Public spending on families in aging populations: a survey experiment

How to cite

Timár, Eszter; Gassmann, Franziska. 2025. "Public spending on families in aging populations: a survey experiment." AEA RCT Registry. AEARCTR-0016312. https://doi.org/10.1257/rct.16312-1.1

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Eszter Timár* , Franziska Gassmann

* Primary investigator

Registry ID
AEARCTR-0016312
DOI
https://doi.org/10.1257/rct.16312-1.1
First registered
Last updated
Status
In development
Trial dates
2025-07-15 – 2025-08-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/16312

Abstract

European welfare states face a dual demographic challenge of sustained low fertility and rapid population aging. While public investment in children is widely recognized as critical for mitigating the long-run fiscal and economic consequences of these trends, the political feasibility of such investment remains uncertain. The conventional view, rooted in overlapping generations models, suggests that aging societies will increasingly prioritize spending on the elderly, sidelining expenditures on families and children. Yet these models typically overlook the intertemporal nature of self-interest and the potential endogeneity of future pension sustainability to present-day child-oriented spending.
This paper studies how the public forms preferences over intergenerational redistribution in aging societies, with a focus on public spending for families with children. We develop a stylized extension of Galasso and Profeta's (2007) model to demonstrate that current expenditures on children—via their impact on future fertility and human capital—can enhance the fiscal sustainability of pension systems. This creates a long-term economic interest for even the childless median voter to support such policies. Nevertheless, these dynamic linkages are often not salient to the electorate, raising the question of whether communication and framing strategies can shift public attitudes.
To address this, we design and implement a large-scale survey experiment in Germany and Hungary (target N = 7,000), two countries with contrasting welfare state models and demographic trajectories. The survey measures respondents’ perceptions of demographic change, their policy preferences, and their responses to randomized informational treatments. These treatments present child and family policies either as a means to support current families (short-term frame) or as investments to ensure the sustainability of the welfare state (long-term frame). This design enables me to test whether emphasizing future economic benefits can broaden support beyond the directly affected population.
This paper contributes to the literature on the political economy of intergenerational redistribution by identifying when and how framing can mobilize support for forward-looking social investment. It also speaks to broader debates on welfare state adaptation and the role of strategic communication in fostering politically viable responses to structural demographic shifts.

Intervention

The experiment will be embedded into the survey questionnaire.
The experimental block will follow a quasi-prepost design. This means that the outcome variables are measured before and after the treatment, but with similar (rather than identical) questions. This is a form of repeated measurement design which aims to minimize the consistency pressure arising from measuring the outcome twice (Clifford et al., 2021). Eliciting a prior first-stage outcome variable allows the researcher to test the heterogeneity of treatment effect dependent on prior beliefs (Stantcheva, 2023). For example, one can check whether the treatment effects are greater for those who received a larger “shock” from the treatment (ibid.). Pre-testing also allows the researcher to control for individual fix-effects, contributing to greater measurement precision (Clifford et al., 2021).
Pre-test. In the pre-test, respondents are asked to rate government expenditure areas based on their importance. Supporting families with children will be listed alongside other conventional functions (such as health care, infrastructure, pensions for the elderly) and functions not typically associated with government duties (e.g., supporting international corporations). Asking about multiple expenditure functions will avoid drawing explicit attention to our topic of interest, which minimizes priming, experimental demand effect, and consistency pressures.
The treatment. In the experimental section, respondents will be randomly assigned to three equally sized groups (one control and two treatment arms). This will be done using Qualtrics’ built-in randomize function. Each experimental arm will be shown a short (2-minute) video about child and family policies. No video will be presented to the control group.
The video treatment administered to one treatment arm will emphasize the immediate benefits of spending on children and families, such as reducing the burden of parents and improving the well-being of children. The expectation is that this will prime respondents to think about their immediate self-interest, which will depend on whether they have children. This frame may also increase support among non-parents and the elderly who are altruistic (i.e., utility through altruism).
The other video treatment will introduce the notion of endogeneity of future pensions to current child benefits. It will thus emphasize the long-term benefits for the economy and for pension sustainability. It is expected to increase support particularly among childless, but young individuals, for whom the direct utility from child benefits is zero. In terms of our utility model developed in the theoretical section, the treatment aims to activate the forward-looking term. Because of discounting or bounded rationality, the intertemporal link may be undervalued without the provision of information on it.
Post-test. Following the treatment, I will measure both the first-stage and second-stage outcomes. It is important to measure the first-stage outcome because it is the only opportunity to check if the treatment has worked (successfully shifted perceptions). The second-stage outcome refers to my main variables of interest, i.e., respondents’ policy preferences. For more details, see the operationalization of key variables in the Analysis Plan section.

Experimental design

The experiment relies on an information/framing treatment using a video, embedded in a survey. Respondents will be randomly assigned to three equally sized groups (one control and two treatment arms). Each experimental arm will be shown a short (2-minute) video about child and family policies. No video will be presented to the countrol group.
The video treatment administered to one treatment arm will emphasize the immediate benefits of spending on children and families, such as reducing the burden of parents and improving the well-being of children. The other video treatment will introduce the notion of endogeneity of future pensions to current child benefits.

Design details

Randomization method
Randomization done by a computer (using the randomization function of the Qualtrics survey software).
Randomization unit
Individual
Clusters
2 countries
Observations
7000 individuals
Arms
2333 control, 2333 treatment condition A, 2300 treatment condition B.
Primary outcomes
Support for public spending family policies, support for public spending on family policies at the expense of pensions, and willingness to pay for family policies via a small increase in taxes.

Cited by

No citation data available for this registration yet.

Keywords