Published online ahead of print October 28, 2011 SCIENCE Articles in Advance, pp. 1–14 issn 0025-1909 — eissn 1526-5501 http://dx.doi.org/10.1287/mnsc.1110.1444 © 2011 INFORMS be .org. not ms y

or On Evaluation Costs in Strategic Factor Markets: ma The Implications for Competition and file Organizational Design The . missions@inf per ibers David Gaddis Ross

to Columbia School, New York, New York 10027, [email protected] subscr his paper uses a formal model to study how evaluation costs affect competition for resources in strategic fac- policy to Ttor markets. It finds that relative scarcity may not always benefit resource sellers. Rather, when competition le

this among resource investors passes a certain threshold intensity, miscoordination among investors increases to the point that sellers’ expected profits decline. This paper extends the model to consider how investors organize to ailab

v overcome managerial agency in resource evaluation. Two organizational designs are considered: (a) incentiviza- a tion, wherein a lower-level manager is motivated by an incentive contract to evaluate resources for an investor, and (b) supervision, wherein evaluation is either handled directly or closely monitored by headquarters. The regarding model suggests that competition among investors will be associated with a greater use of supervision and that made investors using supervision will tend to make lower offers. This paper also finds that supervision will be more is common when valuable resources are rare. Key words: ; organizational studies; design; motivation–incentives questions which History: Received June 16, 2008; accepted July 22, 2011, by Bruno Cassiman, business strategy. Published y online in Articles in Advance. an ersion, v send 1. Introduction are subject to frictions that interfere with the nor-

ance Two central tenets of the resource-based view of mal effects of competition. Examples include imper- fect price discrimination between groups of buyers

Please the firm are (1) that implementing product market Adv . requires valuable resources (Penrose 1959, (Rosenthal 1980, Stahl 1989), externalities in fixed and in variable costs when simultaneously competing for site Wernerfelt 1984) and (2) that many valuable strategic

s resources are acquired in “strategic factor markets” multiple business opportunities (Lang and Rosenthal ticles (Barney 1986). Given the standard view of markets, 1991), the winner’s curse in bidding (Bulow and Ar Klemperer 2002), externalities in the ownership of

author’ in which rents accrue to those sellers and buyers goods (Jehiel et al. 1996), and rent dissipation in this with greater “relative scarcity” (e.g., Williamson 1975,

the downstream markets, which may affect bidding for to Pfeffer and Salancik 1978, Porter 1979, Brandenburger and Stuart 1996, Dyer and Singh 1998), the strategy strategic inputs like intellectual property (Anton and

ight Yao 2008). These frictions endogenously affect the literature has generally adopted the view that the yr decision of how and even whether a competitor par- profits of resource investors will decline, and the prof- including

cop ticipates in a given market. In some circumstances, , its of resource sellers will increase, as the number of competition exacerbates these frictions to the point investors per seller climbs (e.g., Peteraf 1993).1 Indeed, that competitors compete less aggressively and may holds ebsite Adegbesan’s (2009) formal model of a strategic factor

w even choose not to enter the market, to the detriment market reached this conclusion. of upstream and downstream market participants. However, within diverse strands of the This paper’s first contribution is to construct a for- other literature, authors have observed that some markets

y mal model to apply similar reasoning to the workings INFORMS

an of strategic factor markets and thereby demonstrate 1 Accordingly, formal work on resource acquisition has studied how that the effect of relative scarcity on the division of on firms can mitigate competition for strategic resources by, for exam- rents between resource sellers and resource investors yright: ple, focusing on resources that best complement the firm’s exisiting is considerably more subtle than the strategy liter- resources (Adegbesan 2009), developing a competence at deploy- ature has appreciated. This paper then goes on to Cop posted ing a particular kind of resource (Makadok 2001, 2002), collect- ing the most useful information about which resources to acquire explore how competition affects how firms organize (Makadok and Barney 2001), or judiciously scheduling resource to evaluate and acquire resources in strategic factor acquisition across time (Pacheco-de-Almeida and Zemsky 2007). markets.

1 Ross: On Evaluation Costs in Strategic Factor Markets 2 Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS

Central to this paper’s argument is that whereas suggests an important role for strategy in shaping the some strategic resources are acquired gradually structure of strategic factor markets. be through firm behavior (Dierickx and Cool 1989) or The importance of resource evaluation costs raises .org.

not have values that are widely known, other strategic the question of how firms can best organize to ensure ms y

or resources must be obtained through markets and will that managers perform this “entrepreneurial service”

ma have values that are not fully apparent. Examples (Penrose 1959) properly in the face of competition include the acquisition of another company (Barney from other firms. In particular, we wish to understand file 1986, p. 1231) as well as technology patented by an the substantial variation that we observe within and independent inventor or a tract of land adjoining a between firms along this organizational dimension in The .

missions@inf planned new highway exit (Makadok 2001, p. 391). In a variety of strategic factor markets. Examples include these cases, a responsible investor requires a lengthy the following: per ibers and expensive evaluation to ascertain the value of a 1. Mergers and acquisitions (M&As). While firms to resource before acquiring it. This paper’s model shows often use a centralized corporate development func-

subscr formally that when these evaluation costs are consid- tion to vet acquisitions, many firms will, on some

policy occasions, delegate responsibility to vet an acquisition to ered, competition among investors increases sellers’ to business development personnel at lower levels le profits and reduces investors’ profits only until a crit- this ical number of investors is reached. As the number of in the hierarchy (Deloitte Financial Advisory Services ailab investors increases beyond that point, investors’ prof- 2010). GE Capital, a very active acquirer, has been v a its do not change and sellers’ profits decrease. cited as a leading example of this practice (Ashkenas Although counterintuitive at first, this result has et al. 1998). In many companies, the chief financial regarding officer (CFO) takes a role in vetting acquisitions (G.A.

made an intuitive explanation, which we sketch here and Kraut Company 2008), but the CFO’s role may range is develop at greater length in the body of this paper. As one would expect, the presence of more investors from identification and evaluation (e.g., consumer encourages each investor to make a higher offer to products firm Newell (Collis and Johnson 1998)) to questions which a limited financial vetting of recommendations from

y avoid being outbid, to the benefit of sellers. However,

an if the number of investors increases beyond a crit- business units, which have primary responsibility for ical number and they all actively participate in the identifying and evaluating targets (e.g., telecommuni- ersion, v market, many investors will incur the cost of eval- cations firm Polycom (Roberto and Carioggia 2005)). send uating a resource but not succeed in acquiring one, 2. Greenfield expansion. Introducing new products,

ance driving investors’ expected profits below zero. This opening new plants, and entering new markets require firms to acquire resources such as capital equipment,

Please is not sustainable as an equilibrium. Nor is it possi- Adv . ble for a subset of investors to earn positive profits sites for operations, and new managerial personnel. in The autonomy given to business units for organic site and simultaneously block other investors from partic-

s ipating in the market. Thus, absent a collusive coor- expansion varies considerably within and among ticles dination device among investors, the only possible firms (Hedlund 1981, Birkinshaw 1997, Birkinshaw Ar equilibria are those where each investor randomizes et al. 1998). For instance, SK Chemical, a part of author’ between (a) incurring the cost of evaluating a resource Korea’s SK chaebol, frequently relies on the chairman’s this

the and (b) staying out of the market. This creates the pos- office when deciding to acquire strategic assets, but to sibility of miscoordination that inures to the detriment the head of the petrochemical division was given sig-

ight of sellers in two mutually reinforcing ways. First, nificant autonomy in the decision to acquire a site for 2 yr purely by chance, a large number of investors may a large greenfield chemical plant in Poland in 2003. including Likewise, within the same restaurant chain, it is com- cop opt to stay out of the market at the same time with the , result that even a seller with a valuable resource will not mon to observe a chief executive officer (CEO) taking receive any offers for it. Second, the implied ex post an active role in evaluating the acquisition of some holds ebsite potential restaurant sites while delegating evaluation w market power of an investor that does evaluate a given seller’s resource is higher, because the investor of other sites to lower-level managers (Bradach 1998). may correctly guess that it is one of the few—perhaps, 3. Customer acquisition. Acquiring new customers other frequently requires significant expenditures to gener-

y the only—investor in a position to make an offer for INFORMS ate leads and subsequent sales. In some such contexts, an that resource. As a result, the average “best offer” a seller receives declines once the number of actively the resource seller would be a third party, such as a on competing investors passes the critical number. It fol- market firm. In other contexts, the customer yright: lows that at low levels of competition, the interests of may be, in effect, selling a relationship with itself. One investors and sellers are in conflict: investors would example is commercial banking, in which a bank that Cop posted prefer less competition with other investors, and sell- extends a loan to a customer gains a financial claim ers would prefer more. At high levels of competition, however, sellers would prefer less competition, too. This 2 The author advised SK Chemical on the project. Ross: On Evaluation Costs in Strategic Factor Markets Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS 3

on that customer, as well as limited market power because larger, less frequent bonuses are “riskier” because of switching costs (Rajan 1992, Bharath et al. than smaller, more frequent bonuses, the bonus must be 2007). The bank also gains insight into the customer’s rise more than proportionately to compensate the .org.

not need for additional services like cash management, agent for this increase in risk. ms y

or foreign exchange hedging, and securities underwrit- It follows that in a competitive context, an investor

ma ing, which may be necessary for the relationship to using incentivization can economize on wage costs be profitable (Comptroller of the Currency 2003, Koch by making the seller a higher offer—which is more file and MacDonald 2010). Where a loan is approved in likely to be accepted—but no such benefit exists for a bank’s hierarchy varies considerably by lender and an investor using supervision. There are two impli- The . missions@inf loan market (Koch and MacDonald 2010) and even cations. First, in competition with other investors, within the same bank branch (Liberti and Mian 2009). an investor using incentivization will tend to make per ibers To shed light on this variation within and across higher offers than an investor using supervision. Sec- to firms in the of resource evaluation, this ond, as the number of investors competing for a

subscr paper extends the model of a strategic factor market resource increases, so too does the probability that an

policy investor’s offer will be rejected in favor of a competi- to to develop a novel formal analysis of how compe- tor’s, raising the relative cost of incentivization. Thus, le tition affects managerial agency in resource evalu- this ation.3 In the extended model, a resource investor supervision should be associated with more competi- ailab has a principal who acts to maximize the investor’s tive strategic factor markets. v a profits, such as an owner or an executive in corpo- In the context of strategic resource evaluation, rate headquarters (a CEO, CFO, or corporate develop- switching between supervision and incentivization regarding can typically be effected quickly, and, as noted above, made ment manager, for example). To evaluate and possibly the same firm will frequently employ both organiza- is make an offer for a resource, the principal has two organizational design options. One option is “incen- tional designs. For example, senior managers at head- tivization,” wherein the principal hires a risk-averse quarters may focus on evaluating the acquisition of questions which one strategic asset (supervision) while assigning the y agent to act on the principal’s behalf. This gives rise

an to a delegated expertise problem such as those in evaluation of another strategic asset to business devel- Lambert (1986) and Demski and Sappington (1987), opment personnel at lower levels in the hierarchy ersion, v where the agent represents a manager outside of cor- (delegation); this was precisely the case in the exam- send porate headquarters to whom primary responsibil- ple of SK Chemical described above. Even within the

ance ity for gathering information about a resource has same branch, a bank may use a centralized credit con- trol department to screen some loans (supervision)

Please been delegated, such as an acquisition “leader” in GE Adv . Capital’s terminology (Ashkenas et al. 1998), a busi- while granting authority to relationship managers to in screen other loans (delegation). An appealing feature site ness unit manager, or a relationship manager pursu-

s ing a new customer. Another option is “supervision,” of the model in this paper is that it captures this ticles which has two interpretations: either (a) the princi- within-firm variation, which may appear ad hoc to Ar pal monitors the agent in the sense of Mookherjee outside observers. Specifically, the model generates author’ and Png (1992) by employing a costly technology to an equilibrium in mixed strategies among a priori this

the eliminate the agent’s opportunity for agency behav- homogenous investors, wherein investors randomize to ior, or (b) the principal evaluates the resource directly, between organizational designs, using incentivization

ight as when, for example, a CEO takes a direct role in when making high offers and supervision when mak- yr evaluating a potential acquisition. ing low offers. including

cop Finally, just as more buyers per seller can increase

, This paper shows that using an incentives for strategic resource acquisition has an important the cost of incentivization, strategic factor markets in theoretical implication, namely, that as it becomes less which valuable resources are rare—as might occur holds ebsite

w likely that a seller will accept a particular investor’s in an economic recession—will be associated with a offer, the compensation of the investor’s agent will higher relative cost of incentivization. This suggests rise. This happens because an important part of the a clear pattern of organizational change across the other business cycle and in markets for different kinds of y agent’s compensation is a bonus received for acquir- INFORMS

an ing a valuable resource. The more infrequently the strategic resources. seller accepts a particular investor’s offer, the more on this bonus must rise to compensate the investor’s 2. A Model of a Strategic yright: agent for receiving the bonus less often. Moreover, Factor Market Cop posted 3 The formal literature on resource acquisi- 2.1. Setup tion has considered agency behavior in the study of managerial Consider a market for strategic resources composed ability (Makadok 2003). of N investors and P sellers. Each investor has the Ross: On Evaluation Costs in Strategic Factor Markets 4 Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS

capacity to evaluate and acquire a single resource,4 progressively. We need the following notation: Let and each seller has a single resource to sell. A pri- N 0 ≤ N denote the number of investors that partici- be ori, an investor knows (i) that proportion ƒ < 1 of the

.org. pate in the market with positive probability denoted 0 0 not resources are “valuable” and will generate a return by  > 0, let N represent the smallest possible N ms y or of R for the investor that acquires it, (ii) that propor- where investors’ expected profits are 0, given the ma tion 1 − ƒ of the resources are not valuable and will values of the other parameters in the model, and generate a return of 0 for the investor that acquires it, let q4‹5 represent the probability that a given offer file (iii) that R may vary among valuable resources on an ‹ is accepted by a seller with a valuable resource. interval [R1 R¯], where R¯ ≥ R, and (iv) that sellers have 0

The As described below, N , , and q4‹5 are determined ∗ 5

. 0 missions@inf a reservation price of I . An investor may evaluate in equilibrium, and N is a function of the model’s a resource before acquiring it at cost c and thereby 0

per parameters. Note that, as we will see below, N may ibers learn both whether the resource is valuable and, if so, 0

to be greater than, equal to, or less than N in equi- the value of R. librium. We do not index 1 ‹1 or q4‹5, because, in An investor may elect not to participate in the subscr equilibrium, the values of these endogenous param-

policy market. This strategy yields a profit of 0. Otherwise, to eters are common across investors that participate in the investor randomly selects a single seller, chooses le the market. this whether to evaluate the seller’s resource at cost c, and then chooses whether to make the seller a take-it-or- We begin by considering a strategic factor market ailab v with only one investor and one seller (N = 1 and a leave-it offer ‹. A seller may thus be selected by as ¯ many as N investors and as few as 0. Investors can- P = 1), and the assumption that R = R = R takes a

regarding not observe how many other investors have selected single value. In this case, the sole investor evaluates made the same seller or coordinate with other investors.6 the sole seller’s resource. If the resource is valuable,

is ∗ If an investor’s offer ‹ exceeds both a seller’s reser- the investor offers the seller ‹ = I , which the seller vation price of I ∗ and the best alternative offer from accepts, because the seller can do no better by reject-

questions ing the offer. Thus, all the rents generated by a valu- which another investor (with ties resolved randomly), the y seller accepts the offer, and the investor acquires able resource are captured by the investor. Formally, an the resource. The analysis focuses on the region of the we have N 0 = N = 11  = 11 and q4I ∗5 = 1.

ersion, parameter space where it is not optimal to make offers Next, consider a strategic factor market with two v send for a resource without evaluation and where it is opti- investors and one seller (N = 2 and P = 1). Suppose mal to make offers for valuable resources, implying that both investors always participate in the mar- ance (a) E[R]ƒ − I ∗ < 0 and (b) I ∗ < R.7 ket by evaluating the seller’s resource at cost c (for- Please Adv mally,  = 1), and that the seller’s resource is valuable. . 2.2. Competitive Equilibrium in The two investors now find themselves in “cutthroat”

site We describe how the market equilibrium works by competition (akin to Bertrand). If either investor offers s starting with a simple case and adding more detail

ticles a ‹ lower than the highest ‹ associated with non-

Ar negative marginal profit (i.e., the ‹ that solves R − 4 An implication of this assumption is that even if making offers author’ = for multiple resources were possible, it would be a loss-making ‹ 0), the other investor can acquire the resource this by offering slightly more. Therefore, the only equilib-

the strategy, because the investor would lose money, in expectation, to on each resource for which the investor made an offer without rium in this hypothetical subgame is for one investor evaluation. to acquire the resource at an offer of ‹ = R. Yet, at ight 5 I ∗ represents the proprietary value of the resource to the seller yr this offer, an investor is losing money on an overall if the seller retains it. (By contrast, the resource is not valuable to including basis once the cost of evaluation c is considered. (If cop

, other sellers and is worth either R or 0 to the investor, depending on whether the resource is valuable.) For example, if the seller owns R − ‹ = 0, then R − ‹ − c < 0.) We conclude that it can- land near a highway exit (Makadok 2001, p. 391), then I ∗ might not be an equilibrium for both investors to participate holds ebsite represent the present value of the seller’s future farming income in the market with certainty. If, conversely, the first w from the land. investor always participates in the market and the 6 The model can accommodate the case where investors observe second investor never does, the first investor would other how many other investors have selected the same seller by offer ‹ = I ∗; but, then, the second investor could make y setting P = 1. INFORMS = ∗ an positive profit by offering slightly more than ‹ I , 7 We also make the standard assumption that sales are final in the sense that the investor cannot immediately resell the resource effectively restarting the “cutthroat” competition we on to the seller or another party after acquiring it, which would be have just described. We conclude that it cannot be an yright: quite rare in the strategic factor markets discussed above. For equilibrium for either investor to blockade the other instance, an acquisition of a company cannot be easily reversed, from participating in the market. Rather, the only Cop posted and it may be impossible to convince a borrower to repay a loan early. Any attempted resale would also be subject to problems of possible equilibrium is a symmetric one where each adverse selection, forcing would-be buyers to undertake their own investor randomizes over participation: with prob- evaluation. ability  < 1, an investor participates by evaluating Ross: On Evaluation Costs in Strategic Factor Markets Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS 5

the seller’s resource, and with probability 1 − , an is lower when three investors participate in the mar- investor does not participate. ket than when two investors participate. In other be words, sellers’ expected profits are lower if N 0 = 3

.org. We now turn to the offer ‹ an investor makes for 0 not a valuable resource in this equilibrium. As already than if N = 2. ms y or demonstrated, no investor will follow a pure strategy This pattern continues as the number of investors ∗ ma of always offering a particular ‹ ≥ I . Instead, each grows. For any N ≥ 3, there is a different possible investor makes random offers according to a continu- equilibrium for each value of N 0 in 821 0 0 0 1 N − 11 N 9. file ous distribution function F 4‹5 defined on an interval Regardless of the equilibrium, each investor makes an ∗ ¯ expected profit of 0. The profit of each seller is declin- The 6I 1 ‹7. An investor only acquires the resource with an 0 . ∗ missions@inf offer of ‹ = I if the seller receives no other offer, so ing in N , the number of investors that participate in ∗ ¯ the market with positive probability. Thus, as N con- per F 4‹ ≤ I 5 = 0, and any offer at or above ‹ is always ibers ¯ tinues to rise, the profit of each seller at best remains

to accepted by the seller, so F 4‹ ≥ ‹5 = 1. Over the range 6I ∗1 ‹7¯ , F 4‹5 increases (and thus the probability the constant and in most equilibria declines. subscr seller will accept the offer increases) at exactly the rate Let us now discuss the fully general model. The

policy ¯ to needed to compensate the investor for the cost of pay- offer ‹ is clearly an increasing function of R, so where le ing the seller more. Regardless of the value of ‹, an R can take on more than one value, investors tend this investor’s expected profit is 0, just as if an investor to make higher offers for resources associated with a ailab higher R, as intuition would suggest. Where there is v does not participate in the market. Formally, we have a N 0 = N = 2, N 0 = 21  < 11 and q4‹5 = 1 − 41 − F 4‹55 more than one seller (P > 1), investors have greater 0 ∗ ¯ ¯ ∗ market power relative to sellers; therefore, N may be regarding if ‹ ∈ 6I 1 ‹7; q4‹5 = 1 if ‹ > ‹; and q4‹5 = 0 if ‹ < I . =

made greater than 2 and is increasing in P. In general, q4‹5 Add another investor so there are three investors 0 − − N −1 ∈ ∗ ¯ is and one seller (N = 3 and P = 1). In one possible 64P 41 F 4‹555/P7 if ‹ 6I 1 ‹7. An investor’s expected profit is maximized at N 0 = 1 and then equilibrium, two investors actively participate in the 0 0 declines steadily in N until the threshold level of com- questions market (N = 2) as above. In that case,  and F 4‹5 which 0

y petition of N is reached; further competition does not are identical to the model for two investors we have

an affect investors. A seller, by contrast, does best when just analyzed, whereas the third investor is effectively the number of investors is precisely large enough to ersion, blockaded. The reason is that  and F 4‹5 are at a level v reduce investors’ expected profits to 0, but no larger. send where two investors can participate in the market and Because this precise number could fall between integer make expected profits of 0. If the third investor were ance values, which is not possible, the actual integer num- to participate as well, the excess competition would 0

Please ber of investors N that maximizes sellers’ expected Adv drive all three investors’ expected profits below 0. Yet, . profits is either N 0 or N 0 − 1. At N 0 − 1, of course, in this is not the only possible equilibrium. In fact, there

site investors are still making positive profits. is an equilibrium where all three investors do partici- s 0 The following two propositions, which are proven

ticles pate in the market, i.e., where N = 3. The form of this 0 in the appendix, formalize the preceding discussion. Ar equilibrium is the same as when N = 2, but  and

author’ 0 F 4‹5 change in an interesting way. Intuitively, as N Proposition 1. Let i be the probability that investor i

this R rises, we should expect two changes: (a)  should fall evaluates at cost c, and let Fi 4‹5 be the function, conditional the to so that investors’ expected profits remain at 0, and on R, that determines the offer ‹ that investor i makes for R R (b) the potential for miscoordination among compet- a valuable resource. In equilibrium, Fi 4‹5 = F 4‹5 is the ight same for all i and is a continuous probability distribution yr ing investors should rise. The rising potential for mis- ∗ ¯ R ¯ R including coordination means that even sellers with valuable defined on an interval 6I 1 ‹ 7, where ‹ is increasing in R. cop

, N −1 ∗ resources may receive no offers. To see this, note that (i) If ƒ41 − 64P − 15/P7 56E6R7 − I 7 − c ≥ 0, the equi- 0 if N 0 = 1, a seller with a valuable resource receives librium is unique and symmetric, i = 1 ∀ i, N = N , and holds ebsite an offer with probability 1, but that if N 0 = 2, a seller investors make positive profits in expectation. (ii) If ƒ41 − w N −1 ∗ 0 0 0 with a valuable resource only receives an offer with 64P − 15/P7 56E6R7 − I 7 − c < 0, each N ∈ 8N 1 N + probability 1 − 41 − 52 < 1. This pattern continues as 11 0 0 0 1 N 9 is associated with a unique equilibrium where

other 0 N 0 rises above 2. Moreover, because a seller is more N is the smallest integer x ≤ N such that ƒ41 − 64P − y x−1 ∗

INFORMS 15/P7 56E6R7 − I 7 − c < 0, investors make zero profits in

an likely to receive no offers, the implied ex post market 0 power of an investor that does evaluate the seller’s expectation, N investors participate in the market with the on ∈ resource is higher. As a result, investors that do par- same probability  401 15, and the remaining investors do

yright: not participate in the market. ticipate in the market shade their offers lower, that is, the mean of F 4‹5 shifts downward within the range Proposition 2. The expected profit of each seller is Cop posted 6I ∗1 ‹7¯ . Because of these two reinforcing effects, we monotonically increasing in N 0 for all N 0 < N 0 − 1, mono- have the counterintuitive result that the expected best tonically decreasing in N 0 for all N 0 > N 0, and maximized offer ‹ received by a seller with a valuable resource at either N 0 − 1 or N 0. The threshold N 0 is increasing in P. Ross: On Evaluation Costs in Strategic Factor Markets 6 Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS

These results suggest an important role for strategy for an acquisition is usually given authority “to nego- in shaping the structure of strategic factor markets. tiate within a certain range of price and terms and be From an investor’s perspective, the less intense the conditions” (Haspeslagh and Jemison 1991, p. 95), just .org.

not competition with other investors, the better. From a as the head of the petrochemical division at SK Chem- ms y

or seller’s perspective, although it is important that com- ical needed final budgetary approval from the chair-

ma petition be intense enough to force investors to make man to purchase assets in Poland. Similarly, bankers offers significantly above the seller’s reservation price must conform to their bank’s policies on interest file of I ∗, it is likewise important that competition among rates for different classes of borrowers (Comptroller of investors not become excessive. The model suggests

The the Currency 1998). The model’s contracting environ- . missions@inf two channels through which excessive competition ment reflects these requirements, giving the principal among investors could be reduced. One channel is per authority to determine the offer ‹ (if any) the agent ibers to reduce the number of investors that could feasibly should make for a resource after receiving the agent’s to compete for a resource; this corresponds to a reduc- evaluation.

subscr tion in N . The other channel is to reduce the scope for Because the agent is risk averse, the agent’s reser- policy to miscoordination so that only a subset of investors par- vation utility can be met most affordably by paying

le ticipate in the market; this corresponds to a reduction the agent a flat wage that compensates the agent for this in N 0 below N , such that the equilibrium that arises the costly effort e. However, such a wage would not ailab is more favorable to sellers than other possible equi- v be incentive compatible; the agent could shirk, for a libria that could arise whenever N > N 0. We discuss example, by not bothering to evaluate a resource and the practical implications of these ideas below. claiming either that it was not valuable or that the regarding

made seller had rejected the offer made by the agent on 2.3. Organizational Design: Incentivization vs. is the principal’s behalf. So, the optimal incentive con- Supervision tract takes the form of a high wage (wH ) if a valuable The previous section assumed that an investor could

questions resource is acquired and a low wage (wL) if a valu- which evaluate a resource for a generic cost c without speci- y able resource is not acquired. These two wage pay-

an fying who performed that evaluation. In this section, ments can be decomposed into a salary of wL and we assume instead that each investor has a principal an incentive bonus equal to the difference between ersion, v (e.g., an owner or CEO) who maximizes the investor’s send the high and low wages, wH − wL. The agent receives profits. We also assume that, to evaluate the resource this bonus if—and only if—the resource the agent ance and make the offer ‹ to the seller, the principal has evaluates is acquired, which happens with probabil- two organizational design options. The first option Please ity ƒ (the resource is valuable) multiplied by q4‹5 (the Adv . is supervision, wherein the principal either evaluates seller accepts the offer ‹ for the resource). It is for in

site a resource directly or uses a monitoring technology this reason that ƒ and q4‹5 affect the relative cost of s to ensure that an agent (e.g., a lower-level manager) incentivization in a systematic way. As either ƒ or ticles properly evaluates a resource. We continue to stipu- q4‹5 declines, the bonus must rise to meet the agent’s Ar late that the total cost of evaluation under supervision

author’ reservation utility, that is, the agent must receive a is c, although it should be understood that if supervi- this larger bonus to compensate the agent for receiving

the sion represents the principal’s own evaluation efforts, to the bonus less often. Whereas a risk-neutral party c represents the opportunity cost of the principal’s would be indifferent between receiving a larger bonus ight time, and if supervision represents monitoring, c rep- less often and a smaller bonus more often—provided yr resents the sum of the agent’s wages and the cost of the total expected wages are the same—a risk-averse including

cop the monitoring technology. The second organizational , agent will require a larger and larger risk premium as design option is incentivization, wherein the principal the probability of receiving the bonus declines. This uses an incentive-laden wage contract to induce an holds ebsite suggests two implications: (i) incentivization is expen- w agent to evaluate a resource on the principal’s behalf. sive relative to supervision when the proportion ƒ Whereas the principal is risk neutral, the agent of valuable resources is low, as might be the case is risk averse, maximizing a function of the form other in economic recession, and (ii) in a competitive con-

y − u4w5 e, where u4w5 is a concave and increasing text, an investor using incentivization (but not super- INFORMS an function of wages w, and e is the cost to the agent of vision) can economize on wage costs by shading its gathering information about the resource. The agent on ∗ offer ‹ higher, thereby raising the probability that the also has a reservation utility u > 0 and will only q4‹5 yright: seller accepts the offer, . The following proposi- agree to a wage contract that provides that utility. In tion, which is proven in the appendix, formalizes the practice, lower-level managers who handle the pro- Cop posted preceding discussion. curement of strategic resources are frequently sub- ject to some oversight, especially regarding price. For Proposition 3. Let u−14u5 be the inverse of u4w5. example, a business unit manager with responsibility In the incentive contract, the agent receives a wage of Ross: On Evaluation Costs in Strategic Factor Markets Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS 7

−1 ∗ R R wH = u 4u + e/4ƒq4‹555 with probability ƒq4‹5 and a offer according to the function FI 4‹5, where (a) F 4‹5 = −1 ∗ R R wage of wL = u 4u 5 with probability 1−ƒq4‹5. Expected ‚FS 4‹5+41−‚5FI 4‹5, and (b) if ‚ ∈ 401 15, the support of be wages are declining in ƒ and q4‹5. F R4‹5 is 6I ∗ = ‹R1 ‹¯ R7, and the support of F R4‹5 is 6‹¯ R = .org. S S S I S R ¯ R = ¯ R not ‹ 1 ‹ ‹ 7. ms We now use these results to enrich our compet- I I y or itive equilibrium with the choice of organizational Proposition 5. The incidence of supervision (‚) is ma design. Let ‚ denote the equilibrium probability that weakly increasing in N 0 and strictly increasing in the range

file an investor uses supervision. Thus, if investors fol- where ‚ ∈ 401 15. low a pure strategy of using supervision, ‚ = 1,

The and if investors follow a pure strategy of using

. 3. Discussion missions@inf 8 incentivization, ‚ = 0. Clearly, the relative costs of This paper studies strategic factor markets where per ibers incentivization and supervision may be such that one evaluation costs are important. The main results are to organizational design dominates the other, no matter as follows: (1) Competition for resources—as mea- the level of competition among investors. In that case, sured by the number of competing investors and subscr ‚ is always either 0 or 1. However, we have seen that

policy resource scarcity—has a nonmonotonic effect on sell- to the relative cost of incentivization is increasing in q4‹5 ers’ profits. At low levels of competition, more com- le

this and thus in ‹. Suppose, then, that in the absence of petition encourages investors to make higher offers competition, incentivization has a modest cost advan- to avoid being outbid. At high levels of competition, ailab v tage over supervision. For any given level of project a more competition increases miscoordination among 0 scarcity (P), if there is only investor (N = 1), then N investors, so the expected best offer a seller receives

regarding (the number of investors that participate in the mar- declines. This is because the probability that a seller made ket with positive probability) also equals 1, and this receives no offer rises, creating greater ex post market is sole investor will use incentivization. Now, add pro- power for those investors that do evaluate the seller’s gressively more investors to the market, increasing N resource, who can shade their bids lower and acquire 0 questions which and also N . As the intensity of competition among valuable resources for less. (2) As a seller gains mar- y investors increases, the probability of winning with ket power and becomes more likely to reject investors’ an an offer at or near I ∗ drops sufficiently that supervi- offers, the cost of paying incentives to an investor’s

ersion, sion is less costly than incentivization at offers at or agent will increase. Thus, to increase the rate at which v ∗ send near I ; investors then begin to randomize between their bids are accepted, investors using incentiviza- organizational designs, using supervision where the tion will tend to offer sellers more than investors ance probability the seller will accept the offer is low (i.e., using supervision. (3) For the same reason, increasing Please

Adv if ‹ is low) and incentivization where the probabil- the competition for resources increases the probability .

in ity the seller will accept the offer is high (i.e., if ‹ that an investor will be outbid, lowering the propor- site is high). Thus, ‚ rises from 0 to some number less tion of investors using incentivization and raising the s than 1; ‚ continues to rise as further increases in N proportion using supervision. (4) Incentivization is ticles 0

Ar allow for further increases in N . relatively cost efficient compared to supervision when 0 author’ Eventually, N reaches N . From this point forward, the proportion of valuable resources is high, as would 0 0 0 this there is a different equilibrium for each N ∈ 8N 1 N + be expected in good economic times rather than in

the 0 to 11 0 0 0 1 N 9. Further increases in N do not change q4‹5 recession. for any ‹, because q4‹5 is at precisely the level that ight keeps investors’ expected profits at 0 for every offer 3.1. Managerial Implications yr An implication of these results is that managers need

including ‹. However, as demonstrated in the discussion pre- cop 0 , ceding Propositions 1 and 2, we know that as N to consider carefully how strategic factor markets are rises beyond N 0, investors shade their bids lower, and structured. In a strategic factor market where eval- uation costs are important, sellers and investors are holds ebsite lower offers are, in turn, associated with supervision. w Thus, ‚ continues to rise with N 0. The following two engaged in an elaborate dance to structure the mar- propositions, which are proven in the appendix, for- ket in precisely the right way from their individual perspectives—where competition among investors is other malize the preceding discussion.

y initially low, the interests of investors and sellers are INFORMS

an Proposition 4. In the equilibrium of the market, each in conflict, and where competition among investors investor that participates in the market with positive prob- is initially high, there is no conflict. Indeed, sellers on ability uses supervision with probability ‚ ∈ 601 17 and may do better when investors make positive prof-

yright: R makes an offer according to the function FS 4‹5, and, its than when investors do not. Investors also have with probability 1 − ‚, uses incentivization and makes an Cop posted to consider how competition and their own bidding policies affect a manager’s incentive to engage in the 8 In equilibrium, investors always follow the same strategy; thus, entrepreneurial service of strategic resource evalua- ‚ is the same for every investor. tion. A practice of making lowball offers for strategic Ross: On Evaluation Costs in Strategic Factor Markets 8 Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS

resources may drive up the cost of incentivizing man- developers when the CEO takes a direct role in eval- agers to evaluate resources. uation, as often occurs (Bradach 1998). be Managers of resource sellers may employ differ- As noted, one can think of lending in commer- .org.

not ent strategies to influence the structure of a strategic cial banking as, in part, customer acquisition. In that ms y

or factor market and thereby mitigate the coordination context, supervision would be the use of central-

ma problems associated with excess competition. One ized lending procedures like quantitative credit scor- strategy is to develop long-term relationships with a ing, and incentivization would be relationship-driven file subset of resource investors and thereby limit com- lending.9 This paper would predict a greater inci- petition to that subset. Another strategy for reduc- dence of credit scoring in competitive banking mar- The . missions@inf ing miscoordination would be to rotate transactions kets than in markets with only a few banks. It also among investors in a predictable order. Many large follows that informationally opaque segments of the per ibers corporations employ this strategy in raising capi- loan market where quantitative credit scoring is not to tal by borrowing by turns from among a core set feasible (like highly leveraged or small business loans)

subscr of relationship banks. Sellers may also benefit from should have relatively few competing banks and be

policy regional, whereas markets amenable to quantitative to investments in disclosure and transparency that lower techniques (like U.S. home mortgages or credit cards) le evaluation costs for investors. For resource investors, this the goal is more straightforward: limit competition. should have a large number of geographically dis- ailab The literature has identified strategies investors may persed competitors. This is what we observe. In terms v a employ to this end, including developing heteroge- of rates of return, investors earn 4R − ‹5/‹, which neous resource complementarity (Adegbesan 2009) is monotonically decreasing in ‹. Thus, expressed in regarding terms of interest rates in the banking market, the made and acquiring an informational advantage about valu- inverse U-shaped relationship between the expected is able strategic factors (Makadok and Barney 2001). An important contribution of this work is to demonstrate best offer received by a seller and the number of formally that sellers may be investors’ allies in this competing investors would translate to a U-shaped questions which relationship between the expected lowest interest rate

y endeavor if competition for resources crosses a thresh- offered to a borrower and the number of banks com- an old level of intensity. peting in a given market. This may explain why a ersion, v 3.2. Empirical Implications small firm’s cost of borrowing is positively associated send This paper’s results can be used to derive testable with the number of lenders the firm uses (Petersen and Rajan 1994). ance predictions in a number of areas of interest to man- agement scholars. For instance, M&A transactions can There is a wealth of evidence that, in recession, Please Adv . be individually negotiated with a few prospective bank lending flows away (in relative terms) from in informationally opaque borrowers, such as from small

site acquirers or be structured as “sell-side auctions” with

s many potential acquirers undertaking due diligence to large firms (Gertler and Gilchrist 1993) and from ticles and making competing offers. This paper would pre- riskier to safer credit risks (Lang and Nakamura Ar dict that acquirers in auction-like M&A transactions 1995). None of the existing theoretical explanations author’ would be more likely to assign evaluation responsi- for the procyclicality of bank lending provide a this

the bility to senior managers in the corporate office than straightforward explanation for the disproportionate to to business unit managers. Moreover, a prospective impact on informationally opaque borrowers. Herein, the empirical regularity follows naturally from this ight acquirer is likely to make a lower offer if the acquisi- yr tion effort is led by the corporate office. Similar pat- paper’s result that incentivizing bankers to make

including good lending decisions is not cost effective when the cop terns should be observable in bidding for strategic , resources like allocations of the telecom spectrum or proportion of creditworthy borrowers drops in reces- new technologies. sion, and from the fact that informationally opaque holds ebsite borrowers are by definition ill suited to quantitative w In general, firms are more likely to delegate author- ity to acquire strategic assets in a less competitive credit evaluation. environment, especially where evaluation costs are other 3.3. Extensions

y important. Thus, the relative lack of competition SK

INFORMS There are a number of ways that the model might

an Chemical faced in acquiring a site for its chemi- be altered or extended in future research. This cal operations in Poland may have been the rea- on paper posits homogenous investors, but in practice, son decision-making autonomy for the expansion was yright: delegated to the head of the petrochemical division. 9 Likewise, some retailers, such as fast food and hotel Credit-scoring methods rely on “hard,” verifiable information like Cop posted financial statements, as opposed to “soft” information like the chains, are more likely to delegate authority for eval- character of a borrower’s managers, which is more relevant for uating and acquiring sites in less competitive mar- relationship lending. See Allen et al. (2004) for a summary of com- kets, and are likely to make lower offers to real estate monly used quantitative metrics in credit-scoring models. Ross: On Evaluation Costs in Strategic Factor Markets Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS 9

evaluation may reveal that a resource is valuable for theoretical mechanisms that might alter this paper’s one investor but not for another. Such heterogeneity predictions.10 The appendix discusses an imperfect be in resource preferences does not qualitatively change monitoring technology and continuous effort, and .org. 0

not this paper’s results, but it does raise N by relaxing shows that whatever other mechanisms might arise in ms y

or postevaluation competition among investors. models with these features, the mechanism that gives

ma Existing resources and capabilities may effectively rise to the trade-off between supervision and incen- precommit some investors to supervision and oth- tivization in this model would remain. In particular, file ers to incentivization. In one firm, it may be nec- (a) anything that lowers q4‹5 (e.g., more competition essary to delegate evaluation to a manager with among investors) would make the agent’s nonshirk- The .

missions@inf specialized expertise, such as the head of a foreign ing constraint bind more tightly, and (b) the greater “center of excellence” who has worldwide responsi- the intensity of monitoring, the more the agent’s non- per ibers bility for leveraging a particular capability across the shirking constraint is relaxed. to firm (Frost et al. 2002) or a research and develop- The model is designed to apply to situations where

subscr ment project manager with unique technical knowl- the principal retains some control over pricing and policy

to edge (Cassiman and Valentini 2009). In another firm, the agent’s wages, and therefore does not cover situa-

le only the CEO or another senior manager in head- tions in which the agent is effectively an independent this quarters may have the training required to evaluate a business (e.g., a master franchisee). In those cases, 11 ailab resource, or there may be a large cadre of expatriates even state-contingent transfers may be infeasible. In v a at a foreign location to serve as monitors of the for- the model, sellers are passive. In practice, a seller eign managers who conduct evaluations (O’Donnell may adapt its profile over time to the organizational regarding

made 2000). The appendix derives the equilibrium for a case design of investors or more actively seek investors

is in which one investor committed to supervision and when no investor makes an offer for the seller’s one investor committed to incentivization compete in resource. It may be possible at times for investors to a strategic factor market, and the cost of evaluation observe whether other investors make offers for the questions which

y in the absence of competition is the same for each same resource. Incorporating these information exter-

an investor. The main change is that rather than incen- nalities would substantially complicate the analysis tivization being used uniformly for high offers and and may alter the model’s results, although the basic ersion, v supervision being used uniformly for low offers, both agency problem with incentivization would remain. send investors make offers across the same range 6I ∗ = For all N > N 0, there is a different equilibrium for ¯ 0 0 0 ance ‹1 ‹7. The offer function for the investor using super- each N ∈ 8N 1 N +11 0 0 0 1 N 9. Although the multiplic- ∗ ity of equilibria highlights the role of strategy in shap- Please vision, however, has a mass point at I and is more Adv . weighted to low offers than the offer function of the ing market outcomes, this paper does not formally in ∗

site investor using incentivization, which is open at I . analyze different mechanisms for effecting these out-

s Thus, the equilibrium is qualitatively similar to that of comes. The possibilities are numerous and include ticles the base model. explicit collusion, tacit cooperation through repeated Ar Considering other scenarios, we know that in- interaction, and precommitment to participate or not author’ vestors may be heterogeneous in the sense that one participate in a given strategic factor market. Com- this

the investor will be known to have an advantage over paring and contrasting these strategies is a potentially to another investor in evaluation cost c or the return R. fertile area for future research.

ight If P is high enough that the disadvantaged investor For simplicity, we have posited a single-period, yr still makes positive profits in equilibrium, the equilib- single-round bidding game. In a dynamic setting, a including cop rium does not qualitatively change. If P is not high seller with a valuable resource that does not receive , enough for the disadvantaged investor to make pos- any offers could put the resource on the market again. itive profits (e.g., if P = 1), the advantaged investor This might mitigate the effects of miscoordination. holds ebsite

w will always make an offer for a valuable resource, However, it might also affect the incentive of sellers to whereas the disadvantaged investor will only make accept low offers, potentially discouraging investors an offer with probability  < 1. The offer function of other ∗ y the advantaged investor will have a mass point at I 10 Incorporating richer information structures in models of dele- INFORMS

an and be more weighted to low offers than the offer gated expertise is an unsolved problem. Like more conventional function of the disadvantaged investor, which will be principal–agent models (e.g., Evans 1980, Baron and Besanko 1984), on open at I ∗. The effect of competition on the relative Kim (2006) considers an imperfect monitoring technology but, yright: cost efficiency of the two organizational designs does unlike the model herein, does not allow for state-contingent wages. Szalay (2009) has a continuum of effort but does not incorporate not qualitatively change. Cop posted agent risk aversion (or limited liability). The model posits a simple information structure. 11 Aghion and Tirole (1997) and Alonso and Matouschek (2008) con- A richer information structure in effort, resource type, sider delegated expertise problems where state-contingent transfers or monitoring technology could give rise to new are limited. Ross: On Evaluation Costs in Strategic Factor Markets 10 Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS

from evaluating the seller’s resource. The model also fully ascertain whether a firm can acquire resources did not consider different kinds of sales mechanisms, at a cost low enough for profitable use until we can be some of which could potentially moderate the effects account for how that firm organized to acquire its .org.

not of competition and miscoordination. These include resources initially. ms y

or multiple rounds of bidding, ascending or open-bid Our model and analyses contribute insights to

ma auctions, committing to high reserve prices, or set- a number of other literatures as well. For exam- ting participation fees to discourage overparticipa- ple, in recent years, strategy researchers have started file tion, and, conversely, subsidizing evaluation costs. using cooperative to study the theoret- Which of these strategies is most appropriate would ical foundations of value creation and capture (e.g., The .

missions@inf depend on specific contextual factors. Brandenburger and Stuart 1996, 2007; MacDonald and The link between competition and agency devel- Ryall 2004). A standard assumption in this literature per ibers oped in this paper arises because agency affects is that market participants understand the value they to the costs of actively participating in a strategic fac- can create by transacting with other market partici- tor market. Other factors that affect agency costs subscr pants. This paper, by contrast, has studied strategic

policy could also moderate the effects of competition. Exam-

to factor markets, where firms frequently have to incur ples include social factors like norms, ethical stan-

le significant evaluation costs to ascertain the value of this dards, and altruism (Casadesus-Masanell 2004), as potential transactions. This paper shows that these ailab well as the behavioral biases of managers with evaluation costs may significantly affect competitive v a entrepreneurial inclinations (Dushnitsky 2010). Eval- interaction and organizational design. It follows that uating some resources may also require the input of a potentially promising line of inquiry would be to regarding many different managers, creating a more complex

made use a cooperative game theoretic setting to explore agency problem (Mosakowski 1998).

is how evaluation costs affect the conditions that deter- mine how and to what extent firms can appropriate 3.4. Contributions to the Literature and value. Likewise, models in cooperative game theory questions Future Research Directions which

y generally assume that bargaining is unrestricted in The resource-based view has wrestled with the ques- an tion of how a firm can be profitable without “luck” the sense that market participants that can profitably transact with each other will do so. It is for this rea-

ersion, (e.g., Barney 1986). As the argument goes, competi- v son that rents tend to accrue to the side of the market send tion among resource investors should drive up the cost of resources to the value that can be derived with relative scarcity. We find, however, that evalua- ance from them, leaving all residual profit in the hands of tion costs interact with relative scarcity to create the

Please possibility of miscoordination such that potentially

Adv resource sellers. Our analysis suggests this view may . profitable transactions may not occur. In the context

in be too pessimistic. We arrive at this conclusion by

site accounting for investors’ need to evaluate resources of cooperative game theory, this miscoordination is a s before acquiring them. kind of bargaining failure. Another promising line of ticles The cost of evaluating resources introduces enough inquiry, then, would be to explore the implications of Ar bargaining failure for value creation and capture. The

author’ friction to the market that when competition exceeds model of market frictions proposed by Chatain and this a certain threshold, both resource sellers and resource

the Zemsky (2011) is a promising step in this direction. to investors may have incentives to restructure the market. By structuring a strategic factor market to By demonstrating that relative scarcity no longer ight dampen competition, the parties can assure that at benefits resource sellers above a certain threshold, this yr least some rent accrues to both sides of the market— paper establishes a new and important boundary on including cop the benefits of competition among trading partners. , that is, to resource sellers and investors. At the least, our analysis shows that resource sellers should This insight builds on the work of other management not allow competition among resource investors to scholars, such as Denning (1986), who argued that holds ebsite

w become “excessive.” It is perhaps ironic that firms focusing on a small set of suppliers may allow for bet- engaged in actively shaping a strategic factor mar- ter quality control and improve joint research efforts. More recently, Aghion et al. (2005) found that exces-

other ket, as suggested by industry analysis (Porter 1979),

y may be more likely to acquire the resources needed to sive competition may discourage innovation. A qual- INFORMS

an succeed in downstream product markets. In this way, itative difference between those papers and this one industry analysis and the resource-based view would is that the potentially baleful effects of competition on complement rather than contradict one another. in our model ultimately arise from miscoordination yright: Our analysis also shows that the competitive behav- among competitors. Our paper builds on the strategic incentives litera- Cop posted ior of resource investors is intimately linked to how they organize to acquire resources in the presence ture, which studies the implications of organizational of managerial agency, as well as to the characteris- design for competitive behavior. In that literature, tics of strategic factor markets. After all, we cannot organizational design is used as a commitment Ross: On Evaluation Costs in Strategic Factor Markets Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS 11

¯ ¯ device, which, depending on the nature of the strate- ‹ and Fi to economize on notation. Define ‹ as the highest gic interaction, may either increase or decrease the offer that, if accepted with probability 1, would result in an be intensity of competition (Vickers 1985, Fershtman investor earning the same profit the investor would earn .org. with a bid of I ∗ if the latter offer is accepted only if it is

not and Judd 1987, Sklivas 1987, Bonanno and Vickers ms the only one made for that resource. Given the symmetry y or 1988). This paper contributes a new dimension to that Q of the investors and the fact that j6=i64P − j 41 − Fj 4‹555/P7 ma argument by showing that organizational design can be used to maximize the cost efficiency of evalua- must be the same for each investor i, i is some common

file  ∈ 401 17, and F = F for all investors that evaluate with tion, such that incentivization is most cost efficient i positive probability. If there were a mass point in F at any when a buyer makes a high offer for a resource, ¯ ¯ The ‹ < ‹ (at ‹), it would be profitable to transfer the mass to

. ∗ 0 00 missions@inf and supervision is most cost efficient when a buyer ‹ + (to some ‹ > I ). If there were an interval 6‹ 1 ‹ < ‹7¯ 12 makes a low offer. The strategic incentives litera- 0

per where F = 0, then given that there are no mass points in F , ibers ture also finds that organizational heterogeneity may there must be an > 0 such that an investor could profitably to arise from nonconvexities in the production function transfer the probability measure of the interval 4‹001 ‹00 + 5 (Hermalin 1994) or from the joint determination to ‹0, a contradiction. The top of the support of F must subscr ¯ policy of organizational design and managerial contracts be ‹, because otherwise an investor could profitably devi- to (Vroom 2006). This paper demonstrates that a pri- ate by making an offer just above the support of F , and le ∗ this ori homogenous firms may follow ex ante identical the bottom of the support of F must be I , because a seller accepts an offer at the bottom of the support of F if the ailab strategies of randomizing between using incentiviza- v seller receives no other offer. If N < N 0, then an investor a tion for high offers and supervision for low offers, makes positive profit even if all other investors evaluate; so, giving rise to organizational heterogeneity ex post. 0

regarding i = 1 ∀ i and N = N is the unique equilibrium. Otherwise, Taken together, we believe our results encourage 0 0 0 made each N ∈ 8N 1 N + 11 0 0 0 1 N 9 is associated with a unique further research into firm competition in strategic fac- 0 is equilibrium, all investors make zero profits, N investors tor markets. For too long, the strategy field has failed evaluate with the same probability  < 1, and the remain- to recognize the important role that evaluation costs ing investors never evaluate. Now let R possibly vary. We questions which play in the nature of strategic factor markets and will show that ‹¯ R and the mean of F R are increasing in R y how the need to evaluate resources before acquir- and correspond precisely to their values with R fixed. Take an ing them affects organizational design. The subtle any other candidate equilibrium with alternative offer func- 0 0 ersion, intersection of resource evaluation with managerial tions F R and support 6I ∗1 ‹¯ R 7 for some R; then, an investor v 0 send agency, in particular, merits more exploration. How could profitably deviate by making an offer just above ‹¯ R 0 should firms organize to perform entrepreneurial ser- for those values of R for which ‹¯ R < ‹¯ R. If, alternatively,

ance 0 vices (Penrose 1959)? How do organizational deci- ‹¯ R > ‹¯ R1 ∀ R, then investors would earn lower profits with 0 ∗ Please ¯ R Adv an offer of ‹ than with an offer of I , a contradiction. ƒ

. sions influence (and become influenced by) the nature in of a given strategic factor market? We may find that—

site Proof of Proposition 2. Take any given R, and let in the spirit of the resource-based view—it is how s us again drop it as a superscript. Suppose  = 1 for all ticles firms organize to compete in strategic factor markets investors that evaluate with positive probability; then, the 0 Ar that determines how profitable they are in down- probability a seller receives an offer is 1 − 64P − 15/P7N ,

author’ 0 stream product markets. which is increasing in N 0. Let F N be the equilibrium this offer function F when there are N 0 investors. Because the the to Acknowledgments expected profit of each investor is declining in N 0, the top 0 0 0 0 This paper is based on a chapter of the author’s dissertation of the support of F N is above that of F N −1, and F N −1 > F N ight 0 at New York University. He thanks his thesis committee: within the support of F N −1. Suppose  < 1 for all investors yr Heski Bar-Isaac, Kose John, Alexander Ljungqvist, Anthony

including that evaluate with positive probability; then, the probability 0 cop N , Saunders, and Bernard Yeung. He also thanks the depart- a seller receives an offer is 1 − 64P − 5/P7 , which, given 0 ment editor (Bruno Cassiman), the associate editor, two the invariance of 64P − 5/P7N −1, is decreasing in N 0. We anonymous referees, Cristian L. Dezs˝o,Richard J. Makadok, now need the following lemma: holds ebsite and seminar participants at Washington University in St. w Louis and the 2008 Atlanta Competitive Advantage Con- Lemma 1. A function of the form 41 − ax5/41 − bx5, where ference for helpful comments. Errors and omissions are the 0 < b < a < 1, is increasing in x1 ∀ x > 0. other author’s sole responsibility. Proof. Taking logs yields ln41 − ax5 − ln41 − bx5. Differ- y

INFORMS entiating this with respect to x yields an Appendix. Proofs of Propositions x x on Proof of Proposition 1. The proof is by construction. a ln a b ln b − + 0 Take R to be fixed, and let us drop the superscript R on 1 − ax 1 − bx yright: Consider the derivative of the second term of the expression Cop posted 12 Thus, unlike Bonanno and Vickers (1988), delegating decision- with respect to b: making authority to an agent here may lead to more aggressive price competition, because increasing the probability that an offer xbx−1 ln b  bx  1 bx−1  x ln b  + = + 1 0 for a resource will be accepted reduces the agent’s wages. 41 − bx52 1 − bx b 1 − bx 1 − bx Ross: On Evaluation Costs in Strategic Factor Markets 12 Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS

Signing this is a matter of signing the term in parentheses. Expected wages are accordingly Applying L’Hôpital’s rule gives   be e ƒq4‹5u−1 u∗ + + 61 − ƒq4‹57u−14u∗50 .org. x ln b x/b lim = lim − = −10 ƒq4‹5 not x x−1 ms b→1 1 − b b→1 xb y

or Differentiating with respect to q4‹5 yields There is no solution to 44x ln b5/41 − bx55 + 1 = 0 for any ma b ∈ 401 15, because equality requires that x ln b = bx − 1. This    −1 ∗ e −1 ∗ = ƒ u u + − u 4u 5 file occurs at b 1, where the derivatives of the two func- ƒq4‹5 x−1 tions are x/b—b=1 = x and xb —b=1 = x. Otherwise, the first x −1    du ∗ e e The derivative is larger because 1 > b . Thus, x ln b is strictly less − u + < 01

. x x x missions@inf than b − 1 ∀ b ∈ 401 15. We conclude that —a ln a/41 − a 5— > dq4‹5 ƒq4‹5 ƒq4‹5 —bx ln b/41−bx5— and thus that −4ax ln a/41−ax55+bx ln b/41− per

ibers x where the negativity follows immediately from the con- b 5 > 0. ƒ −1 to vexity of u . Differentiating with respect to ƒ is precisely ∗ 1/4N 0−15 We have 4P −5/P = q4I 5 . Let G4‹5 = 1−F 4‹5, and analogous. subscr solve for G4‹5 in terms of q4‹5: Now consider the more general situation where q4‹5 policy to 0− ‹ 1 − q4‹51/4N 15 varies with the offer , which in turn varies in competitive

le = G4‹5 ∗ 1/4N 0−15 0 equilibrium according to an investor’s offer function F 4‹5. this 1 − q4I 5 If ‹ were exogenous, the principal might be able to offer the max ailab Let G 4‹5 be the probability that at least one offer for a agent an incentive-compatible contract at a lower overall v R a resource is greater than or equal to ‹ given that at least one cost, where q4‹5 above is replaced by q4‹5 dF 4‹5. However, offer is made: ‹ is set by the principal. If the agent were to agree to such regarding − − N 0 a contract, the principal could profitably deviate by shifting made 1 64P G4‹55/P7 max ∼ 0 G 4‹5 N 0 to a new offer function F 4‹5 that is more weighted toward is 1 − 64P − 5/P7 low ‹. Accordingly, the only contract that is incentive com- 0 0 1 − q4‹5N /4N −15 patible for the principal is one that specifies a different w = H ∗ 0 0− 0 questions − N /4N 15 for each ‹ and thus for each q4‹5, namely, the precise wage which 1 q4I 5

y payments described above. ƒ This has the form of the function in the lemma, so increas- an ing N 0 lowers Gmax4‹5 ∀ ‹. The shift in Gmax4‹5 ∀ N 0 > Proof of Proposition 4. Take any given R0 and let us 0 ersion, N decreases the expected value of the best offer received again drop it as a superscript. The difference in profitability v send by a seller conditional on an offer being made. Thus, the between incentivization and supervision, I −S , is increas- 0 expected best offer is increasing in N where  = 1 and ing in ‹, so the support of FI is above that of FS ∀ ‚ ∈ ance decreasing in N 0 where  < 1. It follows that the expected 401 15. It remains to show that ‚ is well defined and unique 0 0 0 0 Please best offer is highest at either N − 1 or N . That N for every N . If it is an equilibrium for every investor to Adv . is increasing in P follows because 4P − 15/P is increasing use incentivization, then in another candidate equilibrium in

site in P. ƒ where every investor uses supervision with positive prob-

s ability,  or ‹¯ would be lower, making it profitable to Proof of Proposition 3. Take q4‹5 as given. The princi- ticles deviate to incentivization and an offer of ‹¯ . Likewise, if pal can dissuade the agent from misrepresenting nonvalu- Ar it is an equilibrium for every investor to use supervision,

author’ able resources as valuable or misrepresenting R by setting then in another candidate equilibrium where every investor

this an arbitrarily low wage for resources that are acquired and uses incentivization with positive probability, a deviation to the prove to be nonvaluable or are valuable but have a different to ∗ R than claimed by the agent. Two constraints remain: (i) the supervision and a bid of I would be profitable. Where there is no equilibrium with ‚ equal to 0 or 1, I − S is negative

ight wage contract must meet the agent’s reservation utility, and at an offer of I ∗, positive at an offer of ‹¯ (as defined under yr (ii) the principal must dissuade the agent from not evaluat- 0 incentivization), and zero at q4‹5 = 64P − 41 − ‚55/P7N −1, including ing a resource and claiming that the resource is not valuable cop , or that the offer was rejected. These two constraints are IR defining ‚ ∈ 401 15. ƒ and IC 1, respectively: Proof of Proposition 5. Take any given R, and let us holds ebsite 0 0

w max I = ƒq4‹56R − wH − ‹7 − 41 − ƒq4‹55wL again drop it as a superscript. Suppose N ≥ N . We may w 1 w H L then reapply Lemma 1 to see that as N 0 increases (and thus ∗ 0 s.t. ƒq4‹5u4wH 5 + 41 − ƒq4‹55u4wL5 − e ≥ u 1 (IR) 1/4N − 15 decreases), G4‹5 decreases ∀ ‹. In then follows other from the fact that the support of FI (if it exists) is above y ƒq4‹5u4wH 5 + 41 − ƒq4‹55u4wL5 − e ≥ u4wL50 (IC 1) INFORMS that of FS (if it exists) that the measure of investors who an The IR constraint clearly binds, or the principal could lower use supervision (‚) either remains 0 (all investors use incen-

on both wH and wL. The concavity of the agent’s utility func- tivization) or 1 (all investors use supervision) or increases. tion implies that IC 1 binds as well. We then have the fol- Suppose, instead, that N 0 < N 0. If every investor is using yright: lowing solution: incentivization, increasing N 0 prompts investors to random- Cop posted  e  ize between organizational designs or over the decision to w = u−1 u∗ + 1 evaluate. In the latter situation, we return to the case where H ƒq4‹5 N 0 ≥ N 0; in the former, ‚ has increased. If the investors are −1 ∗ wL = u 4u 50 already randomizing between organizational designs, we Ross: On Evaluation Costs in Strategic Factor Markets Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS 13

know from the proof of Proposition 4 that the probability Aghion, P., N. Bloom, R. Blundell, R. Griffith, P. Howitt. 2005. Com- of winning with an offer at the junction of the supports of petition and innovation: An inverted-U relationship. Quart. J. 0 be N −1 Econom. 120(2) 701–728. FS and FI is 64P − 41 − ‚55/P7 . This probability cannot .org. change as N 0 changes, because it represents the q4‹5 where Allen, L., G. DeLong, A. Saunders. 2004. Issues in the credit risk not

ms modeling of retail markets. J. Banking Finance. 28(4) 727–752. the profitability of the two organizational designs is equal. y or 0 Alonso, R., N. Matouschek. 2008. Optimal delegation. Rev. Econom. This expression is decreasing in N , so ‚ approaches 1 from Stud. 75(1) 259–293. ma below to compensate. ƒ Anton, J. J., D. A. Yao. 2008. Attracting skeptical buyers: Negotiat-

file ing for intellectual property rights. Internat. Econom. Rev. 49(1) 319–348. Competitive Equilibrium with Asymmetric Investors Ashkenas, R. N., L. J. DeMonaco, S. C. Francis. 1998. Making the The By a similar argument to that of Proposition 1, FS and FI are . deal real: How GE capital integrates acquisitions. Harvard Bus. missions@inf ∗ ¯ massless and gapless on 4I 1 ‹], and both investors evalu- Rev. 76(1) 165–178.

per ate with probability less than 1. However, given the greater Barney, J. B. 1986. Strategic factor markets: Expectations, luck, and ibers business strategy. Management Sci. 32(10) 1231–1241. to sensitivity of incentivization to low q4‹5, FS must have a mass point at I ∗, so the profitability of making an offer ‹ Baron, D. P., D. Besanko. 1984. Regulation, asymmetric information, ∗ and auditing. RAND J. Econom. 15(4) 447–470. subscr infinitesimally close to, but greater than, I is zero for the Bharath, S., S. Dahiya, A. Saunders, A. Srinivasan. 2007. So what policy investor using incentivization. Accordingly, F must be open to I do I get? The bank’s view of lending relationships. J. Financial at I ∗. le Econom. 85(2) 368–419. this Birkinshaw, J. 1997. Entrepreneurship in multinational corpora- Continuous Monitoring tions: The characteristics of subsidiary initiatives. Strategic ailab v Suppose that with probability m ∈ 601 17, the principal can Management J. 18(3) 207–229. a detect that the agent has improperly made an offer for Birkinshaw, J., N. Hood, S. Jonsson. 1998. Building firm-specific advantages in multinational corporations: The role of sub-

regarding or rejected a resource, but the principal cannot verify the sidiary initiative. Strategic Management J. 19(3) 221–241. made agent’s effort directly, because an agent who shirks but Bonanno, G., J. Vickers. 1988. Vertical separation. J. Indust. Econom. is makes the right decision by chance could escape detection. 36(3) 257–265. If there is no limit on the penalty for misfeasance, the prin- Bradach, J. L. 1998. Franchise . Harvard Business cipal could guarantee the agent never shirks by setting an School Press, Boston, MA. questions which arbitrarily low wage (e.g., negative infinity). In reality, there Brandenburger, A., H. Stuart. 1996. Value-based business strategy. y are limits to how severely a firm can punish an employee. J. Econom. Management Strategy 5(1) 5–24. an As a normalization, then, let the lowest possible wage be Brandenburger, A., H. Stuart. 2007. Biform games. Management Sci. 53(4) 537–549. ersion, 0. For any given m, the constraint analogous to IC 1 in the v Bulow, J., P. Klemperer. 2002. Prices and the winner’s curse. RAND send base model is then: J. Econom. 33(1) 1–21. Casadesus-Masanell, R. 2004. Trust in agency. J. Econom. Manage- ance ƒq4‹5u4w 5 + 41 − ƒq4‹55u4w 5 − e H L ment Strategy 13(3) 375–404.

Please Cassiman, B., G. Valentini. 2009. Strategic organization of R&D: The Adv ≥ 641 − m5ƒq4‹5 + 41 − ƒq4‹557u4wL5 . choice of basicness and openness. Strategic Organ. 7(1) 43–73. in e Chatain, O., P. Zemsky. 2011. Value creation and value capture with site u4wH 5 ≥ 41 − m5u4wL5 + 1 ƒq4‹5 frictions. Strategic Management J. 32(11) 1206–1231. s Collis, D. J., E. W. Johnson. 1998. Newell company: Acquisition ticles which clearly binds more tightly, the lower m is. strategy. HBS Case 9-794-066, Harvard Press, Ar Boston, MA. author’ Continuous Effort Comptroller of the Currency. 1998. Comptroller’s handbook: Loan this Suppose that effort is continuous and that the greater the portfolio management. Retreived April 18, 2011, http://www the to effort the agent expends, the more accurately the agent can .occ.treas.gov/handbook/lpm.pdf. determine whether a resource is valuable. Then, the per- Comptroller of the Currency. 2003. Today’s credit markets, rela- ight ceived suitability of a resource ƒ4e5 is a function of e, as is tionship banking, and tying. Retreived April 18, 2011, http:// yr the probability d4e5 that a resource deemed to be valuable www.occ.gov/topics/laws-regulations/TyingWhitePaper.pdf. including Deloitte Financial Advisory Services. 2010. Corporate development cop , in fact is not. For any given e, the constraint analogous to 2010: Refining the M&A playbook. Retrieved April 18, 2011, IC 1 in the base model is then http://www.mandachicago.com/corporate_development_refining __m_and_a_playbook.html. holds ebsite E6u4w5 — valuable741 − d4e55 + E6u4w5 — not valuable7d4e5 Demski, J. S., D. E. M. Sappington. 1987. Delegated expertise. w e J. Accounting Res. 25(1) 68–89. ≥ E6u4w5 — resource not acquired7 + 0 ƒ4e5q4‹5 Denning, W. E. 1986. Out of the Crisis. MIT Press, Cambridge, MA.

other Dierickx, I., K. Cool. 1989. Asset stock accumulation and sus-

y As q4‹5 declines (for example, because of a lower ‹ or an tainability of competitive advantage. Management Sci. 35(12) INFORMS 0 1504–1511. an increase in N ), this constraint binds more tightly. Dushnitsky, G. 2010. Entrepreneurial optimism in the market for

on technological innovation. Organ. Sci. 21(1) 150–167. Dyer, J. H., H. Singh. 1998. The relational view: Cooperative strat- yright: References egy and sources of interorganizational competitive advantage. Adegbesan, J. A. 2009. On the origins of competitive advantage: Acad. Management Rev. 23(4) 660–679. Cop posted Strategic factor markets and heterogeneous resource comple- Evans, J. H. 1980. Optimal contracts with costly conditional audit- mentarity. Acad. Management Rev. 34(3) 463–475. ing. J. Accounting Res. 18 108–128. Aghion, P., J. Tirole. 1997. Formal and real authority in organiza- Fershtman, C., K. L. Judd. 1987. Equilibrium incentives in oligopoly. tions. J. Political Econom. 105(1) 1–29. Amer. Econom. Rev. 77(5) 927–940. Ross: On Evaluation Costs in Strategic Factor Markets 14 Management Science, Articles in Advance, pp. 1–14, © 2011 INFORMS

Frost, T. S., J. M. Birkinshaw, P. C. Ensign. 2002. Centers of excel- Makadok, R., J. B. Barney. 2001. Strategic factor market intelligence: lence in multinational corporations. Strategic Management J. An application of information economics to strategy. Manage-

be 23(11) 997–1018. ment Sci. 47(12) 1621–1638.

.org. G.A. Kraut Company. 2008. What makes a great CFO? Does CFO Mookherjee, D., I. P. L. Png. 1992. Monitoring vis-à-vis investigation

not Amer. Econom. Rev. 82

ms quality impact valuation? Retreived April 18, 2011, http:// in enforcement of law. (3) 556–565. y

or www.gakraut.com/thoughts.html. Mosakowski, E. 1998. Entrepreneurial resources, organizational Gertler, M., S. Gilchrist. 1993. The role of credit market imper- choices, and competitive outcomes. Organ. Sci. 9(6) 625–643. ma fections in the monetary transmission mechanism: Arguments O’Donnell, S. W. 2000. Managing foreign subsidiaries: Agents of

file and evidence. Scand. J. Econom. 95(1) 43–64. headquarters, or an independent network. Strategic Manage- Haspeslagh, P. C., D. B. Jemison. 1991. Managing Acquisitions: Cre- ment J. 21(5) 525–548. ating Value Through Corporate Renewal. Free Press, New York. Pacheco-de-Almeida, G., P. Zemsky. 2007. The timing of resource The

. Hedlund, G. 1981. Autonomy of subsidiaries and formalization development and sustainable competitive advantage. Manage- missions@inf of headquarters-subsidiary relationships in Swedish MNCs. ment Sci. 53(4) 651–666.

per L. Otterbeck, ed. The Management of Headquarters-Subsidiary Penrose, E. 1959. The Theory of the Growth of the Firm. Oxford ibers Relationships in Multinational Corporations. St. Martin’s Press, University Press, Oxford, UK. to New York, 25–78. Peteraf, M. A. 1993. The cornerstones of competitive advan- Hermalin, B. E. 1994. Heterogeneity in organizational form: Why tage: A resource-based view. Strategic Management J. 14(3) subscr otherwise identical firms choose different incentives for their 179–191. policy to managers. RAND J. Econom. 25(4) 518–537. Petersen, M. A., R. G. Rajan. 1994. The benefits of lending rela-

le Jehiel, P., B. Moldovanu, E. Stacchetti. 1996. How (not) to sell tionships: Evidence from small business data. J. Finance 49(1) this nuclear weapons. Amer. Econom. Rev. 86(4) 814–882. 3–37. Kim, D. 2006. Capital budgeting for new projects: On the role of Pfeffer, J., G. R. Salancik. 1978. The External Control of Organizations: ailab v auditing in information acquisition. J. Accounting Econom. 41(3) A Resource Dependence Perspective. Harper & Row, New York. a 257–270. Porter, M. E. 1979. How competitive forces shape strategy. Harvard Koch, T. A., S. S. MacDonald. 2010. Bank Management. South- Bus. Rev. 57(2) 137–145. regarding Western Cengage Learning, Mason, OH. Rajan, R. G. 1992. Insiders and outsiders: The choice between made Lambert, R. A. 1986. Executive effort and selection of risky projects. informed and arm’s-length debt. J. Finance 47(4) 1367–1400. is RAND J. Econom. 17(1) 77–88. Roberto, M. A., G. M. Carioggia. 2005. Polycom’s acquisition pro- Lang, K., R. W. Rosenthal. 1991. The contractors’ game. RAND J. cess. HBS Case 9-304-040, Harvard Business School Press, Econom. 22(3) 329–338. Boston, MA. questions which Lang, W. W., L. I. Nakamura. 1995. “Flight to quality” in banking Rosenthal, R. W. 1980. A model in which an increase in the number y and economic activity. J. Monetary Econom. 36 145–164. of sellers leads to a higher price. Econometrica 48(6) 1575–1579. an Liberti, J. M., A. R. Mian. 2009. Estimating the effect of hierarchies Sklivas, S. D. 1987. The strategic choice of managerial incentives. on information use. Rev. Financial Stud. 22(10) 4057–4090. RAND J. Econom. 18(3) 452–458. ersion, v MacDonald, G., M. D. Ryall. 2004. How do value creation and Stahl, D. O. 1989. Oligopolistic pricing with sequential consumer send competition determine whether a firm appropriates value. search. Amer. Econom. Rev. 79(4) 700–771. Management Sci. 50(10) 1319–1333. Szalay, D. 2009. Contracts with endogenous information. Games ance Makadok, R. 2001. Toward a synthesis of the resource-based and Econom. Behav. 65 586–625.

Please dynamic-capability views of rent creation. Strategic Manage- Vickers, J. 1985. Delegation and the theory of the firm. Economic J. Adv

. ment J. 22(5) 387–401. 95 138–147. in Makadok, R. 2002. A rational-expectations revision of Makadok’s Vroom, G. 2006. Organizational design and the intensity of rivalry. site resource/capability synthesis. Strategic Management J. 23(11) Management Sci. 52(11) 1689–1702. s 1051–1057. Wernerfelt, B. 1984. A resource-based view of the firm. Strategic ticles Makadok, R. 2003. Doing the right thing and knowing the right Management J. 4(2) 171–180.

Ar thing to do: Why the whole is greater than the sum of the Williamson, O. E. 1975. Markets and Hierarchies: Analysis and

author’ parts. Strategic Management J. 24(10) 1043–1055. Antitrust Implications. Free Press, New York. this the to ight yr including cop , holds ebsite w other y INFORMS an on yright: Cop posted