A federal defense dollar returns roughly a dollar of GDP quickly; an infrastructure dollar returns less at first but three-to-eight times as much within a decade

Summary

The two spending types produce growth on very different clocks, and averaging them into a single number hides the more useful answer. Defense spending delivers close to a dollar of GDP for a dollar spent, fairly quickly — pooled estimates put the multiplier around 0.8 to 1.0 within the first one to two years, rising toward 1.5 when measured at the regional (state) level rather than nationally. Transport infrastructure spending (highways in particular, the best-studied case) does much less in the short run — some estimates put its immediate multiplier near zero, even negative for several years, because of construction delays and the ease of postponing purchases of durable goods — but delivers far more over time: multipliers of roughly 3 to 8 six to eight years after the money is spent, once the roads are actually built and used. So 'more growth per dollar' depends entirely on the horizon: on a 1-2 year horizon, defense wins; on a 5-10 year horizon, infrastructure wins by a wide margin. National USAspending data (shown above) confirms defense obligations dwarf transportation obligations in absolute size (~$1.0-1.45 trillion/year vs. ~$91-270 billion/year, 2017-2025) but that scale difference says nothing about the multiplier — it is included only for context.

What the literature actually finds

Defense/military spending. The IMF's April 2026 World Economic Outlook (Chapter 2, a dedicated analysis of defense-spending macroeconomics across many countries) reports that once endogeneity concerns are addressed (spending is not simply reacting to the business cycle), the average defense multiplier is close to 1, consistent with standard macro model simulations. The multiplier is smaller in countries that import a lot of their weapons (demand leaks abroad), and larger when the spending increase is deficit-financed rather than offset by cuts elsewhere. A Boston Fed working paper pooling military-spending shocks across advanced and developing economies finds a peak cumulative multiplier of about 0.86 two years after the shock, below 1 in developing economies and above 1 in advanced ones. At the U.S. regional level, Nakamura and Steinsson's widely cited 2014 American Economic Review paper, using historical variation in military procurement across states, estimates an 'open-economy relative multiplier' of about 1.5 — larger than the national estimates because a state doesn't experience the offsetting national monetary/tax tightening a national multiplier does.

Transport/highway infrastructure spending. Sylvain Leduc and Daniel Wilson's 2012 study of state highway spending, 1993-2010 (summarized in a Richmond Fed brief drawing on Valerie Ramey's survey), found that highway spending reduced GDP for up to five years after a spending program began — the money is obligated but the road isn't finished and usable yet — and only then began raising GDP at horizons of six to eight years out, with multipliers estimated at 3 or higher, and in some specifications rising toward 6-8 at the peak. Christoph Boehm's 2020 paper offers a mechanism: government investment multipliers run close to zero in the short run (people can defer purchases of durable/investment-type goods much more easily than of everyday consumption), while government consumption multipliers run around 0.8. This is exactly the opposite short-run ranking from what a naive intuition ('build something real, get more bang for the buck') would predict — the payoff is real but delayed, not immediate.

Why we did not compute our own multiplier from AskAmerica data

We pulled USAspending obligations by awarding agency for Department of Defense and Department of Transportation, 2017-2025 (fiscal.usaspending_by_agency), and national real GDP by year from econ.real_gdp_growth, to check scale and look for a usable signal. That gives only 8-9 annual observations at the national level — far too few, and far too confounded (COVID-19, the 2021-2023 inflation surge, and the IIJA's 2022 DOT obligation spike all fall inside this same short window) to identify a causal multiplier by regression. This is exactly the shape of comparison this analysis is built to flag rather than force: a national annual time series with a single-digit number of years cannot separate 'spending caused growth' from 'growth (or war, or a pandemic) caused spending' or from a coincident shock. The literature estimates above, which use decades of data, cross-country variation, or region-level procurement shocks specifically designed to break that reverse-causality problem, are the trustworthy source for the multiplier itself; the AskAmerica pull is shown only to size the two spending streams, not to estimate their effects.

Bottom line

• Short run (1-2 years): defense spending multiplier ~0.8-1.0 (up to ~1.5 regionally); infrastructure multiplier ~0, sometimes negative.
• Long run (6-8+ years): defense multiplier stays roughly flat near 1; infrastructure (highway) multiplier rises to roughly 3-8.
• Caveat that matters most: these are different kinds of 'growth.' Defense multipliers mostly capture demand-side stimulus (more orders, more jobs, more output now). Infrastructure's long-run multiplier captures a supply-side effect — the completed capital stock makes the private economy more productive for years afterward — which is a structurally different, and generally more durable, channel than defense demand stimulus.