Fiscal policy and the national debt Lesson 8 of 10
Economics · Fiscal policy and government
The fiscal multiplier: how much GDP a dollar of stimulus buys
What the fiscal multiplier measures, how economists estimate it, the evidence from CBO, Ramey and Nakamura–Steinsson, and why it changes with slack, rates and trade.
On this page
The fiscal multiplier answers a question at the heart of every stimulus debate: if the government spends one more dollar, or cuts taxes by one dollar, how much does the economy's total output change? A multiplier of 1.5 means $1 of spending adds $1.50 to GDP. A multiplier of 0.5 means it adds only 50 cents, because something else, such as private spending, falls by the other half.
The number matters because it decides whether stimulus is cheap or expensive. It is also genuinely uncertain. When Congress passed the Recovery Act in February 2009, the Congressional Budget Office (CBO) estimated that each dollar of federal purchases would raise GDP by between $1.00 and $2.50. A decade of research later, Valerie Ramey's survey put most estimates of the average spending multiplier between 0.6 and 1. This lesson explains what the multiplier is, how each kind of estimate is made, what the main studies found and why the answer depends on the state of the economy and the central bank.
What the multiplier measures
Three choices change the number a study reports:
- Impact or cumulative. The first-quarter effect is usually smaller than the effect summed over two or three years.
- What changes. Purchases of goods and services count in GDP directly (GDP explained: how gross domestic product is measured and read counts them in G). Transfers, such as unemployment benefits, and tax cuts work only through what households and firms then spend.
- Holding what fixed. A study that lets the Federal Reserve respond as it normally does will find a smaller multiplier than one that assumes interest rates stay put.
How it works: the rounds of spending
When the government buys $100 billion of goods, firms and workers earn $100 billion more. They spend part of it, which becomes someone else's income, who spends part of that, and so on. What stops the chain are leakages: income that is saved, paid in taxes or spent on imports. If a share s of each extra dollar is respent on domestic output, the rounds add up to a geometric series.
The offsets go by several names. Interest-rate crowding out works through borrowing costs (see Crowding out: do government deficits raise interest rates?). Monetary offset is the central bank deliberately tightening to keep inflation on target. Ricardian effects are households saving more because they expect higher future taxes. Each pulls the multiplier down.
How economists estimate it
Nobody can run the economy twice, once with stimulus and once without, so every method substitutes something for that missing comparison:
- Macroeconomic models. CBO, the Fed and private forecasters plug a policy into a model whose behavior was fitted to past data. Results depend on the model, which is why CBO publishes ranges.
- Time series with "news" shocks. Ramey and others identify changes in spending that were not a response to the economy, such as defense build-ups announced for military reasons, and track GDP afterward. For taxes, Romer and Romer read the legislative record to find tax changes made for reasons other than the business cycle.
- Cross-region comparisons. Compare states that got more federal money with states that got less at the same time. National shocks and Fed policy affect all states equally, so they drop out.
- Household data. Track what households did with a specific payment, such as the 2008 stimulus checks, whose mailing date depended on the last two digits of the taxpayer's Social Security number and was therefore effectively random.
What the evidence shows
CBO's ranges for the 2009 Recovery Act
In its March 2, 2009 letter on the $787 billion Recovery Act (ARRA), CBO grouped provisions by type and gave each a range. Spending that buys output directly ranked highest (1.0 to 2.5). Transfers to people likely to spend them came next (0.8 to 2.2), then tax cuts for lower- and middle-income households (0.5 to 1.7). Provisions aimed at higher-income households or business cash flow ranked lowest, because more of the money would be saved. Putting it together, CBO estimated ARRA would raise GDP by 1.4% to 3.8% by the fourth quarter of 2009 and by 1.1% to 3.4% by the fourth quarter of 2010, and employment by 1.2 million to 3.6 million by late 2010. It also estimated a small long-run cost: beyond 2015 the extra debt would lower GDP by between zero and 0.2%.
The width of those ranges is the honest answer: before the fact, CBO could not narrow it.
Ramey's survey: 0.6 to 1
Ramey's 2019 review of a decade of post-crisis research concluded that, on average, government purchases multipliers are likely to be between 0.6 and 1, and that narrative-based estimates of tax-rate changes imply tax multipliers of −2 to −3: a tax cut worth 1% of GDP raises GDP by 2% to 3%. She found the evidence for multipliers above one in recessions or periods of slack "typically not robust", but the evidence for larger multipliers when monetary policy accommodates fiscal policy, such as at the zero lower bound on interest rates, somewhat stronger, with some estimates of 1.5 or more.
Nakamura and Steinsson: about 1.5 across regions
Using military procurement, which rises and falls across states for national reasons, Nakamura and Steinsson estimated that when a region's federal purchases rise by 1% of its output, its output rises by about 1.5% relative to other regions. That is an open economy relative multiplier. Because all states share one monetary policy and one federal tax system, the Fed's response is "differenced out". The authors argue their estimate is consistent with models in which the national multiplier is large when the central bank does not lean against the stimulus, and smaller when it does.
Household checks
Parker, Souleles, Johnson and McClelland found that households spent about 12% to 30% of their 2008 stimulus payments on nondurable goods in the three months they arrived, and about 50% to 90% including durables such as cars. That speaks to how fast transfers are spent, one ingredient of a multiplier, not to the whole economy's response.
Why the multiplier varies
- Slack. With idle workers and machines, extra demand can raise output without bidding up prices. At full employment it mostly raises prices or displaces other spending.
- Monetary policy. If the Fed raises rates to offset stimulus, the multiplier shrinks. At the zero lower bound, when the Fed is not raising rates, models predict it is larger.
- Openness. Money spent on imports leaks abroad. Small open economies have smaller national multipliers than large, relatively closed ones such as the US.
- Who gets the money. Households that are short of cash spend more of a dollar than those that are not, so targeted transfers usually rank above broad tax cuts in CBO's table.
- Temporary or permanent. A permanent tax cut changes long-run incentives and income expectations; a one-off check mostly does not.
- Debt levels and confidence. Ramey cites cross-country evidence that multipliers may be lower where public debt is high, though the estimates are imprecise.
Pitfalls and limitations
- One number does not exist. "The multiplier" depends on the type of measure, the horizon, the state of the economy and the central bank. Ask which one a claim means.
- A regional multiplier is not a national one. Nakamura and Steinsson's 1.5 is relative to other states with the same Fed; their own framework says the national figure depends on monetary policy.
- Correlation is not the multiplier. Spending rises in recessions because of automatic stabilizers (Automatic stabilizers: taxes and benefits that move with the cycle), so a raw correlation between spending and GDP is biased toward zero or negative.
- Long run differs from short run. CBO expected ARRA to raise GDP for several years and lower it very slightly after 2015 because of the added debt.
Multiplier estimates compared
| Source | What it measures | Estimate |
|---|---|---|
| CBO (2009), purchases | Cumulative, US, 2009 conditions | 1.0 to 2.5 |
| CBO (2009), higher-income tax | Cumulative, one-year AMT relief | 0.1 to 0.5 |
| Ramey (2019), purchases | Average across studies | 0.6 to 1 |
| Ramey (2019), tax rates | Narrative tax-change studies | −2 to −3 |
| Nakamura–Steinsson | Regional, relative to other states | About 1.5 |
How Finabase does it
Finabase does not estimate fiscal multipliers or track stimulus programs. The charts here are ours, built from CBO's published ranges and an invented example. What the app shows is how markets react to the inputs:
- The economic calendar on the Economy page lists each GDP release with the actual figure, the forecast and the previous reading, where any stimulus effect would eventually show up.
- The Treasury page shows yields from 1 month to 30 years, where expectations of more borrowing and of the Fed's response appear.
- News about stimulus bills and spending plans collects under the Economy news topic.
Frequently asked questions
What is a fiscal multiplier in simple terms?
It is how many dollars of GDP one dollar of government spending or tax cuts produces. A multiplier of 1.5 means $1 of stimulus raises output by $1.50; below 1 means some private spending was displaced.
What is the fiscal multiplier in the US?
There is no single figure. Ramey's 2019 survey put most estimates of the average spending multiplier at 0.6 to 1, while CBO's 2009 ranges for federal purchases ran from 1.0 to 2.5 under the conditions of that recession.
Is the multiplier bigger in a recession?
Possibly, but the evidence is mixed. Ramey found that results showing multipliers above one in recessions are not robust, while evidence for larger multipliers when interest rates are stuck near zero is somewhat stronger.
Do tax cuts or spending have a bigger multiplier?
It depends on the tax cut and the spending. CBO ranked direct purchases and transfers to people likely to spend them above tax cuts for higher-income households; narrative studies of permanent tax-rate changes find large effects, which Ramey puts at −2 to −3.
Summary
The fiscal multiplier is the change in GDP per dollar of government spending or tax change. Rounds of respending push it up; leakages into saving, taxes and imports and offsets from higher interest rates pull it down. Estimates span a wide range: CBO's 2009 figures ran from 0 to 2.5 depending on the measure, Ramey's survey puts the average spending multiplier at 0.6 to 1, and regional evidence finds about 1.5 for one state relative to others. The honest reading is that the multiplier is likely larger when the economy has slack and the central bank holds rates down, and smaller when it does not.
Check your understanding
4 questions on this lesson. Each answer comes with the reason, and nothing is scored.
The government spends $50 billion on road repairs. Over two years GDP is $40 billion higher than it would otherwise have been. What was the cumulative multiplier, and what does it imply?
- 0.8: some private spending was displaced, so output rose by less than the spending
- 1.25: the spending was multiplied
- −0.2: the program shrank the economy
Show the answer
A. 0.8: some private spending was displaced, so output rose by less than the spending $40 billion ÷ $50 billion = 0.8. A multiplier below 1 means the program added to GDP, but less than its cost, because something else fell, such as private investment.
Two identical stimulus packages are passed. In country A the central bank holds its rate at zero; in country B it raises rates to keep inflation on target. Where would you expect the larger multiplier?
- Country A, because no rise in interest rates offsets the extra demand
- Country B, because higher rates attract savings
- No difference: the multiplier is a fixed number
Show the answer
A. Country A, because no rise in interest rates offsets the extra demand Monetary offset is one of the main forces that shrink the multiplier. Ramey finds the evidence for larger multipliers is strongest when monetary policy accommodates, such as at the zero lower bound.
A study finds that states receiving more military contracts grew 1.5 dollars per dollar of extra spending relative to other states. Can you conclude the national multiplier is 1.5?
- Yes, adding up all states gives the national effect
- No: the comparison removes the Fed's response and national taxes, which affect the national multiplier
- No: regional data cannot show any effect of spending
Show the answer
B. No: the comparison removes the Fed's response and national taxes, which affect the national multiplier This is the Nakamura–Steinsson design. All states share one monetary policy, so its offset is differenced out. The national multiplier depends on how the central bank responds.
In CBO's 2009 table, why did transfers to people rank above a one-year tax cut for higher-income households?
- Transfers count directly in GDP as government purchases
- People receiving transfers were expected to spend more of each dollar, while higher-income households save more of a temporary cut
- Tax cuts are not counted until the following year
Show the answer
B. People receiving transfers were expected to spend more of each dollar, while higher-income households save more of a temporary cut Transfers are not purchases; they work through spending. The ranking reflects how much of each dollar recipients were expected to spend: 0.8 to 2.2 for transfers to persons versus 0.1 to 0.5 for the higher-income cut.
Exercise
One task to do in Finabase. Nothing is graded or saved.
The lesson says evidence for larger spending multipliers is stronger when interest rates are stuck near zero, because the central bank does not raise rates to offset the extra demand. Open the 3-month Treasury bill yield on its longest range and mark the stretches when it sat near zero. Which of the three big fiscal responses (2001, 2009, 2020–21) happened inside one, and which periods since 2022 did not?
Open the 3-month bill chartWhat you should find
The bill sat near zero from about late 2008 to late 2015 and again from spring 2020 to early 2022, so the 2009 Recovery Act and the 2020–21 relief fell in zero-rate periods, while 2001 did not. Since 2022 the bill has been well above zero, the setting in which a higher policy rate can offset fiscal demand and the multiplier is likely smaller.
Sources and further reading
- CBO — Letter to Sen. Charles E. Grassley on the estimated macroeconomic impacts of the American Recovery and Reinvestment Act of 2009, March 2, 2009 (UNT web archive), retrieved 2026-11-18
- Ramey, V. — Ten Years after the Financial Crisis: What Have We Learned from the Renaissance in Fiscal Research? NBER Working Paper 25531 (2019; Journal of Economic Perspectives 33(2)), retrieved 2026-11-18
- Nakamura, E. and Steinsson, J. — Fiscal Stimulus in a Monetary Union: Evidence from U.S. Regions, NBER Working Paper 17391 (2011, rev. 2013; American Economic Review 104(3), 2014), retrieved 2026-11-18
- Parker, J., Souleles, N., Johnson, D. and McClelland, R. — Consumer Spending and the Economic Stimulus Payments of 2008, NBER Working Paper 16684 (American Economic Review 103(6), 2013), retrieved 2026-11-18
