Relative Performance Evaluation in Management Control · Nyenrode Business University Centre for Accounting, Auditing and Control Straatweg 25 3621 BG Breukelen The Netherlands Tel

NRG POSITIONING PAPER SERIES

RELATIVE PERFORMANCE EVALUATION INMANAGEMENT CONTROL

Hilco J. van EltenMay 2007 no.07-01Nyenrode Research Group

NRG POSITIONING PAPER SERIES

Relative Performance Evaluation in

Management Control

Hilco J. van Elten

May 2007 NRG POSITIONING PAPER NO. 07-01

NRG

The Nyenrode Research Group (NRG) is a research institute consisting of researchers from Nyenrode Business Universiteit and Hogeschool INHOLLAND, within the domain of Management and Business Administration.

Straatweg 25, 3621 BG Breukelen P.O. Box 130, 3620 AC Breukelen The Netherlands Tel: +31 (0) 346 - 291 696 Fax: +31 (0) 346 - 291 250 E-mail: [email protected] This is a positioning paper and meant to stimulate discussion on the author’s behalf. NRG positioning papers can be downloaded at http://www.nyenrode.nl/research/publications

mailto:[email protected]

http://www.nyenrode.nl/research/publications

Abstract Economic theory predicts that performance evaluation gains meaning and accuracy if the performance is compared to peers, as described by Relative Performance Evaluation theory (or: RPE). RPE can be deployed as a means for standard-setting based on peer group performance, by incorporating the performance of a reference group of agents in the compensation plan. Nonetheless, research on RPE suggests that peer comparison is not often a part of the performance evaluation, at least at the CEO level. This research proposal argues that RPE needs also to be studied at lower echelons. Amongst business unit-managers, we believe, RPE can make a significant contribution to opportunism-mitigation. By externally determining the performance standards, targets are less easily influenced by the managers whose compensation depends on them. This extends the interpretation of RPE, a phenomenon that has been analysed primarily from the perspective of efficient risk sharing and informativeness. Building on the literature on RPE, the objective of this research proposal is to study the incidence and form of relative performance evaluation at the business unit level, and to explore contingencies that are associated with empirical (non)-existence of RPE. Extant RPE models do not consider the influence of contingency factors on RPE’s applicability or desirability. This proposal presents a preliminary contingency model which aims to further our understanding of RPE. An additional contribution of this study derives from its direct, survey based approach, as opposed to the mostly indirect, public data based prior research. This allows for a deeper examination of RPE in practice, yielding stronger tests. Keywords Relative Performance Evaluation; Management Control; Business units; Contingency approach; Survey Research; Research Proposal Address for correspondence Drs. Hilco J. van Elten Nyenrode Business University Centre for Accounting, Auditing and Control Straatweg 25 3621 BG Breukelen The Netherlands Tel +31 346 291311 Fax +31 346 291250 Email [email protected] Web www.nyenrode.nl

mailto:[email protected]

NRG 07-01 Relative Performance Evaluation in Management Control – Hilco van Elten - 3 of 24

1. Introduction

One traditional tenet of agency theory is that it makes good economic sense to tie

managerial compensation to performance. Such a compensation scheme would help to

align the goals of managers with those of their principals, and would help to induce the

level of managerial effort the principal seeks. Prior research in incentives has focused on

the choice of performance measures and pay-performance sensitivities, but has largely

ignored standard setting as such, i.e. how firms decide on the level of performance

required to attain one’s bonus (Ittner and Larcker, 2001; Merchant et al., 2003, Anderson

et al, 2006). This is unfortunate, because the performance standard choice is an important

and consequential dimension of the incentive structure (Murphy, 2001).

There are two main ways in which standards can be set (Murphy, 2001): (1) they can be

determined in some internal, administrative process; or (2) they can be externally

determined. The difference between these alternatives lies in the extent to which agents

can influence the standard-setting process. Internally determined standards include

standards based on prior-year performance and standards derived from company plans

or budgets. Such standards are affected by managers’ actions, and can have dysfunctional

effects. For example, when standards are based on prior-year performance, managers

have incentives to avoid unusually positive outcomes, because good performance in the

current period is penalized through an increased standard in the next period. In a similar

vein, budget-based standards provide incentives to negotiate easy standards and

disincentives to beat the budget, especially in a regime of incremental budgeting. In

contrast, externally determined standards are less affected by managers’ actions. For

instance, managers can hardly influence the performance of an industry peer group, and

therefore standards based on such external benchmarks leave less room for managerial

opportunism.

These considerations suggest that the choice of the performance standard matters. More

particularly, they indicate that contractual efficiency can be improved by incorporating

the performance of a reference group of agents in the compensation plan. This extends

the interpretation of relative performance evaluation (RPE), a phenomenon that has

been analysed primarily from the perspective of efficient risk sharing and informativeness

(Holmström, 1982). Extant RPE models argue that as measured performance is almost


always affected by random factors beyond the agent’s control, contracts based solely on

observed individual outcomes expose the agent to substantial risk. However, to the

extent that these factors are common to all firms in an industry or market, RPE allows

the associated uncertainty to be filtered out. RPE, then, provides managers with an

incentive to perform well, while insulating their compensation from the effect of

environmental conditions that also affect the performance of other managers in the same

industry or market. To this reasoning, the current discussion adds considerations of an

opportunism-mitigating kind. This extended version of RPE seems to provide a

compelling theoretical framework for understanding managerial compensation.

This study builds on and extends the literature on RPE –both theoretically and

empirically. Its theoretical contribution lies in the previously introduced opportunism-

mitigation perspective. Furthermore, the research proposal takes an contingency

approach to understand the use of RPE. The empirical contribution relates to this

research’s use of new data sources, namely detailed data on RPE at the business unit-

level. This new data is due to this study’s survey approach; where prior studies mainly

analyzed data from publicly available databases, this study can take a more in-depth look

at RPE-incidence and the variety of forms.

The remainder of this research proposal is organised in four sections. Section 2 reviews

prior research on relative performance evaluation and describes a number of research

opportunities. The research question is presented in section 3. The fourth section of this

research proposal provides the theoretical background of RPE, the preliminary model

and the development of the hypotheses. This proposal concludes with a presentation of a

summary research plan.

2. Research opportunities Despite its promising theoretical basis, the empirical evidence for RPE is at best

inconclusive. Both Jensen and Murphy (1990) and Aggarwal and Samwick (1999) find

that RPE is not a prominent feature of executive compensation contracts. Antle and

Smith (1986) find some weak support for RPE, but they also present results that are

inconsistent with RPE’s implications. Janakiraman et al.’s results (1992) are generally

negative. Gibbons and Murphy (1990) report strong support, but the interpretation of

their results has been criticized quite severely by Janakiram et al. (1992). Overall, the lack


of strong empirical evidence suggests that RPE models have limited descriptive validity.

However, for a number of reasons, I believe that studying RPE can be a fruitful

endeavour for understanding performance evaluation practices. There are opportunities

of a theoretical and a methodological nature. The remainder of this paragraph describes

these opportunities.

2.1 Studying RPE at the business unit-level

Prior research almost exclusively studies RPE at the executive level. But, for a number of

reasons, RPE might not find its ultimate application amongst CEOs. The first reason is

of an opportunism-mitigating kind: RPE at the business-unit level can aid the standard

setting process, by comparing the actual performance to peer-output. Because the target

levels are externally determined, managers cannot easily influence the height of the

performance-standard and negotiate easy target levels for the next period. This leaves less

room for managerial opportunism. Furthermore, RPE-based standard setting helps to set

the target at a realistic level, dealing with unforeseen and unanalyzable circumstances.

With RPE, the principal ex-ante tells the agent what his relative goal will be (e.g.: ‘be the

best’, or ‘perform above average’), but what this means in absolute numbers is

determined ‘after the fact’ (via ex-post comparison with peers), This guarantees a certain

‘realism’ of the target, because it is determined by similar firms or divisions who realize

their targets under the same circumstances. This helps to set challenging but attainable

targets, which enables the organization to motivate its managers with targets.

At the CEO level, standard setting seems to be less of an issue. Here, incentives can be

directly coupled to the interests of the principal (or: shareholders), without the use of

performance standards. This can be accomplished by tying evaluation and compensation

directly to firm value, for example through stock returns. This would provide –potentially

strong– incentives for the CEO to act in the principal’s best interest; to pursue increases

in firm market value. Furthermore, these incentives would be undistorted because they

can hardly yield too much of ‘a good thing’ (namely: increases in firm value). This is

different at the business unit-level, where standards often also function as instrument to

coordinate activities (Leone and Rock, 2002). For example, in case of intra-firm


interdependencies, where rewarding its manager based on some proxy for business unit-

market value could lead to suboptimalization.1

The existence of other means than standard-setting to provide strong incentives does not

disqualify RPE as such for executive compensation. But it can lower RPE’s incidence at

the executive level whilst RPE might still be present at lower echelons.

The second reason relates to the extant efficient risk-sharing perspective on RPE. As

argued in the introduction paragraph, RPE can remove common (e.g. market-specific)

risk from the agent’s compensation. But, it is entirely possible that (relatively wealthy)

executives can undo this undesired market exposure from their incentive contracts

themselves, by hedging this common risk with their own investment portfolios. If so, the

need for a, potentially costly, RPE-contract decreases. Therefore, the possibility of

managerial hedging may limit the reliance on RPE. In support of this argument, Garvey

and Milbourn (2003) found that although RPE is not an important feature in the

compensation contract of the average manager, it is in fact present for younger and less

wealthy executives, i.e. those managers for which personal hedging is likely to be

prohibitively costly. If RPE is part of evaluation practices of executives, who are less able

to hedge their market-exposure, this strengthens the idea that RPE might very well be

part of –on average less wealthy– business unit-managers as well.

Indeed, not much support for RPE has been found at the executive level. Considering

both reasons described in this paragraph, I argue that the inconclusive findings for RPE

at the CEO level need not discourage us, if studying RPE at the business unit level.

2.2 Contingency approach

A second research opportunity lies in another extension of RPE theory, towards a

contingency approach. The extant RPE models that most prior research builds on, do

not consider the influence of contextual factors on the applicability or desirability of

relative performance evaluation. This suggests that RPE is always a good idea, regardless

of the circumstances. However, a still growing body of literature argues that

1 An example of ‘too much of a good thing’ is the case of an overzealous production department, causing congestion in the pipe-line, resulting in high stocking-costs and an overburdened distribution department.


circumstances do influence the choice of methods to assess performance, and –more

broadly– management control systems (Luft and Shields, 2003; Ittner and Larker, 2001).

‘Contingency theories’ contend that there is no universally applicable system of

management accounting and control, and that the choice of appropriate accounting and

control techniques depends upon the circumstances surrounding an organization (Ittner

and Larcker, 2001).

In support of the argument that circumstances matter, the influence of such

contingencies provides an explanation for empirical findings concerning limited RPE use

(e.g.: Bannister and Newman, 2003). A number of contingency factors are described in

the preliminary theoretical model of this proposal, and aim to add to our understanding

of RPE.

2.3 Direct testing vs. indirect testing

The third research opportunity is based on a methodological issue. Most prior studies do

not test RPE-use directly, but rely on indirect testing methods instead2. Typically, indirect

studies attempt to find support for the RPE-hypothesis by employing regression analyses

on publicly available data. Indirect studies do not study the actual contract or the

processes from which this data results. Although indirect testing has a number of

important advantages, such as the ability to gather larger samples, it also has two

drawbacks. The first downside of the indirect approach is the reduced test power. Given

the sample size, indirect tests are inevitably weaker statistically than direct tests, since the

researcher can only speculate about the details of the contract he or she studies. Then,

results’ reliability is threatened by, for example, misspecified peer groups, performance

measures, compensation metrics, etcetera (Bannister and Newman, 2003; Albuquerque,

2005).

The second disadvantage is that indirect testing is not suited to pick up detailed

information on RPE-manifestations. For example, indirect tests can not differentiate

between various forms of RPE, such as formalized, explicit or non-formalized, implicit

RPE. Furthermore, it is entirely possible that RPE mainly exists in subtler ways or by

different means than previously suggested by the literature. For example, it may be that

2 Examples of ‘indirect studies’ include: Gibbons and Murphy (1990), Janakiraman et al. (1992), Aggarwal and Samwick (1999), and Garvey and Milbourn (2003).


the budget itself or the assumptions and estimates underlying the budgeting process are

regularly being ‘marked to market’. If this is the case, it could be that the budget is

reasonably well protected against both managerial manipulation and common

uncertainty. Another possibility is a situation where, instead of peer-comparison using

objective performance measures, subjectivity underlies the performance evaluation. Then, a

manager’s performance evaluation is based on his superior’s ‘feel’ for market conditions,

and thus, the performance evaluation is subjectively linked to market situations.

Considering the potential real life variety of RPE, not only the incidence but also RPE’s

specific manifestations need studying.

Concluding on this research opportunity, exploring the RPE hypothesis with a direct

approach seems desirable, especially given the current state of our knowledge of RPE. By

employing a questionnaire, interviews or reviewing compensation contracts, direct testing

can enable more sophisticated testing of the RPE hypothesis, and go more in-depth.

2.4 Conclusion

Overall, there is currently not much empirical support for the RPE hypothesis. But, a

number of promising opportunities for further research exist. In this paragraph,

opportunities of both theoretical and methodological backgrounds have been discussed.

This study aims to build from these opportunities, seeking further support through

theory development and previously unapplied methods to the topic.

3. Research question

This research proposal aims at theoretically and empirically examining the use of relative

performance evaluation at the business unit-manager level, and associating the use of

RPE with contingency factors. This goal is made explicit in the following research

question:

Can we explain RPE-incidence at the business unit-manager level by associating

incidence with contingency factors?

The proposal contains preliminary theory on when and why RPE is used, associated with

the organizational context (uncertainty, information asymmetry, asset specificity and


managerial power). Besides the theory-driven research question, this study is also partially

exploratory in nature. Besides understanding RPE-incidence, the study wants to explain

the variety of forms of RPE in practise. Concerning this part of the study no preliminary

theory is presented. Currently, the literature does not provide a sufficiently reliable

foundation to build such a theory. Instead, the aim is here to map the forms of RPE, and

gain insights to provide us with building blocks for future research.

4. Theory 4.1 From an insurance perspective to opportunism-mitigation

Performance evaluation is considered to gain meaning and accuracy if the agent’s

performance is compared to the performance of a reference group, in case the agents’

performances are affected by common factors (for example, see: Holmström, 1982;

Janakiraman et al., 1992). This phenomenon has been termed in literature relative

performance evaluation or RPE. This research proposal views RPE as a means for

determining the performance standard and opportunism-mitigation. This approach is

new to the RPE literature. To the best of my knowledge, prior studies exclusively

approached the RPE phenomenon from an informativeness and efficient risk sharing

perspective (see: Holmström, 1982).

Extant RPE models argue that –because measured performance is almost always affected

by factors beyond the agent’s control– contracts based solely on observed individual

outcomes expose the agent to substantial risk. However, when these risk-inducing factors

are partially typical to a reference group, RPE-based contracts would filter out these

uncertainties. This is how the performance of peers is informative about the agent’s

actions; by filtering out common external –often uncontrollable– factors, relative

performance evaluation provides the principal with better insights in the agent’s true

effort (Janakiraman et al., 1992). The principal and the agent now can write a more

efficient contract, with both parties being less exposed to risk.

However, the preliminary theory presented here views RPE from a different angle, that

of opportunism-mitigation and target-setting, where RPE contributes to the

determination of performance-standards. This approach comes into play at lower

hierarchical levels, especially at the business unit level. At the business unit-level,


performance standards (in the form of budgets and targets3) play an important role in

management control. Despite their central place in management accounting and control

practise, budgets and targets are widely known to potentially cause severe damage to

organizations (Onsi, 1973; Hansen et al., 2003; Jensen, 2003). Budgets’ and targets’

negative effects on control have often been traced back to dysfunctional performance

measures and non-linearities in the pay-performance relation. Less studied, but also a

cause of these effects, is the performance standard, and how it is determined. Literature

claims that there is a limited number of ways to determine performance standards (see,

for example: Milgrom and Roberts, 1992; Murphy, 2001). Murphy (2001) distinguishes

two main approaches. He argues that targets can either be 1) internally determined

(amongst others, standards based on budgets, prior-years performance, or board

discretion) or 2) externally determined, for example where standards are compared to a

benchmark.

The main difference between internally and externally determined standards lies in the

extent to which managers can influence their performance standard. Internally

determined standards lie in the sphere of influence of the manager. This can lead to

severe dysfunctional behaviour. For example, when performance standards are budget-

based, this generally provides incentives to negotiate easy standards for next year by

gaming the system, instead of beating the budget, especially in case of incremental

budgeting and ratchet effects.

In contrast to internally determined standards, externally determined standards are not

(as easily) affected by managerial actions. Consider the case of an RPE-based contract,

where contemporaneous performance is compared to peer-performance: the

performance standard is externally defined ex-post by ‘the market’. This market consists

of a reference group with agents whose performances are affected by the same

uncertainties, and thus, who are exposed to the same risk (preferably both in nature and

level). Since peer performance is relatively hard to influence for the manager, the

manager cannot influence the performance standard to which he is evaluated. This

3 When, in this research proposal, the term budget is used, it refers to budgets deployed as targets. Other reasons-to-budget are, for example, for operational planning, communication of goals, and strategy formulation (Hansen and Van der Stede, 2004)


complicates engaging in dysfunctional behaviour, and mitigates (from the principal’s

perspective) undesired managerial opportunism.

4.2 Provision of powerful incentives and an exploration of a coordination role.

This paper views RPE as a means to induce motivation of organizational members (in

this study specifically: business-unit managers)4. This research proposal claims that RPE

can be helpful to motivate managers, because it can aid the calibration of the incentive-

zone in performance-contracts, where the managers performance will lead to a bonus.

The motivating character of targets in compensation schemes has been studied

extensively. When targets are used to motivate managers, typical dysfunctionalities arise.

These dysfunctionalities are mostly related to the calibration of the incentive-zone.

Performance evaluation (and subsequent rewarding) motivates to provide high effort, as

long as the manager believes his effort has an effect on his rewards. Such compensation

systems only have strong motivational properties when the actual performance lies in the

incentive zone, because only then, the manager will receive a variable bonus, dependent

on his performance. Consider the compensation scheme in figure 1 (conform: Murphy,

2001; Jensen, 2003) The incentive-zone refers to the part of the reported performance,

for which the manager receives a bonus. In figure 1, this zone is marked as I. In this area,

the system is motivating the manager to provide high effort, without increased risk of

managerial opportunism.

Around the edges of the incentive-zone (indicated as zones II.1 and II.2), the threat of

opportunistic behaviour increases. The system strongly induces gaming. In zone II.1,

reporting a higher then actual performance is stimulated by the system, in order to just

reach the hurdle bonus. In zone II.2, it serves the manager’s interests better to not report

4 Management control instruments in general, and targets specifically, can also be deployed as means for coordination of organisational activities. Concerning this other role of management control instruments, RPE’s fitness for coordinating activities is assumed to be low. Since RPE abandons preset absolute performance goals –which enable forecasting required capacities– the planning and allocation of resources becomes more difficult. Especially for organisations with much or complex interdependencies (reflected in high asset specificity), this is likely to lower reliance on relative performance evaluation. The link between interdependencies and asset specificity will be discussed in the operationalization of the model (not included in this research proposal).


performance that exceeds the target with more then –in the example– 20 percent. Again,

fraudulent reporting is induced. The zones marked III.1 and III.2 do not have

motivational properties, since the compensation is not dependent on (small) changes in

effort. In this case, dysfunctional behaviour will not take form as fraudulent reporting,

but agents are likely to shirk.

Figure 1: typical compensation scheme

If an organization chooses to use performance evaluation for the provision of incentives,

it is critical to calibrate the performance standard, so that realized performances lies

within the boundaries of the incentive zone. However, it is difficult to ex-ante define the

appropriate height of the performance standard, especially in case of external

shocks/uncertainties.

Relative performance evaluation aids the calibration of the incentive zone by an ex-post

determining the target, comparing with what was attainable for the reference-group. This

solves the main part of the problem: the attainable performance lies within the

boundaries of the incentive zone, hence the manager receives incentives to maximize his

effort. Nonetheless, the performance that is feasible for the reference-group, is not

necessarily feasible for the evaluated manager. In case of highly idiosyncratic risk, RPE-


based contracts may still be too noisy. Then, actual performance can still lie within

‘danger-zones’ II.1 or II.2, or even in III, since RPE can only correct the performance

standard for shared uncertainties, and not for firm-specific risk. Nonetheless, in many

cases, RPE can make a contribution to opportunism-mitigation, depending on the

circumstances. The circumstances which are considered to be critical to RPE-use are

discussed in the next paragraph.

4.3 Contingency approach

This study introduces a preliminary framework of contingency factors that furthers our

understanding of RPE at the business unit-manager level. Contingency theories contend

that there is no universally applicable system of management accounting and control, and

that the choice of appropriate accounting and control instruments or techniques depends

upon the circumstances surrounding an organization (Ittner and Larcker, 2001).

Prior RPE literature generally did not take on the position that circumstances matter, but

rather builds on the assumption that performance evaluation always benefits from peer-

comparison, regardless of the organizational context. In contrast to prior studies, this

research proposal presents preliminary theory to argue that certain contextual factors

influence whether or not organizations deploy RPE. Although the literature associates

numerous contingency factors with management control choices5, the theory presented

here deliberately describes only a small number of factors, which are fundamental to

RPE-use. Why I believe that specifically these factors describe the core of the problem,

will be argued in the next sub-paragraphs.

Figure 2 illustrates the associations between the RPE-incidence and the contingency

variables. The remains of this paragraph describes the individual contingency factors.

First, the independent variables which primarily relate to firms’ need for RPE will be

discussed (the desirability variables). Subsequently, two ability-decreasing factors will be

presented. The latter set of variables are modelled as moderators, as shown in the figure

below.

5 See Chenhall (2003) for an overview of the usual suspects in contingency research.


Environmentaluncertainty

Informationasymmetry

RPE-use

Positive (H1)

Positive (H2)

Assetspecificity

Managerialpower

Negative (H3b)

Negative (H4a)

Negative (H4b)

Desirability

Ability

Negative (H3a)

Figure 2: theoretical model concerning RPE-incidence and contingency variables

The desirability variables

Target setting and motivating managers would be significantly easier in a world of perfect

information and certainty, where future events would not surprise us, and where the

principal would have unlimited access to the specific knowledge of the agent.

Unfortunately, most organizations do not operate in a hyperrational Nirwana, and

therefore need instruments to face the challenges of unforeseen externalities and

incomplete, costly information. Both desirability variables (environmental uncertainty

and information asymmetry) can cause a shortage of information: the principal will have

insufficient information to determine an attainable performance standard. This

potentially leaves room for managerial opportunism. Because RPE can provide

informative performance standards, it makes a contribution to solving this information

gap. Already in the earliest RPE papers, RPE set out to address these two factors,

described in terms of informativeness and (common) uncertainty (Holmström, 1982).

More specifically, Holmström describes how peer-comparison can be informative –and

hence, reduce the information insufficiencies concerning the attainability of performance

standards– by filtering out common uncertainty. Environmental uncertainty and

information asymmetry form the basis of the desirability of relative performance

evaluation. (conform: Holmström, 1982; Murphy, 2001).


Environmental uncertainty

Uncertainty can have many origins, of which the external environment is amongst the

most widely researched aspects. Uncertainty is present in situations in which probabilities

cannot be attached to future events, or when even the future events themselves may not

be predictable (Chenhall, 2003). Uncertainty is found to make up front programming

difficult, and high levels of uncertainty require maintaining flexibility to allow adaptation

to events as they unfold and to information as it accrues becomes imperative (Speklé,

2001; 2004).

Uncertainty is often due to the external environment. So called environmental

uncertainty is a powerful contextual variable that is at the foundation of contingency-

based research (Chenhall, 2003). It is this specific form of uncertainty which’s

consequences are addressed by relative performance evaluation. RPE is only suited to

filter out common risk. This common nature of the uncertainties is essential for RPE for

informativeness-reasons (Holmström, 1982); because also the performance of peers is

influenced by these common shocks, common risk can inform the principal the

circumstances under which the agent had to perform. Only external circumstances are

shared by peers, and affect the performance of all reference-group members.

Indeed; also other shades and colours of uncertainty can ‘cause problems’ by

complicating forecasting, hinder target setting, and affect performance, but only

environmental uncertainties are common, and can be reduced by RPE.

In a highly uncertain environment, RPE not only gains desirability, but non-RPE types of

performance evaluation also lose theirs. This is because performance evaluation based on

traditional, ex-ante determined targets requires forecasting, and forecasting becomes

more difficult with increasing levels of uncertainty, since it involves highly uncertain

future circumstances. This seriously reduces the applicability of non-RPE types of

performance evaluation. But, RPE does not require forecasting, since it relies on ex-post

target setting. As opposed to traditional, absolute performance evaluation, RPE is better

able to deal with an unpredictable, uncertain environment. Therefore, with high

environmental uncertainty, more use of RPE is expected. Hence, the first hypothesis is

the following:


H1: High levels of environmental uncertainty positively influence the use of relative performance

evaluation.

Information asymmetry

Information asymmetry, relating to the information gap between principal and agent, is

often described (both theoretically and empirically) to be the core of agency problems: it

is the combination of information asymmetry and the agent’s aversion both to work and

risk that steer him away from cooperative behaviour (Holmström, 1982; Kunz and Pfaff,

2002). This sub-paragraph argues that RPE can make a substantial contribution to

solving the target-setting problems caused by information asymmetries.

Setting an appropriate and reliable target can be problematic when a subordinate has

better information than a superior about factors that influence performance, especially

when the agents pay depends on his performance vis-à-vis the target (conform: Chow et

al., 1988). These information asymmetries occur when lower-level managers have specific

knowledge about the functioning of the division that is either unavailable to corporate

management, or is too costly for the principal to obtain (Abernethy et al., 2004; Christie et

al., 2003; Dunk, 1993).

In contrast to situations of high uncertainty –where both the principal and the agent

cannot judge the attainability of a performance target up front–, high levels of

information asymmetry hinder only the principal6. The agent –who is considered to have

superior insights about the mechanisms that drive performance– might rely on self-

interest seeking behaviour with guile, which can result in misrepresentation towards the

principal and building budgetary slack (Dunk, 1993; Chow et al., 1988, Fisher et al., 2002).

Through relative performance evaluation, the principal can let the peer-group ‘decide’ the

height of the performance standard, without the need to understand the ins and outs of

the business unit and the factors that drive its performance. This way, the principal works

6 Information asymmetry can go both ways: it is also possible that the principal has information about the business unit, which the agent has not. However, this study specifically looks into information asymmetries where the agent has superior information (conform: Abernethy et al., 2004)


around the business unit to assess its manager’s effort, instead of obtaining the –costly or

just unavailable– information required to in-depth evaluate the business unit.

The prediction this study makes is that RPE will be used especially amongst

organizations facing high information asymmetry. Therefore, the effect of information

asymmetry is explicated as:

H2: Information asymmetry positively influences the use of relative performance evaluation.

The ability variables

Whereas the previous section described ‘desirability-enhancing’ factors for organizational

reliance on RPE, this section discusses some important necessary conditions for RPE-

incidence. These factors, both from economical and behavioural backgrounds, are

expected to disable the use of RPE for target setting and motivational purposes, and are

modelled as moderating variables.

Both factors are, similar to the desirability variables, argued to be at the core of RPE. Asset

specificity relates to the level of comparability amongst agents. Without comparability (or

here: with high levels of asset specificity), no meaningful, informative comparison

amongst agents can be made. The managerial power variable has a broader background;

this factor does not solely stem from agency theory or contingency literature, but is also

supported by behavioural literature and wide recognition by practitioners. Especially the

latter argue the urgency of the inclusion of managerial power when explaining

performance evaluation and compensation.

Asset specificity

Asset specificity concerns the (opportunity costs of the) investments made to support the

products, activities, or processes of a business unit (conform: Speklé, 2004). Based on

Transaction Cost Economics reasoning, Speklé (2001; 2004) argues that the

informativeness of externally determined performance standards depends on asset

specificity; the lower the degree of asset specificity, the better these standards represent

efficient performance. If a division produces highly specific products or services, or relies

heavily on specific processes, the division’s uniqueness reduces the comparability of

performance with other firms or business units. This is problematic for the appliance of

RPE, since comparability is at the very heart of RPE. A sufficient number of reasonably


comparable peers is needed to build an objective point of reference to form the target

level to which the agent’s performance is measured.

In other words, only a large enough reference-group can substitute the market including

its external shocks, which RPE aims to eliminate from the agent’s performance.

Although the principal might want to tie the targets to the market, high levels of asset

specificity can frustrate the much needed comparability. As a result, internally determined

target setting might be the only feasible option.

But, asset specificity does not drive RPE-incidence in a direct manner. The sheer ability to

compare with peers (i.e.: low asset specificity) does not cause peer-comparison, it just

makes it possible. Therefore, no main effect between asset specificity and RPE-use is

hypothesized. Instead, this study predicts moderating effects of asset specificity on the

two main relations hypothesized in H1 and H2. High values of asset specificity will

weaken the relation between environmental uncertainty and RPE-use, and between

information asymmetry and RPE-use.

Therefore, the claim is that asset specificity negatively influences reliance on RPE, not

directly but only via environmental uncertainty and information asymmetry, formalized

as:

H3a: Asset specificity negatively influences the positive relation between environmental uncertainty and the

use of relative performance evaluation.

H3b: Asset specificity negatively influences the positive relation between information asymmetry and the


Managerial power

Managerial power is defined here as the ability of managers to influence or exert their will

or desires to others, for example on the remuneration decisions made by the board of

directors, or by the compensation committee of the board. This definition is in line with

Finkelstein (1992), Lambert et al. (1993), and Grinstein et al. (2004), who generally find

significant and positive effects of managerial power on CEO compensation. These

empirical findings are theory-consistent: under the managerial power approach, there


should be a positive correlation between the level of compensation and the level of

managerial power in the firm; managerial skill and performance should play a secondary

role in explaining the variability in compensation (Grinstein et al., 2004).

However, two main differences exist between the application of the managerial power

approach sketched above, and the view which this proposal adopts. Firstly, the

managerial power approach has been exclusively studied at the top-management level,

whereas this research proposal has a different level of analysis. This study applies

managerial power at the business-unit level, where nonetheless, I assume, similar forces

are at work. Nonetheless, it is likely that lower in the hierarchy managerial power levels

are significantly lower.

Secondly, prior studies deploy managerial power to explain the height of compensation,

not the structure of the compensation contract behind it. Explaining the height of

compensation seems evident, given the unwritten assumption that everyone –including

CEOs and business unit-managers– prefers more compensation over less. However, this

study is concerned with explaining RPE-use, instead of the height of compensation. It

poses a different assumption.

This research proposal argues that managers generally prefer a non-RPE based contract

over an RPE-based contract. RPE is not necessarily a manager’s first best solution,

because of the risk it exposes the manager to (who is generally considered to be risk-

averse). Although RPE is can shield the manager from risk, compared to a ‘regular’

performance-based compensation contract (as discussed in paragraph 4.1 from an insurance

perspective to opportunism-mitigation), it is not as secure as a fixed-pay contract. This lowers

the attractiveness of RPE from a managerial point of view. Also, managerial preference

for non-RPE based contracts receives support from the behavioural literature.

Fershtman et al. (2003) also claim that managers prefer non-RPE contracts, whereby the

authors rely on the human tendency to derive utility from earning more than others. The

general idea is that managers care not only about their own wage, but also prefer their

compensation to be positively related to the wages of other managers (Fershtman et al.,

2003). This suggests that –for as far as it lies within their power– managers will negotiate

a contract that yields more compensation than their peers, regardless the circumstances.


Casual conversation with managers and compensation consultants7 provides support for

these ideas. Interviewees argue that at top and middle-manager levels, managerial power

negatively influences the use of RPE.

The applicability of relative performance evaluation –as a means for standard setting and

remuneration– is seriously weakened when managers ultimately influence their own pay

via managerial power instead of effort. Managerial power would allow them to negotiate

a contract which makes this possible. Therefore, RPE will lose it motivating properties

when managerial power is high, despite it still provides an objective performance target.

Similar to the discussion about asset specificity, managerial power does not cause RPE-

incidence directly. Again, this study predicts no main effect between this managerial

power and RPE-use. Instead, this study hypothesizes moderating effects of managerial

power on the two main relations hypothesized in H1 and H2. High values of managerial

power will weaken the relation between environmental uncertainty and RPE-use, and

between information asymmetry and RPE-use.

Consequently, the following hypotheses are formulated:

H4a: Managerial power negatively influences the positive relation between environmental uncertainty and

the use of relative performance evaluation.

H4b: Managerial power negatively influences the positive relation between information asymmetry and the


5. Research plan The aim of this research proposal is to highlight the relevance of studying RPE at lower

hierarchical levels, and to build theory that helps us to explain the (non-)use of RPE at

the business unit-managers. Such a preliminary theory is presented in paragraph 4.

Furthermore, as brought forward in the research question, this study sets out to explain

the variety of forms of RPE in practise. This part of the research question will be studied

inductively, since no solid theoretical expectations can be build from prior literature.

7 These conversations include discussions with the CEO of a major Dutch housing association and an executive compensation specialist from Towers Perrin.


To answer the research question, this project involves a quantitative, large sample

empirical study. Given the paucity of prior research in this area, it is unlikely that I can

rely fully on existing datasets. To ensure a substantial and high-quality response, I plan to

administer a survey through the NIVRA-Nyenrode network of professional controllers

(current students of the MSc in Controlling programme). Recent experience shows that

this is in fact a feasible solution to mitigate response problems.


References Abernethy, M.A., Bouwens, J., and Van Lent, L. 2004. Determinants of control system

design in divisionalized firms. The Accounting Review, 79: 545-570. Aggarwal, R.K. and Samwick, A.A. 1999. Executive compensation, strategic competition,

and relative performance evaluation: Theory and evidence. Journal of Finance, LIV: 1999-2043.

Anderson, S.W., Dekker, H.C., and Sedatole, K.L. 2006. The impact of pay-for-performance and negotiated goal setting on performance-to-goal, goals and performance. Working paper.

Albuquerque, A. 2005. Who are your peers? A study of Relative Performance Evaluation. Working paper.

Antle, R. and Demski, J.S. 1988. The controllability principle in responsibility accounting. The Accounting Review, LXIII: 700-718.

Antle, R. and Smith, A. 1986. An empirical investigation of the relative performance evaluation of corporate executives. Journal of Accounting Research, 24: 1-39.

Bannister, J.W. and Newman, H.A. 2003. Analysis of corporate disclosures on Relative Performance Evaluation. Accounting Horizons, 17: 235.246.

Chenhall, R.H. 2003. Management control systems design within its organizational context: findings from contingency-based research and directions for the future. Accounting, Organizations and Society, 28: 127–168.

Chow, C.W., Cooper, J.C., and Waller, W.S. 1988. Participative budgeting: effects of a truth-inducing pay scheme and information asymmetry on slack and performance. The Accounting Review, 63: 111-122.

Dunk, A.S. 1993. The effect of budget emphasis and information asymmetry on the relation between budgetary participation and slack. The Accounting Review, 68:400-410.

Fershtman, C., Hvide, H.K., and Weiss, Y. 2003. A behavioural explanation of the relative performance evaluation puzzle. Annales d'économie et de statistique, 71-72: 349-361.

Finkelstein, S. 1992. Power in top management teams: dimensions, measurement, and validation. Academy of Management Journal, 35: 505-538.

Fisher J.G., Maines L.A., Peffer S.A., and Sprinkle G.B. 2002. Using Budgets for Performance Evaluation: Effects of Resource Allocation and Horizontal Information Asymmetry on Budget Proposals, Budget Slack, and Performance. The Accounting Review, 77: 847-865.

Garvey, G. and Milbourn, T. 2003. Incentive compensation when executives can hedge the market: Evidence of relative performance evaluation in the cross section. Journal of Finance, LVIII: 1557-1581.


Gibbons, R. and Murphy, K.J. 1990. Relative performance evaluation for chief executive officers. Industrial and Labor Relations Review, 43: 30S-51S.

Grinstein, Y., and Hribar, P. 2004. CEO compensation and incentives: evidence from M&A bonuses. Journal of Financial Economics, 73: 119–143.

Hansen, S.C., Otley, D.T., and Van der Stede, W.A. 2003. Practice developments in budgeting: an overview and research perspective. Journal of Management Accounting Resrach, 15: 95-116

Hansen, S.C., and Van der Stede, W.A. 2004. Multiple facets of budgeting: an exploratory analysis. Management Accounting Research, 15: 415-439.

Holmström, B. 1982. Moral hazard in teams. Bell Journal of Economics, 13: 324-340. Horngren, C. T., G. Foster, and S. M. Datar. 2000. Cost Accounting: A Managerial

Emphasis. Upper Saddle River, NJ: Prentice Hall. Ittner, C.D. and Larcker, D.F. 2001. Assessing empirical research in managerial

accounting: a value-based management perspective. Journal of Accounting and Economics, 32: 349-410.

Janakiraman, S.N., Lambert, R.A., and Larcker, D.F. 1992. An empirical investigation of the relative performance evaluation hypothesis. Journal of Accounting Research, 30: 53-69.

Jensen, M.C. and Murphy, K.J. 1990. Performance pay and top-management incentives. Journal of Political Economy, 98: 225-264.

Jensen, M.C. 2003. Paying people to lie: the truth about the budgeting process. European Financial Management, 9: 379-406.

Kunz, A.H. and Pfaff, D. Agency theory, performance evaluation, and the hypothetical construct of intrinsic motivation. Accounting, Organizations and Society, 27: 275-295.

Lambert, R.A., Larcker, D.F., and Weigelt, K., The structure of organizational incentives. Administrative Science Quarterly, 38: 438-461.

Leone, A.J. and Rock, S. 2002, Empirical tests of budget ratcheting and its effect on managers’ discretionary accrual choices, Journal of Accounting and Economics, 33: 43-68.

Luft, J. and Shields, M.D. 2003. Mapping management accounting: graphics and guidelines for theory-consistent empirical research. Accounting, Organizations and Society, 28: 169-249.

Merchant, K.A., Van der Stede, W.A., and Zheng, L. 2003. Disciplinary constraints on the advancement of knowledge: the case of organizational incentive systems. Accounting, Organizations and Society, 28: 251-286.

Moers, F. 2006. Performance Measure Properties and Delegation. The Accounting Review, 81: 897–924.

Milgrom, P., and Roberts, J. 1992. Economics, Organization & Management. Prentice Hall, New Jersey.


Murphy, K.J. 2001. Performance standards in incentive contracts. Journal of Accounting and Economics, 30: 245-278.

Onsi, M. 1973. Factor analysis of behavioural variables affecting budgetary slack. The Accounting Review, 48: 535-548.

Speklé, R.F. 2001. Explaining management control structure variety: A transaction cost economics perspective. Accounting, Organizations and Society, 26: 419-441.

Speklé, R.F. 2004. Configurations of control: A transaction cost approach. Working paper.

Nyenrode Research Group

Nyenrode Business UniversiteitNyenrode Research Group

Straatweg 25Postbus 130, 3620 AC BREUKELEN

T +31 346 291 696F +31 346 291 250E [email protected]

www.nyenrode.nl/nrg

Documents

Relative Performance Evaluation in Management Control · Nyenrode Business University Centre for Accounting, Auditing and Control Straatweg 25 3621 BG Breukelen The Netherlands Tel