Measuring Attitudes towards Public Spending using a Multivariate Tax Summary Experiment

The forthcoming article “Measuring Attitudes towards Public Spending using a Multivariate Tax Summary Experiment” by Lucy Barnes, Jack Blumenau and Benjamin E. Lauderdale is summarized by the author(s) below. 

What do citizens want to spend public money on? How much do they want to spend, if the consequences are increased taxes? Given the size and centrality of public budgets in contemporary western post-industrial democracies, these questions are of crucial importance.  

However, measuring public opinion on these preferences is difficult. There is a tension between the complexity of budgetary realities and the limits of citizens’ engagement with public policy detail. How much information can we extract about public views on taxation and government spending, when few people have ever considered their ideal budget? Existing surveys often ask simple questions which make life easy for respondents, but neglect real and complex budget constraints. This risks eliciting unrealistic demands. Alternatively, we might ask respondents to describe a comprehensive spending allocation while imposing budgeting constraints. However, such a task is likely to be extremely difficult for the average respondent, and diverges from the way real budget proposals are experienced by voters. 

We thus propose a strategy for inferring coherent budget preferences from accessible questions about spending – specifically, a multivariate forced choice task. In our experiment, respondents are asked to choose between the status quo spending allocation and an alternative which maintains budget balance. These alternatives include proposals for re-allocations of spending across different areas (such as health, welfare, criminal justice), and changes to the overall level of spending and taxation. We combine the results of these choices with a model of the respondents’ decisions to infer the distribution of underlying preferences which is consistent with these choices, made under the strict budget constraint.  

We apply this method to data from the UK in 2018. British taxpayers receive annual summaries indicating how their taxes are allocated across a variety of spending categories, which we use to describe the status quo. The results indicate that the British public endorse spending increases in many areas, including some of the most expensive (pensions, education, and health). In only a few, small, spending areas do UK respondents favour cuts (overseas aid, contributions to the EU budget). This snapshot of preferences suggests that British taxpayers would accept tax increases of 7% to support these spending increases.  

Our model also allows us to estimate average preferences as a function of individual-level covariates, with three central results. First, we recover sensible partisan differences on the level and allocation of spending. Second, we confirm the fiscal conservatism of younger voters. In line with recent scholarly work but cutting against popular descriptions of “millennial socialists”, younger people endorse lower spending in all categories except culture, overseas aid, and the EU budget. Finally, we recover a multidimensional structure within spending preferences: support for greater spending on areas except defence, business, pensions and justice describes a dimension oriented from Conservative-Leave to Labour-Remain voters, while support for spending on areas excluding the EU budget, culture, and overseas aid constitutes a second dimension oriented from Conservative-Remain to Labour-Leave. 

We validate our approach by using our results to generate a proposal for the average preferred level of taxation and spending across categories, and to predict levels of support for this proposal versus the status quo. We evaluate this prediction in a validation experiment, asking respondents to directly compare status quo taxation and spending levels to the proposal generated by our model. Our model indicates that 73% of respondents should support the proposal; in fact, 58% of the new survey respondents do. By our pre-registered criteria, we count this as a “moderate success.” We also show that the demographic groups predicted to be more supportive of the proposal are indeed more likely to support it in the validation experiment.  

Beyond our substantive UK results, the approach to measurement should be useful in many other applications. For example, our improvements to the trade-off between real-budget complexity and respondent task difficulty can be implemented across other complex budget-preference measurement tasks. The method also generates data suitable for other important research questions, such as social choice analyses to identify winning proposals, or welfare analyses to assess which groups are best served by particular proposals. 

About the Author(s): Lucy Barnes is Associate Professor in Comparative Politics, Department of Political Science at University College London, Jack Blumenau is Lecturer in Political Science and Quantitative Research Methods, Department of Political Science at University College London and Benjamin E. Lauderdale is Professor of Political Science, Department of Political Science at University College London. Their research “Measuring Attitudes towards Public Spending using a Multivariate Tax Summary Experiment” is now available in Early View and will appear in a forthcoming issue of the American Journal of Political Science.