For decades, the textbook answer to whether fiscal stimulus works was a qualified no. Government spending might lift output in the short run, but higher interest rates would discourage private investment and leave the net effect modest. The fiscal multiplier, the change in output per dollar of public spending, was widely thought to sit somewhere below one.
The 2008 financial crisis complicated that view. Policy rates were pinned near zero across the United States, Europe, and Japan. Central banks could no longer cut rates to fight the downturn. For the same reason, they had no reason to raise rates in response to fiscal expansion. The mechanism that normally neutralises stimulus had been switched off.
This raises a structural question rather than an ideological one. If the size of the multiplier depends on what the central bank does, then fiscal policy has no fixed potency. Its effectiveness is conditional. Anyone trying to anticipate how economies will respond to policy in the next downturn needs to understand those conditions in theory, in the data, and in how institutions behave.
Crowding Out vs Crowding In
Start with the conventional channel. When government increases spending, aggregate demand rises, output expands, and inflationary pressure builds. A central bank following something like a Taylor rule responds by raising its policy rate. Higher rates make borrowing more expensive for firms and households, so private investment and spending on durable goods fall. In an open economy, higher rates also attract capital inflows. The currency strengthens and net exports decline.
This is crowding out, and its size depends largely on how aggressively the central bank leans against the fiscal impulse. Suppose monetary policy fully offsets the demand boost to keep inflation on target. Then the multiplier can fall well below one, because public spending largely replaces private spending. Milton Friedman's skepticism about discretionary fiscal policy rested partly on this logic: in his view, the monetary authority, not the treasury, ultimately governs nominal demand.
At the zero lower bound, the chain breaks. The central bank would like rates below zero but cannot push them meaningfully lower, so a fiscal expansion does not trigger tightening. In New Keynesian models, higher spending also raises expected inflation. With nominal rates fixed, rising inflation expectations lower real interest rates, which are what matter for investment and saving decisions. Private spending may then rise alongside public spending rather than falling. This is crowding in.
Christiano, Eichenbaum, and Rebelo (2011) and Woodford (2011) formalised this result. Their models show multipliers that can exceed two while the economy is stuck at the bound. The critical caveat is timing: the effect is strongest when spending occurs while the constraint actually binds. Spending that arrives after rates have lifted off meets the conventional monetary response and its usual offsets. A programme that takes three years to implement may land in a very different policy regime than the one that justified it.
TakeawayA fiscal multiplier is not a property of government spending itself. It is a property of the monetary regime that spending meets.
What the Empirical Evidence Shows
Testing this theory is harder than stating it. Zero-lower-bound episodes are rare and tend to coincide with severe recessions. Fiscal decisions are rarely random either, because governments spend more precisely when economies are weak. Separating the effects of slack, of the monetary constraint, and of the spending itself requires careful identification. Researchers typically rely on military spending shocks or on narrative records of policy announcements.
Some evidence supports the idea that the multiplier depends on economic conditions. Auerbach and Gorodnichenko (2012) estimated US multipliers of roughly 1.5 to 2 during recessions, compared with figures well below one in expansions. Miyamoto, Nguyen, and Sergeyev (2018) studied Japan's long stretch near zero rates and found multipliers around 1.5 at the bound, versus roughly 0.6 outside it. Blanchard and Leigh (2013) showed that forecasters systematically underestimated the output cost of European austerity after 2010. That pattern is consistent with multipliers larger than assumed at a time when monetary offsets were limited.
Other work is more cautious. Ramey and Zubairy (2018) used US data stretching back to the late nineteenth century. They found multipliers generally between about 0.6 and 1, with little robust difference between periods of high and low slack. Their zero-bound estimates were sensitive to specification and rose above one only in certain samples, such as those excluding the World War II years. Results also hinge on whether spending was anticipated, since markets often react to announcements well before money is spent.
A balanced reading is that the evidence leans toward larger multipliers when monetary policy is constrained, but less dramatically than the boldest models suggest. Estimates of 1 to 1.5 at the bound appear more defensible than estimates of 3. That still matters for policy design. A dollar of spending that adds a dollar and a half of output behaves very differently from one that adds sixty cents.
TakeawayWhen theory predicts a large effect and the data show a moderate one, the gap itself is informative. It often points to frictions, timing problems, or expectations the model left out.
Policy Coordination and Overall Effectiveness
If the multiplier depends on the central bank's response, then fiscal and monetary policy are not separate levers. They form a joint system. The relevant question is not simply how large the fiscal multiplier is, but how large it is given this particular central bank's reaction function. Suppose the central bank credibly commits to keeping rates low for an extended period. That commitment effectively prolongs the zero-bound regime and can raise fiscal potency even after recovery begins.
Expectations are central to this interaction. Forward guidance promising low future rates reinforces crowding in by keeping real rates depressed. If instead markets expect the central bank to tighten as soon as recovery appears, stimulus loses much of its force before it reaches the economy. The 2009 American Recovery and Reinvestment Act arrived while the Federal Reserve was signalling extended accommodation. That alignment, intentional or not, plausibly strengthened the Act's impact.
Coordination carries risks as well. The pandemic response of 2020 and 2021 combined large fiscal transfers with near-zero rates and extensive asset purchases. Output recovered quickly, but inflation climbed to levels unseen in four decades, and central banks eventually tightened sharply. The episode showed that crowding-in logic has limits. Once supply constraints bind and the economy nears capacity, the coordination that amplifies demand also amplifies inflation.
There is also an institutional tension. Central bank independence exists partly to prevent monetary policy from accommodating deficits indefinitely, a situation economists call fiscal dominance. Any coordination at the zero bound therefore needs to be framed as temporary and conditional. Otherwise it risks eroding the credibility that gives forward guidance its power. The design challenge is to allow cooperation when constraints bind while preserving separation when they do not.
TakeawayFiscal and monetary policy are best understood as a single system with two operators. The effect of either depends on what the other is expected to do next.
The zero lower bound reveals something general about macroeconomic policy: its effects depend on the regime in which it operates. Fiscal multipliers are not fixed constants waiting to be measured. They emerge from the interaction between spending, interest rates, and expectations.
The evidence suggests stimulus is more effective when monetary policy cannot or will not offset it, though the margin is closer to modest than spectacular. Timing, credibility, and the economy's spare capacity determine whether that added effectiveness actually shows up.
For anyone tracking the next downturn, the size of a spending package alone is not the most useful signal. The monetary environment it enters matters more: where rates sit, what the central bank has promised, and how close the economy is to its limits.