80 C. Lo¨fgren, H. Ohlsson Economics of Education Review 18 1999 79–88
associated with ability. The negative effect of writing a second thesis has to do with preferences and so does the
decreasing probability over time. The lower probability of women is because of preferences, not ability.
The paper is structured as follows: In Section 2 we present a theoretical model of how the decision to com-
plete is made. The choice of variables and the descriptive facts are discussed in Section 3. Section 4 described the
estimation method. Empirical results are reported and discussed in Section 5. In Section 6 the main results are
summarized and discussed.
2. Theoretical framework
The educational achievement of a student is determ- ined both by the student’s ability and his or her prefer-
ences for making an effort studying instead of doing something else. To structure our discussion on the deter-
minants of thesis completion time we have extended a model of student behavior from Costrell 1994. In the
model, students are assumed to differ in preferences as well as in ability. Suppose that student i has the utility
function:
U
i
5 U
i
L
i
,w
i
1 where L
i
is the effort studying and w
i
is future earnings. It is assumed that
∂ U
i
∂ L
i
, 0 and ∂
U
i
∂ w
i
0. The student’s educational achievement, which is defined as
the student’s future productivity, is given by the edu- cational production function:
y
i
5 y
i
L
i
2 assuming
∂ y
i
∂ L
i
0 and ∂
2
y
i
∂ L
2 i
, 0. The production function reflects how the ability of the student deter-
mines his or her educational achievementfuture pro- ductivity. The zero level of study effort yields the pro-
ductivity y 5 y
i
L . This is assumed to be the same for
all students. Suppose that employers can identify the productivity
of the individual. Future earnings would then, in a well- functioning market economy, correspond to productivity,
w
i
5 y
i
. Choosing effort to maximize utility subject to the production function gives the first-order condition:
∂ y
i
∂ L
i
5 2 ∂
U
i
∂ L
i
∂ U
i
∂ y
i
3 where
∂ y
i
∂ L
i
is the marginal product of effort, ∂
U
i
∂ L
i
is the marginal disutility of study effort, and ∂
U
i
∂ y
i
is the marginal utility of future earnings. According to Eq.
3 a student in the situation above will increase his or her study effort as long as the utility from one extra hour
studying — through the resulting increase in future earn- ings — is larger than the disutility associated with the
extra hour of studies. In Fig. 1 a student with the pro- duction function y
1
would choose to study L
1
hours ther- eby reaching a maximum utility of U
1 1
. Suppose instead that employers, because of infor-
mation costs, cannot identify the productivity of the indi- vidual. What the employers can do is to distinguish
between students who have earned a degree and those who have not. In this situation there will be no individual
variation in productivity among those graduating. Future earnings will equal the productivity level associated with
the degree. Students have no incentive to study more than necessary to graduate, since such extra efforts
would not yield extra earnings. Suppose that the edu- cational achievement necessary to graduate is yˆ and that
the study effort resulting in this standard is Lˆ
i
. Assume also that the student has full information; i.e.,
he knows his production function. The student will then choose between two effort–earnings combinations:
max[U
i
Lˆ
i
,yˆ,U
i
L ,y
] 4
The student, whose situation is depicted in Fig. 1, will attain the standard yˆ by spending Lˆ
1
hours studying. The maximum standard that the student will meet is y˜
1
. The student would not choose to meet a higher standard than
this because he or she would then experience a lower utility than by choosing y
, which can be seen in Fig. 1. The time a student studies, therefore, depends both on
preferences and ability. However, the model produces the following important result: If we only study students
who have completed their thesis i.e., by excluding those that have chosen L
,y , the variation in effort L
i
can be attributed to differences in ability only. This is seen
Fig. 1. A student’s choice of study effort and future earnings.
81 C. Lo¨fgren, H. Ohlsson Economics of Education Review 18 1999 79–88
in Fig. 2, which shows the situation for three students. Student 1 meets the standard yˆ, completes his thesis, by
making an effort of Lˆ
1
hours. Student 2, on the other hand, does not complete her thesis. She chooses L
,y .
These two students have the same ability, i.e., the same production function, but different preferences, i.e., stud-
ent 2 has higher preferences for doing something else than studying. Student 3 also chooses to complete her
studies with an effort of Lˆ
3
. This student differs from the other two in both preferences and ability. It is obvious
then that when comparing all three students’ preferences and ability both determine effort. But when only compar-
ing completers — students 1 and 3 — the difference in effort Lˆ
3
2 Lˆ
1
is attributable only to the different pro- duction functions y
3
and y
1
, i.e., to the difference in ability.
Let us summarize: Student Production
Indifference Effort Thesis
function curve
1 y
1
U
1
Lˆ
1
Completed 2
y
2
U
2
L Not
completed 3
y
3
U
3
Lˆ
3
Completed So far the assumption has been made that a student
can only choose between meeting the standard yˆ or not. A more realistic description is that a student has a choice
between different standards for different grades. Suppose there are two grades other than failing as at Swedish
universities: passed yˆ
p
and passed with distinction yˆ
pd
. Compared to Eq. 4 the student now will choose among three effort–earnings combinations:
Fig. 2. Choice of time for studies and future earnings under
different production and utility functions.
max[U
i
Lˆ
pd i
,yˆ
pd
,U
i
Lˆ
p i
,yˆ
p
,U
i
L ,y
] 5
Preferences and ability will determine the choice in the way illustrated by Fig. 3. The figure shows two stu-
dents with identical production functions but different indifference maps. Student 2 will choose to meet yˆ
p
while student 1 will choose yˆ
pd
. The difference in study effort between these two students is determined by pref-
erences since they have the same production function. In general, it is the case that if we compare students who
have completed with the same degree, the study effort is a matter of ability only. However, if we compare com-
pleters both those who have passed with distinction and those who have passed it is not possible to attribute the
difference in study effort to ability only.
In this study the focus is on study time thesis com- pletion time. Denote this by l
i
and say it is measured in weeks. The completion time is by definition equal to
study effort, L
i
, divided by study intensity, L
i
l
i
the number of study hours per week so that l
i
; L
i
L
i
l
i
. In the model the student is assumed to make a choice of
the utility-maximizing effort. But there is also a choice of how to allocate this effort between study intensity and
study time. This means that for two students with the same ability the same production function and the same
L
i
for a given standard one of them may choose a rela- tively low effort per week L
i
l
i
and thereby have to use more weeks to complete the thesis. This is a question of
differences in preferences, not in ability. It follows that to fully distinguish, in accordance with the theoretical
model, between preference and ability effects one needs information on study effort, L
i
.
Fig. 3. Choice of study time with two different standards.
82 C. Lo¨fgren, H. Ohlsson Economics of Education Review 18 1999 79–88
3. Descriptive facts
