Government student loan default: differences between graduates of the liberal arts and applied fields in Canadian colleges and universities.
Wright, Laura ; Walters, David ; Zarifa, David 等
INTRODUCTION
DESPITE GREATER PROPORTIONS of youth pursuing postsecondary
education in the last few decades, student debt has reached
unprecedented levels. In 1982, individuals with student loans who
graduated from a community college typically borrowed $4,000. (1) In
1995, this figure climbed to nearly $10,000 (Finnie 2001a), and $12,600
in 2000 (Allen and Vaillancourt 2004). Similarly, graduates of
undergraduate programs owed on average $6,000 in 1982 and nearly $14,000
in 1995 (Finnie 2001a). By 2000, borrowing students typically owed
$19,500 in government student loans, where approximately 14 percent of
university and 5 percent of college borrowers had government student
debts exceeding $25,000 (Allen and Vaillancourt 2004).
Rising student debt levels have been attributed to two key factors.
First, reductions in government funding of postsecondary institutions
have resulted in higher tuition fees. Between 1989 and 1997, for
example, the average tuition increase exceeded 90 percent (see Michael
and Kretovics 2005; Schwartz and Finnie 2002). Second, beginning in the
1990s, federal and provincial government expenditures on postsecondary
education have been redirected toward providing students with loans, as
opposed to student or institutional grants (Schwartz and Finnie 2002).
Hence, the financial burden of postsecondary education has shifted from
the government to students. Increases in the cost of schooling and
levels of borrowing have also coincided with rising enrollment levels
and tuition fees since these data were collected in Canada (Statistics
Canada 2009). Likewise, similar trends have also been identified in the
United States (Arum and Roksa 2011; Brint and Rotondi 2008; NCES 2009).
The Canada Student Loans Program (CSLP) is the primary form of
government student assistance in Canada and is part of the Government of
Canada's Human Capital Agenda (Clift, Hawkey, and Vaughn 1998;
HRSDC 2009). The CSLP is intended to enable low socioeconomic
individuals to defer the costs of their postsecondary education by
borrowing against their future earnings (Sweet and Anisef 2005).
Students apply for both federal and provincial loans using a single
application by which the students' financial needs are assessed.
During the repayment term, the provincial and federal loans are
consolidated and students repay both portions of their loan through the
National Student Loans Service Centre (NSLSC 2009). CSLP provides over
$1.9 billion in loans to over 330,000 qualifying students annually
(HRSDC 2004).
The repayment of student debt in the form of government loans is an
important social issue that requires special attention. During the 2000
to 2001 academic year alone, over 400 million tax dollars were spent
absorbing defaulted loans (HRSDC 2007b). (2) Moreover, important changes
have been made to the CSLP in the 2005 to 2006 academic year that have
potential implications for the future. First, the federal loan limit has
been increased from $165 to $210 per week of study in order to keep pace
with rising tuition fees (HRSDC 2007a). As in the past when limits
increased, student debt levels will likely rise for those graduating in
2010, as students generally borrow the maximum allowable (Finnie and
Schwartz 1996). Second, the eligibility requirements have become more
lax as a result of the smaller amount of money parents are expected to
contribute to their child's education (HRSDC 2007a). Since students
will likely borrow even more money for their education in the future,
student loan default rates could potentially increase.
The purpose of this paper is to identify which postsecondary
graduates are most likely to default on their government student loans
within two years after graduation. This study provides a timely analysis
of a topic that is of central concern to students, guidance counselors,
academics, and policymakers. It will be especially informative for new
cohorts of students as they make financial decisions about their
postsecondary education payment options that play an important role in
their financial well being over the course of their careers. Loan
defaults are a vital concern to students because defaulting on a
government student loan affects one's credit rating, which, in
turn, has negative and lasting implications for graduates who would like
to obtain a mortgage or credit for other consumer goods.
This study draws on Statistics Canada's 2005 National
Graduates Survey (NGS) to examine student loan defaults among recent
college and university graduates. The only research available utilizing
data relating to student loan defaults from the most recent NGS cohort
is based on descriptive tables and profile reports of the 2005 cohort
provided by Statistics Canada (Bayard and Greenlee 2009). Thus, the
latest NGS survey represents an extremely valuable, yet untapped
resource for examining debt repayment among recent postsecondary
graduates. Further, the analyses make an important contribution to our
understanding of how student loan default rates vary by gender, level of
schooling, and field of study. (3) We are unaware of any existing
studies that have used the NGS to examine default rates by gender, and
nearly all research utilizing the NGS data includes only reported
repayment difficulties, leaving defaults largely unstudied. (4) This
paper seeks to bridge this knowledge gap, and provide important insights
on default rates. In addition, we believe our findings have implications
for traditional human capital and rational action approaches to
examining debt repayment, and discuss alternative approaches that
suggest social differences in student loan uptake and repayment as well
as a rising "culture of student debt." We begin by reviewing
the relevant literature in this area before turning to our findings and
broader implications.
THEORETICAL FRAMEWORK
In order to examine the factors influencing students'
probability of defaulting on their student loans, we consider four major
perspectives. The first perspective draws on rational action and human
capital theory; the second perspective examines borrowers' ability
to pay; the third perspective frames debt decisions within a culture of
student debt; and the fourth perspective examines the characteristics of
field and level of education. We also highlight the importance of
examining the relationship between gender and loan repayment behaviors.
Rational Action and Human Capital
In the existing literature, the issue of student debt and loan
repayment has largely been interpreted through economic frameworks. For
example, rational action perspectives largely see individuals as
behaving logically and rationally. With its roots in classical economic
theory, individuals are seen to weigh the costs and benefits of their
decisions (see White, Marshall, and Wood 2005). That is, individuals
make decisions in efforts of minimizing the associated costs. This line
of reasoning has characterized human capital theories (see Becker 1993),
where human capital consists of knowledge and skills often acquired
through formal education and deemed necessary in today's
knowledge-based economy (see Lamb and Sutherland 2010; Riddell 2008;
Servage 2009). The acquisition of human capital is viewed as an
investment activity, as individuals make rational choices about pursuing
higher levels of education based on the direct and indirect costs of
schooling, as well as the anticipated benefits (see Fever, Rees, and
Gorard 1999; Lamb and Sutherland 2010). Human capital theories would
suggest that despite the increased costs of postsecondary education,
students continue to invest in their human capital in order to compete
in the new knowledge-based economy (Schwartz and Finnie 2002; Servage
2009). By extension, student debt can be viewed as an investment in
human capital that will yield dividends down the road.
Ability to Pay
Another perspective that has been applied directly to
borrower's repayment and defaulting decisions is called the ability
to pay theory (Cabrera, Nora, and Castaneda 1992; Cabrera, Stampen, and
Hansen 1990; Volkwein and Szelest 1995). Ability to pay theory assumes
that students' income levels as well as the incomes of their
families have substantial influences on repayment and default behaviors.
A major premise of this line of research is that graduates who find
themselves in danger of defaulting on their loans may turn to their
parents for financial assistance. This rationalization supports efforts
to provide the greatest amount of funding for students who have the
lowest incomes (see Flint 1997). The theory assumes that the first
priority for students is to cover their essential subsistence
experiences (e.g., rent, food, clothing, taxes), and costs related to
leisure, education, and savings are secondary. Individuals who find
themselves in financial difficulty may turn to family or friends for
financial assistance. According to Flint (1997), one major limitation of
this theory is its inability to explain why borrowers, who seemingly can
afford to pay their loans, refrain from doing so. For example, some U.S.
studies reveal mixed evidence of recent graduates' incomes on their
likelihoods of defaulting. Some studies indicate incomes are tightly
related to defaults (e.g., Ryan 1993; Volkwein and Szelest 1995), while
others do not (e.g., Flint 1997; Hesseldenz and Stockham 1982; Spencer
1974). Similarly, total family income has also shown mixed results in
the United States where some studies find that those from lower
socioeconomic backgrounds (precollege income) are more likely to default
on their loans, while others find no effects (for a discussion, see
Flint 1997). It is unclear, however, what these relationships may look
like in Canada.
A Culture of Student Debt: Investing in the "Postsecondary
Experience"
The mixed effects of personal and family income on debt repayment
actions are difficult to explain. One recent theoretical framework for
understanding student debt may contribute to our understanding of the
"irrationalities" of student debt repayment. Brint and Rotondi
(2008) test the relevance of traditional economic approaches to explain
student debt in the United States. The authors acknowledge that much of
the increased reliance on student loans and debt to finance
postsecondary education can be attributed to economic factors. First,
the cost of obtaining a postsecondary degree and diploma has increased
dramatically. Second, "stagnating family incomes" and
increased pressures to acquire greater levels of human capital over the
same period have created a substantial increase in student debt. Yet
Brint and Rotondi (2008) suggest that economic theories fall short in
predicting the behavioral consequences of increased debt. Based on
interviews at a Southern California public university, Brint and Rotondi
(2008) suggest an alternative framework--the culture of student debt.
Contrary to rational choice approaches, Brint and Rotondi (2008) argue
that the recent debt behaviors of students have been greatly influenced
by institutional behaviors (e.g., loan financing has become more
convenient and easily accessible) and a rising middle-class phenomenon
to partake in the "full college experience." Contrary to
seeing loans as primarily investment instruments, Brint and Rotondi
(2008) suggest that loans may also provide a means for consumption. That
is, middle-class students may be using their student loans to engage in
the "full college experience." Student loans may provide a
means for students to reduce the stress in having to worry about money
not only to finance their schooling but also to spend time with friends,
engage in campus activities, and enjoy the postsecondary life.
Similarly, Arum and Roska (2011) found some support for students
engaging in a particular lifestyle during their college years. When
students were asked why they were working concurrent with pursuing
postsecondary education, about 37 percent of students reported to pay
for tuition costs and an additional 6 percent reported that they
intended to send the money back home (Arum and Roska 2011:86). Yet the
authors note that this leaves nearly half of all students who were
working with reasons outside the "investment" scope.
Level of Schooling and Field of Study
Previous research has also pointed to educational characteristics
as being strongly related to loan repayment. But, to what extent can
education type and field of study predict default behaviors beyond the
individual characteristics of the borrowers? In terms of level of
education, community college graduates typically earn less on average
than bachelor's level university graduates (Allen and Vaillancourt
2004; Bayard and Greenlee 2009; Kapsalis 2006; Walters 2004). However,
community college programs are generally less expensive than are
university undergraduate programs. For example, the average tuition fee
for undergraduate students in 2008 to 2009 was $4,724, while the average
college fee was around $1,900 (see Statistics Canada 2009). Further,
community college programs are generally shorter in length than are
undergraduate-level university programs; a typical community college
program is about 21 months in comparison to 40 months for most
undergraduate-level university programs (Allen and Vaillancourt 2004).
Hence, graduates of community colleges generally have an additional two
years in the labor market to advance through their careers, repay
student loans and potentially contribute to retirement savings. In fact,
it has been argued that differences in earnings between university and
college graduates are partly mitigated by the higher cost of a
university education combined with a bachelor's level
graduate's lost earning potential during his/her additional two
years in school (Drewes 2006).
Although research drawing on labor force surveys has repeatedly
shown that individuals holding a postsecondary credential fare better in
the labor market than those holding only a high school diploma (Allen
and Vaillancourt 2004; Bayard and Greenlee 2009) average graduate
earnings have not kept pace with debt levels. In fact, average real
earnings have remained stable since 1986, while student debt levels have
risen dramatically (Finnie 2001a). Thus, the return on postsecondary
educational investments is a critical concern for those students who
must borrow to finance their education, especially as recent graduates
are experiencing a greater debt burden as a result of decreased
government funding and rising tuition fees. Moreover, it is not
surprising that students from low socioeconomic backgrounds are
especially apprehensive about borrowing for their education (Chapman
1997; Christie and Munro 2003; Clift et al. 1998; Michael and Kretovics
2005; Sweet and Anisef 2005).
Many studies also reveal that recent college graduates generally
have more difficulty repaying their student loans than graduates from
baccalaureate-level university programs. For example, Dubois (2006)
examined the 2000 NGS cohort and found that 41 percent of college
graduates with government student debt reported repayment problems, in
comparison with 31 percent of undergraduate-level university graduates.
However, past studies have not analyzed in detail the relationship
between field of study and repayment difficulties. (5) This is a
significant oversight as field of study is considered to be a critical
marker for stratifying students in the labor market, especially in
Canada (Davies and Guppy 1997; Davies and Hammack 2005; Finnie 2001b;
Walters 2006).
Postsecondary graduates of applied and technical fields have
consistently experienced more favorable labor market outcomes than
graduates of the so-called "generalist" programs (Allen and
Vaillancourt 2004; Finnie 2002; Finnie and Frenette 2003; Kapsalis 2006;
Lavoie and Ross 1999). This finding is generally consistent across
college and bachelor's level university graduates (Walters 2004).
However, while these studies have examined the relationship between
field of study, earnings, and repayment difficulties, the comparisons
did not make distinctions between community college and university-level
graduates. (6) Given that both field of study and level of schooling are
important factors in the school-to-work transitions of postsecondary
graduates, a comprehensive analysis of current patterns in student loan
repayment should incorporate both.
Gender and Student Loan Defaults
The relationship between gender and student loan default is another
issue that must also be considered in student loan and repayment
experiences. First, the proportion of females enrolled in higher
education has increased dramatically since the 1960s, and by 1988 female
participation in postsecondary education was at par with male
participation (see Andres and Adamuti-Trache 2007). By 2005 women made
up the majority of students at both the college and undergraduate
university level, representing 58 and 63 percent of these students,
respectively (Bayard and Greenlee 2009). Second, while there have been
some promising shifts, women are still generally overrepresented in the
lower paying liberal arts fields and are underrepresented in the more
lucrative applied fields which greatly influences their labor market
outcomes (Andres and Adamuti-Trache 2007). Therefore, despite gains in
women's educational attainment, a gendered earnings gap still
remains (Adamuti-Trache et al. 2006; Finnie 2000; Penner 2008).
Moreover, females also typically face greater debt-to-earnings ratios
(7) than males because they tend to borrow the same amount of money for
their education, but generally earn lower wages. Thus, it is not
surprising that women have generally reported greater difficulties
repaying their loans (Dubois 2006; Finnie 2001a).
METHODS
This study employs data from Statistics Canada's 2005 NGS. The
survey contains a series of questions relating to educational history
and the employment profiles of the respondents. It is especially well
suited for this study because it also includes a variety of questions
relating to student borrowing and repayment experiences in Canada. The
survey population is composed of all graduates of Canadian postsecondary
educational institutions who had completed the requirements for degrees,
diplomas, or certificates during the 2005 calendar year. Approximately
40,000 postsecondary graduates of various programs across all provinces
and territories were sampled for the survey.
Sample
The subsample of postsecondary graduates used in this study
includes graduates with community college diplomas or certificates and
university baccalaureate-level degrees. These programs represent the
majority (88 percent) of program choices among students who enter the
postsecondary education system immediately out of high school (Bowlby
and McMullen 2002). University graduates of professional degree programs
(e.g., dentistry, DDS; veterinary medicine, DVM; law, LLB; and medicine,
MD) were removed from the analyses because these programs are generally
not accessible to students directly out of high school. (8) Similarly,
respondents with degrees from graduate-level university programs (e.g.,
MA and PhD) are also excluded because students in these programs
typically rely on additional sources of funding such as teaching
assistantships and graduate scholarships. We also removed respondents
who received an additional postsecondary credential or studied full time
within two years of graduation, because we are interested in examining
the experiences of those who graduated in 2005 and did not return to
full-time study during the survey period. As the purpose of this study
is to examine issues relating to government student loan repayment,
respondents who did not receive financial assistance in the form of
government student loans are excluded from the analyses. Finally, a
small number of observations (9) were removed as a result of missing
data, leaving a total of 7,535 cases for our analysis.
Variables and Procedures
This study employs logistic regression to identify which
postsecondary graduates are most likely to default on their government
student loans. The key explanatory variables used for our analysis are
level of schooling and field of study. The level of schooling variable
distinguishes between graduates of community college diploma or
certificate programs and university baccalaureate-level degree programs.
The field of study variable is divided into eight categories based on
the Classification of Instructional Programs. This is the same field of
study classification system developed and used by the National Center
for Education Statistics in the United States. The field of study
categories are as follows:
1. Education.
2. Liberal arts (including the fine arts, humanities, and social
sciences). (10)
3. Commerce, management, and business administration. (11)
4. Mathematics and physical and biological sciences.
5. Engineering and computer science.
6. Health-related fields.
7. Other (including not specified, undeclared and fields that do
not fit meaningfully in any of the above categories).
The statistical analyses control for the primary sociodemographic
characteristics assessed at the time of the survey such as sex, marital
status, age, region of habitation, presence of dependent children in the
household, and visible minority status. In addition to the amount of
outstanding government loan, we also included variables that identify
whether the respondents received bursaries, grants, or other
scholarships over the course of their programs as well as whether or not
the respondents have other debt in addition to their government student
loans. Combined, these variables capture issues relating to overall debt
burden, socioeconomic status, and ability (scholarships), which are
factors that will likely impact repayment outcomes. Finally, we also
include the earnings variable derived by Statistics Canada, which
represents an estimate of the respondents' gross annual earnings
during the 2007 calendar year. (12)
The NGS uses two separate but related measures to capture the level
of difficulty respondents have repaying their student debt; the level of
difficulty reported by the respondent, and whether the respondent had
defaulted on his/her government loan. Defaulting is a more pressing
issue because defaulted loans are transferred to the Canada Revenue
Agency for collection and can result in a compromised credit rating. All
of the NGS surveys have measured the subjective level of difficulty
reported by recent graduates but a question explicitly addressing
default was only included in more recent NGS surveys. Still, most
research employing the NGS data includes only the question capturing
reported repayment difficulty. (13) For this study, we employ the
self-reported measure of whether the respondent had defaulted on his or
her government student loan as the response variable. Respondents who
reported "yes" are coded as 1, while those who reported
"no" are coded as 0. The categories and descriptive statistics
(proportions and means) relating to each variable in this study can be
found in Table 1.
DESCRIPTIVE RESULTS
Descriptive statistics for graduates of college and university
undergraduate-level programs who received government student loans are
provided in Table 1. Females made up the majority of both college and
undergraduate-level graduates of 2005 (61.78 and 62.6 percent,
respectively). The average age at graduation was approximately 28 years
for both college and undergraduate-level university graduates,
indicating that college graduates typically enter their programs at an
older age. Nearly two-thirds of both college and undergraduate-level
students were single at graduation. Similarly, most respondents did not
report having any dependent children at the time of graduation (73
percent of college graduates and 84 percent of bachelor-level
graduates).
Table 1 also provides a breakdown of the proportion of college and
bachelor's level university graduates from each field. As expected,
at the undergraduate level the largest proportion of graduates come from
the liberal arts (29 percent). The smallest proportion of
undergraduate-level respondents graduated from engineering/applied
science fields (8 percent), along with fields classified as
"other" (1 percent). In comparison, among college graduates,
17 percent of respondents come from fields classified as engineering or
applied sciences, while 20 percent of respondents were from the liberal
arts field. Seventeen percent of bachelor's level respondents and 2
percent of college-level respondents graduated from the education
related fields. A higher proportion of college-level graduates (22.5
percent) than bachelor's level graduates (14 percent) received
health-related credentials. The same can be said to a lesser extent for
graduates of the business related fields, which represented 22 percent
of college graduates and 19 percent of undergraduate-level graduates.
The data in Table 1 also reveal that respondents whose parents had
obtained postsecondary schooling were more likely to graduate from
bachelor's level university programs than from college-level
programs. Approximately 48 percent of bachelor's level graduates
reported having a mother who had attained a postsecondary credential,
whereas only 35 percent of college-level graduates' mothers had
completed tertiary education. As shown in Table 1, a nearly identical
pattern is apparent regarding the educational attainment of the
graduates' fathers.
Graduates of undergraduate-level programs were much more likely to
obtain grants or bursaries (45 percent) than graduates of college-level
programs (35 percent). Likewise, graduates of bachelor's level
programs were also more likely to receive scholarships (42 percent) than
their counterparts from community colleges (24 percent).
Baccalaureate-level university graduates were much more likely to
borrow from both government and other sources than college graduates.
Approximately 23 percent of bachelor's level graduates had borrowed
up to $4,999 in additional loans in comparison with only 14 percent of
community college graduates who had done so. The figures for those who
had borrowed more than $5,000 from other sources are 13 and 12 percent
for university and college graduates, respectively.
University-level graduates had borrowed considerably more from
government sources than their counterparts from community colleges. The
average government student loan owed at graduation by
undergraduate-level university graduates of the class of 2005 was
$19,212 while those graduating from community colleges owed $12,648.
Despite the lower average debt levels of college graduates, these
individuals are more likely to default on their government student loans
two years after graduation. Among college graduates, 9.41 percent
reported that they had defaulted on their student loans within two years
after graduation. The corresponding figure for university graduates is
7.72 percent. Finally, the average yearly earning in 2007 dollars is
$37,250 for community college graduates and $46,031 for university
graduates.
REGRESSION RESULTS
The response (dependent) variable distinguishes between graduates
who did and did not default on their student loans within the two-year
period after graduation. This variable may mildly underestimate the
total number of defaults since our data only pertain to the first two
years after graduation. However, it has been well documented in the
literature that most defaults occur within the first two or three years
of the repayment period; therefore the two year default rate, while
conservative, is a good indicator of total default (Dynarski 1994; HRSDC
2007a; Kapsalis 2006). Since the response variable consists of two
discrete categories, we employ a series of logistic regression models.
(14)
The purpose of the regression analysis is to identify which
postsecondary graduates are likely to default on their government
student loans, while controlling for other factors that potentially
confound this relationship. Table 2 provides the parameter estimates for
the regression of the response variable on the explanatory variables
discussed above for all respondents who borrowed from the government for
their schooling. The regression estimates provided in Model 1 reveal
that females are significantly less likely to default on their
government student loans than are males (p < .05), controlling for
the other variables in the model. (15) Likewise, those who are married
are less likely to report defaulting on their government student loans
than are those who are not married (p < .001), controlling for the
other variables in the model. The parameter estimate for age reveals
that older respondents are more likely to default on their government
loans than are younger respondents (p < .001). One possible
explanation for this unexpected finding is that younger graduates may be
more likely to live with their parents and remain dependents during the
early part of their careers. Therefore, younger graduates may experience
lower costs of living relative to older graduates, which, in turn,
provides them with extra money to repay their student loans.
Interestingly, respondents who have dependent children are more likely
to default on their government student loans than are respondents who do
not have any dependent children (p < .001). Thus, the financial
burden associated with supporting dependent children appears to have a
detrimental impact on the ability to repay a student loan.
The effects of both variables relating to visible minority status
and mother's education are not statistically significant. However,
graduates who report that their father does not have a postsecondary
education are more likely to default on their government loans than
graduates who indicate that their father does have a postsecondary
education (p < .001).
While the estimates from Model 1 reveal that receiving scholarships
or bursaries does not impact default rates, graduates with larger
student loans are more likely to default than graduates with smaller
student loans (p < .001). Similarly, graduates who also borrowed from
other (nongovernment) sources to support their education are more likely
to default on their government student loans than graduates who did not
borrow from other sources.
The "main" effects of both level of schooling and field
of study are statistically significant (p < .001), controlling for
the other variables in the model. Significance tests for variables are
for the effect of the entire explanatory variable (e.g., multicategory
categorical variables) on the response variable, whereas significance
tests for dummy variables compare the effect of each category of the
explanatory variable relative to the reference category. (16)
The interaction term in Model 2 is used to assess whether the
effect of field of study on student loan defaults is related to level of
schooling (college versus university), and vice versa. (17) The effect
of the interaction is statistically significant (p < .01),
controlling for the other variables in the model. (18) To improve the
interpretability of these results, the log-odds estimates from Model 2
are converted into predicted probabilities and plotted in Figure 1,
holding the control variables constant at typical values. (19) The
estimates are accompanied by their corresponding 95 percent confidence
intervals to provide additional guidance when interpreting the
differences between the groups (for more information on the construction
of these effects displays, see Fox and Andersen 2006).
Figure 1 reveals that the probability of defaulting on a government
student loan varies strongly across the fields of study. Among
university graduates, the probability of defaulting on a government
student loan is lowest for engineering graduates (.03), and highest for
liberal arts graduates (.11). The probability of defaulting on a
government student loan is also quite low for university graduates with
health (.05) and math and science related degrees (.05).
Among community college graduates, those of fields relating to
engineering and health have the lowest probability of defaulting on
their government student loan; their predicted probabilities are .05 and
.06, respectively. Community college graduates with highest predicted
probabilities come from fields relating to the liberal arts (.10),
business (.11), education (.12), and those classified as
"other" (.17).
Probably the most interesting finding in Figure 1 is that community
college graduates of engineering and health-related fields are
significantly less likely to default on their government loans than are
university graduates of liberal arts fields. The findings displayed in
Figure 1 are particularly informative because they reflect differences
that are attributable to credentials (e.g., field of study and level of
schooling), and not to the characteristics of the respondents with those
credentials (i.e., the statistical controls included in the regression
model).
The models estimated in Table 3 apply only to graduates who have
full-year employment and are employed full-time (>30 hours per week).
Aside from the selection criteria, the only difference between these
models and those provided in Table 2 is the inclusion of earnings (using
the natural logarithm) as an explanatory variable. Interestingly, the
effect of earnings on whether a postsecondary graduate will default on
his or her government student loans is not statistically significant
when controlling for the other variables in the model. (20) If
individuals were acting rationally, then we might expect that those who
were earning the most would presumably make their loan payments.
However, this may not be the case.
[FIGURE 1 OMITTED]
This finding resembles that of some U.S. studies that suggest
borrowers' incomes may not be tightly related to their likelihood
of defaulting (see Flint 1997). The respective coefficients for all the
other explanatory variables in terms of their relative strength and
direction (positive or negative) are quite similar in Tables 2 and 3.
One discrepancy worth reporting is that the effect of the variable
capturing the amount of government student loans is statistically
significant in the models for all graduates who borrowed government
money (Table 2), but not statistically significant for the models
estimated for graduates who are employed full time (Table 3).
In Model 2, the effect of the interaction between level of
schooling and field of study is statistically significant (p < .05).
In order to provide a meaningful interpretation of the interaction
effect, the regression coefficients relating to level of schooling and
field of study from Model 2 of Table 3 are converted into predicted
probabilities and plotted in Figure 2, holding the control variables at
their means (for quantitative variables) and proportions (for
categorical variables). The pattern of predicted probabilities displayed
in Figure 2 are very similar to those presented in Figure 1. In other
words, the impact of the interrelationship between field of study and
level of schooling on whether graduates will default on their government
student loan does not change when focusing only on respondents employed
full time, year round. Likewise, when restricting analyses to
respondents who are employed full time (two years after graduation) the
probability of defaulting on a government student loan falls slightly
for all groups of graduates, but remains above. 10 for university
graduates of the liberal arts, and community college graduates of fields
classified as "other." The implications relating to these
findings will be discussed below.
DISCUSSION
The results of this study are consistent with past research
indicating that a sizable minority of graduates with government student
loans have difficulty repaying (e.g., Clark 1999; Dubois 2006; Finnie
2001a; Kapsalis 2006). (21) When controlling for the sociodemographic
characteristics and holding them constant at typical values, the
probability of defaulting ranges from .03 to .17, depending on level of
study and field of study. Interestingly, our results also show that
females are less likely to default on their student loans than males,
regardless of field of study or level of program. This finding runs
counter to past research that indicates that females have more
difficulty repaying their student loans (Adamuti-Trache et al. 2006;
Finnie 2000; Penner 2008). One possible explanation for females'
lower levels of reported loan defaults and high levels of reported
repayment difficulties is that females may be more likely than males to
continue repaying their loans despite experiencing challenges, perhaps
because they are more averse to the risks and consequences of default
(Chen and Volpe 2002).
[FIGURE 2 OMITTED]
Our findings support existing evidence that default rates to a
large extent depend upon one's level of education, where college
graduates face greater repayment difficulties (Dubois 2006; Finnie 200
la). (22) At the outset, this finding appears to support theories of
human capital and rational action, as university programs require
additional years of education, and on average yield higher earnings. By
extension, providing graduates act rationally with their higher earnings
and make loan repayments a priority, we would expect defaults to be much
lower for these graduates.
However, our findings reveal that field of study may be a more
important predictor of repayment experiences. Both college and
university graduates from more technical and applied fields, credentials
deemed necessary in today's knowledge economy, were less likely to
default on their loans. In fact, baccalaureate-level graduates with
liberal arts degrees are significantly more likely to default than are
community college graduates with diplomas in health and engineering
related fields, controlling for the other variables in our models.
Considering that approximately 29 percent of baccalaureate-level
university students who borrow from the government to finance their
education are from the liberal arts, it is discouraging that more than
one in every 10 of these graduates can be expected to default on their
loans within two years of graduation, even though they are employed full
time, full year.
Rational action and human capital theories suggest that liberal
arts graduates would experience greater repayment problems largely
because of their relatively lower earnings. Yet these theories fall down
in one major respect. The findings of this study indicate that the
higher level of student loan default among liberal arts graduates, for
the most part, is attributable to factors beyond individual earnings.
Previous studies have indicated that the amount a student borrows is
generally not related to his/her field of study because most students
borrow the maximum allowable amount (Finnie 2001a; see also Folloni and
Vittadini 2010). Thus, students' borrowing practices may not be as
tightly coupled with anticipated labor market returns.
One alternative explanation is that students may be consuming
beyond their means even after graduating. Brint and Rotondi (2008)
suggest that during one's educational careers, students are
willingly taking on greater levels of debt to enjoy the "college
experience." Perhaps these patterns of consuming beyond one's
means extend into their repayment behaviors. For example, issues
surrounding student debt and student loan defaults represent part of a
larger problem relating to poor financial planning, consumer spending
habits, and consumer debt in the population as a whole. In fact, drawing
on data from the Survey of Financial Security, conducted in 1999 and
2005, a recent report by the Bank of Canada revealed that household debt
(mortgage, credit card, and line of credit) increased dramatically
across all classes and age groups during this period (Meh et al. 2009).
Drawing on ability to pay theories, a second explanation could be
that students from higher socioeconomic backgrounds are relying on their
parents to step in and provide financial assistance. Ability to pay
theories also helps explain our positive effect of being married and the
negative effect of having children on default rates. Single individuals
and larger families would have greater financial constraints, which may
lead to higher defaults. While our results do control for the effects of
parents' education levels, the NGS data do not allow us to rule out
the possibility that students from wealthier backgrounds may rely on
their family as a social safety net to avoid defaulting. In fact, the
consistently significant effect of father's education reducing the
chances of defaulting may reflect this possibility. It may be the case
that individuals from certain fields of study have parents with more
financial resources. Certainly, a growing number of studies are showing
that in expanded systems of postsecondary education, field choices are
also influenced to some degree by family background characteristics
(Davies and Guppy 1997; Davies and Hammack 2005; Goyette and Mullen
2006; Zarifa 2008).
Finally, field of study differences may be reflective of the
characteristics of the graduates and their respective programs. Arum and
Roksa (2011) suggest that financial literacy across campuses is quite
low, and that students often fail to consider the consequences of
student debt. The great majority of students simply do not know the
repayment terms of the loans (e.g., duration of the loan, repayment
amount, repayment schedule). The relatively high rates of student loan
default found in this study are surprising considering that the federal
Interest Relief Program is available, which suspends loan payments for
six months at a time for those students who fall below an income
threshold prorated for family size (HRSDC 2007b). Thus, a significant
portion of individuals are defaulting on their loans within two years
after graduation despite the availability of this program. One
possibility is that the program has not been sufficiently promoted. In
fact, in 2006 a Client Satisfaction Survey of Canada Student Loan
Borrowers revealed that a large minority of borrowers are not even aware
of any repayment assistance options available to them. (23) This lack of
awareness may be further evidence of the larger financial literacy
problem.
This financial literacy may vary across fields. For example,
students of fields such as computer science, engineering, and business
are likely to graduate with proficient skills in numeracy and data
management that are easily transferable to financial planning. They are
probably also more likely than their liberal arts counterparts to have
taken courses relating to accounting, if not personal finance. The
skills acquired by liberal arts graduates in both college and
university, on the other hand, are likely much less helpful in preparing
them for matters relating to money management, which might explain their
relatively high default levels. Thus, it might be valuable for
government and institution officials involved in the CSLP to consider
making additional courses or workshops relating to financial planning
available, if not mandatory, for students who require financial
assistance.
The authors would like to thank Reza Nakhaie and the anonymous
reviewers for their constructive comments. We also thank Paul-Phillippe
Pare for translating the abstract into French. Funding for this research
was provided by a grant from the Social Science and Humanities Research
Council of Canada.
References
Adamuti-Trache, M., C. Hawkey, H.G. Schuetze and V. Glickman. 2006.
"The Labour Market Value of Liberal Arts and Applied Education
Programs: Evidence from British Columbia." Canadian Journal of
Higher Education 36(2):49-74.
Allen, M. and C. Vaillancourt. 2004. "Class of 2000: Profile
of Postsecondary Graduates and Student Debt." Culture, Tourism and
the Centre for Education Statistics Division, Statistics Canada.
Andres, L. and M. Adamuti-Trache. 2007. "You've Come a
Long Way, Baby? Persistent Gender Inequality in University Enrollment
and Completion in Canada, 1979-2004." Canadian Public Policy
33(1):93-116.
Arum, R. and J. Roksa. 2011. Academically Adrift: Limited Learning
on College Campuses. Chicago, IL: The University of Chicago Press.
Bayard, J. and E. Greenlee. 2009. "Graduating in Canada:
Profile, Labour Market Outcomes and Student Debt of the Class of
2005." Culture, Tourism and the Centre for Education Statistics
Division, Statistics Canada.
Becker, G.S. 1993. Human Capital: A Theoretical and Empirical
Analysis, with Special Reference to Education. 3d ed. Chicago, IL: The
University of Chicago Press.
Bowlby, J.W. and K. McMullen. 2002. "At a Crossroads: First
Results for the 18 to 20-Year-Old Cohort of the Youth in Transition
Survey." Human Resources and Skills Development Canada. Service
Canada, Ottawa, Canada.
Brint, S. and M.B. Rotondi. 2008. "Student Debt, the College
Experience, and Transitions to Adulthood." Paper presented at the
annual meetings of the American Sociological Association, July 31 to
August 4, Boston, MA.
Cabrera, A.F., A. Nora and M.B. Casteneda. 1992. "The Role of
Finances in the Persistence Process: A Structural Model." Research
in Higher Education 33(5):571-93.
Cabrera, A.F., J.O. Stampen and W.L. Hansen. 1990. "Exploring
the Effects of Ability to Pay on Persistence in College." Review of
Higher Education 13(3):303-36.
Chapman, B. 1997. "Conceptual Issues and the Australian
Experience with Income Contingent Chargers for Higher Education."
The Economic Journal 107(442):738-51.
Chen, H. and R.P. Volpe. 2002. "Gender Differences in Personal
Financial Literacy among College Students." Financial Services
Review 11:289-307.
Christie, H. and M. Munro. 2003. "The Logic of Loans:
Students' Perceptions of the Costs and Benefits of the Student
Loan." British Journal of Sociology of Education 24(3):621-36.
Clark, W. 1999. "Paying off Student Loans." Perspectives
on Labour and Income 11(1):28-33. Statistics Canada, Catalogue No.
75-001-XPE.
Cliff, R., C. Hawkey, A.M. Vaughn. 1998. "A Background
Analysis of the Relationship between Tuition Fees, Financial Aid, and
Student Choice." Retrieved July 23, 2009
(http://www.cufa.bc.ca/Clift/A_Background_Analysis.html).
Davies, S. and N. Guppy. 1997. "Fields of Study, College
Selectivity, and Student Inequalities." Social Forces 73(4):
131-51.
Davies, S. and F.M. Hammack. 2005. "The Channeling of Student
Competition in Higher Education: Comparing Canada and the U.S." The
Journal of Higher Education 76(1): 89-106.
Drewes, T. 2006. "Returns to College Education: Evidence from
the 1990, 1995, and 2000 National Graduates Surveys." Human
Resources and Skills Development Canada. Service Canada, Ottawa, Canada.
Dubois, J. 2006. "Trends in Student Borrowing and Pathways:
Evidences from the 1990, 1995 and 2000 Classes." Human Resources
and Skills Development Canada. Service Canada, Ottawa, Canada.
Dynarski, M. 1994. "Who Defaults on Student Loans? Finding
from the National Postsecondary Student Aid Study." Economics of
Education Review 13(1):55-68.
Fever, R, G. Rees and S. Gorard. 1999. "Some Sociological
Alternatives to Human Capital Theory and Their Implications for Research
on Post-Compulsory Education and Training." Journal of Education
and Work 12(2):117-40.
Finnie, R. 2000. "From School to Work: The Evolution of Early
Labour Market Outcomes of Canadian Postsecondary Graduates."
Canadian Public Policy 26(2): 197-224.
Finnie, R. 2001a. "Student Loans: The Empirical Record."
The Canadian Journal of Higher Education 31(3):93-142.
Finnie, R. 2001b. "Fields of Plenty, Fields of Lean: The Early
Labour Market Outcomes of Canadian University Graduates by
Discipline." The Canadian Journal of Higher Education 31(1):141-76.
Finnie, R. 2002. "Student Loans, Student Financial Aid and
Post-Secondary Education in Canada." Journal of Higher Education
Policy and Management 24(2):155-70.
Finnie, R. and M. Frenette. 2003. "Earning Differences by
Major Field of Study: Evidence from Three Cohorts of Recent Canadian
Graduates." Economics of Education Review 22:170-92.
Finnie, R. and S. Schwartz. 1996. Student Loans in Canada: Past,
Present and Future. Toronto, Canada: C.D. Howe Institute.
Flint, T.A. 1997. "Predicting Student Loan Defaults."
Journal of Higher Education 68(3): 322-54.
Folloni, G. and G. Vittadini. 2010. "Human Capital
Measurement: A Survey." Journal of Economic Surveys 24(2):248-79.
Fox, J. and R. Andersen. 2006. "Effect Displays for
Multinomial and Proportional-Odds Logit Models." Sociological
Methodology 36:225-55.
Goyette, K.A. and A.L. Mullen. 2006. "Who Studies Arts and
Sciences? Social Background and the Choice and Consequences of
Undergraduate Field of Study." The Journal of Higher Education
77(3):497-538.
Hesseldenz, J. and D. Stockham. 1982. "National Direct Student
Loan Defaulters: The Ability to Repay." Research in Higher
Education 17(1):3-14.
Human Resources and Skills Development Canada. 2004.
"Formative Evaluation of the Canada Student Loans Program--October
20047' Retrieved August 14, 2009
(http://www.rhdcchrsdc.gc.ca/eng/publicationsresources/evaluation/2007/sp_ah_671_10_04e/page02.shtml).
Human Resources and Skills Development Canada. 2007a. "Annual
Report 20042005." Retrieved July 22, 2009
(http://www.hrsdc.gc.ca/eng/learning/canadastudentloan/
Publications/annual_reportJ2004-2005/execsummary.shtml).
Human Resources and Skills Development Canada. 2007b. "Canada
Student Loans Program Statistics." Retrieved August 14, 2009
(http://www.hrsdc.gc.ca/eng/hip/cslp/statistics/
08_st_DefaultRatesCanadaProvinces.shtml).
Human Resources and Skills Development Canada (HRSDC). 2009.
"The Canada Student Loans Program." Retrieved July 22, 2009
(http://www.hrsdc.gc.ca/eng/learning/canada student_loan/cslp.shtml).
Kapsalis, C. 2006. "Factors Affecting the Repayment of Student
Loans." Culture, Tourism and the Centre for Education Statistics
Division, Statistics Canada.
Lamb, M. and M. Sutherland. 2010. "The Components of Career
Capital for Knowledge Workers in the Global Economy." The
International Journal of Human Resource Management 21(3):295-312.
Lavoie, M. and F. Ross. 1999. "Is it Worth Doing a Science or
Technology Degree in Canada? Empirical Evidence and Policy
Implications." Canadian Public Policy 251(1):101-22.
Meh, C.A., Y. Terajima, D. Xiao Chen and T. Carter. 2009.
"Household Debt, Assets, and Income in Canada: A Microdata
Study." Discussion Paper 2009-7, Bank of Canada.
Michael, S.O. and M. Kretovics, eds. 2005. Financing Higher
Education in a Global Market. Higher Education Series. New York: Algora
Publishing.
National Centre for Education Statistics (NCES). 2009.
"Average Undergraduate Tuition Fees and Room and Board Rates
Charged for Full-Time Students in Degree-Granting Institutions, by Type
and Control on Institution: Selected Years, 1980-81 to 2007-08."
U.S. Department of Education. Retrieved September 2, 2009
(http://nces.ed.gov/programs/ digest/d 10/tables/dt10_346.asp).
National Student Loans Service Centre (NSLSC). 2009. "National
Student Loans Service Centre: On-Line Services." Retrieved July 26,
2009 (https://nslsc.canlearn.ca/eng/ default.aspx).
Penner, A.M. 2008. "Race and Gender Differences in Wages: The
Role of Occupational Sorting at the Point of Hire." The
Sociological Quarterly 49:597-614.
Riddell, W.C. 2008. "Investing in Human Capital: Policy
Priorities for Canada." IRPP.
Ryan, L.D. 1993. "California State University Loan
Defaulters' Characteristics." Journal of Student Financial Aid
23(3): 29-42.
Schwartz, S. and R. Finnie. 2002. "Student Loans in Canada: an
Analysis of Borrowing and Repayment." Economics of Education Review
21:497-512.
Servage, L. 2009. "Alternative and Professional Doctoral
Programs: What is Driving the Demand?" Studies in Higher Education
34(7):764-79.
Spencer, L.E. 1974. "Risk Measurement for Short-Term
Loans." Journal of Student Financial Aid 4(3):30-35.
Statistics Canada. 2009. "Average Undergraduate Tuition Fees
for Full Time Canadian Students, by Discipline, by Province."
Retrieved August 9, 2009 (http://www40.statcan.gc.ca/
101/cst01/educ50a-eng.htm).
Sweet, R.A. and P. Anisef, eds. 2005. Preparing for Post-Secondary:
New Roles for Governments and Families. Kingston, ON:
McGill-Queen's University Press.
Volkwein, J.F. and B.P. Szelest. 1995. "Individual and Campus
Characteristics Associated with Student Loan Default." Research in
Higher Education 36(1):41-70.
Walters, D. 2004. "The Relationships between Postsecondary
Education and Skill: Comparing Credentialism with Human Capital
Theory." The Canadian Journal of Higher Education 34(2):97-124.
Walters, D. 2006. "An Examination of the School-to-Work
Transitions of Male and Female College and University Graduates of
Applied and Liberal Arts Programs in Canada." Higher Education
Policy 19:225-50.
White, J.M., S. Marshall, and J.R. Wood. 2005. "Family
Structure, Child Well-Being, and Post-Secondary Saving: The Effect of
Social Capital on the Child's Acquisition of Human Capital."
Pp. 222-48 in Preparing for Post-Secondary Education: New Roles for
Governments and Families, edited by R.A. Sweet and P. Anisef. Kingston,
ON: McGill-Queen's University Press.
Zarifa, D. 2008. "Emerging Forms of Stratification in Higher
Education: Comparing Canada and the United States." Department of
Sociology, McMaster University, Ontario, Canada. Unpublished doctoral
dissertation.
LAURA WRIGHT
University of Western Ontario
DAVID WALTERS
University of Guelph
DAVID ZARIFA
Nipissing University
(1.) All dollar figures throughout the paper are reported in
Canadian dollars.
(2.) During the 2000 to 2001 academic year, over $1.5 billion was
lent to students across Canada.
(3.) Finnie (2001a) examined payback rates by field and gender for
bachelor's level graduates using 1995 NGS data. This is the latest
data we could find. No similar research could be found examining
college-level graduates.
(4.) Clark (1999) was the only study found using the NGS that
addresses student loan default explicitly.
(5.) Finnie (2001a) compared reported repayment difficulties for
both college and university graduates in general. However, he provided a
comprehensive examination of the debt-to-earnings ratios and reported
repayment difficulties for university graduates of different fields, but
not those of community college graduates of different fields.
(6.) Using Statistics Canada's Longitudinal Administrative
Database, combined with the CSLP administrative records, Kapsalis (2006)
explored the default rates of graduates of different fields and levels
for the class of 1995. However, the analysis is limited to descriptive
statistics and does not include statistical controls.
(7.) In Finnie's (2001a) analysis, debt burdens are calculated
by dividing the amount a student owes at graduation by his/her annual
earnings such that a higher ratios depicts a larger debt burden.
(8.) Technically these are undergraduate programs, however, they
are typically classified as "professional" because they are
required for access into highly regulated professions. Admission to
these programs is considerably more competitive than typical
undergraduate degree programs, as they generally require high
grade-point-averages for at least two-year undergraduate schooling, and,
in some instances, competitive standardized test results.
(9.) Less than 10 percent of cases were missing.
(10.) Due to issues relating to sample size graduates of the fine
arts, humanities, and social sciences had to be grouped together as the
"liberal arts." We had to pat these graduates together because
there were too few respondents of community colleges with credentials in
the fine arts and humanities. Separate analyses for university graduates
revealed that fine arts graduates are more likely to report defaulting
on their student loans than social science. The difference was
statistically significant (p < .05), with and without earnings in the
model. The difference in loan default between graduate of the humanities
and each of the fine arts or social sciences was not statistically
significant in the models we estimated.
(11.) To be consistent with previous research (e.g., Finnie 2001a;
Finnie and Frenette 2003; Schwartz and Finnie 2002) graduates of fields
relating to economics were removed from the social sciences and placed
with graduates of commerce and business administration.
(12.) This measure is based on the respondents' reported
salary, how it was paid (hourly, weekly, monthly, or yearly) and the
number of hours usually worked.
(13.) Clark (1999) examined the 1995 NGS cohort and found that 4
percent of graduates with student loans had defaulted by 1997.
(14.) The two quantitative explanatory variables capture age and
the amount of money borrowed in the form of government student loans.
Indicator (0-1 dummy) coding is used for the categorical variables. The
reference categories are labeled in Table 2.
(15.) We report all coefficients that are statistically
significant. Only the key estimates in our analyses relating to
postsecondary programs will be converted into predicted probabilities
for a better assessment of the practical impact of these variables on
loan default.
(16.) Significance tests for the parameter estimates are based on
the z-test (e.g., the regression estimate divided by its standard
error). The significance tests for variables involving multiple
parameter estimates are obtained via the likelihood ratio chi-square
test.
(17.) The inclusion of the interaction term did not have a
substantial impact on the other parameter estimates in the model.
(18.) We included a three-way interaction among level of schooling,
field of study, and gender. The effect of this interaction was not
statistically significant (nor were two-way interactions involving
gender), indicating that the default levels for males and females do not
vary across postsecondary type.
(19.) Means are used for quantitative variables and proportions are
used for categorical variables.
(20.) We also estimated a model where we replaced the log of
earnings with earnings, and the effect of earnings was not statistically
significant.
(21.) These findings likely underestimate the seriousness of the
student loan default problem, as we could only account for those who
successfully completed their programs. The overall estimates of student
loan default would likely be much higher if borrowers who were
unsuccessful in completing their programs could have been included in
the analyses.
(22.) It is important to note that our data do not contain measures
of family income. Thus, it is possible that part of the level of
education and field effects may be attributable to the family background
effects on those student choices.
(23.) See http://www.hrsdc.gc.ca/eng/learning/canada_student_loan/Publications/annual_report/2005.2006/ part2-e.shtml#31.
Laura Wright, Department of Sociology, University of Western
Ontario, London, Ontario, N6A 5C2. E-mail:
[email protected]
Table 1
Descriptive Statistics for the Variables in the Analysis
College University
Percentage Percentage
n = 3,379 n = 4,156
Sex
Female 61.78 62.60
Male 38.22 37.40
Marital status
Married 35.98 37.27
Not married 64.02 62.73
Dependent children
Yes 27.19 16.03
No 72.81 83.97
Region
Atlantic provinces 9.79 9.71
Quebec 22.20 21.41
Western provinces 28.51 29.75
Ontario 39.49 39.13
Visible minority status
Visible minority 20.35 23.40
Nonminority 79.65 76.60
Mother has postsecondary education
No 64.50 51.79
Yes 35.50 48.21
Father has postsecondary education
No 65.00 52.57
Yes 35.00 47.43
Bursaries/grants
Yes 35.13 44.99
No 64.87 55.01
Scholarships
Yes 24.49 41.71
No 75.51 58.29
Otherloans
<$5,000 13.93 23.15
$5,000+ 11.87 13.29
Did not borrow from other sources 74.20 63.56
Field of study
Education 2.46 17.40
Business 22.09 19.02
Math/sciences 8.90 11.35
Engineering/Applied science 16.65 8.34
Health 22.48 14.15
Other 7.21 1.00
Liberal arts 20.20 28.74
Default on government student loans 9.41 7.72
Mean Mean
Age 27.77 27.75
Amount of government student loans $12,648 $19,212
Table 2
Logistic Regression Model Predicting Government Student Loan
Default for College and University Graduates of Various Fields of
Study
Model 1
n = 7,535 b SE (b)
Constant -3.434 .249
Sex
Female -.201 .093 *
Male -- --
Marital status
Married -.468 .103 ***
Not married -- --
Dependent children
Yes .435 .122 ***
No -- --
Age .044 .007 ***
Region
Atlantic provinces .378 .149 **
Quebec .040 .131
Western provinces .091 .105
Ontario -- --
Visible minority status
Visible minority .189 .104
Nonminority -- --
Mother has postsecondary education
No -.028 .094
Yes -- --
Father has postsecondary education
No .302 .094 ***
Yes -- --
Bursaries/grants
Yes .171 .091
No -- --
Amount of government student loans/100 .007 .003 ***
Scholarships
Yes -.081 .095
No -- --
Other loans ***
<$5,000 .545 .117 **
$5,000+ .367 .108 ***
Did not borrow from other sources -- --
Field of study ***
Education -.405 .151 ***
Business -.338 .119 ***
Math/sciences -.728 .170 ***
Engineering/Applied science -1.123 .184 ***
Health -.823 .143 ***
Other .163 .208
Liberal arts -- --
Level of schooling
University -.351 .099 ***
College -- --
Field of study * level of schooling
Education * university
Business * university
Math/science * university
Engineering * university
Health * university
Other * university
Model 2
n = 7,535 b SE (b)
Constant -3.818 .273
Sex
Female -.203 .093 *
Male -- --
Marital status
Married -.483 .104 ***
Not married -- --
Dependent children
Yes .427 .123 ***
No -- --
Age .044 .007 ***
Region
Atlantic provinces .349 .149 *
Quebec .072 .132
Western provinces .077 .105
Ontario -- --
Visible minority status
Visible minority .217 .104 *
Nonminority -- --
Mother has postsecondary education
No -.019 .094
Yes -- --
Father has postsecondary education
No .295 .094 **
Yes -- --
Bursaries/grants
Yes .183 .091 ***
No -- --
Amount of government student loans/100 .007 .003 ***
Scholarships
Yes -.089 .095
No -- --
Other loans
<$5,000 .538 .117 ***
$5,000+ .366 .108 ***
Did not borrow from other sources -- --
Field of study ***
Education .217 .416
Business .173 .193
Math/sciences -.040 .265
Engineering/Applied science -.605 .247 ***
Health -.456 .219 ***
Other .627 .250 ***
Liberal arts -- --
Level of schooling
University .177 .170
College -- --
Field of study * level of schooling **
Education * university -.800 .445
Business * university -.805 .250 ***
Math/science * university -1.071 .346**
Engineering * university -.947 .388 ***
Health * university -.507 .294
Other * university -1.121 .604
* p < .05; ** p < .01; *** p < .001.
Table 3
Logistic Regression Model Predicting Government Student Loan
Default Controlling for Earnings
Model 1
n = 5,858 b SE (b)
Constant -2.572 1.399
Sex
Female -.237 .108 *
Male -- --
Marital status
Married -.594 .124 ***
Not married -- --
Dependent children
Yes .486 .151 ***
No -- --
Age .046 .009 ***
Region
Atlantic provinces .402 .181 *
Quebec -.069 .160
Western provinces .208 .122
Ontario -- --
Visible minority status
Visible minority .310 .121 *
Nonminority -- --
Mother has postsecondary education
No -.121 .111
Yes -- --
Father has postsecondary education
No .423 .110 ***
Yes -- --
Bursaries/grants
Yes .197 .107
No -- --
Amount of government student loans/100 .005 .004
Scholarships
Yes -.219 .113
No -- --
Other loans **
<$5,000 .261 .145
$5,000+ .352 .122 **
Did not borrow from other sources -- --
Field of study ***
Education -.478 .186 **
Business -.417 .143 **
Math/sciences -.784 .203 ***
Engineering/Applied science -1.241 .226 ***
Health -.617 .170 ***
Other .158 .242
Liberal arts -- --
Level of schooling
University -.279 .120 *
College -- --
Log of earnings -.086 .135
Field of study * level of schooling
Education * university
Business * university
Math/science * university
Engineering * university
Health * university
Other * university
Model 2
n = 5,858 b SE (b)
Constant -3.173 1.428
Sex
Female -.232 .109
Male -- --
Marital status
Married -.612 .125 ***
Not married -- --
Dependent children
Yes .486 .151 ***
No -- --
Age .046 .009 ***
Region
Atlantic provinces .376 .182 *
Quebec -.035 .162
Western provinces .193 .123
Ontario -- --
Visible minority status
Visible minority .332 .122 **
Nonminority -- --
Mother has postsecondary education
No -.111 .111
Yes -- --
Father has postsecondary education
No .422 .111 ***
Yes -- --
Bursaries/grants
Yes .209 .107
No -- --
Amount of government student loans/100 .005 .004
Scholarships
Yes -.211 .114
No -- --
Other loans **
<$5,000 .258 .146
$5,000+ .355 .123 **
Did not borrow from other sources -- --
Field of study **
Education .089 .574
Business .122 .238
Math/sciences -.052 .314
Engineering/Applied science -.744 .306 *
Health -.105 .255
Other .640 .292 *
Liberal arts -- --
Level of schooling
University .293 .207
College -- --
Log of earnings -.070 .137
Field of study * level of schooling *
Education * university -.747 .605
Business * university -.819 .299 **
Math/science * university -1.145 .411 **
Engineering * university -.819 .456
Health * university -.799 .336 *
Other * university -1.074 .717
* p < .05; ** p < .01; *** p < .001.
