From Foresight, No. 49
published 2007
Table A.1 and Table A.2 present the economic outcomes and the additional estimated amounts associated with the number of degree holders produced in each policy scenario. These estimates reveal the potential amount of income and revenue Kentucky stands to gain by investing in the education of its workforce. These are selected benefits however and do not provide a complete picture of the full benefits associated with doubling the number of working-age adults with a Bachelor’s degree or higher level of education.
Table A.1 shows that as the number of degree holders increases with each step of the intervention plan, the 2020 PCI increases accordingly, as does the 2020 aggregate personal income (API), the cumulative API over the time period 2000 to 2020, and the associated general fund (GF) revenue amounts. Table A.2 shows the additional amounts associated with each discrete change in the number of degree holders produced by that step.
Kentucky stands to gain up to $139.5 B more in additional API and $9.0 B more in additional GF revenue if it achieves reaching the national average in the percent of its working-age population with a bachelor’s degree or higher. Steps 1 and 5 account for over half (69 percent) of the additional degree holders needed to double the number of degree holders in the state by 2020. The associated additional API and GF revenue amounts over the time period 2000 to 2020 are $95.9 B and $6.2 B, respectively. The remaining steps are important, however, in that if they are not fulfilled, Kentucky will fall short of its goal by 31 percent and miss out on approximately $43.6 B more in API and $2.8 B more in GF revenue.
This technical appendix describes the data and methodology used to explain variations in past and future estimates of state-level per capita income relative to education. Table A.3 shows information regarding the variables used in the analysis.
Table A3: Names, Description, and Source of State-Level Variables, 1984-2003
A log-linear model explained the relationship between PCI and each of the independent variables:
ln (PCIit)=ait + B1tEDUit + B2tPCEMPit + B3tURBANit +uit
for all i = 1 to 50 for each of the 50 states, t = 1 to 20 for each of the 20 years from 1984 to 2003, and error term u~ N(0,1).
Each variable is hypothesized to have a positive relationship with PCI:
H0: B1 > 0B2 > 0
B3 > 0
HA: B1 = B2 = B3 = 0
All estimated coefficients for all 20 models were positive and statistically significant at approximately the 1 percent level. The adjusted R2 values for each of the models ranged from a low of 0.59 in 1994 to a high of 0.77 in 1988. Table A.4 shows the elasticities of each variable given by each model year.
The policy scenarios come from projections made by the Kentucky Council on Postsecondary Education (CPE or the Council). Their report “Double the Numbers: 2020 Targets and Projections Methodology” explains how they arrived at their estimated goals. The numbers provided in the intervention plan were obtained from the Council’s “Double the Numbers” brochure. Both reports are available online at http://cpe.ky.gov/planning/2020projections.
The compound annual growth rate (CAGR) of the elasticities listed in Table A.4 were used to project out to 2020 the estimated relationship between the independent variables of educational attainment, per capita private employment, and the percent of people living in an urban area, and the dependent variable, PCI, which are listed in Table A.5. The CAGRs for both per capita private employment and the urban population for Kentucky were used to project the growth in each of these variables to 2020. Using these changes and the ones provided in each of the policy scenarios, per capita income was predicted annually for the years 2000 to 2020. The predicted populations were used to obtain aggregate personal income for the state. Finally, a revenue elasticity of one was assumed in predicting the subsequent changes in revenue growth.
* Dr. Amy L. Watts is an economist at the Center. Return to text.
1 In 1984, 1989, and 1992 the levels of statistical significance for the estimated coefficients on EDU were slightly greater than 1 percent, but still well below the 5 percent statistical significance level.