The Political Economy of Progressive Tax Reform. Part 1: Citizens

To understand the determinants of citizens' preferences over property taxes, we randomly assign survey respondents to receive combinations of 6 survey-experimental interventions.
1- In the Correcting Misperceptions treatment respondents are shown photographs of 5 properties that are representative of Lahore’s regressive schedule and we ask citizens to assess the tax liability associated with each property. We then provide respondents with the actual average tax rates faced by each of the properties and show them the actual shape of the tax schedule in Lahore.
2- The placebo intervention acts as a placebo for the interventions discussed below. This group receives an informational video message about the different tiers of government and the assignment of revenue and spending functions to each tier.
3- The Public Goods intervention provides respondents with information on local public good deficits in Lahore and argues that the lack of financing, which is a consequence of low property tax utilization is a big constraint on the government’s ability to meet the public service delivery needs of citizens
4- The Revenue Leakage intervention provides respondents with information on the magnitude of the property tax compliance challenge in Lahore and the potential financing that can become available if there was full compliance. It ends with the message that raising adequate financing for local public good provision in the city will be difficult for government in the absence of improved compliance
5- The Spending Leakage intervention provides respondents with information on citizen perceptions in Lahore about tax reciprocity, i.e. the proportion of taxes that are spent on the provision of public services in the city. It ends with the message that raising adequate financing for local public good provision in the city will be difficult for government in the absence of measures that can strengthen tax reciprocity
6- The Elite Capture intervention provides respondents with examples of recent cases where opposition from high value property owners in Lahore successfully delayed the introduction of reforms designed to raise more property taxes from the wealthy. It ends with the message that raising adequate financing for local public good provision in the city will be difficult for government in the absence of cooperation from the wealthy elite of the city.

2024-05-27

2025-02-18

Preferences for the level and progressivity of property taxes. We elicit these through an android app that shows respondents 9 randomly selected properties and elicits respondents' preferred tax rates for them.

Our most important primary outcome is survey respondents' desired degree of tax progressivity. We use four measures of progressivity that are commonly used in the literature. Each measure is normalized so that 0 means a proportional tax system, positive numbers mean progressive tax systems, and negative numbers mean regressive tax systems. We also combine the four measures into an index of progressivity since they each capture slightly different aspects of the progressivity of the overall tax schedule, and so that we can use a single measure of progressivity when we explore heterogeneity of the treatment effects.

Our progressivity measures are:
1- Tax Elasticity The tax elasticity is the regression coefficient from the regression of log tax liability on log property value.
2- Kakwani Index. This index is based on the Lorenz curves of property wealth and of taxes paid.
3- Feldstein tau. This measure derives from Feldstein(1969) The measure is the estimated tau from the non-linear regression T = V - b V^(1-tau)
4- Top Tax Rate. Since much of the literature focuses on progressivity at the very top of the distribution (e.g. Piketty & Saez, 2007), our fourth measure focuses on the top 10\% of the property value distribution: It is the ratio of the average tax rate among the 10% wealthiest properties to the average tax rate overall in Lahore.
5- Progressivity Index. Following Kling, Liebman, and Katz (2007), our final progressivity measure is an equally-weighted index of our four measures.

please see our pre-analysis plan for details on all secondary outcomes and how we adjust inference for them.

We randomly assign survey respondents to one of six treatment arms. This is described in detail in our attached pre-analysis plan.

randomization done on a computer using stata.

survey proeprty.

No

7,577

7,577

1,263 in each of 6 treatment arms.

please see our attached pre-analysis plan

Trust_Taxation_Pre_Analysis_Plan.pdf