Organizational Economics Meeting

November 18-19, 2011
Robert S. Gibbons of MIT and Elizabeth Martinez of Massachusetts General Hospital (MGH), Organizers

Ian Larkin and Desmond Ang, Harvard Business School; Matthew Chao, California Institute of Technology; and Tina Wu, New York University

How (and How Much) Are Physicians Influenced By Pharmaceutical Sales Calls? A Large-Scale Quasi Experiment Using Archival Data

Larkin, Ang, Chao, and Wu investigate the mechanisms by which pharmaceutical sales people influence the prescribing behavior of physicians. They use a detailed database of physician prescribing behavior and pharmaceutical marketing behavior, and quasi-exogenous changes to policies at some university hospitals around permitted interactions between pharmaceutical salespeople and physicians. They find that limiting or banning personal gifts to doctors (such as meals) leads to large reductions in prescribing heavily marketed drugs, while limiting educational gifts or enacting policies that restrict access to doctors by sales people has no incremental effect. They also find that banning product samples leads to increases in prescriptions for marketed drugs, suggesting that product samples substitute rather than create demand for branded, heavily marketed drugs. These results suggest that sales people primarily influence physicians by inducing reciprocity, not by providing information.


Joseph Doyle and Jonathan Gruber, MIT and NBER; John Graves, Vanderbilt University; and Samuel Kleiner, Cornell University and NBER

Do More Expensive Hospitals Deliver Better Care? Evidence from Ambulance Referral Patterns

This paper is part of a project that aims to measure performance differences across hospitals. A main limitation is that patients sort to hospitals, which can confound comparisons. Doyle, Gruber, Graves, and Kleiner use the idea that the ambulance company assigned to an emergency call is effectively randomly assigned within ZIP codes. Typical variation comes from mutual aid agreements if the primary ambulance company is busy or rotational assignment of companies to 911 calls. The second idea compares patients on either side of ambulance dispatch-area boundaries. The authors show that these ambulance companies differ in where they take patients, despite similar patient characteristics and locations. They also find that high-cost hospitals - hospitals that provide more treatments to patients - have significantly lower mortality. These hospitals are more likely to be at the technological frontier, and are more likely to be teaching hospitals. Conditional on technology adoption or teaching status, higher costs continue to be related to lower mortality.


Katherine Kellogg, MIT
Operating Room: Relational Spaces and Micro-institutional Change in Surgery

One of the great paradoxes of institutional change is that even when top managers in organizations provide support for change in response to new regulation, the employees whom new programs are designed to benefit often do not use them. This 15-month ethnographic study of two hospitals responding to new regulation demonstrates that using these programs may require subordinate employees to challenge middle managers with opposing interests. argues that relational spaces-areas of isolation, interaction, and inclusion that allow middle-manager reformers and subordinate employees to develop a cross-position collective for change-are critical to the change process. These findings have implications for research on institutional change and social movements.


Anna Levine Taub, Northeastern University; Anton Kolotilin, MIT; Robert S. Gibbons; and Ernst R. Berndt, MIT and NBER

The Heterogeneity of Concentrated Prescribing Behavior: Theory and Evidence from Antipsychotics (NBER Working Paper No. 16823)

Physicians prescribing drugs for patients with schizophrenia and related conditions are remarkably concentrated in their choice among ten older typical and six newer atypical anti-psychotic drugs. In 2007, the single anti-psychotic drug most prescribed by an average physician accounted for 59 percent of all anti-psychotic prescriptions written by that physician. Moreover, among physicians who concentrate their prescriptions on one or a few drugs, different physicians concentrate on distinct drugs. Levine Taub, Kolotilin, Gibbons, and Berndt construct a model of physician learning-by-doing that generates several hypotheses amenable to empirical analyses. Using 2007 annual anti-psychotic prescribing data on 15,037 physicians from IMS Health, the authors examine these predictions empirically. While prescribing behavior is generally quite concentrated (varying somewhat depending on the concentration measure being used), prescribers with greater prescription volumes, with training in psychiatry, men, and those not approaching retirement age all tend to have less concentrated prescribing patterns, particularly among the newer generation of anti-psychotics.


Amitabh Chandra, Harvard University and NBER, and Douglas O. Staiger, Dartmouth College and NBER
Expertise, Underuse, and Overuse in Healthcare

The reasons for variation in treatment rates across hospitals serving similar patient populations are not well understood. Differences in the use of a treatment across hospitals may be attributable to greater benefits of treatment in some hospitals (expertise), withholding of beneficial treatment in some hospitals (underuse), or providing harmful treatment in other hospitals (overuse). Chandra and Staiger develop an empirical model that can distinguish between these explanations, based on a behavioral model in which hospitals choose to treat patients if the benefit from treatment exceeds a hospital-specific threshold. Expertise, underuse, and overuse are identified based on differences across hospitals in both their treatment rates and the treatment effect on patient survival. Using data on heart attack treatments, they find that expertise varies considerably across hospitals, but a substantial amount of variation in treatment is due to overuse.

James M. Malcomson, Oxford University
Relational Incentive Contracts with Persistent Private Information

Malcomson investigates relational incentive contracts with private information about agent types drawn from an interval and persistent over time. For a sufficiently productive relationship, a pooling contract exists in which all agent types continuing the relationship choose the same action. Necessary and sufficient conditions are given for some separation to be feasible; the parties can then do better than with full pooling. When future actions are optimal, however, separation of all types is never possible; the finest separation achievable is into partitions containing a non-degenerate interval of types. Separation always involves lower output initially than after separation has occurred.


Maria Guadalupe, Columbia University and NBER, and Hongyi Li and Julie Wulf, Harvard University
Functional Centralization and the Division of Labor in Management

In contrast to the widely-held view that flatter hierarchies are associated with the delegation of decisions, Guadalupe, Li, and Wulf show that the trend towards flattening in large U.S. firms since the mid-1980s has been accompanied by increased centralization of activities at the top of the organization. In particular, the number of functional managers (for example, Chief Financial Officer or Chief Marketing Officer) reporting directly to the CEO has increased relative to the number of general managers. Using panel data on senior management positions in large U.S. firms (1986-99) and exploiting variation within firms over time and across position types, the authors document how the centralization of functional activities relates to the firm's IT investments and business diversification. Centralization increases with IT intensity for "administrative" functions (for example, finance, law, HR); yet, the same relationship only holds for "product" functions (for example, marketing or R and D) in firms with related businesses. Firms in related businesses are more likely to centralize product functions, but the authors find no relationship with administrative functions. They also document how pay changes with firm organizational structure for the different types of managers. These findings suggest that the nature of the information associated with the different functions and the degree of business diversification are important forces driving the centralization decision.


Iwan Barankay, University of Pennsylvania
Gender Differences in Productivity Responses to Performance Rankings: Evidence from a Randomized Workplace Experimen

Performance rankings are a pervasive feature of life. Behavioral theories suggest that knowing one's rank may shape effort directly because of its effect on self-image. In a randomized control trial with full time employees (n=527) for whom rankings convey no direct financial benefits, Barankay studies the long-run productivity consequences of privately informing them about their performance rank. First, perhaps surprisingly, showing employees their rank actually reduces their performance and this result is driven by the demoralizing effect of being informed of a worse than expected rank. Second, the treatment effect is gender specific, as only men but not women reduce their performance which, upon further analysis, is because women are less heedful of their rank than men. Rankings are a plausible candidate for a behavioral incentive scheme, as they speak to well established theories of interpersonal comparisons and self-image. Yet this study documents their detrimental effect on performance. This paves the way for further research to improve the design of rank feedback to exploit rank preferences in a way that it raises performance.


Edward P. Lazear and Kathryn L. Shaw, Stanford University and NBER, and Christopher T. Stanton, University of Utah
The Value of Bosses

As more productivity data become available, it is possible to examine the effects of people and practices on productivity. Arguably, the most important relationship in the firm is between worker and supervisor. The supervisor hires and fires, assigns work, instructs, motivates and rewards workers. Models of incentives and productivity build at least some subset of these functions in explicitly, but because of lack of data, little work exists that demonstrates the importance of bosses and the channels through which the productivity enhancing effects operate. Using a unique company-based dataset, Lazear, Shaw, and Stanton estimate supervisor effects and find them to be large for technology-based services workers. The three most important findings are: The only "peer" that matters is the boss. In this environment, peers have little or no effect on output, but bosses affect workers significantly. Second, there is substantial variation in boss quality, measured by their effect on worker productivity. Replacing a boss who is in the lower 10 percent of boss quality with one who is in the upper 10 pecent of boss quality increases a team's total output by about the same amount as would adding one worker to a nine member team. Third, the marginal product of a boss is about 60 percent greater than that of a worker, commensurate with the ratio of their wages. Additionally, good bosses should be sorted to the best workers: although good bosses increase the productivity of both good and bad workers, they increase it by more for the firm's top performers.


Yanhui Wu, University of Southern California
Authority, Incentives and Performance: Theory and Evidence from a Chinese Newspaper

Wu examines the incentive theory of authority using personnel data from about 200 journalists in a Chinese newspaper. Theory suggests that restricting workers' authority can alleviate the multi-tasking problem, but may depress their initiative. Relying on an unexpected organizational reform from delegating to centralizing editorial decision rights in some divisions of the newspaper, Wu finds two main results: 1) centralizing authority improves the reporters' performance of their journalistic task, in terms of the newspaper' internal assessment and the external measures of news content, while reducing their activities for private gain; 2) centralizing authority decreases the number of articles originated by division editors, a measure of their initiative. Consistent with the theory, the results shed light on the central trade-off between better control over opportunistic behavior and depressing initiative in a multi-tasking setting and in a multi-layered hierarchy.


Silke J. Forbes, University of California, San Diego; Mara Lederman, University of Toronto; and Trevor V E. Tombe, Wilfrid Laurier University
Quality Disclosure and Gaming: Do Employee Incentives Matter?

Forbes, Lederman, and Tombe investigate gaming of a public disclosure program and, in particular, whether gaming depends on the incentives provided to the employees who are most likely to carry out the gaming. They do this in the context of the government-mandated disclosure of airline on-time performance. While this program collects data on the actual minutes of delay incurred on each flight, it ranks airlines based only on the fraction of their flights that arrive 15 or more minutes late. This creates incentives for airlines to game the program by reducing delays on specifically those flights they expect to arrive with about 15 minutes of delay. In addition, several airlines have introduced employee incentive programs based explicitly on the airline's performance in the government program. The empirical analysis finds no evidence of gaming by airlines without incentive programs, or with incentive programs with targets that are unrealistically hard to achieve. On the other hand, there is strong evidence of gaming by airlines that implemented incentive programs with targets that could be - and were - achieved. Specifically, their flights that are predicted to arrive with about 15 minutes of delay have significantly shorter taxi-in times and are significantly more likely to arrive exactly one minute sooner than predicted. These findings highlight that gaming of a disclosure program will depend not only on the design of the program but also on if and how the measured quality dimensions can be manipulated, and whether those who are in a position to manipulate them have incentives to do so.


Luigi Guiso, EUI, and Luigi Zingales, University of Chicago and NBER
The Value of Social Networks in Bank Lending

Guiso and Zingales link a directory of the members of a social club in Italy to information from the credit register to study the effect of social networks on bank lending. They find that a bank whose officers are part of the same social network as officers of a firm has two and a half times the probability of extending a line of credit to the firm than does a random bank. As a new bank enters the club, chances increase that a new relation with a firm in the club is set up; as the bank exits the club, existing relations are discontinued. The effect of club membership is stronger for firms that are more likely to suffer from imperfections in credit markets, either because they are too small or because they are illiquid. Effects of membership are additive, and the total amount of credit a firm gets is larger the higher the number of banks with representatives in the same club. This analysis indicates that social interactions facilitate market relations. Yet, the establishment of social relationships is itself subject to frictions, which may have repercussions on the functioning of the market for credit.