Understanding tax distortions and Ricardian equivalence
Krugman is writing today in his column: “Federal debt isn’t at all like a mortgage on America’s future doesn’t mean that the debt is harmless. Taxes must be levied to pay the interest, and you don’t have to be a right-wing ideologue to concede that taxes impose some cost on the economy, if nothing else by causing a diversion of resources away from productive activities into tax avoidance and evasion.”
I can only agree with this statement as long as the government is using distortionary taxes and there is no obvious reason to do so. It’s about the breakdown of Ricardian equivalence. Under distortionary taxation, debt and taxes are no longer equivalent. Barro (1979) showed that public debt should then be used to smooth out the distortions of taxation over time.
However, distortionary taxes are levied for a reason: income redistribution. If governments would not be interested in redistribution of income at all, they would levy non-distortionary non-individualized lump-sum taxes instead. But for lump-sum taxes we know that Ricardian equivalence can hold.
Most standard macro-economic analyses, however, leave distributional concerns out of the picture and assume a representative agent, so that there are no distributional concerns to begin with. Why, then, use distortionary taxes if everyone is equal? In representative-agent models, all representative agents would surely prefer the government to levy lump-sum taxes on them, rather than distortionary taxes. Since everyone is identical, there is no distributional gain of distortionary taxation, but only efficiency costs of doing so. And all representative agents would agree on this. Thus, most papers only consider the efficiency cost of taxation, but not its associated distributional benefits.
Allowing for heterogeneous agents to motivate distortionary taxes has dramatic consequences for Barro’s (1979) normative theory of optimal debt management. In a recent paper of mine (Jacobs, 2012), I argue that – if the government sets taxes optimally – the distributional gains of taxation cancel against the dead weight losses of taxation. As a direct consequence, Ricardian equivalence would be restored – even in a world with distortionary taxes – while all other assumptions to obtain Ricardian equivalence are assumed to hold. The reason is, well, as long as taxes are optimally set, all distortions from taxation are exactly compensated by distributional gains.
On a net basis, there are no inter-temporal distortions caused by taxation, which would normally result in a breakdown of Ricardian equivalence. Of course, there are still intra-temporal distortions associated with taxation as individuals’ labor supplies will be distorted downwards.
Hence, at the optimal tax system the government is indifferent between using debt, distortionary taxes and lump-sum taxes to finance its outlays, whether these are on public goods or interest payments. As Ricardian equivalence is restored, debt and taxes are equivalent and the optimal time-path of public debt becomes indeterminate. The level of distortionary taxation at each moment in time is only dependent on the desire to redistribute income at that moment in time.
These conclusions were also reached by Ivan Werning (2007), and in my paper I demonstrate that the same logic extends to policy rules for public goods provision, corrective taxation, regulation, and so on.
Of course one can debate if governments set taxes optimally in the real world. They probably do not. But this criticism would be equally applicable to Barro (1979), which is also an optimal-tax analysis (in a Ramsey setting). I still think that the argument is valid to an approximation by invoking Becker’s efficient redistribution hypothesis. Political competition in elections may ensure that there are no unexploited gains for income redistribution.
The point I am trying to make is that the optimal time-path of public debt is not determined by tax distortions associated with the financing of the debt; in models with heterogenous agents Ricardian equivalence is restored. That does not imply that by allowing for the usual other arguments (liquidity constraints, finite lives, and so on), Ricardian equivalence may break down.
Krugman concludes: “But these costs [the costs of tax distortions – BJ] are a lot less dramatic than the analogy with an overindebted family might suggest.” Indeed, if the government would optimize the tax system, these costs would be zero. On a fundamental level, tax distortions cannot motivate why the government needs to use public debt to smooth the tax burden over time.
Barro, Robert J. (1979), “On the Determination of the Public Debt”, Journal of Political Economy, 87, (5), 940–971.
Jacobs, Bas (2012), “A Contribution to the Theory of the Marginal Cost of Public Funds”, Erasmus University Rotterdam.
Werning, Ivan (2007), “Optimal Fiscal Policy with Redistribution” Quarterly Journal of Economics, 122, (3):925–967.