A Positive Theory of Fiscal Deficits and Government Debt

Why do so many democracies run deficits year after year, even when the economy is calm and there’s no obvious shock to pay for? Here’s the unsettling answer Alesina and Tabellini propose: debt isn't just a budget outcome, it’s a weapon. When two parties disagree about what government should buy, the party in power can use debt today to box in its rival tomorrow. You don’t just hand over the keys after an election; you leave the tank half empty. Their model makes that idea precise with a simple but sharp setup. Think of an economy with two public goods, which we’ll call g and f. Both parties value both goods, but not equally. That disagreement is summarized by a parameter, a, that tilts preferences toward one good or the other, and a second parameter, l, that sets the baseline elasticity of demand for public spending. They focus on cases where l is greater than a and a is positive—so the parties are genuinely different, but not from separate planets. Taxes are distortionary, meaning raising an extra dollar of revenue hurts somewhere else in the economy. And there’s a realistic twist: sometimes government can't cut certain services below a minimum. That downward rigidity—think basic policing or schools that can’t shrink overnight—turns out to matter because minimums make tomorrow’s choices less flexible. Under the hood, the problem is dynamic. The state variable is the debt stock you inherit at the start of your term, b today. Your controls are what you’d expect: set the tax rate, split your budget between g and f, and decide how much debt to leave for the next period, b tomorrow. The analysis cleanly separates the "within-period" allocation—given today’s debt, pick taxes and the two goods to maximize your party’s objective—from the "across-period" choice, which is how hard you lean on debt to shift resources into today from the future. Two structural facts organize the politics. First, inherited debt hurts the incumbent and, perversely, helps the opposition party. If you start your term with a heavy interest bill, you have to raise taxes or cut your favorite programs right away. From the other side’s perspective, every extra dollar of your starting debt limits what you can do, which they like. Second, the intertemporal choice of debt obeys one crisp condition. Set the marginal value of postponing another unit of taxation—the immediate political and utility gain from borrowing one more dollar today—equal to the discounted expected marginal cost of carrying that extra dollar into tomorrow, when someone (maybe you, maybe them) must pay for it. In their notation, that’s marginal value equals marginal cost. The marginal value slopes downward because the within-period payoff is concave: the more you already borrowed, the less extra benefit you gain from another dollar. Marginal cost typically slopes up: each extra unit of debt ratchets up tomorrow’s interest bill and the pain that follows. Where those two meet determines how much debt you leave at the end of the term. Now zoom in on the simplest version: two periods. In the last period, games are over. Whoever wins must raise whatever taxes are necessary and cannot push debt beyond the horizon, so end-of-period debt is zero. That last fact does two things. It makes the current interest rate independent of who will be in office later, and it freezes the mapping from end-of-period debt to payoffs. So, in period zero, the incumbent chooses b for next period by solving that marginal value equals marginal cost condition, with one key comparative static: if the probability you’ll be re-elected—let’s call it P—goes up, the optimal b goes down. Intuitively, a higher P means you internalize more of the future tax distortion and the public goods you’ll be forced to cut. The cost curve effectively shifts up, so the intersection with marginal value moves left. That single line—b falls as P rises—does a lot of work. It explains why politicians with tenuous prospects have the strongest temptation to overspend now and send the bill forward. It also sets up the central contrast with a world without elections. A social planner—one decision-maker choosing policy for the whole society—would balance the budget every period and carry zero debt in steady state. The planner’s objective is a weighted average of citizens’ welfare, and the optimal composition of g and f equalizes their social marginal utilities. If you raise the share of g by a little, the social benefit you get from that is exactly matched by the social benefit from a marginal increase in f, so any further tilt would be wasteful. With distortionary taxes in the background, the planner also sees that extra steady-state debt just raises the interest bill that must be financed forever. More debt means higher taxes or fewer public goods tomorrow. So zero is the sweet spot. Politics breaks that equivalence. With turnover and disagreement, debt becomes strategic. Alesina and Tabellini’s steady-state analysis shows that, under a mild technical condition in their appendix, a positive level of debt can be locally stable in the political equilibrium. You can push on the system, and it returns to a neighborhood around some positive P. And you can sense the logic: when sides disagree about the composition of spending, the party in office uses debt to get more of what it likes now and to leave less room for its rival to remodel the budget later. That pressure is stronger when the minimum-provision constraint bites, because cutting your rival’s favorite program tomorrow might be literally off-limits. Before we go long-run, one symmetry result from the short-run is too useful to ignore. In the baseline, near the steady state and with mirror-image parties, taxes and the overall deficit today look the same no matter who holds office; only the mix between g and f differs. That means voters’ expectations about future taxes don’t fluctuate with election polls, which is why the value functions that summarize each party’s incentives are well-behaved in that neighborhood. The action is in composition and in debt, not in the size of government per se. Push the horizon to infinity and the politics settle into a stationary rhythm. Each government conditions only on the inherited debt stock when choosing taxes, the mix between g and f, and the amount of new debt. Reputation games are deliberately left out. In this Markov world, the same marginal value equals marginal cost logic characterizes the intertemporal choice: tilt spending into today until the marginal value of borrowing matches the expected marginal cost of servicing a slightly bigger debt stock tomorrow. The standout comparative static carries over intact. In a neighborhood of the steady state, the higher your party’s chance of being reappointed, the lower the steady-state debt you are willing to sustain. Chen and colleagues would call this "internalization of the shadow of the future," but you don’t need the label to get the point. If you expect to be the one stuck with higher interest payments, you borrow less. Two more pieces round out the infinite-horizon picture. First, there exists a rational political equilibrium that ties debt choices to election probabilities in a coherent way. Election probabilities themselves flow from voter preferences—the distribution of median-voter ideals over g versus f determines how likely each party is to win—and those probabilities feed back into how much each side borrows. Second, polarization magnifies everything. As the gap between l and a widens—meaning the parties care more differently about the two goods—the fiscal policies they pick move farther from the planner’s balanced benchmark. The deficit bias grows. Layer in those minimum-provision constraints and debt gets more entrenched. If you’re forced to keep both g and f above some floor, tomorrow’s government has a smaller set of dials to turn, so the pain of today’s borrowing is harder to shift away from your rival’s priorities, and the political motive to preempt grows. There’s a nice echo here with work by Persson and Svensson from the 1980s. They looked at a related setup and found the same instinct: if rivals disagree over what to spend on, incumbents who expect to lose have an incentive to leave deficits; if they expect to win, the incentive weakens. Different model, same human move. What about the intraperiod allocation—the choice of g versus f when you’re actually in office? When the minimum constraints don’t bind, each party equalizes the marginal utility per dollar across the two goods within its own objective. If constraints do bind, that equalization breaks: one good is stuck at its floor, the other adjusts, and the shadow value of relaxing the constraint shows up in today’s first-order condition. That mechanical change shifts the marginal value and marginal cost balance too. The result can be more or less current borrowing depending on how tight the floor is, but the direction of the strategic force—use debt to shape your successor’s options—doesn’t change. Put the pieces together and the political economy points to three concrete things. One, compared to the planner, competitive democracies with disagreement over the composition of spending exhibit a persistent deficit bias. Not because politicians are short-sighted in general, but because they’re rational in a world without binding commitments. Two, that bias is state-dependent. It’s stronger when reelection chances are low, when polarization is high, and when minimum spending floors limit tomorrow’s flexibility. Three, there is no convergence to a single, voter-pleasing policy. The equilibrium is time-consistent—nobody wants to change their mind once in office—but it involves systematic deficits relative to the planner and persistent fights over composition rather than size. If you’re wondering how this aligns with the observation that some countries keep debt in check for decades, the model gives you levers to consider. Countries with low polarization, with institutions that make it easier to adjust spending, or with political environments where incumbents are confident of reappointment should, all else equal, carry less debt. The United States—polarized, with lots of downward rigidity in big programs—lands near the opposite corner. That’s not a proof, but it’s a testable pattern the paper points to. There are caveats. The framework is two-party, voters are rational, and parties can’t commit. Real-world politics features coalitions, renegotiation, and reputations. And even in this model, steady-state stories differ depending on parameters: the planner sits at zero debt; the political steady state can support positive and locally stable debt. One technical way to see the welfare cost is to look at what higher steady-state debt does inside the model. As debt climbs, the interest bill rises; financing it forces higher taxes or smaller g and f. In other words, more debt crowds out the very public goods the parties are fighting over. Step back, and the intellectual contribution is clear. Debt isn't just smoothing shocks or financing wars. It's also a strategic commitment device in a recurring contest over what government should do. Once you see it that way, the persistence of deficits in calm times looks less like a policy error and more like the equilibrium of a game with the rules we’ve set. If you want a different outcome, you have to change the rules—raise the probability of reelection, reduce polarization, soften rigidities—or give the planner back some of the commitment technologies politics took away. That’s a conversation for institutions and reforms. The paper’s gift is to show, with disarming clarity, why the status quo looks the way it does.