Loss Aversion and Framing

I recently conducted a test of loss aversion and framing in the context of environmental regulation. The simple result is that in one frame, a mortality risk of 1/100,000 was valued at $25; in another frame, a mortality risk of 1/100,000 was valued at $100. The difference is substantial and important, because the first suggests a value of a statistical life of $2.5 million, and the second a value of $10 million.

In the first frame, a large group of law students was asked the following question:

Under existing regulation of arsenic in drinking water, you face an annual cancer risk of 1 in 500,000. The government is proposing to tighten the regulation, in a way that would decrease the risk to 1 in 600,000. How much would you be willing to pay each year, in increased water bills, to favor the government’s proposal?

(a) $25
(b) $50
(c) $100
(d) $200
(e) $400
(f) over $400

The median answer was $25.

Under the second frame, a large group of law students was asked the following question:

Under existing regulation of arsenic in drinking water, you face an annual cancer risk of 1 in 600,000. The national government is proposing to loosen the regulation, in a way that would increase the risk to 1 in 500,000. How much would you have to be paid each year, in reduced water bills, to favor the government’s proposal?

(a) $25
(b) $50
(c) $100
(d) $200
(e) $400
(f) over $400

The median answer was $100.

From the standpoint of conventional economic theory, an apparent oddity in these results is that people were willing to pay far less to reduce a risk from 1/500,000 to 1/600,000 than otherwise identical people were willing to demand to allow a risk to be increased from 1/600,000 to 1/500,000. The disparity seems to be a tribute to the power of loss aversion: A loss, from the status quo, is perceived as far more undesirable than a gain, from the status quo, is perceived as desirable. (Loss aversion is a central finding in behavioral economics. It remains a bit of a puzzle why people show loss aversion.)

A general implication is that in contingent valuation studies, frames greatly matter. A more particular implication is that the value of a statistical life, in such studies, will much depend on whether the change is described as a loss or a gain.

By the way, the EPA uses a value of a statistical life of about $6.1 million -- very close to halfway between the two figures in this little study.

Comments

[DWAnderson]

Isn't it possible that something else explains this discrepancy? Namely, a distrust of government proposals. In each case the respondents had a bias in favor of the status quo.

In the first question, people are skeptical of the real risk reduction in exchange for their tax dollars. In the second question, people are skeptical that the risk increase would be limited to the amount set forth in the question, so want to be compensated more to overcome that skepticism.

I suggest this because the above thinking reflects my instinctive reactions (as such a skeptic) to the two proposals.

This may be a framing issue as well, just not the one that you identify.

[saul levmore]

It would be interesting to see how the group responds if simply given a choice among two or three regulatory options with associated costs. It could be framed as new regulation, or they might simply not be told of the status quo.

In some contexts the framing of the reduction in risk might lead to a different result because the listener (especially law students trained to expect follow-up, Socratic questions) imagines the next question as "Well, if you are willing to spend $x for that extra safety why don't you spend less than $x for that other safety step in some other arena." New investments are somehow (and perhaps just as puzzling)more readily compared to alternative investments, than are investments already in effect.

[David Kalow]

It may be worth a glance at the Nature review of "Why Choose This Book? How We Make Decisions" (445:711, Feb 15, 2007), which I read as suggesting the brain needs shortcuts - perhaps like being loss/risk adverse - in order to compute efficiently

[LAK]

Yea, so what? This is nothing new, you taught me the same 5 years ago. I still am trying to figure out what one is supposed to do with this information, why it is important (other than providing further support for the very general and obvious claim - to everyone other than the most alienated egg head ivory tower classical economists - that human beings do a poor job a valuaing things) and why you keep devoting yourself to these kind of behavioral studies.

To show that "frames greatly matter?" No sh*t. To show that in valuing a statistical life the manner in which you frame the valuation matters? One need only look at some of the things human beings value to KNOW they do a poor job at placing a value on almost everything, and that such valuations will vary by all sorts of factors.

For such a smart man with so much to say, I still, years later, wonder why you care about demonstarting and quantifying these flaws in human valuation which are obvious to almost anyone and clearly unquantifyable in any meaningful way (You'll have to tell me how the results from a sample of law students is meaningful to the rest of humanity and in other social contexts)

Try conducting the same study with law partners who make over a $ million a year, you'll get different variation there too. Try the same study with some welfare mothers too, then get back to me and explain why you do these studies, other than to show your classical economist friends for the 100th time that not everyone is a rational ivory tower white male. Thanks but the rest of us in the real world already know this.

[Erasmussimo]

First off, I'd like to offer two minor criticisms for the study. The first is that the first question, which received a median answer of $25, is flawed because there was no answer lower than $25. How do you know that, had there been an option for $10, the median response would not have been $10? This is a quibble given your interpretation of the results.

Second, there's an assumption of linearity here. If people are willing to pay $100 for a 1/100,000 reduction in risk, that doesn't mean that they're willing to pay $1,000 for a 1/10,000 reduction in risk or $10,000 for a 1/1,000 reduction in risk. What do we know about nonlinearity in this matter?

Lastly, I'd like to explain the value of studies such as this to Mr. LAK. Yes, it is true that people aren't rational. It is also true that people are sometimes rational. Indeed, some decisions people make are surprisingly rational. Human rationality is not a black-and-white trait; we are not divided into one group of cool rationalists and another group of slobbering impetuous fools. Rationality is a multi-dimensional trait, and we are just now beginning to explore the dimensions and styles of human rationality. Indeed, I suspect that, as we learn more about this behavior, we'll find that even the most apparently irrational behaviors have some context in which they turn out to be rational.

For example, economists are just starting to figure out the "herd mentality" aspect of stock markets. It turns out that there is a rational basis for playing the sheep. Of course, you already knew all about that rational basis, so you have no need for this research. But for the rest of us dummies, research like this is illuminating.

[Joan A. Conway]

Hope springs eternally with the people under 40, who are the people most affected by the cost of owning property and paying for their water!

People don't think it can happen to them!

Therefore under loss aversion the increase in the cancer risk has more support than the decrease in the cancer risk.

People generally see others as the opponent for public services, and pay accordingly to risk the lives of their neighbors, believing somehow that they are immune to these risks.

Human nature is not that flattering to many responsible humans, such as the communities responsible.

They encourage individual responsibility that they believe allows them to tighten their purse strings.

They simply do not trust big government regulations, even on essential services.

This is a dividing line between many conservatives and liberals.

Until regulations are proven to be faulty or fair, many feel regulations violates their freedom.

Few are willing to pay for fair government, and prefering conditions on fairness, such as the Fair Housing Act, where a fair amount of harassment is accepted, under a socio/economic theory.

It is the other guys problem!

This of course is unfair, but to make it costly for the majority to provide saver water has the unintended consequence of being seen as an unfair tax upon the many in support of the few.

[LAK]

Eras,

It is certainly valuable to qualify the ways in which human beings are irrational. Identifying heuristics is valuable to all, and the behavioralists do a good job. It is especially important to do so given that so much economic theory and worse, actual policy, is based on assumptions of rational action and decisions made with adequate knowledge.

My quiblle is when you try to quantify these behaviors in some sort of meaningful way, which is on its face ridiculous to me. Measuring law students to come up with some kind of variation is useless. Worse it is kind of like playing the conservative economists' game to inject some bad math into studies of human behavior, usually done as an attempt to give a more rational air to unethical policies.

[Joan A. Conway]

I don't know what you are saying LAK, but your last sentence: Worse it is kind of like playing the conservative economists' game to inject some bad math into studies of human behavior, usually done as an attempt to give a more rational air to unethical policies, impresses me. Do you care to explain this to me?

Posted by: LAK | February 26, 2007 at 11:59 AM

My previous comments were done without the support of the Harvard Negotiation Law Review, Multidisciplinary Journal on Dispute Resolution, Spring 1999, Psychological Principles in Negotiating Civil Settlements, by Richard Birke, and Craid R. Fox. The following idea is posted in light of reading this pamphlet many years ago.

Remeber those Clinton Dems when the 2000 election was over, and Al Gore lost, those Dems that exercised or had the power of memory through their misappropriations of White House office stuff.

Could our Revoluntionary Framers of our constitution contrive for those necessary checks and balances between the parties after a closely lost election relying on a Supreme Court Review, under the Separation of Powers Act, or between the branches with the inherent disputes between the President of the out-going party, and the Congress of the in-coming party.

Such a Plan has and does lack a means of socializing all sides of the constitutional norm that elections and party shifts can be very ugly, relying on an historical Supreme Court review of the results, and that wars must be authorized by Congress, that the Supreme Court refuses to review.

Both was planned for presidential defiance or party destruction of governmental property as defiant acts unprecident in our history.

Congress refuses to be accountable by authorizing the Supremem Court to review the Executive, and Congressional, and excesses of administrative agencies, without Article III standing, and having 'no juticiable' case or controversy.

I am a little busy just now, but I will try to catch up as soon as possible.

[fed]

A minor mathematical error: the change is in fact 1/3000000, not 1/100000 (a change from 1/500000 to 1/600000 is a change from 6/3000000 to 5/3000000). Not that it changes the upshot of the informal survey.

And while loss aversion may account for the seeming irrationality, status quo does account for some of the disparity in a rational way; the proposed gain and loss are not equivalent percentages of the status quo. Which isn't to suggest every student did the math and adjusted his or her answer, but people may at least subconsciously consider what portion of their status quo is affected, and loss aversion (at least in some cases) could be a mistaken handling of the intuitive estimates of percentage change.

[LAK]

Joan,

let's just say I encounter a lot of expert reports done by conservative economists at the behest of major corproations who engage in anticompetitive activity. They are chalk full of mathematical equaltions that most of the time are idealism that have very little to do with what actually goes on in the real world, and are only included to give the air of importance and mathematical analysis to often complex dyanamic issues that can't be captured by idealized matehmatical models.