We use panel data regressions on US federal and state R&D funding at public universities during 1985-2012 to estimate whether federal academic R&D subsidy is a complement or substitute for state funding of academic research. The results indicate a statistically significant but modest substitution effect. We further find that there is essentially no substitution during periods of rapid growth of federal R&D spending. However, during periods of relatively slow growth in federal funding, the substitution effect is substantial--a one dollar cut below the average growth rate of federal funding leads to an increase of about 65 cents above the average growth rate of state funding.
Keywords: Federal subsidy, higher education, research and development
In the U.S., the federal government is the primary supporter of academic research and development (R&D). However, state governments have increased financial support for academic research since the 1980s in pursuit of technology-based economic development. The total federally financed R&D expenditures by public doctorate-granting universities (2) grew by an annualized rate of 6.8% in 1985-1990, 4.4% in 1990-2000, 5.5% in 2000-2010, and the total federal funding of academic R&D nearly quadrupled in 2012 as compared with 1985. (3) The total state and local government financed academic R&D expenditures also show substantial growth: an annualized rate of 8.6% in 1985-1990, 3.2% in 1990-2000, and 2.5% in 2000-2010. The total state funding grew by 150% in the period 1985-2012.
[FIGURE 1 OMITTED]
The involvement of both federal and state governments in funding academic R&D reveals an important facet of fiscal federalism in the science and technology policy arena. This empirical study is intended to examine the relationship between the funding of public academic R&D from federal and state governments. Because federal funding has been growing more rapidly than state funding of public university R&D, the average state share of total government-financed public university R&D expenditures has declined since 1985 from slightly over 20% in 1985 to 18.5% in 1995 and to about 12.5% in 2010.
We further examine data on the annual growth of state public university R&D expenditures financed by federal and state governments from 1985 through 2012. Our empirical analyses employ panel study econometric techniques to explore the effect of federal grant funding on state funding of public university R&D. The model is applied to all fifty states for the time period 1985-2012. The statistical estimates indicate that there is a statistically significant but modest substitution between federal and state funding of R&D at public higher education institutions. A relatively strong substitution effect mostly appears when the growth of federal support is relatively low. In years when the growth of federal funding is high, state governments maintain virtually the same growth rates of their funding, regardless of changing growth of federal support of academic R&D.
This empirical study is intended to improve understanding of state budgetary responses to the varying growth of federal government grant support of academic research. A substitution effect implies not only a potential crowd-out of federal funding on state appropriations, but also stabilization of total funding through increased state support when federal support dwindles. The substitution response becomes stronger in the case of relatively low growth of federal funding. This may suggest that public universities only actively pursue state funding of academic research when there is pressure to do so because of proportionately slow growth or cuts in their federal funding.
REVIEW OF EMPIRICAL LITERATURE
Given quite limited research of the relationship between federal and state funding of academic research, we first look at the empirical literature on estimating the effect of various other intergovernmental grants to state and local governments. The majority of empirical studies suggest that intergovernmental grants rarely stimulate additional spending from recipients' own sources on targeted programs. For instance, in an early review by Hines and Thaler (1995), eight of the ten most cited empirical studies report that every grant dollar increases the spending of recipient jurisdictions by less than one dollar (as low as 25 cents), which shows that intergovernmental grants very likely crowd out state and local expenditures that would have been made without the grants. This crowd-out effect is consistent with economic theory. Many economic studies have found that the demand for public services is generally price- and income-inelastic (Fisher, 2007). Therefore, a price effect (from a matching grant) or an income effect (from a non-matching grant) will not be large enough to increase demand by the amount of the grant. As a result, a fraction of the grant is diverted to support other government programs or to reduce taxes. The crowd-out effect of federal aid on state funding exists in many federal grant programs. Following is a summary of literature on three specific federal grant programs: Title 1 educational funding; alcohol, drug abuse, and mental health block grants; and highway aid. (4)
Federal involvement in education finance has been quite modest compared to state and local expenditures. Most of the federal grants for education have been categorical grants targeted for specialized education programs. The largest federal education grant program is Title I, which has provided compensatory education to children from low-income families since 1965. (5) Feldstein (1978) analyzed the total education expenditures of 4,690 school districts across the country in 1970. He reports that for every additional dollar of Title 1 federal aid, a school district increases its total educational expenditures by 72 cents. Using annual financial data at the school district level from 1991 to 1995, Gordon (2004) estimated the effect of a change in the grant amount on the change in targeted expenditure at the state or local level. The article finds that Title I grants initially boost total school district revenue and instructional spending about dollar for dollar. However, by the third year following Title I changes, the effects are no longer significantly positive because local government reactions counter the effects of Title I grant money.
Another federal grant program is the alcohol, drug abuse, and mental health (ADAMH) block grant. This federal grant was created in 1981 to support state programs to address issues of alcoholism, drug abuse, and mental health. Concerned about the possible substitution effect motivated the federal government to impose state maintenance of effort requirements. A study by Jacobsen and McGuire (1996) suggests that federal alcoholism and drug abuse (ADA) grants were used by states almost entirely as a substitute for states' own funding of substance abuse prevention and treatment services before 1990. However, they find that the federal ADA grant had a strong positive effect on states' own spending on substance abuse programs after 1990, when the federal government stepped up its efforts to enforce the requirements of state funding levels. A later study by Gamkhar and Sim (2001) reports no statistically significant effects of the current year ADA grant on state and local substance abuse expenditures, even after 1990. However, they find that the one-period-lagged ADA grant had a large effect on state and local government spending throughout the period of 1985-1995: on the margin, a dollar increase in ADA block grant generated an additional 80 cents in total state and local government spending.
Federal highway aid is among the largest federal grant programs to the states, providing billions of dollars to support state highway construction and maintenance. Federal highway grants are made through the Federal Highway Trust Fund, financed by a federal tax on gasoline sales. Using a panel data set of 45 states in the U.S. for the years 1976-1990, Gamkhar (2000) shows that when federal highway aid increases for three consecutive years, every $1 increase in federal highway grant obligations leads to an 87 cent increase in the highway expenditures of state and local governments.
After controlling for endogeneity of federal grants, Knight (2002) reports that the impact of a dollar of federal highway grants on state highway spending varies from 12 to 33 cents. Knight's findings indicate that federal grants do crowd out state spending in an economically and statistically significant manner, leading to little increase in net spending on highway programs. In a more recent study, Gamkhar (2003) estimates the effect of a dollar of federal highway grants at 37 cents. The recent, endogeneity-corrected studies on federal highway aid clearly suggest that it is rare to find places where federal grant money does not substitute state spending on highway programs.
The crowd-out effect is one side of the substitution relationship when grants are expanding. Some empirical studies examine whether the substitution relationship is symmetric in the case of federal grant retrenchment. Gramlich (1987) finds that state and local governments tried to maintain the existing levels of program spending by increasing their own sources of revenue in response to reduced federal grants. Gamkhar and Oates (1996) later provided empirical evidence consistent with the hypothesis that state and local government responses to fluctuations of federal grants are symmetric. The symmetric substitution relationship seems to work in two distinct ways: a crowd-out effect when federal grants rise and a fillin-gap effect when federal grants fall.
Compared with other federal grant programs, federal grants to support R&D at universities have not been much explored. This is probably because a number of federal agencies sponsor academic research, and the grants...