The Governance of Contractual Relations Incentives and Transaction

The Governance of Contractual Relations: Incentives and Transaction Costs Steve Tadelis UC Berkeley Prepared for ESNIE Cargese, France May 2008

The Role of Contracts • The “broad” economic problems: – – • The Role of Contracts – – • choice and decision-making in a world with scarcity production, distribution, and consumption of goods and services Facilitate the production/delivery of goods and services E. g. , sales, rentals, licensing… Activity of Focus: Procurement – Delivery of goods and services between parties in the “shadow” of an existing institutional backdrop

Procurement Contracts • The economic problem of procurement: – – – • In practice, procurement is either: – – • Identify the need and understand the deliverable Find capable sellers and anticipate their costs Make sure that sellers deliver on promise Off the shelf, well specified products/services (boring!) Custom design, uniquely built products/services Problems with custom design – – – Design and specs may be (and are often) incomplete Design problems/evolving needs require adaptation “Adaptation costs” may be severe

The Questions: • What contacting structures are most efficient? – Uncover the costs/benefits of different contract regimes – Identify empirically measurable attributes for policy • How should contracts be awarded? – Award mechanisms are tied to contract regimes • What are the economic costs of adaptation? – Measure the impact of change on costs above and beyond the efficient cost of “change” • What are the resulting policy implications? – Common concerns of private information may be less important than mitigating adaptation costs

The Approach • Ingredients: – In the spirit of Williamson (TCE): adaptation cost focus (simple, right, plausible, testable) – Agency rigor (incentives and contracting costs) – Focus on empirically measurable trade-offs • Tradeoff: productive efficiency vs. explicit contracting costs. • Fresh interpretation: “make” or “buy” cast in terms of input versus output contracting.

Theory: A simple model • Based on Bajari and Tadelis (2001) and Levin and Tadelis (2008) • Technology: – One unit of service to be procured – quality depends on labor inputs: q = ( +e)t. – t ¸ 0 is time on the job, > 0 is baseline productivity and e ¸ 0 is “effort” intensity. • Endowments: – Agent is endowed with time T to be allocated between time on the job and some outside activity that pays w > 0 (outside job or value of leisure)

Model, cont. • Preferences: – Agent (seller): Likes money, not work (needs incentives) – Principal (buyer): Likes “performance” and money • Contracting: – Principal can contract on performance or time – Costs of writing/enforcing contracts: • On time: small (some > 0) • On performance: costly – Contracting costs increase in “complexity” – Complexity is hard to pin down (costly to describe, monitor, measure; need for flexibility…)

A Helpful Conclusion Proposition: The optimal contract either contracts only on time or only on performance, but not on both

Results and Comparative Statics costs V (q, s) W (q |EC) Production costs with employment contracts specifying “time” W (q |PC) + k(q, m) adding performance contracting costs W (q |PC) Production costs with Performance contracts specifying quality

Introducing Adaptation and Decisions • Technology: – Ex post noise ε calls for making decisions to best adapt the production to achieve quality q: q = ( +e)f(d, ε)t. – f(d, ε) a productivity coefficient. For any realization of ε there is a unique maximizer d. • Preferences: – Agent’s preferences are given by – Notice the potential conflict over decisions.

Contracting Assumptions • Principal can contract on three dimensions: – Minimal performance standards: – Minimal time on job: – Decision rights: which party chooses d • Costs of writing/enforcing contracts same as before (cost of contracting over who makes decisions is trivial) • Note: Assumptions on contracting and information are symmetric vis-à-vis ex ante and ex post. (q and d)

Results: General idea • The technology of a transaction: Labor (time) PRT (Hart) Labor (“effort”) Material and Machines Function q “Decisions” • Question: Optimal contract/Governance? – Buy: Buy performance/function – Make: Buy inputs and control decisions • Similar complementarties and comparative statics

Implications for Public Procurement Basic comparative statics predictions. 1. Higher costs of specifying, measuring, or adjusting desired performance less performance contracting (more C+ and more “make”). Scale economies: city size and location. 2. Small/rural cities are more constrained (less economies of scale and less markets) modeled trade-offs most relevant for larger cities. 3. Small cities may use public contracts as imperfect substitute for inhouse. Political economy effects. 4. Cities with more politically motivated decisions -- mayor rather than manager, older cities with established unions less contracting. 6. These cities should also place less emphasis on economic trade-offs model trade-offs less relevant. Contracting scope economies. 7. City managers must learn to write/manage contracts, so contracting may be more likely in cities that provide a lot of services • These predictions are tested and verified in Levin and Tadelis (2007) (Survey was key!).

Award Mechanism Policies • When transactions are simple, FP contracts should be selected and awarded by auction • When transactions are complex, C+ contracts should be selected and negotiated with reputable sellers – Another reason to negotiate complex projects is to extract information ex ante from the selected seller • These predictions are verified in the private sector (Bajari, Mc. Millan and Tadelis, 2008) but not in the public sector. Why? • FARs prohibit negotiated contracts • What are the implied adaptation costs?

Trying to Measure Adaptation Costs • Do anticipated changes affect seller bids? – Need to Incorporate changes into a bidding model – Back out the effect of expected changes on bids • What are the economic costs of adaptation? – Need to know contract’s initial condition – Need to know what the changes were – Need to measure the impact of change on the project’s costs above and beyond the efficient cost of change • What are the resulting policy implications? – Need to try and stack up the adaptation costs against fear from corruption • BHT (2007) address the first two questions

Unit Price Auctions: A Hybrid Contract • Highway construction: – Gravel, asphalt, sidewalk, measured in some units – A specification of a project can be in how much of each input (output) is needed – (Timber selling auctions: Haile; Athey-Levin) • flexibility built into the contract: – Task is known but quantities are not – If there are changes to the quantities ex post (incomplete design) then the payment mechanism for changes is clear.

Unit Price Auctions: Example • A specification is a vector of quantities for the measurable units expected to be used • A bid is a vector of unit prices per quantities • A price is the dot product of these two vectors Item Description Est. Quantity Unit Bid Est. Item Bid 1. asphalt (tons) 25, 000 25. 00 625, 000. 00 2. sidewalk (sq. yds) 10, 000 90, 000 3. rumble strip 5. 00 250. 00 Bid Price: $715, 250. 00 50 • Winning bid is evaluated by price (usually lowest)

Adjustments, Extras and Deductions • Adjustments – – Engineer’s estimated quantities may be off actual quantities differ from estimated ones. Large deviations: adjustments to final payment Our data includes estimated quantities, actual quantities and adjustments to payments • Extra Work – Unanticipated problems in the design – Our data includes payments for extra work • Deductions – Contractor screw ups (time, specs, etc. ) – Our data includes deductions

Adaptation Costs • τ = proportional adaptation costs so total adaptation costs are: K = τA+A+ - τA-A- + τXX - τDD • With adaptation costs, R(bi) = ∑Tt=1 btiqta + A + X + D – K • If changes and adaptation costs are anticipated, we can use bids to back out the implied adaptation costs! • BHT (2007) do exactly this: adaptation costs are very large!

Concluding Remarks • Simple model of procurement Contracts – View as choice of contractual form (employment/C+ vs. specific performance/FP). – When “output” contracting is too expensive then replace it with a partial control of inputs and decisions – Trade-off: productive efficiency vs. contracting costs. • Designed with empirical predictions in mind – Consistent with previous empirical studies – Useful to guide public (and private) sector • Offers some “Williamsonian” guidance into the choice of procurement regimes as they relate to transaction characteristics. • Lesson is that adaptation costs can be severe, so contractual choice should be mindful of these.

Thank You