Contract enforcement, social efficiency, and distribution: some experimental evidence.

Taylor, C.R. 2002. "Restoring Economic Health to Contract Poultry Production." Agricultural and Resource Policy Forum, Auburn University, May. Available at http://www.auburn. edu/~taylocr/topics/poultry/poultryproduction. htm.

Telser, L. 1980. "A Theory of Self-Enforcing Contracts." Journal of Business 53:27-44.

Vukina, T., and P. Leegomonchai. 2006. "Oligopsony Power, Asset Specificity, and Hold-Up: Evidence from the Broiler Industry." American Journal of Agricultural Economics 88: 589605.

Wu, S., and B. Roe. 2006. "AJAE Appendix: Contract Enforcement, Social Efficiency, and Distribution: Some Experimental Evidence." Unpublished manuscript. Available at http:// agecon.lib.umn.edu/.

(1) Some states have sought to improve transparency of performance measures. For example, Georgia passed HB 648 in 2004. This bill requires processors to provide "any statistical information and data used to determine compensation paid" at a grower's request. The bill also allows growers to be present when birds and feed are being weighed.

(2) In personal communication, senior members of the California Processing Tomato Growers Association suggested to one of the authors that, in any given year, processors will re-sign 90% of growers from the previous year. Thus, repeat trading is quite common. As another example, Hamilton (2001, pp. 3-7) suggests that, while most broiler contracts are flock-to-flock, some contracts implicitly make relationships continuous until terminated. This suggests that the same parties may trade repeatedly across many flocks. Moreover, the fact that growers must make relationship specific investments in long-term assets such as new buildings that are specific to a particular processor implies an expectation that trading will persist over multiple flocks or seasons. Additionally, the fact that there is physical specificity in the broiler industry (Vukina and Leegomonchai 2006), which limits the number of trading partners in specific geographical regions, suggests that a certain amount of repeat trading with the same partner may be unavoidable.

(3) Under the Uniform Commercial Code, oral agreements are considered enforceable contracts if growers are deemed to be "merchants" (Hamilton 1995).

(4) One might question whether the use of students rather than farmers weakens our results. We regard the use of students as a strength because growers' attitudes toward contracting issues may be politicized by recent controversies about contracts. Most university students are unfamiliar with these political entanglements and may respond more neutrally.

(5) Firms often establish "private trades" by contacting specific suppliers with whom they have good relationships in order to avoid costly public solicitations when the desired supplier is already known.

(6) Camerer and Fehr (2006) report that in similar types of experiments with excess sellers, seller competition tends to drive buyers to make less favorable offers to sellers.

(7) For example, C preceded RC1 as often as RC1 preceded C across the thirteen experiments.

(8) Consistent with standard experimental economics practice, one session was randomly chosen via a public roll of a die to be the paying session. Subjects were informed that actual earnings depend upon the rules of the game and the participant's and other participants' actions. Average earnings were $23 per subject and ranged from $13 to $41.

(9) We cluster on experiments rather than buyers, sellers, and/or sessions because, due to group composition, all composite errors within an experiment might be correlated, not just observations associated with buyers, sellers, and sessions.

(10) Our probit is similar to BFF's probit in their Table III. How ever, our results differ dramatically because BFF estimated contract renewal where the dependent variable is the probability of private contracting with the same seller as last period. We estimate the probability of private regardless of whether it is with the same seller. This is because we are interested in what explains private trading, not just what explains contract renewal. Moreover, BFF use data only from RC1 sessions whereas we pool C, RC1 and RC2 data.

Steven Y. Wu is assistant professor and Brian Roe is associate professor, Department of Agricultural, Environmental, and Development Economics, The Ohio State University. This material is based upon work supported by the USDA/NRICGP program for Markets and Trade or Rural Development, Sponsor ID (40040100): 2005-35400-15963, Award No. GRT00001044/60003271, and the Ohio Agricultural Research and Development Center. Table 1. Summary of Treatments, Sessions, Rounds, Participants, and Trades

C RC1 RC2 1. Third-party enforcement of price? Yes Yes No 2. Third-party enforcement of quality? Yes No No 3. Total no. of sessions 13 7 6 4. No. of buyers per session (a) 5 5 5 5. No. of sellers per session (a) 7 7 7 6. Rounds per session 15 15 15 7. No. of possible trades per session 75 75 75

(Row 4 x Row 6) 8. Total no. of possible trades across all 975 525 450

sessions (Row 3 x Row 7) 9. Total no. of actual trades executed by 942 512 449

subjects (% of total possible) (97%) (98%) (99.8%) (a) Note that there were two sessions in each experiment. Thus, each group of five buyers and seven sellers (twelve subjects) who were recruited for an experiment actually participated in two sessions. Hence, there were a total of 156 subjects (twelve subjects per experiment) who participated in the twenty-six sessions. Table 2. Discretionary Adjustments Made by Buyers According to Performance in Treatment RC1

Discretionary Price Adjustment by Buyer

Contract Type Reward No Adjust. Q > Q * Public 2% 0.7%

Private 4.4% 0.9% Q = Q * Public 3.3% 12.9%

Private 1.1% 13.8% Q < Q * Public 0.4% 5.3%

Private 0.2% 2.5% Overall Public 5.7% 18.9%

Private 5.7% 17.2%

Total 11.4% 36.1%

Discretionary Price Adjustment by Buyer

Deduct Overall Q > Q * 3.3% 6.0%

1.3% 6.6% Q = Q * 11.4% 27.6%

1.6% 16.5% Q < Q * 29.4% 35.2%

5.3% 8% Overall 44.1% 68.9%

8.2% 31.1%

52.3% 100% Note: O* = quality supplier agreed to deliver in the contract. Results are reported as % of total number of transactions. Total number of trades is 449. Table 3. Summary Statistics-Evidence of Opportunistic Behavior

C RC1 RC2 1. % of public trades where Q < Q * in -- 81% 39% period following termination. 2. % of public trades where Q < Q * in -- 69% 52% all other periods. 3. % of private trades where Q < Q * in -- 73% 43% period following termination. 4. % of private trades where Q < Q * in -- 32% 23% all other periods. 5. Avg. promised profit to seller in 14.4 13.5 21.5 public trading. 6. Actual mean seller profit in public 14.4 19.8 12.9 trading. 7. Avg. promised profit to seller in 26.5 27.7 26.9 private trading. 8. Actual mean seller profit in private 26.5 31 22.9 trading. 9. % public trades where seller profit 6% 3% 24% fell below reservation. 10. % private trades where seller profit 1.9% 0.7% 7.9% fell below reservation. Note: Promised profit is the amount the seller earns if both parties stick to the terms of the contract. Table 4. First Stage Regression: Probability of Private Trade Regressors Coefficients RC1 0.96 **

(0.21) RC2 0.24

(0.17) Reservation -0.06

(0.05) Lagged positive surprise -0.025

(Max{Q - E(Q), 0}) -0.05 Lagged negative surprise 0.020

(Min {Q - E(Q), 0}) (0.03) Lagged quality 0.06

(0.04) Length of private relationship 0.51 **

prior to current period (in (0.13)

periods) Lagged price deviation (P * - P) -0.004

(0.008) Period -0.001

(0.05) Period2 -0.001