Discounting -- why psychology matters too

A reader of my Bloomberg essay from last week (and related post here) on economic discounting emailed me to make a point that deserves some brief discussion. He asked me not to "ignore the psychology of discounting," i.e. the realistic aspects of human behaviour that really determine how we discount the future, whether or not that behaviour conforms to some "rational" paradigm.

My view is that I couldn't agree more. I didn't mention psychology in the Bloomberg piece for two reasons. First, for lack of space (900 words), and second, so as to give greater emphasis to the mathematical angle, this being where traditional economists feel they are on the firmest ground. Point to the fact that real people (and many other animals as well) in experiments don't seem to follow exponential discounting, but something weaker, and the hard-nosed economist will simply respond by saying this only shows that "people are irrational, and we can learn to act more rationally with logic."

Hence, whatever the importance of psychology, I thought it was important to bring out this one argument against exponential discounting, as it rests on the same dry logic that economists thought supported their position. It doesn't. What economists are currently doing is irrational. That should raise some serious questions about how cost-benefit analyses may be leading us to undervalue the future in a hundred settings.

But that's not at all to say that psychology isn't important. My reader pointed to the work of psychologist Shane Frederick of Yale University. I happen to know of Frederick as I use one of his past experiments on framing effects in some writing seminars I give (along with my colleague Justin Mullins) to Ph.D. students. The puzzle is:

A bat and a ball together cost $1.10. The bat costs one dollar more than the ball. How much is the ball?

Most of us feel an initial inclination to say 10 cents, even though the correct answer is 5 cents. The immediate, intuitive part of our brain pulls us toward the 10-cent response, and we have to use the slow deliberate part of our brain to get the right answer. If I recall the numbers correctly, Frederick did this experiment with students at Princeton University and University of Michigan, giving them something like 15 minutes to respond, and roughly half still gave the wrong answer. I use this to instill in students how important the structure of their writing is -- it's not only what you say, but how you say it, and saying it the wrong way creates a puzzle for your readers.

But back to discounting -- my reader (thanks!) pointed me to this paper by Frederick which looks at how similar framing effects influence how people respond to discounting questions. Very briefly: Some earlier experiments suggested that people, when asked how they value a life now versus one in the future, give a lot more value to the present life. The numbers ranged from 45 lives to more than 200 lives 100 years from now being worth the same amount as one life now. This was taken as evidence of strong discounting in the psychological make-up of people.

In contrast, Frederick showed in further experiments that the results you get depend very strongly on how you ask the question. Frame it in one way and you get evidence of strong discounting, frame it in another and people weigh lives 100 years in the future equal to those today. As he notes, the results are all down to the "elicitation procedures" used by the experimenter:

... different elicitation procedures yield widely varying results because they evoke (or suppress) several distinct considerations or criteria relevant to the evaluation of such life saving programs (e.g., uncertainty, efficiency, and distributional equity), and because they produce, to different extents, experimental demand effects: cues about what a reasonable answer should be.

Perhaps the most important observation made in this paper is that the previous experiments purporting to find evidence of strong discounting actually don't show such evidence. Referring to one of the most prominent such papers, Frederick notes that the experimenters afterward asked participants to explain their responses. Their answers followed the pattern below:

• Technological progress provides means to save people in the future 31.3%
• One should live day by day 31.7%
• Future is uncertain 15.4%
• The life I save may be my own 6.5%
• Present-oriented program saves more lives 1.6%
• Saving lives now means more lives in the future 2.8%
• Other 7.7%
• Do not know 2.9%

Frederick's comments on this I think are quite important:

Notably, there is no category labeled “I care less about future generations than this generation” or anything that suggests “ethical values” or “kinship” or a diminished concern for future people. In the study presented here, respondents were not requested to explain their answers, but were invited to comment on the questions or their answers if they wished. These comments suggest reasons similar to those listed above. Many respondents refused to believe that the future deaths would actually occur (e.g., “We’ll figure out a way to save lives in the future,” “Technology will change and guarantee higher survival rates,” “In 100 years, a solution might be found to save the life.”). Others were dubious of the long term commitment by the government needed to ensure that the future programs would be instituted (e.g., “I don’t trust long term projects in the hands of government agencies which are subject to political whims.”). None of the 29 people who offered written justifications for their choice indicated that they felt less concern for, or empathy toward, or kinship with future people.

I received a good number of very similar comments in the more critical responses to my Bloomberg essay. As Frederick sums up:

The results of this study cast doubt on previous claims that the public values future lives much less than present lives.