Are pornography and marriage substitutes for young men?

Apparently they are, according to a recent IZA Discussion Paper (PDF) by Michael Malcolm (West Chester University of Pennsylvania) and George Naufal (IZA). Using data on young males from the 2000, 2002 and 2004 waves of the U.S. General Social Survey (for which the data are freely available online for most waves), the authors investigated whether the number of hours spent on the internet each week and self-reported use of the internet to view pornography lowered the probability that respondents were married at the time of the survey.

There are a couple of obvious problems here, which the authors acknowledge. The first is reverse causality – married people are likely to have less opportunity to view pornography (what with their wife looking over their shoulder at what they are working on). Second, people with poor interpersonal skills are less likely to get married, and more likely to use the internet, which would confound the issue.

To overcome these problems, the authors use instrumental variables analysis (which I have earlier discussed here): they essentially find some variable that is expected to be related to pornography or internet usage, but shouldn’t plausibly have a direct effect on marriage rates. In this case, the authors use father’s education level as an instrument for hours of internet usage (because more educated people used the internet more, particularly in the early 2000s), and use urbanisation as an instrument for pornography usage (as more urbanised areas have better internet, and hence pornography, access).
The results show that pornography is indeed a substitute for marriage for young men:

“For pornography consumption specifically, the magnitude of the marginal effect also varies across models, but again all are negative and significant at the 10% level. Using the bivariate probit model, each 1% increase in propensity to look at pornography is associated with a 0.6% decline in the probability of being married. Across other models, the estimated marginal effect ranges from 0.07% to 5.3%.”

Can we be sure this is the effect of pornography, and not just general internet usage (which also had negative and significant effects on marriage)? Based on other results from the paper it seems likely, since self-reported use of the internet for other things (visiting finance sites, news sites, education sites, health sites, or sports sites) each had smaller marginal effects than visiting porn sites. Interestingly, visiting religious sites was associated with a higher probability of being married.

What does all this mean? One interpretation is that greater access to pornography via the internet lowered the cost of sexual gratification. As we know from simple demand theory, if ‘goods’ are (close) substitutes and you lower the cost of one, the demand for the other will fall. To the extent that marriage is also a source of sexual gratification, the demand for marriage has reduced, and this is consistent with marriage and pornography being substitutes. I tell a similar story in my ECON110 class, about the increasing availability of casual sex, and its effect on the demand for commercial sex services over time (the evidence is in the decrease in the price of commercial sex services, an example described in the Levitt and Dubner book SuperFreakonomics.

[HT: Bill Cochrane]