Final Report Summary - WSCWTBDS (The Welfare State in a Complex World Taxes and Benefits in a Diverse Society)
The welfare state encompasses much of government activity – including personal taxation, unemployment and disability insurance and public pensions. Making these benefits available is more costly than shown by the financial outlays as the imposition of taxation and the provision of benefits reduce the incentive for individuals to seek and retain employment. On the other hand, the value of the welfare state to citizens, by providing financial assistance in old age or when economic shocks hit, can be very large. Balancing these costs and benefits is a key challenge for policy-makers. The aim of this project was to make substantial contributions to the evidence base on the welfare state – how it affects household behavior and how governments should design tax and transfer programmes. Our work has both theoretical and empirical components. We very briefly discuss both strands below.
Our theoretical work has considered how features of the welfare state should be designed. One example relates to how entitlements to benefits should be assessed. In aiming to implement their redistributive objectives, governments, in almost all cases, levy taxes and provide transfers that are deterministically conditioned on a set of household characteristics. We study whether random transfers can improve outcomes and we provide necessary and sufficient conditions for the existence of socially-useful randomizations in entitlements. They require that those that the government wants to help with the random transfers be less risk averse than the rest of the population. Another example of our theoretical work relates to the taxation of couples. An important question when considering couples without children is whether the second earner is taxed less when the other member of the couple works than when (s)he does not work. Internationally, many tax systems exhibit this property. We investigate the conditions under which this is justified. In other applications we study the implications of how income taxation and public pension schemes interact and how they can be designed to complement each other to achieve a government’s redistributive objectives.
The empirical side of our project has tackled questions focussed on understanding why individuals make the particular labour supply and savings decisions that they do. This can inform how they would respond if the government were to change tax and transfer policy. One example develops a rich labour supply model to identify why individuals work the number of hours that they do. This involves separating the effect coming from the choice they make over the hours they would like to work from the effect of employers offering them only a limited number of hours choices. A new source of identification (which comes from the non-linearity of budget constraints implied by the tax and transfer schedule) is used to separately identify these. The estimated model is used to show (for example) that women likely to be subject to the choice constraints belong to significantly poorer households than average and work shorter hours. In other work we put forward a new method for estimating the patience of individuals (formally, their discount rates). The extent to which members of a household are patient determines the extent to which they will accumulate a financial buffer to protect themselves against life’s risks, and will bear upon how they will respond to certain aspects of government tax and transfer policies. The level of patience exhibited by households needs to be taken into account by policy-makers in designing programmes (particularly public pensions).
The project also considered how changes to policy pertaining to the welfare state should be evaluated. We give two examples here. The first relates to analyzing the distributional impact of tax and benefit changes. These are typically assessed by comparing their impact on current income - we show how evaluating the effect on lifetime income can lead to quite different conclusions. The second concerns what is considered to be the gold-standard in programme evaluation – a social experiment where participants are randomly assigned to getting a treatment or not. We empirically investigated the importance of ‘randomisation bias’ - whereby the act of randomisation induces a change in behavior of those participating in the experiment.
As part of our dissemination and knowledge transfer activities, we organized and hosted two large conferences at the Institute for Fiscal Studies in London which attracted leading academics working on issues related to the welfare state (as well as attendees from outside of academia). Alongside the second conference, we organised an event for senior economists in the UK civil service to highlight some of the policy-relevant work undertaken by academics attending the conference. The programmes for these events can be found at https://www.ifs.org.uk/events/791 and https://www.ifs.org.uk/events/1452
Our theoretical work has considered how features of the welfare state should be designed. One example relates to how entitlements to benefits should be assessed. In aiming to implement their redistributive objectives, governments, in almost all cases, levy taxes and provide transfers that are deterministically conditioned on a set of household characteristics. We study whether random transfers can improve outcomes and we provide necessary and sufficient conditions for the existence of socially-useful randomizations in entitlements. They require that those that the government wants to help with the random transfers be less risk averse than the rest of the population. Another example of our theoretical work relates to the taxation of couples. An important question when considering couples without children is whether the second earner is taxed less when the other member of the couple works than when (s)he does not work. Internationally, many tax systems exhibit this property. We investigate the conditions under which this is justified. In other applications we study the implications of how income taxation and public pension schemes interact and how they can be designed to complement each other to achieve a government’s redistributive objectives.
The empirical side of our project has tackled questions focussed on understanding why individuals make the particular labour supply and savings decisions that they do. This can inform how they would respond if the government were to change tax and transfer policy. One example develops a rich labour supply model to identify why individuals work the number of hours that they do. This involves separating the effect coming from the choice they make over the hours they would like to work from the effect of employers offering them only a limited number of hours choices. A new source of identification (which comes from the non-linearity of budget constraints implied by the tax and transfer schedule) is used to separately identify these. The estimated model is used to show (for example) that women likely to be subject to the choice constraints belong to significantly poorer households than average and work shorter hours. In other work we put forward a new method for estimating the patience of individuals (formally, their discount rates). The extent to which members of a household are patient determines the extent to which they will accumulate a financial buffer to protect themselves against life’s risks, and will bear upon how they will respond to certain aspects of government tax and transfer policies. The level of patience exhibited by households needs to be taken into account by policy-makers in designing programmes (particularly public pensions).
The project also considered how changes to policy pertaining to the welfare state should be evaluated. We give two examples here. The first relates to analyzing the distributional impact of tax and benefit changes. These are typically assessed by comparing their impact on current income - we show how evaluating the effect on lifetime income can lead to quite different conclusions. The second concerns what is considered to be the gold-standard in programme evaluation – a social experiment where participants are randomly assigned to getting a treatment or not. We empirically investigated the importance of ‘randomisation bias’ - whereby the act of randomisation induces a change in behavior of those participating in the experiment.
As part of our dissemination and knowledge transfer activities, we organized and hosted two large conferences at the Institute for Fiscal Studies in London which attracted leading academics working on issues related to the welfare state (as well as attendees from outside of academia). Alongside the second conference, we organised an event for senior economists in the UK civil service to highlight some of the policy-relevant work undertaken by academics attending the conference. The programmes for these events can be found at https://www.ifs.org.uk/events/791 and https://www.ifs.org.uk/events/1452