The (Honest) Truth About Dishonesty

The other participants played the role of advisers, and their task was to advise the estimators on their guesses. (Think of someone akin to your stock adviser, but with a much simpler task.) There were two interesting differences between the estimators and the advisers. The first was that whereas the estimators were shown the jar from a distance for a few seconds, the advisers had more time to examine it, and they were also told that the amount of money in the jar was between $10 and $30. That gave the advisers an informational edge. It made them relative experts in the field of estimating the jar’s value, and it gave the estimators a very good reason to rely on their advisers’ reports when formulating their guesses (comparable to the way we rely on experts in many areas of life).

The second difference concerned the rule for paying the advisers. In the control condition, the advisers were paid according to the accuracy of the estimators’ guesses, so no conflicts of interest were involved. In the conflict-of-interest condition, the advisers were paid more as the estimators overguessed the value of the coins in the jar to a larger degree. So if the estimators overguessed by $1, it was good for the advisers—but it was even better if they overguessed by $3 or $4. The higher the overestimation, the less the estimator made but the more the adviser pocketed.

So what happened in the control condition and in the conflict-of-interest condition? You guessed it: in the control condition, advisers suggested an average value of $16.50, while in the conflict-of-interest condition, the advisers suggested an estimate that was over $20. They basically goosed the estimated value by almost $4. Now, you can look at the positive side of this result and tell yourself, “Well, at least the advice was not $36 or some other very high number.” But if that is what went through your mind, you should consider two things: first, that the adviser could not give clearly exaggerated advice because, after all, the estimator did see the jar. If the value had been dramatically too high, the estimator would have dismissed the suggestion altogether. Second, remember that most people cheat just enough to still feel good about themselves. In that sense, the fudge factor was an extra $4 (or about 25 percent of the amount).

The importance of this experiment, however, showed up in the third condition—the conflict-of-interest-plus-disclosure condition. Here the payment for the adviser was the same as it was in the conflict-of-interest condition. But this time the adviser had to tell the estimator that he or she (the adviser) would receive more money when the estimator overguessed. The sunshine policy in action! That way, the estimator could presumably take the adviser’s biased incentives into account and discount the advice of the adviser appropriately. Such a discount of the advice would certainly help the estimator, but what about the effect of the disclosure on the advisers? Would the need to disclose eliminate their biased advice? Would disclosing their bias stretch the fudge factor? Would they now feel more comfortable exaggerating their advice to an even greater degree? And the billion-dollar question is this: which of these two effects would prove to be larger? Would the discount that the estimator applied to the adviser’s advice be smaller or larger than the extra exaggeration of the adviser?

The results? In the conflict-of-interest-plus-disclosure condition, the advisers increased their estimates by another $4 (from $20.16 to $24.16). And what did the estimators do? As you can probably guess, they did discount the estimates, but only by $2. In other words, although the estimators did take the advisers’ disclosure into consideration when formulating their estimates, they didn’t subtract nearly enough. Like the rest of us, the estimators didn’t sufficiently recognize the extent and power of their advisers’ conflicts of interest.

The main takeaway is this: disclosure created even greater bias in advice. With disclosure the estimators made less money and the advisers made more. Now, I am not sure that disclosure will always make things worse for clients, but it is clear that disclosure and sunshine policies will not always make things better.

So What Should We Do?

Now that we understand conflicts of interest a bit better, it should be clear what serious problems they cause. Not only are they ubiquitous, but we don’t seem to fully appreciate their degree of influence on ourselves and on others. So where do we go from here?