The Government Spending Multiplier: Less Bang for the Buck

A survey of the empirical literature on the spending multiplier

As European governments increase military spending in response to the Trump administration's freeze on financial support for Ukraine's defense, an unexpected question has resurfaced: How much will this new spending boost the economy?

The question of how much bang for the buck government spending reaps for the private economy is one that economists have been debating for several decades. My newly published working paper, The Government Spending Multiplier: A Survey of Empirical Literature, examines model assumptions, theoretical innovations, state-dependent factors, and dataset choices within the literature.

In simple terms, the fiscal multiplier is a method of measuring the effect of government spending on the nation’s economic output, or gross domestic product (GDP). A multiplier of 1.00 means that every additional $1 of government spending increases GDP by an equal amount ($1). A multiplier of 1.50 means that every additional $1 of government spending increases GDP by a larger amount ($1.50).

During the COVID-19 pandemic and subsequent economic downturn, many economists argued that the fiscal multiplier of government stimulus spending was very large. Paul Krugman claimed that “based on the experience of the past decade, the multiplier would probably be around 1.5,” while Federal Reserve Bank economists noted that “the multiplier may be as high as 1.5 to 2.0.”

My newly published working paper reviews the model assumptions, theoretical innovations, state-dependent factors, and dataset choices from 67 empirical studies. Multipliers are generally within the range of 0.50 to 0.90, with higher estimates during economic slack and at the zero lower bound (ZLB) and lower estimates for regimes with high public-debt ratios. These findings are broadly consistent with the literature of Valerie Ramey, who finds that the bulk of estimates for average spending multipliers lies in a narrow range of 0.60 to 1.00. They are also largely consistent with recent meta-analyses of large datasets.

Why, then, during economic crises do so many economists claim that spending multipliers are 1.50 or 2.00? One explanation is an over-reliance on a handful of studies that find misleading high multipliers based on subtle assumptions. For example, some studies use linearized models, which result in estimates as high as 1.50 or 2.00. When data-consistent assumptions are used instead, multipliers tend to max out at around 0.80 or 0.90.

Other studies that find multipliers as high as 2.00 using a cross-sectional approach tend to significantly overestimate national-level multipliers. Once weighted by initial state population and total spending, the multiplier falls to less than 0.90. What’s more, many studies overlook the notion of rational forward-looking agents who optimize their consumption over time, contradicting the Ricardian equivalence principle.

Another factor explaining the tendency of economists to cite high multiplier estimates is the fact that the degree of state dependence is more modest than suggested by earlier research. For example, ad hoc conversion facts bias estimated multipliers upward during recessionary periods. Correcting for these biases lowers the estimated spending multiplier from as high as 2.00 in some specifications to less than 0.80.

When interest rates are near zero (the zero lower bound, or ZLB), some economists argue that government spending has a larger impact because monetary policy can't counteract it by raising rates. However, the most robust empirical studies find little evidence that fiscal multipliers are significantly higher at the ZLB, with most estimates staying below 1.00. While a few models suggest larger multipliers, they often rely on assumptions that may overstate the effect of government spending.

High levels of public debt can blunt the impact of fiscal policy because the debt crowds out private investment and increases long-run interest rates. Fiscal multipliers have been found to be smaller in heavily indebted economies, with some estimates even turning negative over long horizons. While low-debt countries will typically have multipliers between 0.60 and 1.00, highly indebted countries will often face multipliers below 0.40 or even negative, which suggests fiscal stimulus is less potent with higher levels of debt.

Some economists argue that government investment spending, particularly on infrastructure, should have higher fiscal multipliers than consumption spending because it boosts both short-term demand and long-term productivity. Although some theoretical models predict higher fiscal multipliers for government investment spending compared with consumption, robust empirical data often shows minimal differences, with investment multipliers frequently constrained by significant crowding out of private investment.

In the aggregate, government spending multipliers broadly fall within the range 0.50 to 0.90—the degree of state dependence is more modest than suggested by earlier research. As policymakers increasingly turn to fiscal stimulus during periods of economic uncertainty, understanding the nuanced effects of government spending on economic output remains critical.

You can find the full paper here.