Fiscal Studies最新文献

The use of randomised control trials (RCTs) has become widespread in economics and other social sciences, and is likely to grow further as new digital tools and increasingly rich data facilitate the design and implementation of experiments. When rigorously designed and implemented, they have the potential to offer the most reliable empirical evidence on the causal impact of an intervention. But for that potential to be realised, RCTs need to be sufficiently powered to detect a meaningful effect, or to say confidently that the effect is negligible. This symposium offers practical insights for researchers on designing more powerful experiments and computing the required sample size, accompanied by tools that researchers can use in designing their own RCTs.

The first paper, by David McKenzie, discusses how to improve power at each stage of an RCT – design, implementation and analysis. While increasing sample size is the default option, McKenzie offers guidance on many other options available to researchers and why they work. The second paper, by Brendon McConnell and Marcos Vera-Hernández, dives into detailed aspects of implementing sample size calculations for different randomisation designs, and offers the formulae, tools and computer code necessary to implement them in practice. The final paper, by Brandon Hauser and Mauricio Olivares, studies hypothesis testing in randomised experiments, and its consequences for sample size calculations. The paper shows how small deviations from the most standard assumptions invalidate standard randomisation-based inference, and provides useful results and guidance for how to adapt power analysis to ensure that calculations remain valid.

Basic methods to compute required sample sizes are well understood and supported by widely available software. However, researchers often oversimplify their sample size calculations, overlooking relevant features of their experimental design. This paper compiles and systematises existing methods for sample size calculations for continuous and binary outcomes, both with and without covariates, and for both clustered and non-clustered randomised controlled trials. We present formulae accommodating panel data structures and uneven designs, and provide guidance on optimally allocating sample size between the number of clusters and the number of units per cluster. In addition, we discuss how to adjust calculations for multiple hypothesis testing and how to estimate power in more complex designs using simulation methods.

This paper revisits the problem of power analysis and sample size calculations in randomised experiments, with a focus on settings where inference on average treatment effects is conducted using randomisation tests. While standard formulas based on the two-sample $�t$-test are widely used in practice, we show that these calculations may yield misleading results when directly applied to randomisation-based inference – unless certain assumptions are met. We demonstrate that differences in potential outcome variances or unequal group sizes can distort the behaviour of the randomisation test, leading to incorrect power and flawed sample size calculations. However, a simple adjustment – studentising the test statistic – restores the validity of the randomisation test in large samples. This adjustment allows researchers to safely apply standard power and sample size formulas, even when using randomisation inference. We extend these results to a range of experimental designs commonly used in applied economics, including stratified randomisation, matched pairs and cluster-randomised trials. Throughout, we provide practical guidance to help researchers ensure that their design-stage calculations remain valid under the inferential methods they plan to use.

This paper offers practical advice on how to improve statistical power in randomised experiments through choices and actions researchers can take at the design, implementation and analysis stages. At the design stage, the choice of estimand, choice of treatment, and decisions that affect the residual variance and intra-cluster correlation can all affect power for a given sample size. At the implementation stage, researchers can boost power through increasing compliance with treatment, reducing attrition and improving outcome measurement. At the analysis stage, power can be increased through using different test statistics or estimands, through the choice of control variables, and through incorporating informative priors in a Bayesian analysis. A key message is that it does not make sense to talk of ‘the’ power of an experiment. A study can be well powered for one outcome or estimand but not others, and a fixed sample size can yield very different levels of power depending on researcher decisions.

Why do individuals fail to support tax reforms that serve their material self-interest? Using an original online survey experiment spanning eight countries and 12,000 respondents across Latin America, one of the most unequal regions in the world, we find evidence for a previously unexplored explanation: misperceptions regarding the current incidence of the taxes to be reformed. Treated respondents who are informed that the value-added tax (VAT) is regressive are significantly more likely to prefer reforms that make it more progressive. Treatment effects are driven by the large fraction of respondents who underestimate the regressivity of the VAT. They are disproportionately right-leaning and more likely to attribute success to individual effort than luck; treatment effects are largest among individuals who hold these views of the world. Many respondents exhibit inconsistent preferences, violating the generalised axiom of revealed preferences; treatment effects are significantly stronger among consistent respondents. These findings expand the potential for information interventions to shift support for fiscal policy reforms protecting the most vulnerable.

We study the effect of the introduction of the UK Shared Parental Leave policy in 2015 on both the uptake and the length of leave taken by fathers. Using data from the UK Household Longitudinal Study and a regression discontinuity in time design, we find no evidence that the reform increased either uptake or length of paternal leave, reinforcing questions about its effectiveness.

In this paper we examine different channels through which poverty affects child outcomes, as well as the evidence regarding the magnitude of their impacts. We begin by discussing the family investment model, which highlights the constraints that poverty or lack of income pose on a family's ability to purchase goods or services that contribute to the child's overall development, and the family stress model, which emphasises the emotional toll that experiencing poverty can have on parents and (directly and indirectly) on children. We then devote special attention to a more recent perspective on the family stress model, originating at the intersection of cognitive and developmental psychology and behavioural economics, which posits that another pathway through which poverty-induced stress can affect family well-being is through the effect of poverty on parental cognitive functioning.

We document new stylised facts about substantial gifts and loans made in Great Britain. The vast majority of gifts and loans are made from parents to their adult children, and we find substantive differences in receiving by ethnicity, region and gender. Gifts are only weakly related to individuals' economic resources once their parents' socio-economic status is accounted for, but loans are made more frequently to those with lower wealth. Transfers modestly increase inequalities in total inflows (income plus transfers) over early adulthood. Over an eight-year period, cumulative transfers received are worth 0.5 per cent of income for those in the bottom income fifth compared to 2.6 per cent of income for the top fifth. Over half of the value of gifts is given by the wealthiest fifth of individuals and transfers increase inequalities between those with parents of higher and lower socio-economic status. However, transfers decrease relative inequalities in wealth, being largest as a share of wealth for those at the bottom of the wealth distribution.

Demand aggregation through Framework Agreements (FAs) has emerged as a promising tool to support the efficient expansion of affordable healthcare in developing countries. However, the effectiveness of FAs in achieving lower costs may be hindered if prompt payment is not enforced. This paper estimates the impacts of a reform implemented in Chile in 2014, which introduced a prompt payment enforcement procedure in the FAs that supplied public hospitals. Under this reform, firms were allowed to suspend dispatches until overdue bills were paid. The results from a difference-in-differences estimation indicate that hospitals with greater exposure to the reform, measured by their larger share of late bill payments, reduced their payment delays after 2013 compared with less-exposed hospitals, without compromising health-care quality. The resulting decrease in the average financial cost of FAs led to lower prices, highlighting the importance of prompt payment enforcement for realising savings through FAs.