Here is a quick heuristic about social policy. Suppose it is in the social interest to decrease pregnancy rates in some population of teenagers. Then my quick heuristic is this:
- One cannot expect the promotion of a contraceptive to lower the pregnancy rate very much below the typical-use failure rate of that contraceptive for that population.
What is the reason for the heuristic? It is obviously difficult to simultaneously provide contraception, and instruction on their use, while promoting abstinence. Intuitively (and I'm just giving a heuristic) we do not expect mixed messages to work well. But the only way we're going to get a pregnancy rate below the typical-use failure rate of the contraceptive is by having some people abstain totally or to decrease their sexual frequency significantly below the number used in the typical-use failure rate calculations. But we would expect, for rational choice reasons, the availability of contraception to decrease the rate of total abstinence by reducing the costs of intercourse, and we would also expect it to increase sexual frequency. Here is a rough estimate. Suppose Sally disvalues pregnancy in her circumstances at somewhere between 1.2 and 20 times the value of a year of sex (very roughly: she'd be willing to abstain about 1.2 years to avoid a pregnancy in her circumstances but she wouldn't be willing to abstain 20 years to avoid it). It seems intuitively right to me that many teens will fall in this range. Then if pregnancy is the only consideration, it is decision theoretically rational for Sally to max out her sexual activity if she is on the pill but to abstain totally if she is not using any contraceptive (this is a pretty easy calculation using an 85% no-contraceptive annual conception rate).
The U.S. teen pregnancy rate in 2006 was 7%. Promotion of the pill could imaginably lower that somewhat, if the 5% figure is correct, but not if the 9% figure is. However, unless these teenagers use both the pill and condoms, there will be significant health risks for sexually transmitted infections. Because of these, any contraceptive-based policy would likely involve condoms as well. But intuitively one does not expect all that many teens to double up and use both the pill and condoms. If condoms are promoted, we'd expect a significant percentage of the sexually active population to use only condoms. If we have a half-and-half mix between condoms and the pill, and no overlap, there'll be a failure rate around 10-12%. Which is significantly higher than the pregnancy rate, at least with 2006 data.