The Long-Run Economic Impact of an Institution of Higher Education

Abstract

One of the main contributions of higher education institutions is human capital. In the context of regional universities, their primary impact can be measured through the future income stream of alumni who remain in the local area for work. This constitutes a long-run impact and a substantial part of the overall economic impact of an institution to the local, regional, and national economies. This study provides an important contribution to the existing literature by developing a methodology that takes into account the counterfactual, allowing for the more precise definition of economic impact. The methodology is applied to a public regional university.

Keywords

economic impact human capital higher education short run long run multiplier

Economic impact is defined as “the difference between existing economic activity in a region given the presence of the institution and the level that would have been present if the institution did not exist” (Beck, Elliott, Meisel, & Wagner, 1995, p. 246). Two approaches are proposed in the literature—the short-run expenditures approach and the long-run human capital contribution approach. The short-run approach is well established and is utilized in numerous studies to estimate the economic impact of different types of institutions and programs. In estimating the economic impact of a university, the short-run approach is the standard method used. In contrast, the long-run approach is not well established and has not been estimated extensively because of data requirements that are difficult to satisfy. A comprehensive economic impact study should include both the short- and long-run impacts.

This study will concentrate on the long-run approach and propose a methodology for the long-run human capital contribution approach that addresses weaknesses and problems of previous research. The theory of economic impact will be summarized, including a brief discussion of the short-run approach, followed by a section on the proposed long-run methodology. The long-run methodology will then be applied based on data from a public regional university.

One of the main challenges of estimating the economic impact from the creation of human capital is the ability to take into account the counterfactual—what would have occurred had the student not attended the institution. This study addresses the counterfactual and allows for the estimates to be a closer approximation of the definition of economic impact as originally conceptualized by Bluestone (1993). It should also be noted that economic impact studies only capture benefits and ignore the costs; thus, studies do not capture net benefits. This is in contrast to benefit–cost analysis, fiscal impact analysis, or return on investment analysis that captures the net impact of an investment/project. This study concentrates on the long-run estimation of economic impact; that is, estimating the long-term benefits of an educational institution to the local region.

The Theory of Economic Impact

Short-Run Approach

A majority of economic impact studies of educational institutions estimate an institution’s short-run economic impact. These institution-based studies treat an increase or decrease in expenditures by a university as comparable with the expansion or withdrawal of an industry to a region (Brown & Heaney, 1997).¹ Four different methods were used in estimating short-run economic impact, as summarized in Stokes and Coomes (1998): income expenditure analysis, economic base analysis, input–output analysis, and the Caffrey–Isaacs approach. These four methods estimate the short-run impact by applying the multiplier to the specific economic activity contributed by the university to the economy of the region. The multipliers and the specific economic activities utilized differ among the four methods, and the choice of which approach is employed primarily depends on data availability and researcher preference.²

Income expenditure analysis developed by John Maynard Keynes employs university spending as its economic data. The multiplier is based on a regional marginal propensity to consume and the marginal propensity to import. To estimate the short-run economic impact the multiplier is applied to university spending. According to Stokes and Coomes (1998), several studies were completed using the income expenditure approach, including Cooke (1970), Brownrigg (1973), Dick and Wood (1980), Lewis (1988), and Armstrong (1992).

Economic base analysis was proposed in the research of Werner Sombart, a German sociologist, during the 1920s. In this methodology, the economy of the region where the university is located is divided into two sectors: the basic sector, which consists of entities that provide goods and services outside the region; and the nonbasic sector, which consists of entities that contribute goods and services to the local economy. The university is considered an exporting entity, in that the output (goods and services) of the university is in demand by outside regions and therefore university spending is considered an export. The income derived from university spending contributes to the economic impact of the university to the local region. Therefore, only the sources of income from outside the region are included in the economic impact. The economic impact multiplier is based on the ratio of total income to export income. Several studies, summarized by Stokes and Coomes (1998), have used the economic base method, including Kraushaar (1964), Moore (1979), and Smith and Bissonette (1989).

The Caffrey–Isaacs approach was developed in 1971 for the American Council of Education to provide a framework for estimating educational economic impact. Economic information required by this model includes spending by faculty, staff, students, the university, and constituents. Other economic activities occurring because of the university location and facilities are also estimated. Local government income and spending changes from the existence of the university adds to the impact. The multiplier depends on local business effects, the value of property related to the university, the costs and benefits to the local government, wages, and jobs created. There are several university economic impact studies where this method was used, including Indiana University, Marquette University, and Loyola University of Chicago.

Input–output (I–O) models, developed first by Wassily Leontief in 1936, estimate interindustry relationships in a region by measuring the distribution of inputs purchased and output sold by each industry. The I–O models calculate how the impact of one dollar “ripples” throughout the regional economy, creating additional expenditures and jobs. This is more commonly referred to as the “multiplier effect.” A matrix of industry transactions, including industry production, final demand, and value added, is developed to determine multipliers. Because industries are related through the transactions matrix, economic activity by sectors affects every other sector.

Total effect multipliers, which are used in all four methods, can be divided into a direct effect, an indirect effect, and an induced effect. The direct effect is the amount of money that a university spends in the economy; that is, the university purchases goods and services from firms located in the economy. Those businesses that receive money from the university also purchase goods and services, and hire people who will spend their wages and salaries in the economy. This additional amount of spending by businesses that receive income as a result of university spending is the indirect effect. Employees of the university and employees of university vendors also spend a portion of their wages and salaries locally; that is, the induced effect. Essentially, dollars “ripple” through the economy producing this multiplier effect. Hefner (1997) provides a comprehensive guide to using I–O analysis to measure economic impact.

In practice, I–O analyses are most commonly employed using commercially available software packages including IMpact Analysis for PLANning (IMPLAN; n.d.), Regional Input-Output Modeling System (RIMS II), or Regional Economic Models, Inc. (REMI; n.d.). IMPLAN has been widely used in estimating the local and regional economic impact of universities. Use data for both RIMS II and REMI are unavailable because of privacy restrictions of their clients.

Long-Run Approach

The theoretical foundation of the human capital approach was developed by Becker (1962, 1975). Education is considered as an investment in human capital and an indicator of future well-being of the individual. Investment in human capital is beneficial when the present value of the marginal future income exceeds the present value of the marginal costs. This is true for both individuals and society as a whole. Mincer (1974) developed an earnings regression that estimated the return on investment of education level or “schooling.” Commonly known as the “schooling model,” years of education are regressed against earnings.

\ln (earnings) = α + (Schooling) β + ε

Many studies have used a form of this equation to estimate the value of human capital.

The basic theoretical model of education as an investment in human capital suggests increased lifetime earnings of graduates. Many studies have found a positive relationship between level of education and lifetime earnings, including Romano (1986), Kane and Rouse (1995), Day and Newburger (2002), Kantrowitz (2007), and Julian and Kominski (2011). Employment data from the U.S. Bureau of Labor Statistics (BLS) also find that the unemployment rate is negatively related to the level of education earned. Data from the U.S. Census Bureau likewise support the economic value of education. For instance, an estimate of close to $1 million work life earnings differential between a college graduate versus a high school graduate was reported from the American Community Survey Briefs in 2011.³ A majority of the research investigates the value of human capital in the aggregate and not for individual institutions. For the existing literature that has tried to capture the value of human capital, estimates were generated by simply adding total earnings by alumni (e.g., Pittsburg State University Economic Impact Statement, 2002). These estimates, however, represent the scale of the human capital contribution but not the economic impact. This particular study adjusts for the counterfactual and earnings differential between a high school and a college graduate to estimate the value of human capital.

Relatively few studies have studied the economic impact of an individual educational institution using the human capital approach. The methodology requires “quantifying future activity levels and comparison of that effect on what would have occurred had the student not attended the institution” (Blackwell, Cobb, & Weinberg, 2002, p. 89). Bluestone (1993) and Berger and Black (1993) are two early studies that estimated the economic impact of universities based on the human capital approach. Both Bluestone and Berger and Black are the most cited articles, in the context of Becker (1975) and Mincer (1974), for providing a framework to the human capital approach.

Bluestone (1993) estimated the economic impact of the University of Massachusetts at Boston. Regression analysis utilizing cross-sectional income data for four different groups of students was developed to determine “age-earnings profiles.” High school graduates, students who completed 1 to 3 years of college, college graduates, and graduate students were distinguished by gender and age. Income was estimated for these groups for their work life from age 25 to 65. The present value of these streams of income was then compared with current costs of education to evaluate the economic impact of a university education. Results did not consider the possibility that graduates had innate ability that the university did not influence (i.e., the ability bias). They also did not consider that wages and salaries increase over time because of increases in productivity (i.e., productivity bias).

Berger and Black (1993) estimated the impact of a public education in Kentucky using a methodology that attempted to adjust for the ability bias and productivity bias. A version of the Mincer model using cross-sectional data, with adjustments for education level, gender, and experience, was estimated specific for the state of Kentucky.

\ln (Annual earnings) = f (Education, Experience, Gender)

The level of education using dummy variables was distinguished by high school, some years of college, associate degrees, bachelor’s degrees, graduate degrees (both at the master’s and doctorate levels), and professional degrees. Differences in the job markets between the state of Kentucky and the United States were also taken into account in estimating annual earnings. Productivity growth was considered by using growth rates of predicted income over time.

A more recent approach to the long-run human capital measurement was proposed and used by Economic Modeling Specialists Intl. (EMSI). According to EMSI, during the last 15 years, 1,700+ economic impact reports were completed for educational institutions in the United States. These reports consist of two components: an analysis of the economic impact of an educational institution and the return on investment of the education obtained. The economic impact is a benefit analysis, whereas the return on investment is a cost–benefit analysis. The economic impact methodology of EMSI is a long-run human capital approach and is most comparable with the methodology presented in this study. The methodology used by EMSI calculates the impact of student productivity, consists of the added labor income resulting from the education obtained and from the added profits that employers obtain because of the students employed.

Unlike other studies, EMSI bases its measure of education not on degrees obtained but by credit hours earned. Earnings differences are determined based on the number of credit hours earned and the average wages by degrees earned, referred to by EMSI as an “education ladder.” The ability bias is adjusted for by reducing the marginal differences in wages by 10%. The aggregate increase in annual earnings is calculated as

Δ E = \sum_{t = 1}^{n} (e_{i} h_{i})

where n = steps in the “education ladder,” ei = change in earnings at each step i, and hi = the number of credit hours completed at each step i. EMSI also assumes that earnings increase over time. Utilizing a Mincer earnings function, aggregate earnings are adjusted to reflect increases at a decreasing rate. The additional income earned by businesses that employ graduates is estimated using industry-specific earnings data and a distribution of where graduates are employed. This is then adjusted by the “education ladder” steps and aggregated to obtain the total additional income earned by businesses. The sum of the aggregate labor income and the aggregate profits earned is the total economic impact. EMSI also adjusts for the counterfactuals of alternative education, substitution effects, and alternative uses of funds. A more complete description of its methodology can be found in the EMSI (2014) report, Where Value Meets Values: The Economic Impact of Community Colleges, Analysis of the Economic Impact and Return on Investment of Education.

While the short-run expenditures approach is the generally accepted standard for measuring the economic impact of universities with its methodology more established, the human capital approach has produced results that are subject to empirical criticism. The difficulty arises because measuring the human capital contribution requires “quantifying future activity levels and comparison of that effect on what would have occurred had the student not attended the institution” or the counterfactual (Blackwell et al., 2002, p. 89). The framework of this study is consistent with these works, but more importantly, addresses the counterfactual.

Methodology

Institutions of higher education produce education, an investment in human capital, and thus generate regional economic impact through increased work life earnings of graduates who choose to stay and work in the local area. This study defines the local area as the state where the regional public institution under consideration is located. Defining the local area is critical in estimating the long-run economic impact since it affects the magnitude of the impact. A broader definition of the geographic area will imply students have more options to choose from in terms of competing institutions. One of the challenges of estimating economic impact is the counterfactual. The more precise definition of an economic impact should only include the human capital attributed to graduates who would not have been present in the local area anyway if the institution did not exist (Beck et al., 1995). In addition, only incremental incomes of college graduates who stay in the local area should be included (Bluestone, 1993). This study is able to capture the more precise definition of economic impact that is not captured in previous economic impact studies. For the public regional university under consideration, there are potentially six public universities to compete with, in addition to a number of private institutions.

The human capital contribution approach adds to the short-run expenditure approach by capturing the contribution to the state economy of the human capital produced by the university. By providing education, the university enhances the skills and thus the productivity and income of its graduates. These graduates contribute to the state economy by spending their earned income, paying taxes, providing services, and so forth. The primary long-run human capital contribution can be measured by the income stream of graduates who stay to work in the state.

Since one of main contributions of the study is the ability to capture the counterfactual, it is likewise important to mention the potential bias introduced from migration of high school students when attending college and migration of college graduates. Schmidt (1998) indicates there seems to be a perception that talented students leave a state for college and do not return. Students who leave states and enroll elsewhere, on average, take $30,000 to $50,000 per student per year out of their state economies (Postsecondary Education Opportunity, 2003). As for college graduates, their economic impact in the local area (state) will be larger if attending college in a state that encourages them to work in the state. Survey data from the U.S. Department of Education show that 72.4% of 1993 graduates were still in the state where their degree-granting institution was located 1 year after graduation, and 66.7% were still in the same state 4 years after graduation. More recent data reveal, however, that graduates are more mobile, with 69.2% of year 2000 graduates still in the same state in which they attended college. The study state’s share of 1993 and 2000 graduates remaining in the state is lower than the national average.⁴ Using data from the 1979 to 1996 wave of the National Longitudinal Survey of Youth, Kodrzycki (2001) found that a student who had gone to college out of state was 54% more likely to be out of state 5 years after graduation compared with someone who went to school in state. That migration is explained more by individual characteristics rather than overall employment opportunities in the state where they graduated. In addition, young graduates are more likely to move if they are in a state that has low employment growth, high unemployment, or low pay for college graduates. Groen (2004) finds a modest link between attending college and subsequently working in the state. His empirical estimates suggest that, of a potential 100 additional students to attend college in state versus out of state, no more than 10 would be working in the state 10 to 15 years after college. Data constraints limit our ability to explicitly factor in migration throughout the worker’s work life. We are only able to roughly approximate migration for one out of the five groups of alumni utilized to estimate economic impact accounting for the counterfactual. If migration is extensive, the empirical estimates of economic impact will likely be biased upward.

Data

The human capital estimation methodology is applied using 2010 data from the regional public university in consideration. The alumni office provided information on the number of alumni living in the state as of 2010, which stood at 26,069. All the data utilized in the empirical models (income regressions) came from a survey of alumni who currently reside in the state based on the list provided by the alumni office. The survey was administered online (initiated through email) by the university’s Center for Survey Research to collect more specific information regarding current income, employment status, occupation, and the alumni’s next best alternative option if he/she had not attended the university (to capture the counterfactual).⁵ The most critical piece of primary information from the survey was the data that allowed us to directly estimate the counterfactual. Blackwell et al. (2002) accounted for import substitution, what we consider counterfactual in this study, by extrapolating data from their institution’s alumni office. Based on the proportion of both local and nonlocal graduate locations 6 to 7 years following graduation, they estimate approximately 22% of their alumni remain in the local area because they attended the institution. The data in this study allowed us to expand on Berger and Black (1993), who estimated the human capital impact of their institution but did not explicitly account for the counterfactual, and Blackwell et al. (2002), who estimated the counterfactual for their institution but not the economic impact. This study does both.

To capture the counterfactual, the actual question on the survey was the following:

If you did not attend University X, would you have: (A) attended another university in the state, (B) attended a trade school in local area X (where the university is located), (C) attended another university out of state, (D) not attended college at all, and (E) others.

The total alumni earnings represent the scale of the contribution of the university graduates to the human capital in the state. However, the more precise economic impact should only include the incremental alumni earnings that can be uniquely attributed to the university. To estimate accurately, two corrections need to be made to (a) only include earnings differential to that of a high school graduate and (b) not include earnings from alumni who would have gone to another university in the state if the university did not exist (adjusting for the counterfactual).

The three groups who qualify for the unique contribution of University X (i.e., because of the existence of University X) are (C), (D), and (E) as indicated above. Group (C) are those who would have attended another university out of state. For this particular group, the economic impact of the university, the counterfactual, is not a matter of college–high school wage differential but estimated simply as earning the college wage adjusted for the proportion of return migration; that is, those who would return in the state. We use the proportion of interstate college migrants who returned to their home state on graduation as reported by Ishitani (2011) from the 2000 National Educational Longitudinal Study and Postsecondary Education Transcript Study provided by the National Center for Education Statistics as the adjustment percentage. This stands at 39%. Ideally, this would be adjusted further for the proportion who went to high school in another state and would never have settled in the state in the absence of University X. Unfortunately, data were not available and to this extent, and this portion of the estimate could be slightly overestimated. For Group (D), those who would not have attended college at all, the college–high school earnings differential is used. It can be argued that people with only a high school degree are less likely to migrate and would have stayed in the state under the counterfactual. Thus, our inability to account for migration in this group is likely to have a very minimal impact on the economic impact estimate. Group (E), those who responded with “other,” are also included with the implicit assumption that they would have remained in the state and earned a high school wage if University X had not existed. Unfortunately, no further data are available to check the robustness of this assumption. Given the location of University X (relatively rural and isolated) and the overall student population (a large number of first-generation students), it seems reasonable to assume that these individuals would have remained in the state. The average percentage of the respondents who are in this group across all the age cohorts is only about 5%.

Two of the remaining groups, (A) and (B), are excluded in the economic impact estimates. These groups are assumed to have identical earnings and geographic location with or without the existence of the university, and thus their contribution to the local area is not attributable to the university. Group (A) includes those who would have attended another university in the state. The benefits from this group represent only a reallocation of to where the benefits should be attributed and do not constitute additional benefits. Group (B) are those who would have attended a trade school in the state. The exclusion of this group in estimating the economic impact assumes that the earnings for a trade school graduate approximates that of a graduate from University X. It is likely that a trade school graduate earns less than a college graduate. The BLS reports that the median U.S. weekly earnings in 2013 of a college graduate was $1,108 versus that of an associate’s degree holder, which was $777. The underestimation of economic impact resulting from this assumption is likely smaller than these numbers would indicate given the relative location and type of educational institution of University X.

For the overall sample, approximately 40% of graduates fall under Groups (C), (D), and (E), representing the total overall counterfactual percentage. For precision, we further disaggregated the data to find the actual counterfactual percentages for each age cohort. Thus, the subsequent estimates of economic impact use the appropriate percentages in Groups (C), (D), and (E) under each age cohort.

Empirical Estimation

The methodology developed to estimate the long-run human capital economic impact involves a two-step process. The first step was to estimate a version of the Mincer (1974) earnings equation that controls for occupation, field of study, gender, age, experience, ability, and other relevant factors. The second step then utilizes the results from Step 1 to generate predicted earnings by age group, which will ultimately be used to estimate the long-run economic impact. Earnings is defined as the annual net income (after taxes and other deductions) only.⁶

An earnings regression was estimated since the income of all relevant alumni was not available. Estimating an earnings model using primary information from the survey allowed for a more accurate extrapolation of earnings for the full alumni population of interest, using the appropriate age distributions than does directly utilizing the average income as reported in the survey and then multiplying with the total number of relevant alumni. The empirical model developed for the earnings equation is specified in Equation (1) as

\ln E a r n i n g s = β_{1} + {\sum_{2}^{6} β}_{i} A G E + β_{7} G E N + \sum_{8}^{12} β_{i} G E N_A G E + β_{13} E X P + β_{14} E X P^{2} + β_{15} A B I L I T Y + β_{16} \ln S T A R T I N C + β_{17} J O B_M A J O R + β_{18} H S_S T + β_{19} C A R E E R_S T + \sum_{20}^{32} β_{i} O C C U P + \sum_{33}^{37} β_{i} M A J O R + e_{i}

Age is included in the model as dummy variables following the age group classifications of the U.S. Census Bureau when estimating work life earnings. Interaction terms between age and gender were included to capture for variation in potential gender-based differences in income for different age groups. It is expected, for instance, that potential gender-based discrimination in income is more likely prevalent for the older age group given the developments in the labor market over time. Experience is modeled in quadratic form to capture the inverted U profile of earnings through the work life cycle. An important addition to Equation (1) that differentiates it from previous studies is the inclusion of the dummy variable ABILITY⁷ (if alumni graduated with honors), which attempts to correct for ability bias. In general, the talents and skills of those who pursue a college education would have resulted in higher incomes even if they did not attend college. Occupation (OCCUP) and undergraduate major (MAJOR) are included in the model as dummy variables following the categories utilized by the U.S. Census Bureau (see Table 1).

Table 1.

Variable Definition and Means.

Variable	Definition	Mean
Earnings	Current annual net income ($)	57669.656
AGE_1	=1 if age between 21 and 24; 0 otherwise	0.025
AGE_2	=1 if age between 25 and 34; 0 otherwise	0.205
AGE_3	=1 if age between 35 and 44; 0 otherwise	0.168
AGE_4	=1 if age between 45 and 54; 0 otherwise	0.209
AGE_5	=1 if age between 55 and 64; 0 otherwise	0.239
AGE_6	=1 if age >65; 0 otherwise	0.154
GEN	=1 if male; =0 if female	0.48
GEN_AGE	Interaction term between AGE and GEN
EXP	Years working for current employer	12.32
ABILITY	=1 if graduated with honors; =0 otherwise	0.30
STARTINC	Annual net income when started career ($)	20191.157
JOB_MAJOR	=1 first job in major; =0 otherwise	0.74
HS_ST	=1 attend high school in state; =0 otherwise	0.93
CAREER_ST	=1 start career in state; =0 otherwise	0.87
OCCUP_1	=1 if occupation in business; =0 otherwise	0.228
OCCUP_2	=1 if occupation in math; =0 otherwise	0.044
OCCUP_3	=1 if occupation in social sciences; =0 otherwise	0.045
OCCUP_4	=1 if occupation in legal; =0 otherwise	0.022
OCCUP_5	=1 if occupation in education; =0 otherwise	0.318
OCCUP_6	=1 if occupation in entertainment; =0 otherwise	0.025
OCCUP_7	=1 if occupation in health; =0 otherwise	0.094
OCCUP_8	=1 if occupation in protection; =0 otherwise	0.007
OCCUP_9	=1 if occupation in services; =0 otherwise	0.006
OCCUP_10	=1 if occupation in sales; =0 otherwise	0.040
OCCUP_11	=1 if occupation in office; =0 otherwise	0.039
OCCUP_12	=1 if occupation in farming; =0 otherwise	0.032
OCCUP_13	=1 if occupation in production; =0 otherwise	0.022
OCCUP_14	=1 if occupation in others; =0 otherwise	0.078
MAJOR_1	=1 if undergrad major is business; =0 otherwise	0.230
MAJOR_2	=1 if undergrad major is agriculture; =0 otherwise	0.095
MAJOR_3	=1 if undergrad major is math; =0 otherwise	0.045
MAJOR_4	=1 if undergrad major is liberal arts; =0 otherwise	0.238
MAJOR_5	=1 if undergrad major is biological sciences; =0 otherwise	0.084
MAJOR_6	=1 if undergrad major is others; =0 otherwise	0.308

It should be noted that earnings equation only captures the measureable (in $) benefits of relevant alumni and does not include the positive externalities generated in the local area from these groups of college-educated individuals. It has been shown that college-educated individuals generate positive externalities in their area of employment, effectively improving productivity and earnings of nonalumni (Iranzo & Peri, 2009; Moretti, 2004; Shappiro, 2006; Winters, 2011).

On estimating Equation (1), predicted earnings are calculated and utilized to calculate total earnings in 2010. Total earnings for the current year, which represent the scale of the human capital contribution of the university to the state, was estimated by multiplying the number of alumni in the local area in each age group by the estimated workforce participation rate and estimated average earnings for each age group, then summing the earnings across all ages. The data required for these adjustments were obtained from the alumni survey. As indicated earlier, the economic impact should only include the incremental alumni earnings that can be uniquely attributed to the university. To value the human capital economic impact, the additional earnings are estimated. This is done by subtracting the earnings from a high school degree to that of the predicted earnings of college graduates as estimated from Equation (1), which captures earnings differential from a high school degree.⁸ A final adjustment was made to capture the counterfactual, which only counts alumni whose attendance of the school is uniquely attributed to the university, as discussed earlier.⁹ There is a wide variation in counterfactual percentages by counterfactual group and age cohort ranging from 3% to 28%. In contrast, Blackwell et al. (2002) only utilized a single estimate, which was 22%, for their institution (location was in the Cincinnati, OH, area). University X is in a relatively rural area in the Midwest.

The benefits gained from obtaining a college education are long term and accrue toward an individual’s work life. Work life earnings were estimated for the relevant proportion of the alumni as of 2010. The methodology utilized by the BLS and the U.S. Census Bureau is adopted to estimate work life earnings,¹⁰ and adjusted for survival rates and workforce participation rate to take into account potential changes throughout the work life of the individual. Ideally, the long-run impact should also take into account the work life mobility of workers and should only include the proportion who stay. The literature in migration has established age as an important factor—that age reduces the net present value of migration (Goss & Paul, 1986). We do not have data on worker migration for the particular state in consideration for each age cohort and are unable to take this into account in the long-run estimate. In 2010, however, the American Community Survey indicates that more than 70% of the Midwest’s current population are born in their current state, implying a relatively lower lifetime mobility compared with the national average.

The resulting estimate represents what alumni could expect to earn on average in 2010 dollars during a “hypothetical” 54-year work life.¹¹ This estimate is the long-run human capital economic impact of the university in the local economy.

Results

The results for the earnings Equation (1) is presented first (see Table 2). These were mostly consistent with the theoretical predictions in terms of causal relationships between income and individual characteristics. The model provided a relatively good overall fit given that the data were cross-sectional with an R² of 40.9%. There is evidence of gender-based difference in salary, which is more evident in the older age group. Ceteris paribus, male alumni in the age group of 21 to 24 years earn about 4.3% less than their female counterparts. As we examine the older age groups, however, the earnings differential switches in favor of the males, becoming more pronounced as we move further down the age groups. Males earn approximately 13.2% more than females for the age group 25 to 34 years and increases to about 37% higher for the age group greater than 65 years. These results are consistent with the a priori expectations, given the developments in the labor market and the movement to create a more equal playing field across gender. Alumni who graduated with honors (ABILITY), on average earn about 9% more relative to those who did not graduate with honors, holding other factors constant. Starting one’s career in his/her major field of study (JOB_MAJOR) yields a current income that is 5.1% higher relative to someone who did not start out a career in his/her major. The parameter estimates for the variables capturing different occupations indicate business-related occupations (the reference category) command higher salaries. For instance, alumni in the social sciences occupation (OCCUP_3) earn, on average, 33% less than alumni in the business-related occupations. On the other hand, the variables capturing undergraduate majors were not significant. This may be partially captured by the variables capturing occupations (OCCUP) and JOB_MAJOR, all significant predictors of current income.

Table 2.

OLS Parameter Estimates of the Earnings Equation (Standard Errors).

Variable	Parameter estimate
Constant	6.851*** (0.324)
AGE_1	−0.193* (0.130)
AGE_2	0.010 (0.107)
AGE_3	0.216** (0.106)
AGE_4	0.220** (0.105)
AGE_5	0.274*** (0.105)
GEN	0.679*** (0.120)
GEN_AGE1	−0.722*** (0.191)
GEN_AGE2	−0.547*** (0.128)
GEN_AGE3	−0.399*** (0.130)
GEN_AGE4	−0.307** (0.129)
GEN_AGE5	−0.306** (0.129)
EXP	0.20*** (0.004)
EXP2	0.00005*** (0.00001)
ABILITY	0.090*** (0.025)
STARTINC	0.360*** (0.031)
JOB_MAJOR	0.055** (0.026)
HS_ST	0.039 (0.053)
CAREER_ST	−0.039 (0.036)
OCCUP_2 (math)	−0.089* (0.058)
OCCUP_3 (social)	−0.330*** (0.058)
OCCUP_4 (legal)	−0.012 (0.077)
OCCUP_5 (education)	−0.291*** (0.036)
OCCUP_6 (entertainment)	−0.436*** (0.076)
OCCUP_7 (health)	−0.113** (0.051)
OCCUP_8 (protection)	−0.240* (0.132)
OCCUP_9 (services)	−0.415*** (0.150)
OCCUP_10 (sales)	−0.153*** (0.058)
OCCUP_11 (office)	−0.550*** (0.060)
OCCUP_12 (farming)	−0.096 (0.074)
OCCUP_13 (production)	−0.273*** (0.079)
OCCUP_14 (others)	−0.351*** (0.048)
MAJOR_2 (agriculture)	0.032 (0.045)
MAJOR_3 (math)	−0.084 (0.062)
MAJOR_4 (liberal arts)	−0.041 (0.036)
MAJOR_5 (biological sciences)	0.051 (0.053)
MAJOR_6 (others)	−0.044 (0.034)
F stat = 33.585; p value = 0; N = 1,782

***

Significant at the 1% level. **Significant at the 5% level. *Significant at the 10% level.

The parameter estimates from Table 2 were utilized to predict average earnings per age group (column 2, Table 3). The total alumni earnings in 2010, which represents the scale of the human capital contribution of the university to the state, were estimated at $1.1 billion. The economic impact for the year 2010 in terms of human capital, on the other hand, was estimated at $130.85 million. Table 3 illustrates the steps (as initially discussed in the Methodology section) to estimate the scale (column 5) and ultimately the economic impact from human capital brought by the university to the state economy (column 12). Essentially, the predicted earnings from Equation (1) was adjusted for the number of relevant alumni (column 3) by age group, workforce participation rate (column 4), earnings differential from a high school graduate (column 6), and, finally, the adjustment for capturing the counterfactual percentages (columns 8 and 9).

Table 3.

Estimated Scale of Human Capital Contribution and Economic Impact in 2010.

(1) Age group	(2) Average estimated earnings	(3) Number of relevant alumni	(4) % Employed	(5) Total earnings [(2) × (3) × (4)]	(6) State high school graduate 2010 average income (U.S. Census Bureau)	(7) Additional earning capacity [(2)-(6)]	(8) Counterfactual % for Group C^a [Survey]	(9) Counterfactual % for Groups D and E [Survey]	(10) 2010 Economic impact for Group C^a [(2) × (3) × (4) × (8)]	(11) 2010 Economic impact for Groups D and E [(7) × (3) × (4) × (9)]	(12) 2010 Economic impact all groups [(10) + (11)]
21-24	$26,418.40	782	84%	$17,353,720.50	$20,033	$6,385.40	6%	5%	$1,041,223.23	$209,722.17	$1,250,945.40
25-34	$39,237.87	5,214	96%	$196,402,799.51	$25,258	$13,979.87	6%	7%	$11,784,167.97	$4,898,277.72	$16,682,445.69
35-44	$53,486.01	4,432	95%	$225,197,480.08	$28,273	$25,213.01	3%	12%	$6,755,924.40	$12,738,820.91	$19,494,745.31
45-54	$56,336.38	5,475	95%	$293,019,572.55	$29,075	$27,261.38	3%	19%	$8,790,587.18	$26,940,713.47	$35,731,300.65
55-64	$59,100.46	6,257	83%	$306,927,009.40	$28,315	$30,785.46	3%	20%	$9,207,810.28	$31,975,687.35	$41,183,497.63
>65	$69,607.31	3,910	26%	$70,762,793.28	$19,067	$50,540.31	3%	28%	$2,122,883.80	$14,386,198.70	$16,509,082.50
				$1,109,663,375.32							$130,852,017.18

Group C: Those who would have attended another university out of state. The counterfactual percentage for this group was adjusted for return migration of college graduates and economic impact is not college–high school wage differential, but simply college earnings of the corrected proportion of graduates.

The full long-run economic impact requires estimating additional work life earnings of relevant alumni, and Table 4 presents the steps involved. The estimated 2010 economic impact for each age group is multiplied to the number of work life years associated with the age group, adjusted for the survival rate, and then summed for all age groups. As of 2010, the long-run economic impact (from human capital) of the university to the state economy is estimated at approximately $1.29 billion (in 2010 dollars). As stated earlier, data constraints limited our ability to factor in work life mobility of workers, which introduces a likely upward bias in the long-run estimate.

Table 4.

Estimated Value of Long-run Human Capital Economic Impact in 2010 Dollars.

(1) Age group	(2) 2010 Economic impact [Column 12, Table 3]	(3) Survival rate [Centers for Disease Control]	(4) Long-run economic impact [(1) × (2) × (3)]
21-24 (4 years)	$1,250,945.40	0.999294	$5,000,248.94
25-34 (10 years)	$16,682,445.69	0.998979	$166,654,129.13
35-44 (10 years)	$19,494,745.31	0.998205	$194,597,522.41
45-54 (10 years)	$35,731,300.65	0.9958	$355,812,291.86
55-64 (10 years)	$41,183,497.63	0.991287	$408,246,658.18
65-74 (10 years)	$16,509,082.50	0.980715	$161,907,048.44
			$1,292,217,898.95

The long-run economic impact estimated here can be juxtaposed with a more conventional short-run economic impact estimate to illustrate their relative magnitudes. These two estimates capture different facets of the benefits of an institution of higher education. The existence of the university in the local area generates additional economic activities from the expenditure of the university itself, its employees, its students, and visitors. These make up the short-run impact in the local area. The short-run estimate is on a yearly basis and its magnitude depends on the expenditures from the aforementioned for that particular year. The long-run economic impact, the main focus of the study, is derived from the human capital produced by the university estimated as the college–high school earnings differential throughout the alumni’s work life, but only includes graduates who choose to stay and work in the local area. The short-run economic impact estimate for the university in consideration using the I–O approach with IMPLAN for 2010 was estimated at $163.07 million. There were five spending components included in this estimate: (a) university operating expenditures, (b) personnel expenditures (employee compensation), (c) student expenditures, (d) visitor expenditures for students and faculty, and (e) additional spending by the university from its other activities not included in its operating expenditures (e.g., university bookstore, special athletic events, foundation). Furthermore, the economic impact from the creation of human capital, one of the major contributions of the university to the local economy, as estimated here was at $130.85 million for 2010. This 2010 human capital contribution likewise represents a short-run impact. The long-run economic impact, which constitutes the additional work life earnings of alumni who choose to stay and work in the local area, was estimated at $1.29 billion (in 2010 dollars). These short- and long-run estimates provide a more comprehensive representation of the university’s economic impact to the local area. A majority of the benefits from institutions of higher education arise from the creation of human capital alone and thus the ability to estimate this is critical.

Conclusions

Institutions of higher education typically utilize economic impact studies to articulate their value. It is an important public relations tool, particularly in the current environment of tightening state budgets. The most common economic impact studies conducted by institutions of higher education are those of the short-run approach—the additional economic activity and spending generated in the local area from the spending of the university and its constituents. This approach has very minimal data requirements and a relatively standard methodology is established and can thus be routinely carried out. As such, this type of approach can be easily done and a yearly economic impact estimate can be calculated. This is useful in illustrating the benefits of the existence of an institution in the local area in terms of additional wages and salary, jobs, and tax collections—benefits that are widely reported in the popular press and are good publicity for the institution. This allows the institution to have a dollar measure of its value readily available to various stakeholders and likewise for marketing purposes. It should be noted, however, that the principal contribution of state colleges and universities is the creation of human capital—a benefit that accrues toward the graduate’s work life and thus a long-run contribution. Although the human capital contribution was acknowledged in previous economic impact studies, data constraints have limited the ability to provide an accurate estimate.

This study provides an important contribution to the existing literature by developing a methodology to estimate the long-run human capital contribution through a combination and adaptation of the seminal methodology as introduced by Berger and Black (1993) and the U.S. Census Bureau’s methodology of estimating work life earnings. In addition, the methodology is applied to a public regional university with access to mostly primary data collected mainly for the purpose of this study, again allowing for a more precise estimate. The most critical piece of information is the counterfactual—the incremental income of the alumni who would have gone elsewhere in the local area “but for” the university. The additional challenge is estimating the increased work life earnings, which require adjustments for labor force participation rate and survival rate throughout a work life cycle. The overall methodology involves a two-step process: (a) estimating an earnings equation that controls for occupation, gender, age, ability, and other relevant factors; and then (b) predicting average earnings per age cohort adjusting for labor force participation rate and survival rate to be used to estimate additional work life earnings (the long-run economic impact) and, most important, also correcting for the counterfactual to remain true to the definition of economic impact.

The ability to take into account the counterfactual is important since it can have a significant impact on the size of the long-run economic impact. For instance, a university in an isolated rural area with the local area defined as the regional location of the institution might more likely be credited with virtually the human capital of all its alumni. This study defines the local area as the state where the institution is located, a broader definition, which meant more competition for an alternative university/college where its alumni could have attended. The estimate from the alumni survey indicates that, depending on the age cohort, the percentage of graduates who should be counted as a unique contribution of the university, range from 3% to 28% after also adjusting for potential migration (an approximation given data availability). Therefore, correcting for earnings differential and the counterfactual, the long-run human capital economic impact of the university under consideration to its state economy as of 2010 is estimated at approximately $1.29 billion in 2010 dollars.

Given the very different methodology and components of benefits that are captured from the short-run approach (from additional spending and its multiplier effect estimated on a yearly basis) versus the long-run approach (from the human capital that is a lifetime benefit), it is not surprising to see a substantial difference in the magnitudes of the economic impacts. These two approaches jointly provide a more complete illustration of the overall economic impact of the institution; ideally, all economic impact studies should include both. From a more practical standpoint, however, the short-run approach may be enough if the main purpose is to provide evidence of additional economic activity created by the university in the community. If the main purpose, however, is to justify the existence of the institution, it is imperative that the long-run human capital approach must be done in tandem with the short-run approach. After all, the primary purpose and contribution of institutions of higher education is the creation of human capital.

Given the need for the counterfactual to get an accurate estimate of the long-run economic impact, it is likewise important to take into consideration the potential effects of migration of college graduates and worker mobility throughout the work life. If not taken into account, the long-run economic impact is likely to be overestimated. This study made an attempt to capture migration but in a very limited capacity, given data availability. This is a limitation of the study. On the other hand, it offers a relatively straightforward methodology of estimating the long-run impact by adopting the work life earnings estimation methodology by the U.S. Census Bureau and identifying critical data necessary to carry out the procedure.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Tom Johansen is a clinical associate professor of finance at Oklahoma State University, where he also received his PhD in finance. His consulting and research interests focus on valuation, economic impact, and value investing strategies.

Kathleen Arano received her PhD in applied economics from Mississippi State University. She is an assistant professor of economics at Indiana University Southeast. She has published research on faculty retirement decisions and regulatory reform in the natural gas industry.

References

Armstrong

H. W.

(1992). The local income and employment impact of Lancaster University. Urban Studies, 30, 1653-1668.

Beck

Elliott

Meisel

Wagner

(1995). Economic impact studies of regional public colleges and universities. Growth and Change, 26, 245-260.

Becker

(1962). Investment in human capital: A theoretical analysis. Journal of Political Economy, 70(5), 9-49.

Becker

(1975). Human capital (2nd ed.). Chicago, IL: University of Chicago Press.

Berger

M. C.

Black

(1993). The long-run economic impact of Kentucky public institutions of higher education. Lexington: University of Kentucky.

Blackwell

Cobb

Weinberg

(2002). The economic impact of educational institutions: Issues and methodology. Economic Development Quarterly, 16, 88-95.

Bluestone

(1993). UMASS: An economic impact analysis. Boston: University of Massachusetts.

Brown

K. H.

Heaney

(1997). A note on measuring the economic impact of institutions of higher education. Research in Higher Education, 38, 229-240.

Brownrigg

(1973). The economic impact of a new university. Scottish Journal of Political Economy, 30, 123-129.

10.

Cooke

(1970). Analysing university student contribution to the economic base of the community. Annals of Regional Science, 4, 146-153.

11.

Day

J. C.

Newburger

E. C.

(2002). The big payoff: Educational attainment and synthetic estimates of work-life earnings. Washington, DC: U.S. Census Bureau.

12.

Dick

Wood

(1980). The impact of the Newcastle-upon-Tyne Polytechnic upon employment and income in the local economy. Newcastle upon Tyne, England: Newcastle upon Tyne Polytechnic.

13.

Economic Modeling Specialists Intl. (2014). Where value meets values: The economic impact of community colleges. Moscow, ID: Author.

14.

Goss

Paul

(1986). Age and work experience in the decision to migrate. Journal of Human Resources, 21, 397-405.

15.

Groen

J. A.

(2004). The effect of college location on migration of college-educated labor. Journal of Econometrics, 121, 125-142.

16.

Hefner

F. L.

(1997). Using input-output models to measure local economic impacts. International Journal of Public Administration, 20, 1469-1487.

17.

Impact Analysis for Planning. (n.d.). Home. Retrieved from https://implan.com/

18.

Iranzo

Peri

(2009). Schooling externalities, technology, and productivity: Theory and evidence from U.S. states. Review of Economics and Statistics, 91, 420-431.

19.

Ishitani

(2011). Exploration of interstate college and post-graduation migration in the United States. Paper presented at the 50th Annual Forum of the Association for Institutional Research, Toronto, Ontario, Canada.

20.

Julian

Kominski

(2011). Education and synthetic work-life earnings estimates: American Community Survey Reports. Washington, DC: U.S. Census Bureau.

21.

Kane

Rouse

(1995). Labor-market returns to two- and four-year college. American Economic Review, 85, 600-614.

22.

Kantrowitz

(2007). The financial value of a higher education. NASFAA Journal of Student Financial Aid, 37(1), 19-27.

23.

Kodrzycki

Y. K.

(2001). Migration of recent college graduates: evidence from the National Longitudinal Survey of Youth. New England Economic Review, January/February, 13-34.

24.

Kraushaar

J. L.

(1964). How much of an asset is a college? College and University Business, 36, 434-435.

25.

Leontief

(1936). Quantitative input and output relationships in the economic system of the United States. Review of Economics and Statistics, 18(3), 105-125.

26.

Lewis

J. A.

(1988). Assessing the impact of Woverhampton Polytechnic on the local community. Urban Studies, 25, 53-61.

27.

Mincer

(1974). Schooling, experience and earnings. Cambridge, MA: National Bureau of Economic Research.

28.

Moore

G. A.

(1979). Local income generation and regional income redistribution in a system of public higher education. Journal of Higher Education, 50, 334-348.

29.

Moretti

(2004). Workers’ education, spillovers and productivity: Evidence from plant-level production functions. American Economic Review, 94, 656-690.

30.

Pittsburg State University Economic Impact Statement. (2002). Pittsburg, KS: Pittsburg State University.

31.

Postsecondary Education Opportunity. (2003). Interstate migration and geographic mobility of college graduates 1977 to 2002. No. 130, 1-8.

32.

Regional Economic Models, Inc. (n.d.). Home. Retrieved from www.remi.com

33.

Rickman

D. S.

Schwer

(1995). A comparison of the multipliers of IMPLAN, REMI, and RIMS II: Benchmarking ready-made models for comparison. Annals of Regional Science, 29, 363-374.

34.

Romano

(1986). What is the economic payoff to a community college degree? Community/Junior College Quarterly, 10(3), 153-164.

35.

Shappiro

J. M.

(2006). Smart cities: Quality of life, productivity, and the growth effects of human capital. Review of Economics and Statistics, 88, 324-335.

36.

Schmidt

(1998, February 20). More states try to stanch ‘brain drains’, but some experts question the strategy. Chronicle of Higher Education, p. A6.

37.

Smith

Bissonette

K. K.

(1989). The economic impact of nonresident students on West Virginia’s economy. Research in Higher Education, 30, 229-238.

38.

Stokes

Coomes

(1998). The local economic impact of higher education: An overview of methods and practice. Association for Institutional Research for Management Research, Policy Analysis, and Planning, 67, 1-16.

39.

Winters

J. V.

(2011). Human capital, higher education institutions, and quality of life. Regional Science and Urban Economics, 41, 446-454.