Shirking air traffic controllers

It sure sounds like it would improve safety. Everybody knows that being an air traffic controller can be pretty stressful. All those planes up there, have to keep them from bumping into each other. So it just makes sense that you'd want to provide stress leave for those controllers who are obviously suffering, right? What could go wrong?

Doug Allen, currently visiting at Canterbury's Economics department as Erskine Fellow, points me to this classic 1982 piece in the American Economic Review.

The Federal Employees' Compensation Act (FECA), which provides income insurance for government workers injured or disabled on the job, changed in 1974. Employees with at least one dependent would receive 75% of their base salary as tax-free compensation if out on leave. Because air traffic controllers earned a lot, their net take-home pay was higher if they were out on disability. On-the-job stress is pretty hard to disprove.

FECA amendments in 1974 let employees submit evidence from their private clinical psychologists as evidence of disability; prior to that, the government selected the doctor. And, unlike things like "leg cut off" where most doctors would agree, it's harder to verify whether any doctor's diagnosis of stress is all that valid. So Staten and Umbeck predicted that we'd see the most action post-FECA change in claims for non-verifiable things like stress.

They also note that OWCP examiners, who check the veracity of claims, were told to check for specific indicators of on-the-job stress. Deterioration in job performance over time counts. And everybody's monitoring air traffic controller performance all the time, making sure that planes don't get too close together.

So what happens? After the rule change, a bunch of air traffic controllers figure out that they can earn more by demonstrating on-the-job stress, choose times of the day when it's safest to breach minimum separation between aircraft so there aren't actual collisions, and encourage planes to get just close enough to each other to trigger their near mid-air collision sensors. Controllers with more experience have much stronger incentive to do it, as the compensation is based on the employee's salary, and there's a reasonably steep salary progression. Before the law change, inexperienced staff and those with more than 10 years' service (actual burnout) reported the most 'system errors' (violation of separation rules). After the law change, those with 5-10 years' service started experiencing the most system errors.

They conclude:

The steady increase in controller compensation awards and retirements during the 1970's has been cited recently as evidence of stress in the occupation. However, it seems clear that this is in large part a product of making compensation and retirement less costly to obtain. Concern has also been expressed over the relation between job pressures placed on controllers and aviation safety. If one is willing to take reported System Errors as a measure of workforce efficiency, it is clear that measured efficiency deteriorated following the FECA amendments, but not because of traffic increases or equipment failures. We have argued that in attempting to formalize the criteria for adjudicating controller psychological claims, OWCP created an incentive for controllers to demonstrate deteriorating performance. 

We have attempted in this paper to constrain theory of shirking behavior to generate refutable propositions. The methods described in the paper can be employed in the investigation of any type of contract. However, the nature of the FECA provisions and the air traffic control profession provided the elements essential for demonstrating the effectiveness of this approach. That we chose air traffic controllers in particular is, of course, no reflection on the integrity of the profession. The maximization postulate upon which micro theory relies leaves no profession exempt from marginal responses to gains from shirking. We offer this study as a demonstration that economics as a behavioral science can provide useful insights into behavioral patterns likely to result from incentives to shirk.

In later work, they show that air traffic controllers became a lot more likely to do this kind of thing after finding a doctor who's willing to certify stuff as being stress-related.

Regulations should always pay some heed to that people respond to incentives.