An Assessment of the Community College's Influence on the Relative Economic Development of a County.

Abstracts

The study investigated the impact of establishing a community college upon a county's relative economic development. Census data from 1940 through 1990 were used in graphical analysis of short-time series, ratio comparisons, and ordinary least square regression analysis. The results indicated that establishing a community college potentially contributed 8% to 11% of a county's economic development. This estimated "catalyst" influence was accounted for by neither payroll, population growth, nor economic advantages that existed prior to establishment of the community colleges.

In an election year, perhaps more so than in other years, the topic of economic development takes centerstage. Every politician has a view of the best way to direct county, state, or national level economic development. Invariably, education, especially workforce education, occupies a prominent role in any proposal. Since the inception of the statewide community college system in 1963, North Carolina has included career education, workforce preparation, and transfer education as the three core missions of the community college. The state's efforts to include the community college system in a statewide economic development strategy have been recognized beyond its borders (Blake, 1981; Campbell & Faircloth, 1982; Scott, 1987; Melville & Chmura, 1991; Gracie, 1998).

Recently, community college leaders within the state have suggested that no other institution has played a more significant role in supporting economic development at local and state levels than the community colleges (Lancaster, 1998). Are such claims justified, or are they just rhetoric? It is easy to document a direct economic orientation. Historically, two-year colleges have stressed the importance of workforce and economic development in their missions and curricula (Eells, 1931; Koos, 1924). More recently, these same purposes have been echoed (Commission on the Future of Community Colleges, 1988; Bosworth, 1997; Garza & Eller, 1998). While this certainly qualifies as intent, is there evidence to support a direct link between the presence of community colleges and economic development?

Studies of community colleges' economic impact have focused largely upon documenting economic impact through the provision of jobs (payrolls), the translation of payrolls into the purchase of goods and services in the local economy, and the direct purchase of goods by the institution (Wellsfry, 1988; Jefferson College, 1998). While such an influence is not inconsequential, one of the community college system's goals is to improve the existing economic climate through the educational skills of its students. This reflects the mythical "value-added" component and the equally shadowy "catalyst" influence of the community college. As has been lamented by Katsinas (1994), existing studies have considered these only minimally.

The current study addresses this catalyst influence by assessing the community college's impact upon the relative economic development of the county in which it was established. We did not use college payroll and expenditures as the primary focus in determining economic impact. Rather, we designed the study to estimate the effect of a community college's presence and contributions in stimulating the general business climate of a county. Additionally, we investigated the presence of a community college on the potential improvement in the general educational level of a county's population.

Method

In our investigation of community colleges' economic and educational impacts, we used county-level data available through the national census. Counties represent a basic administrative and economic policy-developing unit. We reasoned that if community colleges do have economic and educational influences on a large scale, they would be evident at the county level.

The North Carolina Community College System came into existence about 1960. Of the state's 100 counties, 58 contain community colleges. All except 4 were established between 1958 and 1969. In the analyses, we compared selected economic and educational variables in counties with community colleges and in those without community colleges. Differences in educational and economic growth rates between "community college counties" and "noncommunity college counties" for the decades prior to 1960 and the decades following it might be attributed partially to the effect of community colleges being developed.

We used a graphical approach, suggested in Cook and Campbell (1979), to analyze the short-time series data. In this, aggregated data for counties on the individual educational and economic variables were compared from 1940 through 1990 by establishing trend lines. Trends in the 58 counties with community colleges were compared with those in the 42 counties without community colleges on all relevant variables. By looking at the patterns from 1940 to 1990, it was possible to observe and chart differential growth rates. Considering and eliminating some alternative explanations for trend differences created a stronger argument for the potential influence of community college development on the selected economic and educational results.

Based upon these initial results, follow-up analyses focused on a subset of counties and a single economic variable--total retail sales in the county. The consequent subset involved 39 counties--18 containing community colleges and 21 without community colleges. The rationale for this second set of analyses rested upon the following. In looking at the data for all 100 counties, it was apparent that community colleges generally had been established in counties that were somewhat more economically developed than other counties. Thus, noted differences in the economic development between the two groups of counties may have been attributable to initial economic advantages of the "community college" counties. To control for this initial economic advantage, we restricted the second set of analyses to those counties that had per capita retail sales between $85 and $149 in 1940. The per capita retail sales in 1940 for all counties in the state ranged from $27 to $336. The selected range of values encompassed the top 50% (less one county) of the noncommunity college counties and the bottom 33% (less three counties) of the community college counties. This range represented the functional overlap in the distributions of the per capita retail sales figures for the two sets of counties for that year; consequently, it controlled the variable.

A final group of analyses used ordinary least squares regression. The time-series analysis could demonstrate graphically a difference between the two groups of counties; the regression analysis statistically described the same scenario. The result provided a different perspective of the same phenomena, yet allowed for a statistical control of economic differences.

Variables Included in the Analysis

The variables included in the analysis were extracted from the U.S. Census data for 1990, as well as from the County and City Data Books (U.S. Census Bureau, 1983; 1972; 1967; 1956; 1947)available for each census 1940 through 1980. The year 1940 served as a cut-off for establishing a precommunity college baseline. This produced a balanced time series due to the three census periods following the development period for community colleges and the selection of 1940 resulting in three census periods prior to the development period.

The variables in the census and the County and City Data Books that seemed most appropriate for this study were those that represented specific economic and educational elements of a county's population. The census contained data on the percent of the population aged 25 and over who have graduated from high school and those who have graduated from college. The percent who graduated from college was not used because the necessary data were not available for 1940, 1950, and 1960 in the data sources we used. The only educational indicator available over the requisite period of time was the percent of the county's population that graduated from high school.

The economic indicators available were more varied and more numerous. These indicators included total employment, as well as employment in various sectors of a county's economy. While there were four to six such employment areas or industries listed for each census year, only two (agriculture and manufacturing) were identified with the required consistency. To provide a view of the changing workforce conditions over the six decades, we used the percentage of the county's population employed, the percentage of the county's population employed in agriculture, and the similar figure for employment in manufacturing. Family income was another economic indicator included. This provided a gross estimate of the economic well-being within the county.

Three additional variables allowed a closer look at the county's economic vitality: (1) the value of the agricultural products sold; (2) the value of the manufactured products produced within the county; and (3) the total retail sales in the county. These three variables reflect the economic vibrancy of a county in different sectors. For some analyses, an adjustment was made to these three variables to control for general population growth. This adjustment was accomplished by dividing the requisite economic indicator by the population of the county. For example, the total value of agricultural products was divided by the population of the county. This resulted in a value of agricultural products sold per capita within the county. These per capita figures produced a clearer picture of economic gains within a county that were more than just a function of population growth.

The selected variables provided an overview of the county's economic and educational "landscape" prior to and following the development of the community college system within the state. Changes that occurred in this landscape coinciding with and following the development of the community colleges may reflect the impact of these educational institutions.

Limitations

Certain limitations are associated with the study's methods and conceptual foundation. Economic climate analyses were restricted to using data from two production areas--manufacturing and agriculture--and a global retail sales indicator. The importance of the service sector, except as reflected in retail sales, was not addressed due to the unavailability of the requisite data. Likewise, the only information about the community college included in the study was its presence or absence. The presence or absence of a community college does not take into account higher education attendance patterns, the size of the community college, its budget, nor its curriculum and how these might contribute to economic development. Similarly, it is possible that there exist social, economic, or political influences--unrelated to economic factors present before the development of the community colleges--that were not reflected in the study's conceptual model. All these limitations represent a background against which the meaningfulness of the presented conclusions must be determined.

Results and Discussion

The results of the short time-series analyses were clear for the different variables. The trend lines indicated no consistent differences in the counties containing community colleges and those without them for percent of the population graduated from high school, family income, percent of the population employed, percent employed in agricultural occupations, and the percent employed within manufacturing occupations. There was no definite break within any existing pattern following the 1960-1970 decade.

The trend lines for these variables tended to be identical, and in some cases they practically were overlaid. An example of this is presented in Figure 1 for the percent of the population employed in manufacturing. As can be seen in the figure, the two trend lines are almost identical. In 1940, those counties that eventually contained a community college had a slightly higher percentage of the population employed in manufacturing than those counties that would not have a community college. In 1950, this discrepancy shrank and in succeeding decades the proportion became even less. In 1980 and 1990, the proportions were the same.

[ILLUSTRATION OMITTED]

The results from this initial set of analyses suggest that the presence of the community colleges did not make any difference in terms of the overall educational level (as indicated by the high school graduation rate). Likewise, it did not seem to have a substantial effect on either the total family income, the percent employed, or the percentages employed within agriculture and manufacturing. However, there was a noticeable difference in the value of agricultural products sold between those counties containing community colleges and those not containing community colleges. Uniformly, counties not containing community colleges had lower agricultural sales. Unlike the two variables discussed next, the relative magnitude of the difference between community college and noncommunity college counties has decreased across the six decades. Such a trend would not suggest a community college influence.

The results for the other two economic variables--the value of manufacturing products sold and the total retail sales in the county--seemed to suggest that community colleges potentially contributed to increases in the value of manufactured products sold and total retail sales. In the interest of journal space, only the graphical results for retail sales (Figure 2) will be presented. It should be noted that the trend lines for the value of manufactured products were almost identical to those for total retail sales.

[ILLUSTRATION OMITTED]

In comparison to the value of manufactured products sold, the retail sales trend may be the more accurate economic development indicator since it is more sensitive to the total economic climate within the counties and the state. Between 1940 and 1990, the importance of manufacturing decreased in the state while the importance of service industries increased. As previously indicated, service industry data for all the required years were not available in the data sources that were used.

The trend lines for total retail sales indicated a discrepancy in 1940 between the two sets of counties. This small difference increased slightly in 1950 and was maintained through 1960. Following that, the differences in the trend lines increased. Beginning in 1970, this widening gulf suggests that something was taking place within the counties that made the discrepancy in retail sales greater than it had been before that time. Since this corresponded with the development of the community colleges, it is conceivable that the presence of the community college contributed to a general business climate that materialized in a stronger local economy and hence greater retail sales.

An alternative interpretation to the "community college effect" is that the differences might be a function of population--counties with larger populations naturally would be expected to have larger retail sales than counties with smaller populations. Thus, if a county had a slight economic advantage prior to creating a community college and this economic advantage resulted in more jobs and more people, then later retail sales differences could stem from simple population growth. Such an interpretation seems reasonable on the surface; however, it is not as plausible when viewed from another perspective.

The "influence" of population on retail sales and the value of manufactured products can be removed by viewing these two variables as a function of population (a per capita figure). To achieve this, each of the two economic indicators was divided by the population of the county for the appropriate year. The per capita indicator for retail sales was used to generate the graphical results presented in Figure 3. For the per capital retail sales and the value of manufactured goods per capita, a general pattern similar to that observed in Figure 2 was evident. There was a small difference in the indicators across the two sets of counties from 1940 through 1960. Beginning with the 1970 census figures, this difference began increasing. The discrepancy was more pronounced for the value of manufactured products than for retail sales, but the same general pattern was present in both.

[ILLUSTRATION OMITTED]

These results suggest that during the decade of the 1960s, forces began acting upon the counties' economies that were reflected in an increasingly discrepant rate of retail sales. Since the development and establishment of the community colleges occurred concurrently, these institutions may have contributed to an economic environment that produced more development than was present in the counties without the community colleges.

Since a potential community college impact was evident in the total retail sales and the value of manufactured goods, a second set of analyses was performed. In these, only those counties that were relatively equal in their per capita retail sales in 1940 were used. As previously stated, our rationale for such an approach was to control for existing economic advantages within counties. Generally, community colleges were established within counties that were "better off" than counties that did not get community colleges (per capita retail sales of $788 versus $580 in 1960). The observed trends from the previous two analyses may have been attributable to inherent economic advantages as opposed to advantages "spawned" by the community college. To control for the existing economic advantages of a county, we homogeneously selected per capita retail sales in 1940. As a result of using per capita retail sales rather than the total retail sales, we controlled for population size as well. Thirty-nine counties were part of this analysis--18 contained community colleges and 21 did not.

Figure 4 contains the results from the trend line analysis on per capita retail sales for the two sets of counties. The pattern is somewhat like that observed in Figure 3, in which population size had been controlled. In Figure 4 where both population size and "economic advantage" had been controlled, the trend lines almost perfectly overlay until the 1990 census year. From 1980 to 1990, the community college counties began to have greater relative per capita sales than did the noncommunity college counties.

[ILLUSTRATION OMITTED]

The same analysis for per capita value of manufactured goods yielded results somewhat intermediate between those seen in Figures 2 and 3. The trend was much more indicative of a pronounced economic impact than was observed for the per capita retail sales in Figure 4. Beginning with the 1970 census figures, a difference materialized in the trend lines, and this increased until there was a large discrepancy for the 1990 census figures. Perhaps this trend reflected a difference in the developing economic bases of community college and noncommunity college counties.

Another analytical perspective provided additional insight into the possibility of a "community college effect." Given that a short time series was being observed, it is possible that the divergence in the slopes of the two lines beginning in 1970 may be purely a function of a geometric link between the variables. Rather than expecting the counties' trend lines to remain parallel, it may be that an initial economic advantage would be "doubled" or "tripled" over the course of a decade. Hence, with each passing decade the initial discrepancy would increase geometrically. With the 1940 differences between the two sets of counties being small, it is plausible that it took two additional census periods for the inherent discrepancy to become obvious.

We addressed this possibility by creating an index of economic comparability for each decade. A ratio was developed for each decade by dividing the average of the appropriate economic indicator in the community college counties by the average of the corresponding figure in the noncommunity college counties. The result indicated how much greater the economic variable was in the counties that ultimately had a community college than in those without a community college. These were done for all 100 counties and for the reduced set of 39 counties in which economic potential in 1940 had been controlled. Any geometric pattern would be reflected in these ratios. Table 1 contains these ratios.

Table 1

Trend in the value of manufactured goods and total retail sales of
these values in the community college counties and the noncommunity
college counties

                       Ratio in the value of

                                        Per Capita
Census   Manufactured    Retail Sales   Manufacture-   Per Capita
Year     Goods (N=100)   (N=100)        d Goods        Retail Sales
                                        (N=39)         (N=39)

1940         9.96            4.93           1.60           1.02

1950         6.59            4.96           1.26           1.02

1960         2.29            4.90           1.33            .99

1970         9.13            5.05           1.46            .98

1980         5.61            5.16           1.49           1.02

1990         5.59            5.35           1.56           1.10

For the counties that ultimately contained a community college, these figures show that in 1940 the value of manufactured goods was approximately 10 times greater than the corresponding figure for the noncommunity college counties. This ratio dropped in 1950 and 1960, rose in 1970, and dropped again in 1980 and 1990. Data from the 39 county subset echoed the absence of a well established "community college" pattern. Beginning with the 1950 census, the per capita value of manufactured goods indicates a 3% to 13% increase each decade. If the presence of the community college was having a marked influence on the economy, then one might anticipate that these two sets of ratios would reflect a stable pattern up to the 1970 census figure and then show a marked increase. Such was not the case for the variables based on the value of manufactured goods.

For total retail sales, the picture was different. As indicated before, the retail sales figure may convey a better idea of what is taking place in the total economy of a county. As reflected in the ratios of retail sales from 1940 to 1960, the counties containing a community college had retail sales that were approximately five times greater than the total retail sales in counties without a community college. The ratio remained rather stable with figures of 4.93, 4.96, and 4.90 for the period. Commencing in 1970 and continuing afterward, the ratio began to increase consistently over the next three-decade period. This is the same pattern that was seen in short time-series analyses presented in Figures 2 and 3. The ratios based upon per capita retail sales figures suggest that once population size and the state of the economy in 1940 (the 39 counties) had been controlled, the economies for the two sets of counties functioned very much on par until the 1980's. In the 1980's, the community college counties posted a 10% advantage over the noncommunity college counties.

Since these latter figures take into account the influence of population growth and economic levels in 1940, they may provide the best picture of the community college's potential influence on a county's economic development. In considering this, the following point could be made: Since the community college counties had a 2% greater retail sales per capita in 1940, the 10% difference in 1990 may represent nothing more than realizing the greater economic potential reflected in the 1940 figure. One response to this is that if the 10% difference represented some untapped economic potential, then why would it take 50 years (approximately 15 to 20 years after community colleges were developed) to manifest itself? On the other hand, a 15-year lag in the economic influence of establishing community colleges could be expected.

The results of the short time-series analyses and the ratio comparisons suggest that community colleges potentially contribute to a county's economic development (represented by total retail sales and the value of manufactured goods). Multiple regression analyses were performed to investigate this possibility further. These were performed on the data to determine if a single regression line best represented the two sets of counties or whether separate regression lines were more appropriate. The homogeneity condition was met. Summary results of the regression analyses are presented in Table 2. In the analyses, the community college variable accounted for from 0.0% to 2.2% (M = 0.9%) additional variance that was not attributable to the precommunity college economic climate.

Table 2

Summary results of regression analyses estimating community colleges'
potential influence on economic indicators

                           R-Square for   R-Square
                           Correspond.    Change     Beta for
Depende-    Census         1940, 1950,    for        Comm.
nt          Year     N     and 1960       Comm.      College
Variable                   Independent    College    Variable
                           Variables      Variable

              1970    39           .591       .007      -.082
Retail
Sales Per     1980    39           .436       .002       .037
Capita
              1990    39           .344       .022       .152

              1970    39           .814       .000       .019
Value
Man.          1980    39           .673       .016       .134
Goods Per
Capita        1990    39           .384       .006       .080

              1970   100           .804       .001       .036
Retail
Sales Per     1980   100           .742       .008       .110
Capita
              1990   100           .631       .019       .170

              1970   100           .840       .000      -.015
Value
Man.          1980   100           .582       .011       .117
Goods Per
Capita        1990   100           .391       .014       .132

Summary

It could be argued that the noted economic differences between the community college counties and the noncommunity college counties may have been attributable to the presence of political, social, and economic forces within those counties that were present prior to founding the community colleges. Further, these influences resulted in both community colleges being placed in the county and an economic foundation being built that began to materialize with the 1970, 1980, and 1990 data. Since community colleges were established in counties that in general were more economically well off in comparison with the counties in which they were not established, this has certain intuitive appeal.

The rebuttal rests in the nature of the data analyzed and what they show. In the analyses, the included economic data reflected development before the community colleges were established. None of the data indicated suggest economic forces existing prior to 1960 that would produce the observed post-1960 trends. This is especially true in the regression analyses. In these, the existing pre-1960 economic indices used in the study were considered first and following this, the potential community college impact was assessed. If the economic gains linked to community colleges' presence were an outgrowth of preexisting social, political, and economic forces, the following must be true. Of necessity, these influences would have been independent of the pre-1960 economic state. Thus the intuitive appeal of the "existing forces" argument for both the development of the community colleges and the expansion of county economies rests upon the presence of some forces that were tied in with the state of post-1960 economies, but were not related to the state of the pre-1960 economies. This does not seem very plausible.

In looking at the total picture, it seems safest to conclude that the mere presence of the community college has not taken struggling economies and transformed them into booming ones. Rather the data suggest that the presence of community colleges may have positively impacted the economic climate within a county. This statement is true even after taking into account the economic climate within the county over the 30 years prior to the major development period of the state's community colleges.

What is the magnitude of this potential influence? The ratio comparisons suggest an 8% difference (10% less the initial 2% advantage). This estimate employs the 1990 figure at which point the discrepancy between the community college and noncommunity college counties became evident. The regression analysis for the same 1990 per capita retail figures indicated that 2.2% of the variation in per capita retail sales for 1990 was associated with the presence of a community college. The generated regression equation was used to estimate the influence of adding a community college to counties without them on per capita retail sales. The resulting figure suggested an average 11% increase being produced by the presence of a community college.

Based upon these, the presence of a community college was associated with as much as 8% to 11% of the economic development within a county, as indicated by retail sales. Since the community college payrolls within the state represent only 0.6% of the payrolls within counties with community colleges, the 8% to 11% estimate of the community colleges' potential economic influence is not a function of adding another payroll. In North Carolina, community colleges tended to have been established in counties that were more economically advantaged than counties in which they were not established. Even taking this initial economic advantage into account, the total set of results suggests a community college potentially has a small but important impact on the economic climate within a county.

References

Bosworth, B. (1997). Economic development, workforce development, and the urban community college. Community College Journal, 67, 8-13.

Blake, L. J. (1981). Up front in economic development. Community and Junior College Journal 52. 3,16 - 17.

Campbell, D. F. & Faircloth, D. M. (1982). State models for economic development. Community and Junior College Journal 52. 7, 18-19.

Cook, T. & Campbell, D. (1979). Quasi-experimentation: Design and analysis issues for field settings. Chicago, IL: Rand McNally College Publishing Company.

Commission on the Future of Community Colleges. (1988). Building communities: A vision for a new century. Washington, D.C.: American Association of Community and Junior Colleges.

Eells, W. C. (1931). The Junior College. Boston: Houghton Mifflin.

Garza, H. & Eller, R. (1998). The role of rural community colleges in expanding access and economic development. New Directions for Community Colleges, 103, 31-42.

Gracie, L.W. (1998). Measurable outcomes of workforce development and the economic impact of attending a North Carolina community college. New directions for community colleges 104. 26 53 - 60.

Jefferson College. (1998). The economic impact of Jefferson College on the community and state--FY 1997. Hillsboro, MO: Author. (ERIC Document No. ED 434 688).

Katsinas, S. (1994). A review of the literature related to economic development and community colleges. Community College Review, 21, 67-80.

Koos, L.V. (1924). The Junior College. Minneapolis, MN: University of Minnesota Press.

Lancaster, H. M. (1999, March). A major force in economic development: A challenge for the North Carolina community college system. National Institute for Leadership Institutional Effectiveness (NILIE) Conference keynote address Asheville, NC, March 29-31, 1998 (ERIC Document No. ED 432 319).

Melville, J. G. & Chmura, T. J. (1991). Strategic alignment of community colleges and state economic policy. New Directions for Community Colleges 19. 3, 7-15.

Scott, R. (1987). Proven partners: Business, government, and education. Community, Junior, and Technical College Journal 57. 3 16 - 19.

U.S. Bureau of the Census (1983). County and city data book, 1983. Washington, DC. U.S. Government Printing Office.

U.S. Bureau of the Census (1972). County and city data book, 1972. Washington, DC: U.S. Government Printing Office.

U.S. Bureau of the Census (1967). County and city data book, 1967. Washington, DC: U.S. Government Printing Office.

Kevin L. Pennington is an assistant professor in the Department of Educational Leadership & Foundations at Western Carolina University. [e-mail]

Robert B. Pittman is a professor in the Department of Educational Leadership & Foundations at Western Carolina University. [e-mail]

J. Casey Hurley is a professor in the Department of Educational Leadership & Foundations at Western Carolina University. [e-mail]