ECONOMIC ANALYSIS OF DEMAND FOR DISTANCE EDUCATION IN CANADA BY Edward Hans Kofi Acquah, PhD. Senior Institutional Analyst & Academic Expert ATHABASCA UNIVERSITY CANADA AIR 2009 ANNUAL FORUM MAY 30-JUNE 03, 2009 ATLANTA, GA. USA. INTRODUCTION AND BACKGROUND Distance Education • Has come of age • Become a significant part of postsecondary education Definitions: 1. “a formal educational process in which the students and the instructor are not in the same place” (Prasad & Lewis, 2008) Definition implies that instruction may be: • Synchronous: real time or simultaneous • Asynchronous: not real time or simultaneous And may involve: INTRODUCTION AND BACKGROUND (continued) • Communications: through the use of video, audio or by computer and internet technologies, or • Communications: by written correspondence and use of technologies, e.g. CD-ROM 2. “An educational practice promoting a learning process that brings knowledge closer to the learner” (Deschenes & et all, 1996) 3. Distance Education Courses and Programs Are classified as ff: • Online Courses/Programs: all instruction is online; • Hybrid/Blended Online Courses/Programs: combines online and in-class instructions with a reduced in-class seat time for students; • Other Distance Education Courses/Programs: postal correspondence RESEARCH OBJECTIVES • To analyze the demand for distance education in Canada • To determine which factors influence this demand • To determine gender differences in terms of the factors that influence the demand • To determine the policy implications for Canadian universities and colleges offering distance/online education HISTORICAL PERSPECTIVES • DE has experienced growth and expansion in North America in recent years in terms of: • DEMAND: program enrolments and course registrations; • SUPPLY: Institutions, learning management systems (LMS), delivery modes, faculty, and innovative learning resources; • In the fall of 2006, 3.5 million students (19.8% of PSE enrolments) in the US took at least one course online; • The Table below gives a better perspective on growth in DE (Babson Survey Research Group & Sloan Foundation, 2008): GROWTH IN THE DEMAND FOR DE IN THE US ENROLMENT DEMAND Period Fall 2002 Fall 2006 2006 Increase # # # Doctoral/Research 258,489 566,725 308,236 Compound Annual Growth Rate% 21.7 Master’s Baccalaureate Community Colleges Specialized Programs 335,703 130,677 806,391 686,337 170,754 1,904,206 350,634 40,077 1,097,905 19.6 6.9 24.0 71,710 160,268 88,558 22.3 PROGRAMS HISTORICAL PERSPECTIVES (CONTD) • Parsad & Lewis (2008) have shown that 66% of 1,600 Title IV Degreegranting PSE institutions offered Online, Hybrid/Blended Online or other Distance Education courses in the 2006-07 academic year. These are the pioneers of DE in Canada: • • • • • • • The Queen’s University, 1889 University of Saskatchewan, 1907 Xavier University, 1935 The University of British Columbia, 1950 Memorial University of Newfoundland, 1967 University of Waterloo, 1968 Ryerson Polytechnic University, 1970 PIONEERS OF DE IN CANADA • • • • • Simon Fraser University, 1975 University of Victoria, 1979 British Columbia Institute of Technology, 1985 McGill University, 1987 Salt College of Applied Arts & Technology, 1988 • Athabasca University, Canada’s Open University (1973) & • Tele University & Open Learning Institute (1975) were fashioned on the British Open University (1971) model • In 1994, there were 200,000 college and university enrolments in DE in Canada (Canadian Studies Directorate, 1994) • DE enrolments at Athabasca University increased from 10,874 (1994/95) to 12,853 (1997/98), 18.2% or 5.7% per year PIONEERS OF DE IN CANADA • Course registrations increased from 20,641 (1994/95) to 25,312 (1997/98), 22.6% or 7.0% per year -(Athabasca University, 1997/98 Calendar) Other trends in DE in Canada: • Drop in the average age of distance learners • Increase in registrations and course loads • Increase in the number of female students Other DE Settings: • In-House training for employees and professional associations, e.g. Institute of Canadian Bankers, Certified General Accountants of Canada; • Alberta Distance Education Training Association (ADETA) FACTORS UNDERLYING DE GROWTH • Economic growth • Rising Incomes • Increasing public expenditures on PSE • Population Growth • Geographic separation of linguistic minority groups • Continuing education needs of populations living far from urban centres • Flexibility inherent in DE, e.g. any time anywhere • Computer and Internet innovations IMPORTANCE OF DE • Empirical research has shown that academic achievements of DE learners are comparable to that of on-campus taught face-to-face • Higher enrolments in DE means economic effectiveness of resource use since DE institutions don’t need additional expenditures like new classrooms in order to expand Value-added by DE: • Increasing student access • Serving rural communities • Expanding student educational choices • Ability of DE to transcend geographical boundaries IMPORTANCE OF DE • These developments make it all the more imperative to devote time and resources toward research to learn more about the increasing enrolment demand, institutional and general factors fuelling this growth, and the individual characteristics of the students who are being served. DETERMINANTS OF DE DEMAND Past research indicates that demand for post-secondary education is influenced by a complex set of factors, including: • Expected stream of future benefits (Shultz, 1961; Becker, 1964; Bishop, 1977; Campbell and Siegel, 1971; Fiorito and Dauffenbach, 1982; Freeman, 1986; Leslie and Brinkman, 1988; Willis and Rosen, 1979); • Family income as part of student’s investment capital (Bishop, 1977; Gorman, 1983; Galper and Dunn, 1969; Schwartz, 1986; Spies, 1978) • Price (Tuition & Fees) (Funk, 1972; Heller, 1997; Campbell and Siegel (1967; Radner and Miller, 1975; Funk, 1984; Ehrenberg, Sherman; and Schwartz, 1986; Leslie and Brinkman, 1987; Jackson and Weathersby, 1975). • Employment expectations and • Family background characteristics (Albert, 2008). • . DTERMINANTS OF DE DEMAND However the following determinants have not been explored: • Number of Programs • Number of Distance & Online Courses, • Marketing Expenditures on advertising and recruitment activities, • The Canadian University Participation Rates (UPR) ANALYTICAL FRAMEWORK THE MODEL: • A formal statement of the general model is given as: • Qdt = f1 (Pt, UPRt, GDPt, MktExpt, #DistCrst, Unempt) • Where: • Qdt is the demand for distance and online education in year t • Pt is the real tuition & fees in year t (money tuition deflated by the Consumer Price Index, CPI) • UPRt is the proportion of the 18-24 year old in post-secondary education in year t in Canada • GDPt per capita, here represents average household income as well as an indicator of how well the Canadian economy is doing in year t. • MktExpt is the average expenditure on marketing and recruitment activities in year t • DistCrst is the number of distance and online courses available in year t. • Unempt is the unemployment rate in year t in Canada. THE REGRESSION FUNCTION To estimate the general model, the following multiple regression version was used: • Qt = b0 + b1x1t + b2x2t + b3x3t + b4x4t + b5x5t + b6x6t + et Where: • Qt = response or dependent variable, i. e. enrolments/registrations in fiscal year t (= 1975/76, 1976/77 …….2007/08) • b0 = intercept of the regression model, which is the mean value of the response variable when all the predictor variables are zero • x1t = represents tuition & fees or the price per a 3-credit course paid by students in a fiscal year t deflated by the Consumer Price Index • x2t = represents the Canada University Participation Rate in fiscal year t • x3t = represents the effect of the Gross Domestic Product, GDP, that is the state of the economy, on enrolments/registrations in fiscal year t. The GDP may also stand for the role of income in demand for education THE REGRESSION FUNCTION • x4t = represents marketing expenditures on advertising and other recruitment activities in fiscal year t • x5t = represents number of distance education courses available in fiscal year t • x6t = represents the Canadian unemployment rate in fiscal year t • et = the stochastic error term in fiscal year t, that is, the effect of potential variables not included here in the model under consideration • b1, b2, b3, b4, b5, b6 are the coefficients or parameters of the explanatory or predictor variables to be estimated MODEL ASSUMPTIONS • The model is based on the following classical linear regression assumptions: • E (et) = 0 for all t, that is the expected value of the errors is zero for all possible sets of given values of x1, x2, x3, x4, x5 and x6., that is: E |ei| = 0 for i = 1, 2, 3, 4, 5, 6. • The error term e is independent of each of the m independent variables x1, x2, x3, x4, x5 & x6 i.e. E (xktet) = 0 for all k = 1, 2, 3, 4, ..m • The errors, e, for all possible sets of given values of x1, x2, x3, x4, x5 & x6 are normally distributed. • Any two errors ek and ej are independent. Their covariance is zero: E (ekej) = 0 for k ≠ j • The variance of the errors is finite, and is the same for all given values of x1, x2 ...xm. That is V |ei| = s2 is a constant for I = 1, 2, n HYPOTHESES The regression model was estimated using STATA statistical software to test the following hypotheses : • That the price (tuition) effect upon demand is negative (b1<0) • That the UPR effect upon demand is negative (b2<0) • That the income effect upon demand is positive (b3>0) • That the marketing effect upon demand is positive (b4>0) • That the distance courses effect on demand is positive (b5>0) • That the unemployment effect on demand is negative (b6<0) ESTIMATED MODELS AND MODEL DIAGNOSTICS The Macro Perspectives • The following model was estimated at the macro level: • Qt = b0 + b1x1t + b2x2t + b3x3t + et The Micro Perspectives • The following model was estimated at the micro level: • Qt = b0 + b1x1t + b2x2t + b3x3t + b5x5t + et MODEL DIAGNOSTICS The estimated models were diagnosed and tested for the presence of: • Autocorrelation (serial correlation: potential values of the residuals follow a particular pattern): Residual plots & D-W test • Heteroscedasticity (V (et) = s2 for all j): residual plots against the predicted values of the dependent variable & Brausch-Pagan Test • Multicollinearity (if independent variables are dependent upon each other or are collinear): Tests: rx1x2 & VIF Results: • No compelling evidence of a serious presence of any of these data problems were found • No strong evidence of model misspecifications were found ESTIMATED MACRO MODELS Table 1 Estimated Results of the Macro Enrolment Demand Model (All Students) Variables Estimated Coefficients Standard Error Standardiz ed Beta (b) t-value Sign p>|t| Constant (b0) 317,892.5*** 29,703.45 0.000 10.702 0.0000 Tuition & Fees (b1) -3,927.07** 1,262.56 -1.1191 -3.110 0.0077 13.150 Income (GDP) (b2) #Courses(b3) 85.0738** 21.22 0.6903 4.009 0.0013 3.006 109.7501** 32.15 Mean Variance Inflation Factor (V. I. F.) 1.0302 3.414 0.0042 9.245 R2 = 0.862 Adjusted R2 = 0.833 Correlation R = 0.929 F-value= 29.19 p-value= 0.0000 D.W.=1.48 *p<0.05; ** p<0.01; ***p<0.001 VIF 8.467 ESTIMATED RESULTS Table 2 Estimated Results of the Macro Enrolment Demand Model (Male Students) Estimated Coefficients Standard Error Standardize d Beta (b) t-value Sign p>|t| VIF 121,551.44*** 12,186.49 0.000 9.974 0.0000 13.150 -1,466.68** 517.99 -1.2022 -2.831 0.0133 3.006 30.59*** 8.71 0.7133 3.513 0.0034 9.245 #Courses(b3) 46.58** 13.19 Mean Variance Inflation Factor (V. I. F.) 1.2571 3.531 0.0033 13.150 Variables Constant (b0) Tuition & Fees (b1) Income (GDP) (b2) R2 = 0.808 Adjusted R2 = 0.767 Correlation R = 0.899 F-value= 19.66 p-value= 0.0000 D.W.=1.60 *p<0.05; ** p<0.01; ***p<0.001 8.467 ESTIMATED RESULTS Table 3 Estimated Results of the Macro Enrolment Demand Model (Female Students) Estimated Coefficients Standard Error Standardize d Beta (b) t-value Sign p>|t| VIF Tuition & Fees (b1) 196,900.00 -2,491.04 18,060.28 767.66** 0.000 -1.0721 10.902 -3.245 3.17E-08 0.0059 13.150 3.006 Income (GDP) (b2) 54.35 12.90*** 0.6663 4.213 0.0009 9.245 #Courses(b3) 64.00 19.55** 0.9072 3.274 0.0055 13.150 Variables Constant (b0) Mean Variance Inflation Factor (V. I. F.) R2 = 0.884 Adjusted R2 = 0.859 Correlation R = 0.940 F-value= 35.47 p-value= 0.0000 D.W.=1.39 *p<0.05; ** p<0.01; ***p<0.001 8.467 PERFORMANCE OF THE MACRO MODELS • The macro model provides a reasonably very strong fit to the data: R2 =0.86; 0.81 & 0.88 for All, Male & Female students • Adjusted-R2 =0.83; 0.77 & 0.86 for All, Male & Female students • The large F-values (significant far beyond 0.001, that is p<0.001) implies that it is a very strong model • The estimated multiple correlation coefficients R (0.93; 0.89 & 0.94) indicate very strong correlation • Results are consistent with a priori expectations • The estimated coefficients, b’s, possess the necessary signs and are statistically significant. • This indicates that the influence of tuition and fees (price), income and number of courses on enrolment demand are all significant. ESTIMATED MICRO MODELS Table 4 Estimated Results of the Micro Demand Model (All Students) Variables Constant (b0) Estimated Coefficients Standard Error Standardize d Beta (b) t-value Sign p>|t| 808.26 1,308.1489 0.000 0.618 .5418 953.6708 -0.5453 -4.368 .0002 24.178 Tuition & Fees (b1) -4,165.31*** VIF Income (GDP) (b2) 18.77*** 3.6084 0.7041 5.201 1.78E-05 28.457 #Courses(b3) 29.21** 9.9555 0.5762 2.934 .0068 59.845 MktgExp (b4) 0.0127*** 0.0025 0.2634 4.995 3.09E-05 4.321 Mean Variance Inflation Factor (V. I. F.) R2 = 0.983 Adjusted R2 = 0.980 Correlation R = 0.991 F-value= 381.40 p-value= 0.0000 D.W.=0.76 *p<0.05; ** p<0.01; ***p<0.001 29.2 ESTIMATED MICRO MODELS Table 5 Estimated Results of the Micro Demand Model (Male Students) Estimated Coefficients Standard Error Standardize d Beta (b) t-value Sign p>|t| 671.21 445.39 0.000 1.507 .1434 Tuition & Fees (b1) -1,167.66** 324.70 -0.5472 -3.596 .0013 24.18 Income (GDP) (b2) 5.65*** 1.23 0.7581 4.596 .0001 28.46 #Courses(b3) 7.33* 3.39 0.5173 2.162 .0396 59.85 MktgExp (b4) 0.0036*** 0.00 0.2654 4.127 .0003 4.32 Variables Constant (b0) Mean Variance Inflation Factor (V. I. F.) R2 = 0.974 Adjusted R2 = 0.970 Correlation R = 0.987 F-value= 254.87 p-value= 0.0000 D.W.=0.67 *p<0.05; ** p<0.01; ***p<0.001 VIF 29.2 ESTIMATED MICRO MODELS Table 6 Estimated Results of the Micro Demand Model (Female Students) Estimated Coefficients Standard Error Standardize d Beta (b) t-value Sign p>|t| -481.88 -481.88 0.000 -0.722 0.4766 Tuition & Fees (b1) -1,984.30*** -1,984.30 -0.4342 -4.077 0.0000 24.18 Income (GDP) (b2) 14.99*** 14.99 0.9401 8.142 0.0000 28.46 #Courses(b3) 4.89 4.89 0.1614 0.963 0.3443 59.85 MktgExp (b4) 0.0098*** 0.01 0.3383 7.519 0.0000 4.32 Variables Constant (b0) Mean Variance Inflation Factor (V. I. F.) R2 = 0.987 Adjusted R2 = 0.985 Correlation R = 0.994 F-value= 526.64 p-value= 0.0000 D.W.= 1.03 *P<0.05; ** p<0.01; ***p<0.001 VIF 29.2 ANALYSIS OF RESULTS AND POLICY IMPLICATIONS Introduction • The estimated results are consistent with all our hypotheses • The estimated coefficients, b’s, possess the necessary signs and are statistically significant. This indicates that the influences of: • Price (Tuition and Fees) • Income (GDP) • Number of Courses • Marketing Expenditures on advertising and recruitment on enrolment demand are generally all significant. ANALYSIS OF RESULTS AND POLICY IMPLICATIONS PRICE (Tuition & Fees) • The impact of price on demand for DE is negative and statistically significant in both models; • When price rises, demand for DE declines, all other thins remaining constant • Price is the second most important predictor of demand by male students (b=1.202: macro) & second most important predictor of DE (b=0.547: micro), • Price is the second most important predictor of demand by female students (b=1.072: macro) & second most important predictor of DE (b=0.434: micro) • This means that price changes are of greater concern to male students than to female students at Canadian distance institutions in general and the typical distance institution in particular. • Increases in price result in the loss of more male enrolments than female enrolments for DE. • These results confirm the economic hypothesis that demand for education is inversely related to price (Jackson & Weathersby, 1975; Bishop, 1977; Funk, 1972; Corman, 1983). ANALYSIS OF RESULTS AND POLICY IMPLICATIONS INCOME (GDP) • Income has positive effect on demand for DE (Mueller & Rockerbie, 2005) • It is statistically significant for Canada (macro models) and for the typical distance education institution in Canada (micro models). • Income is the third most important predictor of demand by male students (b=0.713: macro) and first most important predictor of demand by male students (b=0.758: micro) • Income is the third most important predictor of demand by female students (b=0.666: macro) but the first most important predictor of demand by female students (b=0.940: micro) • The influence of income on demand for DE is greater for male students than for female students in Canada in general, but greater for female students than male students in the typical institution • This means that increase in income attracts more demand from female students than from male students in the typical institution. ANALYSIS OF RESULTS AND POLICY IMPLICATIONS MARKEING EXPENDITURE • The marketing expenditure variable has a positive effect on demand for distance education and is statistically significant • This means that the more we spend on advertising and other recruitment activities, the more students will enrol at a typical distance education university • Thus a $1.00 increase in marketing expenditures will lead to 0.0127 new enrolments; $10,000 increase will lead to 127 new enrolments; and a $100,000 increase will lead to 1,270 new enrolments. • Marketing and recruitment expenditures are more important to female students (b=0.338) than to male students (b=0.265) as exemplified by the estimated standardized beta coefficients • This means that any dollar amount spent on marketing and recruitment attracts more female enrolments than male enrolments. ANALYSIS OF RESULTS AND POLICY IMPLICATIONS NUMBER OF COURSES • The estimated course coefficients for both macro and micro models are consistent with the a priori expectations and are statistically significant • The availability of distance and online courses appear more important to male students than female students • The availability of courses is the first most important predictor of demand by male students (b=1.257: macro) but the third most important predictor of demand by male students (0.517) in the micro model • Availability of courses is the second most important predictor of demand by female students (b=0.907) in the macro model, but the fourth most important predictor of demand by female students (b=0.161) in the micro model • This means that increase in the availability of distance and online courses attract greater demand for distance education from male students than from female students SUMMARY AND CONCLUSIONS • The overall results are illuminating and offer some interesting implications for enrolment demand for distance education in Canada. • The impact of tuition and fees (price) on demand for distance education is negative and statistically significant, confirming the economic hypothesis that demand is inversely related to price. • Price changes are of greater concern to male students than to female students at Canadian distance institutions, implying that increases in the price of distance education would result in more male enrolment losses than female losses • The impact of income on the demand for distance education is greater for male students than for female students in Canada in general. • However, in a typical distance education institution, the impact of income on demand for distance education is greater for female students than for male students • This suggests that increase in income would attract more demand from female students than from male students. SUMMARY AND CONCLUSIONS • At the typical distance university, marketing and recruitment expenditures are more important to female students than to male students • This means that any dollar amount spent on marketing and recruitment would attract more female enrolments than male enrolments • Availability of distance and online courses appear to be more important to male students than to female students. • Increased availability of courses would attract greater male demand for distance education than female demand • The overall importance of the study is its ability to provide a theoretical and empirical framework for the analysis of demand for distance education at both national and institutional levels. THE END THANK YOU