Papers Flashcards
Lazear (2000), Performance Pay and Productivity, American Economic Review.
Discuss their approach and findings, as well as any limitations of their study:
RQ: How sensitive is worker behavior to incentives, and what specific changes in behavior are elicited?
Approach: Several sophisticated models show that workers respond to incentives. However, the models are largely untested because of a lack of data. A newly available dataset allows the authors of this paper to answer how sensitive worker behavior is to incentives. The analysis in the paper is based on data from a large auto glass company (Safelite Glass Corporation), which gradually changed its compensation method for its workforce during 1994 and 1995. The new compensation equals max[W, be-k] where W is the former wage, b is the piece rate, e is the effort level and k is the implicit charge for the job. The main motivation behind changing the compensation method to a piece-rate payment scheme is to increase worker effort. The data used in the paper come from the real world rather than a laboratory, and therefore, they are well suited to the purpose of the study. The effect of changing the compensation method was estimated using worker-level monthly output data.
Findings: A switch to a piece rate pay resulted in a 44% increase in productivity. Some of the effect (22%) results from the average worker producing more because of incentive effects, and some of the effect results from an ability to hire the most productive workers and possibly from a reduction in quits among the highest output workers. Switching to a piece-rate pay also increases the variance in output, as more ambitious workers have less incentive to differentiate themselves when hourly wages are paid.
Limitations: results may not be generalisable to other occupations, as managerial and professional jobs may not be as well suited to piece-work due to less measurable output.
Fryer (2013), Teacher Incentives and Student Achievement: Evidence from New York City Public Schools, Journal of Labor Economics.
Discuss their approach and findings, as well as any limitations of their study:
RQ: What are the potential impacts of teacher incentives on student achievements?
Approach: If teachers lack motivation, financial incentives for student achievement may have a positive impact on teachers’ effort. However, if teachers do not know how to increase student achievement, such incentives may have no impact on teachers’ effort nor the students’ achievements. Such incentives can also have a negative impact on students’ achievements if the teacher cheats or focuses on specific tested objects rather than more general learning. Some also argue that such incentives can decrease a teacher’s intrinsic motivation or lead to harmful competition between teachers. The empirical evidence on the efficiency of teacher incentives is ambiguous. This article describes a school-based randomized trial in over 200 New York City public schools designed to better understand the impact of teacher incentives. In 2008 - 2010, the United Federation of Teachers and the New York City Department of Education implemented a teacher incentive program in about 200 high-need schools. Each school could receive specific rewards if it met an annual performance target, but each school had discretion over their incentive plans. The author combined student-level administrative data and teacher-level human resources data. He used a rather complex regression model to reveal any causal impact of participating in the incentive program. The main outcome variable is an achievement test unique to New York, called ELA. He compared participating- and non-participating schools that would have the same academic outcomes had they both participated in the program.
Findings: The authors found no evidence that teacher incentives increase student performance. Potential explanations are: incentives not high enough, group-based incentives inefficient, no direct relationship between effort and output. If anything, student achievement (in math and reading) declined. Neither does he find evidence that the incentives change student or teacher behavior.
Limitations: (from student presentation) There is a difference between developed and developing countries; in developing countries, incentives have proven to be successful at increasing achievement. The type of exam/qualification/
school system matters. The experiment only targeted the worst performance school in NYC, bias?
DeVaro, J. and Kurtulus, F. A. (2010). An empirical analysis of risk, incentives and the delegation of worker authority, ILR Review.
Discuss their approach and findings, as well as any limitations of their study:
RQ: (1) Is there evidence of a risk-incentives tradeoff as predicted by the principal-agent model? (2) Is there evidence of a positive relationship between incentive pay and the delegation of worker authority? (3) Is there evidence of a positive relationship between risk and authority? (4) Can we find support for the main testable implication of Prendergast’s model, namely that the evidence favoring a risk-incentives tradeoff should strengthen when authority controls are added to the empirical model?
Approach: The authors empirically test Prendergast’s theory that incorporates the delegation of worker authority into the principal-agent model to explain the lack of consistent empirical support for the tradeoff between risks and incentives. Using data from the 1998 WERS, the authors investigate their four research questions. Four points are addressed empirically: (1) In the absence of a control for delegation of worker authority, the sign of the risk-incentive relationship is ambiguous. (2) Incentive pay and delegation of worker authority should be positively correlated. (3) Authority and risk should be positively correlated. (4) When a control for worker authority is included in a regression of incentive pay on risk, the risk coefficient should decrease. The authors estimate a probit model with the probability of performance pay being equal to 1 as the dependent variable.
Findings: The results are affirmative for all four research questions above. The authors believe their results suggest that Prendergast’s theory at least partially explains why previously published empirical literature has failed to uncover a risk-incentives trade- off; at the same time, this does not rule out the possibility that alternative theories may also play a role.
Limitations: We see nothing peculiar to Britain in the fundamental workplace issues Prendergast’s model addresses, and we therefore expect the empirical support for his model in Britain to generalize to data sets from other countries. Furthermore, though our binary measure of incentive pay proved to be quite informative, more detailed information concerning how the intensity of incentive pay varies across organizations would also be valuable.
Nagar (2002), Delegation and incentive compensation, Accounting Review.
Discuss their approach and findings, as well as any limitations of their study:
RQ: How does top management balance delegation of authority and the extent of incentive compensation awarded to lower-level managers?
Approach: Top-managers face two organisational design choices: (1) how much authority to delegate to lower-level managers and (2) how to design incentive compensation to ensure that these managers do not misuse their discretion. The author develops a simultaneous model of retail banks’ delegation and incentive compensation choices for brand managers. He uses two sources of data: (1) Wharton Financial Institutions Center’s 1994 survey of retail banks and (2) the bank call reports filed with the Federal Deposit Insurance Corporation (FIDC) that contain quarterly balance sheets, income statements, and other financial data used to monitor banks’ conditions. He uses a simultaneous equation model with delegation as the dependent variable, and independent variables as incentives, growth and size.
Findings: The model results indicate that high growth, volatile and innovative banks delegate more authority to managers. In turn, managers with more authority receive more incentive based pay. However, incentive pay does not affect the firms’ delegation choices.
Limitations: No empirical study can simultaneously model all organisational design choices. Some design variables such as delegation and managers’ risk aversion are measured with error. Lastly, the study uses observations from one industry polled at the same time, making cross-sectional dependencies a severe concern.
De Varo and Prasad (2015). The relationship between delegation and incentives across occupations: Evidence and theory. The Journal of Industrial Economics.
Discuss their approach and findings, as well as any limitations of their study:
RQ: how does the relationship between incentive pay and delegation vary across occupations?
Approach: data drawn from both the management and worker questionnaires in the 1998 British Workplace Employee Relations Survey. The data are a nationally representative stratified random sample. Variables: Complex = 1 if the establishment’s largest occupational group is “Professional occupations” or “Technical, scientific occupations”. Incentive pay = 1 if any employees received payments / dividends from individual performance-related shames. Delegation = 1 if the worker responds “a lot” on the amount of influence he/she has about the range of tasks in his/her job. They estimated the coefficients using a probit model.
Findings: The effect of delegation turns negative when controlling for complex occupation. Thus, they find a negative relationship between incentive pay and delegation in complex jobs (cardiac surgeons), but a positive relationship between incentives and delegation in simple jobs.
Limitations: Any binary cut of occupations into broad categories such as “complex” and “simple” is arbitrary. Classified as using incentive pay even if only one worker received such pay. Model omits interaction between industry and delegation.
Elsaid, Davidson, Benson (2009), CEO compensation structure following succession: Evidence of optimal incentives with career concerns, The Quarterly Review of Economics and Finance.
This is in the sample exam, so might be less important.
Discuss their approach and findings, as well as any limitations of their study:
RQ: optimal incentives and career concerns for newly hired CEOs
Approach: using a sample of 169 succession announcements between 1991 and 2005, they test the Gibbons and Murphy (1992) model by comparing succession announcement period abnormal returns to the compensation packages of successor CEOs.
Findings: find that abnormal returns are negatively related to the percentage of performance-based pay of newly hired CEOs when companies announce CEO successions (consistent with G & M’s model). The stock market reacts less positively to contracts with considerable performance-based pay for new CEOs whose career concerns should be high. It may be better to allow newly hired CEOs to be paid in human capital increases from the managerial labor market than to have their pay too closely tied to performance.
Limitations: (…)
Koch, Morgenstern, Raab (2009), Career concerns incentives: An experimental test, Journal of Economic Behavior & Organization.
Denne var litt vanskelig å forstå…
Discuss their approach and findings, as well as any limitations of their study:
“RQ”: a laboratory experiment testing the empirical viability of the underlying signal jamming mechanism in the career concerns model.
Approach: First, they test whether the qualitative predictions on signal jamming from the career concerns model hold up in the laboratory. Then, they elicit subjects’ first- and second-order beliefs, and investigate whether players correctly conjecture each others’ actions and beliefs and engage in behaviour consistent with this.
Findings: Career concerns are effective in providing effort incentives (signal jamming successfully creates incentives on the labor supply side). Find evidence that subjects adjust their beliefs in the direction predicted by theory (agent discounts signals from principal).
Limitations: Lab experiments cannot capture the entire complexity of the reality and people’s real decision making (…)
Konings, J. and S. Vanormelingen (2015), The impact of training on productivity and wages: firm-level evidence. Review of Economics and Statistics.
Discuss their approach and findings, as well as any limitations of their study:
RQ: How does on-the-job training impact productivity and wages?
Approach: Use firm-level longitudinal data with information on measures of training. For each firm: the number of employees who received some kind of formal training, the training costs and the hours spent on training between 1997 and 2006. Apply and extend the control function approach for estimating production functions and wage equations. They estimate coefficients of the production function and the wage equation by running a regression model.
Findings: Productivity premium of a trained worker is substantially higher compared to the wage premium. An increase in the share of trained workers by 10 percentage points is associated with 1.7% to 3.2% higher productivity. The average wage increases only by 1% to 1.7% in response to the same increase in training. The marginal product of a trained worker receiving an average amount of training hours is 27.9% higher than that of an untrained worker, while the worker’s wage is only 16.1% higher.
Limitations: restricted data set (location, time, private sector), no information on whether training is general or firm-specific, upward biased estimates of the wage premiums (but consistent with results).
Card, D., Mas, A., Moretti, E., & Saez, E. (2012). Inequality at work: The effect of peer salaries on job satisfaction. American Economic Review.
Discuss their approach and findings, as well as any limitations of their study:
RQ: The effect of disclosing peer’s salaries on workers’ job satisfaction and job search intentions.
Approach: A randomly chosen subset of employees at the University of California (the treatment group) were informed about a new website listing the pay of their coworkers. All employees were surveyed on their job satisfaction. The employees’ job satisfaction was then compared to the satisfaction of a control group who had not seen the website.
Findings:
First of all, workers are interested in coworkers’ pay.
Pay below median: not satisfied and relatively high likelihood of looking for a new job.
Pay above median: no change in satisfaction or likelihood of looking for a new job.
Job satisfaction depends on relative pay comparisons, and the relationship between job satisfaction and pay is not linear. Negative comparisons matter more than positive comparisons. Employers have an incentive to maintain pay secrecy, since the costs of revealing information for lower-paid employees exceed the benefits of revealing information for their high-wage peers.
Limitations: Survey information is limited to self-reported outcomes, raising the question as to whether the effects of the information treatment translated into changes in observable economic behaviour (this limitation is accounted for).
