these policy instruments cannot be judged in an environment with a single representative agent. We show four main results. First, longevity increases aggregate consumption inequalities substantially in both pension systems, whereas the effect of a pension system reform works to reinforce the consumption inequalities and reduce the wealth inequalities. Second, the contribution cap has negligible effect on inequalities, but the role for minimum pension benefit guarantee is more pronounced. Third, the reduction in inequalities due to minimum pension benefit guarantee is achieved with virtually no effect on capital accumulation. Finally, the minimum pension benefit guarantee addresses mostly the inequalities which stem from differentiated endowments and not those that stem from differentiated preferences.

    Our data are shared here.

    Joanna
    Tyrowicz
    Krzysztof
    Makarski