Public Finance — Public Expenditure: Redistribution, In-Kind Benefits and Public Choice

Public Finance — Spending, Like Taxes, Changes Behaviour

The public goods and externalities of Chapter 1 were the efficiency grounds for government spending. In real budgets, the largest share goes to redistributive spending such as social security and welfare. This chapter looks at how the design of redistributive spending changes recipients’ choices, and how the size of spending is set in the political process.

1. Cash or in kind?

For the same amount, a recipient’s utility from cash is at least as great as from an in-kind benefit, because cash can be spent on anything while an in-kind benefit must be used on that good.

Suppose a household with monthly income of 1 million won spends 200,000 won on food. If the government gives a food voucher worth 300,000 won, the budget line shifts right in parallel up to 300,000 won of food and beyond that is the same as 300,000 won in cash. For a household that would spend at least 300,000 won on food if given 300,000 won in cash, the voucher and cash make no difference. For a household that, given cash, would spend only 250,000 won on food and the rest on rent, the voucher forces it to consume 50,000 won more food than it wants, and its utility is lower than with cash.

In-kind benefits are nonetheless common, for three reasons.

• Paternalistic preferences: taxpayers willingly pay for “money that goes to children’s school meals” but are less willing to fund cash that can be used freely.
• Self-selection: in-kind benefits of lower quality or that require queuing, such as public rental housing or free meals, are applied for only by those who really need them. They become a targeting tool when income is hard to verify.
• Price effects: bulk purchasing can lower unit costs. Conversely, giving vouchers in a market with inelastic supply (rental housing) raises prices, and the benefit leaks to suppliers.

2. Top-up benefits tax earnings at 100%

A top-up benefit sets a minimum living standard and fills the gap between it and income. If the standard is 1 million won a month, a household with no earnings receives 1 million won and one earning 400,000 won receives 600,000 won. Both end up with the same income of 1 million won.

Each won earned reduces the benefit by one won, so below the standard the effective marginal tax rate is 100%. Working does not raise income, so the incentive to work near the standard disappears. The marginal rate can be lowered by withdrawing the benefit only partially as income rises, but then the income range over which benefits are paid widens and costs rise.

Of the guarantee level $G$, the withdrawal rate $\tau$ and the fiscal cost, only two can be chosen. Raising the guarantee and lowering the withdrawal rate widens the eligible range $G/\tau$, and costs balloon.

The earned income tax credit (EITC) approaches from the opposite direction. It has a phase-in range where the credit rises with earnings, a flat plateau and a phase-out range where it declines. In the phase-in range people receive more the more they work, which raises labour force participation. In the phase-out range the effective marginal rate rises by the withdrawal rate, which can reduce the hours of those already working. Empirical studies generally report a large participation effect and a small effect on hours.

3. The size of spending is decided by voting

The Samuelson condition of Chapter 1 tells us the efficient quantity of a public good, but government does not know individuals’ marginal benefits. Actual quantities are set by voting and budget negotiations. The field that analyses this process with economic tools is public choice.

The median voter theorem: if the choice is one-dimensional (say, the size of a library budget) and every voter’s preferences are single-peaked (the further from the favourite level, the less liked), majority voting yields the median voter’s preferred level.

Suppose three residents want library budgets of 1 billion, 3 billion and 8 billion won. Against 1 billion, 3 billion wins the votes of those who prefer 3 and 8 billion; against 8 billion, it wins the votes of those who prefer 1 and 3 billion. The outcome is 3 billion won. The average preference is 4 billion won, but majority voting picks the median, not the mean. Because intensity of preference is not counted, there is no reason for the median level to coincide with the efficient level of the Samuelson condition.

If preferences are not single-peaked, majority voting can cycle (the Condorcet paradox): A beats B, B beats C and C beats A. The outcome is then decided by the order in which motions are put — that is, by whoever controls the agenda.

4. Incentives inside government

Rent seeking is spending resources to capture the excess profits created by regulation or subsidies. If an import licence yields 1 billion won of excess profit a year, firms have an incentive to spend close to 1 billion won on lobbying, lawyers and advertising to obtain it. That spending produces nothing, so the social cost of the regulation is the monopoly deadweight triangle plus a large part of the excess-profit rectangle (Tullock).

Check your understanding

Under a linear benefit with a guarantee of $G=1.2$ million won and a withdrawal rate of $\tau=0.4$, what are the benefit and final income of a household earning 1.5 million won? $B=1.2-0.4\times 1.5=0.6$ million won, so final income is 2.1 million won. Benefits end at an income of $1.2/0.4=3$ million won. Raising the withdrawal rate to 0.6 moves the end point down to 2 million won and cuts costs, but households in that range face an effective marginal rate of 60%.

References

• Jonathan Gruber, Public Finance and Public Policy, ch. 9, 17
• Harvey Rosen and Ted Gayer, Public Finance, ch. 6, 13
• William Niskanen, Bureaucracy and Representative Government (1971)
• Gordon Tullock, “The Welfare Costs of Tariffs, Monopolies, and Theft,” Western Economic Journal (1967)