Verma 1

December 4, 2020

Himani Verma

ECO 441K: Introduction to Econometrics

Instructor: Dr. Helen Schneider

The University of Texas at Austin

The Effect of Increase in Minimum Wage on Community College Enrollment

Final Paper

Abstract

In this paper, we use a difference in differences model to estimate the effect increase in

minimum wage on community college enrollments in the states of Maryland (treatment group)

and Pennsylvania (control group). Our dataset comprised of a total of 120 observations across 30

community colleges in the two states from 2013 to 2017 (excluded 2015 to denote policy

change). We found that the p-value of our difference in differences estimator was statistically

insignificant (p-value > 0.05). The paper did report significance of tuition and financial aid

variables. The results of this paper can be used by policymakers to devise legislation pertaining

to funding of postsecondary education.

 Verma 2

Introduction

Community colleges account for a significant share of America’s postsecondary education

institutions. In fall 2014, 42% total and 25% of full-time undergraduate students were enrolled in

community colleges (College Board 2016). Among other factors, community colleges are

preferred institutions for low-income students due to comparably lower costs of attendance than

its public and private counterparts. In 2015-2016, the average tuition was about $3,435 for

attending full-time public two-year in-district institutions which compares to about 37% of the

cost of attendance of public four-year in-state colleges and 11% of the full-year tuition of private

nonprofit four-year institutes (College Board 2016). Despite lower tuition costs, enrollment rates

of community colleges have declined from 29% to 25% in line with the overall contraction in

undergraduate enrollment by 4% between 2010 and 2014 (College Board 2016; NCES 2020).

Reduction in state budgets for higher education is a key contributor to lower enrollment rates

(CBPP 2019). Budget cuts are associated with further tuition hikes and less financial aid

disbursements. Less college financing opportunities becomes a significant concern for

prospective college students in the low-income categories.

Minimum wage has consistently been a key legislative topic in the United States. While

the federal minimum wage has not been raised from $7.25 since 2009, many states have

implemented their own adjusted rates to account for changes in purchasing power and cost of

living (NCSL 2020). Stagnant low levels of income cause great social and economic harm to the

society. It results in perpetuating income inequality (Militaru et al. 2019), sustaining low worker

productivity (Levine 1992), and causing additional budgetary stress due to redeployment of

funds towards issues of public health and social repair. Therefore, increasing minimum wage

rates is an important mission for policymakers.

 Verma 3

Literature Review

Existing literature measuring the increase in minimum wage with enrollment in educational

institutions is primarily centered around the teenage labor market. In addition to this, these

studies offer little or ambiguous evidence to make any extrapolative conclusions for our

population of interest. On one hand, Neumark et al. (1995) find that teenagers respond positively

to dropping out of schools with increase in minimum wage rates. These results are further

replicated in the empirical studies of Chaplin et al. (2003) and Pachecoa et al. (2007). A $1

increase in minimum wage raises the non-continuation ratio of high school students by about 10

percent (Chaplin et al. 2001). Pachecoa et al. (2007) further evidence that the labor market

enqueuing of school going population in the age group of 16 to 19 years rose by 1.5 percent in

response to a 10 percent hike in wage rates. However, their results varied across adults in the age

group of 20 to 24 years old. Higher wage rates did not significantly impact the educational

enrollment efforts of the older subgroup (Pachecoa et al. 2007). This model was a considerable

development of the earlier rudimentary analysis of Card (1992) that concluded no direct

association between rise in minimum wage and enrollment rates when isolated from the overall

increase in employment.

The proposed study contributes to the previous literature by narrowing its focus to

postsecondary education enrollment rates of a comparably income sensitive subgroup (College

Board 2016). We are interested in estimating the effect of minimum wage increases on student

enrollment rates in community colleges. Our study uses a difference-in-differences framework

that measures a change in the minimum wage in a state with community college enrollments in

both the treatment and control states. In January 2015, Maryland increased its minimum wage by

10.3% from $7.25 to $8.00, with additional step increases planned to reach the target wage rate
 Verma 4

of $10.10 in 2018. Maryland’s neighboring state, Pennsylvania maintained its minimum wage at

the federal rate of $7.25. We use community colleges of these two states as our unit of analysis

and collect data at both the institutional and state level.

Empirical Model

We use a difference-in-differences model to estimate the relationship between community

college enrollment and minimum wage rates. Our two groups of interest are the states of

Maryland and Pennsylvania. We chose these states because they shared similar postsecondary

education statistics. Currently, there are 14 community colleges in Maryland and 16 community

colleges in Pennsylvania. The total community college enrollment figures for the two states

differed by merely 1,190 students in the year 2013. In this paper, we collect data for two years

before and after the year of policy intervention, i.e. from 2013 to 2017 with the year 2015 used

as the baseline year of policy intervention in our model. In this paper we estimate the following

regression model:

𝐸𝑛𝑟𝑜𝑙𝑙 = 𝛼 + 𝛽1 𝑀𝐷 + 𝛽2 𝑀𝑖𝑛𝑤𝑎𝑔𝑒 + 𝛽3 (𝑀𝐷 ∗ 𝑀𝑖𝑛𝑤𝑎𝑔𝑒) + 𝛽4 𝑇𝑢𝑖𝑡𝑖𝑜𝑛 + 𝛽5 𝐴𝑖𝑑 + 𝛽6 𝑃𝑒𝑙𝑙

+ 𝛽7 𝑈𝑟𝑎𝑡𝑒 + 𝛽8 𝐼𝑛𝑐𝑜𝑚𝑒 + 𝛽9 𝐵𝑙𝑎𝑐𝑘 + 𝛽10 𝐴𝑠𝑖𝑎𝑛 + 𝛽11 𝐻𝑖𝑠𝑝𝑎𝑛𝑖𝑐 + 𝛽12 𝑂𝑡ℎ𝑒𝑟

+ 𝛽13 𝐹𝑒𝑚𝑎𝑙𝑒

In the model above, Enroll variable measures fall enrollment for community colleges in

Maryland and Pennsylvania. Our treatment variable MD is a dummy variable which is coded 1

for the state of Maryland which is our treatment group. The coefficient of the treatment variable,

𝛽1 is the estimated pre-treatment mean difference between community colleges in Maryland and

Pennsylvania. Our time variable is Minwage is also a dummy variable with value 1 for years

after 2015 when Maryland increased its minimum wage. We obtain data for two years before and
 Verma 5

after our policy intervention year, i.e. from 2013 to 2017 with 2015 excluded from our dataset.

The coefficient of the treatment variable, 𝛽2 is the pure effect of the passage of time in the

absence of the actual intervention, i.e. minimum wage increase in 2015 in Maryland. The

coefficient 𝛽3 is the difference in differences estimator. It tells us whether the expected mean

change in community college enrollment from before to after Maryland’s minimum wage

increase in 2015 was different in Maryland and Pennsylvania.

Tuition variable measures the published full-time undergraduate in-state tuition and

required fees in dollars. Aid variable captures the percentage of enrolled full-time undergraduate

students receiving any grant or scholarship aid. The Pell variable represents the maximum

Federal Pell Grant in a given year. The variable Urate is the seasonally adjusted annual

unemployment rate in percentage for each state. Income variable includes the state annual

median income. We further control for race of enrolling students; white is excluded. Finally, we

control for gender by including total female student enrollments.

Data

Community college enrollment data was obtained from the Integrated Postsecondary Education

Data System (IPEDS) and is based on submissions to the National Center for Education

Statistics. The data was collected for 14 Maryland and 16 Pennsylvania community colleges

from 2013 to 2017 with data from 2015 excluded to account for the policy intervention year. The

data is available from https://nces.ed.gov/ipeds/datacenter/InstitutionByName.aspx.

Data for tuition, aid, race, and gender variables is also sourced from IPEDS. Information

about the IPEDS is available at https://nces.ed.gov/ipeds/use-the-data. We use total tuition and

required fees reported in dollars for full-time, in-state undergraduate students for the full
 Verma 6

academic year starting from the fall semester in our dataset. The data for aid is measured as the

percentage of total enrolled undergraduate students receiving any type of grant or scholarship aid

from the federal government, state or local government, institution, or institution known sources,

or availing federal or state student loans. Race and gender data are recorded as a percentage of

the total student enrollment for a given year.

The data for maximum Pell grants for each year is obtained from the publicly available

data books of the U.S. Department of Education. The Federal Pell Grant Program data reports

can be accessed at https://www2.ed.gov/finaid/prof/resources/data/pell-data.html. State

characteristics include the unemployment rate and annual median income. Data for these were

obtained for the years from 2013 to 2017 (2015 excluded) from the Federal Reserve Economic

Research. State-level data can be found at: https://fred.stlouisfed.org/categories/27281.

Table 1 below presents descriptive statistics.

 Verma 7

Table 1. Descriptive Statistics

Variable Mean Minimum value Maximum value

(st. dev)
Enrollment rates 8,591.84 621 26,155
(6,420.61)
Maryland (MD) 0.53 0 1
(0.50)
Minimum wage 0.50 0 1
(0.50)
Interaction 0.27 0 1
(Maryland * Minimum wage) (0.44)
Tuition rates 7,093.75 2,462 11,310
(1,665.70)
% of students receiving aid 71.15 39 97
(15.16)
Pell grants 5,775.50 5,645 5,920
(102.32)
Unemployment rates 5.54 4.20 7.30
(1.003)
Median income 67,318.65 55,156 82,093
(9,475)
% of Black students 18.84 3.04 79.14
(17.37)
% of Asian students 3.14 0 11.82
(2.82)
% of Hispanic students 7.76 0.98 34.28
(7.12)
% of students of other races 9.06 2.11 35.04
(5.04)
% of female students 60.06 47.72 80.45
(4.33)

Table 1 above shows a wide variance in enrollment across community colleges from 621

students in Garrett College, Maryland in 2017 to 26,155 students in Montgomery College,

Maryland in 2013 with the average of 8,591.84 students. Tuition rates also varied significantly,
 Verma 8

with the lowest tuition rates recorded at $2,462 for Baltimore City Community College in

Maryland in 2013 and 2014. Tuition rates were the highest at $11,310 for Community College of

Beaver County in Pennsylvania in 2017. The mean tuition rate across the two states over the four

years was $7,093.75. The percentage of students receiving financial aid also differed greatly.

Only 39% of students in Anne Arundel Community College in Maryland received financial aid

in 2016 and 2017. On the other hand, Garrett College in Maryland had 97% of students

supported financially in 2013, 2016 and 2017. Average of 71.15% of students received some sort

of financial aid. Furthermore, race and gender demographics also varied significantly across

colleges.

Empirical Results

Regression results in Table 2 below reject our difference in differences null hypothesis that

community college enrollments did not change in Maryland and Pennsylvania from before and

after the minimum wage change in Maryland in 2015 (Prob > F = 0.0000). Our regression model

has therefore statistically significant explanatory power.

 Verma 9

Table 2. Regression Results

However, neither of our treatment (MD), time (Minwage), or their interaction variables

are statistically significant at the 10% significance level (p-value > 0.1). We also have

statistically insignificant results for Federal Pell Grants and our state characteristics, i.e.

unemployment rates and annual median income (p-value > 0.1). Additionally, our gender

demographic variable was also an insignificant determinator (p-value > 0.1).

Important determinants of community college enrollment include tuition and financial

aid. The p-value of tuition is less than 0.01 and therefore the variable is significant at the 1%

level (p-value < 0.01). This means that for the given dataset, a 1% increase in tuition increases

community college enrollment by 1.39. The positive sign on the variable coefficient of the

variable may also be indicative of good quality of education. With an increase in improvement of
 Verma 10

school quality by 1%, enrollment rates rise by 1.39. Financial aid is another statistically

significant variable but at the 10% level (p-value < 0.1). College enrollment rates decrease by

61.93 with every 1% increase in financial aid. Furthermore, our results were also significant for

select racial groups. The p-value for the groups Black and Asian were statistically significant at

the 1% level (p-value < 0.01). The racial group labeled as “other” included students identifying

as either Native Hawaiian or Pacific Islander, American Indian or Alaskan Native, multiracial, or

foreign. The results were statistically significant for this group at the 10% level (p-value < 0.1).

We can conclude from these findings that people identifying as Blacks, Asians, or Other are

more likely to enroll in community colleges in comparison with the omitted racial group of

Whites. The paper however does not find any statistical significance for Hispanics (p-value >

0.1).

Diagnostic Tests

Our empirical model makes multiple assumptions of errors. We run a series of tests to check for

these errors, namely heteroscedasticity (Breusch-Pagan Test), multicollinearity (Variance

Inflation Factor or ‘VIF’ Test), normality (Skewness-Kurtosis All Normality Test), and omitted

variables (Ramsey Regression Equation Specification Error Test or ‘RESET’). Results of these

tests along with necessary corrections have been presented as follows.

Breusch-Pagan Test

We use the Breusch-Pagan Test to test for heteroscedasticity of errors in our regression model.
 Verma 11

Table 3. Breusch-Pagan Test

Since our F-statistic is significant at the 5% level of significance, we reject the null hypothesis

that the residuals are homoscedastic. To correct for the failure of our test, we use a log

transformation of our dependent variable and rerun our regression.

 Verma 12

Table 4. Breusch-Pagan Test with log transformation

VIF Test

The Variance Inflation Factor or the VIF Test is used to detect multicollinearity.
 Verma 13

Table 5. VIF Test

Since our mean VIF value is marginally greater than the tolerance value of 10, our model fails

the test. We will report this failure as a limitation to our study.

Skewness-Kurtosis All Normality Test

We use the Skewness-Kurtosis Test to check our model assumption of the normality of residuals.

Table 6. Skewness-Kurtosis Test

 Verma 14

We see that the Prob>chi2 statistic is significant at the 5% level, and hence we reject the null

hypothesis that our data is normally distributed. We repeat the test with a log transformation of

our dependent variable as a correction to the failure.

Table 7. Skewness-Kurtosis Test with log transformation

Ramsey Test

Finally, we check whether the non-linear combinations of the fitted values explain the response

variable of our regression model with the Ramsey Test.

 Verma 15

Table 8. Ramsey Test

We note that the p-value is significant at the 5% level, implying that our model failed the RESET

diagnostic. As a rectification to this inadequacy, we repeat the test on our log transformed

regression model.

Table 9. Ramsey Test with log transformation

 Verma 16

We see that our model fails test even with the log transformation (p-value < 0.05). This is

another notable limitation of our study.

Conclusions and Policy Implications

Empirical results reject the null hypothesis of our differences in differences model. Community

college enrollments did not remain consistent in the states of Maryland and Pennsylvania from

before and after the minimum wage change in Maryland in 2015. However, our difference in

differences model coefficients were not statistically significant. Therefore, minimum wage

increase is not an important determinant of community college enrollment. This result is

consistent with Pachecoa et al. (2007) that found an insignificant impact on the enrollment levels

of students in the 16–24 years old cohort due to changes to minimum wages. The population

group is the aforementioned study is relevant to our paper since 78% of total full-time

undergraduate student enrollments in two-year public institutions are under the age of 25

(College Board 2017). Policymakers should instead focus on lowering tuition rates and

expanding financial aid disbursements. Such policy interventions may include improvement of

the Free Application for Federal Student Aid (FAFSA) program, additional appropriations by

state governments to colleges to allow institutions to charge lower tuition rates and/or to offset

listed tuition prices through institutional grant aids, expansion of scholarship opportunities at all

three levels of government, and revisions in interest rates for student loans. Additionally, our

empirical results also represent the variance in enrollment across racial groups, i.e. greater

enrollment of Blacks, Asians, and other races which may be indicative of their low-income and

cultural backgrounds (Park et al. 2019).

Results of this study are not without limitations. First, our selection of treatment and

control groups is subject to bias and may have very different trends in their outcomes. Second,
 Verma 17

minimum wage increased in our treatment state increased stepwise after our baseline intervention

year. This continuous increase in minimum wage may introduce error in our results. Finally,

difference in differences might not be the most suitable model for our study. We noted that our

model marginally failed the diagnostic test for multicollinearity (VIF test) and also could not be

corrected for general specifications even with a log transformed model (Ramsey Test). A fixed

effects model may offer better insights about our question of interest.
 Verma 18

References

Card, D. (1992). Do Minimum Wages Reduce Employment? A Case Study of California, 1987-

89. Industrial and Labor Relations Review, 46(1): 38-54.

Center on Budget and Policy Priorities (2019). https://www.cbpp.org/research/state-budget-and-

tax/state-higher-education-funding-cuts-have-pushed-costs-to-students. Accessed

November 10, 2020.

Chaplin, D.D., Turner, M.D., and Pape, A.D. (2003). Minimum wages and school enrollment of

teenagers: a look at the 1990’s. Economics of Education Review, 22: 11-21.

Levine, D. (1992). Can Wage Increases Pay For Themselves? Tests with a Productive Function.

The Economic Journal, 102(414): 1102-1115.

Militaru, E., Popescu, M.E., Cristescu, A., and Vasilescu, M.D. (2019). Assessing Minimum

Wage Policy Implications upon Income Inequalities. The Case of Romania.

Sustainability, MDPI, 11(9): 1-20.

National Center for Education Statistics: The Condition of Education (2016).

https://nces.ed.gov/programs/coe/pdf/Indicator_CHA/coe_cha_2016_05.pdf. Accessed

November 10, 2020.

National Conference of State Legislatures. (2020). https://www.ncsl.org/research/labor-and-

employment/state-minimum-wage-chart.aspx. Accessed November 10, 2020.

Neumark, D., and Wascher, W. (1995). Minimum Wage Effects on Employment and School

Enrollment. Journal of Business & Economic Statistics, 13(2): 199-206.

 Verma 19

Pachecoa, G.A., and Cruickshank, A.A. (2007). Minimum wage effects on educational

enrollments in New Zealand. Economics of Education Review, 26: 574-587.

Park J.J., and Assalone A.E. (2019). Over 40%: Asian Americans and the Road(s) to Community

Colleges. Community College Review, 47(3): 274-294.

The College Board (2016). https://research.collegeboard.org/pdf/trends-community-colleges-

research-brief.pdf. Accessed November 10, 2020.

The College Board (2017). https://research.collegeboard.org/pdf/trends-student-aid-2017-full-

report.pdf. Accessed November 10, 2020.

Wescher L, Hutchinson T, and Rannou A. (2019). Minimum Wages, Employment, and College

Enrollment. The American Economist, 64(1): 3-18.