5. Labor Law
American labor law is not often thought of in terms of default rules, but defaults are very important in this field. More precisely, the default provision under the National Labor Relations Act is that workers are not unionized. If a group of workers mount a unionization drive and a majority of the workers (or, in some cases, a majority of a subset of the nonmanagement workers) within a workplace vote to unionize, then a union will be certified, and it will be authorized to bargain collectively on behalf of all the workers as a whole. (129) Union certification efforts can be cumbersome, expensive, and contentious. At the same time, it seems plausible that American law's chosen default rule is an appropriate one on majoritarian grounds--most American workers are nonunionized and have been for quite some time. (130)
Psychological studies have shown that personality characteristics correlate strongly with membership in a voluntary union. In particular, the Big Five traits of extraversion and neuroticism both predict union membership, and the interaction of these two traits predicts union membership very strongly. (131) Big Five personality characteristics also predict which industries individuals are likely to be drawn to and which individuals are most likely to thrive and retain their jobs in particular industries. For example, nurses who report high levels of neuroticism are much more likely to experience emotional exhaustion and burnout, which may cause them to leave nursing, whereas nurses with high levels of extraversion are likely to avoid burnout. (132) And while politicians score very high on extraversion and openness, bureaucrats do not. (133) Managers and sales representatives show high levels of extraversion, (134) and the unemployed commonly evince high levels of neuroticism. (135)
This kind of data suggests a radical possibility, which is that certain workplaces or industries, especially those containing high numbers of very extraverted and neurotic individuals, might be deemed unionized by default. (136) Given the underrepresentation of highly neurotic individuals in the workforce, the nonunionized default plausibly makes sense for most workplaces.
At this point, we want to identify this kind of workplace profiling to determine the default rule as a theoretical possibility rather than as something we are advocating. Correlation and causation are distinct, and the factors that drive union membership continue to be debated. (137) For example, it is plausible that extraversion and neuroticism explain the success of unionization campaigns rather than workers' underlying preference for union membership. It is even conceivable that correlation runs in the opposite direction and that participation in a union makes workers more extraverted and neurotic. We would need to get a fuller sense of these causal variables before offering prescriptions for labor law. That said, depending on the results of future research, a prounionization default rule could be appropriate in some contexts.
Big Data Guinea Pigs
Countries with enormous populations ought to take advantage of economies of scale. In this case, that would mean forgoing the careful monitoring of all their citizens' choices and perhaps sidestepping some of the problems from inefficient social norms in the process. We therefore propose that American law ask 1 million guinea-pig residents to make active choices about their preferences, which the law would then data mine to identify the ways in which the other 314 million individual Americans are similar to the 1 million guinea pigs. (138) The law would provide modest compensation to the guinea pigs for the costs they incurred in the process. The guinea pigs' active choices would then become the personalized default choices for the people most similar to them across a variety of observable metrics. These surveys could be conducted through a governmental agency, like the Census Bureau or Consumer Financial Protection Bureau, or through an industry consortium.
A great deal of contract law scholarship concerns the extent to which consumers are rushed or inattentive and pay little attention to contract terms as a result. (139) Yet, if one in every 314 people is a compensated contract-law guinea pig, (140) then the law might reasonably devote substantial resources to making sure that these guinea pigs are very well informed and have adequate time to consider the contractual options and associated tradeoffs. The guinea pigs would spend time reading the fine print so others do not have to. Once an entity--presumably a governmental agency--has assembled a large dataset to track the choices of these guinea pigs, the entity can identify behavioral patterns and facilitate efforts by firms to give each consumer a set of default contractual terms that mimic those chosen by the guinea pigs with the personalities and attributes most similar to hers. We envision a "clustering" approach for identifying coherent groups of people to whom particular personalized default rules will apply. (141) Only the choices made by the guinea pigs prior to the time the contract at issue was executed would matter.
In a recent article, Ayres and Schwartz propose that firms be required to survey their customers about whether particular terms in a contract are consistent with their expectations. (142) Terms that surprised many consumers or that had surprising and very bad consequences for a smaller number of consumers would need to be set apart in a special box designed to prompt consumers to pay more attention to such terms. (143) A consumer-voting mechanism could ensure that the most surprising or most disadvantageous terms appear most prominently in the special box. (144) Ayres and Schwartz's proposal somewhat resembles our approaches to defaults and disclosure. (145) But whereas Ayres and Schwartz propose an impersonal approach to determining which contract terms are problematic, our approach is personalized. It recognizes that different terms will be problematic to different types of people, so the "boxes" or defaults that different sorts of people are shown should differ systematically.
Our "sampling" strategy mirrors the sorts of extrapolations that demographers and survey researchers routinely use in their work. (146) And the private sector already uses such strategies for predictive purposes. For example, Netflix's Cinematch algorithm for movie ratings (a) analyzes the one- to five-star ratings provided by its users after they have seen a movie; (b) matches each user's ratings with the ratings of other users in the Netflix database; and (c) uses these similarity scores to predict how much users will like particular movies. Users can then employ these predictions in deciding which films to rent or stream. (147) The more films a user rates, the better the algorithm can personalize the user's movie recommendations and the recommendations of similar Netflix customers.
Of course, rating each movie on Netflix entails an active choice. Many Netflix users do not bother to evaluate movies they have seen, perhaps because it is time consuming. (148) And many Netflix users similarly do not use the "taste preferences" features, which permit users to specify how often they watch movies that can be characterized as "absurd," "bawdy," "cerebral," "dark," etc. (149) One of the potential benefits of personalized default rules in a world of Big Data is that much of the data used to generate similarity scores and personalized defaults will be generated automatically, without requiring the user to do anything. It is almost tantamount to Netflix monitoring how many times a viewer laughed during a comedy, cried during a tragedy, or gasped during a horror flick.
A more modest alternative to using guinea pigs would be to generate information necessary for personalizing default rules by asking individuals about their general preferences, characteristics, and traits, as well as about their past behaviors, and using this information to tailor default rules for them. An agency might distribute questionnaires to consumers, explaining that the answers will be used for personalizing default rules in their interactions with merchants. We predict that many consumers will answer the questionnaires, which should not be too intrusive, with the understanding that their answers would facilitate their receiving deals better adapted to their true preferences. The gist of the approach is to use information culled from a survey to modify defaults that a consumer will encounter. This blanket approach to personalizing default rules seems far more efficient than selective modifications of contractual boilerplate on a transaction-by-transaction basis. We propose that individuals should be able to see the "profile" constructed for them and change this profile if it does not fit their true preferences. (150)
In any event, in modern, high-stakes transactions, it is becoming increasingly common for sellers to have information about the consumers they are dealing with, which enables them to decide on pricing and service quality, pinpoint potentially fraudulent transactions, and evaluate the effectiveness of their marketing strategies. (151) As the information age proceeds, it will be reasonable to assume that sellers "know their customers" and either already are or can easily become familiar with the personalized default rules that correspond to particular customers.
Consumers are less likely to have this sort of information about individual firms' propensities, although in the case of large national firms or local firms with extensive Yelp profiles, the information asymmetries may be less pronounced. Imposing on consumers a burden to "know their sellers" is less justifiable, particularly when they are dealing with small-scale sellers in non-repeat-play environments. (152)
POSSIBLE OBJECTIONS AND LIMITATIONS
Part II articulated a rather bold vision of...