Brian D. Galle (Georgetown University Law Center), David Gamage (Indiana University Maurer School of Law), & Yulia Kuchumova (National Research University Higher School of Economics) have posted Tax Base Diversification as an Enforcement Tool on SSRN. Here is the abstract:
We examine when it is optimal to employ sales or VAT-type taxes as complements to a labor income tax. We find a Ramsey-type result in which each tax instrument should be imposed in inverse proportion to the combined elasticity of real and avoidance responses to the respective tax. Contrary to some prior results, we find that sales-type taxes are optimally non-zero across a variety of settings, and in particular when the (weighted) elasticity of taxable income with respect to the wage tax is greater than the cross-elasticity of taxable income with respect to the sales tax. We argue this parameter should be of key interest for empirical research, and conduct some exploratory studies in which we estimate it using changes in California local sales taxes and EU VAT rates. Our estimates generally suggest non-zero consumption taxes would be efficient.
Nontechnical Summary:
Economic theory holds that taxes on wages are, with certain limited exceptions, generally preferable to taxes based on transactions, such as the U.S. sales tax or global value-added taxes. If we set aside savings, all income earned is also spent, so that taxing earnings and taxing spending can be economically equivalent. But if taxes on sales have exceptions or otherwise impose different rates on different items, as they almost always do in the real world, the sales tax is less efficient, because it distorts consumer choices in ways the tax on wages would not. At the same time, some authors have recognized that this general proposition may no longer hold once tax avoidance is taken into account. For example, development consultants often observe that a VAT can help nations with weak wage-tax institutions collect revenue. The intuition behind these claims, as we understand it, is that a sales-type tax is effectively another way of collecting wage taxes. If the two are economically equivalent, but a sales tax is harder to avoid than the income tax, it might be preferable to collect the “wage” tax at the cash register.
In this paper, we develop this intuition more formally. When, if ever, will these enforcement gains ever exceed the economic distortion caused by taxing some sales differently than others? We develop a model that allows us to identify the “sufficient statistics” or economic variables a policy maker would need to observe to decide whether to impose a sales tax. Generally speaking, we show that a sales tax is more likely to be a good policy choice when what we call the cross-elasticity of taxable income with respect to the sales tax (“XETI”) is smaller in absolute magnitude than the elasticity of taxable income with respect to the wage tax (“ETI”). That is, if sales taxes distort taxpayers' reported wages less than a revenue-equivalent wage tax, the sales tax is more likely to be a good choice. Since prior work has not identified this important role for XETI, there are no prior estimates of it. We therefore offer some preliminary evidence, based on taxpayer responses to California municipal sales tax elections and EU VAT changes, on whether XETI in fact is smaller in magnitude than ETI. Our estimates of XETI suggest that governments should likely collect at least some revenue through a sales tax or VAT.