the experimental evaluation of (contractual solutions to) the holdup problem randolph sloof...

37
The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Post on 22-Dec-2015

213 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

The experimental evaluation of (contractual solutions to) the

holdup problem

Randolph Sloof

Amsterdam School of Economics

Page 2: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Focus of this workshop

1. Focus on one particular topic: the holdup under-investment problem

2. Focus (predominantly) on one research tool: controlled laboratory experiments

Page 3: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Key ingredients of holdup problem

1. Specificity of the investment

2. Incomplete contracts– Allows for renegotiation

• Classic example: GM and Fisher bodies (see Klein et al., 1978)

Page 4: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Relevance of holdup

• Boundaries of the firm / vertical integration– TCE: Coase, Williamson …

– Property rights theory: Hart, Grossman, Moore ...

• Human capital acquisition– General vs. specific training: Becker, MacLeod,

Malcomson …

• …

Page 5: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Empirical evidence on holdup

• Evidence based on field data scarce or indirect

– Crude measure of the size of the specific investment / “appropriable quasi-rent”; e.g. Malcomson (1999) provide rough estimates of firms’ recruitment, exit and training costs:

“Incomplete as they are, the figures certainly indicate the existence of turnover costs and, hence, the potential relevance of holdup.” (p. 2313)

Page 6: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Empirical evidence on holdup (II)

• Evidence based on field data scarce or indirect

– Asset specificity appears positive determinant of vertical integration in a number of TCE studies

• But does not necessarily imply that asset specificity increases transaction costs of non-integration (i.e. holdup); transaction costs could decrease to greater extent under integration (Gibbons, 2005, p. 221)

– High wage returns to training suggest underinvestment• But estimates are much smaller when you correct for

selectivity bias (Leuven and Oosterbeek, 2007)

Page 7: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Problems with field data

• Direct field data evidence on holdup is difficult to obtain:

– Root of the problem is non-verifiability of the specific investment; investment is difficult to objectively measure and observe

– Assessment of the extent of holdup problem requires counterfactual information (e.g. what would have been a worker’s productivity without investment in training)

Page 8: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Controlled laboratory experiment

• “…Create a manageable microeconomic environment in the lab in which real people participate for real and substantial profits…” (Plott, Smith)

– Adequate control and accurate measurement of relevant variables

– Institutional rules are set by experimenter (and no deception!)

– Induced valuation: decisions determine monetary payoffs

Page 9: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Remainder of talk

• Sketch some experiments on holdup and solutions to it:– Promotion policies within firms– Privacy rights / informational rents– (Elaborate contracts that create ‘attractive’

outside option)

• Implications and limitations

Page 10: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Promotion rules and skills acquisition

• Up-or-out promotion policy provides strong incentives to invest in firm-specific skills and may therefore solve the holdup problem– It comes at a cost by wasting acquired skills of people not

promoted– There may be a mismatch between the matching and

incentive motive of promotions

• Alternative: up-or-stay– Choice between up-or-out and up-or-stay represents

tradeoff between inefficient matching and inefficient investments in skills acquisition

Page 11: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

An experimental test setup: does the tradeoff exist?

• Two agents: firm and worker• Worker is either of low or high productivity, but type is unknown at

start of the relationship (and not verifiable)• Investment stage:

– During probation period worker can make investment of 25, to increase the probability that he is of high productivity from ¼ to ¾

• ‘Bargaining’ stage:– After worker’s productivity becomes known, firm can offer worker one of

two jobs, or fire the worker– Difficult job pays wage equal to wd=110, easy job wage we

– Worker decides whether to accept the firm’s job offer or not

• Neutral framing in experiment, 30 rounds, random matching, fixed roles

Page 12: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Production environmentLow-productivity High-productivity

firm worker firm worker

Difficult 0 – 110 110 220 – 110 110

Easy 100 – we we 175 – we we

Out 0 0 0 0

• Efficient matching:• Low prod. worker Easy job• High prod. worker Difficult job

• Efficient to invest: (¾–¼)·(220–100) = 60 > 25

100 points = 1 euro

Page 13: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

First best not attainable

• Firm must have incentive to:– Keep Low type in easy job: we < 100

– promote High type to difficult job:• 220 – 110 > 175 – we we > 65

• Given promotion of high type, worker must have incentive to invest:• Low is fired: (¾–¼)·(110 – 0) = 55 > 25 if we > 100

• Low is kept: (¾–¼)·(110 – we) > 25 we < 60

Page 14: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

% of efficient decisionsUp-or-out (we=110)

Up-or-stay (we=70)

Stay-or-stay (we=50)

Investment: Predicted

Actual

100%

85%

0%

46% †

0%

43% †

Assignment:

HighDiff. Predicted

Actual

100%

100%

100%

96%

0%

69%

LowEasy Predicted

Actual

0%

19%

100%

92%

100%

97%

† Single insignificant difference (=0.05)

Page 15: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Up-or-out (we=110)Low-productivity High-productivity

firm worker firm worker

Difficult – 110 110 110 110

Easy – 10 110 65 110

Out 0 0 0 0

• In experiment, 19% of low types is offered easy job• For low types that did not invest this is 3%• For low types that invested this is 27%

• Reciprocity: willingness to sacrifice in order to reward or punish • Positive reciprocity by the firm occurs, but not often given that it is cheap

•Overall surplus realized = 135 < 140 = predicted

Page 16: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Up-or-stay (we=70)Low-productivity High-productivity

firm worker firm worker

Difficult – 110 110 110 110

Easy 30 70 105 70

Out 0 0 0 0

• In experiment, 8% of low types are dismissed• For low types that did not invest this is 2%• For low types that invested this is 9%

• This negative reciprocity by the firm rationalizes investment• Overall surplus realized = 140 > 130 = predicted

Page 17: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Main experimental findings (I)

• Support for comparative statics predictions…:

– under up-or-out more investment, so holdup (underinvestment) problem is mitigated;

– under up-or-out more inefficient dismissals, so up-or-out comes at the cost of wasting skills;

– under stay-or-stay less promotions

Page 18: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Main experimental findings (II)

• …but contracts differ in the extent to which they give scope for reciprocity

– Reciprocity: refers to the motivation to reward kind actions of others and punish unkind ones

– up-or-out: almost no scope for reciprocity– up-or-stay: negative reciprocal reaction of firm increases efficiency;

punish non-investment– As a result, up-or-stay more efficient than up-or-out

• For reciprocity mechanism to work, investment must be observable

Page 19: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Holdup and investment (un)observability

• Theory: holdup occurs because investor does not have full bargaining power:

– Private information about e.g. the investment made (or outside options) may allow the investor to obtain an informational rent;

– Through this informational rent, the investor captures a larger (marginal) return on investment;

– Privacy rights may thus be used to boost investment incentives

Page 20: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Gul (2001, Ecometrica)

“…[one] objective of this paper is to emphasize the role of allocation of information as a tool in dealing with the hold-up problem. Audits, disclosure rules or privacy rights could be used to optimize the allocation of rents and guarantee the desired level of investment. Controlling the flow of information in organizations may prove to be a worthy alternative to controlling bargaining power in designing optimal organizations” (p. 344)

• Here other controls of bargaining power refer to (p. 360): – allocation of ownership rights – detailed specification of non-cooperative renegotiation game

Page 21: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Crowding out

• Many lab experiments (e.g. trust game of Berg et al., 1995) find that holdup is alleviated by fairness and reciprocity motivations

• Private information may interfere with informal reciprocity mechanism when it becomes impossible to observe whether the investor behaved ‘kind’ or not

• Private information about investment made– Informational rent investment – Obstruction of reciprocity mechanism investment – Net effect ?? experiment of Sloof et al. (2007, JEMS)

Page 22: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

An experimental test setup: Does unobservability boost investments? • Two stage game between buyer and seller

1. Investment stage: Buyer chooses between pie of 50, or pie of 130 at costs C (with C{20,40,60})

2. ‘Bargaining’ stage: Seller chooses price demand P [0,130]. Trade takes place at this price iff P actual pie.

• Two types of sessions:– Obs: Buyer’s choice is observable

– Unobs: Buyer’s choice unobservable

• Neutral framing, 36 (6 x 6) rounds, C varied within sessions, random matching within matching groups, switching roles

Page 23: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Standard theory

Demand=50 Demand=130

Invest=0 0, 50 0, 0

Invest=1 80–C, 50 –C, 130

• Unobservable treatment: equiv. to simultaneous move game• Unique mixed strategy NE: iun = Pr(Invest) = 50/130 0.385

C{20,40,60}

• Observable treatment: unique SPE has iobs = Pr(Invest) = 0

iun – iobs > 0 and independent of investment costs C

Page 24: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Social preferences

• Using model of intention-based reciprocity a la Rabin (see paper) or inequality-aversion a la Fehr-Schmidt (see earlier WP version, or Ewerhart (2006)):

– Scope for these motivational factors is large when investment costs C are low relative to the return on investment (130 – 50 = 80)

– With large scope, iobs is high (iobs 1), just like iun is

– With large scope, i.e. low C, private information thus has no impact

iun iobs when C is low and iun > iobs for C is high

Page 25: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Mean investment levels (rounds 19-36)

Unobservable Observable p-value (indiv.)

p-value (group)

C=20 64.4% 57.2% 0.434 0.221

C=40 45.6% 18.9% 0.000 0.016

C=60 38.3% 7.8% 0.000 0.003

St. Theory 38.5% 0%

in C in C

• Investment levels in line with social preferences predictions• For low C motivational crowding out• For high C private information boosts investment incentives • Investment levels decrease with C (scope for reciprocity )

Page 26: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Main conclusions this study

• Unobservability does boost investment incentives, but only when there is limited scope for fairness and reciprocity– Scope is limited when the profitability of the

investment is low (C is close to the gross return of 80)

• With sufficient scope for these motivations, crowding out of informal reciprocity mechanism– Social preferences have a larger impact on investment

incentives when the investment is observable

Page 27: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Holdup and the outside option principle

• The exact form of the post-investment bargaining stage determines the marginal returns on investment

• Intuitive theoretical idea: use contracts to effectively re-structure the ex post bargaining process, such that investor becomes residual claimant– Create a binding ‘outside’ option endogenously through the initial

contract; e.g. a one-sided option to extend the contract

• In an abstract experiment you can test the underlying mechanism

Page 28: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Outside option principleInvestor

Other

100·I

S=3000 S=9000

10,000

• Outside option s of Other acts as constraint on division• Other gets: max {5000+50·I, S}

• Investor gets: min {5000+50·I, 10,000–S+100·I}

• S high: investor gets full return on investment

Page 29: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Experimental setup

• Two stage game between buyer and seller1. Investment stage: Buyer chooses investment I in between 0 and 80, at

investment costs I2

2. Bargaining stage: Multiple-pie alternating-offer bargaining game for 10 rounds, buyer makes first offer, as responder you can opt out

• Two types of treatments:– S=3000: Buyer is predicted to under-invest (25 < 50)– S=9000: Buyer is predicted to invest efficiently (50)– Comparing treatments, investment should increase with S

• Neutral framing, 20 (4 x 5) rounds, S varied within sessions, random matching within matching groups, fixed roles

Page 30: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Mean investment levels (rounds 11-20)

S=3000 S=9000 p-value (indiv./grp)

Predicted 25 50

Observed 36.96 39.46 0.27 / 0.26

“Optimum” 29.60 44.13

(5.11) (2.92)

• Investment levels do not vary with S• Holdup occurs for both values of S• For S=3000 buyers overinvest from a selfish point of view (given actual bargaining outcomes)

Page 31: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Main conclusions this study

• No support for predicted relationship between investment incentives and outside option principle– Potential explanation: self-serving bias of buyer lead

them to ask for (and to feel entitled to) almost the full return on investment, independent of S

– For low S this demand is not honoured, therefore overinvestment in this case

• No support for underlying mechanism: casts doubts on contractual solutions that rely on this

Page 32: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Main conclusions general

• Underinvestment typically much less of a problem than theory predicts– In the lab subjects typically behave less

opportunistically; fairness and reciprocity motivations play a role

• Some suggested solutions do alleviate holdup, but less so than predicted

Page 33: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Potential criticisms

• Laboratory is an artificial setting, in reality people will behave differently because e.g.:

– Higher stakes: but fairness and reciprocity play important role in high stake lab experiments as well (Camerer, 2003)

– In reality decisions are not taken by students: yet lab tests of subject pool effects indicate that professionals are also motivated by fairness / reciprocity (Fehr and List, 2004)

– In reality personal interaction and communication: lab experiments show that promises and threats alleviate holdup (Ellingsen and Johannesson, 2004)

Page 34: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

External validity

• Many ‘external validity’ issues can be pre-tested in the lab

• But in the end, verification by means of field data is needed:– Field experiments to directly verify lab results

• e.g. Gneezy and List (2006) and Falk (2007) examine gift exchange in the field

• but this may be difficult for holdup

– Incorporate insights from the lab into field studies• Leuven et al. (2005) on worker reciprocity and employer investment

in training

Page 35: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Importance of reciprocity in practice

• Reciprocity may provide implicit commitment and limit opportunism– Reciprocal workers may abstain from bargaining for a

wage increase (or abstain from quitting) after the firm has invested in their skills

– Firms may therefore be more willing to pay for training of those workers who are reciprocal

• 2001 survey among 3127 dutch employees– Info about participation in (firm-sponsored) training – Include question to measure reciprocity motivations

Page 36: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Reciprocity question

• “If someone does something that is beneficial to you, would you be prepared to return a favour, even when this was not agreed upon in advance?”– Not at all (1%)

– No (3.3%) Low

– Maybe (9.1%)

– Yes (60.8%) Medium

– Certainly yes (25.8%) High

Page 37: The experimental evaluation of (contractual solutions to) the holdup problem Randolph Sloof Amsterdam School of Economics

Training participation rates by reciprocity type

Low Medium High

Firm-sponsored 32.5%* 40.4%* 47.5%*

No firm support 7.5% 10.0% 10.5%

* Significantly different (1% level, ranksum tests)

• Same conclusion follows when controlling for personal and firm characteristics in a probit analysis