One married couple. Two smokers. One goes cold turkey. What happens to the other? Quit entirely? Smoke less? How much less? A paper by David Cutler and Edward Glaeser at Harvard’s Kennedy School investigate spousal influence on smoking. Measuring this influence on any behavior has always been tricky because both spouse’s decisions influence the other. To get around this problem, Cutler and Glaeser use a complex statistical model in which they use the presence of workplace bans for one spouse to measure the reduction in smoking by the other spouse.
Cutler and Glaeser estimate that the effect of one spouse’s quitting on the other is, on average, 40 percent. However, the influence is not symmetric. Wives have a bigger impact on husbands than husbands have on wives. Impacts are larger among spouses with higher education levels. The two also try to determine whether smoking bans have spillover effects beyond individual marriages. These effects show up at aggregation levels beyond individuals, which means that cities and states with workplace smoking bans have lower smoking rates overall. This need not be the case. People could simply shift their smoking patterns, which would not change state smoking rates dramatically. Since the evidence does not suggest this happens, they conclude that broader smoking bans that go beyond individual workplaces could reduce smoking. They write:
These results suggest that policy interventions that impact an individual’s smoking habit will have both direct effects and also indirect effects through on the smoking of peers. Workplace bans seem not only to have reduced worker smoking but also the smoking of the worker’s spouse. Our results also suggest that interventions are likely to have larger impacts when they are imposed at higher levels of aggregation, although we found little evidence suggesting that social interactions can explain the shape of the time series of smoking rates.
Nudge blog 