Does government spending help the economy?
A spending dollar either multiplies into more output or it doesn’t. The honest answer turns out to depend on four things — and most arguments ignore all of them.
The multiplier promise
Four million views, and the claim underneath it is simple: a government that prints its own currency can never run out of money, so spending is basically free. The first half is almost right. The second half is where the trouble starts, because “can’t run out” and “costs nothing” are two completely different statements, and the gap between them is the whole subject of this walkthrough.
Start with the result that made the case for spending in the first place. When the government buys a dollar of goods, that dollar becomes someone’s income. They spend most of it, which becomes someone else’s income, who spends most of that. The dollar circulates, and total output rises by more than the dollar the government put in. This is the fiscal multiplier, and it is the engine of every stimulus argument ever made.
How big is it? That depends on how much of each new dollar people spend rather than save — the marginal propensity to consume. Suppose households spend 80 cents of every extra dollar ($MPC = 0.8$). The first round adds \$1 of government purchases; the second adds 80 cents; the third adds 64 cents; and so on down a shrinking chain. Sum the whole series and the textbook number is startling.
The rounds form a geometric series. With government spending $G$ and a marginal propensity to consume $c$, total output rises by
$$\Delta Y = G \cdot \big(1 + c + c^2 + c^3 + \cdots\big) = \frac{G}{1 - c}$$With $c = 0.8$, the multiplier is $1/(1-0.8) = 5$. One dollar of spending, five dollars of output. On paper, the deficit pays for itself many times over.
Think of it as a chain reaction. The government hires a construction crew. The crew spends its paychecks at restaurants and shops. Those owners hire more staff and order more supplies. Each round is smaller than the last, because some money leaks out into savings at every step, but the rounds add up to far more than the original dollar. That is why a multiplier of five looked, for a generation, like the closest thing economics had to free money.
Then the qualifier arrives, and it is a serious one. When the government borrows to spend, it competes for the same pool of savings that private firms borrow from. That competition pushes interest rates up, and higher rates discourage private investment. Economists call this crowding out: every dollar of public spending displaces some private spending that would otherwise have happened. If crowding out were total, the multiplier would collapse to zero, and the government would just be moving the same dollar from one pocket to another. The naive multiplier of five was always an upper bound. The real number is whatever survives the crowding-out subtraction.
The full apparatus lives in the economics textbook’s intro-macro chapter: the Keynesian cross that generates the multiplier, and the IS-LM model that prices in the interest-rate feedback. Peek at the cross to see where $1/(1-c)$ comes from; peek at IS-LM to see crowding out happen on a diagram.
“A country that issues its own currency can never run out of money the way you or I can. It can always afford to buy what is for sale in its own currency.”
— Stephanie Kelton, The Deficit Myth, 2020
Can a government that prints money just spend freely?
The viral claim says a currency issuer can never go broke, so deficits don’t matter. Half of that is correct. The other half hides the entire question of what spending actually costs.
The multiplier at full volume vs. crowding out at full strength
“The point is that nobody has come up with a coherent argument against a fiscal stimulus — the multiplier really is greater than one, and government spending really does raise output and employment in a depressed economy.”
— Paul Krugman, New York Times, January 2009
Krugman was arguing for the \$1.2 trillion stimulus in early 2009, and his case ran straight through the Keynesian-cross arithmetic: with the economy depressed and idle resources everywhere, a spending dollar circulates without bidding anything away, so the multiplier exceeds one and output rises by more than the cost. He is operating inside the tradition that John Hicks built into the IS-LM apparatus from Keynes’s 1936 General Theory, the lineage the History of Economic Thought book traces in its chapter on the Keynesian revolution. The depressed-economy condition is doing the work, and Krugman knows it. It is the same condition Stage 3 will make precise.
“Every dollar of increased government spending must correspond to one less dollar of private spending. Jobs created by stimulus spending are offset by jobs lost from the decline in private spending.”
— John Cochrane, “The Grumpy Economist”, 2009
This is crowding out at maximum strength: if savings are fixed, public borrowing displaces private borrowing dollar for dollar and the multiplier is exactly zero. Cochrane is writing from the monetarist and rational-expectations counter-revolution that spent the 1970s and 1980s dismantling naive Keynesianism, the program the History of Economic Thought book covers in its chapter on the counter-revolution. The argument is too strong as stated: it assumes perfect crowding out, which requires savings to be completely fixed and resources fully employed, neither of which holds in a slump. But the core insight survives the exaggeration: borrowing to spend has a cost, and the multiplier is whatever is left after you subtract it.
Where this leaves us
The multiplier is real, but it is smaller than the naive $1/(1-c)$ formula promises, because crowding out claws some of it back. How much it claws back depends on the state of the economy: in a slump with idle resources, crowding out is weak and the multiplier is large; at full employment, crowding out bites and the multiplier shrinks. So the number you carry into Stage 2 is not a point estimate. It is a question: under what conditions? And the first condition to interrogate is the one the multiplier quietly assumed: that consumers spend their share mechanically, round after round, without ever thinking about the bill.
But what if consumers are smarter than that? What if they look at the government borrowing \$100 billion and think: someone is going to have to pay for this in higher taxes later, and that someone is me? If they do, they might just save the stimulus instead of spending it. And then the multiplier doesn’t shrink. It vanishes.
The rational consumer problem
“Our national debt just crossed \$33 TRILLION. That’s over \$99,000 of debt for every single American. We are committing fiscal child abuse.”
— Sen. Rand Paul, October 2023
The debt-per-citizen framing assumes a model of government debt that most economists reject: the government is not a household, and the debt is not a bill that gets posted to your account. But it captures a real intuition, one with an airtight piece of theory behind it: if today’s deficit is tomorrow’s tax, maybe rational consumers already know that, and behave accordingly.
The Stage 1 multiplier rests on a hidden assumption: that people spend their share of each dollar mechanically, this year, without thinking about next year. Robert Barro took that assumption apart in 1974, and the result is one of the most unsettling theorems in macroeconomics.
His argument runs in four steps:
- The government cuts taxes or spends more, funding it by borrowing. Households have more money in hand today.
- But the debt must eventually be repaid through higher future taxes. The government has not created wealth; it has rescheduled when the bill arrives.
- Forward-looking households see the future tax coming. So they save the entire windfall to pay it, leaving their consumption unchanged.
- The multiplier is zero. Deficit-financed stimulus changes the timing of taxes, nothing else. This is Ricardian equivalence: debt and taxes are equivalent ways of paying for the same spending, and rational consumers are indifferent between them.
Formally, the household faces a lifetime budget constraint in which the present value of consumption equals the present value of after-tax income. A tax cut today financed by borrowing raises current disposable income but raises the present value of future taxes by exactly the same amount — the bond the government issues is matched, dollar for dollar, by the household’s future tax liability. The present value of lifetime resources is unchanged, so optimal consumption is unchanged. The household simply buys the new government bond with the tax cut and waits for the tax bill it knows is coming.
Imagine your employer hands you a \$5,000 “bonus” and tells you, in the same breath, that it will be docked from your salary over the next two years. You wouldn’t go on a spending spree. You’d set the money aside, because you know the clawback is coming. Ricardian equivalence says a deficit-financed tax cut is exactly that bonus: not new income, just an advance against a future deduction you can see on the horizon.
Barro’s theorem is airtight given its assumptions. The trouble is the assumptions. It requires households to be forward-looking over infinite horizons, to have unlimited access to credit, and to care about the taxes their grandchildren will pay. Relax any one of these and the result weakens. The decisive relaxation is the second. Many households are credit-constrained: they would borrow against future income if they could, but no bank will lend to them, so they spend every extra dollar they get. Campbell and Mankiw, in 1989, estimated that roughly half of all consumption comes from these “rule-of-thumb” or hand-to-mouth households. For that half, a tax cut isn’t a clawback they save against. It is cash they spend immediately, MPC near one. That single empirical finding is what rescued the fiscal multiplier from Ricardian oblivion.
The intertemporal-choice machinery behind all of this lives in the economics textbook’s intermediate-macro chapter: the lifecycle and permanent-income models that say consumption should track lifetime resources rather than current income. The formal Ricardian-equivalence proof, and the precise conditions under which it fails, sit in the monetary-and-fiscal-theory chapter.
“Our national debt just crossed \$33 TRILLION. That’s over \$99,000 of debt for every single American. We are committing fiscal child abuse.”
— Sen. Rand Paul, October 2023
Was the 2009 stimulus too small?
The debt-per-citizen panic and the “it should have been bigger” complaint are the same fight from opposite ends. Both turn on the same question Barro raised: how much of a spending dollar actually reaches the economy?
Ricardian equivalence vs. the hand-to-mouth half
“In the absence of distorting taxes, a marginal change in the financing of a given amount of public expenditure — by debt issue versus taxation — has no effect on aggregate demand, interest rates, or capital formation.”
— Robert Barro, “Are Government Bonds Net Wealth?”, Journal of Political Economy, 1974
Barro’s paper is airtight given its assumptions, and that is precisely why it mattered: it became the forcing function for every fiscal-policy paper written afterward. You could no longer assume a mechanical multiplier; you had to say why Ricardian equivalence fails in your model before you could claim any effect at all. Barro is writing from the same Lucas-Sargent rational-expectations turn that produced Stage 1’s Cochrane, the recasting of macroeconomics on neoclassical micro-foundations that the History of Economic Thought book covers in its chapter on the counter-revolution. The theorem doesn’t describe reality. It disciplines every claim about reality that came after it.
“We find that about 50 percent of income accrues to households that consume their current income rather than their permanent income. The behavior of aggregate consumption is well described by a model in which half the population follows a rule of thumb.”
— John Campbell & N. Gregory Mankiw, NBER Macroeconomics Annual, 1989
This is the empirical finding that rescued the fiscal multiplier from Ricardian oblivion. Campbell and Mankiw didn’t dispute Barro’s logic — they showed his key assumption fails for roughly half the economy. A household living paycheck to paycheck cannot smooth consumption against a future tax it has no way to save against; it spends what it gets, now. With half of consumption coming from such households, a targeted tax cut or transfer has a multiplier well above zero, because half the recipients behave nothing like Barro’s infinitely-patient optimizer. Ricardian equivalence is right about the savers and wrong about everyone the savers can’t see.
Where this leaves us
Pure Ricardian equivalence is wrong as a description of reality — too many households are credit-constrained for the savers’ logic to govern the whole economy. But it taught the most important lesson in fiscal policy: the multiplier depends on who gets the money. A dollar to a household that can’t borrow is a dollar spent; a dollar to a wealthy saver is a dollar parked. Targeting is not a detail. It is the variable that moves the multiplier from near zero to comfortably above one. That is consumption theory speaking, not politics.
So the multiplier isn’t zero, but it isn’t the five that Stage 1 promised either. Is fiscal policy even worth the trouble? The answer depends on something we haven’t put on the table yet: what the central bank is doing. And there is one specific condition, rare and dramatic and increasingly common, where fiscal policy stops being one tool among several and becomes the only tool that works.
The ZLB exception
“The Committee decided today to establish a target range for the federal funds rate of 0 to 1/4 percent.”
— Federal Open Market Committee, statement, December 16, 2008
Fourteen words that announced the most powerful central bank on Earth had run out of ammunition. The Fed’s main tool is the interest rate, and it had just driven it to zero. It could not go lower. For the institutional chronology — the Lehman week, this zero-rate moment, the ARRA-and-QE policy mix that followed, and the inflation return of 2021–2023 that closed the era — the Economic History book’s chapter on the 2008 crisis and after is the spine. What matters here is the theory. When the interest rate hits zero, the whole Stage 1 crowding-out story inverts.
In normal times, the central bank and the Treasury are playing tug-of-war. The Treasury spends to stimulate; the central bank, watching inflation, raises rates to cool things back down. Economists call this monetary offset: the Fed cancels out fiscal moves it doesn’t want, so the net multiplier from a spending push is small. This is the deeper reason crowding out bites. It isn’t just private borrowers competing for savings; it’s the central bank actively leaning against the fiscal wind.
The zero lower bound changes the game. Nominal interest rates can’t fall much below zero. If a bank charged you to hold deposits, you’d take your cash and put it under the mattress. So when the economy is so weak that the Fed wants negative rates but can’t deliver them, monetary policy is stuck. And here is the key consequence: at the zero bound, both forces that shrink the multiplier switch off at once.
At the zero lower bound, two things happen simultaneously:
- Crowding out disappears. With slack everywhere and rates pinned at zero, public borrowing doesn’t push rates up; there is no upward pressure to resist. Private investment isn’t displaced.
- Monetary offset disappears. The Fed wants more demand; it isn’t leaning against the fiscal push, it’s cheering for it. It won’t raise rates to cancel the spending, because it’s desperate for exactly the inflation the spending might create.
With both subtractions gone, the multiplier doesn’t just survive; it expands. Christiano, Eichenbaum, and Rebelo (2011) estimated ZLB multipliers of roughly 1.5 to 2.0, against normal-times estimates of 0.6 to 1.0.
Picture a highway. In normal times it’s near capacity, so when the government drives more cars onto it (spending), traffic slows for everyone else (crowding out), and the highway patrol (the Fed) meters the on-ramps to keep things moving. At the zero bound, the highway is nearly empty, because a recession has cleared the traffic, and the patrol has thrown the on-ramp gates wide open and is waving cars through, since an empty highway is the disaster it’s trying to end. Now every car the government adds is pure additional traffic, with nothing pushed off the road. That is why the same spending dollar buys far more output at the zero bound than it does in a boom.
This is the strongest theoretical case for fiscal stimulus: not as a general policy for all seasons, but as a crisis tool deployed precisely when monetary policy is exhausted. The three-equation New Keynesian model that makes this rigorous — and the zero-lower-bound analysis that pins down the multiplier expansion — lives in the economics textbook’s New Keynesian chapter. The intellectual lineage of that model — the long argument from Keynes through the monetarist challenge to the New Keynesian synthesis that finally formalized the ZLB case — is the subject of the History of Economic Thought book’s chapter on the New Keynesian synthesis and the modern monetary consensus.
“The Committee decided today to establish a target range for the federal funds rate of 0 to 1/4 percent.”
— Federal Open Market Committee, December 16, 2008
Is the Fed always in control of the economy?
The folk belief is that the central bank can always steer the economy back to health. December 2008 is the date that belief broke — and the date fiscal policy stopped being optional.
The strongest case for stimulus vs. the ZLB skeptics
“When the nominal interest rate is constrained by the zero bound, the government-spending multiplier can be much larger than one. The multiplier is large precisely in the situations in which output is most depressed.”
— Lawrence Christiano, Martin Eichenbaum & Sergio Rebelo, Journal of Political Economy, 2011
This is the canonical result, derived inside a standard New Keynesian model — not a heterodox one. Christiano, Eichenbaum, and Rebelo showed that the multiplier isn’t a constant; it’s a function of the monetary regime, and it climbs sharply when rates are stuck at zero. Valerie Ramey’s 2019 survey of the empirical literature cross-confirms the structure: state-dependent multipliers, larger in slack-heavy, zero-bound conditions than in booms. The theoretical case and the empirical case point the same direction. Stimulus earns its keep exactly when the economy is most depressed.
“The large-multiplier result depends on the zero bound binding for an extended, expected period. If markets believe a rate hike is around the corner, the result weakens or disappears — and by the time the spending lands, the special conditions may be gone.”
— ZLB-multiplier skeptics, after Valerie Ramey, Journal of Economic Perspectives, 2019
The skeptics’ case is about timing and political economy, and it bites. The large multiplier requires the zero bound to bind for a sustained, expected stretch; if everyone thinks a rate hike is imminent, the expectations channel that drives the result evaporates. And fiscal policy is notoriously slow to legislate, appropriate, and build, so by the time the spending arrives, the special conditions may have passed, leaving ordinary crowding out to claw the multiplier back down. The counter to the counter is the historical record: the post-2008 zero bound lasted roughly seven years, far longer than “temporary,” and the 2020 response showed that fiscal deployment can be fast when the political will exists. The skeptics are right that the conditions are demanding. They’re wrong that the conditions are rare.
Where this leaves us
The zero lower bound is where fiscal policy earns its keep. When the central bank is pinned at zero and committed to staying there, crowding out and monetary offset disappear together, and the spending multiplier expands to something like 1.5 to 2.0. This is a conditional endorsement, not a blank check: the result depends on the bound actually binding for a sustained, expected period, and a credible “rate hike around the corner” can dissolve it before any spending lands. For the central-bank side of this story, what the Fed was doing with its unconventional tools while fiscal policy carried the load, see the companion walkthrough on how central banks actually steer the economy; for the post-2008 monetary regime as the live policy environment, see did economics cause the 2008 crisis.
We’ve built a conditional answer: fiscal policy works, and works hardest at the zero bound. But a radical minority says the whole apparatus of multipliers, crowding out, and Ricardian equivalence is asking the wrong question. What if deficits don’t need to be “paid for” at all? And what if the thing that actually pins down the price level isn’t the central bank, but the government’s budget?
Fiscal theory and the open questions
Stephanie Kelton, the most prominent voice for Modern Monetary Theory, tells Stephen Colbert that the deficit is not what we think it is. The audience laughs, and then stops laughing. MMT and its rival, the Fiscal Theory of the Price Level, are the live theoretical edges of this whole debate: the places where the apparatus from Stages 1 through 3 either gets reframed or gets challenged outright.
Two frameworks sit at the edge of the discipline, and they disagree with each other as sharply as either disagrees with the mainstream. Start with Modern Monetary Theory. Its core claim is the one from Stage 1, taken to its conclusion: a government that issues its own currency cannot involuntarily run out of it. The real constraint is not finance but inflation: you can keep spending until you run out of idle resources, at which point prices rise. Taxes, in this view, don’t “fund” spending; they regulate demand by pulling money back out of the economy. The deficit is just an accounting residual, the difference between what the government put in and what it taxed back.
The Fiscal Theory of the Price Level (Cochrane, Leeper, Sims) starts from the same government budget but draws the opposite conclusion. It says the price level adjusts to make the real value of government debt equal the present value of future surpluses. If the government runs deficits it never plans to repay with surpluses, the price level rises until the real debt is back in line. Deficits don’t just risk inflation, they cause it, through a channel that has nothing to do with the money supply.
The FTPL valuation equation makes the claim precise. The real value of government debt equals the expected present value of future primary surpluses:
$$\frac{B_t}{P_t} = \sum_{s=0}^{\infty} \beta^s\, E_t\big[\,\text{surplus}_{t+s}\,\big]$$If expected surpluses fall while the nominal debt $B_t$ is fixed, the price level $P_t$ must rise to keep the equation balanced. The price level is determined by fiscal expectations, not by the central bank’s money supply — which is exactly the claim that puts FTPL at odds with the standard monetary-dominance regime.
Think of government bonds like shares in a company, and the “dividends” are future budget surpluses. If investors stop believing the government will ever run surpluses to back its debt, the bonds are worth less in real terms, and since the bonds are denominated in dollars, the only way for their real value to fall is for the dollar itself to buy less. The price level is the share price of the government’s fiscal credibility. Lose the credibility, and the price level reprices, exactly as a stock would.
So the two heterodox frameworks split cleanly: MMT says deficits are fine until inflation shows up, and then you tax it back; FTPL says deficits are the inflation, transmitted through the price level the moment expectations about future surpluses shift. The mainstream apparatus is laid out across the economics textbook’s monetary-and-fiscal-theory chapter: the government budget constraint, Ricardian equivalence revisited, the FTPL formalism itself, and the fiscal-multiplier estimates. And the deeper disagreement here, whether the price level is anchored by an active monetary policy or by the fiscal regime, is the live lineage contest the History of Economic Thought book stages in its chapter on the New Keynesian synthesis and the modern monetary consensus, where the standard monetary-dominance regime is exactly the orthodoxy FTPL argues against.
“The federal government can’t run out of money. It can’t be forced to default. The only relevant constraint is inflation — the limit is the economy’s real productive capacity, not the size of the deficit.”
— Stephanie Kelton, The Deficit Myth, 2020
Is MMT right about deficits?
MMT’s diagnosis — that a currency issuer faces a resource limit, not a financial one — is more right than the mainstream long admitted. Its prescription is where the trouble starts.
MMT vs. the fiscal theory of inflation
“The deficit isn’t evidence of overspending. It might be evidence of underspending. The right question is never ‘how will we pay for it?’ but ‘will it stoke inflation?’”
— Stephanie Kelton, The Deficit Myth, 2020
Kelton’s strongest empirical card is 2020: a deficit-financed response that drove unemployment down from 14.7 percent to 3.5 percent in 28 months, with no financing crisis, exactly as the MMT framework predicted. The honest accounting has to include the other half of the ledger, and Kelton’s critics press it hard: the same era produced 9.1 percent inflation by mid-2022. An MMT defender answers that the inflation was largely supply-shock driven, from energy and supply chains, and that the framework predicts inflation when capacity binds; the failure was in the political response, not the diagnosis. Whether that answer holds is the live experiment we’re inside.
“Inflation, when it comes, is not caused by money alone or by a wage-price spiral. It comes when people don’t expect the government to repay its debts — when fiscal policy loses its anchor. Inflation is, in this sense, always and everywhere a fiscal phenomenon.”
— John Cochrane, The Fiscal Theory of the Price Level, 2023
Cochrane’s FTPL is the sharpest theoretical rival to both MMT and the monetarist orthodoxy. He reads 2021–2022 as a near-textbook fiscal-inflation episode: a multi-trillion-dollar deficit with no credible plan for future surpluses, followed by a price-level adjustment exactly as his valuation equation predicts. It is a genuinely different mechanism from “too much money chasing too few goods.” The qualification that keeps FTPL honest: it generates distinctive, testable predictions only under a specific regime, active fiscal policy paired with passive monetary policy. When the central bank is willing and able to defend the price level by raising rates, as the Fed did in 2022, the standard monetary-dominance story reasserts itself and FTPL’s distinctive predictions fade. FTPL is right that the fiscal-monetary interaction matters more than standard models admit; it’s an open question how often its specific regime actually holds.
Where this leaves us
MMT is right that currency-issuing governments face different constraints than households: resources, not solvency. FTPL is right that the fiscal-monetary interaction matters more than standard models typically admit. But neither framework replaces the core insights from Stages 1 through 3. The multiplier is real and conditional; targeting decides its size; the zero bound is where it peaks; and the inflation constraint is the binding one once capacity is reached. MMT’s prescription overestimates political responsiveness; FTPL’s distinctive predictions hold only under a specific regime. Which way the 2020s resolve the question — was the post-2020 inflation a vindication of the fiscal hawks, the MMT framework working as designed, or an FTPL repricing? — is the live experiment we’re all inside of right now. For the deeper question of what money is and how its regime changes across eras, the companion walkthrough on what money actually is is the closest neighbor.
Where this leaves us
We started with a viral claim that a government which prints its own money can spend freely. The textbook answer arrived first, and it was a real number: the fiscal multiplier, $1/(1-c)$, the engine that turns a spending dollar into more than a dollar of output. But every stage took something away from the unconditional version. Crowding out shrinks the multiplier below its naive ceiling. Ricardian logic says a dollar to a saver is nearly wasted, which means the multiplier depends on who gets the money. The zero lower bound says the multiplier expands precisely when monetary policy is exhausted. And the fiscal-theory frameworks at the edge say the whole thing might be reframed once you ask what actually anchors the price level. The honest answer is not yes or no. It is a list of conditions.
Four conditions decide whether a spending dollar helps:
- The credit-constrained share. A dollar to a hand-to-mouth household is spent; a dollar to a wealthy saver is parked. The roughly 50 percent hand-to-mouth share rescues the multiplier from Ricardian collapse.
- The monetary regime. In normal times the central bank offsets fiscal moves and crowding out bites, holding the multiplier near 0.6 to 1.0. At the zero bound both forces switch off and it climbs to 1.5 to 2.0.
- Targeting. Stimulus aimed at credit-constrained households has an MPC near one; broad tax cuts to savers approach the Ricardian zero. The composition of a bill can matter more than its size.
- The fiscal-monetary regime. Once real capacity binds, the inflation constraint is the one that holds, and whether MMT’s “tax it back” or FTPL’s price-level adjustment describes the response is the unresolved frontier.
So does government spending help the economy? Yes, conditionally, and the conditions are the whole answer. The mainline story from Stages 1 through 3 is the most reliable guide we have: real multipliers, larger when targeted and larger still at the zero bound, smaller when the money goes to people who save it or when the economy has no slack left to absorb it. The heterodox edges sharpen the question without overturning it. The next time someone tells you “stimulus always works” or “deficits are always reckless,” you have the tools to push past both slogans to the conditions each one ignores.