The strength of fiscal multipliers and spillovers have been the subject of intense debate and considerable empirical research in recent years. An important part of this literature (Chodorow-Reich 2018) has provided estimates of ‘local’ fiscal multipliers, based on the effects of differential fiscal shocks at lower levels of aggregation within larger entities (e.g. states within the US or countries within the EU or the OECD). But translating local multipliers into national ones is not straightforward, in part because of the potential for large fiscal spillovers, especially among entities that are strongly integrated economically.
In a recent paper (Auerbach et al. 2019), we contribute to the literature on local fiscal multipliers while also providing estimates of the strength of fiscal spillovers, both geographic and across industries. We do this by taking advantage of a unique, detailed dataset on US Department of Defense (DOD) contracts that allows us to shed considerable light on the relationship between government spending and local economic activity. The dataset contains information on the zip code and industry of the contractor and the zip code of primary place of work performance. Using information on each contract’s amount and duration, we construct contract-level, time-specific measures of DOD outlays. We aggregate these contracts to construct city-level measures of DOD spending since 1997 and combine this information with disaggregate data on output, earnings, and employment to examine the direct effects of DOD spending and spillovers across industries and geographies.
A recent influential paper by Cohen et al. (2011) finds that an influx of federal funds into a congressional district, attributable to likely exogenous political shifts, causes a reduction in private-sector investment in that location. They propose a mechanism whereby the local spending reduces labour available to the private sector, which drives down investment. It remains to be determined whether an influx of federal funds (such as defence contracts) indeed crowds out local labour supply faced by the private sector (either in other industries or nearby locations).
Spending shocks in the Dallas-Fort Worth metropolitan area
Dallas and Fort Worth are part of the fourth largest metropolitan area in the US, and approximately equal in size. In 2001, Fort Worth received a huge government defence spending shock when, on 26 October, the DOD announced it was awarding Lockheed Martin a $200 billion contract to build a new fighter jet.
Panel A of Figure 1 shows that before this contract was awarded, Dallas and Fort Worth received similar DOD funding. After 2001, there is a sharp divergence in military funding, with Fort Worth receiving approximately $100 billion (in 2001 prices) cumulatively between 2001 and 2015. In contrast, Dallas received approximately $40 billion over the same period. Panel B of Figure 1 plots the dynamics of employment (in logs, normalised to zero in year 2001). The pre-contract trends were similar for Dallas and Fort Worth. After 2001, a large gap in employment opens across the two cities. We observe similar dynamics for normalised log earnings (Panel C). Since no other major shock hit the cities differentially, we can attribute these differential dynamics to the government spending shock, consistent with a positive effect of government spending on the economy of the location receiving the spending shock.
While suggestive that defence spending has a large impact, this case study does not allow us to distinguish direct multipliers from spillover effects, i.e. the extent to which the growth in Fort Worth relative to Dallas reflects net overall growth. Such simple analysis also excludes other potential explanations for differential growth in the two cities. Therefore, we rely on a more systematic empirical analysis that allows us to distinguish such factors.
Figure 1 Government spending, employment and earnings in Dallas and Fort Worth
Notes: The figure plots time series of employment, earnings, and government spending for Dallas, TX and Fort Worth, TX. Employment and earnings series are in logs and normalized to zero in year 2000. Employment and earnings series are seasonally adjusted. Government spending is cumulative since 1997, in billions of 2001 dollars.
Our baseline estimates imply that a dollar of DOD spending in a city increases GDP in that city by a dollar and increases labour earnings by $0.35, and that an increase of DOD spending equal to a percent of local earnings increases employment by 0.2%. These estimates are close to the city-level multiplier estimates from Demyanyk et al. (2018), who focus on a purely cross-sectional setting. We believe that these estimates imply substantial within-location positive spillovers. The spending shocks also have positive effects on nearby localities, increasing earnings in proximate cities by about half of the own-city effect and increasing GDP in other cities across the same state by between half and a whole of the own-city effect. Our estimated state-level GDP multiplier effect of around 1.5 is consistent with the state-level estimates in Nakamura and Steinsson (2014). The positive geographic spillovers imply that any negative spillover effects that operate through factor markets (e.g. pulling in labour from nearby locations) are outweighed by positive demand spillovers (e.g. input-output linkages or induced consumer spending). These findings imply that the increase in local demand is not accommodated by a net reallocation of labour from nearby locations.
Rather than pulling in labour from nearby locations, it is possible that defence spending pulls in labour from other local industries. To understand how defence spending spills over across local industries, we compute three distinct industry-location-specific measures: (1) direct spending from the DOD on goods produced by an industry, (2) indirect DOD spending on that industry based on input requirements of other industries in that location that receive DOD spending, and (3) total DOD spending in other industries, net of input requirements for the observed industry.
Using these measures, we examine industry-level output multipliers. The first measure yields an estimate of the value added by the local industry’s workers (if using the earnings measure) and capital (when examining GDP). The second measure yields an estimate of the local sourcing for DOD contracts and is analogous to estimates of backward linkages previously explored in the context of foreign direct investment (e.g. Gorodnichenko et al. 2014). An estimate of zero would imply that industries source intermediate inputs almost exclusively from other locations, as would be expected in a world with no trade costs, while positive estimates signal the importance of local sourcing. The third measure yields an estimate of what we interpret as general equilibrium effects, which include indirect effects such as reallocation of production factors or positive demand externalities other than through direct backward linkages.
We find that local industries benefit from all three forms of spending. DOD spending has positive direct effects on recipient industries, positive effects on other local industries that supply intermediate inputs, and positive effects on industries that are not directly linked to the recipient industry. Perhaps most surprisingly, net general equilibrium effects are positive. A given industry with no direct production linkages does not (on average) experience a decline in labour when the DOD purchases goods or services from other industries. We also examine the effects of each form of industry-level spending on industries in nearby locations and find similarly positive effects. In summary, we find no evidence that average spillovers are negative, either across industries or within industries across nearby geographies.
Taken in isolation, each piece of evidence we provide is consistent with demand-induced spillovers that operate either through input linkages, consumer spending, or even reductions in borrowing costs (perhaps due to perceptions of lower default risk). But taken together and placed within a general equilibrium framework, the body of evidence is quite puzzling in that net employment effects (across industries and locations) are all positive. Increased production (induced by DOD spending) in an industry and location does not require (on average) an outflow of labour from other industries or locations.
Our evidence suggests that the economy features slack (and hence private activity is not crowded out by public purchases) on average over the business cycle. A natural follow-up question is whether multipliers are higher during periods of higher slack. We find that multipliers are substantially higher for cities experiencing unemployment above their own sample 25th percentile level, but with no additional strength of multipliers increases as slack increases beyond the 25th percentile. One interpretation of these results is that multipliers are high when there is some slack, but beyond that the amount of slack doesn’t matter (perhaps short-run aggregate supply curves are relatively flat and rapidly steepen only as the economy approaches capacity). The fact that the threshold level of unemployment is around the 25th percentile suggests that multipliers on average are high and don’t lead to crowding out. This is consistent with our industry-level results that, on average (across the business cycle) there is no systematic evidence of crowding out.
Overall, our results point to strong positive spillovers, both within and across locations. While our findings are contrary to conventional wisdom and the predictions of neoclassical theory, our evidence points to the relevance of Keynesian-type models that feature excess capacity, either at the firm level or the level of individual workers (e.g. Michaillat and Saez 2015, Murphy 2017). These models even appear relevant for low-frequency (annual) demand shocks, suggesting that excess capacity could be more than a temporary phenomenon.
Auerbach, A J, Y Gorodnichenko, and D Murphy (2019), “Local Fiscal Multipliers and Fiscal Spillovers in the United States,” NBER Working Paper No. 25457.
Chodorow-Reich, G (2018), “Geographic Cross-Sectional Fiscal Spending Multipliers: What Have We Learned?”, forthcoming in American Economic Journal: Economic Policy.
Cohen, L, J Coval, and C Malloy (2011), “Do Powerful Politicians Cause Corporate Downsizing?” Journal of Political Economy 119(6): 1015-1060.
Demyanyk, Y, E Loutskina, and D Murphy (2018), “Fiscal Stimulus and Consumer Debt”, forthcoming in Review of Economics and Statistics.
Gorodnichenko, Y, J Svejnar, and K Terrell (2014), “When Does FDI Have Positive Spillovers? Evidence from 17 Transition Market Economies,” Journal of Comparative Economics 42(4): 954-969.
Michaillat, P and E Saez (2015), “Aggregate Demand, Idle Time, and Unemployment.” Quarterly Journal of Economics 130 (2): 507-569.
Murphy, D P (2017), “Excess Capacity in a Fixed-Cost Economy.” European Economic Review 91: 245-260.
Nakamura, E and J Steinsson (2014), “Fiscal Stimulus in a Monetary Union: Evidence from U.S. Regions”, American Economic Review 104(3): 753-792.