vol. 17, no. 2 (February), 1999, pp 3 & 5-8
Do Deficits Matter?
Lawrence H. White* and Roger Garrison**
Creative accounting by the Clinton administration has taken the government's budgetary imbalances out of the media's spotlight. But there is no basis for believing that we are entering a new era of fiscal responsibility. Deficits are likely to dominate future decades just as they dominated the past three.
Lingering issues are still in dispute, even among market-oriented economists, particularly the relationship between government borrowing and interest rates. Steven E. Landsburg and Lauren J. Feinstone, writing in the influential book The Armchair Economist, suggest that deficits do not affect interest rates. The common belief that larger deficits cause interest rates to rise, they say, is "reinforced by two essentially fallacious arguments" that "break down under careful scrutiny." They grant that their critique of the two arguments "doesn't prove that deficits don't affect interest rates," but it does mean that proponents fail to prove their case.
Landsburg and Feinstone refer to the first supposedly fallacious argument as The Goliath Myth: "According to this theory, the country is populated by little 'Davids,' competing against the 'Goliath' of the federal government.... This competition for a limited supply of money drives up interest rates to the point where David can't even afford to finance a slingshot."
They continue: "The analogy is entirely without foundation. Government does not consume money by the act of borrowing it; dollars borrowed by government are immediately available to be borrowed again by individuals.... Goliath consumes no money; he just moves it around a little."
The argument proves too much. By the same logic, neither does the act of taxing "consume money." Just what sort of an act, one wonders, does "consume money"? Putting Federal Reserve notes through a shredder? Taken literally, Landsburg and Feinstone's statement suggests though they presumably did not mean to say this that acts that do not consume money are, in some sense, socially costless.
Government borrowing indeed does not "consume money" in the sense of making the money stock shrink. But it does consume credit, and it does crowd out private borrowing. By way of illustration, consider why Landsburg and Feinstone's conclusion that banks have just as much to lend is wrong.
Suppose that the government sells a newly issued $1 bond to David's bank. David's bank now has an additional $1 of its asset portfolio tied up in government bonds and cannot lend that dollar to David. No other bank has any more to lend. End of story.
The scenario Landsburg and Feinstone describe is a bit more complicated, but it leads to the same conclusion. Suppose that the government sells a $1 bond to Jack, who reduces his bank balance by $1 to pay for the bond. In Landsburg and Feinstone's telling, Jack's part of the story stops here, but in reality it doesn't. If Jack's portfolio mix of bonds and money holdings was previously in equilibrium, he will want to restore the mix by reducing his holding of some other bonds by $1. He won't just passively accept an extra $1 in bonds, and $1 less in money balances, without some increase in the reward for holding bonds relative to money. Ditto for any person to whom Jack sells $1 in bonds.
Interest is not a reward for parting with money as such. Interest is the reward for parting with present goods. To see the essential nature of a government bond sale more clearly, consider a non-monetary economy where Jacks, Jills, Davids, and Goliaths trade present wheat for promises of future wheat. There prevails a market-clearing exchange rate: one bushel of wheat today exchanges for a stream of i bushels per year of future wheat. (We can call i the interest rate.) Now the government decides to shift from acquiring its last bushel of wheat through taxes to buying it with a bond promising future wheat. (Assume that the government gives the present wheat to people who would otherwise consume a bushel less.) To induce wheat-consumers in the market to give up voluntarily an additional bushel of present wheat, the price must rise (all other things equal). The amount of future wheat offered the intertemporal exchange rate or interest rate must rise to increase the reward for waiting to consume.
Landsburg and Feinstone do not deny this, but instead argue that when government borrows, it ipso facto also lends. When government demands present goods, it simultaneously supplies them. In our wheat case, holding total government wheat purchases constant, the decision to issue a bond to finance one bushel in present wheat purchases is implicitly a decision to cut taxes by one bushel today and to raise taxes by i bushels per year in perpetuity (to service the bond), which amounts to lending taxpayers one bushel at interest.
Landsburg and Feinstone label the failure to appreciate this subtlety The Myth of Dick and Jane: "The fallacious argument here runs like this: 'If the government wants to increase its borrowing, it must induce people to lend to it. This means it must offer higher interest rates. Then everyone else must offer a higher interest rate in order to remain competitive.' The mistaken notion underlying this argument is that if Dick wants Jane to lend him a dollar at the prevailing rate of 10 percent, and if she is reluctant to do so, then Dick must offer a higher interest rate to get Jane to change her mind.
"Not so. There is another way to change Jane's mind. Dick can offer to lend Jane a dollar at 10 percent interest, in exchange for her making an identical loan to him. Indeed, Dick can convince her to lend him any amount at all as long as he lends her the same amount, at the same interest rate without producing any upward pressure on that rate.
"This example is not as fanciful as it sounds. Whenever the government wants to borrow a dollar, it simultaneously lends a dollar, just as Dick does. After all, why does the government borrow? It does so to avoid raising your taxes for the time being in effect lending you back the taxes it would ordinarily assess.
"Unlike the borrowing of an individual, government borrowing is always accompanied by an implicit loan to the taxpayers. The government, like Dick, borrows from the public (or Jane), while simultaneously lending the same amount at the same rate. Like Dick and Jane, the government and the public can carry this on at any level without having any effect on the rate of interest."
Economists call this the "Ricardian Equivalence" argument. It asserts that taxpayers recognize that bond-financed deficit spending implies the present-value-equivalent of future taxes. With any increase in government borrowing, taxpayers will increase their savings just enough to negate any effect on interest rates. Taxpayers will purchase any additional amount of government bonds without reluctance at the existing yield, because they recognize that if government is selling more bonds, then the taxpayers need to acquire more bonds in order to afford their (or their heirs') higher future taxes. If the government sells $1 billion in perpetual bonds at 10 percent per annum, then it must increase taxes by $100 million per year. The anticipation of those taxes creates the demand for exactly $1 billion in government bonds paying 10 percent per annum. Landsburg and Feinstone assert the savings imperative forcefully: "Knowing that you are committed to making payments of [$100 million] a year forever, you will be forced to set aside a fund from which to make these payments."
Ricardian Equivalence (with intergenerational bequests that vary one-for-one with the size of the inherited government debt) provides a useful theoretical benchmark. It gives us a logically possible limiting case in which deficits would not affect interest rates. Those who argue that "the government must offer higher interest rates when it borrows more" would thus be uttering a fallacy (as Landsburg and Feinstone say) if by "must" they were to mean that no other result is logically possible. But if, as we believe is the case, they only mean, "must, given the world as it actually is," then the argument is not fallacious. The Landsburg Feinstone critique is misplaced.
Does Ricardian Equivalence actually hold in the world as it is? That's an empirical question, and there is an ongoing debate among economists over which side's evidence is weaker. Perfect Ricardian Equivalence seems unlikely on the face of it. When additional federal debt is issued today, the future taxes to service or retire it will fall largely on future generations. No doubt some taxpayers do rationally plan to leave bequests, do take note each time the national debt grows by a dollar, and can and do calculate just how much more to save and bequeath in order to offset their heirs' likely future tax bills. But it seems implausible that all the taxpayers in the economy act this way (or that the demand for government bonds varies precisely as if they do). Some taxpayers have no heirs; some with heirs leave no bequests. Even a moderate amount of realism suggests that we view the typical bequest as a residual left behind at death, rather than some precisely calculated optimum.
Landsburg and Feinstone seek to minimize the difference between deficits and taxes, holding the level of spending constant. There are two reasons for believing that the difference matters. First, the level of spending may in fact rise with the extent of deficit financing. Taxpayers who do not have heirs, or do not plan future-tax-covering bequests for them, will rightly consider a future tax less burdensome to them personally than a current tax. Because a larger deficit means lower taxes today on all taxpayers, not just on Ricardian bequesters, it shifts some of the burden of current government spending onto future voters who are inadequately represented in today's borrowing decisions. Deficit finance therefore makes a higher level of government spending politically palatable.
Second, when Ricardian academics argue that the choice between borrowing and taxing is a complete toss-up, they have in mind a choice between near-neutral taxes in the present and equivalently near-neutral taxes in the future. But there is a powerful asymmetry that the Ricardian argument fails to consider. Borrowing, almost by its very nature, is unfinished business. We can't make a fair comparison between the two methods of finance until we hear "the rest of the story."
How will the government accommodate its deficit? Will it borrow from the domestic public? Will it borrow from citizens of other countries? Will it borrow from its own central bank? Our not knowing just how the government's borrowing will impinge on the private sector (through interest rates? exchange rates? the inflation rate?) can add heavy doses of uncertainty to credit markets, securities markets, and capital markets.
Once the Federal Reserve and its discretionary ability to monetize debt are brought into consideration, we can see that government borrowing can create distortions that may well exceed those associated with ordinary taxation. Will the government service and roll over the debt indefinitely, or begin to pay it back at some date, and if so when? Exactly which tax rates will it raise, by what degree, at what dates? What new taxes will it impose, and when? If we don't know the answers to all these questions, we can't be secure in our economic planning. A larger budget deficit adds greater uncertainty to all plans that might be affected by future taxes and the interim flows of funds.
The Ricardian case for indifference between current taxes and borrowing would therefore apply at most only to the case where all future tax codes are as precisely specified today as the current tax code is. When government borrowing is expanded, the "repayment plan" for the "implicit loan to the taxpayers" would have to be made explicit, so that informed private plans could be made around it. And the specification of all future tax codes would have to include the rates of taxation on base money to be imposed by the Federal Reserve via monetary expansion. In a nutshell, the key contrast between taxing and borrowing is the contrast between the existent federal tax code and the nonexistent federal deficit code.
Landsburg and Feinstone are right that the size of the public sector is
the primary concern. Accordingly, all government spending might be best
financed by a tax payable in cash the day before each national election.
Such a tax would certainly be onerous for both the taxpayers and the politicians.
Rather than making the burden obvious, politicians prefer to borrow for
now and to leave us all to guess about just what will happen after election
day. If we can understand why politicians are not indifferent to the method
of financing government spending, we can also understand why the rest of
us shouldn't be either.
* Lawrence H. White is professor of economics at the University of Georgia.
**Roger W. Garrison is professor of economics at Auburn University.