Domestic Currency Denominated Government Debt as Equity in the Primary Surplus

“There are many dollar-denominated intertemporal budget constraints in the world, including those of, say, Orange County California, IBM…and (hypothetically) a dollarized Mexico. Isn’t it true that any one of these intertemporal budget constraints ‘determines’ the price level the same way that of the US federal government does?” (Section I, p. 1). Sims calls this reaction “itself fallacious” but worth engaging carefully, both as a defense of the fiscal theory of the price level (FTPL) and because the answer illuminates real policy questions – indexed government debt, dollarization, and the fiscal foundations of the lender-of-last-resort function.

“The important difference between the US government and me is not our relative size or the fact that it is hard to sue the US government and not hard to sue me. The important difference is that when I issue ‘dollar-denominated’ government debt I am making a promise to pay in units of a commodity (dollars) that I have a limited capacity to produce or obtain in the market, while the government is promising only to pay new paper that it issues at nearly zero cost” (Section II, pp. 2-3). Because a private borrower’s future capacity to pay is finite, creditors can see when existing obligations exhaust it and simply refuse further lending; a government facing the same news does not lose the ability to borrow, because there is no analogous “commodity” constraint – “the price level rises to make the real value of the future primary surpluses match the real value of the outstanding debt.”

First, the “too big to fail” case: if the government is committed to maintaining the value of a systemically important firm’s liabilities, “the company…in effect gains the ability to create and sell government liabilities,” so that if the government lacks the fiscal capacity to offset the drain, “the company may indeed control the price level” (Section II, p. 3). Second, a hypothetical firm that pays no dividends, instead running a fluctuating buyback program funded from profits (with periodic stock splits to prevent the share count from shrinking): such a security “would have positive market value,” equal to the present value of expected future buybacks – a firm-level analogue of a government financing itself with fixed-nominal-rate bonds while running a positive, fluctuating primary surplus, whose “price level,” in units of the firm’s own stock, “will rise over time” if the fixed rate exceeds real returns elsewhere (Section II, pp. 3-4).

“It is…privately issued equity, not privately issued dollar debt, that is the closest analogue to publicly issued dollar-denominated debt” (Section II, p. 4), because both fiat government debt and equity “make no promise to pay in units other than the security itself, and…therefore have value determined by expectations of the future resources that will be devoted to backing the security.” Sims is careful to note the analogy’s limits: unlike a firm’s shareholders, a government does not typically have managers single-mindedly maximizing a fiscal “surplus” for personal benefit over a secure, long horizon – a condition he says “seems plausible perhaps only for the case of certain types of colonial governments or stable personal dictatorships,” useful nonetheless as a simplifying starting point (Section III, pp. 4-5).

Just as a firm facing rising bankruptcy risk from too much debt may go public to gain a larger capital cushion, Sims argues nominal debt functions for governments as “absorber of fiscal risk”: “the capacity of a government to borrow abroad in dollars could be enhanced by the presence of substantial outstanding nominal…debt,” because unpredictable shocks to revenue and expenditure (crop failures, disasters, wars, mineral discoveries) can be absorbed in the fluctuating value of nominal debt “with no inefficient adjustment of fiscal parameters,” rather than forcing costly changes to tax rates or spending if all debt were dollar- or foreign-currency-denominated (Section III, pp. 5-6). Sims summarizes: “stability of the real value of dollar debt is guaranteed by the instability of the value of the nominal debt.”

Sims connects his point to “two earlier strands of research in public finance”: Barro’s argument for tax-smoothing (avoiding fluctuations in tax rates), and Judd’s emphasis that “unanticipated capital taxation is not distorting,” which implies fiscal shocks should ideally be absorbed through the rate of capital taxation if that were administratively feasible (Section III, p. 6). “The implicit tax on holdings of government debt that arises from unexpected shocks to fiscal balance…is automatic, with no administrative costs, and it is non-distorting for the same reason that unanticipated capital taxation is non-distorting” – so nominal debt lets a government achieve, at essentially zero administrative cost, the same non-distorting risk-absorption that direct unanticipated capital taxation would provide if it were practical.

Extending Barro’s tax-smoothing model to allow for a stochastic primary surplus and endogenous surprise inflation, Sims shows that if fiscal policy commits to a fixed tax-to-output ratio tau/Y while monetary policy holds the nominal interest rate fixed, “the price level will adjust to maintain balance between the real value of outstanding debt and the discounted present value of future tau - G,” letting tau/Y remain “absolutely fixed despite stochastic variation in G” (Section IV, pp. 7-8). Working through a two-state Markov model of government spending G, Sims derives the minimum tax rate consistent with a stationary equilibrium with positive debt, and shows that “setting tau so that debt retains some real value, then using the debt to absorb the surprises in G, produces a better outcome in this model than does Barro’s original prescription of keeping tau_t equal to the discounted present value of real primary surpluses while prices remain constant” – because Barro’s constant-price prescription implies “an instantaneous complete devaluation of outstanding debt via an infinite price level” the moment tax revenue is set equal to expenditure once and for all, which is suboptimal once future shocks to G are uncertain.

Sims computes the “unanticipated real return” to holders of U.S. government debt for 1950-89 as the realized change in the market value of debt minus expected debt service (using a Bayesian VAR on the commercial paper rate, CPI, and industrial production to estimate anticipated inflation), following a method similar to Hall and Sargent (1997) (Section V, pp. 9-10). He finds that “between 1973 and 1980 all but two years produced negative unanticipated returns, as would be expected of an optimizing government offsetting the negative fiscal shocks of the oil crises,” while “the 80’s…are dominated by positive unanticipated returns, as inflation is brought under control again.” The magnitudes – “on the order of $40 billion as the maximum annual capital loss in the 70’s” – are, in his words, “neither huge nor trivial,” and their timing pattern (losses following the oil shocks specifically) suggests the fluctuations reflect purposeful risk absorption rather than pure policy erraticism, though Sims flags that “further research is needed to check this more carefully.”

Sims frames the entire analysis as bearing directly on two live policy debates: arguments in the U.S. “for increased use of indexed government debt,” and arguments in other countries “for abandonment of the domestic currency in favor of the dollar, which would result in all government debt being dollar-denominated” (Section I, p. 1). Because nominal debt is what lets a government absorb unpredictable fiscal shocks non-distortingly, “a primary cost of eliminating fiat-denominated government debt is that thereby a route by which the fiscal system can efficiently share risk with holders of the debt is eliminated” – his empirical estimates, though rough, are meant to establish that this route has been used to a “neither huge nor trivial” extent by the actual postwar U.S. fiscal system, so the efficiency loss from foreclosing it via indexation or dollarization is a real, not merely theoretical, cost to weigh against whatever benefits those proposals offer.

The manuscript’s final section is headed “VI. Conclusion” but contains no text before the references – consistent with its status as a conference paper (prepared for the August 1999 Latin American meetings of the Econometric Society in Cancun, this version dated December 1, 1999) rather than a fully polished journal article. The paper’s substantive argument is nonetheless complete by the end of Section V: the equity analogy (Sections II-III), the formal shock-absorber model (Section IV), and the empirical estimates of realized fiscal risk-sharing (Section V) together make the case, without a separate summary conclusion appended.