Tuesday, September 4, 2007

Safety

Peter's summing up of the standard neo-classical approach to job safety is exactly right. There are some exceptions. Robert Frank, for instance. Frank, whose work I admire, is a little bit like the proverbial carpenter who, having only a hammer, thinks everything is a nail. In his case, the hammer is the idea that people care not just about income, but also relative income. In *Choosing The Right Pond* he uses his hammer on the work-place safety issue. Here is a little example based on his ideas that I use in the Principles course. (By the way, I teach economics at Bowling Green State University in Ohio: hello fellow bloggers!).

Suppose there are two types of jobs, safe and risky, and 2 workers. Safe jobs have a safety index of 2 and pay $20,000, while risky jobs have a safety index of 1 and pay $30,000. Further suppose workers utility is the product of three factors: income measured in thousands of dollars, safety( measured by the index), and relative income. Now we have a standard prisoner's dilemma. If you take the safe job, taking the safe job as well gives me utility of (20)(2)(1) = 40. Taking the risky job gives me (30)(1)(3/2) = 45, so I take the risky job. If you take the risky job, I get (20)(2)(2/3) = 26.67 if I take the safe job and (30)(1)(1) if I take the risky job, so I take the risky job in this case too. Each of us does better choosing the risky job whatever the other does, but when we choose the risky job we are worse off, with utility of 30 each, than had we both taken the safe job and gotten utility of 40 each.

Making safety level 2 mandatory makes both workers better off and has no effect on employers (the pay differential is assumed to reflect the cost of making the workplace safer) and is thus a Pareto-improvement. Also note that if we were to infer from the fact that workers reject the safe job (without regulation) that the added safety is worth less than $10, 000, and go on to use this as input into an estimate of the value of life, we would be seriously wrong: deciding collectively, workers would be willing to pay up to $15,000 each for the safer job, since an income level of 15 and safety level 2 gives the same utility, 30, as the risky job when relative income isn't a factor.

The neat thing about Frank's hammer is that the result is consistent with perfectly competitive labor markets and no lack of knowledge on the workers' part of the true risks they face. When these conditions are not met - as they aren't - then the case for regulation is a fortiori. But Frank shares with the neoclassicals a vocabulary that is incapable of talking about injustice.

6 comments:

Peter Dorman said...

Hi Kevin!

A couple of assumptions jumped out at me from Frank's example. First, the risky job pays more. Second, workers care about relative income but not relative safety. The first is, in my opinion, mostly counterfactual; the second would need evidence since there is no reason to suppose one type of comparison should be so much more salient than the other. Overall, I get the impression of a "just so" story.

From the standpoint of psychology (which all of us economists are obviously experts in), I think the critical issue is cognitive dissonance, commonly known as denial. Workers in dangerous jobs will try to suppress awareness of their situation. This buys an increase in "utility" at the cost of greater vulnerability to real physical harm. I give the issue some space in Markets and Mortality.

Robert D Feinman said...

Just a procedural request.
1. There are two Robert Frank's writing prominently at the moment (Newsweek and Cornell), so saying which one is meant each time would be helpful.
2. For those who missed the original article and/or the original essay links back to one or both would also be appreciated.

Anonymous said...

pd

frank assumes a risk premium wage
big deal
so that ain't real ....
he's in neo-econcon wonderland
why not
increase the good story line

even forget the weights biz
and the lower wage effect of both competing for the job
that has one or two openings

its really the partial equalibrium problem isn't it
the marshall's law fiddle
that flies out at ya
even after crossing thru the mirror


this is a magic hat trick
despite producing out of that hat a result
we soft souled humanists
obviously would like to reach

paine

kevin quinn said...

I'm sorry not to have specified the Robert Frank I meant - the Cornell economist, not the journalist. I'm new at this and having trouble incorporating links at the moment, but I will learn, I hope.


Peter, yes, the assumption is that it is relative *income*, not relative safety - or relative utility, for that matter- that is important. In general all of the arguments for the inefficiency attached to the pursuit of "status goods" posit a set of non-status goods, the under-allocation of resources to which constitutes the inefficiency. In other work, Frank makes both future goods ( via saving) and even leisure (pace Veblen) non-positonal goods. So, there are issues here.

It is a just-so story, but I think it is useful to consruct counter-examples to the claim that well-informed rational agents in competitive markets will achieve an efficient level of safety.

Peter Dorman said...

Kevin,

There is so much to say about all of this, but I will try not to get carried away. First topic: is it enough to assume competition and rationality to get a model of wage differentials that fully compensate for risk? If you think the employment relationship can be modeled as a moment-in-time, anonymous contractual exchange the answer is yes. If you think it is ongoing and the parties know each other’s identities you need to set up a repeated game model, and no one has yet done that in a way that reproduces those nifty neoclassical wage offsets. Until we have a model to talk about, we can’t say whether competition and rationality are sufficient.

To see this from a different angle, suppose instead of occupational safety we were studying sexual harassment in a world in which it is costly for companies to set up monitoring and enforcement systems to control harassing behavior by supervisors. Is it the case that the assumptions of “rationality” (a lot to unpack there) and competition would give you the result that an efficient level of harassment is realized, with harassed workers getting enough of a wage boost to leave them just as well off as the others? I would expect that being the victim of harassment would have long-lasting effects on the employment relationship that would have to be represented in the model. (Like a good economist, I am temporarily ignoring the ethical aspects of the situation.)

Second topic: There is no a priori reason to suppose that interpersonal comparisons play a greater role in income than they do in health. Yes, I know there is a large body of evidence on the income side, but there is also quite a bit on health. The most recent example is the much ballyhooed Framingham restudy, focusing on the spread of obesity through social networks. It has been interpreted as telling us that my conception of what it means to be overweight depends on how chunky my friends and family are. And biometrics are normally calibrated at a community level, no?

So why should Frank assume interpersonal comparisons along one dimension of well-being but not any others?

kevin quinn said...

Peter, I don't think we disagree as much as you think. In the example, the workers take the risky job, but the $10,000 premium is *not* a compensating variation: only a $20,000 premium would be compensatory. Frank's ideas are a critique of the practice of taking the differntials between safe and risky jobs as compensatory - no?

What if we put the case in the following way: in the example, suppose we gave the workers the right to a safe workplace. Suppose that the employers can save more than $10,000 but less than $20,000 by running an unsafe workplace. Would the workers in the example sell their right to safety for a wage boost of $10,000? Yes. Would that indicate that they had been compensated for the extra risk? No! Frank has no truck with the CD stuff.

I agree with you that is is dubious to make only relative income matter, though.