July 8, 2011 10:52AM

New Paternalist Surprises

The front pages of Thursday's Wall Street Journal and Washington Post (links below) both featured stories on the unexpected consequences of the sort of "nudging" policies recommended by so-called "libertarian paternalists." Mario Rizzo, an economist at New York University who often blogs at ThinkMarkets, sent along this commentary to share with C@L readers:

As Glen Whitman and I have repeatedly argued, new paternalism faces a knowledge problem similar to that uncovered by F.A. Hayek in his critique of socialist calculation. In our view, paternalists cannot acquire the knowledge they need to implement policies that are effective according to their own standards.

The new paternalism purports to nudge people toward the better satisfaction of their own preferences than people can achieve themselves. We are too ignorant, too weak-willed, and computationally too incompetent to satisfy our real or underlying preferences. We save too little for our retirement because we are overly impatient and cannot postpone spending.  We eat too many calories because we underweight the costs of future illness due to obesity.

The new paternalists thought they had at least a partial solution to the first problem of undersaving. For those people who have employer-sponsored retirement savings programs we can use a common “defect” in decisionmaking to help them out.  People are prone to status-quo bias, that is, they tend to leave in place whatever situation they may find themselves in. So when employees are automatically enrolled in retirement savings unless they opt out, more people are enrolled than when the default is non-enrollment unless they opt in. What could be simpler?  Make the default automatic enrollment, and voila more retirement savings!

Now comes the annoying data.

According to a recent study commissioned by the Wall Street Journal more people are indeed enrolled in 401k programs as predicted. However, those who would have chosen enrollment under the old opt-in system (around 40% of new workers) tended to remain in a lower salary allocation in the default (frequently 3%) than they would have chosen on their own. So instead of using the status-quo bias to increase savings it turned out that the automatic enrollment decreased savings among this group.

What was the result overall?  Savings in these plans have gone down. Vanguard estimates that half the decrease in 401k savings rates in its plans from 7.3% in 2006 to 6.8% in 2010 was due to automatic enrollment. Of course, this was not entirely unexpected, as Whitman and I pointed out in our 2009 law review article.

Moving on to obesity. Mandatory posting of the caloric content of food in chain restaurants has been another new paternalist light-touch intervention in the market. Although restaurants are coerced, consumers are simply given information. With the calorie-postings staring at people at the moment of sale it was thought that many people would cut down on calorie consumption. An article in Thursday’s Washington Post indicates that the local laws mandating calorie-postings have not had the desired effect. Instead of concluding that perhaps people simply value the pleasure of caloric food more than the projected negative consequences, paternalists conclude that a stronger policy may be necessary. Many of them stand ready to impose a heavier touch: an excise tax on “junk food.”

Each of these cases illustrates the same problem. Trying to determine what people’s true or underlying preferences are (Do they want more 401k savings? Do they want to reduce calorie consumption? Do they want to incur the opportunity costs of each?) is more difficult than it may seem at first.  Trying to engineer the appropriate response from individuals requires a detailed knowledge of the interaction of many conflicting incentives. New paternalists do not take seriously the very biases literature they often cite. For one thing they choose to analyze only one — or perhaps two — biases at a time. The reality is one of many conflicting complex-interacting biases. Biases are also very context-dependent and so they may vary in direction and quantitative magnitude from case to case and from year to year.

The moral of the story: Expect more surprises and unintended consequences from new paternalist policies.