Is VAT really neutral?

Do VAT rates affect the competitive behaviour of firms?

When designing a tax system, a key goal is efficiency: the minimisation of distortions which lead to wasted resources. For taxes on purchases, maintaining efficiency in production requires not taxing intermediate input transactions. This has been one of the key motivations behind the recommendation for the use of VAT systems, which avoid distorting production in a perfectly competitive economy. This advice has been taken up by 166 countries in the UN (see OECD’s Consumption Tax Trends 2014).

At the same time, a growing body of evidence suggests that worldwide, economies deviate significantly from the perfectly competitive benchmark and that important distortions in production are prevalent especially in developing countries (Restuccia and Rogerson, 2008; Hsieh and Klenow, 2009; Nishida et al., 2017; Peters, 2020). In particular, there is growing evidence that lack of competition and firms’ market power are an increasingly important source of distortion (see De Loecker et al., 2018). In such settings, it will not necessarily be the case that VAT avoids distorting production. Further, we might want to explicitly design the tax system to counteract firms’ market power and improve productive efficiency.      

With this in mind, a key question is whether VAT rates have an effect on the competitive behaviour of firms, and in particular, firm markups: the ratio between what they charge and the cost of production? Secondly, if so, how should we choose commodity tax rates for systems with imperfect competition?

The efficiency effects of commodity taxes

Our findings suggest that changes in VAT rates affect markups – and productive distortions – in two ways. First, there is a downstream effect: the firm’s markup is affected by its buyers. Second, there is an ‘upstream effect’: the firm’s markup is affected by its suppliers. The downstream effect is such that an increase in the firm’s own price makes their potential buyers more willing to switch to alternatives in the market. Formally, it increases the firm’s elasticity of demand. Given this, when a firm’s buyer pays a tax on their purchases, the firm will charge a lower markup to keep their consumers. The upstream effect is such that an increase in costs from a firm’s suppliers implies an increase in costs of production for a firm. When the costs of production for a firm rise, firms might try to pass through the cost increases to prices. Yet, when doing so, potential buyers become more willing to switch to alternatives, and to avoid that, firms reduce their markups.

In equilibrium, the effects of the VAT can be illustrated in a simple network where firm F1 sells to firm F2, who sells to F3, and so on, all the way till they sell to a wholesaler, who sells to a retailer, who sells to a final consumer. In this network, we have:

1. A downstream effect: if final consumers are more price sensitive as prices rise, then a change in sales taxes makes them more willing to leave high priced firms. This leads firms to reduce the markups they charge and quantities produced. In turn, this change in demand, price and quantities affects input demands by retailers. As a consequence, the wholesaler faces customers (i.e., the retailer) in different circumstances, who are more price sensitive, so more willing to abandon high priced firms. This implies the high-priced wholesalers need to reduce their markups. This keeps moving upwards in the production network, all the way until it reaches firm F1.
2. An upstream effect: in most VAT systems, there are some firms outside the VAT system, typically based on their size but also due to non-compliance. An increased VAT rate on firm F1 translates into higher input prices for firm F2 if F2 is outside the VAT. VAT is implemented by firms who purchase products being able to claim credit for the VAT already paid, but this only applies to firms in the VAT system. In this case F2 cannot claim VAT tax credits as it is outside the system. As a consequence, F2 will have higher costs of production, and will charge higher prices and (typically) lower markups. The higher prices by firm F2 translate into higher input costs for firm F3, and firm F3 will have higher costs. This translates into higher prices charged by firm F3 and typically lower markups by firm F3. This moves downward in the production network all the way until it reaches the consumer.

This intuition can be used to work out when a VAT has no production efficiency consequences. First, if all firms are inside the VAT, the upstream effect is absent. But, as above, in real-world settings there are usually some minimum firm size requirements before VAT registration is required. And informality – non-compliance with the tax regime – is prevalent in low- and middle-income countries.  Second, if the introduction of the VAT does not make consumers more sensitive to firms' prices, then a change in VAT rates won’t change optimal markups by retailers, and the downstream effect disappears. This absent downstream effect is prominent in a lot of work on economic networks, and comes from the assumption of a ‘constant elasticity of substitution’ (CES) demand structure. However, CES demand is inconsistent with much of the evidence on markups and market power of firms, as it would imply that a 1% increase in costs leads to a 1% increase in prices and no changes in markups (see, for instance, De Loecker et al., 2016). By allowing for a more realistic structure for demand and realistic tax systems, our framework is the first framework allowing for a general equilibrium analysis of the distortive effects of taxes in economic networks.

In a very general model, it is hard to say much more than this: in principle, the downstream effect could lead some firms to charge higher markups and some firms to charge lower markups. In particular, retailers outside the VAT would face less competition by retailers inside the VAT, and this can lead them to charge higher markups; similarly, retailers inside the VAT would face more competition by retailers outside the VAT, and be led to charge lower markups. To say more than this, we need to (i) work out what the model says under more restrictive assumptions, to develop extra intuition, (ii) to quantify the model to understand the quantitative relevance of these effects.