We model the behaviour of a mortgagor considering to evade the real estate transfer tax. We build an observable measure of over-appraisal that is inversely related to tax evasion and conclude that the tax authority could focus auditing efforts on low-appraisal transactions. Our empirical analysis confirms the predictions using a unique database for Spain, where we directly observe: real payment, value declared to the authority and appraisal. We also argue that individual characteristics (e.g. education of the buyer) and local characteristics (e.g. local levels of corruption and trust) explain part of the observed heterogeneity in evasion. The theoretical model is augmented with guilt and stigma to be consistent with the observed heterogeneity.

Introduction

Transfer taxes (i.e. taxes on real property transactions) are common in most OECD countries and yet they remain understudied (Best and Kleven 2018). More generally, the empirical literature on tax evasion has been facing the challenge of obtaining reliable data.1 Difficulty understanding and observing fraud has obvious consequences on the effectiveness of audits performed by tax authorities. Our paper aims to help fill these gaps by focusing on home buyers’ strategic behaviour, whereby we test our prediction and provide some policy recommendations. To this end, we present a model where a mortgagor decides their housing expenditure, together with the share of the latter they declare to the tax authority. By under-declaring to the tax authority, buyers can reduce their expected tax burden. However, when they are short on assets, they may also have an interest in inflating the declared appraisal of the house, in order to facilitate obtaining a mortgage. The model exploits the tension between obtaining a larger mortgage and paying fewer taxes and it uncovers the relationship between tax evasion, access to cash (or other untraceable payment systems) and housing declared over-appraisal (that is, the ratio between declared appraisal and declared transaction).2 In particular, when agents are liquidity-constrained, potential mortgagors and financial institutions have an incentive to collude and inflate the appraisal that is declared. Typically, those same potential mortgagors are unlikely to fraud.3 Our result has an interesting policy implication: declared over-appraisal is observable, hence the tax authority could use it as a screening device to determine which transactions to audit.