Person focused pay plan

Human Resource Management, July–August 2016, Vol. 55, No. 4. Pp. 697–719

© 2015 Wiley Periodicals, Inc.

Published online in Wiley Online Library (wileyonlinelibrary.com).

DOI:10.1002/hrm.21740

Correspondence to: Sanghee Park, Assistant Professor, School of Management and Labor Relations, Rutgers,

The State University of New Jersey, Piscataway, NJ, 08854, Phone: (848) 445-1051, Fax: (732) 445-2830,

E-mail: [email protected]

EVALUATING FORM AND

FUNCTIONALITY OF

PAY-FOR-PERFORMANCE PLANS:

THE RELATIVE INCENTIVE

AND SORTING EFFECTS OF MERIT

PAY, BONUSES, AND LONG-TERM

INCENTIVES

S A N G H E E P A R K A N D M I C H A E L C . S T U R M A N

Using two-year longitudinal data from a large sample of US employees from a

service-related organization, the present study investigates the relative effects of

three forms of pay-for-performance (PFP) plans on employees’ job performance

(incentive effects) and voluntary turnover (sorting effects). The study differenti-

ates between three forms of pay: merit pay, individual-based bonuses, and long-

term incentives. By defi nition, these PFP plans have different structural elements

that distinguish them from each other (i.e., pay plan form) and different charac-

teristics (functionality), such as the degree to which pay and performance are

linked and the size of the rewards, which can vary both within and across plan

types. Our results provide evidence that merit raises have larger incentive and

sorting effects than bonuses and long-term incentives in multi-PFP plan environ-

ments where the three PFP plans are operating simultaneously. Only merit pay

has both incentive and sorting effects among the three PFP plans. The implica-

tions for the PFP-related theory, as well as for the design and implementation of

PFP plans, are discussed. © 2015 Wiley  Periodicals, Inc.

Keywords: pay-for-performance plans, incentive effect, sorting effect, compensation

698 HUMAN RESOURCE MANAGEMENT, JULY–AUGUST 2016

Human Resource Management DOI: 10.1002/hrm

Even with a

substantial body of

research discussing

the effects of

particular PFP

interventions, PFP

research has typically

failed to consider the

complex, multiplan

environments

in which many

organizations invest

and many employees

face.

examinations of individual PFP plans. Most stud- ies examine a single form of PFP at a time, par- ticularly in laboratory studies (e.g., Bandiera, Barankay, & Rasul, 2007; Cadsby et al., 2007; Eisenberger, Rhoades, & Cameron, 1999; Kwong & Wong, 2014), but also in organizational settings (e.g., Banker, Lee, Potter, & Srinivasan, 1996, 2001; Dunford, Boudreau, & Boswell, 2005; Eisenberger et al., 1999; Pearce et al., 1985; Schaubroeck, Shaw, Duffy, & Mitra, 2008). Other work provides only broad overviews of PFP plans’ effects, such as in strategic HR management research that typically asks general questions about the extent to which employees are covered by PFP (e.g., Bhattacharya, Gibson, & Doty, 2005; Delery & Doty, 1996; Gerhart & Milkovich, 1992; Toh, Morgeson, & Campion, 2008; Wright, Gardner, Moynihan, & Allen, 2005). This previous work has certainly been valuable for providing information on the nature of PFP effects; however, the generalizability of theory and findings from single-plan focal stud- ies to multiplan environments is questionable.

Many companies use multiple types of PFP simultaneously (Cohen, 2011; Gerhart & Fang, 2014; Gerhart et al., 2009; Rynes et al., 2005). A 2010 WorldatWork Survey showed that 92 percent of companies use merit raises, 80 percent provide some form of individual-based variable pay pro- gram (not including sales commissions or merit raises), and 57 percent use some sort of perfor- mance-sharing plan. The same survey conducted in 2012 (WorldatWork, 2012) showed this to be an increasing trend, with 95 percent offering merit pay, 84 percent with some form of individual- based variable pay program, and 58 percent using some form of performance sharing. While these surveys do not explicitly report the number of dif- ferent incentive plans covering the same employ- ees, mathematically, we can extrapolate that at least three-quarters of companies use at least two forms of PFP, and over one quarter are simultane- ously using three different PFP plans. Despite this prevalent complexity, though, there is minimal research considering the relative effectiveness of different PFP plans.

Studying PFP explicitly within the more com- plex environment of multiple PFP plans is critical to gain a better understanding of the relative effec- tiveness of different PFP forms. This study makes several contributions to our understanding of the effectiveness of PFP. First, this study considers multiple PFP plans simultaneously in multi-PFP plan environments. As previously noted, most prior PFP research has considered a single PFP plan at a time, with it either being explicitly on a single plan, or unstated or unexplored if other PFP plans were operating simultaneously. It is unclear

T heory and empirical evidence indicate that, in general, pay-for-performance (PFP) plans have positive effects on employee job performance (e.g., Gerhart & Fang, 2014; Gerhart & Rynes, 2003; Jenkins,

Mitra, Gupta, & Shaw, 1998; Lawler, 1971; Zenger, 1992). A common component of compensation systems, PFP plans are referred to as “pay that varies with some measure of individual or orga- nizational performance” (Milkovich, Newman, & Gerhart, 2013, p. 335). Theory attributes the influences of PFP plans to two broad sets of effects: incentive effects and sorting effects (Cadsby, Song, & Tapon, 2007; Gerhart & Fang, 2014; Gerhart & Milkovich, 1992; Gerhart & Rynes, 2003; Gerhart, Rynes, & Fulmer, 2009; Rynes, Gerhart, & Parks,

2005). Incentive effects represent the influence of PFP plans through employee motivation, based on the premise that PFP plans can increase employee motivation and, hence, employee performance. Sorting effects alter the composition of the workforce, in that PFP plans can affect the quality of workers who apply for jobs (Lazear, 1986; Rynes et al., 2005) and the performance level of those leaving the organiza- tion (Salamin & Hom, 2005; Shaw & Gupta, 2007; Trevor, Gerhart, & Boudreau, 1997). While there are still some examples of ineffective PFP plans (e.g., Beer & Cannon, 2004; Kahn & Sherer, 1990; Lawler, 2000; Pearce, Stevenson, & Perry, 1985; Pfeffer, 1998), the prepon- derance of evidence shows that PFP plans have positive effects (cf., Gerhart & Fang, 2014; Gerhart et al., 2009). Yet, even with a substan- tial body of research discussing the effects of particular PFP interven-

tions, PFP research has typically failed to consider the complex, multiplan environments in which many organizations invest and many employees face (Gerhart et al., 2009; Rynes et al., 2005). This lack of consideration of more multifaceted envi- ronments presents a theoretical gap for under- standing and testing how relevant PFP theories apply in more complex environments, and a practical gap for organizations needing to predict the sort of effects they should expect from their multiplan environments.

While the literature on PFP plans is quite extensive (for reviews, see Gerhart & Fang, 2014; Gerhart et al., 2009; Guthrie, 2007; Rynes et al., 2005), prior PFP research is largely based on specific

Human Resource Management DOI: 10.1002/hrm

EVALUATING FORM AND FUNCTIONALITY OF PAY-FOR-PERFORMANCE PLANS 699

What makes the study

of compensation

systems complex is

that some aspects

of PFP plans are

different by definition

(e.g., the reward

is permanent, a

one-time payment,

or will take time

before the reward is

vested), while other

characteristics can

vary both within and

across plan types

(e.g., the strength

of the relationship

between performance

and rewards, the

award size).

should differ in terms of both their incentive and sorting effects (Gerhart et al., 2009). What makes the study of compensation systems com- plex is that some aspects of PFP plans are differ- ent by definition (e.g., the reward is permanent, a one-time payment, or will take time before the reward is vested), while other characteristics can vary both within and across plan types (e.g., the strength of the relationship between performance and rewards, the award size). While prior use of theories regarding a single type of plan has typi- cally yielded general predictions that PFP plans should have posi- tive effects, such a holistic approach misses important characteristics of PFP plans and has questionable (or at least untested) generalizability to considering the simultaneous effects of multiple PFP plans. Based on pay plans’ mechanisms, we can differ- entiate between pay form and func- tionality, which can vary depending on different pay practices, thus delineating where hypotheses can be created based on the type of PFP being provided (i.e., pay plan defi- nition, or form) and those based on the specific characteristics of the PFP plan (i.e., pay plan functionality).

What’s in a Name Anyhow: The Effect of Different Pay Forms

PFP plans come in a variety of forms, both in terms of the level of the per- formance metric (e.g., individual, team, unit) and the type of award it provides (e.g., recognition, non- monetary awards, lump-sum cash awards, long-term incentives [LTIs], and permanent pay increases). It is beyond the scope of any one study to contrast every potential PFP plan, and so we begin to address the noted gap in compensation research by considering three increasingly common PFP plans: merit pay, indi- vidual-based annual performance bonuses, and LTIs. The first two of these are indi- vidually based and rewarded; the third is awarded to an individual and the size of the award depends in part on individual performance, but ultimately the value of the award depends on the overall market performance of the organization and vest- ing requirements.

Merit pay is a form of reward in which individ- uals receive permanent pay increases (i.e., raises) as a function of their individual performance

whether the predictions for individual PFP plans would generalize when it is explicitly known that other PFP plans are in operation. Furthermore, by applying PFP-related theories to the context of multiple PFP plans, we are examining a previously unexplored set of processes. It is not immediately evident what the effects of one PFP plan would be after controlling for the effects of other PFP plans, especially if the plans have related effects. This study extends PFP theories to consider the context of multiple PFP environments, where a portfolio of PFP plans cover employees.

Second, we contribute to the compensation lit- erature by considering a gap between research and practice. Practically, we inform managers about how different PFP plans should be combined to increase employee performance in their organi- zations. It is clear that organizations are invest- ing significant sums of money into multiple PFP forms (WorldatWork, 2010, 2012). With a sizable and growing number of employees being covered by two or more PFP plans, the lack of research on the relative effectiveness of such plans represents a notable gap in applicable research knowledge. While many argue that practitioners should take an evidence-based approach to management pol- icy (e.g., Rousseau, 2006; Rousseau & McCarthy, 2007), the lack of research addressing this spe- cific situation represents another notable discon- nect between research and practice (e.g., Cascio & Aguinis, 2008; Rynes, Giluk, & Brown, 2007), a particular problem in the area of compensation (Deadrick & Gibson, 2007; Rynes et al., 2007).

The purpose of this article is to apply existing PFP-relevant theory to differentiate between the effects of multiple PFP plans implemented simul- taneously. We propose that a structural approach to understanding PFP plans can be used to form predictions on the relative effectiveness of dif- ferent PFP plans for both incentive effects and sorting effects. By considering the specific char- acteristics of PFP plans, we can build theory to predict not just the general (directional) effects of PFP plans, but the relative effectiveness of plans. Furthermore, we can extend theory to the purpose of considering the simultaneous effects of multi- ple PFP plans.

A Structural Approach to Compensation Plans

Multiple types of PFP plans are often used through a combination of individual-based rewards (e.g., merit pay, lump-sum bonuses, and individual incentives) and/or group-based rewards (e.g., gain sharing, profit sharing) (Gerhart et al., 2009; Milkovich et al., 2013). Every pay form has advan- tages and disadvantages, and these programs

700 HUMAN RESOURCE MANAGEMENT, JULY–AUGUST 2016

Human Resource Management DOI: 10.1002/hrm

A pay plan’s name

reveals information

about its award,

but simply calling

something a PFP plan

does not necessarily

mean it links pay with

performance.

it links pay with performance. A number of theo- ries suggest that the strength of the PFP link will lead to beneficial incentive and sorting effects (Lambert, Larcker, & Weigelt, 1993); thus, the degree to which pay and performance are linked is a critical characteristic of any PFP plan (Milkovich et al., 2013; Zenger, 1992).

Expectancy theory proposes that employees make rational decisions based on the character- istics of the incentives they are facing (Bartol & Durham, 2000; Fusilier, Ganster, & Middlemist, 1984; Vroom, 1964), hence, positing that, all else being equal, motivation will be stronger if there is a stronger link between performance and rewards (Bartol & Durham, 2000; Bonner & Sprinkle, 2002; Kahn & Sherer, 1990; Lawler, 1971). Thus, finan- cial rewards that are strongly tied to individual performance increase employees’ effort, and this increased effort is supposed to lead to increases in performance (Bonner & Sprinkle, 2002; Lawler, 1971). Similarly, agency theory predicts that if per- formance can be monitored and tied to awards, then the rewards system can improve individual performance (Bartol & Locke, 2000; Eisenhardt, 1989). Agency theory also posits that a strong PFP plan can help solve the risk-sharing problem that organizations often experience in agency relation- ships by leading people who are highly risk-averse and less productive to leave their jobs (Cadsby et al., 2007; Eisenhardt, 1989). Tournament the- ory suggests that employees compete for higher rewards, and so a stronger link between perfor- mance and rewards should be associated with greater effort to achieve the higher awards (Becker & Huselid, 1992). At the same time, the competi- tion among individuals attracts high performers but increases voluntary turnover of poor perform- ers (Bloom & Michel, 2002; Shaw & Gupta, 2007). Even though some have argued that equity the- ory is counter to PFP, it has been recognized that equity does not mean equality (Brown, Sturman, & Simmering, 2003; Trevor, Reilly, & Gerhart, 2012). To maintain equity across employees, it is necessary to link pay and performance so that individuals’ ratios of performance to rewards are maintained across performance levels.

To understand the potentially different effects of PFP plans, we must, therefore, specifi- cally examine the strengths of the associations between performance and rewards. Research has provided examples of widely disparate relation- ships between pay and performance under nomi- nal PFP plans. For example, research has shown varying relationships between raises and perfor- mance under merit plans (e.g., Harris, Gilbreath, & Sunday, 1998; Kahn & Sherer, 1990; Markham, 1988). Similarly, some research has examined

ratings (Heneman & Werner, 2005). The pay plan is usually based on an individual’s performance, assessed by an employee performance appraisal (Rynes et al., 2005; Schwab & Olson, 1990). Merit pay shares elements of both variable pay and fixed pay. It is variable in that the pay raise depends on individual performance, and thus new raises must be re-earned each year. It is fixed, though, in that any given merit raise increases base pay, and thus regardless of future performance levels, that new base pay will continue to be received even if performance changes (barring employee termination).

Bonus pay is a monetary reward given in addi- tion to employees’ fixed compensation (Milkovich et al., 2013). Bonuses are ostensibly based on indi- vidual performance but do not increase employ- ees’ base pay (Sturman & Short, 2000). This type of pay plan has been widely used in organizations to motivate employees’ performance, and surveys report that the popularity of bonus pay is increas-

ing (cf. Sturman & Short, 2000; WorldatWork, 2012). Individual- based performance bonuses are attractive from the company’s per- spective because the one-time cash rewards link pay to performance but do not increase fixed labor costs (Sturman & Short, 2000).

LTIs are rewards linked to a firm’s long-term growth as well as employee retention (Rousseau & Ho, 2000), generally in the form of cash or stocks (Moynihan, 2013). LTIs allow a link between pay and performance, and like bonuses must be re-earned each year. Their

rewards, however, are not immediate. Employees must wait until such awards are vested before their value can be used. While companies have histori- cally offered LTIs mostly to executives, many firms have begun applying LTI plans to nonexecutive employees (Core & Guay, 2001; National Center for Employee Ownership, 2012; Oyer & Schaefer, 2005). LTI plans are also PFP plans because the award itself may be a function of individual per- formance, and the value of the incentives change based on the performance of the organization. This helps tie employee rewards to overall orga- nizational performance, although such PFP is no longer solely linked to individual performance.

Getting What You Pay For: The Link Between Pay and Performance

As reviewed earlier, a pay plan’s name reveals information about its award, but simply calling something a PFP plan does not necessarily mean

Human Resource Management DOI: 10.1002/hrm

EVALUATING FORM AND FUNCTIONALITY OF PAY-FOR-PERFORMANCE PLANS 701

When we

consider the plans

simultaneously, the

situation is more

complex. We cannot

simply assume that

the incentive effects

from all three plans

combine linearly,

because their effects

are not independent.

multiple pay plans, one must therefore consider the independent effects of each pay plan. That is, we must ask: what is the effect of a given PFP plan after controlling for the effects of the other PFP plans that are present? For example, to know the effect of merit pay in a multi-PFP environment, we must look at the relationship between merit pay and performance after controlling for the relationships between pay and performance from the other PFP plans. This represents the incen- tive effects uniquely attributable to the particu- lar PFP plan. Stated in more statistical terms, this means we are looking at the partialed effects of each pay form: the relationship between pay and performance for a given pay form after control- ling for the effects of the relationship between pay and performance for all other pay forms. When considering multiple PFP plans simultaneously, we only expect a given PFP plan to have an effect if the plan still has a relationship with performance after controlling for the relationships from the other pay forms. Hence, we predict:

Hypothesis 1: When considering the incentive effects of multiple pay plans simultaneously, the strength of the connection between individual perfor- mance and associated rewards, after separating the PFP effects (i.e., control- ling for effects) associated with other pay plans, will be positively related to future employee performance.

Relative Effects of PFP Plans,

Considered Simultaneously

Prior research has paid little atten- tion to the valence (i.e., the attractiveness of rewards) of the monetary awards across PFP plans (cf. Gerhart, Minkoff, & Olsen, 1995). Yet with the multiple pay forms that are the focus of this study—merit pay, bonuses, and LTIs—one cannot assume that valences are equal. Individuals should value rewards differently due to their particular characteristics. With unequal valences, the incen- tive effects of different pay plans should likewise be unequal.

The theories reviewed earlier have essentially the same key takeaway: that more is better. As has been most typically applied, that “more” has been considered within the context of a single pay form; yet, the same basic concept would seem to apply if there are multiple pay forms. Expectancy theory would predict that if one is covered by mul- tiple pay plans, and assuming the plans operated

bonus plans that have only a modest correlation between performance evaluations and bonuses (e.g., r = .15 in Mizruchi, Stearns, & Fleischer, 2011), where others have shown stronger rela- tionships (e.g., r = .42 in Salamin & Hom, 2005). Research on LTI has been more limited. One exception (Cappelli & Conyon, 2011) examined how stock incentives relate to employees’ future job performance. In the study, all employees at the same administrative level received the same amount of shares with the same vesting require- ments. The results showed that higher profits led individual employees to better performance. This research shows that LTIs can influence indi- vidual employee performance, but more research is noted to understand how strong this effect is, particularly in relation to other PFP options.

In this study, we consider complex envi- ronments where more than a single PFP form is provided. First, we look at incentive effects of multiple PFP plans that have been implemented simultaneously.

Incentive Effects of Pay-for-Performance

Incentive Effects of PFP Plans, Considered

Simultaneously

The fundamental premise behind PFP plans is that by tying pay to higher performance levels, such plans will motivate higher performance. All else equal, and based most directly on expectancy and tournament theory, stronger connections should be associated with greater performance gains. The context of a multiple PFP environment, though, presents an untested theoretical question. In field settings, it is often unclear if a specific PFP plan under consideration was the sole PFP plan or if other PFP plans might have been in place. The theory, though, is quite general in its proposi- tions, suggesting that if a given PFP plan creates a link between performance and pay, it should be associated with improved individual performance. While prior findings predicted that each form of PFP should be related to higher performance lev- els, this has been framed when considering a com- parison to a null effect (e.g., Banker et al., 1996, 2001; Lazear, 2000; Pearce et al., 1985) or relative to the other plans (Kahn & Sherer, 1990; Nyberg, Pieper, & Trevor, 2013).

When we consider the plans simultaneously, the situation is more complex. We cannot simply assume that the incentive effects from all three plans combine linearly, because their effects are not independent. Particularly, if we are consider- ing how pay is tied to performance, the way in which pay and performance are linked may essen- tially overlap across plans. When considering

702 HUMAN RESOURCE MANAGEMENT, JULY–AUGUST 2016

Human Resource Management DOI: 10.1002/hrm

idea that a permanent increase has greater valence than a one-time payment, or from a tournament theory perspective stemming from a raise consti- tuting a larger pay differential than a bonus, on a unit-per-unit basis (e.g., a $1 raise versus a $1 bonus), the incentive effect for merit pay should be greater than the incentive effect for bonuses.

LTIs represent one-time payments; however, while bonuses are immediate payments, LTIs are not. LTIs require a vesting period (cf. Dunford, Oler, & Boudreau, 2008) and so are not immedi- ately liquid. Stock awards are also more risky, as the value of the award can fluctuate with time and even become zero (Hull, 2012). As noted earlier, all else being equal, individuals typically prefer immediate rewards to delayed rewards (e.g., Green & Myerson, 2004). In addition, the liquidity of cash bonuses causes such rewards, on a dollar-per- dollar basis, to have a greater present value than a comparably sized stock award. Together, these characteristics indicate a lower valence for LTIs than both raises and bonuses. As our focus is on predicting individual-level outcomes, we predict:

Hypothesis 2: If the separated PFP relationship (i.e., the effects for each plan, after controlling for the PFP effects of the other plans) for each plan has effects (is greater than zero), the incentive effect for merit pay on individual job performance should be greater than the incentive effect for bonuses, which should be greater than the incentive effect for LTI.

However, if the PFP relationship for any pay form is zero after controlling for the effects of the other pay plans, then regardless of the pay form, the effect of that PFP relationship should be unre- lated to performance. Thus, we predict:

Hypothesis 3: For any plan type where the separated PFP relationship (i.e., the effect of the plan, after con- trolling for the PFP effects of the other plans) has no effects (is not signifi cantly different from zero), the incentive effect of the connection between pay and per- formance for that plan on individual job performance should be zero.

Sorting Effects of Pay-for-Performance

Sorting Effects of PFP Plans, Considered

Simultaneously

In addition to incentive effects, PFP plans can play an important role in attracting and retaining highly productive employees (Bartol & Durham, 2000; Gerhart & Fang, 2014). Research has shown that the relationship between performance and turnover is curvilinear, such that high perform- ers and low performers are most likely to leave

independently (i.e., each relationship between pay and performance effect was independent of the other PFP relationships) and the valences for those pay plans were equal (i.e., the rewards from each plan were valued equally), the motivational effects from multiple pay plans should be cumula- tive. For considering multiple pay plans, however, such simplifications are unlikely. This is due to two fundamental issues that must be considered to form hypotheses regarding multiple pay forms: the different levels of valence across plans, and the nonindependence of PFP relationships across plans.

While there certainly may be individual differ- ences with regard to pay preferences and valence (Mitchell & Mickel, 1999), all else being equal, a larger reward should be perceived more posi- tively than a smaller reward. Similarly, individu- als’ preferences for rewards are a function of delay (i.e., immediate versus delayed rewards) and risk. According to decision-making literature, future uncertain rewards are less valued than immediate assured rewards (Green & Myerson, 2004; Steel & König, 2006). Immediate rewards should be per- ceived more positively than a future reward of the same amount. Likewise, a guaranteed reward should be perceived more positively than a risky reward. Steel and König (2006) addressed that individuals are more likely to value immediate but smaller rewards than large but distant ones when they need to choose some behaviors that lead to rewards. Indeed, people tend to undervalue future events. Thus, the three pay plan types we are examining—merit pay, bonuses, and LTIs— clearly differ with regard to their value, immedi- acy, and risk. The objective characteristics of the rewards can, depending on individual differences, be interpreted as the attractiveness of the rewards (valence).

A key characteristic of merit pay is that it per- manently increases employees’ base pay. Although new merit raises have to be re-earned each year, once a raise is given, the individual will continue to receive that reward as long as the individual remains with the organization. This characteristic differentiates merit pay from the other forms of PFP that we discuss. Bonuses are one-time pay- ments, and thus do not change an individual’s level of base pay (Sturman & Short, 2000). As a one-time payment, the economic value of a bonus is always less than that of a raise for any person staying beyond one year. Due to the characteristics of merit pay, the permanent increase from merit pay has a greater lifetime value than the one-time rewards granted by other pay plans (Shaw, Duffy, Mitra, Lockhart, & Bowler, 2003). Whether from an expectancy theory perspective based on the

Human Resource Management DOI: 10.1002/hrm

EVALUATING FORM AND FUNCTIONALITY OF PAY-FOR-PERFORMANCE PLANS 703

Each PFP plan will

have an effect on

reducing employee

voluntary turnover

when considered

simultaneously

because, to the

extent that each

plan links pay and

performance, each

plan is reinforcing the

equity relationship

that high outputs

(performance) are

tied to high inputs

(rewards).

high performers (Allen & Griffeth, 2001; Salamin & Hom, 2005; Schwab, 1991; Trevor et al., 1997; Williams, 1999). The degree of this contingency should moderate the relationship between per- formance and desirability of movement. This prediction held in instances when raises (Trevor et al., 1997) and bonuses (Salamin & Hom, 2005) were related to performance, but not when raises were unrelated to performance (Salamin & Hom, 2005). Thus, we expect that PFP, no matter what the form, should help reduce the probability of high-performer turnover, but also that the nature of the reward (raise, bonus, or LTI) makes relative differences in PFP effectiveness. It is again more complex to consider multiple plans operating simultaneously.

The way PFP influences turn- over is based on the supposition that it moderates the relationship between performance and the desir- ability of turnover (Allen & Griffeth, 2001). This is based more on theo- ries of equity and fairness than expectancy. As such, partialing out the relationship between pay and performance should not necessar- ily have the same effect on turnover as it does on performance. We still expect that each PFP plan will have an effect on reducing employee voluntary turnover when consid- ered simultaneously because, to the extent that each plan links pay and performance, each plan is rein- forcing the equity relationship that high outputs (performance) are tied to high inputs (rewards).

The enduring nature of merit pay indicates a potentially strong sorting effect. Because a raise has a permanent effect on base pay, once it is earned in a given year, it will be repeatedly earned, even if perfor- mance declines. Furthermore, current raises can lead to more future value because raises are com- pounded. That is, as raises are typically expressed as a percent of salary (Milkovich et al., 2013), a raise creates more value for future raises. It may also make future bonuses and LTI larger if they are based on the size of the individual’s salary. Thus, in comparison to a one-time bonus or LTI, a raise should have the greatest potential for retain- ing employees in contrast to a comparably sized bonus or LTI. We, therefore, predict:

Hypothesis 4: When considering merit pay, bonuses, and LTI plans simultaneously, the PFP for merit pay

an organization (Salamin & Hom, 2005; Sturman, Shao, & Katz, 2012; Trevor et al., 1997). A higher PFP relationship, however, should decrease high- performer turnover because the high reward con- tingency leads to lower desirability of movement (e.g., Allen & Griffeth, 2001; Jackofsky, Ferris, & Breckenridge, 1986; Trevor et al., 1997).

Empirical research specifically into how differ- ent forms of PFP influence sorting effects, though, is limited (Gerhart et al., 2009). Trevor et al. (1997) looked at mean salary growth over a four- year period. They found that higher salary growth reduced the overall likelihood of turnover and the turnover of high performers. Similarly, Salamin and Hom (2005) looked at individuals’ mean pay increase and latest bonus as moderators of the performance-turnover relationship. They found that bonuses reduced the probability of turn- over. In contrast to Trevor et al. (1997), though, Salamin and Hom (2005) found that raises had no significant effect on how performance related to turnover.

Both Trevor et al. (1997) and Salamin and Hom (2005) considered how pay influences the effect of performance on turnover; they did not, however, specifically consider how strongly pay and perfor- mance were linked. In Trevor et al. (1997), the cor- relation between performance and average salary growth was 0.30. In Salamin and Hom (2005), the correlation between the mean pay increase and performance was only 0.05, whereas the correla- tion between performance and the latest bonus was 0.42. Thus, the discrepant results associated with pay increases between these two studies may be due to the different strengths of the connection between pay and performance (Zenger, 1992). In both studies, when there was a higher correlation between pay and performance (i.e., .30 or .42), the pay system did improve retention of high performers.

Research on LTI is more limited. Some research on executive compensation has shown that stock awards and other LTI are associated with reduced turnover (e.g., Batt & Colvin, 2011; Mehran & Yermack, 1996). Other research has examined how repricing underwater stock options influ- ences turnover (e.g., Carter & Lynch, 2004; Daily, Certo, & Dalton, 2002; Dunford et al., 2005). No research has yet addressed if LTI affects the perfor- mance-turnover relationship.

Research also has yet to specifically exam- ine the degree to which a PFP link for multiple types of pay forms influences individual turn- over. Relevant theory, however, can provide some insights into the nature of the sorting effects that we might expect. Tying pay to performance should reduce the desirability of turnover for

704 HUMAN RESOURCE MANAGEMENT, JULY–AUGUST 2016

Human Resource Management DOI: 10.1002/hrm

bonuses actually had a positive effect on intent- to-turnover. Thus, because bonuses seemingly can create both incentives and disincentives for turnover when considered simultaneously with other PFP plans, we have no a priori hypotheses regarding their effects when considered in con- junction with merit raises and LTI.

Method

Sample

The data for this study were obtained from a large service-related company that was a subsidiary of a larger, diversified publicly traded American corpo- ration. The company that is the focus of our study offers broad-based business services to companies in the global travel industry, providing technol- ogy and support to global travel companies and managing technology related to online travel. The business, thus, focuses on issues of technol- ogy and pricing, but has positions in account sales and service, accounting, customer training and support, finance, human resources, legal, mar- keting and communications, and product and technology development. The company provided data on performance ratings, gender, organization tenure, salary, and percentage of three financial rewards (merit pay, bonuses, and LTIs) associated with 2001 and 2002. We only examined employ- ees who had performance ratings in 2001, did not leave involuntarily in 2002, and were eligible for all three forms of compensation. This resulted in an original potential sample of 900 employees from 17 locations in the United States. All posi- tions were white collar, exempt, full time, and required a college degree. The most common job titles were developer, senior developer, systems engineer, travel supervisor, account manager, and project manager.

It should also be noted that the references to the calendar year are not entirely precise in that they refer to the relevant time period and not the date of the award or decision. Performance reviews in a given year are intended to reflect the individual’s performance for that calendar year (so the 2001 performance rating is designed to cap- ture performance during the 2001 year but was determined in early 2002). Compensation awards associated with a given year reflect the award of each type given for that year. Thus, the 2001 bonus is actually awarded in 2002, but is awarded to the individual for service over the 2001 year. Similarly, the 2001 raise is awarded in early 2002, as is the 2001 LTI. We refer to 2001 rewards to represent the awards associated with 2001 perfor- mance. Note that 2001 salary, though, was the sal- ary at the beginning of 2001.

will negatively moderate the performance-turnover relationship.

While we again predict the strongest effect for merit pay, the sorting effects of bonuses versus LTI should differ from their incentive effects. A key purpose of deferring compensation is to foster employee retention (e.g., Core & Guay, 2001; Jones & Kato, 1995; Oyer & Schaefer, 2005). Providing LTI increases the cost of turnover for employees because leaving the organization requires them to forfeit any rewards that are not vested. Thus, we predict:

Hypothesis 5: When considering merit pay, bonuses, and LTI plans simultaneously, the PFP for LTI will negatively moderate the performance-turnover relationship.

Relative Effects of PFP Plans, Considered

Simultaneously

Both merit pay and LTI, by increasing the cost of turnover, should reduce the likelihood of high- performance turnover. Yet because a merit pay award has greater value than an equally sized LTI award, we again expect merit pay to have a stron- ger effect. We, therefore, predict:

Hypothesis 6: When considering merit pay, bonuses, and LTI plans simultaneously, the effect of PFP on the performance-turnover relationship will be stronger (i.e., more negative) for merit pay than for LTI.

The nature of bonuses, though, makes their effects less clear. This is because bonuses have characteristics that both decrease the desirabil- ity of movement but increase the ease of move- ment. By having a connection between pay and performance, bonuses will increase equity and fairness perceptions of high performers, thus decreasing the desirability of movement. At the same time, the large monetary influx from a large bonus may actually increase the individ- ual’s ease of movement. This can occur because either (1) the monetary reserves make leav- ing one job and finding another more feasible, or (2) if the individual had an intent to leave, was a high performer, and was expecting a large bonus, it would be most logical to wait for the bonus before leaving the position. Consistent with this, Sturman and Short (2000) showed that bonus satisfaction was negatively related to intent to turnover when not considering other aspects of pay satisfaction or other attitudes; however, after controlling for the other pay satisfaction dimensions (including for raises),

Human Resource Management DOI: 10.1002/hrm

EVALUATING FORM AND FUNCTIONALITY OF PAY-FOR-PERFORMANCE PLANS 705

“turn into” actual shares of the company’s real stock. In other words, when granted, the restricted stock had no immediately realizable monetary value (although it was expressed as such, based on the number of shares and its current price). The company used a five-year gradual vesting schedule, with 20 percent of the stocks becoming vested each year (and the value of that award con- stituting income). Once vested, the award had the same market value as any other share of common stock from the organization. Over the span of this study, the value of company stock did vary, but increased an average of 1.08 percent per month. Employees of the company were educated about the financial rewards system via intranet, written communication, and training workshops.

Measures

Employee Job Performance

Employee job performance ratings from 2001 and 2002 were used as independent and dependent variables. The ratings used a 4-point scale: signifi- cantly exceeds expectations, exceeds expectations, meets expectations, and is below expectations. The ratings were transformed to indicator vari- ables from 1 (lowest performance) to 4 (highest performance). The company used “management by objectives” to create performance ratings, and significant time was allocated by the organization for these purposes.

For predicting turnover, we needed to exam- ine potential nonlinear effects of performance (Salamin & Hom, 2005; Sturman et al., 2012; Trevor et al., 1997). For these analyses, we con- sidered both a linear effect, computed as a mean- centered measure of 2001 performance, and a quadratic effect. Because we wanted to specifi- cally differentiate between linear and nonlinear effects, we needed to ensure that the two terms were orthogonal. Although mean-centering (cf. Aiken & West, 1991) is common and can reduce the correlation between linear and squared terms, it does not yield orthogonal terms. Indeed, in this case, mean centering would still result in the linear and quadratic terms being correlated 0.21. We, therefore, used residual centering (Lance, 1988). That is, we regressed the squared term on the linear term and used the residual to represent the quadratic effect. This residual is uncorrelated with the linear term but captures the nonlinearity associated with the squared term (Lance, 1988). Residual centering has been shown to be effective in cases like this one, in which (1) the main and quadratic effects are correlated, (2) sample sizes are moderate to large, and (3) decomposition of the effects is desired (Lance, 1988).

To provide an estimate of the PFP relation- ships for each employee (as will be explained in greater detail below), we calculated the relation- ship that existed between 2001 performance and each resultant form of compensation for 2001 performance under each supervisor. Thus, we eliminated all individuals in the sample who did not have the same supervisor in both 2001 and 2002 and (for the purposes of being able to com- pute PFP relationships for each supervisor, as we will explain below) all employees under supervi- sors who did not have at least three subordinates. This resulted in a sample of 720 employees under 88 supervisors. Supervisors had an average of 8.1 subordinates (SD = 7.58; range = 3 – 59). For pre- dicting 2002 performance, we eliminated subjects who left the organization (and thus did not have 2002 performance ratings), resulting in a sample of 635 employees (also under 88 supervisors).

Organizational Compensation System

The organization provided guidance to supervi- sors regarding the allocation of compensation, but supervisors had discretion in the allocation decisions. This is a typical approach used by many organizations for their PFP plans (cf. WorldatWork, 2012). For merit pay, a range of percent pay increases were specified for each performance rat- ing category. Supervisors had discretion regarding what the specific raise should be within this range.

The company paid bonuses, which reflected the judgment of the supervisor based on the indi- vidual’s performance rating and the individual’s position. Each position had a bonus target, based on its degree of responsibility. Organizational per- formance affected the amount of total rewards that would be distributed, a decision made by execu- tives within the organization (not in our sample). Supervisors were given a budget with which to allocate rewards to their subordinates. They were instructed to use the bonus target and individual performance to guide their decisions, but had dis- cretion with regard to how actual distributions were made. Managers did not have complete discretion, though, in that all pay decisions were ultimately approved by the human resources department.

The distribution of LTI was based on both orga- nizational performance (which determined the budget for this award) and individual job perfor- mance, although the dollar value of actual awards was also affected by the performance of the com- pany’s stock. Each year, the company distributed restricted stock units to their employees based on individual performance and criticality of employ- ees’ job positions. At the time of the award, the restricted stock had no real market value. Rather, these stock grants would, at a future vesting date,

706 HUMAN RESOURCE MANAGEMENT, JULY–AUGUST 2016

Human Resource Management DOI: 10.1002/hrm

This approach to measuring PFP, while new, has a number of advantages. Most notably, it directly assesses the simple and straightforward relationship that exists between pay and perfor- mance for each pay form, which matches the measure to the theoretical premise better than other PFP measures have done. Note also that this approach does not require us to assume that indi- viduals know what others get paid. If a supervi- sor has a strong PFP relationship, a low performer will see a low reward and a high performer will see a higher reward (which is similar to what a ratio score would yield). If the supervisor has no relationship between pay and performance, this measure will yield a zero as opposed to a ratio score what would yield a high value for a low performer and low value for a high performer. Although individuals may or may not know what others receive, people are likely aware of what typical awards are, either in the company overall or in the economy. A measure of slope (which is what our measure is), thus, does not suffer from as much random variance that can occur with ratio scores, even without necessarily assuming that individuals know the performance and rewards of their peers. It should also be noted that this metric has very practical benefits. As an objective measure of PFP, it captures the actual relationship that exists between pay and performance, which is something that can be influenced by organiza- tional policy. It is also a measure that companies can calculate using archival data, and thus allows organizations to critically evaluate their current PFP plans.

Voluntary Turnover

Voluntary turnover included departures from the company during 2002. One concern in turn- over research is that some previous research has failed to distinguish between voluntary turn- over and involuntary turnover (Gardner, Wright, & Moynihan, 2011; Gerhart, 1990; Wright & Cropanzano, 1998). Fortunately, the company provided specific information that distinguished between voluntary and involuntary turnover. Thus, we considered only voluntary turnover. In the sample, 85 (12 percent) of the 720 employ- ees voluntarily left the company. We used a “0” to indicate that an employee stayed with the com- pany and a “1” to indicate voluntary turnover (although it was treated in most analyses as a cat- egorical variable).

Control Variables

Because this study examined the effect of finan- cial rewards on employees’ future performance, previous performance (i.e., 2001 job performance

Measuring the Pay-for-Performance

Relationships

To test our hypotheses, we needed a measure of the link between pay and performance for the three compensation plans. Prior compensation research has examined this by looking at the reward received (e.g., the change in pay or total compensation) divided by the performance score (e.g., a given dollar change in shareholder value or firm return) (e.g., Hall & Knox, 2004; Hayes, 2004) or by calculating the derivative of pay with respect to performance (e.g., Kahn & Sherer, 1990). The former approach is not ideal for our study, as it is simply an individual’s ratio and does not directly capture a relationship between pay and performance. It is unclear if effects from such ratios are due to the numerator, denominator, or the hypothesized combination of the two. It also does not capture if indeed there is a pattern of pay and performance effects across individuals (which is what PFP plans are purported to have). The approach of calculating a derivative (e.g., Kahn & Sherer, 1990) more directly assesses if there is a relationship between performance and pay, but the approach is based on looking at effects asso- ciated with interactions with performance scores (because if there is a straightforward relationship between pay and performance, this becomes a constant that is equal across subjects). We wanted to more directly assess exactly how pay and per- formance are related to each other within groups that should share a similar effect. To estimate this relationship for each employee, we examined the strength of the link between pay and performance under each supervisor.

For each supervisor (as noted above, with at least three subordinates), we computed the three regression coefficients that estimated the relation- ships between 2001 performance and the three forms of 2001 rewards. For each supervisor and pay form, the percent reward was regressed on the employee performance rating. Rewards were pre- sented as percentage points (so 1 = 1 percent), and the raw rating metric (1–4) was used as the inde- pendent variable. For each pay form, we thus ran 88 regressions. This yielded 88 B coefficients associ- ated with each of the three PFP relationships, which we labeled PFP-Merit, PFP-Bonus, and PFP-LTI.

It should be noted that, for subsequent analy- ses, our PFP metrics were not independent. That is, individual observations of PFP were not fully independent because employees under the same supervisor operated under the same PFP relation- ship. We thus needed to use multilevel modeling (Raudenbush & Bryk, 2002) when considering the PFP effects of the various plans.

Human Resource Management DOI: 10.1002/hrm

EVALUATING FORM AND FUNCTIONALITY OF PAY-FOR-PERFORMANCE PLANS 707

relative effects of different PFP plans, our depen- dent variable was 2002 job performance. The level 1 model we used was:

Perf 2002

= B 00

+ B 1 *Perf

2001 + B

2 *Gender

+ B 3 *Tenure + B

4 *ln(Salary

2001 )

+ B 5 *Merit percent + B

6 *Bonus percent

+ B 7 *LTI percent + ε [1]

In this model, the intercept (B 00

) was modeled as a random effect.

The second level of analyses varied depend- ing on what type of PFP plans we tested. As noted earlier, we first examine each PFP plan separately. The purpose of this test is to examine if prior PFP research conducted on a single plan generalizes to a situation where one is examining a single plan within a multi-PFP context. While these models are not used to test our hypotheses, it is impor- tant to see if prior research (in which we do not know whether a plan was operating in a multi-PFP plan environment) is consistent with a purposely underspecified model (i.e., when we examine a single PFP plan at a time, but we do know that the plan is operating in a multi-PFP plan environ- ment). For these models, the PFP effect for each pay form was entered as the sole level 2 variable. So, to test the effect of merit pay (Model 2a), the level 2 equation was

B 00

= G 00

+ G 01

*PFP-Merit + ξ [2a]

whereas to test the effect of bonus pay (Model 2b), the level-2 equation was

B 00

= G 00

+ G 01

* PFP-Bonus + ξ [2b]

and to test the effect of LTI (Model 2c), the level 2 equation was

B 00

= G 00

+ G 01

* PFP-LTI + ξ [2c]

This study’s hypotheses consider the role of multiple PFP plans operating simultaneously. To test our hypotheses related to incentive effects of all three plans simultaneously (Model 3), the level 2 equation was as follows:

B 00

= G 00

+ G 01

*PFP-Merit + G 02

*PFP-Bonus + G

03 *PFP-LTI + ξ [3]

Sorting Effects

For assessing sorting effects, because turnover is a dichotomous variable, we used the Bernoulli model in HLM. Thus, our analyses are multilevel but analogous to logistic regression (Raudenbush

rating) was used as a control variable. Using prior performance as a control variable helps par- tial out the effects of stable characteristics that caused employees’ performance (Sturman, 2003; Sturman, Cheramie, & Cashen, 2005) and unmea- sured effects that are attributable to omitted fac- tors (e.g., ability, job knowledge, motivation levels, or opportunities to perform) that might affect per- formance and pay (Kahn & Sherer, 1990; Sturman, 2007). While performance ratings certainly are not perfect measures (Viswesvaran, Ones, & Schmidt, 1996), its inclusion does help address the alterna- tive explanation that high performers get rewarded but also remain high performers. We want to know what effect PFP has beyond knowing that, in gen- eral, past performance is the best predictor of future performance (Sturman et al., 2005).

Organization tenure was used as a control variable because it could interfere with test- ing the main effects of the different characteris- tics of financial rewards on future performance (Sturman, 2003). Gender differences have been considered a potentially important factor caus- ing pay differences (Milkovich et al., 2013), and so were also controlled for (with men coded as 0, and women coded as 1). Additionally, salary from 2001 was controlled in this study, as was the level of the most recent raise, bonus, and LTI. Because the salary data was skewed, we used a log transfor- mation to reduce the leverage of high values. The raise, bonus, and LTI were expressed as a percent of salary (before log transformation).

Analyses

Before we tested our hypotheses, we wanted to replicate prior findings and thus see if our sam- ple provides similar results to prior research. Our analyses are conducted in a series of steps. First, we create a baseline model without PFP variables, to serve as a point of reference (Model 1). We then run a series of models in which we consider a single pay plan at a time. This allows us to repli- cate prior research that has examined a single pay plan at a time but did not consider if other PFP plans were in effect. Thus, we run a model exam- ining merit pay (Model 2a), bonuses (Model 2b), and LTIs (Model 2c). Then, we test our hypotheses with the model that includes all three plans simul- taneously (Model 3).

Incentive Effects

In all of our models, because individuals were nested within supervisors, we used hierarchical linear modeling (Raudenbush & Bryk, 2002) using the HLM7 statistical package (Raudenbush, Bryk, Cheong, Congdon, & du Toit, 2011). To test the incentive effects of independent PFP plans and

708 HUMAN RESOURCE MANAGEMENT, JULY–AUGUST 2016

Human Resource Management DOI: 10.1002/hrm

by Gerhart and Fang (2014). The mean PFP rela- tionship for merit pay was 0.62 (SD = 0.14) and ranged from 0 to 0.95; for bonuses, the mean was 0.73 (SD = 1.9) and ranged from –1.1 to 10.5; and for LTI, the mean was 0.57 (SD = 1.48) and ranged from 0.02 to 9.2. This means, for example, a one- point increase in performance was associated with, on average, a 0.62 percentage point higher raise. The average raise for a high performer (4) is thus 1.86 percentage points higher than that for low performer (1). For some supervisors, larger bonuses were actually granted to lower perform- ers, as evidenced by a minimum PFP-Bonus value of –1.1. However, as noted, the average relation- ship was 0.73.

There are also some strong correlations among the PFP variables and the pay amount percent vari- ables. For example, the relationship between PFP- Bonus and PFP-LTI was strong (r = .73), showing that managers are fairly consistent in how they distribute bonuses and LTI. Of course, although a high correlation, it still indicates that nearly half the variance between these variables is unex- plained, and thus there clearly are differences in how managers allocate awards.

Variance components for all models are pre- sented in Tables III and IV. For both the prediction of 2002 job performance and voluntary turnover, the analyses revealed significant level 2 variance (p < .05) for the intercepts in the random-inter- cepts models.

Replicating Previous Findings

Independent Incentive Effects

Models 2a, 2b, and 2c in Table III present the results of the HLM analyses, which examine a sin- gle pay plan at a time. These results are consistent with existing theory and prior empirical work. Note that these are not nested models, not fully comparable, and thus not used for our hypothesis tests. The purpose of these models is to illustrate what the results would look like if a researcher were examining a single PFP plan at a time in a multi-PFP environment. The results show that the strength of a PFP plan’s connection between individual performance and rewards is positively related to future employee performance when considering a single PFP plan at a time. As shown in Models 2a, 2b, and 2c (for merit pay, bonuses, and LTI, respectively), all three PFP relationships were significantly related to future performance (all at p < .001).

Independent Sorting Effects

Table IV presents the analyses predicting 2002 vol- untary turnover, and specifically how the strength

& Bryk, 2002). For our test of independent sorting effects, our level 1 model was as follows:

Prob(turnover) = B 00

+ B 10

*Perf 2001

+ B 20

*Perf2 2001

+ B

3 *Gender + B

4 *Tenure

+ B 5 *ln(Salary

2001 ) + B

6 *Merit

percent + B 7 *Bonus percent

+ B 8 *LTI percent + ε [4]

For this model, the intercept and the effects of performance could potentially vary across super- visors, as these coefficients could be affected by the degree of PFP resulting from each supervisor’s decisions. Testing revealed, though, that there was only significant level 2 variance for the intercept term; for B

10 and B

20 , the variance component was

not significant (at p > .50). Thus, only the inter- cept was modeled as a random effect. Nonetheless, for all three of these coefficients, the level 2 equa- tions were equivalent to Equations 2a, 2b, and 2c above, with the key differences being that (1) there are three equations, with the dependent variables being B

00 , B

10 , and B

20 , and (2) there was no level

2 error term in the equations predicting B 10

and B

20 . Thus, to test the hypotheses related to sorting

effects, the level 2 equations were as follows:

B 00

= G 00

+ G 01

*PFP-Merit + G 02

*PFP-Bonus + G

03 *PFP-LTI + ξ [5a]

B 10

= G 00

+ G 01

*PFP-Merit + G 02

*PFP-Bonus + G

03 *PFP-LTI [5b]

B 20

= G 00

+ G 01

*PFP-Merit + G 02

*PFP-Bonus + G

03 *PFP-LTI [5c]

Results

Summary statistics are presented in Tables I and II. Table I presents the summary statistics for the portion of the sample (N = 635) used in the analy- ses predicting job performance; Table II presents the summary statistics for the sample used in the prediction of turnover (N = 720). Note that the means of merit percent, bonus percent, and LTI percent (shown in Table II) were 3.08 percent, 7.02 percent, and 1.19 percent, respectively. Thus, the average additional compensation received by employees was 11.29 percent, although only 3.08 percent was an increase in base pay.

For merit pay, across the four performance rat- ings, the average pay increases were 1.9 percent (below expectations), 2.8 percent (average perfor- mance), 3.3 percent (exceeding expectations), and 4.6 percent (far exceeding expectations). These levels are similar to the average merit increases in these or similar performance categories reported

Human Resource Management DOI: 10.1002/hrm

EVALUATING FORM AND FUNCTIONALITY OF PAY-FOR-PERFORMANCE PLANS 709

T A

B L

E

I

S u

m m

a ry

S ta

ti s

ti c

s f

o r

A n

a ly

s e

s o

f J

o b

P e

rf o

rm a

n c

e

M e a n

S D

L e

v e

l 1

V a

ri a

b le

s L e v e l 2 V

a ri

a b

le s

1 2

3 4

5 6

7 8

9 10

11

L e v e l 1

V a ri

a b

le s

1 .

2 0

0 2 P

e rf

o rm

a n

c e

2 .4

6 .6

9 —

2 .

2 0

0 1 P

e rf

o rm

a n

c e

2 .5

2 .6

1 .5

6 —

3 .

G e n

d e r

.5 0

.5 0

– .0

6 – .0

6 —

4 . T

e n

u re

1 4 .2

6 7.

2 6

.0 1

.1 0

.1 2

—

5 .

L n

( 2 0

0 1 S

a la

ry )

1 1 .0

4 .3

8 .2

9 .2

2 – .2

9 .1

0 —

6 .

M e ri

t P

a y (

% )

3 .0

9 .7

2 .7

7 .6

1 – .0

3 .0

1 .1

5 —

7. B

o n

u s P

a y (

% )

6 .8

5 5

.5 8

.2 2

.1 7

– .0

7 .0

8 .4

5 .1

9 —

8 .

L T

I P

la n

( %

) 1 .0

4 5

.1 7

.2 3

.1 2

.0 0

.0 0

.3 8

.1 3

.6 1

—

L e v e l 2

V a ri

a b

le s

9 .

P F P -M

e ri

t .6

2 .1

4 .4

5 .2

0 .0

5 – .0

3 .0

7 .4

5 .2

1 .1

9 —

1 0 .

P F P -B

o n

u s

.7 3

1 .8

9 .2

8 .1

7 .0

2 .0

0 .4

3 .1

8 .7

7 .7

6 .2

6 —

1 1 .

P F P -L

T IP

.5 7

1 .4

8 .2

3 .1

4 .0

4 – .0

1 .3

5 .1

5 .7

1 .8

1 .2

2 .7

3 —

N o

te s : N

= 6

3 5 . C

o rr

e la

ti o

n s g

re a te

r th

a n

. 0 8 a

re p

< . 0 5 .

710 HUMAN RESOURCE MANAGEMENT, JULY–AUGUST 2016

Human Resource Management DOI: 10.1002/hrm

T A

B L

E

I I

S

u m

m a

ry S

ta ti

s ti

c s

f o

r A

n a

ly s

e s

o f

Tu rn

o v

e r

M e a n

S D

L e

v e

l 1

V a

ri a

b le

s L e v e l 2 V

a ri

a b

le s

1 2

3 4

5 6

7 8

9 10

11

L e v e l 1

V a ri

a b

le s

1 .

2 0

0 1 P

e rf

o rm

a n

c e

2 .4

9 .6

3 —

2 . T

u rn

o v e r

.1 2

.3 3

– .1

4 —

3 .

G e n

d e r

.5 0

.5 0

– .0

6 .0

1 —

4 . T

e n

u re

1 3 .8

0 1 1 .0

4 .1

2 – .1

7 .0

9 —

5 .

L n

( 2 0

0 1 S

a la

ry )

1 1 .0

4 .3

9 .2

1 – .0

2 – .3

2 .1

2 —

6 .

M e ri

t P

a y (

% )

3 .0

8 .7

4 .5

9 – .0

4 – .0

2 .0

0 .1

2 —

7. B

o n

u s P

a y (

% )

7. 0 2

6 .4

6 .1

3 – .0

7 – .1

0 .0

6 .4

7 .1

4 —

8 .

L T

I P

la n

( %

) 1 .1

9 6

.4 1

.0 8

– .0

6 – .0

4 – .0

1 .4

0 .1

0 .6

4 —

L e v e l 2

V a ri

a b

le s

9 .

P F P -M

e ri

t .6

2 .1

4 .1

3 – .0

3 – .0

7 – .0

3 .0

4 .4

3 .1

8 .1

7 —

1 0 .

P F P -B

o n

u s

.7 3

1 .8

9 .1

5 .0

1 .0

1 .0

0 .4

1 .1

7 .7

5 .7

3 .2

4 —

1 1 .

P F P -L

T IP

.5 7

1 .4

8 .1

2 – .0

2 .0

3 – .0

1 .1

3 .1

4 .6

8 .7

8 .2

1 .7

2 —

N o

te s : N

= 7

2 0 . C

o rr

e la

ti o

n s g

re a te

r th

a n

. 0 8 a

re p

< . 0 5 .

Human Resource Management DOI: 10.1002/hrm

EVALUATING FORM AND FUNCTIONALITY OF PAY-FOR-PERFORMANCE PLANS 711

Trevor et al., 1997; specifically, the probability of turnover was, on average 52 percent, 14 percent, 7 percent, and 16 percent for performance scores 1 to 4, respectively). Adding the PFP variables to the second level of analysis was likewise consistent with prior research.

The results also show that the strength of each plan’s PFP link negatively moderated the performance-turnover relationship. As with the replication of the incentive effects, this pertained to considering individual PFP plans separately. As

of the various PFP links influence the perfor- mance-turnover relationship. Again, these results show that our sample produces results consistent with prior research.

A model with no level 2 variables (Model 1 in Table IV) showed that performance had the expected negative linear effect (G

10 = –.54; p < .05)

and positive nonlinear effect (G 20

= .71; p < .001), thus replicating the predicted inverted U-shape relationship between performance and turnover (Salamin & Hom, 2005; Sturman et al., 2012;

T A B L E I I I Separate and Partialed Effects of PFP Plans

Variable Model 1 Model 2a Model 2b Model 2c Model 3

For Intercept (Random Effect)

Intercept –3.06 –2.38 –2.48 –2.84 –2.93

(.68)*** (.62)*** (.71)*** (.65)*** (.62)***

PFP-Merit 10.98 88.22

(28.16)*** (28.94)**

PFP-Bonus 8.09 8.09

(1.86)*** (4.81)*

PFP-LTI 8.63 –2.06

(2.31)*** (5.84)

2001 Performance .14 .15 .14 .14 .15

(.048)** (.048)** (.047)** (.048)*** (.048)**

Gender .0087 .0034 –.0093 –.0051 –.012

(.034) (.031) (.034) (.034) (.032)

Tenure –.0034 –.0027 –.0021 –.0026 –.0017

(.0027) (.0026) (.0026) (.0027) (.0025)

Ln (2001 Salary) .34 .33 .29 .32 .28

(.067)*** (.064)*** (.060)*** (.064)*** (.057)***

Merit pay (%) 61.61 57.35 57.98

(5.71)*** (6.62)*** (6.60)***

Bonus Pay (%) –.54 –1.07 –1.27

(.95) (.65) (.94)

LTI (%) .082 –1.22 –.47

(1.01) (.96) (.93)

Fit Statistics

Model Likelihood –341.50 –325.60 –331.17 –333.53 –315.73

Sigma2 .1490 .1473 .1478 .1477 .1464

% LV 1 Var 68% 69% 69% 69% 69%

Explained

Random Effects .0319*** .0229*** .0255*** .0291*** .0196***

Variance Component

Percent Variance 79% 95% 95% 94% 96%

Component Explained

Notes: N (level 1) = 635; N (level 2) = 88. *p < .05; **p < .01; ***p < .001. For the null model, Sigma2 = .4693. For the random intercepts

model, likelihood = –624.98, Sigma2 = .3510 (25% explained), the random effects variance component = .1458 (signifi cant at p < .001), and

ICC(1) = .31.

712 HUMAN RESOURCE MANAGEMENT, JULY–AUGUST 2016

Human Resource Management DOI: 10.1002/hrm

T A B L E I V Prediction of Turnover

Variable Model 1 Model 2a Model 2b Model 2c Model 3

For Intercept (Random Effect) Intercept –1.45 –3.42 –1.90 –5.70 –2.00

(4.82) (4.31) (4.70) (4.63) (5.51)

PFP-Merit –262.34 –227.10

(143.36)* (115.77)*

PFP-Bonus –2.77 92.21

(13.92) (26.40)***

PFP-LTI –76.05 –242.17

(50.27)† (78.77)**

2001 Performance (linear)

Intercept –.54 –.73 –.53 –1.14 –2.05

(.27)* (.27)** (.19)** (.48)** (.59)***

PFP-Merit –823.81 –793.07

(223.88)*** (185.76)***

PFP-Bonus –30.33 21.29

(16.30)* (27.15)

PFP-LTI –162.77 –298.61

(88.55)* (123.49)**

2001 Performance (Quadratic)

Intercept .71 .64 .60 .20 –.083

(.25)** (.22)** (.21)** (.33) (.41)

PFP-Merit –497.56 –332.38

(241.34)* (253.46)†

PFP-Bonus –31.67 38.69

(18.93)* (43.14)

PFP-LTI –120.65 –222.78

(53.87)* (86.00)**

Gender .25 .27 .32 .39 .34

(.25) (.27) (.27) (.27) (.31)

Tenure –.088 –.086 –.094 –.098 –.095

(.020)*** (.021)*** (.022)*** (.022)*** (.023)***

Ln (2001 Salary) .023 .20 .070 .39 .020

(.43) (.39) (.43) (.42) (.49)

Merit pay (%) 11.38 31.21 23.55

(28.49) (23.10) (23.16)

Bonus Pay (%) 7.51 2.70 –1.50

(5.56) (3.69) (6.39)

LTI (%) –10.23 4.44 8.67

(8.92) (6.31) (10.72)

Fit Statistics

Model Likelihood

Sigma2 .0944 .0912 .0939 .0938 .0916

% LV–1 Var 10% 13% 11% 11% 13%

Explained

Random Effects .3406 .3606 .3872 .4063 .1877

Variance Component

Percent Variance 17% 12% 5% 1% 54%

Component Explained

Notes: N (level 1) = 720; N (level 2) = 88. †p < .10; *p < .05; **p < .01; ***p < .001. For the purpose of reporting level 1 variance explained in

the turnover model, as no level 1 sigma is given for a dichotomous outcome, the table reports the sigma2 for a normal (continuous) outcome

variable. All other turnover analyses are based on the more appropriate Bernoulli outcome model.

Human Resource Management DOI: 10.1002/hrm

EVALUATING FORM AND FUNCTIONALITY OF PAY-FOR-PERFORMANCE PLANS 713

shown in Models 2a, 2b, and 2c of Table III, if we were studying any single pay plan at a time that was operating in a multi-PFP environment, the effects of the PFP relationships all had negative effects on the linear and/or quadratic performance variables (all at p < .05), thus indicating that a stronger PFP relationship was associated with a lower probabil- ity of turnover as performance increased. So, as with incentive effects, were this a study about a single PFP plan, we would have yielded conclu- sions similar to prior research about PFP plans.

Tests of Relative Incentive Effects (Hypotheses 1–3)

Tests of our hypotheses regarding relative incen- tive effects are shown in Table III, supporting our hypotheses. The three hypotheses pertained to considering all three plans simultaneously. Testing these hypotheses involved considering the separated effects of each plan’s PFP relationship. This is important because the PFP relationships for the three pay practices are correlated, with a par- ticularly high correlation between PFP for bonuses and PFP for LTI.

We first computed the partial correlation for each of the three PFP variables on 2001 perfor- mance (i.e., the correlation between each PFP vari- able and 2001 performance with the effects of the other two PFP variables separated out). We found that, while the raw correlation of each PFP rela- tionship was significantly related to performance, the partial correlation coefficient for merit pay was largest (r

My.B,L = .16), followed by a smaller but

still significant effect for bonuses (r By.M,L

= .09), and no significant relationship for LTI (r

Ly.M,B = –.04).

Hypothesis 1 predicted that when consider- ing the incentive effects of multiple pay plans simultaneously, the strength of the connection between individual performance and associated rewards, after separating out the PFP relationship associated with other PFP plans, would be posi- tively related to future employee performance. Therefore, we expected a positive effect in our full model from merit pay and bonuses, but no effect for LTI. Model 3 in Table III indeed supports this. The effect for PFP-Merit was significant (B = 88.22, p < .01), as is the effect for PFP-Bonus (B = 8.09, p < .05); the effect of LTI was nonsignificant (B = –2.06, p = .73).

Hypothesis 2 predicted that for each plan with separate significant PFP relationships, the effect for merit pay would be greater than the effect for bonuses, which again was supported (p < .01). Because PFP-LTI had no separated effect (i.e., the partial correlation coefficient was not significant), Hypothesis 2 did not pertain to it. Hypothesis 3 predicted that, for any plan type where the

separated PFP relationship was zero, which is true here for LTI, the effect of the connection between pay and performance for that plan should be zero. Indeed, as noted above, the effect of PFP-LTI is nonsignificant (p = .73).

Tests of Relative Sorting Effects (Hypotheses 4–6)

The second half of our hypotheses considered the relative sorting effects of PFP plans, and spe- cifically how the strength of the various PFP links would influence the performance-turnover rela- tionship. Model 3 in Table IV shows the analyses predicting 2002 voluntary turnover and our tests of Hypotheses 4 through 6.

The three hypotheses considered the effects of all three PFP plans when analyzed together. Hypothesis 4 predicted that, when considering merit pay, bonuses, and LTI plans simultaneously, the PFP relationship for merit pay would nega- tively moderate the performance-turnover rela- tionship. This was supported by a negative effect of PFP-Merit on the linear effect of performance (p < .001) and a marginally non-significant effect on the quadratic terms (p = .095). Hypothesis 5 predicted that, when considering merit pay, bonuses, and LTI plans simultaneously, the PFP relationship for LTI would negatively moderate the performance-turnover relationship, which was supported by both significant effects on the linear (p < .01) and quadratic terms (p < .001). Finally, Hypothesis 6 predicted that when consid- ering merit pay, bonuses, and LTI plans simulta- neously, the effect of PFP on turnover would be stronger (i.e., more negative) for merit pay than for LTI. This was supported by the negative effect of PFP-Merit being significantly more negative than the effect for LTI on the linear performance term (p < .05). The effect on the quadratic term was more negative as predicted, although the dif- ference did not approach statistical significance (p = .35).

As noted earlier, we had no a priori predictions regarding the effect of bonuses when considered simultaneously with merit pay and LTI. Our analy- ses revealed that, in the analysis with all three PFP effects, bonuses increased the overall probability of turnover through its significant positive effect on the intercept (p < .001; note that merit pay and LTI had significant negative effects). Bonuses had no effect on either the linear or quadratic perfor- mance terms.

Discussion

Prior work on PFP has generally shown positive incentive and sorting effects, yet this work has not explicitly considered what effects we should

714 HUMAN RESOURCE MANAGEMENT, JULY–AUGUST 2016

Human Resource Management DOI: 10.1002/hrm

The general findings

of incentive and

sorting effects

do hold when

considering PFP

plans independently,

even when other PFP

plans are operating

but are not controlled

for in the analyses.

do indeed support the applicability of expectancy theory for making such predictions. In the rep- lication, the results show that merit pay is more valuable than a bonus or LTI on a dollar-per-dollar basis, and indeed it has stronger incentive and sorting effects.

Third, we expanded on prior theory to con- sider the implications of multiple pay plans being implemented simultaneously. While some prior work has analyzed situations with multi- ple pay plans (Kahn & Sherer, 1990; Salamin & Hom, 2005), both of those studies had one pay plan where rewards were unrelated to perfor- mance; additionally, these studies did not explic- itly consider how multiple pay plans operating simultaneously might be different from plans’ independent effects. We specifically predicted and supported that considering partialed effects is important for incentive effects, while not so for sorting effects.

Our study thus provides a theoretically consis- tent explanation for the mixed results previously observed for merit pay. The effectiveness of merit pay has been repeatedly questioned (Gerhart et al., 2009; Gerhart & Rynes, 2003; Heneman & Werner, 2005). A key concern is that differences in awards between the best and the worst performers are often not large (Gomez-Mejia & Balkin, 1989); others have shown examples where there is actu- ally no relationship between pay and performance in a nominal merit pay plan (e.g., Kahn & Sherer, 1990). These concerns, though, are not completely generalizable to all implementations of merit pay. Rather, when viewed through the lens of expec- tancy theory, they suggest that the merit plans are often poorly implemented because they fail to generate a PFP link. Our findings provide a better understanding of the mechanisms that PFP plans should have so as to yield the desired results. Our results show that it is an overgeneralization to sug- gest there is a single positive effect for any type of PFP plan. Instead, PFP plan effectiveness depends on how strongly pay and performance are linked.

Furthermore, the practical effects of any pay plan will also depend on the budget for the awards, for without resources it is difficult to cre- ate a plan with a strong PFP link. Thus, while our results showed that merit pay has the strongest effects on performance and turnover in multi-PFP environments, if an organization fails to create a link between raises and performance, even if they call it a merit plan, we would not expect merit pay to be an effective tool.

Our study is also one of the few studies to examine the effect of LTIs on individual employ- ees. When considered independently, PFP for LTIs was related to increased performance; however,

expect from PFP plans when employees are per- forming in a multiple PFP plan environment. Considering how different PFP plans operate in the same environment requires us to consider the relative relationships we should expect from PFP plans, thus requiring us to add to our theoretical precision (Edwards & Berry, 2010). It also requires us to consider how the relevant theory is appli- cable to partialed effects—the sorts of effects we expect for one PFP plan when controlling for the effects of other PFP plans. Our findings show that prior PFP research, which has generally focused on a single plan at a time, generalizes to more com- plex environments. Furthermore, the predictions of the relative effectiveness of PFP plans from theory generally hold, and, hence, the expansion of theory to multiplan environments does have some external validity.

This study provides three general forms of theoretical contributions. The first is the replication and confirmation of the generalizability of prior research. Our results show that prior PFP research—be it on a single-PFP plan or in multi-PFP plan environments but with the other plans not consid- ered—replicates and generalizes to multi-PFP plan environments. The general findings of incentive and sorting effects do hold when consid- ering PFP plans independently, even when other PFP plans are operat- ing but are not controlled for in the analyses.

Second, our results provide a test of the theoretical precision of theories that have been related to PFP plans. Because of the differ-

ent sorts of rewards associated with different PFP plans, expectancy theory in particular would pre- dict different effects. Expectancy theory remains one of the dominant decision-making theories (Vancouver, Weinhardt, & Schmidt, 2010), and continues to play an important role in its own right (Cadsby et al., 2007; Gerhart et al., 2009; Kepes, Delery, & Gupta, 2009) in addition to being incorporated into more sophisticated current theories of motivation (e.g., Schmidt & DeShon, 2007; Steel & König, 2006; Vancouver et al., 2010). While the internal validity of expectancy theory has generally been supported (i.e. general posi- tive effects found for expectancy, valence, or the interaction; see Van Eerde & Thierry, 1996), there is far less research testing the external validity of the theory (its ability to make accurate predictions in new contexts) or if its tenets hold for predicting the relative effects of expectancy or valence. We

Human Resource Management DOI: 10.1002/hrm

EVALUATING FORM AND FUNCTIONALITY OF PAY-FOR-PERFORMANCE PLANS 715

Future theoretical

development relevant

to PFP plans requires

attention to both

content (i.e., the

characteristics of

the plan) and context

(i.e., examining a plan

in light of other PFP

plans that may be in

place).

motivation; rather, the predictions were based on approximations of the relationships between pay and performance and from the characteris- tics of the plans and supervisory decisions. While this is not the first study to estimate PFP relation- ships mathematically (e.g., Kahn & Sherer, 1990; Salamin & Hom, 2005; Trevor et al., 1997), it is not a direct test of the internal validity of the related theories. It would certainly be valuable to see how individuals perceive the sort of PFP linkages in which they are operating. While our use of theory and supported hypotheses provide evidence of the external validity of relevant theory, and particu- larly expectancy theory, our article does not con- tribute to testing of the theories’ internal validity.

Our single context also limits the generaliz- ability of our findings. Other forms of PFP and other simultaneous combinations of PFP plans should be examined to provide greater precision in our understand- ing of the effectiveness of compen- sation plans. There are many other types of PFP plans, and even more potential PFP portfolios. Because organizations are more likely to use “hybrid plans” than independent pay plans (Gerhart, 2000; Gerhart & Fang, 2014; Gomez-Mejia & Balkin, 1992; Gomez-Mejia, Berrone, & Franco-Santos, 2010; Milkovich et al., 2013), understanding how the characteristics of multiple PFP plans simultaneously affect performance and voluntary turnover is crucial for organizations to design effective compensation systems.

There are also limitations to the nature of our data. Unaddressed in this article, there may exist inter- actions between PFP relationships. It is also pos- sible that pay systems have effects beyond one year. In our analyses, we examine the effect of pay outcomes on performance or turnover in the subsequent year. It is possible, for example, that long-term incentives, may have effects that occur in subsequent years, or the strength of effects may change over time. It is beyond the scope of our study, in addition to the capabilities of our data, to consider the potential multiyear effects of hybrid pay systems, and thus our results may not be fully capturing the set of effects associated with these plans.

In short, our article represents a single case of a multi-PFP environment, and more research on more and different plans is needed. While performing such research will obviously require significant industry-academic cooperation to

in this context, the link between pay and perfor- mance after partialing out the links with the other pay plans was not significant. This limited the potential incentive effect of LTI when considered in conjunction with the other PFP plans, although it still had a sorting effect. Further research on LTIs would be useful, as our results show that, in gen- eral, the effectiveness of a compensation plan is a function of its characteristics. We only examined a single LTI plan in this article; other plans may vary in terms of their vesting requirements and the specific reward granted, and thus can be more complex (Moynihan, 2013).

Motivational theories have strongly sup- ported the underlying mechanisms of PFP plans regarding the extent to which financial rewards can motivate employees to higher performance and the desirable behaviors that organizations expect. Situations have become, for both organi- zations and employees, more multifaceted due to organizations providing more complex compen- sation system environments and employees being covered by multiple PFP plans. Future compensa- tion research needs to consider more carefully the effectiveness of PFPs. Indeed, each PFP has a dif- ferent form and set of characteristics, and all of the different combinations of multiple PFPs that organizations provide will have relative effects on various important outcomes. This creates more complex decision making and motivational pro- cesses that need greater research attention.

Overall, a key theoretical contribution from this paper is our demonstration of the potential and utility associated with developing greater theoretical precision (Edwards & Berry, 2010). Simply calling a plan PFP is insufficient; and while general directional effects are valid, theory can be extended to predict the relative effective- ness of PFP plans. Future theoretical development relevant to PFP plans requires attention to both content (i.e., the characteristics of the plan) and context (i.e., examining a plan in light of other PFP plans that may be in place).

Limitations

This article has a number of advantages over pre- vious PFP studies. We used longitudinal data con- trolling for prior performance to examine both the incentive and sorting effects of PFP plans. The study also considered the different effects of the characteristics of multiple PFP plans simultane- ously. Like all research, though, this study is not without limitations, and it is important to point out the key issues that threaten the potential gen- eralizability of our findings.

From a theoretical perspective, this study did not directly assess individual perceptions of

716 HUMAN RESOURCE MANAGEMENT, JULY–AUGUST 2016

Human Resource Management DOI: 10.1002/hrm

Companies may

want to avoid raises

because of the

increase to fixed

labor costs, but our

findings show that

minimizing merit

pay means giving

up a powerful PFP

tool. Raises had

larger effects than

bonuses and LTI, and

only raises had both

incentive and sorting

effects.

and rewards, and adjust policy as necessary to ensure stronger links. Most companies using PFP have individual performance data (WorldatWork, 2012). A related contribution of our study is our demonstration on how companies can use HR data to see how strongly their plans link pay and performance, and thus change policy based on using their HR data.

Third, our results raise interesting questions about the use of bonuses. While bonuses consid- ered independently did have positive incentive and sorting effects, after controlling for the effects of other PFP plans, there was actually a positive effect on turnover. This is consistent with the finding by Sturman and Short (2000), who found a positive effect of bonus satisfaction on turnover intentions after controlling for satisfaction with other pay dimensions. Organizations may ben- efit by using their available data to make similar tests in their own organizations to see if their pay plans, and bonuses in particular, are having unin- tended consequences (Pfeffer, 1998).

Finally, this study emphasizes the importance of the relative effectiveness of different types of PFP plans in multi-PFP plan environment. It is very common today for organizations to provide their employees with more than one type of PFP. As many organizations are focusing more on PFP plans, implementing a single or multiple PFP plan(s) is not a differentiator among organiza- tions. With the results of our study, organizations must identify the complexity of pay environ- ments and distinguish between forms and charac- teristics of different PFP plans and across PFP plan types to better understand how these factors influ- ence employees’ motivation and decision-making processes.

provide the sort of data needed to conduct such tests, there is great practical and theoretical value that could be provided by such work.

Implications for Practice

Given the prevalence of mul- tiple PFP plan environments, our research into the effects of multi- ple PFP plans operating simultane- ously has important implications for practice. First, we show that employees do, on average, respond rationally to incentives. Companies may want to avoid raises because of the increase to fixed labor costs, but our findings show that mini- mizing merit pay means giving up a powerful PFP tool. Raises had larger effects than bonuses and LTI, and only raises had both incentive and sorting effects. Our findings do show that companies can use other pay forms to get the same effects as raises, but it would require stronger PFP links and larger rewards. The trade-off between higher one-time costs versus greater fixed labor costs thus becomes a cost-benefit decision (cf. Sturman, Trevor, Boudreau, & Gerhart, 2003).

Second, our study shows that companies can apply PFP-related theory to the design of PFP plans,

and thus take advantage of evidence-based management (e.g., Rousseau, 2006; Rousseau & McCarthy, 2007). Companies can specifically look at the degree to which managers link pay

SANGHEE PARK is an assistant professor of human resource management in Rutgers University’s School of Management and Labor Relations. She received her PhD in human

resources from the Hotel School at Cornell University. Her primary research interests are on

the infl uence of compensation systems, particularly looking at the intersection of pay and

motivation, and the dynamics of the multiple dyadic relationships within multiple hierarchies

in organization. Her research has been published in the Journal of Applied Psychology.

MICHAEL C. STURMAN (PhD, Cornell University) is the Kenneth and Marjorie Blanchard Professor of Human Resources, and the associate dean for faculty development at Cornell

University’s School of Hotel Administration. There, he teaches courses on human resource

management and compensation. His research focuses on the prediction of individual job

performance over time and the infl uence of compensation systems. His work is published

in journals such as the Journal of Applied Psychology, Academy of Management Journal,

Personnel Psychology, and Journal of Management. Michael is also a Senior Professional of

Human Resources as certifi ed by the Society for Human Resource Management.

Human Resource Management DOI: 10.1002/hrm

EVALUATING FORM AND FUNCTIONALITY OF PAY-FOR-PERFORMANCE PLANS 717

Carter, M. E., & Lynch, L. J. (2004). The effect of stock option

repricing on employee turnover. Journal of Accounting

and Economics, 37, 91–112.

Cascio, W. F., & Aguinis, H. (2008). Research in industrial and

organizational psychology from 1963 to 2007: Changes,

choices, and trends. Journal of Applied Psychology, 93,

1062–1081.

Cohen, K. (2011, September). Salary budget increases going

for a slow ride: WorldatWork 2011–2012 budget survey

results. Workspan, 54, 30–37.

Core, J. E., & Guay, W. R. (2001). Stock option plans for non-

executive employees. Journal of Financial Economics,

61, 253–287.

Daily C. M., Certo S. T., & Dalton D. R. (2002). Executive stock

option repricing: Retention and performance reconsid-

ered. California Management Review, 44, 8–22.

Deadrick, D. L., & Gibson, P. A. (2007). An examination of the

research-practice gap in HR: Comparing topics of interest

to HR academics and HR professionals. Human Resource

Management Review, 17, 131–139.

Delery, J. E., & Doty, D. H. (1996). Modes of theorizing in

strategic human resource management: Tests of univer-

salistic, contingency, and confi gurational performance

predictions. Academy of Management Journal, 39,

802–835.

Dunford, B. B., Boudreau, J. W., & Boswell, W. R. (2005).

Out-of-the-money: The impact of underwater stock

options on executive job search. Personnel Psychology,

58, 67–101.

Dunford, B. B., Oler, D. K., & Boudreau, J. W. (2008). Under-

water stock options and voluntary executive turnover: A

multidisciplinary perspective integrating behavioral and

economic theories. Personnel Psychology, 61, 687–726.

Edwards, J. R., & Berry, J. W. (2010). The presence of some-

thing or the absence of nothing: Increasing theoreti-

cal precision in management research. Organizational

Research Methods, 13, 668–689.

Eisenhardt, K. M. (1989). Agency theory: An assessment and

review. Academy of Management Review, 14, 57–74.

Eisenberger, R., Rhoades, L., & Cameron, J. (1999). Does

pay for performance increase or decrease perceived self-

determination and intrinsic motivation? Journal of Per-

sonality and Social Psychology, 77, 1026–1040.

Fusilier, M. R., Ganster, D. C., & Middlemist, R. D. (1984). A

within-person test of the form of the expectancy theory

model in a choice context. Organizational Behavior and

Human Performance, 34, 323–342.

Gardner, T. M., Wright, P. M., & Moynihan, L. M. (2011). The

impact of motivation, empowerment, and skill-enhancing

practices on aggregate voluntary turnover: The mediat-

ing effect of collective affective commitment. Personnel

Psychology, 64, 315–350.

Gerhart, B. (1990). Voluntary turnover and alternative job

opportunities. Journal of Applied Psychology, 75,

467–476.

Gerhart, B. (2000). Compensation strategy and organiza-

tional performance. In S. L. Rynes & B. Gerhart (Eds.),

Compensation in organizations (pp. 151–194). San

Francisco, CA: Jossey-Bass.

Gerhart, B., & Fang, M. (2014). Pay for (individual) perfor-

mance: Issues, claims, evidence and the role of sorting

effects. Human Resource Management Review, 24, 41–52.

References

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing

and interpreting interactions. Newbury Park, CA: Sage.

Allen, D. G., & Griffeth, R. W. (2001). Test of a mediated per-

formance-turnover relationship highlighting the moderat-

ing roles of visibility and reward contingency. Journal of

Applied Psychology, 86, 1014–1021.

Bandiera, O., Barankay, I., & Rasul, I. (2007). Incentives for

managers and inequality among workers: Evidence from

a fi rm-level experiment. Quarterly Journal of Economics,

122, 729–773.

Banker, R. D., Lee, S. Y., Potter, G., & Srinivasan, D. (1996).

Contextual analysis of performance impacts of outcome-

based incentive compensation. Academy of Management

Journal, 39, 920–948.

Banker, R. D., Lee, S. Y., Potter, G., & Srinivasan, D. (2001). An

empirical analysis of continuing improvements following

the implementation of a performance-based compensa-

tion plan. Journal of Accounting and Economics, 30,

315–350.

Bartol, K. M., & Durham, C. C. (2000). Incentives: Theory

and practice. In C. L. Cooper & E. A. Locke, (Eds.), Indus-

trial and organizational psychology (pp. 1–33). Oxford,

England: Blackwell.

Bartol, K. M., & Locke, E. A. (2000). Incentives and

motivation. In S. L. Rynes & B. A. Gerhart (Eds.), Com-

pensation in organizations: Current research and practice

(pp. 273–310). San Francisco, CA: Jossey-Bass.

Batt, R., & Colvin, A. J. S. (2011). An employment systems

approach to turnover: Human resources practices, quits,

dismissals and performance. Academy of Management

Journal, 54, 695–717.

Becker, B. E., & Huselid, M. A. (1992). The incentive effects

of tournament compensation systems. Administrative

Science Quarterly, 37, 336–350.

Beer M., & Cannon M. D. (2004). Promise and peril in

implementing pay-for-performance. Human Resource

Management, 43, 3–20.

Bhattacharya, M., Gibson, D. E., & Doty, D.H. (2005). The

effects of fl exibility in employee skills, employee behav-

iors, and human resource practices on fi rm performance.

Journal of Management, 31, 622–640.

Bloom, M, & Michel, J. (2002). The relationship among

organizational context, pay dispersion, and managerial

turnover. Academy of Management Journal, 45, 33–42.

Bonner, S. E., & Sprinkle, G. B. (2002). The effects of mon-

etary incentives on effort and task performance: Theories,

evidence, and a framework for research. Accounting,

Organizations and Society, 27, 303–345.

Brown, M., Sturman, M.C., & Simmering, M. (2003). Com-

pensation policy and organizational performance: The

effi ciency, operational, and fi nancial implications of

pay levels and pay structure. Academy of Management

Journal, 46, 752–762.

Cadsby, B., Song, F., & Tapon, F. (2007). Sorting and incentive

effects of pay for performance: An experimental investi-

gation. Academy of Management Journal, 50, 387–405.

Cappelli, P., & Conyon, M. J. (2011). Stock option exercise

and gift exchange relationships: Evidence for a large US

company. NBER working paper series. Retrieved from

http://www.nber.org/papers/w16814

718 HUMAN RESOURCE MANAGEMENT, JULY–AUGUST 2016

Human Resource Management DOI: 10.1002/hrm

Kepes, S., Delery, J., & Gupta, N. (2009). Contingencies in

the effects of pay range on organizational effectiveness.

Personnel Psychology, 62, 497–531.

Kwong, J. Y. Y., & Wong, K. F. E. (2014). Fair or not fair? The

effects of numerical framing on the perceived justice of

outcomes. Journal of Management, 40, 1558–1582.

Lambert, R. A., Larcker, D. F., & Weigelt, K. (1993). The struc-

ture of organizational incentives. Administrative Science

Quarterly, 38, 438–461.

Lance, C. E. (1988). Residual centering, exploratory and

confi rmatory moderator analysis, and decomposition of

effects in path models containing interactions. Applied

Psychological Measurement, 12, 163–175.

Lawler, E. E. (1971). Pay and organizational effectiveness: A

psychological view. New York, NY: McGraw-Hill.

Lawler, E. E. (2000). Rewarding excellence: Pay strategies for

the new economy. San Francisco, CA: Jossey-Bass.

Lazear, E. P. (1986). Salaries and piece rates. Journal of

Business, 59, 405–431.

Lazear, E. P. (2000). Performance pay and productivity.

American Economic Review, 90, 1346–1361.

Markham, S. (1988). Pay-for-performance dilemma revisited:

Empirical example of the importance of group effects.

Journal of Applied Psychology, 73, 172–180.

Mehran, H., & Yermack, D. (1996, May). Stock-based com-

pensation and top management turnover. NYU Working

Paper, No. FIN-96-035. Retrieved from SSRN: http://ssrn

.com/abstract=1298300

Milkovich, G. T., Newman, J. M., & Gerhart, B. (2013). Com-

pensation (11th ed.). New York, NY: McGraw-Hill, Irwin.

Mitchell, T. R., & Mickel, A. E. (1999). The meaning of money:

An individual difference perspective. Academy of

Management Review, 24, 568–578.

Mizruchi, M. S., Stearns, L. B., & Fleischer, A. (2011). Getting a

bonus: Social networks, performance, and reward among

commercial bankers. Organization Science, 22, 42–59.

Moynihan, D. P. (2013). Long-term incentives: Best practice

vs. best fi t. Workspan, 58(5), 58–62.

National Center for Employee Ownership. (2012). A statisti-

cal profi le of employee ownership, Oakland, CA: Author.

Nyberg, A. J., Pieper, J. R., & Trevor, C. O. (2013).

Pay-for- performance’s effect on future employee

performance: Integrating psychological and eco-

nomic principles toward a contingency perspective.

Journal of Management. Advance online publication.

doi:10.1177/0149206313515520.

Oyer, P., & Schaefer, S. (2005). Why do some fi rms give stock

options to all employees? An empirical examination of

alternative theories. Journal of Financial Economics, 76,

99–133.

Pearce, J. L., Stevenson, W. B., & Perry, J. L. (1985). Manage-

rial compensation based on organizational performance:

A time series analysis of the effects of merit pay.

Academy of Management Journal, 28, 261–278.

Pfeffer, J. (1998, May–June). Six dangerous myths about

pay. Harvard Business Review, 76, 109–119.

Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical

linear models: Applications and data analysis methods

(2nd ed.). Thousand Oaks, CA: Sage.

Raudenbush, S. W., Bryk, A. S., Cheong, Y. F., Congdon, R.

T., & du Toit, M. (2011). HLM 7. Lincolnwood, IL: Scientifi c

Software International Inc.

Gerhart, B., & Milkovich, G. T. (1992). Employee compensa-

tion: Research and practice. In M. D. Dunnette & L. M.

Hough (Eds.), Handbook of industrial and organizational

psychology (2nd ed., pp. 481–569). Palo Alto, CA:

Consulting Psychologists Press.

Gerhart, B., Minkoff, H. B., & Olsen, R. N. (1995). Employee

compensation: Theory, practice, and evidence. In G. R.

Ferris, S. D. Rosen, & D. T. Barnum (Eds.), Handbook of

human resource management (pp. 528–547). Cambridge,

MA: Blackwell.

Gerhart, B., & Rynes, S. L. (2003). Compensation: Theory,

evidence, and strategic implications. Thousand Oaks, CA:

Sage.

Gerhart, B., Rynes, S. L., & Fulmer, I. S. (2009). Pay and per-

formance: Individuals, groups, and executives. Academy

of Management Annals, 3, 251–315.

Gomez-Mejia, L. R., & Balkin, D. B. (1989). Effectiveness

of individual and aggregate compensation strategies.

Industrial Relations, 28, 431–445.

Gomez-Mejia, L. R., & Balkin, D. B. (1992). Compensation,

organizational strategy, and fi rm performance. Cincinnati,

OH: Southwestern.

Gomez-Mejia, L. R., Berrone, P., & Franco-Santos, M. (2010).

Compensation and organizational performance. Armonk,

NY: M. E. Sharpe.

Green, L., & Myerson, J. (2004). A discounting framework

for choice with delayed and probabilistic rewards.

Psychological Bulletin, 130, 769–792.

Guthrie, J. P. (2007). Remuneration: Pay effects at work. In P.

Boxall, J. Purcell, & P. Wright (Eds.), Oxford handbook of

human resource management (pp. 344–363). New York,

NY: Oxford University Press.

Hall, B. J., & Knox, T. A. (2004). Underwater options and the

dynamics of executive pay-to-performance sensitivities.

Journal of Accounting Research, 42, 365–412.

Harris, M. M., Gilbreath, B., & Sunday, J. A. (1998). A longi-

tudinal examination of a merit pay system: Relationships

among performance ratings, merit increases, and total

pay increase. Journal of Applied Psychology, 83, 825–831.

Hayes, R. M. (2004). Discussion of underwater options and

the dynamics of executive pay-to-performance sensitivi-

ties. Journal of Accounting Research, 42, 413–421.

Heneman, R. L., & Werner, J. M. (2005). Merit pay— linking

pay to performance in a changing world (2nd ed.).

Greenwich, CT: IAP.

Hull, J. C. (2012). Options, futures and other derivatives (8th

ed.). Upper Saddle River, NJ: Prentice Hall.

Jackofsky, E. F., Ferris, K. R., & Breckenridge, B. G. (1986).

Evidence for a curvilinear relationship between job

performance and turnover. Journal of Management, 12,

105–111.

Jenkins, G. D., Mitra, A., Gupta, N., & Shaw, J. D. (1998).

Are fi nancial incentives related to performance? A meta-

analytic review of empirical research. Journal of Applied

Psychology, 83, 777–787.

Jones, D. C., & Kato, T. (1995). The productivity effects of

employee stock-ownership plans and bonuses: Evidence

from Japanese panel data. American Economic Review,

85, 391–414.

Kahn, L. M., & Sherer, P. D. (1990). Contingent pay and

managerial performance. Industrial and Labor Relations

Review, 43, 107S–120S.

Human Resource Management DOI: 10.1002/hrm

EVALUATING FORM AND FUNCTIONALITY OF PAY-FOR-PERFORMANCE PLANS 719

Rousseau, D. M. (2006). Is there such a thing as “evidence-

based management? Academy of Management Review,

31, 256–269.

Rousseau, D. M., & Ho, V. T. (2000). Psychological contract

issues in compensation. In S. L. Rynes & B. A. Gerhart,

(Eds.), Compensation in organizations: Current research

and practice (pp. 273–310). San Francisco, CA: Jossey-

Bass.

Rousseau, D. M., & McCarthy, S. (2007). Educating

managers from an evidence-based perspective.

Academy of Management Learning and Education, 6,

94–101.

Rynes, S. L., Gerhart, B., & Parks, L. (2005). Personnel psy-

chology: Performance evaluation and pay for perfor-

mance. Annual Review of Psychology, 56, 571–600.

Rynes, S. L., Giluk, T. L., & Brown, K. G. (2007). The very sep-

arate worlds of academic and practitioner periodicals in

human resource management: Implications for evidence-

based management. Academy of Management Journal,

50, 987–1008.

Salamin, A., & Hom, P. W. (2005). In search of the elusive

U-shaped performance-turnover relationship: Are high

performing Swiss bankers more liable to quit? Journal of

Applied Psychology, 90, 1204–1216.

Schaubroeck, J., Shaw, J. D., Duffy, M. K., & Mitra, A. (2008).

An under-met and over-met expectations model of

employee reactions to merit raises. Journal of Applied

Psychology, 93, 424–434.

Schmidt, A. M., & DeShon, R. P. (2007). What to do? The

effects of discrepancies, incentives, and time on dynamic

goal prioritization. Journal of Applied Psychology, 92,

928–941.

Schwab, D. P. (1991). Contextual variables in employee

performance-turnover relationships. Academy of

Management Journal, 34, 966–975.

Schwab, D. P., & Olson, C. A. (1990). Merit pay practices:

Implications for pay-performance relationships. Industrial

and Labor Relations Review, 43, 237S–255S.

Shaw, J. D., Duffy, M. K., Mitra, A., Lockhart, D. E., & Bowler,

M. (2003). Reactions to merit pay increases: A longitu-

dinal test of a signal sensitivity perspective. Journal of

Applied Psychology, 88, 538–544.

Shaw, J. D., & Gupta, N. (2007). Pay system characteristics

and quit patterns of good, average, and poor performers.

Personnel Psychology, 60, 903–928.

Steel, P., & König, C. J. (2006). Integrating theories of

motivation. Academy of Management Review, 31,

889–913.

Sturman, M. C. (2003). Searching for the inverted U-shaped

relationship between time and performance: Meta-

analyses of the experience/performance, tenure/

performance, and age/performance relationships.

Journal of Management, 29, 609–640.

Sturman, M. C. (2007). The past, present, and future of

dynamic performance research. Research in Personnel

and Human Resource Management, 26, 49–110.

Sturman, M. C., Cheramie, R. A., & Cashen, L. H. (2005). The

impact of job complexity and performance measurement

on the temporal consistency, stability, and test-retest reli-

ability of employee job performance ratings. Journal of

Applied Psychology, 90, 269–283.

Sturman, M. C., Shao, L., & Katz, J. (2012). The effect of cul-

ture on the curvilinear relationship between performance

and turnover. Journal of Applied Psychology, 97, 46–62.

Sturman, M. C., & Short, J. C. (2000). Lump-sum bonus sat-

isfaction: Testing the construct validity of a new pay satis-

faction dimension. Personnel Psychology, 53, 673–700.

Sturman, M. C., Trevor, C. O., Boudreau, J. W., & Gerhart, B.

(2003). Is it worth it to win the talent war? Evaluating the

utility of performance-based pay. Personnel Psychology,

56, 997–1035.

Toh, S. M., Morgeson, F. P., & Campion, M. A. (2008). Human

resource confi gurations: Investigating fi t with the organiza-

tional context. Journal of Applied Psychology, 93, 864–882.

Trevor, C. O., Gerhart, B., & Boudreau, J. W. (1997). Voluntary

turnover and job performance. Curvilinearity and the

moderating infl uences of salary growth and promotions.

Journal of Applied Psychology, 82, 44–61.

Trevor, C. O., Reilly, G., & Gerhart, B. (2012). Reconsidering

pay dispersion’s effect on the performance of interde-

pendent work: Reconciling sorting and pay inequality.

Academy of Management Journal, 55, 585–610.

Van Eerde, W., & Thierry, H. (1996). Vroom’s expectancy mod-

els and work-related criteria: A meta-analysis. Journal of

Applied Psychology, 81, 575–586.

Vancouver, J. B., Weinhardt, J. M., & Schmidt, A. M. (2010).

A formal, computational theory of multiple-goal pursuit:

Integrating goal-choice and goal-striving processes.

Journal of Applied Psychology, 95, 985–1008.

Viswesvaran, C., Ones, D. S., & Schmidt, F. L. (1996). Com-

parative analysis of the reliability of job performance

ratings. Journal of Applied Psychology, 81, 557–574.

Vroom, V. H. (1964). Work and motivation. New York, NY:

Wiley.

Williams, C. R. (1999). Reward contingency, unemployment

and functional turnover. Human Resource Management

Review, 9, 549–576.

WorldatWork. (2010). Compensation programs and practices

2010. Washington, DC: Author.

WorldatWork. (2012). Compensation programs and practices

2012. Washington, DC: Author.

Wright, P. M., Gardner, T. M., Moynihan, L. M., & Allen, M.

R. (2005). The relationship between HR practices and

fi rm performance: Examining causal order. Personnel

Psychology, 58, 409–446.

Wright, T. A., & Cropanzano, R. R. (1998). Emotional exhaus-

tion as a predictor of job performance and voluntary

turnover. Journal of Applied Psychology, 83, 486–493.

Zenger, T. R. (1992). Why do employers only reward extreme

performance? Examining the relationships among per-

formance, pay, and turnover. Administrative Science

Quarterly, 37, 19 .

Copyright of Human Resource Management is the property of John Wiley & Sons, Inc. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.