When Does Introducing a Value-Added Tax Increase Economic Efficiency? Evidence from the Synthetic Control Method

The theoretical prediction that a value-added tax (VAT) does not distort firms' production decisions has led to its rapid adoption worldwide, but there is surprisingly little empirical evidence. This paper provides one of the first causal estimates of the efficiency gains (i.e., an increase in GDP per worker) of introducing a VAT in a worldwide sample of countries using the synthetic control method. The synthetic control is a weighted average of countries without a VAT that closely resembles the economic structure and outcomes of the country with a VAT for several years before the adoption of a VAT. In line with previous studies, I find that the VAT has, on average, a positive and economically meaningful impact on economic efficiency. However, this result is driven by richer countries only. There is no significant impact of the VAT on poorer countries. I find similar results when estimating the impact of the VAT on total factor productivity and capital stock per worker, two important channels through which a VAT affects GDP per worker. This paper provides evidence that a success of VAT almost entirely depends on the initial level of income of a country, which, in turn, determines whether a country is able to properly design and enforce a VAT. The findings are robust across a series of placebo studies and sensitivity checks.