•k
[
\
LIBRARIES
•
Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium IVIember Libraries
http://www.archive.org/details/certificationoftOOacem
oB/m^^
HB31
.M415
'''^
IV
/
working paper
department
of
economics
Certification
Of Training And
Training Outcomes
Daron Acemoglu
Jorn-Steffen Pischke
No. 99-28
October 1999
massachusetts
institute of
technology
50 memorial drive
Cambridge, mass. 02139
WORKING PAPER
DEPARTMENT
OF ECONOMICS
Certification
Of Training And
Training Outcomes
Daron Acemoglu
Jorn-Steffen Pischke
No. 99-28
October 1999
MASSACHUSEHS
INSTITUTE OF
TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASS. 02142
MASSACHUSETTS iWSTITUTE
OF TECHNOLOGY
NOV
2 G 1999
LIBRARIES
Certification of Training
and Training Outcomes
Daron Acemoglu and Jorn-Steffen Pischke
MIT
July 1999
Abstract
External certification of workplace
is
widespread
in
many
countries.
skills
This
may
obtained through on-the-job training
indicate that training
is
financed by
workers, and certification serves to assure the quality of the training offered by the
is financed by firms, esGermany. We show in this paper that external certification of training
may also be necessary for an equilibrium with firm-sponsored training. Firm financing of training is only possible if firms have monopsony power over the workers
after training. If the training firm can extract too much of the employment rents,
firm.
However, other evidence shows that general training
pecially in
however, workers
ing.
may
not have sufficient incentives to put forth effort during train-
Certification increases the value training to the outside market,
workers, making firm-sponsored training possible.
and hence to
Introduction
1
Standard theory of human capital as developed by Becker maintains that workers should
pay
for investments in general skills.
by independent bodies,
especially
It
makes sense that such
when the
skills
are often certified
skills
are provided by firms at the workplace.
However, various pieces of evidence suggest that in a variety of circumstances firms rather
than workers pay
for investments in the general training of their
example. Bishop, 1997, Acemoglu and Pischke, 1999a).
that this
may be because other employers do not
Many
due to external
if
may be
certification of skills, there
for
observe whether an employee has received
become more
training programs
(see,
economists have suggested
training with his current employer (Katz and Ziderman, 1990,
According to this view,
employees
less
Chang and Wang,
easily identifiable, for
1996).
example
investment in training by firms.
In discussing proposals for occupational certification in the U.S., Heckman, Roselius, and
Smith (1994),
for
example, raised this
and argued that increased
issue,
certification
would
discourage training.
In this
light,
Germany's training system
is difficult
to understand.
Germany has an
extensive apprenticeship system which most non-college-bound youths complete before
becoming
full-time workers.
Many
studies have
documented that employers pay a
icant part of the financial costs of these training programs (see for
presented in Acemoglu and Pischke, 1998).
exams given by the chambers
of the training period,
and receive
received
The presence
certificates
of certificates that
by a worker observable to
example the evidence
However, apprenticeships are highly regu-
lated; apprentices take
skilled jobs.
of
commerce and
crafts at the
which play an important
make the amount and
end
role in access to
quality of the training
potential employers suggests that
about the quality of training that underlies
signif-
it is
not uncertainty
firms' incentives to invest in general skills in
Germany.
In Acemoglu and Pischke (1998), we developed an alternative theory of firm-sponsored
training based
on an asymmetry of information between current and potential employers
regarding the ability of young workers.
Valuable information about the abilities and
aptitudes of young workers will be revealed diiring the early years of their career.
Much
of this information will be directly observed by the current employer, but not necessarily
by other
skills
firms.
This informational monopsony
of their employees.
We
program, which they often
induce firms to invest in the general
argued that this theory accounts
the apprenticeship programs in Germany.
puzzling for this theory as well.
will
for the prevalence of
Nevertheless, certification of training skills
is
Why do firms operate within the regulated apprenticeship
criticize as rigid
and outdated, rather than developing
their
own
training programs without certification?^
We
In this paper, we attempt to answer this question.
tion of training in a labor market where
example
tion of
in the
form of
We
skills.
effort, is
discuss the role of certifica-
some investment by the workers themselves,
necessary during training for the successful accumula-
show that contrary to conventional wisdom,
certification
be necessary to encourage firm-sponsored training. The institutions that
ticeship skills therefore
may
actually
certify appren-
emerge as an integral part of the German training system, and
would not increase firm-sponsored
their removal
for
training, but likely
undermine the whole
system.^
The underlying
intuition for our results
employees because they have
sufficient
is
simple. Firms invest in the training of their
monopsony power, which makes them,
While labor market imperfections,
part, the residual claimant for the returns to training.
and
in particular
at least in
asymmetric information, increase the monopsony power of firms and
encourages firms to invest in training, they also reduce the workers' incentives to invest
in their skills, as
the worker
is
most of the returns
will
be appropriated by the
effort,
commit to reward workers
effort.
But
the problem woiild be solved.
diligence that workers exert in training programs
firms could
If
like
is
effort activities,
hard to monitor.
is
and
the
Alternatively,
if
conditional on their effort, the right incentives
often not possible in equilibrium either. This
in the labor
firms could observe
most
could be given to workers. Since the efibrt choice of the worker
this
by
If effort
necessary to complete the training, the labor market arrangements have
to give sufficient incentives to workers to exert this
monitor such
firm.
is
not observed, however,
means that asymmetric information
market might undermine the existence of training programs by not giving
enough incentives to workers. In
this setting, certification arises as
a method of ensuring
that workers receive more of the return to their general training. Certification therefore
helps to balance the power between workers and firms evenly, encouraging workers to
exert effort. In fact, in our simple model, firm-sponsored training arises as an equilibrium
with
certification,
We
but without certification training
next outline the basic environment.
in the absence of certification.
firms caimot
We
is
In section
never an equilibrium.
3,
we
characterize the equilibrium
simplify the analysis in this section
commit to future wages, and show that
by assuming that
in the absence of certification, there
^Although the German firms are currently not allowed to design their training programs, this is a
though they
could do so.
^Burtless (1994) makes an informal argument that more certification would encourage workers to
obtain more training but does not clearly articulate the reasons why there is too little training without
relatively recent development. In the past, they chose not to develop such programs, even
certification.
be no training in equilibrium. In section 4 of the paper, we solve
will
for the equilibrium
with certification, and show that there now exists an equilibrium in which firms invest in
the
skills of their
we
5,
workers, and workers exert effort diiring the training period.
relax the assumption of no
commitment
to future wages,
In section
and show that even
presence of such commitment, certification facilitates training. In section
6,
we
in the
discuss
further extensions of our results.
The Environment
2
The economy
are low ability,
At
i
and
=
0,
number
consists of a large
and the remaining
—p
1
no one knows a worker's
A
firms.
fraction are high ability.
fraction
There are two periods.
but other firms do not. This asymmetry of
ability,
information between the current employer and potential employers
monopsony power that
will lead to
Workers are not productive
output of a worker in period
where
/i
=
1 for
i
at
=
i
=
=
1 is
=
low ability workers, and
/i
worker.
upon.
o;
—
1
The effort
The
>
The
is
e
=
choice of the worker
analysis
If
is
unaffected
cost of training
is
or
=
>
1 for
high ability workers, r denotes
a worker receives no training, r
1
effort to benefit
where
e
=
1
is
=
0,
otherwise
We
denote
has a disutility cost of c for the
and cannot be contracted
and low
ability workers,
no discounting.
7 per worker and
t
=
from training.
costs differ across high
There
workers are unable to bear this cost at
77
his private information
is
if effort
the return to training.
a^^h
some
also needs to exert
the effort choice of the worker by
the source of the
but they can be trained during this period. The
0,
y
The worker
1.
is
firm-sponsored training in our economy.
whether the worker was offered training.
T
p of workers
but at the end of the period a worker himself
ability,
employer learn this
his current
and
of workers
0,
is
incurred by the firm.
so
all
We
assrune that
training has to be firm-sponsored.^
Moreover, since firms cannot distinguish between high and low ability workers at the
begirming of
i
—
0,
training has to be offered to
possible because at this point workers do not
at the beginning of
A
fraction
i
=
A of workers
1,
all
know
workers.
their
Self-selection
own
ability either.
workers decide whether to stay with the
will quit irrespective of the
is
initial
also not
Finally,
firm or not.
wages in the outside market. The
^See Acemoglu and Pischke (1999a, b) for a discussion of possible reasons for
unable to buy training from their employers.
why workers may be
remaining fraction
wage
offered
1
—A
of workers will stay in their current job
by the current firm and
;;
is
if
training depending
on the
rangements. In the
first,
fore, outside firms will
off.^
no
is
w
abilities
is
the
and whether
They may observe whether the workers
institutional arrangements.
there
where
v,
the wage offered by the outside market.^
Firms in the second hand labor market never observe workers'
they have quit or have been laid
w>
We
certification of skills
distinguish between
receive
two
ar-
by independent bodies. There-
not observe whether a worker has successfully completed a training
program. The second institutional setup, which
is
similar to the
system, involves an outside body that runs examinations and
German
certifies
apprenticeship
the successful com-
pletion of training programs.
The
outside labor market
make
their information, firms
beginning of time
^
=
is
W to ensure zero
wage
due to
is
competitive, so wages will be such that conditional on
zero
profits.*^
We
also
competitive, so firms
profits. Intuitively, if
assume that the labor market
may have
the firm will
ex post monopsony power, then competition at
its
this profit in the
form of upfront wages,
to
pay a
make
t
=
first
at the
period training
positive profits at
will force it to
t
=
1
pay out
W.
Equilibrium Without Certification
3
We
start
there
by showing that
no
is
in the absence of certification, there will
certification, all
a unique wage
To prove
v. It is
this,
workers in the outside labor market are
be no training. As
alike,
so there will be
straightforward to see that workers will not exert training effort.
suppose that firms offered training and workers exerted
effort.
In this
case the outside wage would be
,^
where
qi is
+ (l-p)g.r?
^
pqi + {l-p)qh
Pg/
(1)
the probability that a low ability worker separates from the current firm (due
to a quit or layoff)
and
qn
is
the probability that a high ability worker separates from
the firm. This equation ensures that firms in the outside market, which do not observe
worker ability or training, make zero profits in expectation. Since worker training
observed, a worker
W+v
(first
*This
^A
is
who
is
not
deviates and chooses not to exert effort will get utility equal to
period wage,
W, and
outside wage, v). Also notice that the
maximum wage
the formulation of the asymmetric information model used in Acemoglu and Pischke (1999b).
is vacuous in this simple model anyway, since the firm can always set wages so as
distinction that
to induce
all
^Formally,
(1998).
low ability workers to
we
quit.
are looking for a Perfect Bayesian Equilibrium of this game. See
Acemoglu and Pischke
.
that the current firm will pay in the second period
time;
if
the firm paid a higher wage,
necessarily
make
W+v—
most
is
w =
the outside wage at the
v,
would not retain any more workers, so would
This implies that the return to a worker exerting
less profits.
which
c,
it
is
strictly less
than the
utility
from not exerting any
effort is at
effort,
W+
v.
not exert
effort,
and so there
unable to reward workers for exerting
effort;
outside firms cannot
This implies that without certification workers
will
be
will
no training.
Intuitively, firms are
distinguish trained and untrained workers, so
all
workers will earn the same wage in the
This in turn means that the training firm will also pay a single wage
outside market.
and untrained workers.
to both trained
It
could promise ex-ante to pay a higher wage
Absent a possibility
to trained workers, but such a promise would not be credible.
committing to
this higher wage, the firm
We
reneging on this promise.
would always benefit
assume such a commitment
Also notice that there
to this issue in section 5 below.
in the
for
second period from
not possible here but retiirn
is
no
is
role for training firms to
engage in certification of training themselves. Since the firm benefits from paying a lower
wage,
it
has an incentive not to certify any workers as trained.
Equilibrium
rf)
—
now
is
Employers at
straightforward to characterize.
V to their high ability employees, and at most
w{h
=
1)
=
i
=
1 offer
all
low ability workers
will
=
to low ability workers.^
1
1,
and
where the
last
Since a fraction A of high ability workers quit, outside wages v will be greater than
therefore
w{h
indeed quit. Outside wages are
p+(l-p)A
Firms make second period
equality defines
11,
profits in period
4
which
=
^
0,
is
profits of IIo
=
(1
—
A) (1
—p)
(77
—
u)
=
11
a term that will reoccur throughout the paper. To ensure zero
firms will pay upfront wages VT
=
11.
Equilibrium with Certification
Now
consider the
same economy with an outside body
training. In this case outside firms
do not observe
certifying successful completion of
ability,
but they observe whether the
worker has successfully completed training. Notice that certification implies that the firm
has offered training and the worker has exerted
wages now, one
^When
t;„
offering a
strategy as a "layoff"
for
effort.
There
will
be two different outside
workers without certified training, the other
wage w{h
=
1)
=
1 firms realize
that workers will quit.
vt for
We
a worker with a
therefore refer to this
certificate of training.
With the same reasoning
+ {1 - p)qiv ^
pqj + (1 - p)ql
+ {l-p)qhV
pqf + (1 - p)ql
pq?
pqj
where g"
is
the probability that a low ability worker
because the firm did not
The
the firm.
Now
we have
as above
who
does not have training (either
because he did not exert
ofTer it or
.„.
effort)
has separated from
other g's are defined similarly.
consider an allocation (a candidate equilibrium) in which firms offer training to
all
their workers,
to
all
and
who
workers
workers exert
all
effort. Specifically,
the firm offers a wage of Wt
are high ability and have exerted effort, and lays off
all
= Vt
other workers.
In this case, the outside wage for workers with the certificate (who are necessarily trained)
will
be
Vt
The wages
in the
no
ability
certification case.
1.
the outside wage
firm
may want
A
(3)
-,
who have
for workers
produces
= P+{1-p)Xt]
r-r(^-
worker
not exerted any effort are going to be the
who has
same
as
not exerted effort and turns out to be low
Since some high ability workers will be in the outside market again,
is ?;„
>
1,
so the initial firm will not retain any low ability workers.
to keep high ability workers without training, by offering
if
the productivity of these workers
rj
>
77
is
greater than
Vn will always be the case, so the outside wage
?;„.
It is
them Wn
The
=
Vn,
straightforward to see that
is
p+(l-p)A
and a firm keeps some
of high
and low
of the high ability workers
ability
who have not exerted effort. The fraction
workers in the outside market are
and do not depend on behavior because workers make
their type.
This implies
Vt
still
the population fractions
their effort decision before
knowing
= aVn.
Next, we can determine whether workers prefer to exert effort during training.
will
do so
if Vt
>
Vn
+
c.
Substituting from (3) and
(4),
we
They
find that this inequality
is
equivalent to
that
is,
the rettirn to training should be sufficiently high, especially as compared to the
cost of effort,
c.
We assume this condition holds in the rest of the analysis. The presence
6
of training certificates
to their
hence
skills,
now
Is it
now
ensures that the outside market rewards the workers according
effort,
ensuring the correct incentives.
Consider the profits of a firm that
also profitable for firms to offer training?
offers training in this equilibrium,
=
lit
where
(1
-
was defined above.
11
-
p) {arj
In contrast,
if it
-
vt)
- 7 = an -
chooses not to train,
employees
will prefer to invest in the training of their
{a
a and
and
so,
77
are sufficiently large
and 7
is
As
before,
The main
—
> n„
and
be
Wt
=
all
—
therefore that, as long as conditions (5)
is
an equilibrium with training when the training
satisfied,
certification.
7.
and
(6)
results are externally
Given conditions
To
and
(6),
see this,
add
(5)
obtain
(a-l)(^;
+ n) >
^Wt + avt- c
where we have exploited the
hand
if
an equilibrium with firm-sponsored
will exist
but no training in the absence of
(6) to
or
- l)n > 7
welfare will be higher in the equilibrium with certification and training.
(5)
are
(6)
Q wages adjust to ensure zero profits, hence
result of this section
hold, there will be
certified,
t
if Ilf
sufficiently small, this condition will
with certification of training, there
training.
its profits
> 7
(a-l)(l-A)(l-p)(7y-i;„)
If
7,
= (i-A)(i-p)(77-?g = n
n„
Firms
A) (1
fact that
side of the second line of (7)
while the right hand side
is
is
>
W = U,
7
+c
(7)
W+v
Wt
=
all
—
7,
and a
>
1.
left
the utility of a worker in a training equilibrium,
the utility in a no training equilibrium. Since firms
profits, (7) establishes that welfare in
The
the training equilibrium
is
greater.
make
zero
Certification
is
therefore welfare enhancing.
Training Avith
5
Our
at
analysis above
time
i
=
1,
assrmiption for
Wage Commitments
was simplified by the assumption that firms
ruling out
many
commitments to wages.
firms, for large
Although
German companies,
it
set the post-training
this
may be a
wages
reasonable
might be more plausible to
assume that they can
vise their
reputation to commit to a wage path.^
It is
important to
investigate the role of this assumption since in the presence of such commitments, training
may be an equihbrium
even without
exploited the fact that the
maximum wage
worker had nothing to gain by exerting
without exerting
The
t
=
1,
Intuitively,
certification.
our analysis in section 3
the firm will pay at
effort as
t
=
1
is
w =
v,
so the
he could get the outside wage v even
effort.
alternative
is
a situation in which the firm commits to a higher wage in period
say Wc, and encourages workers to exert
however. Because
we do not have
effort.
worker
certification,
There
effort is
is
an important constraint,
not verifiable, and therefore
contracts cannot be contingent on a worker's effort choice. This implies that Wc has to
be high enough so that the expectation of getting
exert effort.
be
laid off
Only high
and receive
effort choice, so
v.
this higher
ability workers will receive
wage encourages workers to
Wc while low
Workers do not know their
ability
ability
workers will
when they have
to
make
they will base their decision on the expected gain from exerting
still
their
effort.
Since low ability workers will be laid off and a fraction A of high ability workers will quit
voluntarily, the
maximum
which needs to be
expected gain from exerting
at least as great as
lay off high ability workers,
run interest of the
who have
—
p) (1
—
A) {wc
—
v),
assume that the firm can commit to
effort,
even when
it is
not in the short
firm.^
all
workers exert
for retained
effort,
Vt
Therefore,
also
not exerted
Suppose the firm commits to a wage
This implies that
We will
c.
effort is (1
=
workers sufficient to encourage
effort.
hence
p+{l-p)Xr]
r-rCt
p + {l-p)X
;
we need
'"'=(l-p)(l-A)+'"
Without
certification, the firm will only follow the training strategy if profits
Het
=
(1
-
A) (1
- p)
{arj
-
w,)
from training
-j
we could imagine a repeated game similar to our static economy where firms are
but workers live only two periods. If workers can observe the past wage offers of firms,
there could exist a self-sustaining equilibrium where firms would be punished if they renege on their wage
promises. To keep the exposition simple, we do not discuss this more complicated model in this paper.
"In a repeated game framework, if the firm were to retain a worker who has not exerted effort, then all
workers in the future would prefer not to exert effort. Therefore, retaining such workers is incompatible
with a wage commitment strategy.
*More
formally,
infinitely lived,
are larger than the profits
of
commitment but no
<
Uct
IT
effort.
it is
unprofitable for firms to provide training
This condition together with
7< (aWhen
(5)
and
both
(8) are
a set of parameter values, defined by
certification, there exists
n, which ensures that
worker
from not training workers. Thus, even with the possibility
side certification.^*^ Thus,
and encourage
(6) implies
l)n <
satisfied, firms will
c
+ 7.
(8)
provide training only
when
there
wage commitments do not necessarily ensure equilibrium
sponsored training without
certification,
and
certification institutions
may
still
out-
is
firm-
be neces-
sary for high investments in training.
Concluding Remetrks
6
We
have kept the model above deliberately simple to clearly
may be
of our analysis: external certification of training
illustrate the
necessary to provide
incentives for workers to contribute their part to training investments
invest in worker training.
the qualitative results.
Many
choose to exert
siifficient
and ensure that firms
of the assumptions are very special but not essential to
It is useful at this
point to discuss briefly a few of the key ones.
The assumption that workers do not know their type ex-ante,
ble in this strict form, is not crucial. Our basic results still hold
their type ex-ante.
main point
while relatively implausiif
workers perfectly knew
In the case with commitment, however, low ability workers will not
effort
anymore, since they do not get a payoff from
More important,
it.
the firm will be able to induce high ability workers to exert effort with a lower wage com-
mitment. The range of parameters where wage commitments can ensure firm-sponsored
training will therefore be larger.
In the case of Germany, however, where workers enter
apprenticeships at a relatively young age, like 15 or 16,
good knowledge about
all
their
We have also assrmied that
who have
and
own
may be
unlikely that they have very
aptitudes and comparative advantages.
certification allows perfect discrimination
exerted effort during training and those
ability
it is
who have
not.
between workers
In practice, effort
more able workers may be able
substitutes in examinations, so that
to obtain the training certificate with less effort than low ability workers.
^°In fact, the condition for
commitment
to ensm'e training
is
This makes
more stringent than the one given
in
the text. Equihbrium commitment requires the discounted present value of a profit stream, Ilct/r, to be
greater than the alternative, where r is the discount rate. The firm will receive a larger profit now from
reneging on the promised wage w^ to trained workers now.
premium, but
also receive only profits 11
from then on.
This condition boils down to the one in the text when r
It will
pay them
Vt instead,
saving the wage
This means commitment requires
—
>
0.
Ilct
>
11
-I-
re.
putting in no effort more attractive for workers in the certification case, and tlierefore
more
leads to a
In
more
t
=
stringent condition than (5).
Acemoglu and
interesting
1 is
Pisclike (1998),
is
not completely exogenous.
Instead,
and
i.e.
a probability distribution function.
are possible.
One
equilibrium
lots of training, while
wage and
little
may be
we model the
for
a fraction
We
1
— F{w —
show that
It is
workers quits, where
in this setup multiple equilibria
characterized by few quits, low outside wages,
similarly possible to obtain multiple equilibria in the
an economy without such
and no training may not necessarily lead to a high training
initial
turnover rate of the economy
in this particular economy, even
highlights that certification
in
v) of
an economy
like
on
another equilibrium exhibits a high quit rate, a high outside
or no training.
example, the
model becomes
quit decision as depending
certification case. This implies that introducing certification in
credentials
selection
when the decision of workers to quit voluntarily at the beginning of period
the relative inside and outside wages,
F
we show that the adverse
is
though
it
is
equilibriiun.
too high, training
may
If,
not arise
could be sustained as an equilibrium.
This
only one institutional feature which helps support training
Germany.
10
References
[1]
Acemoglu, Daron and Jorn-Steffen Pischke (1998)
"Why do
firms train?
Theory and
evidence," Quarterly Journal of Economics 113, 79-119.
[2]
Acemoglu, Daron and Jorn-StefFen Pischke (1999a) "Beyond Becker: Training
perfect labor markets,"
[3]
Economic Journal Features
109,
in im-
February 1999, F112-F142.
Acemoglu, Daron and Jorn-Steffen Pischke (1999b) "The Structure of Wages an
In-
vestment in General Training" Journal of Political Economy, 107, June 1999, 539-572.
[4]
Burtless,
Gary (1994) "Meeting the
skill
demands
of the
new economy,"
Solomon and Alec R. Levenson Labor markets, employment
policy,
in:
Lewis C.
and job
creation.
Boulder, Co: Westview Press, 59-81.
[5]
Bishop, John H. (1996)
of the literature," in
"What we know about employer-provided
Solomon Polachek
(ed.)
training: a review
Research in Labor Economics,
vol. 16,
Greenwich, CT: JAI Press, 19-87.
[6]
Chang, Chun and Yijiang
Wang
The Pigovian
information:
(1996)
"Human
capital investment
under asymmetric
conjecture revisited," Journal of Labor Economics 14, 505-
519.
[7]
Heckman, James
and training
consensus"'
policy,
[8]
J.,
policy:
in:
and job
Rebecca
A
L. Roselius,
and
Jeffrey A.
Smith (1994) "U.S. education
re-evaluation of the underlying assumptions behind the 'new
Lewis C. Solomon and Alec R. Levenson Labor markets, employment
creation. Boulder, Co:
Westview Press, 83-121.
Katz, Eliakim and Adrian Ziderman (1990) "Investment in general training:
of information and labour mobility,"
Economic Journal
11
100, 1147-1158.
The
role
DEC
^nnnDate
Due
Lib-26-67
,MJT
LIBRARIES
3 9080 01917 6285
ip."^'
It:
t
It
I
6
r't"
7
i-fl
8
9
10
;
11
ll|l|l|l
l|l|M
_JoREGON RULE
CO.
l|l|l|M|l|l|l
1
U.S.A.
2
3
4