Public Finance — Value Added Tax and Indirect Taxes

Public Finance — VAT Is Collected at Every Stage but Leaves Its Burden Only on Consumption

If the income tax is levied on those who earn, indirect taxes are levied on those who spend. This chapter first looks at the structure of the largest indirect tax, value added tax, and then turns to the theory of optimal taxation: which goods should be taxed and at what rates.

1. The input tax credit prevents cascading

Suppose a raw-material supplier sells materials to a manufacturer for 100 won, the manufacturer sells the product to a retailer for 300 won, and the retailer sells it to a consumer for 500 won. The tax rate is 10%.

Each business pays its output tax minus the tax it paid on purchases. The total of 50 won paid equals 10% of the final sale price of 500 won. Businesses merely pass on the tax they collect; the burden rests with the final consumer. Because the credit requires an invoice showing the tax paid at the previous stage, businesses end up checking each other’s returns — a self-enforcing mechanism.

A retail sales tax that raises the same revenue only at the retail stage lacks this cross-check, so the incentive to evade grows as the rate rises. Conversely, a turnover tax that taxes every transaction without credits creates a cascading effect in which tax piles up on goods with many stages, distorting firms towards vertical integration.

2. Zero rating and exemption are completely different

Zero rating applies a 0% rate to output tax and refunds input tax. All VAT accumulated up to that stage disappears. The typical use is for exports, so that goods are taxed only in the country of consumption.

Exemption takes the business out of the VAT system. It collects no output tax but cannot recover input tax either. In the table above, if the manufacturer were exempt, it could not recover the 10 won paid on purchases and would add it to its costs. The retailer would receive no invoice from the manufacturer and so would have no input credit. Selling at 500 won, the retailer collects 50 won and, with no credit, pays all 50 won; total revenue becomes 60 won — 10 won at the raw-material stage plus 50 won at retail. Exempting an intermediate stage actually increased revenue. The effect of exemption is clean only at the final stage; exemption at an intermediate stage revives cascading.

3. Is VAT regressive?

Measured against annual income, yes. If a household with income of 30 million won spends it all and one with 100 million won spends 60 million, a 10% VAT amounts to 10% and 6% of income respectively.

On a lifetime basis, however, the conclusion weakens. The 40 million won saved by the high-income household will eventually be consumed, or inherited and then consumed. Consumption relative to lifetime income is much flatter than on an annual basis. Also, a single year’s income has large transitory swings, and some households with low annual income are retirees or households with a temporary dip in income.

Exempting food and similar items or giving them reduced rates is an ineffective way to ease regressivity. High-income households also buy food, and more expensive food, so in absolute terms more of the benefit goes to them. The standard conclusion is that collecting VAT broadly and redistributing through targeted transfers helps low-income households more for the same cost.

4. Which goods to tax, and by how much

The Ramsey rule of Chapter 2 called for higher rates on goods with inelastic demand. Two further results complement it.

The Corlett-Hague rule: labour (or leisure) cannot be taxed directly, so taxing goods complementary to leisure (golf, travel) at higher rates taxes leisure indirectly and reduces the distortion to labour supply.

The Atkinson-Stiglitz theorem: if the nonlinear income tax can be designed optimally and consumption of goods is separable from leisure in utility, there is no need for differentiated rates across goods. It is optimal to leave redistribution to the income tax and keep indirect taxes uniform. Real preferences are not exactly separable, but the theorem is a strong argument against using multi-rate VAT as a redistributive tool.

The Diamond-Mirrlees production efficiency theorem: in an optimal tax system, intermediate goods should not be taxed, because distortions at the production stage reduce output without benefiting anyone. The input tax credit of VAT is this principle put into practice.

5. Corrective taxes: excise duties and carbon taxes

Excise duties on tobacco, alcohol and fuel are most convincingly explained by their corrective purpose rather than revenue — the Pigouvian logic of Chapter 1. The right rate is the marginal external cost, not the demand elasticity.

Carbon taxes are sometimes defended with the double dividend hypothesis. The first dividend is lower greenhouse gas emissions; the second is the efficiency gain from using carbon tax revenue to cut distortionary taxes such as the income tax. But the carbon tax itself lowers real wages and distorts labour supply (the tax interaction effect), so the second dividend is generally not guaranteed. The majority view is that carbon taxes must be justified by the first dividend alone.

Check your understanding

In the table above, what would total revenue be if the final retail stage were exempt? The raw-material supplier pays 10 won and the manufacturer 20 won, and the retailer neither collects nor recovers anything. Revenue is 30 won, and the retailer absorbs the 30 won of input tax as a cost and passes it into the price. Only the 200 won of value added at the retail stage escapes tax.

References

• Jonathan Gruber, Public Finance and Public Policy, ch. 20, 25
• Peter Diamond and James Mirrlees, “Optimal Taxation and Public Production,” American Economic Review (1971)
• Anthony Atkinson and Joseph Stiglitz, “The Design of Tax Structure: Direct versus Indirect Taxation,” Journal of Public Economics (1976)
• Institute for Fiscal Studies, Tax by Design: The Mirrlees Review (2011), ch. 6–9