Job Assignments under Moral Hazard - School of Economics and ...
Job Assignments under Moral Hazard - School of Economics and ...
Job Assignments under Moral Hazard - School of Economics and ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Job</strong> <strong>Assignments</strong> <strong>under</strong> <strong>Moral</strong> <strong>Hazard</strong><br />
Alex<strong>and</strong>er K. Koch a <strong>and</strong> Julia Nafziger b∗<br />
a Aarhus University <strong>and</strong> IZA<br />
b Aarhus University<br />
November 29, 2010<br />
Abstract<br />
Inefficient job assignments are usually explained with incomplete information about<br />
employees’ abilities or contractual imperfections. We show that inefficient assign-<br />
ments arise even with full information about employees abilities <strong>and</strong> complete con-<br />
tracts. Building on this result <strong>and</strong> allowing effort levels to vary across hierarchy<br />
levels, we provide a new perspective on the Peter Principle: the output <strong>of</strong> an <strong>under</strong>-<br />
qualified employee need not decline post promotion because he is pushed to exert<br />
more effort. As a result, an employee moving up the career ladder may produce <strong>and</strong><br />
earn more only because he works harder <strong>and</strong> not because the job match is efficient.<br />
JEL Classification: D82, D86, J33<br />
Keywords: <strong>Moral</strong> <strong>Hazard</strong>; Information; <strong>Job</strong> <strong>Assignments</strong>; Peter Principle.<br />
∗a <strong>School</strong> <strong>of</strong> <strong>Economics</strong> <strong>and</strong> Management, Aarhus University, Building 1322, 8000 Aarhus C, Den-<br />
mark, Email: akoch@econ.au.dk, Tel.: +45 8942 2137. b <strong>School</strong> <strong>of</strong> <strong>Economics</strong> <strong>and</strong> Management, Aarhus<br />
University, Building 1322, 8000 Aarhus C, Denmark, Email: jnafziger@econ.au.dk, Tel.: +45 8942<br />
2048. We thank Mathias Dewatripont, Guido Friebel, Paul Heidhues, Ian Jewitt, Thomas Kittsteiner,<br />
Matthias Kräkel, Johan Lagerlöf, Meg Meyer, David Myatt, Urs Schweizer, <strong>and</strong> John Sutton for helpful<br />
comments. Nafziger is grateful to Nuffield College, University <strong>of</strong> Oxford for its kind hospitality.<br />
1
1 Introduction<br />
Organizations face the problem <strong>of</strong> how to match individuals with the job in which they<br />
are most productive <strong>and</strong> to provide them with incentives. In practice one observes<br />
that job assignments are <strong>of</strong>ten used to motivate employees – leading to an inefficient<br />
allocation <strong>of</strong> employees to jobs. For example, people <strong>of</strong>ten are promoted to a job for<br />
which they are less competent <strong>and</strong> then remain there, a phenomenon known as the Peter<br />
Principle (Peter <strong>and</strong> Hull 1969). The appearance <strong>of</strong> such inefficiencies seems puzzling<br />
to economists. Why do firms not separate job assignment policies from the provision <strong>of</strong><br />
incentives, leaving the latter to pay-for-performance schemes?<br />
The literature on careers in organizations stresses the role <strong>of</strong> incomplete information<br />
about employees’ abilities <strong>and</strong> contractual imperfections in accounting for inefficiencies<br />
in job assignments (see the section on related literature). Our approach is complementary<br />
to this. We start from a moral hazard model <strong>and</strong> show that job assignment decisions<br />
cannot be separated from incentive provision even if the principal is able to write a<br />
complete contract <strong>and</strong> knows about the agent’s productivity.<br />
The driving force behind inefficient job assignments in our model is a static trade-<strong>of</strong>f<br />
between incentive provision <strong>and</strong> efficient allocation <strong>of</strong> agents to jobs. We develop this<br />
trade-<strong>of</strong>f in a moral hazard model, where a risk neutral principal assigns a risk neutral<br />
<strong>and</strong> wealth constrained agent to one <strong>of</strong> two jobs (a high ability <strong>and</strong> a low ability one).<br />
The probability with which the agent succeeds in producing a high output depends in<br />
each job on his ability, <strong>and</strong> his effort. While the principal knows the ability <strong>of</strong> the agent,<br />
she cannot observe his effort <strong>and</strong> thus designs an output-contingent reward scheme to<br />
induce a certain effort. The cost <strong>of</strong> getting the agent to exert this effort level in a given<br />
job depends on how likely it is that success in this job was generated by high <strong>and</strong> not<br />
by low effort.<br />
Because the job in which an agent <strong>of</strong> a certain ability produces the highest expected<br />
output need not be the one in which it is cheapest for the principal to implement effort,<br />
a trade-<strong>of</strong>f between efficient job assignment <strong>and</strong> incentive provision arises – leading to an<br />
inefficient assignment policy. 1 The necessary <strong>and</strong> sufficient condition that we derive for<br />
such an inefficient outcome to occur depends on the relative informativeness <strong>of</strong> output<br />
about effort in the two jobs, i.e., the ranking <strong>of</strong> likelihood ratios across jobs.<br />
The rather technical condition becomes more intuitive once we interpret the high ability<br />
job as a position higher up in the hierarchy than the low ability job. <strong>Job</strong>s further down<br />
1 One might suspect that the root <strong>of</strong> inefficiencies in job assignments are distortions in implemented<br />
effort levels that arise in moral hazard problems. Our binary effort model shows that this is not the<br />
case. It allows us to consider the job assignment problem for a parameter range where the principal does<br />
not distort effort <strong>and</strong> implements the efficient level in both jobs. Our continuous effort model combines<br />
the effects <strong>of</strong> assignments <strong>and</strong> distortions in implemented effort levels.<br />
2
in the hierarchy <strong>of</strong>ten put low dem<strong>and</strong>s in terms <strong>of</strong> the inputs ability <strong>and</strong> effort. Hence,<br />
in the low ability job success might be quite likely even without hard work (i.e., it is is<br />
a relatively easy task). In the high ability job, if the agent does not work hard, there<br />
is a bigger drop in expected output (i.e., it is a more difficult task for him). Seeing<br />
such an agent succeed in what is for him a difficult job, thus, indicates that he must<br />
have worked hard. So the high ability job poses a harder test <strong>of</strong> the agent’s effort than<br />
the low ability job, <strong>and</strong> this makes it less costly to get the agent to exert effort. That<br />
output in the higher level job is <strong>of</strong>ten more informative about the agent’s effort than<br />
output in the lower level job is a view also found in the literature (Baker, Gibbons,<br />
<strong>and</strong> Murphy 1994, Maskin, Qian, <strong>and</strong> Xu 2000, Ortega 2003). As we show, the reduced<br />
cost <strong>of</strong> incentives (due to differences in informativeness about effort across hierarchy<br />
levels) outweighs the negative effect <strong>of</strong> a lower success probability near the ability cut<strong>of</strong>f<br />
level for efficient job assignment. Hence, the principal sets a lower ability threshold for<br />
assignment to the more informative higher hierarchy level than is efficient.<br />
The mechanism at work is a static one: the job assignment decision in one period affects<br />
the contemporaneous cost <strong>of</strong> providing incentives. This static mechanism on its own<br />
<strong>of</strong>fers several new insights into the driving forces governing promotion decisions in firms.<br />
Embedding our model in a stylized dynamic framework <strong>and</strong> extending it to allow for<br />
continuous effort, we show that it may be optimal for a firm to promote an employee to<br />
a higher level job even though it knows that he will be less productive than he could be<br />
in his previously held position. This finding provides a new perspective on possible roots<br />
<strong>of</strong> the Peter Principle: even in ideal circumstances where a firm is perfectly aware <strong>of</strong> its<br />
employees’ abilities one may find employees in jobs for which they are not well suited.<br />
Our analysis highlights that promoting an employee to a job for which he is less com-<br />
petent need not lead to an observed decline in output – a narrow definition <strong>of</strong> the Peter<br />
Principle <strong>of</strong>ten used in the literature (e.g. Lazear 2004). Output may actually rise be-<br />
cause the promoted agent works harder – hiding the fact that he actually is untalented<br />
for his new job. Existing theoretical treatments <strong>of</strong> promotions in firms have ignored<br />
this effect <strong>of</strong> job assignments on implemented effort levels: they either abstract from<br />
incentives altogether or assume that effort is constant across hierarchy levels. This effect<br />
may, however, be an important driving force for promotion decisions, as our analysis<br />
shows. We therefore suggest an extended definition <strong>of</strong> the Peter Principle that includes<br />
both the possibility that a promoted agent works harder or less hard than he would if<br />
he were reassigned to the previous job. Our setting with varying effort levels across hi-<br />
erarchical levels allows to show that an employee moving up the career ladder may earn<br />
more <strong>and</strong> produce more only because he has to work harder than he would in his old job.<br />
The paper is organized as follows. Next we discuss the related literature. Section 2<br />
3
introduces the static job assignment model with discrete effort. Section 3 extends the<br />
model to a continuous effort setting to analyze promotions <strong>and</strong> explore the relation to<br />
the Peter Principle. The final section discusses our findings <strong>and</strong> concludes. All pro<strong>of</strong>s<br />
are in the appendix.<br />
Related Literature<br />
The large literature on careers <strong>and</strong> incentives in organizations emphasizes two roles <strong>of</strong><br />
job assignment <strong>and</strong> promotion decisions. First, learning about the abilities <strong>of</strong> employees<br />
drives the allocation policy. Firms initially place individuals in jobs where they can do<br />
little harm. Those who prove to be able are then promoted to positions that require a<br />
higher level <strong>of</strong> ability. Second, promotions are a source <strong>of</strong> incentives. 2<br />
This literature identifies several reasons for distortions in job assignments. The principal<br />
may create distortions to influence the information about employees’ abilities she herself<br />
(e.g. Meyer 1991) or outsiders receive (e.g. Waldman 1984a, Bernhardt 1995), or what is<br />
revealed to an employee about his own ability (Ishida 2006, Nafziger 2008). Restrictions<br />
on incentive contracts may also lead to distortions. For example, a Lazear <strong>and</strong> Rosen<br />
(1981)-style promotion tournament may select the agent with lower expected ability<br />
when agents have private information about their type (Clark <strong>and</strong> Riis 2001). If job<br />
assignment decisions need to be delegated, distortions may arise because <strong>of</strong> the additional<br />
agency layer (Carmichael 1988, Baliga <strong>and</strong> Sjöström 2001, Fairburn <strong>and</strong> Malcomson 2001,<br />
Carrillo 2003, Friebel <strong>and</strong> Raith 2004).<br />
When asking about the reasons for distortions in job assignments, the literature tries to<br />
address two puzzles. First, why can the role <strong>of</strong> matching individuals with jobs <strong>and</strong> that<br />
<strong>of</strong> providing incentives not be separated; leaving the former to job assignment decisions<br />
<strong>and</strong> the latter to pay-for-performance schemes (Baker et al. 1988)? Explanations for this<br />
are based on contractual imperfections: for example, performance is not verifiable (e.g.<br />
Malcomson 1984), or some aspects <strong>of</strong> performance – such as human capital accumulation<br />
– are not contractible (e.g. Prendergast 1993, Kwon 2006).<br />
Second, even though employees <strong>of</strong>ten turn out to be less productive in their job than<br />
in their former position, why are they not demoted? Lazear’s (2004) model, based on<br />
temporary shocks to productivity, <strong>of</strong>fers an explanation for one interpretation <strong>of</strong> the so-<br />
called Peter Principle. A promotion means that a worker has delivered a high measurable<br />
output, which is the sum <strong>of</strong> permanent <strong>and</strong> transitory components. Because <strong>of</strong> regression<br />
to the mean <strong>of</strong> the temporary productivity shocks, output declines on average after a<br />
2 For an excellent survey <strong>of</strong> the literature see Gibbons <strong>and</strong> Waldman (1999a). Examples for the first<br />
str<strong>and</strong> are: Prescott <strong>and</strong> Visscher (1980), O’Flaherty <strong>and</strong> Siow (1992), <strong>and</strong> Valsecchi (2003); examples<br />
for the second str<strong>and</strong> are: Lazear <strong>and</strong> Rosen (1981), Malcomson (1984), Rosen (1986), <strong>and</strong> MacLeod<br />
<strong>and</strong> Malcomson (1988).<br />
4
promotion. 3 Bernhardt (1995) formalizes an idea from Peter <strong>and</strong> Hull (1969): promotion<br />
is based on the performance in one task (say as a laborer), but the skills required in the<br />
new task (say as a manager) may be quite different. While on average a more able<br />
laborer is also likely to be a better manager, such a laborer may turn out to be a<br />
failure as a manager once promoted. If there is asymmetric information about employee<br />
abilities, the employer may, however, refrain from demotions because she has to pay a<br />
bonus to retain a demoted worker (Bernhardt 1995). Fairburn <strong>and</strong> Malcomson (2001)<br />
include incentive considerations, <strong>and</strong> assume that workers can sway the performance<br />
evaluation <strong>of</strong> their supervisors with bribes. This makes incentive pay ineffective for<br />
the principal: supervisors <strong>and</strong> their workers would simply collude to extract high wages<br />
without any effort in return. Making workers’ pay contingent on their job, <strong>and</strong> managers’<br />
pay contingent on the firm’s pr<strong>of</strong>its, aligns managers’ interests more closely with those <strong>of</strong><br />
the firm’s owners: promoting a very untalented worker reduces pr<strong>of</strong>its, <strong>and</strong> thus reduces<br />
the supervisor’s pay by more than what the supervisor could gain by extracting a bribe.<br />
Still, some workers close to the efficient promotion threshold are promoted even though<br />
they should not be.<br />
Our approach is complementary to this. We start from assumptions that are most<br />
favorable to avoiding inefficiencies in job assignment decisions: there is no uncertainty<br />
about the agent’s ability <strong>and</strong> complete contracts are possible. We show that nevertheless<br />
a principal would not adopt an efficient job assignment policy. Our results contribute to<br />
<strong>under</strong>st<strong>and</strong>ing both puzzles referred to earlier: job assignment <strong>and</strong> incentive provision<br />
need not be separable even with complete contracts; <strong>and</strong> inefficient promotion decisions<br />
(as reflected in the Peter Principle) may result from an optimal policy <strong>of</strong> the principal.<br />
From a theoretical point <strong>of</strong> view, our model is most closely related to Robbins <strong>and</strong><br />
Sarath (1998). They show in a setting with a risk averse agent that the principal may<br />
not always want to choose from among two output systems the one that generates the<br />
highest revenue. Different output systems vary in how informative they are about the<br />
agent’s effort. While a more informative system corresponds to lower implementation<br />
costs, it need not be the one that creates a higher revenue. Output systems in their<br />
paper correspond in our setting to agents with different abilities in the two jobs. Our<br />
model with a risk neutral agent who is protected by limited liability allows us to look at<br />
a continuum <strong>of</strong> output systems. This enables us to show how inefficient job assignments<br />
may emerge even in a complete contracting setting, <strong>and</strong> how this can help <strong>under</strong>st<strong>and</strong><br />
the Peter Principle.<br />
3 This mechanism has recently been tested using personnel records (Barmby, Eberth, <strong>and</strong> Ma 2007)<br />
<strong>and</strong> lab experiments (Dickinson <strong>and</strong> Villeval 2007).<br />
5
2 The static job assignment model<br />
2.1 The model<br />
A risk neutral principal (‘she’) needs to assign an agent (‘he’) to one <strong>of</strong> two jobs, j ∈<br />
{l, h}. He is protected by limited liability <strong>and</strong> has a reservation utility <strong>of</strong> zero. In each<br />
job he produces a verifiable <strong>and</strong> observable output. Output can either be high (revenue<br />
for the principal normalized to 1) – or low (revenue normalized to 0). The probability<br />
<strong>of</strong> high output, fj(e, θ), depends on the job j ∈ {l, h}, the agent’s ability θ ∈ [θ, ¯ θ], <strong>and</strong><br />
his effort e ∈ {e, ē}. We assume that f(·, θ) is twice continuously differentiable in θ, <strong>and</strong><br />
that higher effort leads to a higher success probability, i.e., fj(ē, θ) > fj(e, θ) ∀ θ. The<br />
utility cost <strong>of</strong> effort is c if the agent provides high effort, <strong>and</strong> zero otherwise.<br />
<strong>Job</strong> l entails ‘safe’ tasks, where a low ability agent can do no harm because the success<br />
probability is insensitive to ability. Hence, we refer to it as the low ability job hereafter<br />
<strong>and</strong> simplify notation to fl(e) ≡ fl(e, θ). In the high ability job h, expected output<br />
increases with higher ability, <strong>and</strong> the impact <strong>of</strong> ability on output also (weakly) increases<br />
with effort (i.e., ability <strong>and</strong> effort are weak complements). The following sensitivity to<br />
ability (SA) assumption captures this: 4<br />
0 = fl θ(e, θ) = fl θ(ē, θ) < fh θ(e, θ) ≤ fh θ(ē, θ). (SA)<br />
Additionally, we assume that for a given effort, an agent with the highest ability level<br />
( ¯ θ) is most productive in job h, <strong>and</strong> an agent with the lowest ability level (θ) is most<br />
productive in job l. The following crossing property (SP) expresses this formally:<br />
fh(e, θ) < fl(e, θ) <strong>and</strong> fl(e, ¯ θ) < fh(e, ¯ θ) for e ∈ {e, ē}. (SP)<br />
Together with Condition (SA) this implies that for each pair <strong>of</strong> equal effort levels in the<br />
two jobs (e, e) ∈ {(e, e), (ē, ē)} there exists a unique interior threshold ˆ θ(e, e) such that<br />
all types with θ ≥ ˆ θ(e, e) have a higher success probability in job h than in job l, <strong>and</strong> all<br />
types with θ < ˆ θ(e, e) have a lower one. 5 Figure 1 illustrates this general property with<br />
a linear example. It also shows that the success probability functions for the two jobs<br />
need not cross for unequal effort levels: in the example with θ = 0, there is a crossing<br />
point ˆ θ(ē, e) but no crossing point ˆ θ(e, ē) in the range θ ∈ [θ, ¯ θ].<br />
The timing is as follows. First, the principal <strong>of</strong>fers a contract, specifying the job j ∈ {l, h}<br />
<strong>and</strong> an output-contingent wage scheme. Both the principal <strong>and</strong> the agent know the<br />
4 The subscript θ denotes the partial derivative with respect to θ.<br />
5 On its own, Condition (SP) guarantees that at least one such threshold exists (by the intermediate<br />
value theorem). Condition (SA) gives a unique threshold: it implies that fh(e,θ)<br />
fl(e,θ)<br />
higher ability the agent becomes relatively more productive in job h.<br />
6<br />
is increasing in θ – with
Figure 1: Illustration <strong>of</strong> the single crossing property (SP) with a linear example<br />
[fh(e, θ) = qh(e) + θ, qh(ē) = 0.4, qh(e) = 0; fl(e) = ql(e), ql(ē) = 0.7, ql(e) = 0.35]<br />
agent’s ability θ ∈ [θ, ¯ θ]. The contract fixes a wage ¯wj(θ) after a high output <strong>and</strong> w j(θ)<br />
after a low one. If the agent accepts the contract, he then provides unobservable effort<br />
e ∈ {e, ē}. At the last stage, output <strong>and</strong> pay<strong>of</strong>fs are realized: the principal receives the<br />
revenue net <strong>of</strong> the wage to the agent; the agent receives utility equal to the wage less<br />
his effort cost. If the agent rejects the contract, the game ends with zero pay<strong>of</strong>fs for all<br />
players.<br />
2.2 Observable effort (first best)<br />
In the benchmark case where the principal can observe the agent’s effort, she simply<br />
compensates him for his effort cost. That is, given job j <strong>and</strong> type θ her expected pr<strong>of</strong>it<br />
F B<br />
F B<br />
is either Πj (ē, θ) = fj(ē, θ) − c for high effort, or Πj (e, θ) = fj(e, θ) for low effort. To<br />
make the problem interesting, we assume that first-best pr<strong>of</strong>its for high effort are larger<br />
than the ones for low effort:<br />
Assumption 1 High effort is efficient in jobs j = l, h: fj(ē, θ) − fj(e, θ) ≥ c ∀θ ∈ [θ, ¯ θ].<br />
The optimal job assignment rule is that the principal places an agent with type θ in job<br />
h if <strong>and</strong> only if the maximized pr<strong>of</strong>it in job h is higher than in job l, which is equivalent<br />
to fh(ē, θ) ≥ fl(ē). So, unsurprisingly, the first-best job assignment rule places the agent<br />
in the job where he is most productive: job h if θ ≥ θ F B = ˆ θ(ē, ē) <strong>and</strong> job l otherwise.<br />
7
2.3 Unobservable effort (second best)<br />
As the principal cannot observe the effort choice <strong>of</strong> the agent, the problem she faces is<br />
not only to assign the agent to a job but also to design an appropriate incentive scheme<br />
that makes it optimal for the agent to provide the effort she desires. To solve this, we<br />
apply for each job separately the usual two-step procedure for moral hazard problems.<br />
First, for a given effort level e, we find the cost minimizing wage scheme that implements<br />
e in job j. Second, given these wage schemes, we maximize pr<strong>of</strong>its with respect to effort.<br />
In a third step, we then determine the pr<strong>of</strong>it maximizing job assignment rule.<br />
2.3.1 The incentive scheme<br />
Consider the first part <strong>of</strong> the problem. If the principal wants to implement high effort,<br />
she minimizes the expected wage bill<br />
fj(ē, θ) ¯wj(θ) + [1 − fj(ē, θ)] w j(θ), (1)<br />
taking into account that w j(θ), ¯wj(θ) ≥ 0 (the limited liability constraint), that the<br />
agent’s expected utility must be larger than his reservation utility <strong>of</strong> zero (the participa-<br />
tion constraint), <strong>and</strong> that it must be optimal for him to provide high effort. The latter<br />
is the case if the following incentive constraint is satisfied:<br />
fj(ē, θ) − fj(e, θ) ¯wj(θ) − w j(θ) ≥ c. (2)<br />
The limited liability <strong>and</strong> incentive constraint together imply that the participation con-<br />
straint is satisfied. But then, as a positive wage after a failure would only decrease<br />
incentives, the principal sets this wage equal to zero. The wage after a success is set to<br />
make the incentive constraint bind, i.e., ¯wj(θ) =<br />
c . The principal can get the<br />
fj(ē,θ)−fj(e,θ)<br />
agent to work hard only by leaving him a rent in addition to the compensation for his<br />
effort cost c.<br />
The expected pr<strong>of</strong>it from assigning an agent with type θ to job j <strong>and</strong> implementing high<br />
effort is Π SB<br />
j (ē, θ) = fj(ē, θ) − Cj(ē, θ), where<br />
Cj(ē, θ) =<br />
fj(ē, θ)<br />
c (3)<br />
fj(ē, θ) − fj(e, θ)<br />
is the expected wage payment, referred to in the following as the implementation cost.<br />
To implement e the principal sets w j(θ) = ¯wj(θ) = 0, resulting in expected pr<strong>of</strong>it<br />
Π SB<br />
j (e, θ) = fj(e, θ).<br />
Everything is in place now to consider the second part <strong>of</strong> the principal’s problem, what<br />
effort level to implement. Facing an agent <strong>of</strong> ability θ, it is optimal for the principal to<br />
implement high effort in job j if <strong>and</strong> only if Π SB<br />
j (ē, θ) ≥ Π SB<br />
j (e, θ), or:<br />
[fj(ē, θ) − fj(e, θ)] 2 ≥ fj(ē, θ) c. (Condition 1)<br />
8
If this fails to hold the principal distorts downward the implemented effort to save on<br />
implementation costs.<br />
2.3.2 <strong>Job</strong> assignments given high effort in both jobs<br />
In this section, we characterize the job assignment rule that the principal adopts if she<br />
always implements high effort (i.e., if Condition 1 holds for both jobs <strong>and</strong> for all worker<br />
types). This way, should we find that the job assignment rule <strong>under</strong> moral hazard is<br />
inefficient, we can be sure that this result is not an artefact <strong>of</strong> inefficient effort choices:<br />
we are looking at the parameter range where the second-best effort level is efficient.<br />
The second-best job assignment rule can be described by ability level(s) θ SB for which<br />
Π SB<br />
h<br />
ē, θ SB − Π SB<br />
l<br />
(ē) = 0 holds. To characterize these threshold(s) further, we first<br />
have to look more closely at the behavior <strong>of</strong> the pr<strong>of</strong>it functions (<strong>and</strong> their difference) in<br />
θ. While for job l this is clear – pr<strong>of</strong>its do not depend on θ – there are two potentially<br />
opposing effects in job h. On the one h<strong>and</strong>, a higher ability level increases the expected<br />
output. On the other h<strong>and</strong>, a higher ability level may have a negative effect on imple-<br />
mentation costs. 6 The following lemma shows that the first effect dominates, <strong>and</strong> thus<br />
pr<strong>of</strong>its in job h <strong>and</strong> the difference in pr<strong>of</strong>its across jobs increase with the agent’s ability:<br />
Lemma 1 Suppose Condition 1 holds for θ ∈ [θ, ¯ θ], then<br />
(a) d<br />
d θ ΠSB<br />
h (ē, θ) > 0,<br />
(b) d<br />
d θ<br />
Π SB<br />
h (ē, θ) − Π SB<br />
l (e) > 0 for e ∈ {e, ē}.<br />
From Lemma 1 it follows that there is a unique threshold θ SB . Does θ SB equal the<br />
first-best threshold θ F B = ˆ θ(ē, ē)? This relies on whether the second-best pr<strong>of</strong>its in<br />
the two jobs, Π SB<br />
h (ē, ˆ θ(ē, ē)) <strong>and</strong> Π SB<br />
l (ē), are equal at ˆ θ(ē, ē) or not. At ability level<br />
θ = ˆ θ(ē, ē) the job assignment does not change the expected output, because the agent<br />
is equally productive in both jobs if he works hard. Hence, the pr<strong>of</strong>it difference is<br />
simply the difference in implementation costs, i.e., Π SB<br />
h (ē, ˆ θ(ē, ē)) = Π SB<br />
l (ē) if <strong>and</strong> only<br />
if Ch(ē, ˆ θ(ē, ē)) = Cl(ē).<br />
So the question is when these implementation costs differ. Without loss <strong>of</strong> generality,<br />
let us consider the conditions <strong>under</strong> which Ch(ē, ˆ θ(ē, ē)) < Cl(ē) arises. Using the imple-<br />
mentation cost function given in Equation (3), it follows that the implementation costs<br />
are lower when assigning an agent with ability θ = ˆ θ(ē, ē) to job h rather than to job l<br />
if <strong>and</strong> only if<br />
fh(ē, ˆ θ(ē, ē))<br />
fh(e, ˆ θ(ē, ē))<br />
fl(ē)<br />
> . (4)<br />
fl(e)<br />
6 As can be seen from the pro<strong>of</strong> <strong>of</strong> Lemma 1, Ch θ(e, θ) ≷ 0 depending on whether fh(ē,θ)<br />
fh(e,θ) ≷ fh θ(ē,θ)<br />
fh θ(e,θ) .<br />
9
This boils down to fh(e, ˆ θ(ē, ē)) < fl(e), at ˆ θ(ē, ē) because fh(ē, ˆ θ(ē, ē)) = fl(ē). 7<br />
Each ratio reflects how likely it is after observing high output in a particular job that<br />
the agent provided effort ē instead <strong>of</strong> e. According to Inequality (4), it is more likely<br />
in job h than l that high output <strong>of</strong> an agent with ability θ = ˆ θ(ē, ē) was the result <strong>of</strong><br />
hard work. Thus, if this agent wanted to shirk, he would have a harder time hiding this<br />
in job h than in the low ability job l. In that sense, the principal can put the agent to<br />
a harder test by placing him in the more dem<strong>and</strong>ing job h: success in the high ability<br />
job is a sign that he must have worked hard. And this makes it is less costly to provide<br />
incentives in job h. 8<br />
As Demougin <strong>and</strong> Fluet (1998) argue, such a comparison <strong>of</strong> likelihood ratios in the high<br />
output state is an appropriate criterion for ranking different information systems for risk<br />
neutral agents that are protected by limited liability. And thus – because an information<br />
system corresponds here to an output-type-job combination – Inequality (4) translates<br />
to job h being more informative about effort than job l for an agent with ability θ. 9 So<br />
our model shows that the principal may place an agent into the job with the highest<br />
level <strong>of</strong> “transparency” about effort rather than into the job with highest productivity<br />
<strong>of</strong> effort.<br />
Proposition 1 Suppose the second-best effort levels are efficient (i.e., Condition 1 holds<br />
for jobs j = l, h <strong>and</strong> θ ∈ [θ, ¯ θ]). Then θ SB = θ F B if <strong>and</strong> only if fh(e, ˆ θ(ē, ē)) = fl(e),<br />
where θ SB <strong>and</strong> θ F B = ˆ θ(ē, ē) are the thresholds for assignment to the high ability job h<br />
in the principal’s second- <strong>and</strong> first-best job assignment rules, respectively.<br />
Proposition 1 shows that the roots <strong>of</strong> inefficient job assignments may <strong>of</strong>ten lie deeper<br />
than incomplete information about employees’ abilities <strong>and</strong> contractual imperfections:<br />
we eliminate these ingredients <strong>and</strong> show that the job assignment threshold is nevertheless<br />
inefficient, except for the degenerate case where fh(e, ˆ θ(ē, ē)) = fl(e). Perhaps less<br />
surprisingly, in settings where distortions in effort levels do occur, these additionally<br />
distort the job assignment rule, as shown in Appendix A.<br />
7 If Inequality (4) held for all θ, output in job h would be globally more informative than output in<br />
job l. For our purposes it is, however, not necessary to have such a global ranking <strong>of</strong> jobs in terms <strong>of</strong><br />
their likelihood ratios for all ability levels θ ∈ [θ, ¯ θ]. Whether the second-best job assignment rule is<br />
biased depends only on a much weaker local condition: are the likelihood ratios equal to each other or<br />
not at θ = ˆ θ(ē, ē)?<br />
8 While the principal seems to design the reward scheme as if he needed to estimate the effort put in<br />
by the agent, he <strong>of</strong> course knows the agent’s effort in equilibrium.<br />
9 Note that this definition <strong>of</strong> informativeness differs slightly from those proposed for the case <strong>of</strong> risk<br />
averse agents with unlimited liability, where the whole likelihood ratio distribution matters (Grossman<br />
<strong>and</strong> Hart 1983, Kim 1995, Jewitt 2006). The difference stems from the fact that for risk neutral agents<br />
rewards are concentrated in one state. Because the wage after low output is zero, there is no need to<br />
‘estimate’ the agent’s effort in the failure state.<br />
10
2.4 Discussion: Biases in hierarchical job assignments<br />
Proposition 1 tells us that we can expect an inefficient job assignment threshold to be<br />
the norm. The direction <strong>of</strong> the bias depends on which job is (locally) more informative<br />
about effort for an agent who is equally productive in both jobs. Two cases therefore<br />
may arise:<br />
(i) If job h is the (locally) more informative job, employees with θ ∈ (θ SB , θ F B ) are<br />
assigned to job h for which they are “<strong>under</strong>qualified”.<br />
(ii) If job l is the (locally) more informative job, employees with θ ∈ (θ F B , θ SB ) are<br />
passed over for assignment to job h, even though they are “overqualified” for the<br />
job l in which they are placed.<br />
The issue <strong>of</strong> <strong>under</strong>qualification has received considerable attention in the literature on<br />
careers in organizations. So <strong>under</strong> what circumstances it is reasonable to apply our<br />
model to the <strong>under</strong>qualification case? 10 One channel through which higher level jobs may<br />
become more informative about effort is that higher level jobs may be more sensitive to<br />
effort than lower level ones. Lower level jobs <strong>of</strong>ten entail ‘safe’ tasks that have relatively<br />
low dem<strong>and</strong>s in terms <strong>of</strong> worker inputs ability <strong>and</strong> effort. Because workers cannot do<br />
much harm here, firms assign to such jobs individuals with low expected ability (e.g.<br />
Calvo <strong>and</strong> Wellisz 1979, Rosen 1982, Waldman 1984b). In contrast, higher level jobs<br />
tend to be more sensitive to workers’ inputs, <strong>and</strong> firms therefore assign to such positions<br />
only workers with sufficiently high ability.<br />
In our model, Condition (SA) captures the notion that job h is more sensitive to ability<br />
than job l. In the context <strong>of</strong> a hierarchy, job h would hence be at a higher level than<br />
job l. The property that higher level jobs are more sensitive to effort can be captured<br />
in our setting by fh(ē, θ) − fh(e, θ) > fl(ē) − fl(e). This says that high effort in job h<br />
raises the success probability <strong>of</strong> the project more than in job l. This sensitivity to effort<br />
condition implies that<br />
fh(e, ˆ θ(ē, ē)) < fl(e), (Condition 2)<br />
because fh(ē, ˆ θ(ē, ē)) = fl(ē). If this condition is met, a consequence <strong>of</strong> our earlier<br />
analysis is that the principal sets the threshold for assignment to job h too low: θ SB <<br />
θ F B . In other words, the principal biases the job assignment threshold in favor <strong>of</strong> the<br />
higher level job h.<br />
Condition 2 has an intuitive interpretation: An agent who is almost equally productive<br />
in both hierarchy levels l <strong>and</strong> h if he works hard (i.e., with ability level close to the<br />
10 That jobs further up in the hierarchy are <strong>of</strong>ten more informative about an employee’s inputs is a<br />
view that can also be found in the literature (Baker, Gibbons, <strong>and</strong> Murphy 1994, Maskin, Qian, <strong>and</strong><br />
Xu 2000, Ortega 2003).<br />
11
efficient threshold ˆ θ(ē, ē)) perceives the higher ability job h as more difficult. So the<br />
principal can put the agent to a harder test by placing him in the more dem<strong>and</strong>ing job<br />
h. Hence, it is less costly to provide incentives than in the lower level job l <strong>and</strong> the<br />
principal distorts the assignment in favor <strong>of</strong> job h.<br />
Another channel through which higher level jobs may become more informative about<br />
effort is that more accurate performance measures are available, a notion for which<br />
Baker et al. (1994, p. 1127) provide a succinct illustration: “[. . . ] market-adjusted<br />
stock-price performance may be a useful measure <strong>of</strong> a CEO’s contribution but typically<br />
is an extremely noisy measure <strong>of</strong> the contributions <strong>of</strong> lower-level employees.” With less<br />
noise the performance measures become more informative about effort. This again would<br />
suggest that assignments are distorted in favor <strong>of</strong> the more informative higher level job.<br />
3 Implications for promotions<br />
3.1 Extending the model<br />
In the following we give our model a simple dynamic interpretation <strong>and</strong> extend it to<br />
continuous effort to derive implications for promotion decisions.<br />
3.1.1 A “dynamic” interpretation <strong>of</strong> the model<br />
Assignment to a higher level job is in many cases tantamount to a promotion. We saw<br />
in the previous section that the principal’s (static) job assignment decision interacts<br />
with her problem <strong>of</strong> providing incentives at lowest cost. As the literature on careers in<br />
organizations has not looked at this effect, we now ask what novel insights the trade-<br />
<strong>of</strong>f can deliver for <strong>under</strong>st<strong>and</strong>ing promotion policies. Our purpose is to stay as close<br />
as possible to the static job assignment model from Section 2 <strong>and</strong> isolate these novel<br />
insights by giving our model a simple dynamic interpretation. For this we add a first<br />
period <strong>and</strong> the two features described in the following.<br />
Workers do not start their careers at the top <strong>of</strong> the hierarchy: job l is the entry level job in<br />
the two-period model (Feature 1). Distortions in the job assignment/promotion decision<br />
arise in our model because the principal wants to change the agent’s incentives within<br />
the second period; first-period incentives therefore do not drive inefficiency results. This<br />
point can be made most clearly in a setting where the agent simply earns a fixed wage in<br />
the first work period, during which the principal learns about the agent’s ability as, for<br />
example, in Waldman (1984a) (Feature 2). With these features the promotion decision<br />
in the second work period becomes a static job assignment problem.<br />
12
3.1.2 The continuous effort model<br />
Our setup links promotion decisions to the cost <strong>of</strong> providing incentives: The principal<br />
can use a promotion as a tool to reduce implementation costs or to extract more effort<br />
from the agent. In our two-effort model, however, we do not have much flexibility in<br />
varying effort levels in <strong>and</strong> across jobs. To elaborate on the effects that varying effort<br />
levels across hierarchy levels can have, we extend our model to allow for continuous effort.<br />
Expected outputs in jobs l <strong>and</strong> h are fl(e, θ) = α e <strong>and</strong> fh(e, θ) = max {α e θ − k, 0},<br />
respectively, with k ≥ 0, θ ∈ [θ, ¯ θ] ⊆ R+. The cost <strong>of</strong> effort is c(e) = e 2 /2. Further, we<br />
impose the following technical conditions:<br />
Assumption 2<br />
(i) α ∈ (0, 1) <strong>and</strong> e ∈ [0, 1].<br />
(ii) k/α < ¯ θ < 1/α.<br />
(iii) θ < √ α 2 +2 k<br />
α<br />
< ¯ θ.<br />
The first two parts guarantee proper probability distribution functions. Specifically, (i)<br />
implies that fl(e, θ) ∈ (0, 1). And the upper bound in (ii) ensures that fh(e, θ) < 1 for all<br />
θ ∈ [θ, ¯ θ]. The lower bound guarantees that there exist interior ability levels θ ∈ (θ, ¯ θ)<br />
for which positive output in job h is possible (i.e., fh(e = 1, θ) > 0). Finally, (iii) ensures<br />
that there exists an interior first-best job assignment threshold.<br />
The ‘warm up’ cost k ≥ 0 for the more complex job h allows us to vary parametrically<br />
the difference in informativeness <strong>of</strong> output across jobs. As argued in Section 2.4, a job<br />
higher up in the hierarchy (job h) <strong>of</strong>ten is more informative about effort. Assuming k > 0<br />
captures this. Analogous to the two-effort case, the continuous version <strong>of</strong> the likelihood<br />
ratio says how informative output in job j ∈ {l, h} is about effort: rj(e, θ) = ∂<br />
∂ e fj(e,θ)<br />
fj(e,θ) .<br />
In our setup, the inverse likelihood ratios across jobs have a simple relation:<br />
1<br />
rh(e, θ) =<br />
1 k<br />
−<br />
rl(e, θ) α θ .<br />
So at k = 0 both jobs are equally informative about effort. With increasing k, job h<br />
becomes more informative about effort relative to job l. This can be seen at an intuitive<br />
level by comparing what success in a given job tells the principal about the possible<br />
effort levels that could have led to this outcome. In job l this could have been any effort<br />
level e > 0. In contrast, success in job h is more informative about effort, because a<br />
success indicates that the agent put in at least an effort <strong>of</strong> k/(α θ). Similarly, in job h<br />
output becomes less informative about effort with a higher ability level: the more able<br />
an agent the more likely it is that he succeeds even with little effort.<br />
13
Figure 2: Second-best versus efficient promotion threshold<br />
The following additional assumptions simplify the analysis by eliminating corner solu-<br />
tions for the second-best outcomes. Intuitively, they restrict the warm-up cost k to be<br />
sufficiently small so that positive effort always pays <strong>of</strong>f in job h.<br />
Assumption 3<br />
(iv) θ ≥ √ 3 k<br />
α .<br />
(v) k <<br />
3 α2<br />
4 .<br />
Part (iv) guarantees that it is optimal for the principal to implement a positive second-<br />
best effort level in job h for all θ ∈ [θ, ¯ θ]. And part (v) ensures an interior second-best<br />
promotion threshold.<br />
We focus here on the implications <strong>of</strong> the model as well as the links to previous results,<br />
<strong>and</strong> relegate formal derivations to Appendix C. To generate the figures that illustrate<br />
our results we use the following parametrization: θ ∈ 1<br />
2 , 2 , e ∈ [0, 1], α = 1/2, <strong>and</strong><br />
0 < k < 1/16.<br />
3.2 Results<br />
3.2.1 Optimal effort levels <strong>and</strong> promotion thresholds<br />
We first discuss the properties <strong>of</strong> the optimal effort levels. The first-best effort lev-<br />
F B els are el B<br />
= α <strong>and</strong> eFh (θ) = α θ, respectively. In each job, we observe the usual<br />
downward distortion <strong>of</strong> effort in a moral hazard model. The second-best effort level<br />
in job l is half <strong>of</strong> the first-best one: e SB<br />
l<br />
B = 0.5 eFl . Comparing this to job h, where<br />
14
eSB h (θ) = 0.5 <br />
F B k eh (θ) + , one sees that the distortion relative to the first-best level is<br />
α θ<br />
smaller than in job l, <strong>and</strong> decreases with a more informative output (higher k). For k = 0<br />
the relative distortion <strong>of</strong> effort is equal for both jobs. The closed form solutions for the<br />
optimal effort levels allow us to derive the first- <strong>and</strong> second-best promotion thresholds:<br />
θ F B √<br />
α2 + 2 k<br />
≡<br />
α<br />
<strong>and</strong> θ SB ≡ α + √ α2 + 4 k<br />
.<br />
2 α<br />
Again, it is interesting to see how the difference between first- <strong>and</strong> second-best thresholds<br />
varies with the informativeness parameter k. For k = 0 the second-best promotion<br />
threshold is efficient: because both jobs are equally informative about effort there is<br />
neither a relative distortion in the second-best implementation costs, nor in the second-<br />
best effort. Hence, the principal cannot gain by moving away from the efficient promotion<br />
threshold. This is similar to the case in the two-effort model where the agent works hard<br />
in both jobs (effort is constant across jobs) <strong>and</strong> likelihood ratios are equal (both jobs are<br />
equally informative). Once the jobs differ in informativeness, the picture changes: for<br />
k > 0 we have θ SB < θ F B . The gap between first- <strong>and</strong> second-best promotion thresholds<br />
increases in k as Figure 2 illustrates. The greater k, the greater the distortion in the<br />
promotion threshold. The following proposition summarizes.<br />
Proposition 2 The second-best promotion threshold is lower than the first-best one for<br />
k > 0; the gap between the two increases in k.<br />
3.2.2 Expected output<br />
So far we have seen that the promotion threshold is set inefficiently low. What does<br />
this imply for the agent’s output? An agent promoted to the higher level job produces<br />
more than an agent who remains in job l, as illustrated in Figure 3. But this does not<br />
mean that the agent is more productive in the new job. Figure 4 illustrates this by<br />
comparing the output in job h with the output the agent would produce in his old job<br />
l, when providing the same effort as he does now. One can see that for ability levels<br />
close to the threshold θ SB , the agent would be more productive with his effort if he<br />
remained in job l instead <strong>of</strong> being promoted to job h. The reason for higher output<br />
thus is that the promoted agent works harder than he would work in his old job. The<br />
patterns illustrated in the figures arise in general in our model <strong>and</strong> are summarized in<br />
the following proposition.<br />
Proposition 3<br />
1. Agents promoted to job h work harder than those remaining in job l.<br />
2. Expected output increases for promoted agents relative to those remaining in job l.<br />
15
Figure 3: Expected output<br />
Figure 4: The (extended) Peter Principle<br />
16
3. In the lower ability range <strong>of</strong> job h, the expected output <strong>of</strong> an agent would be higher<br />
in job l than in job h, if he worked as hard as he now does in job h.<br />
What drives the inefficient job assignment? Because output in the higher level job is more<br />
informative about the agent’s effort, the principal can save on costs for implementing a<br />
given level <strong>of</strong> effort by lowering the promotion threshold relative to the first-best one.<br />
These savings allow the principal to get the agent to exert extra effort, which drives<br />
up expected output (similar to the case in the two-effort model where the principal<br />
implements low effort in job l <strong>and</strong> high effort in job h).<br />
3.2.3 Inefficient job assignments <strong>and</strong> the Peter Principle<br />
Inefficiencies in promotion policies are <strong>of</strong>ten attributed to the Peter Principle. It says<br />
that some employees are promoted to their “level <strong>of</strong> incompetence” <strong>and</strong> then remain at<br />
that hierarchy level (Peter <strong>and</strong> Hull 1969). So it refers broadly to cases where, because<br />
<strong>of</strong> his “incompetence” an employee produces less for the same inputs after a promotion<br />
than he would be capable <strong>of</strong> producing in his former job.<br />
Often, however, the Peter Principle is interpreted in a more restricted sense, as an ob-<br />
served decline in output after a promotion (e.g. Lazear 2004). An important caveat<br />
arising from our analysis is that such a comparison <strong>of</strong> actual outputs may be a mislead-<br />
ing test for “incompetence”. Previous theoretical treatments <strong>of</strong> promotion policies <strong>and</strong><br />
the Peter Principle have sidestepped this issue by simply assuming that effort does not<br />
vary across hierarchy levels (see Section 1). 11 Indeed, Proposition 3 shows that output<br />
may rise, simply because a promoted agent works harder – hiding the fact that he is<br />
actually less talented for his new job than he is for his former job.<br />
To capture the possibility that a promoted agent works harder or less hard than he would<br />
if he were reassigned to the previous job, we therefore propose an extended definition <strong>of</strong><br />
the Peter Principle.<br />
Definition 1 The (extended) Peter Principle holds for an agent with ability θ who is<br />
promoted to job h, in which he exerts effort eh, if fh(eh, θ) < fl(eh, θ).<br />
That is, the Peter Principle is said to hold if the agent produces less after a promotion<br />
than he would be capable <strong>of</strong> producing in his former job given the same inputs, i.e.,<br />
holding fixed his effort eh at its post-promotion level.<br />
Our setup links the Peter Principle to the cost <strong>of</strong> providing incentives: the principal<br />
can use a promotion as a tool to reduce implementation costs or to extract more effort<br />
11 Interestingly, Barmby, Eberth, <strong>and</strong> Ma (2007) recognize this problem in their empirical test <strong>of</strong><br />
Lazear’s (2004) model, <strong>and</strong> attempt to control for differences in incentives across hierarchy levels using<br />
differences in mean income.<br />
17
from the agent. According to Proposition 3 the (extended) Peter Principle thus holds<br />
for those who are just above the promotion threshold to job h. As illustrated in Figure<br />
4, such an agent would produce more in his old job if he worked as hard as he does now<br />
after the promotion.<br />
3.2.4 The pr<strong>of</strong>ile <strong>of</strong> expected earnings <strong>and</strong> the agent’s well-being<br />
Figure 5 illustrates the cost saving effect that supports the (extended) Peter Principle.<br />
Panel (a) shows that an agent who is just at the promotion threshold (ability equal to<br />
θ SB ) earns more than he would if he remained in the lower level job. If, however, the<br />
principal wanted to have the agent work as hard in the lower level job as he does in the<br />
new job h, she would need to pay him more, as shown in Panel (b). The gap in earnings<br />
between job h <strong>and</strong> l widens the more informative job h is relative to job l (i.e., the higher<br />
k). Intuitively, the better information in job h reduces the rent the principal has to leave<br />
to the agent. This allows her to squeeze more effort out <strong>of</strong> him.<br />
The higher earnings in job h relative to the lower level job thus are the result <strong>of</strong> harder<br />
work. Panel (a) <strong>of</strong> Figure 6 illustrates this: for a fixed k a more able agent can expect<br />
to earn more than a less able colleague, both within the higher level job <strong>and</strong> across<br />
jobs. That is, the expected wage is always higher in job h than in job l. But, as a<br />
consequence <strong>of</strong> the cost saving effect, an agent in job h with an ability level close to the<br />
promotion threshold enjoys a lower utility than he would in job l (see Panel (b) <strong>of</strong> Figure<br />
6). The expected utility increases with higher θ <strong>and</strong> soon exceeds that from working in<br />
the lower level job. This is in line with the commonly expressed view that a promotion<br />
is a desirable career move for most employees: expected utility increases for most ability<br />
levels in job h. For such an agent a promotion leads to a better job match, where higher<br />
earnings more than compensate for the increased workload. As the following proposition<br />
shows, the patterns illustrated in the figures arise in general in our model.<br />
Proposition 4<br />
1. A promotion leads to a jump in expected earnings relative to those who remain in<br />
the lower level job l. In the higher level job h expected earnings increase with θ.<br />
2. The net effect on expected utility relative to non-promotion for the agents in job h<br />
(a) is negative in the lower ability range; they are overpromoted, i.e., less happy<br />
than if they were in their old job with lower earnings but less work;<br />
(b) is positive for all others; for them the higher earnings compensate for the<br />
increased workload.<br />
The second result is driven by the principal’s willingness to forgo gains in expected<br />
output in order to lower the limited liability rent she has to leave to the agent. This<br />
18
Figure 5: Expected earnings <strong>and</strong> cost saving effect (θ = θ SB )<br />
Figure 6: Expected wage <strong>and</strong> expected utility (k = 1/32)<br />
19
eduction in the limited liability rent makes an overpromoted agent less happy than he<br />
could be in his old job. However, as he still is earning a limited liability rent, he prefers<br />
the promotion over being fired <strong>and</strong> forced to take the outside option.<br />
Clearly, an agent who rejects a promotion will not necessarily be fired in real world em-<br />
ployment relationships. But there are cases where turning down a promotion is costly,<br />
as the following two examples illustrate. In a recent case, the US Federal Circuit up-<br />
held the dismissal <strong>of</strong> a general attorney in the Los Angeles <strong>of</strong>fice <strong>of</strong> the Small Business<br />
Administration for rejecting a promotion. 12 The FBI had a practice <strong>of</strong> cutting the pay<br />
<strong>of</strong> agents who refuse to accept promotion to a supervisory position. 13 Even when the<br />
penalties are less severe, people may accept their role for reasons <strong>of</strong> prestige or status,<br />
as illustrated by the following case: “a man in his early fifties had been overpromoted,<br />
but held on to his job because, as he put it, ‘<strong>of</strong> the status <strong>and</strong> authority <strong>of</strong> the role’”<br />
(Erskine <strong>and</strong> Brook 1971, p.55). Enriching the model with concerns for status (see e.g.<br />
the model <strong>of</strong> Auriol <strong>and</strong> Renault 2008) seems to be an interesting direction for future<br />
research.<br />
For the cases where an agent will not or cannot reject a promotion our model provides a<br />
link between promotions <strong>and</strong> workplace stress. Those at the bottom end <strong>of</strong> the earnings<br />
distribution in the higher level job are kept on toes because they are not sufficiently<br />
talented for the new position: to make up for their (slight) lack in talent they work<br />
harder than they would in their old job for the same expected earnings. Consistent with<br />
this finding, “overpromotion” has been identified as one <strong>of</strong> the factors causing stress<br />
at the workplace (e.g. Cooper <strong>and</strong> Marshall 1976, Murphy 1995). 14 In line with this,<br />
Boyce <strong>and</strong> Oswald (2008) find that promotions significantly worsen the psychological<br />
strain <strong>of</strong> private sector employees, controlling for endogeneity <strong>of</strong> health measures with<br />
longitudinal data.<br />
4 Concluding remarks<br />
The literature on careers in organizations stresses the role <strong>of</strong> incomplete information<br />
about employees’ abilities <strong>and</strong> contractual imperfections for inefficiencies in job assign-<br />
ments. Our model strips away these presumed sources <strong>of</strong> inefficiencies: even though<br />
there is no uncertainty about the agent’s ability <strong>and</strong> complete contracts are possible, a<br />
static trade-<strong>of</strong>f between incentive provision <strong>and</strong> efficient assignment generically pushes<br />
the job assignment threshold away from its efficient level.<br />
12 Robinson v. Small Business Administration, U.S.C.A.F.C. No. 06-3065, 5/4/06.<br />
13 Los Angeles Times, FBI Agents Rebel Over M<strong>and</strong>atory Transfers, May 22, 2006.<br />
14 A person is overpromoted when he “has reached the peak <strong>of</strong> his abilities [...] <strong>and</strong> is given responsi-<br />
bility exceeding his capacity” (Cooper <strong>and</strong> Marshall 1976, p.18).<br />
20
Applying our simple model to a promotion context shows that an organization may find<br />
it optimal to promote an employee to a job for which he is less talented than for the<br />
previously held lower level position (the Peter Principle). Such a promotion decision<br />
makes it cheaper to induce hard work: facing a more challenging job, the employee<br />
has to make up for his lack in talent by exerting more effort than he would in his old<br />
job for the same compensation. In other words, the promotion policy complements the<br />
wage policy as a tool to reduce the rents that employees enjoy because <strong>of</strong> moral hazard<br />
problems in the organization.<br />
The mechanism at work is a static one: the job assignment decision in one period affects<br />
the contemporaneous cost <strong>of</strong> providing incentives. This alone is sufficient to create<br />
inefficiencies in the assignment policy. To not obscure this, we deliberately kept the<br />
promotion model simple. Nevertheless, the framework delivers novel predictions in line<br />
with several stylized facts. First, it <strong>of</strong>fers an explanation for why organizations seem<br />
to be content with having employees in positions where they are less productive than<br />
they would be in their former job (the Peter Principle). Second, it shows that employees<br />
moving up the career ladder may earn more only because they have to work harder than<br />
they would in their old jobs for the same pay. Third, the setup captures the idea that a<br />
promotion can be a good thing for most workers, but at the same time may reduce the<br />
rents enjoyed by the least able workers in a hierarchy level. Because the latter have to<br />
put in extra effort to outweigh their lack in talent, they are less happy than they would<br />
be if they could stay in their old job – a phenomenon that the literature on workplace<br />
stress refers to as overpromotion (e.g. Cooper <strong>and</strong> Marshall 1976, Murphy 1995).<br />
While our setup goes beyond previous theoretical job assignment models by allowing<br />
for implemented effort levels to vary across hierarchy levels, there are many facets <strong>of</strong><br />
careers in organizations that our model does not capture. Indeed, our purpose was to<br />
show what additional insights a tractable model <strong>of</strong> the trade-<strong>of</strong>f between job assignment<br />
<strong>and</strong> incentives yields. Our framework can serve as a building block that could usefully<br />
be added to an integrated promotion model in the spirit <strong>of</strong> Gibbons <strong>and</strong> Waldman<br />
(1999b, 2006), who abstract from incentives. In such an enriched model, for example,<br />
the effective ability <strong>of</strong> an agent may vary with tenure t in the organization: ρt(θ). Or the<br />
expected output in a job may directly depend on the agent’s experience. Taken together,<br />
the success probability may be written as fj(e, ρt(θ), t). This more general formulation<br />
can capture learning by doing <strong>and</strong> human capital accumulation: one possibility is that<br />
the productivity <strong>of</strong> an agent increases over time, fj(e, ρ, t+1) > fj(e, ρ, t); or ability may<br />
increase with experience ρt+1(θ) > ρt(θ). The model then can also encompass learning<br />
about the ability <strong>of</strong> an agent, if ρt(θ) is thought <strong>of</strong> as the expected ability given the<br />
information available. 15 Developing an integrated promotion model along these lines is<br />
15 Note that in our model we would have ρ0(θ) = E[θ] <strong>and</strong> ρt(θ) = θ.<br />
21
an interesting direction for future research.<br />
Appendix<br />
A <strong>Job</strong> assignments given inefficient effort levels<br />
This section builds intuition for additional effects that inefficient second-best effort levels<br />
can have on the job assignment rule. To illustrate these effects, we assume that job h<br />
is locally more informative about effort than job l (Condition 2). We discuss here the<br />
case where the principal implements the inefficient low effort in job l (e SB<br />
l<br />
Condition 1 is violated for job l). The driving forces for the other case (e SB<br />
l<br />
= e, i.e.,<br />
= ē) are<br />
similar to those in Section 2, <strong>and</strong> we leave the details for the pro<strong>of</strong> <strong>of</strong> Proposition 5.<br />
Suppose first that in job h Condition 1 is also violated for all ability levels, i.e., e SB<br />
l<br />
e SB<br />
h<br />
(θ) = e. Then no incentives are needed because the principal implements low effort<br />
in either job. Hence, the job assignment problem reduces to maximizing the expected<br />
revenue fj(e, θ) over j, resulting in threshold θ SB = ˆ θ(e, e) > θ F B = ˆ θ(ē, ē). In other<br />
words, the downward distortion <strong>of</strong> effort in both jobs pushes the job assignment threshold<br />
above the first-best level. As we saw in Section 2, if second-best effort levels in both<br />
jobs are efficient (i.e., Condition 1 holds for all ability levels in both jobs), the principal<br />
distorts the job assignment threshold in the opposite direction: assignment to job h leads<br />
to an implementation cost saving effect that pulls the threshold below the first-best level:<br />
θ SB < ˆ θ(ē, ē).<br />
We expect both forces to be at work whenever Condition 1 holds for job h, but not<br />
for job l. To illustrate that this is indeed the case, <strong>and</strong> to see how these two forces<br />
interact, suppose that for job h Condition 1 holds for all θ. 16 Because no incentives are<br />
needed in job l, placing the agent into job h cannot lead to cost savings. But it may be<br />
cheaper to implement high effort in job h than it would be in job l. For that reason,<br />
assigning the agent to job h <strong>and</strong> then implementing high effort might be worthwhile for<br />
some θ < ˆ θ(e, e). Indeed, it is even possible that θ SB is lower than θ F B . When does this<br />
occur? Assigning the agent to the higher level job <strong>and</strong> implementing high effort increases<br />
output by fh(ē, θ) − fl(e). This has to outweigh the increase in implementation costs <strong>of</strong><br />
Ch (ē, θ). Hence, if there is a strict gain in pr<strong>of</strong>it from assigning an agent with ability<br />
θ = θ F B to job h we know that θ SB < θ F B (by continuity). The following proposition,<br />
which we prove formally in Appendix B, summarizes our results.<br />
16 Dropping the qualifier “for all θ”, implies that multiple thresholds are possible. Then θ SB in the<br />
explanations <strong>and</strong> results below refers to the lowest threshold for assigning the agent to job h (see the<br />
pro<strong>of</strong> <strong>of</strong> Proposition 5).<br />
22<br />
=
Proposition 5 Suppose job h is locally more informative than job l (Condition 2). In<br />
case multiple cut<strong>of</strong>fs exist, θ SB refers to the lowest second-best threshold.<br />
1. Suppose the second-best effort level in job l is inefficient (e SB<br />
l<br />
is violated for job l). Then,<br />
= e, i.e., Condition 1<br />
(a) θSB ≤ ˆ θ(e, e), <strong>and</strong> θSB is higher than it would be if eSB h (θ) = eSB l = ē ∀ θ;<br />
(b) θ SB < θ F B if <strong>and</strong> only if<br />
fh(ē, ˆ θ(ē, ē)) − fl(e) > Ch(ē, ˆ θ(ē, ē)). (Condition 3)<br />
2. Suppose the second-best effort level in job l is efficient (e SB<br />
l<br />
holds for job l), then θ SB < θ F B .<br />
B Pro<strong>of</strong>s<br />
Pro<strong>of</strong> <strong>of</strong> Lemma 1.<br />
= ē, i.e., Condition 1<br />
d<br />
d θ ΠSB h θ (ē, θ) = fh θ(ē, θ) − Ch θ(ē, θ) = fh θ(ē, θ) − fh(ē, θ) fh θ(e, θ) − fh(e, θ) fh θ(ē, θ)<br />
[fh(ē, θ) − fh(e, θ)] 2 c<br />
> fh θ(ē, θ) − fh(ē, θ) fh θ(e, θ)<br />
c > 0,<br />
[fh(ē, θ) − fh(e, θ)] 2<br />
where the last inequality follows from fh θ(ē, θ) ≥ fh θ(e, θ) (Assumption SA) <strong>and</strong><br />
[fh(ē, θ) − fh(e, θ)] 2 > fh(ē, θ) c (Condition 1). This shows Part (a). Part (b) follows.<br />
Pro<strong>of</strong> <strong>of</strong> Proposition 1.<br />
We prove one side <strong>of</strong> the inequality here, the reverse direction is analogous.<br />
Π SB<br />
h (ē, ˆ θ(ē, ē)) − Π SB<br />
l (ē) > 0 ⇔ fh(ē, ˆ θ(ē, ē)) − Ch(ē, ˆ θ(ē, ē)) > fl(ē) − Cl(ē)<br />
⇔ Ch(ē, ˆ θ(ē, ē)) < Cl(ē) ⇔ fh(ē, ˆ θ(ē, ē)) − fh(e, ˆ θ(ē, ē)) > fl(ē) − fl(e)<br />
⇔ fh(e, ˆ θ(ē, ē)) < fl(e),<br />
using repeatedly fh(ē, ˆ θ(ē, ē)) = fl(ē). θ SB < θ F B follows because θ F B = ˆ θ(ē, ē) <strong>and</strong> the<br />
pr<strong>of</strong>it difference is strictly increasing in θ (Lemma 1).<br />
Pro<strong>of</strong>s <strong>of</strong> Propositions 2-4.<br />
See Appendix C.<br />
Pro<strong>of</strong> <strong>of</strong> Proposition 5.<br />
Part (1a): Denote by ˜ θ the lowest ability level where expected pr<strong>of</strong>its in jobs h <strong>and</strong> l<br />
would be equalized if high effort was implemented in both jobs. That is, Π SB<br />
h (ē, ˜ θ) =<br />
23
Π SB<br />
l (ē) (these are not the maximized second-best pr<strong>of</strong>it levels, as implementing high<br />
effort is not optimal in job l, <strong>and</strong> may not be in job h either). From the pro<strong>of</strong> <strong>of</strong><br />
Proposition 1 we know that ˜ θ < ˆ θ(ē, ē). Condition 2 implies that ˆ θ(ē, ē) < ˆ θ(e, e).<br />
Denote by θ SB the lowest threshold for assigning the agent to the high ability job. We<br />
now show that θ SB ∈ ( ˜ θ, ˆ θ(e, e)], for which we will divide the ability interval [θ, ¯ θ] into<br />
subintervals over which Condition 1 may hold or fail for job h. Note that whether or<br />
not Condition 1 holds for θ ≥ ˆ θ(e, e) or θ ≤ ˜ θ is irrelevant: all types above ˆ θ(e, e) are<br />
assigned to job h, all below ˜ θ to job l.<br />
Suppose first that Condition 1 also does not hold for any θ ≤ ˆ θ(e, e) in job h. Then<br />
because ΠSB h (e, θ) = fh(e, θ) ≤ fl(e) = ΠSB l (e) for θ ≤ ˆ θ(e, e) <strong>and</strong> maxe ΠSB h (e, θ) ≥<br />
fh(e, θ) > fl(e) = ΠSB l (e) for θ > ˆ θ(e, e) it follows that θSB = ˆ θ(e, e).<br />
Suppose now that Condition 1 holds for θ ≥ ˜ θ in job h. Then, for these ability levels we<br />
have Π SB<br />
h<br />
(ē, θ) ≥ ΠSB<br />
h<br />
(e, θ) <strong>and</strong> ΠSB h θ (ē, θ) > 0 (Lemma 1). Because Condition 1 does not<br />
hold for job l, ΠSB l (e) > ΠSB l (ē) = ΠSB h (ē, ˜ θ), as ΠSB l (e) = fl(e) > ΠSB l (ē) = fl(ē)−Cl(ē).<br />
Hence, it follows that θSB > ˜ θ. Moreover, ΠSB h (ē, ˆ θ(e, e)) > ΠSB h (e, ˆ θ(e, e)) = ΠSB l (e), so<br />
θ SB < ˆ θ(e, e).<br />
Finally, if Condition 1 holds only for some θ ∈<br />
( ˜ θ, ˆ θ(e, e)].<br />
<br />
˜θ, θ(e, ˆ e) , then a fortiori θSB ∈<br />
Part (1b): To prove the if part suppose that fh(ē, ˆ θ(ē, ē)) − fl(e) > Ch(ē, ˆ θ(ē, ē)).<br />
Together with Condition 2 this implies that fh(ē, ˆ θ(ē, ē)) − fh(e, ˆ θ(ē, ē)) > Ch(ē, ˆ θ(ē, ē)),<br />
so Condition 1 holds at θ = ˆ θ(ē, ē). Thus, the maximized second-best pr<strong>of</strong>it in job h is<br />
Π SB<br />
h (ē, ˆ θ(ē, ē)) > Π SB<br />
l (e). By continuity, θ SB < θ F B = ˆ θ(ē, ē).<br />
To prove the only if part, suppose θ SB < θ F B = ˆ θ(ē, ē). By definition <strong>of</strong> the second-<br />
best threshold Π SB<br />
h (ē, θSB ) = fl(e). The next steps show that Π SB<br />
h (ē, ˆ θ(ē, ē)) > fl(e)<br />
follows from (Condition 3). Condition 2 implies that ˆ θ(ē, ē) < ˆ θ(e, e) so fl(e) > fh(e, θ)<br />
for θ ∈ [θ SB , ˆ θ(ē, ē)]. From the previous step <strong>and</strong> Π SB<br />
h (ē, θSB ) = fl(e) it follows that<br />
Π SB<br />
h (ē, θSB ) > fh(e, θ SB ), which is equivalent to Condition 1 holding at θ SB . By Lemma<br />
1, Π SB<br />
h θ (ē, θSB ) > 0. We now show that the pr<strong>of</strong>it never falls back to or below fl(e)<br />
over the relevant range. By way <strong>of</strong> contradiction, suppose that there exist values<br />
θn ∈ (θSB , ˆ θ(ē, ē)] for which ΠSB h (ē, θn) = fl(e), <strong>and</strong> let θ1 denote the smallest <strong>of</strong> these.<br />
Because ΠSB h (ē, θSB ) = fl(e) <strong>and</strong> ΠSB h θ (ē, θSB ) > 0 it follows that ΠSB h θ (ē, θ1) < 0. This,<br />
however, implies that Condition 1 is violated at θ1, i.e., ΠSB h (ē, θ1) < fh(e, θ1), which<br />
leads to a contradiction because fl(e) > fh(e, θ1).<br />
Part (2): Denote by ˜ θ the ability level where expected pr<strong>of</strong>its in jobs h <strong>and</strong> l would be<br />
equalized if high effort was implemented in both jobs, as in the pro<strong>of</strong> <strong>of</strong> Part (1a). We now<br />
show that if Condition 1 holds for job l, then Condition 1 holds also for job h at θ = ˜ θ:<br />
24
fh(ē, ˜ θ)−Ch(ē, ˜ θ) = fl(ē)−Cl(ē) ⇔ fh(ē, ˜ θ)−fl(e)−Ch(ē, ˜ θ) = fl(ē)−fl(e)−Cl(ē). The<br />
right-h<strong>and</strong> side is nonnegative because Condition 1 holds for job l. Using fh(e, ˜ θ) < fl(e)<br />
therefore yields fh(ē, ˜ θ)−fh(e, ˜ θ) > Ch(ē, ˜ θ) (Condition 1). Thus, the maximized second-<br />
best pr<strong>of</strong>it in job h at θ = ˜ θ is Π SB<br />
h (ē, ˜ θ), <strong>and</strong> by the definition <strong>of</strong> ˜ θ it follows that<br />
θ SB = ˜ θ. Furthermore, we know that θ SB = ˜ θ < θ F B = ˆ θ(ē, ē). We now show that<br />
θ SB is the only job assignment threshold below θ F B , i.e., Π SB (ē, θ) never falls back to<br />
or below fl(ē) − Cl(ē) for θ ∈ (θ SB , ˆ θ(ē, ē)]. By way <strong>of</strong> contradiction, suppose that<br />
there exist values θn ∈ (θSB , ˆ θ(ē, ē)] for which ΠSB h (ē, θn) = fl(ē) − Cl(ē), <strong>and</strong> let θ1<br />
denote the smallest <strong>of</strong> these. Because ΠSB h (ē, θSB ) = fl(ē) − Cl(ē) <strong>and</strong> ΠSB h θ (ē, θSB ) > 0<br />
(by Lemma 1) it follows that ΠSB h θ (ē, θ1) < 0. This, however, implies that Condition 1<br />
is violated at θ1, i.e., ΠSB h (ē, θ1) < fh(e, θ1), which leads to a contradiction because<br />
fl(ē) − Cl(ē) ≥ fl(e) > fh(e, θ1).<br />
C Continuous effort model<br />
First best<br />
F B<br />
The efficient effort level given θ solves maxe fj(e, θ) − c(e). For job l this yields e<br />
l<br />
= α<br />
F B<br />
B<br />
<strong>and</strong> for job h we have eh (θ) = α θ. Hence, the expected pr<strong>of</strong>its are ΠFl α2 = 2 <strong>and</strong><br />
F B Πh (θ) = α2θ2 2<br />
B<br />
− k, respectively. Setting pr<strong>of</strong>its equal defines the efficient cut<strong>of</strong>f θF<br />
implicitly by α2θ2 − 2 k − α2 = 0. Solving for θF B yields<br />
θ F B √<br />
α2 + 2 k<br />
≡ .<br />
α<br />
Second best<br />
To implement effort e in job j the wage after success must be set at ¯wj(e, θ) = ce(e)<br />
fj e(e,θ) ,<br />
where subscript e denotes the partial derivative with respect to effort. That is,<br />
¯wl(e) = e<br />
α <strong>and</strong> ¯wh(e, θ) = e . Denote the expected cost <strong>of</strong> implementing effort in job j by<br />
α θ<br />
Cj(e, θ) = fj(e, θ) ¯wj(e, θ). Maximizing the principal’s expected pr<strong>of</strong>it fj(e, θ) − Cj(e, θ)<br />
over e gives the second-best effort level for a type-θ agent. In job l this yields e SB<br />
l<br />
half the efficient effort level. In job h we have 17<br />
e SB<br />
h (θ) = 1<br />
<br />
α θ +<br />
2<br />
k<br />
<br />
.<br />
α θ<br />
α = 2 ,<br />
17 SB<br />
Note that an effort <strong>of</strong> zero yields an expected utility <strong>of</strong> zero. So for eh (θ) > 0 to be optimal the<br />
incentive constraint requires that EUh(eSB h (θ), θ) ≥ 0. Equation (5) <strong>and</strong> the derivations below show<br />
that EUh(eSB h (θ), θ) ≥ 0 ⇔ θ ≥ √ 3 k<br />
α . This explains A4.<br />
25
So if k = 0, the second-best effort is also exactly half the efficient effort level, <strong>and</strong> with<br />
higher k the effort increases. Expected pr<strong>of</strong>its are:<br />
Πl = fl(e SB<br />
l ) − Cl(e SB<br />
l ) = α2<br />
4 ,<br />
Πh(θ) = fh(e SB<br />
h (θ), θ) − Ch(e SB<br />
h (θ), θ) = (α2 θ2 − k) 2<br />
4 α2 θ2 .<br />
Setting pr<strong>of</strong>its equal defines the second-best cut<strong>of</strong>f θ SB implicitly, <strong>and</strong> solving for θ SB<br />
yields 18<br />
Pro<strong>of</strong>s <strong>of</strong> Propositions 2-4.<br />
θ SB ≡ α + √ α2 + 4 k<br />
.<br />
2 α<br />
We now prove formally the remaining results discussed in the text.<br />
Promotion threshold θ SB < θ FB for k > 0: By way <strong>of</strong> contradiction, suppose θ F B ≤<br />
θ SB ⇔ 2 √ α 2 + 2 k ≤ α + √ α 2 + 4 k. Squaring both sides: 4 (α 2 + 2 k) ≤ 2 (α 2 +<br />
α √ α 2 + 4 k + 2 k) ⇔ α 2 + 2 k ≤ α √ α 2 + 4 k. Squaring again both sides: α 4 + 4 k α 2 +<br />
4 k 2 ≤ α 4 + 4 k α 2 , yielding a contradiction for k > 0. Taking derivatives shows that<br />
|θ F B − θ SB | decreases in k.<br />
Expected output: For an agent placed in job h, the expected output is higher than it<br />
would be if he were instead assigned to job l (with the second-best effort level for the<br />
respective job). The difference in expected output fh(eSB SB<br />
h (θ), θ) − fl(el ) = α2 (θ2−1)−k , 2<br />
is increasing in θ:<br />
∂ <br />
fh(e<br />
∂ θ<br />
SB<br />
h (θ), θ) − fl(e SB<br />
l ) = α 2 θ > 0.<br />
At θSB there is an upward jump in expected output:<br />
fh(e SB<br />
h (θ SB ), θ SB ) − fl(e SB<br />
l ) = α (√ α 2 + 4 k − α)<br />
4<br />
As the output difference is increasing it follows that fh(eSB SB<br />
h (θ), θ) > fl(el ) also for all<br />
θ > θSB .<br />
The Peter Principle: Keeping effort constant at the second-best level for job h, the<br />
expected output for an agent with ability θ SB placed in job h is lower than it would be<br />
18 Without A5 a corner solution would arise whenever α+ √ α 2 +4 k<br />
2 α<br />
> 0.<br />
≤ √ 3 k<br />
α . Then, θSB = θ. We focus<br />
here on the interesting case where promotion is not automatic, but requires ability level <strong>of</strong> at least<br />
θ SB > θ.<br />
26
if he were instead assigned to job l:<br />
fh(e SB<br />
h (θ SB ), θ SB ) − fl(e SB<br />
h (θ SB )) =<br />
−k α<br />
α + √ α 2 + 4 k<br />
< 0.<br />
Effort eSB h (θ) > eSB l for θ > θSB : At θSB there is an upward jump in effort:<br />
e SB<br />
h (θ SB ) − e SB<br />
l =<br />
The effort gap is increasing in θ: d<br />
d 2<br />
d θ2 eSB h<br />
d θ eSB h<br />
2 k<br />
α + √ α 2 + 4 k .<br />
(θ) = 1<br />
2<br />
(θ) > 0 <strong>and</strong> d<br />
d θ eSB<br />
h (θ) θ=θ SB > 0, as is easy to check.<br />
<br />
k α − α θ2 <br />
SB > 0 for θ > θ , because<br />
Expected earnings: For an agent assigned to job h, expected earnings (EE) are higher<br />
than they would be if he remained in job l.<br />
EEh(e SB<br />
h (θ), θ) = fh(e SB<br />
h (θ), θ) × ¯wh(e SB<br />
h (θ), θ) = (α2 θ2 − k) (α2 θ2 + k)<br />
4 (α2 θ2 ,<br />
)<br />
EEl(e SB<br />
l ) = fl(e SB<br />
l ) × ¯wl(e SB<br />
l ) = α2<br />
4 ,<br />
EEh(e SB<br />
h (θ), θ) − EEl(e SB<br />
l ) = α4 θ2 (θ2 − 1) − k2 4 α2 θ2 .<br />
At θ = θ SB there is an upward jump in the expected earnings: 19<br />
EEh(e SB<br />
h (θ SB ), θ SB ) − EEl(e SB<br />
l ) =<br />
Since the difference is increasing in θ,<br />
d<br />
d θ<br />
α k<br />
α + √ α 2 + 4 k<br />
EEh(e SB<br />
h (θ), θ) − EEl(e SB<br />
l ) = α4 θ 4 + k 2<br />
2 α 2 θ 3<br />
> 0.<br />
> 0,<br />
we conclude that the expected earnings difference is positive for all θ > θ SB .<br />
Expected utility: The expected utility (EU) in job j is given by the expected earnings,<br />
calculated above, net <strong>of</strong> the effort cost c(e SB<br />
j (θ)):<br />
EUh(e SB<br />
h (θ SB ), θ SB ) = (α2 θ2 − 3 k) (α2 θ2 + k)<br />
8 (α2 θ2 , (5)<br />
)<br />
EUl(e SB<br />
l ) = α2<br />
8 .<br />
19 Alternatively, this can be seen also from our earlier result that the success probability at second-<br />
best effort is higher in job h, combined with the fact that the wage after success in job h is also higher:<br />
¯wh(e SB<br />
h (θSB ), θ SB ) − ¯wl(e SB<br />
l ) = 8 k<br />
(1+ √ 1+16 k) 2 > 0.<br />
27
Close to the threshold θ SB , the agent has lower expected utility in job h than if he could<br />
remain in job l:<br />
EUh(e SB<br />
h (θ SB ), θ SB ) − EUl(e SB<br />
2 k<br />
l ) = −<br />
2<br />
√ 2 < 0.<br />
α + α2 + 4 k<br />
The expected utility in job h is increasing in θ:<br />
∂<br />
∂ θ<br />
EUh(e SB<br />
h (θ), θ) − EUl(e SB<br />
l ) = α4 θ 4 + 3 k 2<br />
4 α 2 θ 3<br />
> 0.<br />
The gap thus narrows <strong>and</strong> is eventually closed for some ˇ θ < 1/α, as one can show. An<br />
agent with θ > ˇ θ is better <strong>of</strong>f in job h than in job l.<br />
References<br />
Auriol, E., <strong>and</strong> R. Renault (2008): “Status <strong>and</strong> Incentives,” RAND Journal <strong>of</strong><br />
<strong>Economics</strong>, 39(1), 305–326.<br />
Baker, G., R. Gibbons, <strong>and</strong> K. J. Murphy (1994): “Subjective Performance Mea-<br />
sures in Optimal Incentive Contracts,” Quarterly Journal <strong>of</strong> <strong>Economics</strong>, 109(4), 1125–<br />
1156.<br />
Baker, G. P., M. C. Jensen, <strong>and</strong> K. J. Murphy (1988): “Compensation <strong>and</strong><br />
Incentives: Practice vs. Theory,” Journal <strong>of</strong> Finance, 43(3), 593–616.<br />
Baliga, S., <strong>and</strong> T. Sjöström (2001): “Optimal Design <strong>of</strong> Peer Review <strong>and</strong> Self-<br />
Assessment Schemes,” RAND Journal <strong>of</strong> <strong>Economics</strong>, 32(1), 27–51.<br />
Barmby, T., B. Eberth, <strong>and</strong> A. Ma (2007): “Things Can Only Get Worse? An Em-<br />
pirical Examination <strong>of</strong> the Peter Principle,” University <strong>of</strong> Aberdeen Business <strong>School</strong>.<br />
Bernhardt, D. (1995): “Strategic Promotion <strong>and</strong> Compensation,” Review <strong>of</strong> Eco-<br />
nomic Studies, 62(2), 315–39.<br />
Boyce, C. J., <strong>and</strong> A. J. Oswald (2008): “Do People Become Healthier after Being<br />
Promoted?,” IZA Discussion Paper No. 3894.<br />
Calvo, G. A., <strong>and</strong> S. Wellisz (1979): “Hierarchy, Ability, <strong>and</strong> Income Distribution,”<br />
Journal <strong>of</strong> Political Economy, 87(5), 991–1010.<br />
Carmichael, H. L. (1988): “Incentives in Academics: Why Is There Tenure?,” Journal<br />
<strong>of</strong> Political Economy, 96(3), 453–72.<br />
Carrillo, J. D. (2003): “<strong>Job</strong> <strong>Assignments</strong> as a Screening Device,” International Jour-<br />
nal <strong>of</strong> Industrial Organization, 21(6), 881–905.<br />
28
Clark, D. J., <strong>and</strong> C. Riis (2001): “Rank-order Tournaments <strong>and</strong> Selection,” Journal<br />
<strong>of</strong> <strong>Economics</strong>, 73(2), 167–191.<br />
Cooper, C. L., <strong>and</strong> J. Marshall (1976): “Occupational Sources <strong>of</strong> Stress: A Review<br />
<strong>of</strong> the Literature Relating to Coronary Heart Disease <strong>and</strong> Mental Ill Health,” Journal<br />
<strong>of</strong> Occupational Psychology, 49, 11–28.<br />
Demougin, D., <strong>and</strong> C. Fluet (1998): “Mechanism Sufficient Statistic in the Risk-<br />
Neutral Agency Problem,” Journal <strong>of</strong> Institutional <strong>and</strong> Theoretical <strong>Economics</strong>, 127(4),<br />
622–639.<br />
Dickinson, D. L., <strong>and</strong> M. C. Villeval (2007): “The Peter Principle: An Experi-<br />
ment,” IZA Discussion Paper No. 3205.<br />
Erskine, J., <strong>and</strong> A. Brook (1971): “Report on a Two Year Experiment in Co-<br />
Operation Between an Occupational Physician <strong>and</strong> a Consultant Psychiatrist,” Occu-<br />
pational Medecine, 21, 53–56.<br />
Fairburn, J. A., <strong>and</strong> J. M. Malcomson (2001): “Performance, Promotion, <strong>and</strong> the<br />
Peter Principle,” The Review <strong>of</strong> Economic Studies, 68(1), 45–66.<br />
Friebel, G., <strong>and</strong> M. Raith (2004): “Abuse <strong>of</strong> Authority <strong>and</strong> Hierarchical Commu-<br />
nication,” RAND Journal <strong>of</strong> <strong>Economics</strong>, 35(2), 224–244.<br />
Gibbons, M. R., <strong>and</strong> M. Waldman (2006): “Enriching a Theory <strong>of</strong> Wage <strong>and</strong> Pro-<br />
motion Dynamics Inside Firms,” Journal <strong>of</strong> Labor <strong>Economics</strong>, 24(1), 59–107.<br />
Gibbons, R., <strong>and</strong> M. Waldman (1999a): “Careers in Organizations: Theory <strong>and</strong><br />
Evidence,” in H<strong>and</strong>book <strong>of</strong> Labor <strong>Economics</strong>, ed. by O. Ashenfelter, <strong>and</strong> D. Card,<br />
vol. 3B, chap. 36, pp. 2373–2437. Elsevier, Amsterdam.<br />
(1999b): “A Theory Of Wage <strong>and</strong> Promotion Dynamics Inside Firms,” The<br />
Quarterly Journal <strong>of</strong> <strong>Economics</strong>, 114(4), 1321–1358.<br />
Grossman, S. J., <strong>and</strong> O. D. Hart (1983): “An Analysis <strong>of</strong> the Principal-Agent<br />
Problem,” Econometrica, 51(1), 7–45.<br />
Ishida, J. (2006): “Optimal Promotion Policies with the Looking-Glass Effect,” Journal<br />
<strong>of</strong> Labor <strong>Economics</strong>, 24, 857–877.<br />
Jewitt, I. (2006): “Information Order in Decision <strong>and</strong> Agency Problems,” mimeo,<br />
University <strong>of</strong> Oxford.<br />
Kim, S. K. (1995): “Efficiency <strong>of</strong> an Information System in an Agency Model,” Econo-<br />
metrica, 63(1), 89–102.<br />
29
Kwon, I. (2006): “Incentives, Wages, <strong>and</strong> Promotions: Theory <strong>and</strong> Evidence,” RAND<br />
Journal <strong>of</strong> <strong>Economics</strong>, 37(1), 100–120.<br />
Lazear, E. P. (2004): “The Peter Principle: A Theory <strong>of</strong> Decline,” Journal <strong>of</strong> Political<br />
Economy, 112(1), 141–163.<br />
Lazear, E. P., <strong>and</strong> S. Rosen (1981): “Rank-Order Tournaments as Optimum Labor<br />
Contracts,” Journal <strong>of</strong> Political Economy, 89(5), 841–864.<br />
MacLeod, W. B., <strong>and</strong> J. M. Malcomson (1988): “Reputation <strong>and</strong> Hierarchy in<br />
Dynamic Models <strong>of</strong> Employment,” Journal <strong>of</strong> Political Economy, 96(4), 832–54.<br />
Malcomson, J. M. (1984): “Work Incentives, Hierarchy, <strong>and</strong> Internal Labor Markets,”<br />
Journal <strong>of</strong> Political Economy, 92(3), 486–507.<br />
Maskin, E., Y. Qian, <strong>and</strong> C. Xu (2000): “Incentives, Information, <strong>and</strong> Organiza-<br />
tional Form,” Review <strong>of</strong> Economic Studies, 67(2), 359–378.<br />
Meyer, M. A. (1991): “Learning from Coarse Information: Biased Contests <strong>and</strong> Career<br />
Pr<strong>of</strong>iles,” Review <strong>of</strong> Economic Studies, 58(1), 15–41.<br />
Murphy, L. R. (1995): “Occupational Stress Management: Current Status <strong>and</strong> Future<br />
Direction,” Trends in Organizational Behavior, 2, 1–14.<br />
Nafziger, J. (2008): “<strong>Job</strong> <strong>Assignments</strong>, Intrinsic Motivation <strong>and</strong> Extrinsic Incentives,”<br />
Bonn Econ Discussion Papers No. 5/2008.<br />
O’Flaherty, B., <strong>and</strong> A. Siow (1992): “On the <strong>Job</strong> Screening, Up or Out Rules, <strong>and</strong><br />
Firm Growth,” Canadian Journal <strong>of</strong> <strong>Economics</strong>, 25(2), 346–68.<br />
Ortega, J. (2003): “Power in the Firm <strong>and</strong> Managerial Career Concerns,” Journal <strong>of</strong><br />
<strong>Economics</strong> <strong>and</strong> Management Strategy, 12(1), 1–29.<br />
Peter, L. J., <strong>and</strong> R. Hull (1969): The Peter Principle: Why Things Always Go<br />
Wrong. William Morrow & Co. Inc, New York.<br />
Prendergast, C. (1993): “The Role <strong>of</strong> Promotion in Inducing Specific Human Capital<br />
Acquisition,” Quarterly Journal <strong>of</strong> <strong>Economics</strong>, 108(2), 523–534.<br />
Prescott, E. C., <strong>and</strong> M. Visscher (1980): “Organization Capital,” Journal <strong>of</strong><br />
Political Economy, 88(3), 446–461.<br />
Robbins, E. H., <strong>and</strong> B. Sarath (1998): “Ranking Agencies <strong>under</strong> <strong>Moral</strong> <strong>Hazard</strong>,”<br />
Economic Theory, 11, 129–155.<br />
30
Rosen, S. (1982): “Authority, Control, <strong>and</strong> the Distribution <strong>of</strong> Earnings,” Bell Journal<br />
<strong>of</strong> <strong>Economics</strong>, 13(2), 311–323.<br />
(1986): “Prizes <strong>and</strong> Incentives in Elimination Tournaments,” American Eco-<br />
nomic Review, 76(4), 701–715.<br />
Valsecchi, I. (2003): “<strong>Job</strong> Assignment <strong>and</strong> B<strong>and</strong>it Problems,” International Journal<br />
<strong>of</strong> Manpower, 24(7), 844–866.<br />
Waldman, M. (1984a): “<strong>Job</strong> <strong>Assignments</strong>, Signalling, <strong>and</strong> Efficiency,” RAND Journal<br />
<strong>of</strong> <strong>Economics</strong>, 15(2), 255–267.<br />
(1984b): “Worker Allocation, Hierarchies <strong>and</strong> the Wage Distribution,” Review<br />
<strong>of</strong> Economic Studies, 51(1), 95–109.<br />
31