What About Fiscal Policy?

Congratulations to the Central Bank of Chile on its 100th anniversary. It is a remarkable achievement, and an impressive journey. Listening to many of you today, I was reminded of something simple but important: it takes a long time to build institutions and get policy frameworks right. Chile has worked on this for decades. You have gotten many things right, and that matters for what I want to discuss.

I should also warn you from the start: this will be a bit of a “black sheep” talk relative to the papers you have been hearing. Most of the presentations today were quite granular and technical. Mine will be the opposite. It will be “30,000 feet.” It will have no equations, no graphs—only words. At this time of day, perhaps that is not a bad thing.

And finally, it is not going to be about monetary policy. It is going to be about fiscal policy.

The reason I want to do this is that I think the macroeconomics profession—and policy institutions—have done a gigantic amount of work on monetary stabilization over the last decades. We have developed better models, better rules, better institutional designs. We have learned how to conduct monetary policy more systematically. I do not want to exaggerate; we still make mistakes, and new challenges always arise. But I think there is a real sense in which monetary policy has progressed enormously.

Fiscal policy, by contrast, has not progressed as much. Or at least, it has not progressed in the same way. When we talk about stabilization today, the conversation is almost always about monetary policy. Fiscal policy appears in the background, often as an afterthought, or as something that becomes relevant only in emergencies. That is a striking historical reversal. If you go back to many decades of postwar macroeconomics, fiscal stabilization was central. Fiscal policy was the main instrument discussed. It was viewed as powerful, flexible, and effective. And now we have shifted almost completely to the other side. Fiscal policy, in many countries, is treated primarily as a constraint, or as a problem to be managed, rather than as a stabilization tool.

I am not entirely sure why the shift happened. One reason is surely political economy: fiscal policy is inherently political, while monetary policy can be delegated to independent central banks. Another reason is that fiscal policy is slow to implement and often asymmetric. It tends to expand in bad times but does not contract symmetrically in good times, leading to deficit bias and debt accumulation. Another reason is that economists and policymakers have become wary of the risks of public debt and the possibility of fiscal dominance. But whatever the combination of reasons, I think the result is that we have not invested enough intellectual and institutional effort in designing fiscal policy for stabilization in a disciplined, credible way.

When you invited me, you asked what I would speak about. I have written many papers over time on fiscal policy, on stabilization, on debt, on fiscal rules. At first, I thought I should try to put these ideas together, identify where there is agreement and disagreement, and discuss it at the level of policymakers rather than academic debates. That seemed like what I was invited to do. But as I began organizing my thoughts, I realized something: Chile is not a country where I can easily come and deliver a dramatic lecture about fiscal irresponsibility or institutional failure. Chile is actually a country that has built a very impressive fiscal framework, carefully and gradually, over decades.

So, I found myself in the position of “bringing coal to Newcastle.” Many of the issues that are worrisome elsewhere are not worrisome here. The trajectory of institutional development in Chile, from the mid-1980s to today, is extremely impressive. Of course, it would be more entertaining to come to a country where fiscal policy is obviously broken and tell them what they should do differently. That is more fun, even if it is somewhat harsh. But that is not the situation in Chile.

So, I decided to focus on three themes—three issues—where I think there is still something useful to discuss even in a country that is doing many things right.

1. First, how we can design better fiscal stabilization mechanisms—what I will call quasi-automatic stabilizers—to improve the tradeoff between stabilization and political economy constraints.
2. Second, how dangerous public debt really is. Debt is often treated as obviously bad, and the policy discussion sometimes starts and ends there. I think it is important to be more precise about what exactly makes debt costly or dangerous, because that determines how much effort should go into limiting it.
3. Third, if a country wants a modern set of fiscal rules, how one should think about their design using stochastic debt sustainability analysis, or SDSA. This is a practical framework for thinking about risks under uncertainty, and about how to build rules that are credible but not mechanical.

These topics are linked, but they are not one single topic. So let me take them in turn.

1. Quasi-automatic stabilizers: the promise and the problem of fiscal policy

On paper, fiscal policy looks like a very attractive stabilization instrument. It has many tools. It can change taxes, transfers, public consumption, public investment. It can affect disposable income directly. It can target specific groups. It can influence incentives at different margins. It can affect consumption and labor supply in the short run, and it can affect saving and capital accumulation over time. If you look at macroeconomic models, especially models with multiple distortions and shocks, fiscal policy potentially has many instruments that could improve welfare. And yet, despite this theoretical attractiveness, in practice fiscal policy is often treated as something we should use only cautiously and rarely, at least for stabilization.

Why?

The first reason is implementation. Fiscal measures take time. They require democratic choice, political negotiation, legislation, and administrative execution. Even when there is agreement, the process is slow. And slow stabilization is not stabilization: if the recession ends before the stimulus arrives, you have simply created debt without smoothing the cycle.

The second reason is political economy. If you give governments the ability to adjust taxes and spending for stabilization, they may not use that power responsibly. They may run expansionary fiscal policy in recessions—as they should—but they may not reverse it in booms. They may use it for electoral purposes. They may avoid unpopular fiscal consolidation. This creates a systematic bias toward deficits and debt accumulation.

In many discussions, this is where the conversation ends. People say: fiscal policy looks attractive in theory, but if you allow it to be used, it will be misused. Therefore, we should avoid touching it, or bind it tightly. But I do not think we should stop there. I think we should ask: is there a way to get the benefits of stabilization without opening the door to chronic misuse? This is where I want to introduce the idea of quasi-automatic stabilizers.

Countries already have some automatic stabilizers. In a recession, tax revenues fall automatically. Some spending rises automatically—for example unemployment benefits. That leads to an automatic increase in the fiscal deficit, which provides some support to aggregate demand. This is the classic automatic stabilizer mechanism. But the classic automatic stabilizers are often too weak, or not well targeted, or not designed to operate symmetrically.

What I mean by quasi-automatic stabilizers are mechanisms that are triggered by observable economic conditions—unemployment, output gaps, or other indicators. They do not require new discretionary legislation each time. And importantly, they can be designed to expand when needed and disappear when not needed.

To clarify: they are not “fully automatic” in the sense that there is still some operational choice. Someone has to compute and publish the indicators. Someone has to implement the parameter change. But they are automatic enough to avoid the long political delays and the temptation to debate from scratch each time.
Now, if we are going to rely on quasi-automatic stabilizers, we have to design them properly. Two features are essential.

(i) They must be symmetric: They must expand in downturns and contract in booms. If they are one-sided—only expansionary in bad times—they will mechanically generate a debt drift. Symmetry is non-negotiable.

(ii) They must include a feedback mechanism for debt sustainability: Even if shocks are “zero mean,” debt dynamics do not necessarily converge. If the stabilizer alternates between expansion and contraction without any systematic reaction to the level of debt, the debt ratio can wander and eventually become problematic.  (For example, if fluctuations are white noise, and the interest rate is zero, debt will the follow a random walk).

So you need a feedback rule: when debt rises, the fiscal stance must gradually become tighter, in a predictable and rule-like way. This is the intuition behind the Bohn-type sustainability condition: the primary balance should respond positively to the level of debt. This is not easy to design. But the fact that it is not easy does not mean it is impossible. It means it must be done carefully and transparently. 

2. Chile’s volatility: copper, private income, and the limits of stabilizing only the budget

When I first discussed these issues with Chilean colleagues, one reaction was: “Chile’s big macro problem is copper price volatility, and we have already designed a structural balance rule to deal with it.” That is a reasonable first reaction. Chile’s structural balance rule is a major institutional innovation. It isolates the budget from transitory copper revenue changes and helps prevent the government from spending windfall revenues in booms.

But stabilizing the budget with respect to copper prices is not the same as stabilizing the economy. If you decompose fluctuations in Chilean GDP, copper price movements matter—but they are not everything. A rough number, based on the work I have seen, is that copper accounts for something like 10 to 20 percent of output fluctuations. That is meaningful, but it leaves most of the volatility explained by other shocks—domestic demand shocks, external financial conditions, and so on. Moreover, and importantly, most copper production is private. The government collects royalties and taxes, but a large part of the copper boom is an income shock to the private sector. That means that even if the government stabilizes its own revenues, there can still be large fluctuations in private income and private demand. So you can stabilize the fiscal accounts and still have a nontrivial macroeconomic cycle. This is one reason why other fiscal stabilization tools can still be relevant in Chile.

When we ask what quasi-automatic stabilizers might be useful in Chile, the usual candidates come first: unemployment insurance, countercyclical transfers, solidarity funds. These can be designed so that support rises automatically when unemployment rises.

I understand there are Chile-specific issues: labor market structure, informality, or a limited benefit system may limit the macro power of unemployment-based stabilizers. But even if the aggregate effect is not huge, improvements at the margin could still be beneficial. These instruments also have an equity and insurance dimension, not just stabilization. But I want to discuss one more idea—not because I think it is the one solution, but because it illustrates the type of thinking we should encourage.

That idea is a variable VAT.

3. Variable VAT as a stabilization instrument

Chile has a VAT. Many countries do. The United States does not, which is why this is not typically part of U.S. macro thinking. But in a VAT country, you have an instrument that can be adjusted and can have significant demand effects. The idea is simple: make the VAT rate vary systematically with cyclical conditions. When the economy is overheating, raise the VAT rate. When the economy is weak, lower the VAT rate. Make the adjustment rule  state-contingent.

The key channel is intertemporal substitution in consumption.

A VAT affects the effective price consumers face. If you cut the VAT today but announce that it will increase later—according to a rule—then the price of consuming today becomes low relative to consuming tomorrow. Consumers have an incentive to bring forward purchases. This can be powerful, especially for durable goods. This is conceptually similar to monetary policy. Monetary policy works partly by changing the intertemporal price of consumption—through the real interest rate. A variable VAT can mimic that channel, at least on the consumption side.

One reason this could be useful is magnitude. Central banks often move interest rates by 25 basis points, sometimes 50. Those are relatively small changes. A VAT rate can be moved by a few percentage points, and the implied intertemporal incentive can be large. To give a rough sense: a temporary VAT reduction of 2 percent below normal this year followed by an expected increase next year of 2 percent above normal  is equivalent to a 400 basis point decrease in  the effective “consumption discount rate.” In other words, the implied stimulus can be large.

There are caveats. A VAT change affects consumption directly but not investment in the same way, so it may distort the composition of demand. Perhaps that is acceptable in some situations, but it should be recognized. Another issue is inflation. A VAT reduction lowers measured prices; a VAT increase raises them. That mechanically affects headline inflation. But central banks can, in principle, “look through” VAT-induced inflation effects if the inflation target is defined appropriately—focusing on inflation excluding tax changes, or on a measure of underlying inflation.

Where this becomes most relevant is near the zero lower bound. When monetary policy cannot reduce the nominal rate further, the economy may require other tools that shift intertemporal incentives. Chile came close to that situation in 2021. It avoided it, but it is not unthinkable that it could happen again in the future.  A VAT rate cut may well dominate QE. 

I do not present the variable VAT as a must-do recommendation. I present it as an example of the broader point: fiscal stabilization should not be limited to ad hoc discretionary packages. It should be designed as a system—rule-like, automatic or quasi-automatic, symmetric, and sustainable  (for more, see [4]).

4. How dangerous is public debt? Why “magic thresholds” are wrong

Let me now turn to public debt. Chile has a debt ceiling, roughly 45 percent of GDP. A natural question is: is that number reasonable? Where does it come from? How should we think about it?

We have a tendency, in public debates, to search for a magic number. At some point, not long ago, a prominent view was that 90 percent of GDP is a debt threshold beyond which growth collapsed or crises became likely. That view became popular, but it has not held up well empirically or conceptually. Many countries have exceeded such thresholds without immediate disaster; others have faced trouble at much lower debt levels.

The correct conclusion is simple: there is no universal threshold. The sustainable level of debt depends on the country. It depends on institutions, credibility, tax capacity, growth prospects, the structure of debt, exposure to shocks, the nature of investors, and the political ability to generate primary surpluses.

In some work I am doing with coauthors [5], we look at sovereign credit ratings and how rating agencies respond to debt and fiscal variables. What we find is that agencies do respond sensibly to changes: higher debt tends to lower ratings, larger deficits tend to lower ratings. But the country fixed effects are enormous. The same debt ratio is perceived very differently across countries. That matches reality. There is no one debt ratio below which debt is safe”, and above which debt is dangerous.”

How should we think about the cost of debt? A useful starting point is to separate two cases:

1. Safe debt, where default risk is negligible.
2. Risky debt, where default risk is possible, and expectations can shift.

Let me begin with the safe-debt case, because much confusion comes from mixing the two.

5. Safe debt and the role of r minus g (r-g)If we ignore the irrelevant case of Ricardian equivalence, debt crowds out capital in the portfolio of households.  How costly is the reduction in the capital stock? 

The research on the topic, dating back to famous papers by Samuelson, Allais, Phelps, and Diamond points to the central importance of r-g, i.e the difference between the real interest rate and the real growth rate.  If r-g is positive, lower capital implies lower output and lower consumption in the future:  Debt decreases welfare.   If r-g is negative, then lower capital implies lower output but higher consumption in the future:  The economy is said to be “dynamically inefficient “ and debt increases welfare. 

The intuition for this important but perhaps somewhat surprising result is actually straightforward.  Suppose that the interest rate, taken to be the marginal product of capital, is equal to r, so the total return to capital is rK . Suppose that the economy maintains a constant ratio of output to capital, so that, as output grows at rate g, an amount gK needs to be put aside as investment.  If r is less than g, the economy has to put aside more than it actually gets from capital, and thus consumption decreases.  In this case, more debt, which leads to less capital, is good. 

Now, is this a purely theoretical result? Or is it relevant in practice?

I used to think it was mostly a theoretical possibility taught in graduate macro as a curiosity. But over time, I have become more convinced that it is not exotic. We have observed long periods where safe real rates are low relative to growth. Chile has experienced long stretches where average r − g has been negative.  

There are however many rates of return, interest rates on government bonds, or rates of return on shares.  This raises an obvious question:  Which one should we use to assess whether the economy is dynamically inefficient, and debt may be good?  In my research [1], [2], I have concluded that, under reasonable conditions, the right comparison is not between g and the return on risky capital, like equity, which is indeed, on average, higher than g. The right comparison is between g and a risk-adjusted safe rate, which is close to the return on safe bonds. Once you use that, negative r − g becomes more common.

That said, one must be careful. There are debates about the convenience yield on safe assets, about whether the observed safe rate is artificially low because investors value liquidity,  or for regulatory reasons. There are also debates about whether low safe rates reflect an excessively large an equity premium, implying too low a safe rate for a given marginal product of capital.   But the practical point I take from this line of research is: in the safe-debt case, moderate debt is not necessarily very costly, and we should not treat any level of debt as automatically disastrous.

However, this is only half the story, because debt is often not fully safe.

6. Risky debt: political limits, risk premia, and multiple equilibria

In reality, debt can become risky when investors start doubting whether the government can generate the necessary primary surpluses in adverse states of the world.

Debt dynamics are summarized by the standard equation: debt evolves depending on the interest rate, the growth rate, and the primary balance. If interest rates rise, or growth slows, stabilizing debt requires a higher primary balance. But politics imposes limits. There is some maximum primary surplus a government can realistically sustain: There is a limit to how much you can raise taxes. You cannot cut spending beyond what society tolerates. So even if the government wants to adjust, there is a political constraint which eventually binds.

If debt is high enough, or if interest rates rise enough, the required primary surplus may exceed what is politically feasible. At that point, default becomes possible—not certain, but possible. Once default becomes possible, interest rates include a risk premium. That raises the interest burden, worsening debt dynamics. This can generate a feedback loop between the change in debt and the change in the interest rate, leading debt to increase more and more rapidly.

And there is an even more troubling possibility: multiple equilibria. If investors become pessimistic and demand a higher premium, the interest rate rises, making the debt path worse, validating pessimism. If investors are optimistic instead, rates remain low and the debt path remains stable. Both equilibria may be possible at the same level of debt. This is not an abstract theoretical point. It is a real feature of sovereign debt markets. It implies that countries may want to avoid approaching regions where multiple equilibria become plausible, even if  “fundamentals” look acceptable (on the potential location and size of the range of multiple equilibria, see [3])

So the right question for debt policy is not: “is debt above some threshold?” The right question is: what is the probability that debt dynamics become explosive under plausible shocks?

That naturally brings us to stochastic debt sustainability analysis.

7. SDSAs: diagnosing sustainability under uncertainty

If we want to assess debt sustainability seriously, we should not do it deterministically. A deterministic approach—taking point forecasts for growth, interest rates, and primary balances and projecting the debt ratio—gives a false sense of precision and of security. It ignores the fact that future outcomes are uncertain, that shocks occur, and that risks are nonlinear.

What we need is a stochastic framework.

Stochastic Debt Sustainability Analysis (SDSA) does exactly that. It takes distributions for key drivers—growth, interest rates, exchange rate movements if relevant, primary balance shocks—and it simulates the distribution of future debt paths. The output is not a single line, but a fan chart: a distribution of possibilities. This is useful for two reasons. First, it forces discipline. Instead of debating based on anecdotes, we quantify risk: how likely is debt to rise above certain levels, or enter unstable dynamics? Second, it recognizes that sustainability is about tails. Even if the expected path looks fine, the tail risk may be unacceptable. Or the expected path may look moderately bad, but the probability of truly explosive outcomes may still be small.

In practice, I think it is useful to do SDSAs at more than one horizon. A five-year horizon is useful because beyond five years forecasting becomes extremely difficult. Many budget and policy frameworks operate at that horizon. So you simulate five-year debt dynamics and ask: is debt likely to be on an exploding path? But you also want a longer horizon—say twenty years—because long-run pressures matter. Demographics, persistent shocks, slow-moving expenditure programs, and rare disasters do not show up fully in five years. So for strategic fiscal planning, the longer horizon is informative.

For an SDSA to be credible, it must be done by a technically competent and fully independent institution. Governments, if they do it themselves, have incentives to use optimistic assumptions about growth and revenues. You need independence not just in name, but in practice—independence of analysis and independence of forecast assumptions.

Chile has strong institutions in this dimension. But even strong institutions benefit from explicit use of SDSAs as a central organizing framework  (for more on the potential use of SDSAs, see [6]).

8. Public investment and the limits of simplistic rules

One issue that SDSAs help clarify is public investment. There is a recurrent political argument: Investment is good, so it should be exempt from fiscal constraints. That often leads to proposals for “golden rules” that allow deficits to finance investment.

In practice, such proposals frequently fail for obvious reasons. Investment can be misclassified. Current spending can be relabeled as investment. The rule becomes a loophole. But the opposite approach—treating public investment exactly like current spending—also has a cost. It can discourage worthwhile long-term projects. Some public investments raise welfare even if they do not generate an immediate fiscal return.

In some cases, investment may partly pay for itself through higher growth and higher revenues. In other cases, investment is valuable for welfare but does not raise revenues much. Defense spending is a good example: it may be necessary, but it does not generate a fiscal dividend. Climate adaptation can be similar. Many forms of infrastructure investment raise welfare but cannot be  monetized fiscally. SDSAs provides a practical way to discuss this without slogans.

Instead of saying “investment is always good” or “investment should never be treated differently,” one can ask: what does the investment imply for growth, revenues, and the debt path, probabilistically? If you believe an investment raises growth by some amount, simulate it (although do not ignore the uncertainty). See how the debt distribution changes. Evaluate whether the increased debt risk is acceptable given the welfare gain. It may well be, but the discussion must be an explicit one. 

This is a more mature way to discuss investment within fiscal rules.

9. Fiscal rules: targets, escape clauses, and the need for a debt anchor

Let me now move from diagnosis to rules. Suppose a country wants a modern fiscal framework. What should it look like? At a minimum, you need some rule for the deficit—often a structural balance rule or a structural primary balance rule. This helps limit cyclicality and avoid overreaction to transitory revenues, especially commodity revenues. Chile has done this in an impressive way.

But a primary balance rule alone may not be enough. Debt matters as an anchor. Debt is not necessarily a perfect target—there is uncertainty, and it is influenced by shocks—but the framework needs to ensure that debt does not drift upward indefinitely. This is why combining a structural balance rule with a debt target or ceiling as a weak attractor can be useful. Whether the ceiling is a debt ratio of 40%, or 60%, or more, is not essential, as long as it is credible and the system forces the government to eventually return to it.

Escape clauses are also crucial. The world is uncertain. Crises happen. Natural disasters happen. Wars happen. Pandemics happen. If your fiscal rule has no escape clause, it becomes either incredible or procyclical in the worst moments. A well-designed escape clause allows deviation under exceptional circumstances while preserving a credible commitment to return to the framework.

Chile’s framework, as I understand it, combines these elements: structural balance discipline, a debt ceiling, and escape clauses. This is close to what I would design if I were advising a government.

10. Adjustment speed and credibility: why gradualism often dominates

Finally, suppose the SDSA indicates a need for adjustment. The question becomes: how fast should the government adjust?

I generally favor gradualism. In many models, welfare losses are quadratic in deviations from the optimal path. That means the marginal cost of a large adjustment within a given year is high. If you can spread adjustment over time, you reduce welfare losses. The counterargument is credibility. If adjustment is too slow, markets may not believe it, and risk premia may rise. That is the real tradeoff: welfare versus credibility.

But I want to emphasize that credibility is not a simple function of speed. If you say you will close the gap immediately, that may not be credible either. It may be politically impossible, and people may not believe it. On the other hand, if you say you will adjust over ten or fifteen years, that is also not credible. The plan becomes meaningless.

Credibility comes from a coherent medium-term plan and adherence to it, much like inflation targeting. It is the commitment to converge to the target within a reasonable timeframe, supported by institutions and transparency, that creates credibility—not simply front-loading adjustment. This is why a well-designed framework can allow gradual adjustment while maintaining credibility.

Conclusion

Let me conclude. Chile has built an impressive fiscal framework. There is a real sense in which the Chilean system reflects what many economists recommend: discipline through structural targets, an anchor through a debt ceiling, flexibility through escape clauses, and institutional credibility.

Thus, my message is definitely not that Chile needs radical change. My message is that the next steps are on the margin, but they still matter. First, I think there is room to use fiscal policy more effectively for stabilization—not through discretionary packages each time, but through better designed quasi-automatic stabilizers: symmetric, state-contingent, and with feedback for sustainability.

Second, I think we should be more precise about the dangers of debt. Debt is not automatically disastrous, especially in a world where  r − g can be low or negative for long periods. But risky debt dynamics and multiple equilibria are real, and they justify caution. This means debt targets should be justified through rigorous probabilistic analysis rather than arbitrary thresholds.

Third, I think SDSAs should be central to the design of fiscal rules. They  provide a disciplined way to evaluate risks, justify targets, assess investment tradeoffs, and structure adjustment paths.

There are other topics that I could have discussed—multipliers, fiscal-monetary coordination, fiscal dominance. In Chile, given the strength of the framework, and in contrast to some other countries today, such a discussion is less urgent. But, they remain central. So, once again, congratulations. I am impressed by what you have achieved, and I hope these remarks help you think about how to further refine an already strong system.

Thank you.

References to the underlying articles and papers

[1] Olivier Blanchard. “Public debt and low interest rates.” American Economic Review, 109(4):1197–1229, 2019.

[2] Olivier Blanchard. Fiscal Policy under Low Interest Rates. MIT Press, 2023.

[3] Olivier Blanchard, Pedro Martinez Bruera, and Gonzalo Huertas. “The range of multiple equilibria in sovereign debt markets.” mimeo, 2025. 

[4] Olivier Blanchard, “Fiscal Policy as a Stabilization Tool. The Case for Quasi-Automatic Stabilizers, with an Application to VAT”, Peterson Institute for International Economics Working Paper No. 25-6, 2025

[5] Olivier Blanchard, Daniel Leigh, and Prachi Mishra, “Ratings, deficits, and debt; an exploration”, mimeo 2026

[6] Olivier Blanchard, Jeromin Zettelmeyer, and Alvaro Leandro, “Redesigning EU rules.  From rules to standards”, Economic Policy, 2021