Basic research utilization and the R&D employment contract

Transcription

1 ömföäflsäafaäsflassflassflas ffffffffffffffffffffffffffffffffffff Disussion Papers Basi researh utilization and the R&D employment ontrat io ustonen Helsini Shool of Eonomis and HECER Disussion Paper No. 108 ay 006 ISSN HECER Helsini Center of Eonomi Researh, P.. Box 17 (Aradianatu 7), FI University of Helsini, FINLAND, Tel , Fax , E mail info heer@helsini.fi, Internet

2 HECER Disussion Paper No. 108 Basi researh utilization and the R&D employment ontrat Astrat Basi researh results redue a firm s R&D osts or interat with the R&D output in the maret. Utilization of asi researh arries its unertainty over to the firm s revenue. We analyse the ontrat etween the firm s owner and R&D manager. For a given R&D projet, strong asi researh influene to the firm s expeted revenue implies lowpowered inentives. When the owner an hoose the R&D projet, it is more aligned with asi researh than when the manager hooses it and the inentive is lower powered. Contrary to the standard result, the power of the inentive is inreasing in the unertainty of asi researh and in the manager s ris aversion. JEL Classifiation: D3, L11, L15 Keywords: asi researh, R&D, ontrat, moral hazard io ustonen Department of Eonomis Helsini Shool of Eonomis P.. Box 110 FI FINLAND e mail: mio.mustonen@hse.fi The omments of io Leppämäi, Juuso Välimäi and the partiipants of the HECER and NRI seminars to an earlier version are appreiated. I am grateful for the finanial support of the Finnish Aademy. The usual dislaimer applies.

3 1 Introdution Firms R&D projets are affeted y the results of asi researh. A firm utilizes the results to redue R&D projet osts or the results may interat with the firm s R&D output in the maret, either omplementing or sustituting it. A prominent example of the latter is the urgeoning pen Soure movement that has properties of asi researh. We raise the question, how does the existene of a asi researh projet affet firm ehaviour and the R&D manager s inentives to exert unoservale effort? We present a model where the diretions and ojetives of asi researh projets are nown y firms. A firm s R&D projet may either e fixed or the owner or manager an hoose its diretion in relation to a relevant asi researh projet. Firms do tae into aount asi researh projets in their R&D deisions. For example, Ward and Dranove (1995) provided evidene that firms in the pharmaeutial industry inrease their R&D in those therapeuti ategories that exhiit investments in asi researh. For our purposes, the phenomenon we all a asi researh projet has to have the following properties that are widely aepted in the literature 1. First, its properties and eventual output are puli and freely usale. Thus the output arries a zero prie. Seond, the appliaility of a asi researh projet s output in the firm s R&D, even if the firm nows the general outline of the researh projet, is unertain. Third, the sientists have no strategi onsiderations when they reate a asi researh projet. pen Soure programming is an interesting example of ativity reminisent of asi researh in that it fulfils the aove requirements. Basi researh is usually onsidered to sustitute firms R&D effort and aordingly redue osts. However, Steinmueller (1994, p.59) pointed out that asi researh may affet R&D y reduing the expeted returns from some lines of applied researh or inreasing the real returns in other areas. pen Soure programs enhane or ompete with firms programs in the maret. A pharmaeutial asi researh projet tries to reate a moleule that a 1 For surveys of the general eonomi properties of asi researh, see Stephan (1996), Dasgupta and David (1987, 1994), Broos (1994), Salter and artin (001) and Goel and Rih (005). Evenson and Kislev (1976) and Aghion and Howitt (1997) provided models whih analyse the growth effets of the interation etween asi and applied researh. For desriptions of pen Soure see Lerner and Tirole (00), Shiff (00), Heintzman (003), ustonen (003, 005) and The GNU projet (004a,). 1

4 firm s R&D also sees. If the asi researh projet sueeds, its free output sustitutes the firm s produt in the (liensing) maret and thus it is possile that the existene of asi researh redues a firm s expeted revenue. ur analysis allows for oth positive and negative revenue effets. We analyze the employment ontrat of the R&D manager of a firm. The effort of the ris averse manager is unoservale to the owner, who an only oserve the maret revenue. In our analysis, we extend the ontrating model of Holmström and ilgrom (1987) and Holmström (1979). It features unertain revenue, a linear inentive, unoservale ontinuous agent effort and onvex osts 3. The presene of asi researh influenes the ontrat design in two ways. First, asi researh results redue the firm s R&D osts or they omplement or sustitute the firm s R&D in the maret and this has a diret effet to the oservale revenue of the firm. Seond, asi researh projets are inherently unertain and a stronger influene of asi researh also inreases the volatility of the firm s revenue. As a first step, we analyse the employment ontrat assuming that the position of the asi researh projet relative to the firm s R&D projet is given. It is also natural to analyse the ontrat in the ase where the owner an hoose the position of the R&D projet in relation to asi researh. As a third ase, we analyse the ontrat assuming that the firm s ey R&D deisions are under the disretion of the manager. After aepting the ontrat, he hooses oth effort and the relative position of the R&D projet. Both hoies are unoservale. If the relative positions of asi researh and R&D projets are given, the effet of asi researh presene is quite natural. The stronger the expeted revenue impat of the asi researh, positive or negative, the more volatile is the revenue. The owner offers a lower powered inentive eause of this volatility and thus the optimal inentive oeffiient is dereasing in the revenue impat. In aordane with the traditional result, now the inentive dereases also in the variane of the output of asi researh and in ris aversion. Comparison of the ases where the owner or the manager endogenously hooses the relative position of the R&D projet shows that the owner always hooses a loser proximity etween the projets and the inentive is 3 For a survey of inentives in organizations, see Gions (1998). Parallel to our researh, Dasgupta and David (1994) and Lazear (1997) researhed inentives to perform asi researh.

5 lower powered. In star ontrast to reeived literature, in oth ases the optimal inentive oeffiient is inreasing in the asi researh output unertainty and in the manager s ris aversion. Traditionally it has een onsidered that the utilization of asi researh in firm R&D is hampered y frition in nowledge transfer. The aility to asor researh nowledge has een presented as the motivation for ostly asi researh in firms (e.g Rosenerg 1990). ur analysis reveals another ost fator in asi researh utilization: If the firm annot align its R&D with asi researh, the higher is the unertainty over the asi researh projet s outome the lower is the R&D effort. With optimally aligned R&D, inreased unertainty of asi researh surprisingly leads to higher R&D effort. So if owners or managers are ale to adjust, high ris in asi researh reates a positive externality to firm R&D and this ould e taen into aount if the government an influene the ris levels of asi researh. The analysis further shows that firm profits are higher when the owner determines the R&D strategy. This indiates that in researh intensive industries, we should oserve firm owners with sientifi nowledge having an advantage to institutional owners. In all senarios, the loseness etween asi researh and R&D projets and the R&D manager s effort are inversely related, implying that asi researh rowds out R&D due to moral hazard in R&D. Baggs and Bettignies (004) found within a duopoly model of horizontal differentiation that inreased ompetition diretly rings aout stronger inentives 4. Raith (003) reahed the same onlusion, aleit indiretly, in an oligopoly model 5. ur result is reversed in the ase where asi researh sustitutes the firm s R&D in the maret a natural outome in pen Soure programming. The stronger sustitute is the asi researh outome, and thus the more ompetitive the situation of the firm the weaer is the optimal inentive. In the ase of omplement or ost reduing asi researh, stronger omplementarity, whih an e interpreted as redued ompetition, rings aout a weaer inentive. In the literature, the seletion of risiness has een reognized. eth (1996) analysed a senario where the agent an expend produtive effort and effort to redue outome variane, oth unoservale. eth reported onditions under whih the prinipal wants to motivate the agent to 4 In Baggs and Bettignies (004), a lower transport ost redues the prinipals marginal ost of induing effort. In equilirium, they hoose higher inentives. 5 In Raith (003), ompetition indues firm exit, whih in turn reates higher ost redution inentives for remaining firms. 3

6 wor on reduing the variane. In Demsi and Dye (1999), the manager an mae mean variane trade offs in projet seletion. They presented and analyzed a speial ontrat, linear in outome sample mean and variane and inluding penalty terms from deviations etween the announed mean and variane and their realized samples. Ghata and Pandey (000) analysed agriultural ontrating and found that under limited liaility, shareropping is generated only when the tenant simultaneously and independently hooses effort and ris. In ontrast to this literature, in our paper, we assume that the hange in outome variane is an externality of a deision involving the firm s R&D projet position and thus ostless. ur results provide an explanation to empirial findings that are ontrary to the results of the Holmström ilgrom model. In a survey of over twenty empirial studies, Prendergast (1999, 00) reported that a majority of them found a neutral or positive relationship etween measures of ris and pay performane sensitivity. Prendergast (000) offered reasons why the standard theory would not hold. In unertain environments, i) input monitoring, ii) sorting, iii) investigations and iv) areer onerns are less effetive. Adding suh assumptions to the standard model yields results where the inentive is inreasing in ris. Raith (003) in turn showed in a model of oligopoly with free entry how an inrease in ompetition, for example in the sustitutaility of goods, inreases oth the inentive to ost redution and output ris 6, resulting in a positive orrelation etween the two. In our model, the result follows ausally from the manager s or the owner s optimal hoie of the R&D projet position while the maret struture remains given. The analysis proeeds as follows: in setion we develop first the optimal inentive for given asi researh and R&D projet diretions. Then we solve for the optimal inentive and optimal R&D projet diretion when either the owner or the manager an hoose the diretion. In setion 3 we analyse and ompare the outomes and setion 4 onludes. 6 In Raith (003), an inrease in sustitutaility auses firms to exit and the inentive for ost redution is higher for the remaining firms. Simultaneously, the onstant variane of ost redution of eah firm translates to higher output variation as the numer of firms dereases. 4

7 The model Consider a moral hazard senario ( Holmström and ilgrom 1987) with a ontinuous hoie of the manager s effort and a wage ontrat that is linear in revenue. The risneutral owner maximises his expeted profit, whih is the net of maret revenue and wages paid, maxπ = R% w. The manager s unoservale effort is e. We assume that the manager s effort is manifested as the value of the R&D projet. A asi researh projet exists. It s output is unertain, y ~ N( 1, v) %. Consider first that asi researh results affet the osts of the R&D projet. The diretion of the R&D projet relative to the asi researh projet,, α y%, where 0 0 < max, determines the ost of the R&D projet, % %. The α > is the fixed ost. The firm s revenue is R = e ( α y) asi researh projet output may alternatively affet the expeted value of the R&D projet in the maret. A visile example of suh is pen Soure programming. If the asi researh output is a sustitute to the R&D projet (as an S program may well e to a ommerial program), manifested y < 0, the revenue impat of asi researh an also e negative. Consider homogeneous uyers of mass 1 with utility funtions U = V p, where V is the value of the R&D outome and p the prie. When asi researh omplements a monopoly firm s R&D, its value is V = e + y. The optimal prie is thus p = e + y. If asi researh is a sustitute, < 0, there are two goods for uyers to hoose from with values V = e and V r = y. The firm has to set the prie in suh a way that uyers prefer the ommerial good i.e. V p V r 0. The optimal prie is p = e + y, and the ex ante revenue is R% = e + y%. Without loss of generality we an set α = 0 and over oth ases with the expeted revenue eing 7 : R% = e + y% (1) 7 We anowledge the possiility of a negative draw of the S effort, ut astrat from it (as Holmström and ilgrom). 5

8 The total revenue is oservale. The ris averse ( r > 0 ) manager s utility is rw ( e ( )) =, where the ost of effort is ( ) U e e e =, > 0. In the asi analysis, we astrat from the (general) maret unertainty. Later on, in setion 3, we disuss its effet to the results..1 Baseline ase: A fixed R&D projet Let us first assume that the relative position of the R&D projet,, is given. The owner offers a ontrat to the manager, who aepts. The manager exerts effort e. The outome of the asi researh projet, y%, is resolved. These fators determine the firm s revenue. Finally, wages w are paid. Consider the inentive pay ontrat w = a + R %. The manager s inentive ompatiility onstraint follows from expeted 1 1 = + +, whih yields the utility maximization, maxeu a ( e ) r v e familiar IC ondition e e = (IC1) The owner maximizes expeted profit, max ( 1 )[ ] Eπ = e+ a onditional on the inentive ompatiility ondition (IC1) and the individual rationality ondition U u. (IR1) Inserting the onstraints (IC1) and (IR1) yields: Eπ = r v u max 1 ( ) 1 6

9 The first order ondition is 1 rv = 0. Note for referene that the manager s ris premium is inreasing in ris aversion and variane. The FC yields the optimal oeffiient of the inentive in the ontrat, = rv. () Proposition 1: The loser the given asi researh and R&D projets are to eah other, or the stronger omplement or sustitute the asi researh projet is to the R&D projet in the maret, the lower powered is the optimal inentive. Proof: In (), the larger is the asolute value of, the smaller is the optimal inentive. QED. A high asolute value of implies that the variane of the oservale revenue is large. This in turn inreases the ris premium of the manager. The owner taes this into aount and in optimum sets a low inentive. In the ase of maret interation of sustitute asi researh and R&D, the requirement of non zero profits presents a lower ound to the sustitutaility etween them. Being a sustitute to the firm s R&D, the presene of asi researh output redues the firm s profit. Furthermore, it redues the firm s R&D eause the manager s effort is redued, sine y (IC1), the manager s optimal effort and thus the R&D are proportional to the inentive oeffiient.. The owner hooses the R&D projet Suppose the owner an set the relative position of the projets y design deisions or y seleting an optimal projet from the firm s portfolio of R&D projets. We assume that the portfolio ontains projets that differ in their loseness to the asi researh projet ut that the produtivity of the manager is equal in all of them. The ost of the owner s (and later the manager s) effort in seleting and implementing the position is fixed y assumption and we set it to zero. The owner offers a ontrat to 7

10 the manager. The manager aepts and exerts effort e in R&D taing the R&D projet position as given. Again, after the unertainty is resolved, wages are paid. The owner will position the projets in a way that will maximise his profit antiipating the ehaviour of the manager. The hoie of position has a diret effet to the owner s profit via a revenue or ost hange and an indiret effet eause it hanges the optimal inentive oeffiient and thus the agent s effort. As in the aseline ase, the manager s optimal effort onditional on the revenue share is e = from (IC1). The owner hooses the optimal inentive and projet diretion taing the manager s hoie and individual rationality (IR1) into aount 1 maxeπ = ( 1 ) r v u, The first order onditions read π 1 = rv = 0, (3a) π = rv = (3) Condition (3a) is equivalent to the first order ondition in the aseline ase and (3) 1 yields the owner s hoie of the projet position for a given inentive, =. In rv (3), the owner equals the marginal revenue from proximity to asi researh to its marginal ost in the form of the manager s ris premium. Being loser to asi researh inreases the firm s revenue and thus wages. But it also inreases unertainty related to output, for whih the manager must e ompensated through the individual rationality ondition (IR1). Thus a ounded optimum is found. We oserve that it is the owner s voluntary hoie that introdues unertainty to the prolem. 8 We note that the seond order onditions are satisfied. We assume that max is suffiiently small to ensure that the owner s profit is higher in an interior solution than for = ε and a very strong S omplement. 8

11 Solving for the optimal inentive yields = 0, whih develops to 3 rv + = 0. (4) rv 4 3 The optimal inentive oeffiient alanes the marginal revenue from the firm s R&D to the marginal ost from the ris premium and from effort. Given the owner s hoie of, the ris premium is dereasing in..3 The manager hooses the R&D projet Consider an (institutional) owner who annot hoose an R&D projet. The hoie of the relative position etween the R&D projet and asi researh, manager. Having aepted the ontrat, the manager exerts effort position,, is left to the e and hooses the. The manager s prolem is reminisent of the multitas prolem. He hooses optimal effort y trading off wage against disutility of effort. The manager also hooses the optimal position y trading off wage against the ris premium. The manager s prolem is 1 1 max EU = a + e + r v e e, ( ) yielding the familiar IC ondition, is e =. The first order ondition for the optimal EU = rv = 0, (5) 9 The seond order ondition is satisfied for 3 4 >. rv 9

12 whih yields the IC ondition onerning the projet position, 1 = for > (IC ) rv In (5), the manager equals the marginal wage from proximity to asi researh to the marginal ost of the ris premium. Lemma 1: For a given inentive oeffiient, the owner hooses loser proximity etween R&D and asi researh than does the manager, >. The owner wants the projets to e loser eause the marginal revenue from asi researh is 1 ut for the manager it is dulled y the revenue share < 1. In their deisions, oth fae the same marginal ost from the ris premium, whih is inreasing in. Comparison of the first order onditions (3) and (5) shows this. From the owner s point of view, information has value sine the manager s unoservale hoie of projet diretion differs from his profit maximising hoie. The owner antiipates the manager s hoies and sets the inentive pay to maximise profit. Inserting the onditions IC1 and IC and IR1 to the firm s profit funtion yields ( ) rv rv rv maxeπ = r v u The first order ondition for maximum reads: dπ 1 1 = = (6) d rv 10 We note that the seond order onditions are satisfied trivially. 11 The seond order ondition is satisfied for 3 >. rv 10

13 Comparing (6) to the first order ondition of the aseline ase (1) shows that the manager s deision rule (IC) for the projet diretion renders the ris premium independent of the inentive in the owner s optimization. The marginal revenue from asi researh (seond term in (6)), is in turn inreasing in the inentive oeffiient, ris aversion and variane. The optimal inentive alanes marginal revenue from R&D and asi researh against the manager s marginal ost of effort. The ondition (6) develops to + = 0. (7) rv 3 3 Analysis and omparison of regimes We start with Proposition : If the owner an position the R&D projet, he hooses a loser proximity to asi researh than the manager and the optimal inentive of the manager is lower powered. Proof: Under the owner s rule, the optimal inentive from (4) is =. rv 3 4 Rearranging equation (7) for the optimal inentive under the manager s rule yields =. We note that the first and seond order onditions are satisfied rv 3 simultaneously for > and 3 3 >. Graphial illustration in figure 1 shows that 4 when the owner an hoose the projet diretion, the optimal inentive turns out to e 11

14 lower than when the manager hooses the projet, <. (See figure 1). From Lemma 1 we an infer that sine also ( ) <, this implies >. 1 QED rv Piee rate Figure 1. ptimal inentive under manager s and owner s rule ur earlier result in Lemma 1 was that for a given inentive oeffiient, the owner hooses loser proximity to asi researh than the manager. The fat that the owner hooses always a lower inentive oeffiient further inreases the differene. Under the owner s rule, the manager s R&D effort is low and the firm utilizes muh asi researh. If the manager maes the (unoservale) hoie of the projet position, the deision rule is not optimal from the owner s point of view. In the margin, the 1 From the graph we an also note that the inentive prolem has an interior solution under manager s rule when 3 sup( ) rv < translating to 4 0 < < and under owner s rule when rv sup( ) rv <, in turn yielding 7 0 < < rv 56 1

15 manager reeives only the portion of inreased revenue (5) whereas the owner reeives all of it (3). Yet oth have to tae into aount same marginal osts of unertainty and effort. To maximize utility, the manager hooses lesser proximity to asi researh, <. The owner ompensates this hoie y induing higher effort y hoosing a higher powered inentive, >. The optimal hoies of the owner and the manager imply that profits are equal or higher in the ase where the owner hooses the R&D projet, π(, ) π (, ) than when the manager does that. This is ovious sine under the owner s rule, he maximises profits y simultaneously determining and whereas under the manager s rule, he determines to maximise utility. Antiipating this, the owner hooses and these hoies are suoptimal for profit maximization. An interesting impliation arises: In an industry with entry arriers and asi researh presene, we should see more firms with nowledgeale owners that are ale to diret firm R&D. Furthermore, the importane of the owner s expertise in R&D and asi researh provides a motivation for Venture Capital presene in suh industries. Institutional owners share ownership with VCs to e ale to utilize their industry sills also in R&D strategy and ontrat design. Analysis of the optimal inentive oeffiients yields a surprising result: Proposition 3: If the manager or the owner an hoose the position of the R&D projet relative to asi researh, the optimal inentive oeffiient is dereasing in the ost of effort, ut inreasing in the manager s ris aversion and the variane of the asi researh outome. d Proof: Graphi illustration shows that d < 0 and < 0. Thus the optimal d rv d rv ontrat inentive is dereasing in the ost of effort and inreasing in ris aversion and variane. QED. 13

16 The result is in star ontrast to urrent literature and provides one explanation to Prendergast s (00) finding that more than half of some twenty empirial studies on inentives annot onfirm the negative relationship etween ris and the level of inentive. If the projet positions are given, the inentive oeffiient dereases in risaversion and revenue variane as in the standard model (Proposition 1). Now the inentive oeffiient inreases in ris aversion and variane. Why is this? In the standard ase, the ris premium is inreasing in the inentive oeffiient. Under the owner s rule, his optimal hoies, imply that the ris premium is dereasing in the inentive oeffiient and the revenue from asi researh is independent of the inentive. Under the manager s rule, the higher an inentive oeffiient the owner hooses, the more distant an R&D projet position the manager selets, eause the ris premium inreases faster than wage in. The owner taes this into aount in his optimal hoie of the inentive oeffiient. From his point of view, the ris premium turns out onstant ut the expeted revenue from asi researh dereases in the inentive (and in the asi researh outome variane and in the manager s ris aversion) via its effet on hosen y the manager. Sine the manager s effort inreases in the inentive oeffiient, this implies the following: Result: R&D effort is inreasing in the ris level of the asi researh projet provided that the either the owner or manager an align the R&D projet with it. If not, R&D effort dereases in asi researh ris level. Let us analyse the deision rules (, 3, IC) that ind the utilization of asi researh,, to the inentive oeffiient. In all of them, there is an inverse relation etween the variales. The R&D effort is determined y and thus we an say that Result: For > 0, the utilization of asi researh and R&D effort rowd eah other out due to moral hazard in R&D. So far, we have astrated from maret unertainty. We an inorporate it to the analysis y replaing the revenue equation (1) y R= e+ y+ m % % %, where m N ( 0, z ) %. Let us analyse the effet of maret ris to the results. We note that maret ris does 14

17 not affet the manager s inentive ompatiility onditions or the owner s ondition for optimal. However, it enters into the owner s profit maximization prolem via the individual rationality ondition. Under owner s rule, the equation for optimal inentive (4) is transformed to ( rz ) 4 3 (7) turns into ( rz ) = 0 and under the manager s rule, rv = 0. The larger is the term rz, the smaller are the rv terms ( rz+ 1) 3 and ( 1) 3 rz+ 4. From the graph in figure 1 we an dedue that that the optimal inentive is thus dereasing in the term rz. Comining this with the result of proposition 3, we note that an inrease in the maret variane, z, and in effort ost,, lower the optimal inentive as predited y Holmström and ilgrom. The result of proposition 3 regarding the variane of the effort of the S ommunity, v, holds. The effet of ris aversion, r, to the inentive is now amiguous and depends on varianes of the maret and S effort unertainty. 4 Conlusion We analysed the optimal employment ontrat under moral hazard when asi researh exists and its output either lowers the osts of R&D or interats with the firm s R&D in the maret. The trade off etween inreased revenue and inreased unertainty results in a ounded optimum for the loseness etween the asi researh and R&D. The owner hooses a loser proximity than does the manager and this leads to a lower powered inentive. ptimal inentive oeffiients in oth ases are inreasing in ris aversion and variane ontrary to the traditional result. ur wor has poliy impliations. If the position of asi researh is given (Proposition 1), the firm is a monopoly, ut onstrained y two effets: The sustitute asi researh is an invisile (in revenue terms) ompetitor to the firm and redues its profits. In addition to that, the unertainty related to the asi researh outome worsens the moral hazard prolem and lowers the manager s effort, whih results in further redution of profits. A firm may e a monopoly if one loos at the maret researh data, ut at the same time it may also fae ompetition in the maret and severe moral hazard prolems in employment due to asi researh, espeially pen Soure, presene. This may e 15

18 something that, for example, ompetition authorities should tae into aount when assessing maret power of firms in the IT industry. 5 Referenes Aghion P. and P. Howitt, 1997, Researh and development in the growth proess, Journal of Eonomi Growth, 1, Baggs J. and J E de Bettignies, 004, Produt maret ompetition and ageny osts, mimeo, Queen s University, (Aessed Deemer 004). Broos H., 1994, The relationship etween siene and tehnology, Researh Poliy 3, Dasgupta P. and P. David, 1987, Information dislosure and the eonomis of siene and tehnology, CEPR Disussion Paper 73, also in: Feiwel D ed., Arrow and the asent of modern eonomi theory (New Yor University Press, New Yor). Dasgupta P. and P. David, 1994, Toward a new eonomis in siene, Researh Poliy 3(5), Demsi J. and R. Dye, 1999, Ris, return and moral hazard, Journal of aounting researh, 37, Evenson R. and Y. Kislev, 1976, A stohasti model of applied researh, The Journal of Politial Eonomy, 84, Ghata. and P. Pandey, 000, Contrat hoie in agriulture with joint moral hazard in effort and ris, Journal of Development Eonomis, 63, Gions R., 1998, Inentives in organizations, Journal of eonomi perspetives, 1,

19 GNU Projet, 004a, Categories of free and non free software, (Aessed April 004). GNU Projet, 004, What is opyleft?, (Aessed April 004). Goel R. and D. Rih, 005, rganization of marets for siene and tehnology, Journal of Institutional and Theoretial Eonomis 161, Heintzman, D., 003, An introdution to open omputing, open standards, and open soure, www 106.im.om/developerwors/rational/lirary/1303.html (Aessed August 003). Holmström B., 1979, oral hazard and oservaility, Bell Journal of Eonomis, 10(1), Holmström B. and P. ilgrom, 1987, Aggregation and linearity in the provision of intertemporal inentives, Eonometria, 55, Lazear E., 1997, Inentives in asi researh, Journal of Laor Eonomis, 15, S167 S197. Lerner J. and J. Tirole, 00, Some simple eonomis of open software, Journal of Industrial Eonomis 50(), eth B., 1996, Redution of outome variane: optimality and inentives, Contemporary Aounting Researh, 13(1), ustonen,., 003, Copyleft the eonomis of Linux and other open soure software, Information Eonomis and Poliy, 15, ustonen., 005, When do firms support the development of sustitute opyleft programs? Journal of Eonomis and anagement Strategy, 14,

20 Prendergast C., 1999, The provision of inentives in firms, Journal of Eonomi Literature, 37(1), Prendergast C., 000, What Trade ff of Ris and Inentives?, Amerian Eonomi Review, 90(), Prendergast C., 00, The Tenuous Trade ff etween Ris and Inentives, Journal of Politial Eonomy, 110(5), Raith., 003, Competition, Ris and anagerial inentives, Amerian Eonomi Review, 93(4), Rosenerg N., 1990, Why do firms do asi researh (with their own money)?, Researh Poliy, 19, Salter A. and B. artin, 001, The eonomi enefits of pulily funded asi researh: a ritial view, Researh Poliy 30, Shiff A., 00, The eonomis of open soure software: A survey of the early literature, The Review of Networ Eonomis 1(1), Steinmueller E., 1994, Basi researh and industrial innovation. In: Dodgson. and R. Rothwell (Eds.), The Handoo of Industrial Innovation, Edward Elgar, Aldershot, Stephan P., 1996, The eonomis of siene, Journal of Eonomi Literature, 34, Ward. and D. Dranove, 1995, The vertial hain of researh and development in the pharmaeutial industry, Eonomi Inquiry 33,