Abstract
An earlier version of this paper was presented at the 2003 Joint Statistical Meetings and the George Mason Summer Institute for the Preservation of the History of Economics in Economics. We have benefitted from the comments of William Seltzer and Margo Anderson and, especially, this journal's referee. The paper began with a conversation between Levy and Paul David at the Economics of Science Conference at Notre Dame in April 1997 and was greatly stimulated by Perci Diaconis's as yet unpublished 1998 IMS Lecture. Preliminary versions were presented at the George Washington University Economics Seminar, the 1999 Canadian Law and Economics Association meetings, the George Mason Statistics Department Seminar, and the University of Alberta Econometrics Seminar. We have benefitted from comments from James Buchanan, Adolf Buse, Roger Congleton, Tyler Cowen, Mark Crain, Don Gantz, Jim Gentle, Bruce Kobayashi, David Meiselman, John Miller, David Ribar, Robert Tollison, and other participants. The command files to replicate the Monte Carlo study are available upon request.
Similar content being viewed by others
Notes
Gorlin [1999], an 1100-page compendium of “codes of professional responsibility,” indexes professions alphabetically. The list refers to: chiropractic (1067), computing (1068), counseling (1070), dental hygiene (1073), dentistry (1074), direct marketing (1076), dispute resolution (1078), and engineering (1079).
We have developed a technical working definition of sympathy in Peart and Levy [2005]. For our purposes here, it is sufficient to suggest that a researcher who is sympathetic with his client's preferences over estimates may be induced to choose and then offer the client an estimate that the client would want to obtain. Reading Gardiner's description of Deming's practice, one can visualize the ideal expert–client relationship as that in a dictator game. With sympathetic agents, however, sharing results even in dictator games. See Camerer and Thaler [1995], Hoffman et al. [1996; 1999], and Sally [2001].
Rational choice estimation is the “evil twin” of exploratory data analysis (EDA). Whereas EDA supposes that a model changes as one's beliefs move to encompass more of what is true [Levy 1999/2000], rational choice estimation starts with a true model and finds what is profitable to believe.
If the preferences are lexicographical, then J is to be viewed as a pseudo-indifference curve and is marked with pseudo-Roman numbers.
See notes 2, 5, 8, and 10 on p 5, note 6 on p 6, and notes 2, 5–6, 7, 8, 9, 12, 13, and 14 on p. 7 of ASA 2000.
We do not pretend that transparency is sufficient for ethical behavior in this context. It leaves aside a wide range of issues, relating to experimentation and ethics, that provided the impetus for Human Subjects Review Boards. Although we largely leave aside these issues here, we note that there is an informal set of procedures dealing with them in experimental economics. See Houser [2008].
The relationship between expected utility and minimax decision theory is subtle. Savage's contribution to the Princeton Robustness study, the estimator LJS, is a minimax estimator varying a theme due to P. J. Huber [Andrews et al. 1972, p. 2C3]. On Huber's original paper, see Savage [1972, p. 291]: “An important nonpersonalistic advance in the central problem of statistical robustness.”
So defined non-transparency is a case of asymmetric information [Akerlof 1970].
In the discussions leading up to the American Statistical Association [2000], great care was taken to distinguish an estimate in which the bias is transparent, as defended by Bayesians, from an estimate in which the bias is not transparent.
Transparency in principle is not the same as transparency in fact. The difficulty that often arises in obtaining data for replication is common knowledge among working econometricians.
The alphas are all 1; β 1 is 10; β 2 is −1; β 3 is 3.
This idea results from a conversation with Paul David.
Judging from 10,000 experiments the bias persists through N=6,400. If the bias were measured in terms of the median of the estimates instead of the mean, it too would persist. The experiments were repeated with all exogenous variables following a uniform distribution between 0 and 1. Because it is not surprising that amount of the bias is acutely sensitive to the distribution of the omitted exogenous variables, these results are not reported.
We have benefitted from a conversation with Arthur Goldberger about the concerns of the Cowles Commission on pseudo-identification of structural equation estimates and with Adolf Buse on the modern discussion of weak identification.
Posner [1999, p. 1488]: “Because trial lawyers are compensated directly or indirectly on the basis of success at trial, their incentives to develop evidence favorable to their client and to find the flaws in the opponent's evidence is very great and, if it is a big money case, their resources for obtaining and contesting evidence will be ample.”
This idea was suggested by John Miller.
An alternative rule might be to define a range in which the models are equally acceptable. Presumably, the “splitting the difference” rule would then prevail.
Ashenfelter and Bloom [1984, p. 112] point out how the conclusions in Crawford [1979] depend critically on this assumption.
Ashenfelter et al. [1992, p. 1427]: “ … risk averse bargainers can mitigate the risks inherent in FOA [final-offer arbitration] by submitting more reasonable offers while CA [conventional arbitration] offers less scope for mitigation of risk. One way to interpret the conventional wisdom that FOA is riskier than CA is that, by preventing the arbitrator from compromising, the middle of the distribution of arbitrator's preferred outcomes is eliminated as potential arbitration awards. This by itself would increase risk. However, FOA also eliminates the tails of the distribution, and this decreases risk.”
This is perhaps not unrelated to the question of why Bayesians randomize posed by de Finetti [1972] [Diaconis 1998]. ASA 2000 is explicit about responsibilities to the wider community; see pp. 2–4, 10.
References
Akerlof, George A . 1970. The Market for “Lemons”. Quarterly Journal of Economics, 84: 488–500.
American Statistical Association. 2000. Ethical Guidelines for Statistical Practice. Alexandria VA: American Statistical Association.
Andrews, D.F., P.J. Bickel, F.R. Hampel, P.J. Huber, W.H. Rogers, and J.W. Tukey . 1972. Robust Estimates of Location. Princeton: Princeton University Press.
Ashenfelter, Orley, and David E. Bloom . 1984. Models of Arbitrator Behavior: Theory and Evidence. American Economic Review, 74: 111–124.
Ashenfelter, Orley, and David E. Bloom, Janet Currie, Henry S. Farber, and Matthew Spiegel . 1992. An Experimental Comparison of Dispute Rates in Alternative Arbitration Systems. Econometrica, 60: 1407–1433.
Camerer, Colin F., and Richard H. Thaler . 1995. Anomalies: Ultimatums, Dictators and Manners. Journal of Economic Perspectives, 9 (2): 209–219.
Crawford, Vincent P . 1979. On Complusory-Arbitration Schemes. Journal of Political Economy, 87: 131–159.
De Finetti, Bruno . 1972. Probability, Induction and Statistics. New York: John Wiley.
Deming, W. Edwards . 1965. Principles of Professional Statistical Practice. The Annals of Mathematical Statistics, 36: 1883–1900.
Denton, Frank . 1985. Data Mining as an Industry. Review of Economics and Statistics, 57: 124–127.
Diaconis, Persi . 1998. Mysticism in Statistics: From the Bible Codes to the Cult of Fisher, Institute of Mathematical Statistics Presidential Lecture, Dallas.
Feigenbaum, Susan, and David M. Levy . 1996. The Technological Obsolescence of Scientific Fraud. Rationality and Society, 8: 261–276.
Freedman, D . 1984. On Bootstrapping Two-Stage Least-Squares Estimates in Stationary Linear Models. Annals of Statistics, 12: 827–842.
Froeb, Luke M., and Bruce H. Kobayashi . 1996. Naive, Biased, yet Bayesian: Can Juries Interpret Selectively Produced Evidence? The Journal of Law, Economics and Organization, 12: 257–276.
Gardiner, John . 1996. What and Where are Statistical Ethics, Presented at the Joint Statistical Meetings. Chicago. Available at http://www.tcnj.edu/~asaethic/asaweb2.html.
Gorlin, Rena A. ed. 1999. Codes of Professional Responsibility. Washington: Bureau of National Affairs.
Hoffman, Elizabeth, Kevin A. McCabe, and Vernon L. Smith . 1996. Social Distance and Other — Regarding Behavior in Dictator Games. American Economic Review, 86 (3): 653–660.
Hoffman, Elizabeth, Kevin A. McCabe, and Vernon L. Smith . 1999. Social Distance and Other — Regarding Behavior in Dictator Games: Reply. American Economic Review, 89 (1): 340–341.
Houser, Daniel . 2008. A Note on Norms in Experimental Economics, Eastern Economic Journal, 34 (1): 126–128.
International Statistical Institute. 1985. Declaration on Professional Ethics. 1985. Available at http://www.cbs.nl/isi/ethics.htm.
Leamer, Edward E . 1983. Let's Take the Con Out of Econometrics. American Economic Review, 73: 31–43.
Levy, David M . 1999/2000. Non-Normality and Exploratory Data Analysis: Problem and Solution. Econometric Theory, 15: 427–428; 16: 296–97.
Peart J. Sandre, and David M. Levy . 2005. A Discipline without Sympathy: From Median Well-Being to Mean Well-Being, Canadian Journal of Economics 38: 937–954.
Phillips, Peter C.B . 1983. Exact Small Sample Theory in the Simultaneous Equations Models, in Handbook of Econometrics, edited by Zvi Griliches and Michael D. Intriligator. Amsterdam, North Holland.
Posner, Richard A . 1999. An Economic Approach to Legal Evidence. Stanford Law Review, 51: 1477–1546.
Sally, David . 2001. On Sympathy and Games. Journal of Economic Behavior & Organization, 44: 1–30.
Savage, Leonard J . 1972. Foundations of Statistics, 2nd revised edition. New York: Dover Publications.
Tullock, Gordon . 1980. Trials on Trial: The Pure Theory of Legal Procedure. New York: Columbia University Press.
Vardeman, Stephen B., and Max D. Morris . 2003. Statistics and Ethics: Some Advice for Young Statisticians. American Statistician, 57: 21–26.
White, Kenneth J . 1997. Shazam: Econometrics Computer Program (8.0). New York: McGraw-Hill.
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Levy, D., Peart, S. Inducing Greater Transparency: Towards the Establishment of Ethical Rules for Econometrics. Eastern Econ J 34, 103–114 (2008). https://doi.org/10.1057/palgrave.eej.9050007
Published:
Issue Date:
DOI: https://doi.org/10.1057/palgrave.eej.9050007