A complete list of all papers by Bill Dupor is available here.
Recently Accepted and Published
A Cup Runneth Over: Fiscal Policy Spillovers from the 2009 Recovery Act
Co-authored with Peter B. McCrory (Univ. of California--Berkeley).
The Economic Journal, June 2018, Vol. 128, Issue 611, 1476-1508; Data used in paper; Online appendix.
Click here for penultimate (but publicly available) version.
Abstract. This paper studies the effects of interregional spillovers from the government spending component of the American Recovery and Reinvestment Act of 2009 (the Recovery Act). Using cross-county Census Journey to Work commuting data, we cluster U.S. counties into local labor markets, each of which we further partition into two subregions. We then compare differential labor market outcomes and Recovery Act spending at the regional and subregional levels using instrumental variables. Our instrument is the sum of spending by federal agencies not instructed to allocate Recovery Act funds according to the severity of local downturns. Among pairs of subregions, we find evidence of fiscal policy spillovers. According to our benchmark specification, $1 of Recovery Act spending in a subregion increases its own wage bill by $0.64 and increases the wage bill in its neighboring subregion by $0.50 during the first two years following the act's passage. We find similar spillover effects when we replace the wage bill with employment as our measure of economic activity. The spillover effect occurs in the service sector, whereas the direct effect occurs in both the services and goods producing sector
The Economic Journal, June 2018, Vol. 128, Issue 611, 1476-1508; Data used in paper; Online appendix.
Click here for penultimate (but publicly available) version.
Abstract. This paper studies the effects of interregional spillovers from the government spending component of the American Recovery and Reinvestment Act of 2009 (the Recovery Act). Using cross-county Census Journey to Work commuting data, we cluster U.S. counties into local labor markets, each of which we further partition into two subregions. We then compare differential labor market outcomes and Recovery Act spending at the regional and subregional levels using instrumental variables. Our instrument is the sum of spending by federal agencies not instructed to allocate Recovery Act funds according to the severity of local downturns. Among pairs of subregions, we find evidence of fiscal policy spillovers. According to our benchmark specification, $1 of Recovery Act spending in a subregion increases its own wage bill by $0.64 and increases the wage bill in its neighboring subregion by $0.50 during the first two years following the act's passage. We find similar spillover effects when we replace the wage bill with employment as our measure of economic activity. The spillover effect occurs in the service sector, whereas the direct effect occurs in both the services and goods producing sector
Local and Aggregate Fiscal Policy Multipliers
Co-authored with Rodrigo Guerrero (Yale University)
Journal of Monetary Economics, Dec. 2017, Vol. 92, pp. 16-30. Data used in paper; Online appendix.
Click here for penultimate (but publicly available) version.
Abstract. Using a newly constructed panel of state-level defense contracts, this paper studies the effect of defense spending on the U.S. macroeconomy. Summing observations across states, we estimate aggregate income and employment multipliers. Comparing these to local multipliers estimated with the panel provides evidence that local multiplier estimates may be reliable indicators of fiscal policy's aggregate effects. Furthermore, evidence of small positive spillovers is found. Across several speci cations, we estimate income and employment multipliers between zero and 0.5. This result is reconciled with the greater-than-one multipliers found in Nakamura and Steinsson (2014) by analyzing the Korean War years' impact on the
estimation.
Journal of Monetary Economics, Dec. 2017, Vol. 92, pp. 16-30. Data used in paper; Online appendix.
Click here for penultimate (but publicly available) version.
Abstract. Using a newly constructed panel of state-level defense contracts, this paper studies the effect of defense spending on the U.S. macroeconomy. Summing observations across states, we estimate aggregate income and employment multipliers. Comparing these to local multipliers estimated with the panel provides evidence that local multiplier estimates may be reliable indicators of fiscal policy's aggregate effects. Furthermore, evidence of small positive spillovers is found. Across several speci cations, we estimate income and employment multipliers between zero and 0.5. This result is reconciled with the greater-than-one multipliers found in Nakamura and Steinsson (2014) by analyzing the Korean War years' impact on the
estimation.
So, Why Didn't the 2009 Recovery Act Improve the Nation's Highways and Bridges?
Federal Reserve Bank of St Louis Review, Second Quarter 2017, 99(2), pp. 169-182.
Abstract. Although the American Recovery and Reinvestment Act of 2009 (the Recovery Act) provided nearly $28 billion to state governments for improving U.S. highways, the highway system saw no significant improvement. For example, relative to the years before the act, the number of structurally deficient or functionally obsolete bridges was nearly unchanged, the number of workers on highway and bridge construction did not significantly increase, and the annual value of construction put in place for public highways barely budged. The author shows that as states spent Recovery Act highway grants, many simultaneously slashed their own contributions to highway infrastructure, freeing up state dollars for other uses. Next, using a cross-sectional analysis of state highway spending, the author shows that a state’s receipt of Recovery Act highway dollars had no statistically significant causal impact on that state’s total highway spending. Thus, the amount of actual highway infrastructure investment following the act’s passage was likely very similar to that under a no-stimulus counterfactual.
Abstract. Although the American Recovery and Reinvestment Act of 2009 (the Recovery Act) provided nearly $28 billion to state governments for improving U.S. highways, the highway system saw no significant improvement. For example, relative to the years before the act, the number of structurally deficient or functionally obsolete bridges was nearly unchanged, the number of workers on highway and bridge construction did not significantly increase, and the annual value of construction put in place for public highways barely budged. The author shows that as states spent Recovery Act highway grants, many simultaneously slashed their own contributions to highway infrastructure, freeing up state dollars for other uses. Next, using a cross-sectional analysis of state highway spending, the author shows that a state’s receipt of Recovery Act highway dollars had no statistically significant causal impact on that state’s total highway spending. Thus, the amount of actual highway infrastructure investment following the act’s passage was likely very similar to that under a no-stimulus counterfactual.
The 2009 Recovery Act: Stimulus at the Extensive and Intensive Labor Margins
Co-authored with M. Saif Mehkari (University of Richmond)
European Economic Review 85, (2016), pp. 208-228. Click here for penultimate (but publicly available) version.
Abstract. This paper studies the effect of government stimulus spending on a novel aspect of the labor market: the differential impact of spending on the total wage bill versus employment. We analyze the 2009 Recovery Act via instrumental variables using a new instrument, the spending done by federal agencies that were not instructed to target funds towards harder hit regions. We find a moderate positive effect on jobs created/saved (i.e., the "extensive margin") and also a significant increase in wage payments to workers whose job status was safe without Recovery Act funds (i.e., the "intensive margin"). Our point estimates imply that roughly one-half of the wage payments resulting from the act were paid at the intensive margin. To provide a theoretical underpinning for the estimates, we build a micro-founded dynamic model in which a firm meets new government demand with a combination of new hiring and increasing existing workers' average hours. Faced with hiring costs and an overtime premium, the firm responds by increasing hours along both margins. Our model analysis also provides insight into how government spending policy should be structured to lower the cost of generating new jobs. Finally, we catalogue survey evidence from Recovery Act fund recipients that reinforces the importance of the intensive labor margin.
European Economic Review 85, (2016), pp. 208-228. Click here for penultimate (but publicly available) version.
Abstract. This paper studies the effect of government stimulus spending on a novel aspect of the labor market: the differential impact of spending on the total wage bill versus employment. We analyze the 2009 Recovery Act via instrumental variables using a new instrument, the spending done by federal agencies that were not instructed to target funds towards harder hit regions. We find a moderate positive effect on jobs created/saved (i.e., the "extensive margin") and also a significant increase in wage payments to workers whose job status was safe without Recovery Act funds (i.e., the "intensive margin"). Our point estimates imply that roughly one-half of the wage payments resulting from the act were paid at the intensive margin. To provide a theoretical underpinning for the estimates, we build a micro-founded dynamic model in which a firm meets new government demand with a combination of new hiring and increasing existing workers' average hours. Faced with hiring costs and an overtime premium, the firm responds by increasing hours along both margins. Our model analysis also provides insight into how government spending policy should be structured to lower the cost of generating new jobs. Finally, we catalogue survey evidence from Recovery Act fund recipients that reinforces the importance of the intensive labor margin.
Sticky Wages, Private Consumption and Fiscal Policy Multipliers
Co-authored with Jingchao Li (East China Univ. of Science and Technology) and Rong Li (Renmin Univ. of China)
Journal of Macroeconomics (forthcoming)
Working paper. This draft: April 2018. First draft: January 2017.
This paper demonstrates how adding nominal wage rigidity to a standard, closed economy sticky price model can by itself create a mechanism by which increases in government spending cause increases in consumption. The increase in output arising from government purchases puts upward pressure on the price level. At a fixed short-run nominal wage, this bids down the real wage, which leads producers to increase labor demand. Increased labor demand allows households to both nance the tax bill associated with the government spending as well as increase their own consumption. Ourapproach does not rely upon existing ingredients for generating large fiscal multipliers, such as the zero lower bound, borrowing constrained households or an interactionbetween consumption and government purchases in the utility function.
Journal of Macroeconomics (forthcoming)
Working paper. This draft: April 2018. First draft: January 2017.
This paper demonstrates how adding nominal wage rigidity to a standard, closed economy sticky price model can by itself create a mechanism by which increases in government spending cause increases in consumption. The increase in output arising from government purchases puts upward pressure on the price level. At a fixed short-run nominal wage, this bids down the real wage, which leads producers to increase labor demand. Increased labor demand allows households to both nance the tax bill associated with the government spending as well as increase their own consumption. Ourapproach does not rely upon existing ingredients for generating large fiscal multipliers, such as the zero lower bound, borrowing constrained households or an interactionbetween consumption and government purchases in the utility function.
The Expected Inflation Channel of Government Spending in the Postwar U.S.
Co-authored with Rong Li (Renmin University).
European Economic Review 74, (2015), pp. 36-56. Click here for penultimate (but publicly available) version.
Abstract. There exist sticky price models in which the output response to a government spending change can be large if the central bank is nonresponsive to inflation. According to this "expected inflation channel," government spending drives up expected inflation, which in turn, reduces the real interest rate and leads to an increase in private consumption. This paper examines whether the channel was important in the post-WWII U.S., with particular attention to the 2009 Recovery Act period. First, we show that a model calibrated to have a large output multiplier requires a large response of expected inflation to a government spending shock. Next, we show that this large response is inconsistent with structural vector autoregression evidence from the Federal Reserve's passive policy period (1959-1979). Then, we study expected inflation measures during the Recovery Act period in conjunction with a panel of professional forecaster surveys, a cross-country comparison of bond yields and fiscal policy news announcements. We show that the expected inflation response was too small to engender a large output multiplier.
European Economic Review 74, (2015), pp. 36-56. Click here for penultimate (but publicly available) version.
Abstract. There exist sticky price models in which the output response to a government spending change can be large if the central bank is nonresponsive to inflation. According to this "expected inflation channel," government spending drives up expected inflation, which in turn, reduces the real interest rate and leads to an increase in private consumption. This paper examines whether the channel was important in the post-WWII U.S., with particular attention to the 2009 Recovery Act period. First, we show that a model calibrated to have a large output multiplier requires a large response of expected inflation to a government spending shock. Next, we show that this large response is inconsistent with structural vector autoregression evidence from the Federal Reserve's passive policy period (1959-1979). Then, we study expected inflation measures during the Recovery Act period in conjunction with a panel of professional forecaster surveys, a cross-country comparison of bond yields and fiscal policy news announcements. We show that the expected inflation response was too small to engender a large output multiplier.
For a thoughtful discussion of the paper by John Cochrane, see his post on his weblog The Grumpy Economist, linked here.
A few highlights from the paper (click on each image to see a larger version):
The Analytics of Technology News Shocks
Co-authored with M. Saif Mehkari (University of Richmond).
Journal of Economic Theory (2014). Click here for penultimate (but publicly available) version.
Abstract. This paper constructs several models in which, unlike the standard neoclassical growth model, positive news about future technology generates an increase in current consumption, hours and investment. These models are said to exhibit procyclical news shocks. We find that all models that exhibit procyclical news shocks in our paper have two commonalities. There are mechanisms to ensure that: (I) consumption does not crowd out investment, or vice versa; (II) the benefit of forgoing leisure in response to news shocks outweighs the cost. Among the models we consider, we believe, one model holds the greatest potential for explaining procyclical news shocks. Its critical assumption is that news of the future technology also illuminates the nature of this technology. This illumination in turn permits economic actors to invest in capital that is forward-compatible, i.e. adapted to the new technology. On the technical side, our paper reintroduces the Laplace transform as a tool for studying dynamic economies analytically. Using Laplace transforms we are able to study and prove results about the full dynamics of the model in response to news shocks.
The American Recovery and Reinvestment Act: Solely a Government Jobs Program?
Co-authored with Tim Conley (University of Western Ontario).
Journal of Monetary Economics (2013). Click here for penultimate (but publicly available) version.
Abstract. This paper estimates the private and government sector employment effects of American Recovery and Reinvestment Act (ARRA) spending via an instrumental variables strategy. We argue that this aid was effectively fungible and states used it to offset declines in revenue. This enables us to use exogenous variation in states' budget positions to identify the Act's employment effects. We also exploit exogenous variation across states in ARRA highway funding. According to our benchmark estimates, average state and local government employment, during the twenty-four months following the program's inception, was between 156 thousand and 563 thousand persons greater as a result of ARRA spending (90% confidence interval). The corresponding estimate for the private sector ranges from a loss of 182 thousand to a gain of 1.1 million jobs. Our point estimate for the implied cost of creating a job lasting one year is $202 thousand, which is substantially larger than the corresponding estimate from the President's Council of Economic Advisors.
Recent Working Papers
A Local-Spillover Decomposition of the Causal Effect of U.S. Defense Spending Shocks
Co-authored with Tim Conley (University of Western Ontario), Mahdi Ebsim (Federal Reserve Bank of St. Louis), Jingchao Li (East China University of Science and Technology) and Peter McCrory (University of California-Berkeley)
Working paper. This draft: June 2020.
This paper decomposes the causal effect of government defense spending into: (i) a local (or direct) effect, and (ii) a spillover (or indirect) effect. Using state-level defense spending data, we show that a negative cross-state spillover effect explains the existing simultaneous findings of a low aggregate multiplier and a high local multiplier. We show that enlisting disaggregate data improves the precision of aggregate effect estimates, relative to using aggregate time series alone. Moreover, we compare two-step efficient GMM with two alternative moment weighting approaches used in existing research.
Working paper. This draft: June 2020.
This paper decomposes the causal effect of government defense spending into: (i) a local (or direct) effect, and (ii) a spillover (or indirect) effect. Using state-level defense spending data, we show that a negative cross-state spillover effect explains the existing simultaneous findings of a low aggregate multiplier and a high local multiplier. We show that enlisting disaggregate data improves the precision of aggregate effect estimates, relative to using aggregate time series alone. Moreover, we compare two-step efficient GMM with two alternative moment weighting approaches used in existing research.
The 2008 U.S. Auto Market Collapse
Co-authored with Rong Li (Renmin Univ.), M. Saif Mehkari (Univ. of Richmond) and Yi-Chan Tsai (National Taiwan University)
Working paper. This draft: January 2020.
New vehicle sales in the U.S. fell nearly 40 percent during the past recession, causing significant job losses and
unprecedented government interventions in the auto industry. This paper explores three potential explanations for this decline: increasing oil prices, falling home values, and falling household income expectations. First, we use the historical macroeconomic relationship between oil prices and vehicle sales to show that the oil price spike explains roughly 15
percent of the auto sales decline between 2007 and 2009. Second, we establish that declining home values explain only a small portion of the observed reduction in household new vehicle sales. Using a county-level panel from the episode, we find (1) a one-dollar fall in home values reduced household new vehicle spending by 0.5 to 0.7 cents and overall new vehicle
spending by 0.9 to 1.2 cents and (2) falling home values explain between 16 and 19 percent of the overall new vehicle spending decline. Next, examining state-level data for 1997-2016, we find (3) the short-run responses of new vehicle consumption to home value changes are larger in the 2005-2011 period relative to other years, but at longer horizons (e.g. 5 years), the responses are similar across the two sub-periods and (4) the service flow from vehicles, as measured by miles traveled, responds very little to house price shocks. We also detail the sources of the differences between our findings (1) and (2) from existing research. Third, we establish that declining current and expected future income expectations potentially played an important role in the auto market's collapse. We build a permanent income model augmented to include infrequent repeated car buying. Our calibrated model matches the pre-recession distribution of auto vintages and the
liquid-wealth-to-income ratio, and exhibits a large vehicle sales decline in response to a mild decline in expected permanent income due to a transitory slowdown in income growth. In response to the shock, households delay replacing existing vehicles, allowing them to smooth the effects of the income shock without significantly adjusting the service flow from their vehicles. Augmenting our model with a richer set of household expectations allows us to match 65 percent of the overall new vehicle spending decline (i.e. roughly the portion of the decline not explained by oil prices and falling home values). Combining our negative results regarding housing wealth and oil prices with our positive model-based findings, we interpret
the auto market collapse as consistent with existing permanent income based approaches to durable goods purchases (e.g., Leahy and Zeira (2005)).
Working paper. This draft: January 2020.
New vehicle sales in the U.S. fell nearly 40 percent during the past recession, causing significant job losses and
unprecedented government interventions in the auto industry. This paper explores three potential explanations for this decline: increasing oil prices, falling home values, and falling household income expectations. First, we use the historical macroeconomic relationship between oil prices and vehicle sales to show that the oil price spike explains roughly 15
percent of the auto sales decline between 2007 and 2009. Second, we establish that declining home values explain only a small portion of the observed reduction in household new vehicle sales. Using a county-level panel from the episode, we find (1) a one-dollar fall in home values reduced household new vehicle spending by 0.5 to 0.7 cents and overall new vehicle
spending by 0.9 to 1.2 cents and (2) falling home values explain between 16 and 19 percent of the overall new vehicle spending decline. Next, examining state-level data for 1997-2016, we find (3) the short-run responses of new vehicle consumption to home value changes are larger in the 2005-2011 period relative to other years, but at longer horizons (e.g. 5 years), the responses are similar across the two sub-periods and (4) the service flow from vehicles, as measured by miles traveled, responds very little to house price shocks. We also detail the sources of the differences between our findings (1) and (2) from existing research. Third, we establish that declining current and expected future income expectations potentially played an important role in the auto market's collapse. We build a permanent income model augmented to include infrequent repeated car buying. Our calibrated model matches the pre-recession distribution of auto vintages and the
liquid-wealth-to-income ratio, and exhibits a large vehicle sales decline in response to a mild decline in expected permanent income due to a transitory slowdown in income growth. In response to the shock, households delay replacing existing vehicles, allowing them to smooth the effects of the income shock without significantly adjusting the service flow from their vehicles. Augmenting our model with a richer set of household expectations allows us to match 65 percent of the overall new vehicle spending decline (i.e. roughly the portion of the decline not explained by oil prices and falling home values). Combining our negative results regarding housing wealth and oil prices with our positive model-based findings, we interpret
the auto market collapse as consistent with existing permanent income based approaches to durable goods purchases (e.g., Leahy and Zeira (2005)).
Regional Consumption Responses and the Aggregate Fiscal Multiplier
Co-authored with Marios Karabarbounis (FRB of Richmond), Marianna Kudlyak (FRB of San Francisco) and M. Saif Mehkari (University of Richmond)
Working paper. New version coming soon.
Working paper. New version coming soon.
The Aggregate and Local Economic Effects of Government Financed Health Care
Co-authored with Rodrigo Guerrero (Yale University).
Economic Inquiry (Accepted for Publication). This draft: April 2020. First draft: January 2017.
Government-financed health care expenditures, through Medicare and Medicaid, have grown from roughly zero to over 7.6 percent of national personal income over the past 50 years. This paper investigates the stimulative effects of Medicare spending. Using an annual, state-level panel, we regress state income growth on own-state spending and spending in other states, instrumented by unanticipated shocks to aggregate Medicare spending, to estimate local and spillover effects. In our benchmark specification, the own-spending multiplier equals 1.3 and the spillover multiplier equals 0.4. The total Medicare spending multiplier (i.e., local plus spillover) is approximately 1.7.
Economic Inquiry (Accepted for Publication). This draft: April 2020. First draft: January 2017.
Government-financed health care expenditures, through Medicare and Medicaid, have grown from roughly zero to over 7.6 percent of national personal income over the past 50 years. This paper investigates the stimulative effects of Medicare spending. Using an annual, state-level panel, we regress state income growth on own-state spending and spending in other states, instrumented by unanticipated shocks to aggregate Medicare spending, to estimate local and spillover effects. In our benchmark specification, the own-spending multiplier equals 1.3 and the spillover multiplier equals 0.4. The total Medicare spending multiplier (i.e., local plus spillover) is approximately 1.7.
Schools and Stimulus
Co-authored with M. Saif Mehkari (University of Richmond)
Working paper. This draft: March 2015. First draft: October 2014.
This paper analyzes the impact of the education funding component of the 2009 American Recovery and Reinvestment Act (the Recovery Act) on public school districts. We use cross-sectional differences in district-level Recovery Act funding to investigate the program's impact on staffing, expenditures and debt accumulation. To achieve identification, we use exogenous variation across districts in the allocations of Recovery Act funds for special needs students. We estimate that $1 million of grants to a district had the following effects: expenditures increased by $570 thousand, district employment saw little or no change, and an additional $370 thousand in debt was accumulated. Moreover, 70% of the increase in expenditures came in the form of capital outlays. Next, we build a dynamic, decision theoretic model of a school district's budgeting problem, which we calibrate to district level expenditure and staffing data. The model can qualitatively match the employment and capital expenditure responses from our regressions. We also use the model to conduct policy experiments.
Working paper. This draft: March 2015. First draft: October 2014.
This paper analyzes the impact of the education funding component of the 2009 American Recovery and Reinvestment Act (the Recovery Act) on public school districts. We use cross-sectional differences in district-level Recovery Act funding to investigate the program's impact on staffing, expenditures and debt accumulation. To achieve identification, we use exogenous variation across districts in the allocations of Recovery Act funds for special needs students. We estimate that $1 million of grants to a district had the following effects: expenditures increased by $570 thousand, district employment saw little or no change, and an additional $370 thousand in debt was accumulated. Moreover, 70% of the increase in expenditures came in the form of capital outlays. Next, we build a dynamic, decision theoretic model of a school district's budgeting problem, which we calibrate to district level expenditure and staffing data. The model can qualitatively match the employment and capital expenditure responses from our regressions. We also use the model to conduct policy experiments.
Additional Research on the 2009 Recovery Act
Creating Jobs Via the 2009 Recovery Act: State Medicaid Grants Compared to Broadly-Directed Spending
Working paper. This draft: November 2013. First draft: May 2013.
Abstract. Researchers have used cross-state differences to assess the jobs impact of the 2009 American Recovery and Reinvestment Act (the Recovery Act). Existing studies find that the Act's broadly-directed spending (i.e. excluding tax cuts) increased employment, at a cost-per-job of roughly three to five times that of typical employment compensation in the U.S. Other research finds that a particular component of the Act---emergency Medicaid grants to states---created jobs at a cost of 12% to 20% that of broadly-directed spending. This paper shows that these differences across the components' impacts can be explained by omitted variables in the existing work on the emergency Medicaid grants. Adjusting for the omissions, the jobs effect of the Act's Medicaid grants becomes substantially weaker. The omissions are: (i) not controlling the degree of (non-Recovery Act) federal dependency, (ii) not duly controlling for pre-Act housing and labor market conditions, and (iii) not conditioning on Recovery Act funding beyond that from the Act's Medicaid grants. Adjusting for any one of these omissions, by itself, results in a substantial increase in the cost of job creation and/or no statistically significant jobs effect.
Abstract. Researchers have used cross-state differences to assess the jobs impact of the 2009 American Recovery and Reinvestment Act (the Recovery Act). Existing studies find that the Act's broadly-directed spending (i.e. excluding tax cuts) increased employment, at a cost-per-job of roughly three to five times that of typical employment compensation in the U.S. Other research finds that a particular component of the Act---emergency Medicaid grants to states---created jobs at a cost of 12% to 20% that of broadly-directed spending. This paper shows that these differences across the components' impacts can be explained by omitted variables in the existing work on the emergency Medicaid grants. Adjusting for the omissions, the jobs effect of the Act's Medicaid grants becomes substantially weaker. The omissions are: (i) not controlling the degree of (non-Recovery Act) federal dependency, (ii) not duly controlling for pre-Act housing and labor market conditions, and (iii) not conditioning on Recovery Act funding beyond that from the Act's Medicaid grants. Adjusting for any one of these omissions, by itself, results in a substantial increase in the cost of job creation and/or no statistically significant jobs effect.
Comments on "Are State Governments Roadblocks to Federal Stimulus? Evidence on the Flypaper Effect from Highway Grants in the 2009 Recovery Act"
Co-authored with Rodrigo Guerrero (Federal Reserve Bank of St. Louis)
Working paper. May 2017.
Working paper. May 2017.
The content of this site does not reflect the views of the Federal Reserve Bank of St. Louis or the Federal Reserve System.
