An Examination of Factors Affecting the Financing of College Education: an Intercountry Comparison

ABSTRACT - The role of human capital in economic development is well recognized and researched. What is not that well researched is the sources of funds for higher education, in particular, cross cultural comparisons. This study attempts to fill this void by synthesizing the work done separately at the University of Hawaii, Akita University and the University of the Ryukyus. The framework for analysis is the household production function approach to consumer behavior. Results from this study indicate that student age and time allocated to work are positively related to students’ financing their college education themselves in both countries. Other variables which are not significant also point out to some key policy implications.

Citation:

John F. Yanagida and Mahendra Reddy (1998) ,"An Examination of Factors Affecting the Financing of College Education: an Intercountry Comparison", in E - European Advances in Consumer Research Volume 3, eds. Basil G. Englis and Anna Olofsson, Provo, UT : Association for Consumer Research, Pages: 192-196.

European Advances in Consumer Research Volume 3, 1998      Pages 192-196

AN EXAMINATION OF FACTORS AFFECTING THE FINANCING OF COLLEGE EDUCATION: AN INTERCOUNTRY COMPARISON

John F. Yanagida, University of Hawaii at Manoa, U.S.A.

Mahendra Reddy, University of Hawaii at Manoa, U.S.A.

ABSTRACT -

The role of human capital in economic development is well recognized and researched. What is not that well researched is the sources of funds for higher education, in particular, cross cultural comparisons. This study attempts to fill this void by synthesizing the work done separately at the University of Hawaii, Akita University and the University of the Ryukyus. The framework for analysis is the household production function approach to consumer behavior. Results from this study indicate that student age and time allocated to work are positively related to students’ financing their college education themselves in both countries. Other variables which are not significant also point out to some key policy implications.

INTRODUCTION

In 1965, the U.S. Congress passed the Federal Family Education Loan Program (FFELP) which made available federally supported loans for college students. Along with other forms of financial support, e.g., government scholarships, fellowships, Pell Grants etc., government support for higher education has increased substantially since then. Subsequent studies have shown that student financing of higher education in the U.S. has undergone major changes as tuition and other costs have increased (Leslie, 1984). According to the U.S. National Longitudinal Study, 80% of student financing comes from private sources and the remaining 20% from public sources. However, in recent years, the U.S. government has faced huge federal deficits and most states have encountered growing demands for their tax dollar. The econmic situation has caused federal and state governments to shift more of the burden of financing higher education back to the students (Keynes, 1995). Unfortunately, this shift in government support of higher education has occurred at a time when college and university costs have also escalated. Few, if any, cross-cultural studies have been conducted which compare sources of funds for higher education by country, race or nationality. This study attempts to fill this void by using data from a test study to compare student financing of higher education from selective surveys administered in Japan and Hawaii.

An important source of human capital formation is education. As described by Schultz (1971), education is acquired capital and an investment embodied in human beings. It is well recognized that families help create human capital and the acquisition of human capital is affected by the interactions between schooling and families.

Higher education is mostly about investment. Psacharopoulos (1985) has shown that rates of return to higher education differ among countries. His findings show that rates of return to schooling among countries decline as the country’s economic development advances. So for economically advanced countries, rates of return to schooling (including higher education) are lower than those for less advanced countries.

This paper synthesizes the work done separately at the University of Hawaii, Akita University and the University of the Ryukyus. College students enrolled in Home Economics and Human Resources courses completed surveys to compare spending patterns and financing of college education. The methodological framework for analysis is the household production function approach to consumer behavior. Of interest are the effects from socio-economic factors such as gender, age location (Hawaii, Akita or the Ryukyus), marital status, living arrangements and preferences for money.

The next section discusses the household production function approach to consumer behavior. The third and fourth sections describe the data and the econometric technique used for the analysis respectively. Empirical results are presented and discussed in the fifth section. Concluding comments and proposed extensions of this study are summarized in the last section.

PROPOSED METHODOLOGICAL FRAMEWORK: HOUSEHOLD PRODUCTION FUNCTION

In the household production function approach, households are viewed as firms which produce desired products called commodities using household resources (such as member’s time) and purchased market goods and services. The bundle of commodities produced, such as "obtaining a college education", is achieved by combining time spent by family members with goods and services purchased. The commodities produced within the household are determined by family preferences given income and time constraints.

The household production function approach by Becker (1965) and Michael and Becker (1973) provide a comprehensive framework for analyzing decisions involved with the allocation of time and market goods and services. Consumers or households have the following optimization problem:

MAX U = U(Z₁, Z₂, .....,.....,.....Z_m, R) (1)

subject to: Z_i = f_i(x₁, x₂,.....,....,....x_n,t_i)

                t = t₁ + t₂ + .....+......+ t_m

where

Z_i=commodity i, e.g., "obtaining a college education", i=1,2,........m,

R =demographic variables for the household or Individual,

x_j= market goods and services, j=1,2,.........n,

t_{i =} time spent in the production of the ith commodity.

Differentiating equation (1) and solving simultaneously for Z_i, the derived demands for commodities such as "obtaining a college education" can be derived. The functional form for Z_i can be written as:

Z_i = g_i(j_i, x_j, t_i, R) (2)

where j_{i =} price or cost of Z_i

The inverse demand function has the form:

j_i = h_i(Z_i, x_j, t_i, R) (3)

DATA

Data have been collected from three sites (University of Hawaii at Manoa, Akita University and the University of the Ryukyus (Okinawa)). The University of Hawaii sample included students in a Family Financial Planning course (N=27, spring, 1996 and N=25, spring, 1997) and a Human Development university core course (N=22, Summer Session I, 1996). The Akita University students surveyed (N=98, July, 1996) were enrolled in either a Human Development, Fashion or Home Economics Education course while students at the University of the Ryukyus ( N=74, July, 1996) were in Elementary and Secondary Home Economics Education courses.

The intent of the student survey was to collect data on student spending patterns and preferences for money/savings. Student financing of their college education is one aspect that could be analyzed from these data. In particular, the proportion of the cost of college education borne by the student and the family is of particular interest in this study. Ths proportion is assumed to be a function of the student’s age, gender, marital status, living arrangements, preference for spending/saving, and time spent working.

The survey instrument consisted of Likert scaled and fill-in-the-blank questions. There were four sections in the survey: (i) money attitude scale, (ii) consumption and time use preference, (iii) funding sources, and (iv) financial and socio-demographic background. The survey was back-translated for comparability between the English and Japanese versions.

EMPIRICAL METHODOLOGY

Equation (3) proposes that the price or cost of a college education can be determined by market, household and individual variables and characteristics. However, because of limited data available from the survey, the funding of students’ college education is analyzed by the proportion financed by the family and by the student. This type of analysis where the dependent variable or choice variable can take on only limited values is called a binary choice model or qualitative response model (see Pindyck and Rubinfeld, 1981 and Greene, 1993).

An example of a binary dependent model can be specified as follows:

Y_i = a + bX_i + e_i (4)

where

Y_i = 1 if first option is chosen

0 if second option is chosen

X_i = value of attribute (e.g., market goods and services, time spent, and demographic variables)

e_i = random error term.

Application of ordinary least squares techniques to estimate the above model will result in inefficient estimates since the error term is heteroscedastic. Given this problem, a commonly used approach in the econometrics literature is to transform the original model using a cumulative probability function in such a way that the predictions (P) will lie in the (0,1) interval for all X. This study utilizes this concept and adopts the Probit probability model (which utlizes the cumulative normal probability function) for estimation. The probit model can be shown as follows:

03193e05.gif">EQUATION  (5)

where

P_i = probability that the event occurs

e = base of natural logarithm

s_i = random variable with mean zero and unit variance.

The empirical model for this study can be written as follows [Marital status was excluded as an explanatory variable because all students surveyed at the two universities in Japan were not married.]:

MCF_i = b₀+b₁SA_i+b₂GD_i+b₃LA_i+b₄SMA_i+b₅TAW_i+m_i (6)

where

MCF_i = method of college financing by ith student

= 1 if funded by parents (all or largest share)

= 0 if funded by student (all or largest share)

SA_i = ith student’s age

= 1 if<19 years of age

= 2 if 19 years of age

= 3 if 20 years of age

= 4 if 21 years of age

= 5 if 22 years of age

= 6 if 23-29 years of age

= 7 if 30-39 years of age

= 8 if 40-49 years of age

= 9 if>49 years of age

GD_i = gender (1 =male and 0=female)

LA_i = living arrangement (1=living with parents and 0=living away from parents)

SMA_i = student money attitude or spending preference EQUATION [This student preference variable was constructed using information from Masuo and Reddy (1997). They separated the money attitude index into 3 components: cognitive-power prestige, time preference, and attitudes toward credit. The latter component, attitudes toward credit, was deemed inappropriate for this analysis. The time preference component, while consistent with human capital as an investment, was not utilized because only one of the survey questions (selected from factor analysis) had responses from both students in Hawaii and Japan. The power prestige responses were from three survey questions and were added together to derive this study's SMA index. The three survey questions used were (i) I prefer to be less respected and rich rather than respected and poor, (ii) Money is a crucial factor for one's happiness, and (iii) If I have a lot of money, I can easily get power and respect.]

TAW_i = estimated time allocated for work in hours/month [The variable TAW was calculated as monthly earnings by students divided by the average hourly wage rate. For Hawaii, the average hourly wage rate was $8.09/hour. This was calculated as a weighted average of selected occupation from the Hawaii Employers Council (1997) and student part-time employment at the University of Hawaii. For Japan, the average hourly wage rate was $6.16/hour (ministry of Labour, (1996).].

A priori, the expected signs of the above variables are stated as follows. The variables age and hours spent on work are expected to have negative signs. As students grow older, they become less dependent on parental income for education. With respect to time allocated to work, students generally tend to work if there is a need. Therefore, if alternative methods of financing are not available, then students will engage in paid employment to finance their studies. Hence, increased time allocated to work will indicate increased probability that students are financing their own study. The expected coefficient sign for variable LA or living arangement is positive. Students living away from parents are generally een to display greater independence and thus less reliant on parents to finance their studies.

TABLE 1

DIFFERENCES BETWEEN SAMPLE MEANS

The expected sign for variable SMA or student money attitude is difficult to assign a priori because people from different cultural backgrounds may value money differently. A similar analogy can be made for gender (GD). Different cultural groups place different values on females and males and thus influence the expected sign for variable GD. Rather than assign a priori signs for these variables, this study explains the estimated coefficients and their signs in context of the model and understanding of cultural values at each site.

ESTIMATION RESULTS AND DISCUSSION

Appendix 1 summarizes the sample data from Hawaii and Japan. To compare survey responses from Hawaii and Japan, a difference between means t-test was performed (see Table 1). Results indicate that mean responses for MCF, SA and TAW were significantly different (at the 5% significance level) between Japan and Hawaii. In the case of Japan, the method of college financing tends to lean significantly toward parental support. In the case of student age (SA), the Hawaii students surveyed are significantly older than their Japanese counterparts. Students from Hawaii also work significantly more hours per month than students from Japan. This latter result is consistent with MCF results that Hawaii students tend to rely less on parental funding of their college education. As Hawaii students work longer hours, they extend the time it takes to complete their undergraduate programs. This may help explain the significant age difference between students from Hawaii and Japan.

A statistical test was done to determine whether the three sites (two from Japan and one from Hawaii) could be pooled together and a pooled model could be compared against individual models for Hawaii and Japan. The Chow test results indicate that a pooled model was appropriate. [Assume the following two models for Japan and Hawaii:

EQUATION

Data from the two countries were tested to see whether a pooled model (one which combines the two data sets into one) was more appropriate as opposed to estimating the two equations separately. A Chow test was used to establish this result.

The null hypothesis is that (jointly):

b₀=a₀, b₁=a₁, b₂=a₂, b₃=a₃, b₄=a₄, b₅=a₅

The Chow test follows a F distribution which can be stated as follows:

EQUATION

ESS_R = the error sum of square of the model with Hawaii and Japan's data combined.

ESS_UR = ESS_JAPAN + ESS_HAWAII

Both the restricted and unrestricted model were estimated and the F statistic was computed as shown below:

EQUATION

F_5,211,0.05 = 2.21

Since the value of the F statistic is less than the critical value of the F distribution at the 5 percent level, we do not reject he null hypothesis that the pooled model is more appropriate.]

Results from the probit model estimation are presented in Table 2. These maximum likelihood estimates were obtained by using the Shazam Econometrics computer package (White, 1993). The results show that for most cases, variable signs conform to a priori expectations. In cases where the Japan and Hawaii sites differ in response via the sign of the coefficient, this may convey information specific to differences in socio-economic backgrounds of each site.

TABLE 2

ESTIMATED COEFFICIENTS AND SLOPE

The use of an R² statistic to measure goodness of fit of the regression model is not valid in the case of binary dependent variable models. Instead, an alternative measure that is widely used, the likelihood ratio index (McFadden adjusted R²) is estimated. The likelihood ratio indices estimated in this study indicate a goodness of fit of 24.3% for Hawaii and 18.1% for Japan. Likelihood ratio index values within this range are reasonable for studies which utilize cross-sectional data and where the dependent variable is a binary variable.

Two explanatory variables, TAW and SA, are statistically significant at the 5% level for all three cases (models). The slope of the TAW variable implies that with an increase in time allocated to work by one unit, the probability of parents financing students’ education will decrease by 0.005 for Hawaii and 0.002 for Japan. With an increase in time spent at work, the additional income is used to finance college expenses. The slope coefficient of the age variable indicates that with an increase in age by one unit, the likelihood of parents financing their education decreases by 0.186 in Hawaii and 0.083 in Japan. This implies that as students get older, they tend to be less dependent on their parents.

The variables, gender (GD) and living arrangements (LA), have opposite signs for the individual models. The gender variable’s slope coefficient is 0.086 for Japan and -0.224 for Hawaii which indicates that the method of college financing differs with respect to gender for these wo cases. However, this variable is statistically insignificant in both models. A similar analogy applies to the student’s living arrangements variable, which is also insignificant in both models. For the estimated LA coefficient, while the positive coefficient makes sense in Hawaii’s case, it is difficult to interpret the negative coefficient obtained for Japan. A positive coefficient implies that students living away from parents tend to be more independent and have an increased likelihood of funding their own college education. In the case of Japan, the negative (but statistically insignificant) LA coefficient could imply that parents are substituting financial support for college education for other forms of support such as housing, meals, etc. So, as students tend to live away from home, this increases the probability of college funding by parents.

The interpretation of the SMA variable pertains to how students view money and how this view affects the funding of their college education. A positive coefficient implies that as preferences for money increar4 i.e., money is viewed as the root of power and prestige, the probability that parents fund a larger portion of students’ college education increases. Interestingly, this variable’s coefficient is positive for both Hawaii and Japan but statistically significant only for Hawaii. Financial support for college by parents, while a norm in Japan, is less prevalent in Hawaii (see sample description in Appendix 1). For Hawaii, as preferences for money increases, students seek alternative funding sources including support from parents.

SUMMARY AND CONCLUSIONS

As the cost of higher education increases over time, financing college education becomes a greater concern for students as well as their families. This study utilizes the household production framework to examine how college financing is affected by socio-economic factors, time spent working, and spending preferences. Student survey data from universities in Hawaii and Japan were collected. Comparing average responses between Hawaii and Japan, significant differences were found for variables MCF, SA and TAW.

Using the household production function approach to consumer behavior, constrained utility maximization allows derivation of reduced form equations such as the cost of obtaining a college education. Examining the proportion of costs financed by parents and the student requires estimation of a binary choice model. Probit analysis is used and the maximum likelihood estimates, for most cases, conform to a priori expectations. Statistically significant variables which help explain the choice of funding sources (i.e., parents vs. students) include student age and time allocated for work. Also, interesting results from this study are the variables which were not statistically significant in explaining funding sources, namely gender, living arrangements and student money attitudes (for Japan only).

Results indicate that Hawaii students tend to spend more time working and are generally older than students from the two Japanese universities. These results help explain the finding that Hawaii students are less dependent on family financial support for their college education as compared to Japanese students.

APPENDIX 1

SAMPLE DESCRIPTION

As stated earlier, with rising costs for higher education and governments being faced with larger budget deficits, the burden of financing higher education has shifted more to the student. In periods of fiscal crisis, higher education often becomes an easy target for meeting government budget shortfalls. Several reasons contribute to this situation. First, investments in higher education do not produce immediate returns, rather, returns are often realized in the longer term. Second, there are alternatives to government funding of higher education, e.g., private sector scholarships and loans, funding from parents, and student employment while attending collge. For this initial study, students at the University of Hawaii work longer hours and rely less on their parents for funding support than students from two universities in Japan. By working longer hours to finance rising educational costs, this funding option increases the investment time by students, i.e., lengthens the time needed to complete their degree program. In a social welfare context, the returns to higher education take longer to be realized. In Japan, where funding from parents plays a much larger role, students tend to complete their degree programs faster and their investment in higher education is realized sooner. In terms of economic growth for this latter case, ceteris paribus, returns from investment in higher education can positively impact society faster.

This study serves as a useful beginning for future research examining the allocation of time by university students and their families to various household activities such as meal preparation, investment in human capital (obtaining a college degree), recreational activities, etc. The information gathered from an expanded cross cultural study can address contemporary issues such as matching trade and product preferences by country (or location) and the substitution of eating out versus home-made meals.

REFERENCES

Becker, Gary S. 1965. "A Theory of the Allocation of Time", The Economic Journal, 75: 493-517.

Greene, William H. 1993. Econometric Analysis, 2nd Edition, New York: Macmillan Publishing Company.

Hawaii Employers Council. 1997. Pay Rates in Hawaii, Special Publication Number 246.

Hays, William L. and Robert L. Winkler. 1970. Statistics Volume 1, New York: Holt, Rinehart and Winston, Inc.

Keynes, John M. 1995. "Are Students Borrowing Too Much?" Planning for Higher Education, 23: 35-42.

Leslie, Larry L. 1984. "Changing Patterns in Student Financing of Higher Education", Journal of Higher Education, 55(3): 313-346.

Masuo, Diane M. and Mahendra Reddy. 1997. "Comparison of Students’ Money Attitudes: A Cross-Cultural Sampling of Selected U.S. and Japan Universities", Paper presented at the Association of Consumer Research European Conference, Stockholm, Sweden.

Michael, R. T. and G. S. Becker. 1973. "On the New Theory of Consumer Behavior", Swedish Journal of Economics, 75: 378-396.

Ministry of Labour. 1996. "1995 Basic Survey on Wage Structure, Policy Planning and Research", Minister’s Secretariat, Japan.

Pindyck, Robert S. and Daniel L. Rubinfeld. 1981. Econometric Models and Economic Forecasts, 2nd Edition, New York: McGraw-Hill Book Company.

Psacharopoulos G. 1985. "Returns to Education: A Further International Update", Journal of Human Resources, 20(4): 583-604.

Schultz, Theodore S. 1971. Investment in Human Capital. New York: Free Press.

White, Kenneth J. 1993. SHAZAM: Econometric Computer Program

----------------------------------------