Differential Effects of the Components of Higher Education Expenditure on U.S State Economic Growth

Transcription

1 1 Differential Effects of the Components of Higher Education Expendure on U.S State Economic Growth Valeska Araujo* McNair Scholar Universy of Missouri and Bradley R. Curs Educational Leadership and Policy Analysis Universy of Missouri (corresponding author) AERA Conference Paper (Preliminary Please do not ce) March, 2008 Abstract: Recently, the importance of human capal investment in the Uned States has increased leading to increased tuion levels. This dramatic increase in tuion has raised a debate as to what should be the proper public investment in human capal. Empirically, evidence can be found both supporting and refuting the relationship between higher education spending and state economic growth. This paper expands upon this lerature by modeling the expendure components of higher education instutions and whether they asymmetrically relate to economic growth. The findings indicate that investment in the research capacies of higher education yield the highest return on investment, and thus states should create incentives for their instutions to actively pursue research objectives. * Valeska Araujo would like to thank the Universy of Missouri McNair Scholars Program for financial support.

2 2 Differential Effects of the Components of Higher Education Expendure on U.S State Economic Growth Over the past thirty years, the importance of human capal investment in the Uned States has significantly increased leading to a dramatic upward pressure on tuion prices from both the demand and supply side. This dramatic increase in tuion levels has raised a debate as to what should be the proper public investment in human capal and what are the true effects on economic growth. Specifically, the question of whether an investment in human capal through higher education leads to substantial posive externalies is being questioned (Vedder, 2004a). Theoretically speaking, one could argue that greater public funding on education is needed to increase the productivy of labor force. However, on the other hand, there are arguments indicating that the cost of investing in human capal has increased to such a level that s returns no longer outweigh the costs; at least, not at the expected level or in the regions where the investment was made. Economic theory predicts that investment in higher education should promote economic growth if financed by non-distortionary taxes. However, the problem arises when we go to empirics. There is mixed evidence that government expendure on higher education is increasing growth rates but higher education self is increasing per-capa income in the U.S. states. Given that empirical evidence can be found supporting both arguments - that higher investment in human capal does impact economic growth in some cases posively and in others negatively - there is a need to address the question of not how much is being invested in education, but rather of how is being used. This paper tests the hypothesis of whether investment in various aspects of the higher education system, forthcoming from a strategic decision making process rather than s volume,

3 3 have higher correlation wh economic growth rates over time. The findings indicate that investment in the research capacies of higher education yield the highest return on investment, and thus states should create incentives for their instutions of higher education to actively pursue research objectives. The rest of this paper follows this general pattern. The following section outlines the major lerature looking at the relationship between government expendure and economic growth, wh a particular attention to spending on higher education. Following the lerature review we lay out our empirical specifications in the third section, and subsequently present our results of these estimated specifications. In the final section we provide discussion and concluding remarks. Lerature Review This section summarizes the economic theory and previous empirical lerature on the impact of fiscal policy on economic growth. The reason why we focus on economic growth is because Bauer, Schwezer, and Shane (2006) show that convergence of U.S. states towards their steady-state level of income seems to have stalled since the 1970s. If the U.S. states are still on a convergence path, would make sense to focus on growth rates rather than on per-capa income. Economic theory shows clearly that fiscal policy will affect the level of per-capa income but is ambiguous about s impact on the growth. The neoclassical growth models predict that fiscal policy will only impact the level of per-capa income and not the growth (Solow, 1956). In contrast the endogenous growth models predict that fiscal policy will impact both the level of per-capa income and growth (Barro, 1990; Barro & Sala-i-Martin, 1992, 1995).

4 4 Further, Barro and Sala-i-Martin (1995) tend to divide fiscal policy into two categories productive and unproductive expendure and education falls in the productive category. Barro (1990) predicts that productive government expendure will increase growth if that government expendure is financed by non-distortionary taxation. The effect of productive government expendure on growth is ambiguous if financed by distortionary taxation. The effect of government expendure on growth is negative if unproductive government expendure is financed by distortionary taxation. Due to the importance of human capal in promoting economic growth and the extensive level of government expendure for the promotion of human capal formation, the relation between public higher education expendure and economic growth has become of great interest in economic research. Empirical studies on a country level basis find a significant and posive relationship between government spending on education and economic growth (e.g. Cullison, 1993; Stokey, 1991; Barro, 2000; Blankenau & Simpson, 2004). Given this evidence one should feel comfortable to conclude that there is a significant relation between public expendures on human capal and growth at the country level. However, when expendure patterns on education spending are analyzed across U.S. States there is still a lot of uncertainty as to what is the true relationship between state-level educational spending and state-level economic growth. For example, Davis and Wang (2005) analyze the distribution of government expendures across nine areas over the period 1970 through 2000, and find a negative although insignificant relationship between public expendure on education and state economic growth. However, Davis and Wang do adm the importance of public expendure on growth, and hypothesize that the problem may be that resources are not being used efficiently. Two reasons for inefficiency are presented. First, is that often increases in

5 5 educational funding solely arise through the self-serving incentives school administrators face. Second, there is a negative relationship between the accumulation of physical and human capal and population growth. Further evidence of the negative relationship between public higher education expendure and state economic growth is presented by Vedder (2004a, 2004b) who rejects the widely agreed theoretical relationship between higher expendure and productivy. Empirically, Vedder (2004b) finds a very weak or even nonexistent correlation between expendure and productivy. He provides two plausible explanations for this finding. One is that productivy may be determined by personal attributes of people who attend college rather than by higher expendure levels. Second, there is a weak relation between productivy and expendure due to an inefficient use of funds given that a larger portion of public funds is going to the hands of school. This led him to conclude that public funding in education is not the most efficient use of public funds if economic growth is the states goal. Blackwell, Cobb, and Weinberg (2002) argue that many of the studies that find the nonexistent or negative relation between public higher education expendure and state economic growth is due to omission of important aspects in their studies such as import substution effects. By adding import substution effects in their model, they find a posive relation between state economic growth and the existence of a higher education instution in a local area. Related to this omted variable story is a study by Bhandari and Curs (2007) where the authors argue that the non-posive relationship between state expendure on higher education and economic growth can be attributed to the omission of a control for the privatization of the higher education system. They argue that states wh larger private higher education systems do not have to invest in higher education at the same level as states whout a significant private system.

6 6 An area of concern is that state level studies have solely focused on the level of funds given to higher education instutions and not to how these funds are spent. The fact that opposing empirical results exist, and that only the amount of public expendure has been analyzed, leads one to believe that the problem may not be ultimately explained by the amount of subsidies provided to the higher education system, but rather by the efficiency wh which these funds are utilized. Studies interested in identifying a more specific link between higher education spending on research activies and economic development, analyze the levels of impact of different universy activies on economic development and observe how internal and external factors related to higher education instutions, knowledge spillovers and agglomeration economies respectively, can also influence and contribute to economic development (e.g. Jaffe, 1989; Goldstein & Renault, 2004; Goldstein & Drucker, 2006). Likewise, research on education qualy through spending on instructional activies finds that higher qualy education yields labor-market returns to individuals (e.g. Card & Krueger, 1992; Kane & Rouse, 1993; Brewer, Eide, & Ehrenberg, 1999). A strong correlation between public higher education expendure and economic growth was found throughout country-level studies. Even though some negative outcomes in state-level studies are present, a concurring reason why for these negative outcomes suggested by the researchers is the inefficient use of funds. Considering that different universy activies can produce different impacts on economic development outcomes, one should pay close attention to how these funds are being allocated. If universy activies, that for example induce labor to higher productivy or attract more industry R&D to the region, can be identified new policy

7 7 could be developed to help higher education instutions be more efficient in their expendure and would lead to higher rates of return and hence, higher economic growth. Methodology Typically, economists estimate economic growth equations using one of two measures. The first, Gross State Product (GSP) which is similar to the Gross Domestic Product (GDP) of a country, represents total economic output of the state. However, GSP directly includes spending by governmental agencies, in particular educational spending, thus any estimated relationship between higher education spending and GSP are confounded. Thus, for this analysis we choose the second typical growth measure, real per-capa personal income. Data on real per-capa personal income is provided by the Bureau of Economic Analysis. Following Wang and Davis (2005), we estimate cross-state growth regressions wh U.S. state-level panel data from 1980 to 2000 controlling for state-level fixed effects. Specifically, the dependent variable is the change in real per-capa income for the 5-year period. State-level fixed effects allow the researcher to control for any time invariant unobserved heterogeney across states that may be contributing to differential growth patterns. Following previous growth lerature (e.g. Kneller, Bleaney, & Gemmell, 1999; Wang & Davis, 2005) we begin by estimating the relationship between expendure at public instutions of higher education through the following empirical growth model: g = α + β PHEE + X δ + ε (1) i 1 Where g is the growth rate real personal income per-capa, instutional expendure as a percentage of GSP, PHEE is public higher education X are time-varying state-level characteristics, and α i are state-level fixed effects. Our measure of higher education spending differs from similar research as we have defined PHEE to be expendures are public instutions as opposed

8 8 to total higher education appropriations as is tradional in this lerature. We deviate from the tradional measure of public appropriations as our interest is in how the money is spent at the public higher education instutions, not simply how much is given to the instutions by the state governments. The results of the estimation of equation (1) are found in Table 1, column 1. The contribution of this research is in the treatment of the key explanatory variable, higher education spending. While previous research has typically included total higher education spending or higher education spending as a percentage of state GSP (Vedder, 2004b; Wang & Davis, 2005), this analysis breaks down higher education spending into s expendure components, research, instruction, and other. By breaking down higher education spending into expendure components, we are able to better understand not just if higher education spending leads to economic growth, but which components of expendure are better public investments. To investigate whether different expendure components have differential effects on economic growth we alter equation (1) by breaking down PHEE into three main expendure components research ( REGE ), instructional ( IGE ), and miscellaneous ( MEGE ): g = α + β REGE + β IGE + β MEGE + X δ + ε (2) i Previous lerature shows that a larger investment on both universy research and instructional qualy has a favorable effect on economic growth. However, the relative benefs of different expendure categories of economic growth is an empirical question. Expendure data is gathered from the Integrated Post-Secondary Education Data System (IPEDS) for all instutions of higher education that are eligible for Federal financial aid and aggregated into statewide measures of research, instruction, and miscellaneous expendure levels. The results of the estimation of equation (2) can be found in Table 1, column 2.

9 9 For robustness, we also estimate the same model using the Arellano-Bond estimator, an economic growth modeling process that is able to control for the potential endogeney between economic growth and state higher education spending. Theoretically is expected that higher education spending leads to increases in economic growth, however, if governments change their spending allocations in good or bad times, standard regression techniques can yield biased estimates of this relationship. The Arellano-Bond estimator takes into this possibily and yields unbiased estimates of the relationship between higher education spending and state growth. The results of the re-estimation of equation (2) through the Arellano-Bond estimator can be found in Table 1, column 3. A second analysis investigates whether the channel of public investment in higher education, whether directly to instutions or directly to students, has differential effects on growth. Again, following Kneller, Bleaney, and Gemmell (1999) and Wang and Davis (2005), we begin by estimating the following baseline specification: g = α + βtha + X δ + ε (3) i 1 Where g is the growth rate real personal income per-capa, appropriations as a percentage of GSP, THA is total higher education X are time-varying state-level characteristics, and α i are state-level fixed effects. In equation (3) our measure of public higher education expendure, GHEE, is typical of the lerature as measures total state appropriations towards higher education. However, must be noted that this measure differs from that in equation (1). The results of the estimation of equation (3) are found in Table 2, column 1. We extend equation (3) by breaking down higher education spending into two different measures indicating the final recipient of public funding. Specifically, we break down total

10 10 government expendure on higher education into the components appropriations to instutions ( HEA ) and direct appropriations to students ( SA ): g = α + β HEA + β SA + X δ + ε (4) i 1 2 Data for the HEA come from Grapevine, and data for SA is available from the National Association of State Student Grant and Aid Programs (NASSGAP). Of interest is whether tradional higher education finance methods which provide appropriations directly to instutions have differential effects of state economic growth when compared to more recent models which are more similar to the Federal method of providing money to individual students. The results of the estimation of equation (4) are found in Table 2, column 2. The results of the re-estimation of equation (4) through the Arellano-Bond estimator can be found in Table 2, column 3. Results Column 1 of Table 1, reports the estimated results of equation (1) which show that in regards to total higher education spending the findings are similar to previous research which finds that higher education spending is not related significantly related to state economic growth (Vedder, 2004b; Wang & Davis, 2005). Holding all other variables constant, the estimated coefficient of total higher education expendure provides evidence that an incremental increase in total public higher education instutional expendure will have an insignificant effect on percapa personal income. The results of the estimation of equation (2), which breaks down total higher education expendure into three different expendure categories, is reported in column 2 of Table 1. The results indicate that the estimated impact on state-level economic growth from an increase in expendure on instruction is negative and an increase in other universy activies is posive, although both estimates are insignificantly different from zero. In contrast, the estimated

11 11 coefficient of research expendure shows that a marginal increase in research is correlated wh a posive and significant impact on income per capa. Specifically, an increment of 1 million dollars in universy research expendure, would posively affect economic development of a region by producing an increment of approximately 4.7 dollars in annual income per capa. Qualatively similar results are found using the Arellano-Bond estimator, however, the negative impact of expendure on instructional activies becomes relatively stronger and significant. Our findings would indicate that expendure by public instutions on research activies is posively associated wh state-level economic growth. These results are consistent wh lerature that indicates that research activies have a posive effect of local-area growth (e.g. Jaffe, 1989; Goldstein & Renault, 2004; Goldstein & Drucker, 2006). Further, the results would suggest that is a states interest to encourage public intuions of higher education to pursue research activies. Table 2 reports the results of our second research question, where we analyze the difference of impact on per-capa income by breaking down higher expendure by allocation based on who is the final recipient of funds instead of different universy activies. Column 1 reports the results of equation (3), where the variable of interest is the change in total state expendure on higher education. The results indicate that an increase in total state higher education appropriations is posively correlated wh per-capa income. Hence, the results contradict previous research, such as Vedder (2004b) by finding that state higher education appropriations is related to the economic development of a state. The results of equation (4) breaks down higher education expendure into two different categories of expendure, direct appropriations to instutions of higher education and direct appropriations to students through grant aid, are reported in column 2 of Table 2. The results

12 12 indicate that addional expendure through instutional appropriation has a posive and significant impact on personal income. Specifically, the estimates indicate that a 1 million dollar increase in direct instutional appropriations will lead to an increase in personal income of roughly 0.9 dollars. However, the results indicate that the impact produced by a marginal increase in grant aid expendure will be associated wh a negative, although insignificant, effect on per-capa income. Interestingly, when the Arellano-Bond estimator is employed neher direct appropriations nor student appropriations are significantly related to personal income.. Taken on face value, the results of the estimation of equations (3) and (4) would indicate that direct appropriations to instutions have a stronger relationship wh economic growth than aid to students. Policy Implications and Conclusions Previous empirical lerature finds that government expendure on higher education has a negative effect on U.S. state growth rates. The policy implication is that government should reduce s expendure on higher education. This research finds that what may be important is not the amount spent by states on higher education, but how s spent. Specifically, evidence is found that states where universies invest relatively more in research as opposed to other categories enjoy higher economic growth. These finding parallels those of K-12 lerature which find that giving more money to schools does not lead to increased student achievement unless is spent wisely. Thus, may be in the best interest of states to provide guidance for where there money is to be spent, as opposed to blindly giving to instutions as appropriations or to students through large scale financial aid programs. The importance of public higher education expendure to state economic growth found in previous country-level and some state-level studies is reinforced by the results obtained in this

13 13 research. However, our results in relation to the state-level studies of Vedder (2004b) and Davis and Wang (2005) warrant a more detailed explanation. Both Vedder (2004b) and Davis and Wang (2005) obtained results suggesting a negative or insignificant correlation between public higher education expendure and state economic growth. In spe of the negative results, both adm the importance of such investment for the economic development of a state; therefore, they suggest that the inefficient allocation of funds is a possible explanation of why negative outcomes are obtained. Both describe one potential source of inefficiency as the extensive share of funds given to school administrators rather than to students for instruction purposes, harming college attendance, productivy and qualy of education, hence reducing growth. Opposed to these two studies, our results directly find a posive and significant correlation between state expendure on higher education and state economic growth. Reerating their inefficiency argument, our results disagree wh their proposion as to how to make the allocation of funds more efficient. We find that a larger share of funds given as appropriations directly to universies, wh the most efficient use being towards research purposes will produce higher rates of growth as opposed to a larger investment given directly to students or into instruction purposes. Further, an addional finding suggests that there is an insignificant relation between public higher education instutional expendure and state economic growth. However, through breaking down higher education expendure into different allocation categories we observed that investment on different higher education areas produce different impacts on growth. Specifically, expendure on universy research activies yield the greatest effect on state economic growth. Hence, our results emphasize the large influence and importance of an efficient allocation of funds to the local economy.

14 14 Based on our findings, if higher education expendure is intended to produce higher rates of return whin the region where the investment is being made, we consider that is not a matter of solely increasing the volume of the investment but rather strategically distributing the funds to produce higher rates of return. Given that both, instution appropriations and universy research activies are the two factors that we observed to have both a significant and posive effect on state economic growth, we consider that instead of blindly giving aid to students, states should give a larger share of s funds directly to the instution which would later have the responsibily of distributing such resources efficiently by allocating a larger share of their funds to universy research purposes. Moreover, considering our results on the particular significance of universy research to growth and the significance of factors such as qualy and quanty of education observed by Card and Kruger (1992) and Kane and Rouse (1993) respectively, further research could be carried out to analyze the relation between these three factors to see if an alternative allocation pattern, that would yield even higher rates of return, can be detected.

15 15 References Barro, R.J. (1990), Government spending in a simple model of endogenous growth. The Journal of Polical Economy. 98(5), s103-s125. Barro, R.J. (2000), Education and Economic Growth. Harvard Universy, mimeo. Barro, R.J., & Sala-i-Martin, X. (1992). Public finance in models of economic growth. Review of Economic Studies, 59(4), Barro, R.J., & Sala-i-Martin, X. (1995). Economic Growth. New York: McGraw-Hill. Bauer, P. W., Schwezer, M.E., & Shane, S. (2006). State growth empirics: The long-run determinants of state income growth. Federal Reserve Bank of Cleveland Working Paper. Bhandari, B. & Curs, B. R. (2007). The roles of public higher education expendure and the privatization of the higher education on U.S. states economic growth. Universy of Missouri-Columbia Working Paper. Blackwell, M.., Cobb, S., & Weinberg, D. (2002). The economic impact of educational instutions: Issues and methodology. Economic Development Quarterly, 16(1), Blankenau, W., & Simpson, S. (2004). Public education expendures and growth. Journal of Development Economics, 73, Brewer, J.D., Eide, E.R., & Ehrenberg, R.G. (1999). Does It Pay to Attend an Ele Private College? Cross-Cohort Evidence on the Effects of College Type on Earnings. The Journal of Human Resources, 34 (1), Card, D., & Krueger, A.B. (1992). Does School Qualy Matter? Returns to Education and the Characteristics of Public Schools in the Uned States. Journal of Polical Economy, 100 (1), 1-40.

16 16 Cullison, W. (1993). Public investment and economic growth. Federal Reserve Bank of Richmond Economic Quarterly, 79, Davis, O. & Wang, L. (2005). The Composion of State and Local Government Expendures and Economic Growth. Carnegie Mellon Universy, Working Paper. Goldstein, H., & Drucker, J. (2006). The Economic Development Impacts of Universies on regions: Do size and Distance Matter? Economic Development Quarterly, 20, 22. Goldstein, H.A., & Reanult, C.S. (2004). Contributions of Universies to Regional Economic Development: A Quasi-experimental Approach. Regional Studies, 38(7), Jaffe, A.B. (1998). Real Effects of Academic Research. The American Economic Review, 79(5), Kane, J.T., & Rouse, C.E. (1993). Labor-Market Returns to Two- and Four-Year College. The American Economic Review, 85(3), Kneller, R., Bleaney, M. F., & Gemmell, N. (1999). Fiscal policy and growth: evidence from OECD countries. Journal of Public Economics, 74, Solow, R.M. (1956). A contribution to the theory of economic growth. Quarterly Journal of Economics, 71(1), Stokey, N. (1991). Human capal, product qualy and growth. Quarterly Journal of Economics, 106(2), Vedder, R. (2004a). Going broke by degree: Why college costs too much. Washington, DC: The American Enterprise Instute Press. Vedder, R. (2004b). Private vs. social returns to higher education: Some new cross-sectional evidence. Journal of Labor Research, 25(4),

17 17 Table 1: Impact of Investment on Research, Instruction and Other Areas of Higher Education on Economic Growth In all specifications the dependent variable is change in income per capa. Standard errors are in parentheses. Independent Variable (1) (2) (3) Change in instutions total expendure (0.25) Change in research expendure 4.716* 4.658* (2.49) (2.81) Change in instruction expendure *** (0.93) (1.05) Change in miscellaneous expendure (0.50) (0.54) Change in state population 1.431** 2.480*** 3.652*** (0.63) (0.77) (0.92) Change in union membership ** (1.23) (1.31) (4.64) Change in percentage of population wh high school diploma (54.3) (53.8) (61.1) Lagged change in per-capal real income (0.11) Year fixed effects Yes Yes Yes Constant 2134*** 1807*** -1709*** (265) (296) (277) Observations Number of states R-squared Standard errors in parentheses. *Significant at 10%; **Significant at 5%; ***Significant at 1%

18 18 Table 2: Impact of Allocation of Public Funds across Appropriations and Grant-Aid on Economic Growth In all specifications the dependent variable is change in income per capa. Standard errors are in parentheses. Independent Variable (1) (2) (3) Change in state expendure 0.796*** (0.27) Change in appropriations 0.942*** (0.29) (0.36) Change in grant-aid (2.77) (3.24) Change in state population (0.62) (0.65) (0.84) Change in union membership (1.18) (1.17) (4.88) Change in percentage of population wh high school diploma (52.5) (52.3) (63.2) Lagged change in per-capal real income (0.11) Year fixed effects Yes Yes Yes Constant 2204*** 2296*** 844.7*** (250) (258) (230) Observations Number of states R-squared Standard errors in parentheses. *Significant at 10%; **Significant at 5%; ***Significant at 1%