Tuesday, April 14, 2009

Auto Accidents, AIDS, Contraception and the Pope

Suppose you make cars safer by requiring seat belts, collapsible stearing columns, and other changes that make it less likely that an auto accident will kill the car's occupants. The obvious conclusion, and the one many people reach, is that the highway death rate will go down.

Sam Peltzman, in a classic article, pointed out that there was no good theoretical reason to expect that to happen. Auto accidents do not simply happen; they are the result of decisions made by drivers, such as how fast to drive, how much attention to pay to driving and how much to conversations with your passengers or listening to the radio, whether to drive home or take a cab after drinking a little too much. Making cars safer lowers the cost of dangerous driving; on the margin, drivers are more willing to risk accidents the less likely accidents are to kill them. So making cars safer results in fewer deaths per accident but more accidents. There is no theoretical basis to predict whether the net effect will be fewer deaths or more. Peltzman offered statistical evidence that, in the particular case he he was looking at—a collection of safety requirements imposed in the 1960's—the two effects roughly cancelled. Death rates per accident went down, the accident rate went up, and the annual death rate was about what it would have been without the changes.

I was reminded of this by a more recent controversy involving a different issue but the same logic. The Pope has, not surprisingly, come out against the distribution of condoms as a way of dealing with the AIDS epidemice in Africa—and, not surprisingly, been ferociously attacked for doing so.

Just as with auto safety and auto accidents, making sex safer has two effects working in opposite directions. It makes the chance that a given act of sex will result in AIDS transmission lower. But, by lowering that risk, it reduces the incentive to avoid sex entirely, to avoid sexual acts such as anal intercourse that are particularly likely to transmit AIDS, to avoid sex with people likely to give you AIDS, such as prostitutes. On theoretical grounds we have no way of knowing whether the net effect will be more AIDS or less.

It turns out that there is evidence that, just as in the auto case, the two effects roughly cancel. That, at least, was the widely reported conclusion of a Harvard AIDS researcher who had actually looked at the data. “We have found no consistent associations between condom use and lower HIV-infection rates, which, 25 years into the pandemic, we should be seeing if this intervention was working.”

All of which reminds me of another point, relevant to the Catholic church and contraception, which occurred to me quite a long time ago but which I don't think I have ever seen discussed. The church, for doctrinal reasons that are unclear to me, permits contraception via the rhythm method but condemns essentially all alternative methods. Critics of this policy frequently support their criticism with images of poor women who bear ten or twelve children, with terrible effects for themselves and, it is argued, the world.

The problem with that argument is that the particular problem they are concerned with is one—arguably almost the only one—that unreliable forms of contraception such as the rhythm method can solve. If your objective is to have four children instead of eight, a form of contraception that only occasionally fails will do a pretty good job of achieving it. That, presumably, is one reason why, prior to the invention of modern methods of contraception, birth rates responded to factors such as income that affected the desirability of having children, instead of being almost always near the biological maximum—although my guess is that the low tech methods being used were more likely to be coitus interruptus or oral sex than rhythm.

Unreliable forms of contraception can work pretty well for holding down marital birth rates. On the other hand, if your objective is to permit women to have sex with men they aren't married to without a significant risk of pregnancy—to permit, in other words, what has become the normal pattern of sexual behavior in developed societies—there is much to be said for more reliable forms of contraception.

Which leads me to suspect that neither side of that controversy is being entirely honest about its objectives. The Catholic church defends its position on doctrinal grounds, but it can be interpreted, perhaps more plausibly, as social engineering. Limiting contraception to unreliable methods—rhythm, which the church approves of, and interruptus, which it has no way of preventing—makes casual sex considerably riskier without imposing large burdens on marital sex and thus makes the former less attractive as a substitute for the latter. Critics of the church's position claim that their concern is with overpopulation and poverty, but support contraceptive technologies that enable—arguably have created—the modern pattern of sex largely outside of long term relationships.

On general principles, of course, I think contraception should be legal. On the question of whether improved contraception has had, on net, good or bad effects I am agnostic; I can see legitimate arguments in both directions. My point in this post, however, is not to support either side of that question but only to point out reasons to suspect that neither side of the controversy over contraception is being entirely honest about its motives.



At 8:56 AM, April 14, 2009, Anonymous Anonymous said...

As a practicing Catholic, I find your argument that the Church has ulterior motives in forbidding "technological" contraception slightly far-fetched - considering that it makes it stand for traditional marriage and against divorce quite explicit. Indeed, my personal perception is that it is usually mostly critics who time and again draw attention to the Church's stand on condoms, while the Church itself gives much more "airtime" to the defence of marriage than to contraception.

At 9:50 AM, April 14, 2009, Blogger David Friedman said...

Gorgasal raises a legitimate point--that the Catholic church does indeed argue for the defense of marriage as well as against (most forms of) contraception.

But I don't think I have ever seen statements by the church that link the acceptance of rhythm to the argument I suggested in the post. Nor have I seen any clear explanation of why rhythm is permitted while other means of contraception, both technological and non-technological, are banned.

At 10:55 AM, April 14, 2009, Anonymous blink said...

For the narrow question about aids, etc., you make good points that the balance is unclear. To be nearly indifferent between a worlds with and without contraception, you must discount obvious benefits: the pleasure of extra sex. To override a presumption that individuals know their own interests best, you must have in mind some weighty negative externalities.

At 11:02 AM, April 14, 2009, Anonymous Roger Collins said...

I love to explain this to people by saying the best thing we could possibly do for road safety is quit installing air bags and instead install a 9 inch dagger on the steering wheel pointed at the driver's chest. Obviously safety needs to be balanced with other goals in life.

At 11:49 AM, April 14, 2009, Blogger David Friedman said...

Blink correctly points out that sexual pleasure is a positive. One counter-argument is that the availability of good contraception and abortion results in many more children being brought up by single mothers. That result is counter-intuitive, but I go through the logic in some detail in the chapter at:


Look for the subhead "Out of Wedlock Births."

Blink mentions a dagger as a safety device. I think that proposal originated with Gordon Tullock, in which case he should be credited with it.

At 12:25 PM, April 14, 2009, Blogger Arthur B. said...

This comment has been removed by the author.

At 12:25 PM, April 14, 2009, Blogger HispanicPundit said...

Here is a good primer on Catholic social teachings. Janet Smith is one of the most widely read on the subject.

From my limited understanding, the argument against contraception is the same argument against masturbation and even homosexuality - it is a violation of natural law.

At 2:27 PM, April 14, 2009, Blogger Arthur B. said...

This comment has been removed by the author.

At 5:39 PM, April 14, 2009, Blogger Malcolm Kirkpatrick said...

Fifteen years ago, the school which employed me to teach Math insisted that I include "Consumer Math" and "Investment Math" in my courses (I wanted to teach basic Algebra). So I discussed minimum wage legislation and auto safety legislation, among other things. I asked: "Why might mandated auto safety features make roads less safe?" I anticipated answers as mundane as my own: that added features would make new cars more expensive, and so raise the fraction of older, less safe cars on the road. One student, Ellen Willy, however, instantly chirped: "People will just drive faster." That was a much better answer than mine. I read later that some economists studied the impact of mandatory air bags, and found that cars equipped with air bags were more likely than cars without air bags to be the at-fault party in two-car accidents between cars from both categories. Auto accidents also injure pedestrians, bicyclists, and skateboarders. Making drivers safer might well make non-drivers less safe, until they adapt to the new regime.

The authors of the air bag study suggested mounting a spear in the center of the steering wheel.

At 1:48 AM, April 15, 2009, Anonymous Anonymous said...

This comment has been removed by a blog administrator.

At 2:26 AM, April 15, 2009, Anonymous Anonymous said...

This comment has been removed by a blog administrator.

At 8:25 AM, April 15, 2009, Blogger Adam W. Hogan said...

I think the Catholic Church is very clear on their objectives. Their objective concerns the individual's relationship with virtue, and not the over-all effects to society. Any analysis from a utilitarian benevolent-dictator perspective will fail. The Church isn't out to maximize wealth, life span, or worldly happiness.

Even if condoms did lead to a reduction in AIDS the Church, I believe, would still not support their use. Discovering they were successful would not change reasons these birth-control methods are thought of as sinful.

At 12:25 PM, April 15, 2009, Blogger Jonathan said...

HispanicPundit: "From my limited understanding, the argument against contraception is the same argument against masturbation and even homosexuality - it is a violation of natural law."

What, you mean it violates a law of physics? I don't think any natural law deserves the name if it can be violated so easily and so often.

At 2:47 PM, April 15, 2009, Blogger Daveon said...

What work has been conducted on this _since_ Peltzmen's work? That's a 1975 article.

There's been a lot more mandated changes in car design since then and while accident rates might be a push, I don't have the data to claim otherwise, speaking for the UK the death rates now are lower than the were in the 1930s, despite having several orders of magnitude more cars on the roads.

Do you have any current data to support your argument?

Likewise, a quick Googling around other "experts" on the subject of aids suggests that he has something of an outlying opinion on this. The Botswanan experience, where they've gone heavily pushing condom use has led to a dramatic reduction in aids rates compared to next door South Africa.

At 8:33 PM, April 15, 2009, Blogger Kim Mosley said...

So if the auto manufactures both made cars safer and told the public that they were dangerous then more lives would be saved.

At 10:16 PM, April 15, 2009, Blogger Dale said...

I drove my bright red 1959 MGA faster after I installed seat belts. Got into more accidents and received more tickets but fortunately nobody was hurt.

One way of causing folks to drive more carefully would be to outlaw automobile insurance.

At 6:01 AM, April 16, 2009, Anonymous Ron Menich said...

Fatalities per 100,000 population is clearly trending downwards with time (c.f., http://www-fars.nhtsa.dot.gov/Main/index.aspx).

A longer-term trend of interest is that the number of accident fatalities now (against a population of over 300 million people) is about the same as it was back in 1961 (against a population of only 184 million people). (c.f., http://www.nbc13.com/vtm/news/local/article/us_highway_deaths_at_lowest_levels_since_1961/66905). Thus, the rate of fatal crashes per 100k population has come way, way down since the 1960s.

At 6:24 AM, April 16, 2009, Anonymous Jon said...

I believe you're right about the dishonesty towards objectives. There's money and power involved, which usually leads to subjective opinions and the use of incomplete, or innacurate data.

At 8:25 AM, April 16, 2009, Blogger David Friedman said...

"Fatalities per 100,000 population is clearly trending downwards with time "

That's been true for a long time--there was a time trend in Peltzman's regressions. I expect it partly reflects improvement in medical treatment, in part improvements in auto technology.

At 9:52 AM, April 16, 2009, Blogger Kurtis said...

I can only speak to my experience, condoms are only part of the problem. Number one people need to get tested, If they are positive they MUST tell sexual partners. Other pre cautions must be taken, there are dangers from Oral Sex, even brushing your teeth prior to making out pose a problem.

At 5:13 AM, April 18, 2009, Blogger Unknown said...

when you make our roads seem more dangerous they become more safer.

Please google Hans Monderman to learn more about this.

At 9:47 AM, April 18, 2009, Anonymous Anonymous said...

Risk compensation is definitely real (see Prof John Adam's work) in terms of car accidents. I believe there is a free market solution coming to road safety and that's from a combination of insurance companies and technology. Insurers will soon be able to trace all driver actions (GPS and Engine Management data) and of other people (video). Everyone is going to be scared of being caught doing something silly in a car (the insurer will have statistical evidence of what is dangerous) because someone'll report it to their insurance company. Insurers are bound to be keen to know if they've underestimated just how dangerous one of their drivers is. The devices are already on sale but probably not quite cheap enough yet to be backed by the insurers. No police or car safety regulations needed here...

At 11:27 AM, April 18, 2009, Blogger David Friedman said...

Anonymous raises the possibility of using insurance companies as a substitute for traffic regulations. I discuss a version of that in Chapter 7 of my Law's Order, webbed at:


Scroll down to the subtitle "A Market for Legal Rules."

At 4:16 PM, April 18, 2009, Blogger sconzey said...

The incentives are pretty clear cut here: unprotected sex carries a certain probability of giving you AIDS, which will kill you.

Yet people still do it.

Our first assumption should be that to have unprotected sex with someone who may or may not carry the virus is a rational choice, and let's take it from there.

Why would someone make that choice?


At 12:33 PM, April 19, 2009, Blogger Duncan said...

What Daveon said - especially on Botsawana vs. South Africa.

At 1:13 PM, April 19, 2009, Blogger Duncan said...

* ahem * Botswana, that is.

At 2:42 PM, April 21, 2009, Blogger Duncan said...

I mean seriously. "On theoretical grounds we have no way of knowing whether the net effect will be more AIDS or less. It turns out that there is evidence that, just as in the auto case, the two effects roughly cancel."

What additional risk-taking behaviour is being posited here, such that it's sufficient to balance the consequences of having unprotected sex? Your link leads to a Times article in which the chief of the UN AIDS prevention unit is quoted as describing the idea that two effects balance out as "ludicrous". That Times article quotes Edward C. Green, in a National Review interview, as proposing the idea. Here‘s the article he‘s quoted in. Here‘s an op-ed column he wrote. I can’t find any peer-reviewed work by Green on a hasty google, so I‘ll go with the op-ed. He certainly cites risk compensation as a possible factor. He then goes on:

“Another factor is that people seldom use condoms in steady relationships because doing so would imply a lack of trust. (And if condom use rates go up, it's possible we are seeing an increase of casual or commercial sex.) However, it's those ongoing relationships that drive Africa's worst epidemics. In these, most HIV infections are found in general populations, not in high-risk groups such as sex workers, gay men or persons who inject drugs. And in significant proportions of African populations, people have two or more regular sex partners who overlap in time.”

So - it isn’t that condom uses increases risk-taking sexual activity to such an extent that this balances out the effects of condom use (!) - it’s that the sexual relationships that “drive Africa’s worse epidemics” are precisely those in which condoms aren’t used.

Haven’t the knowledge to assess Green's work - not making any general claims. But the idea that risk-compensation might make condom use ineffective in combating HIV/AIDS is pernicious nonsense.

At 9:55 AM, April 26, 2009, Blogger Dessert Survivor said...

People who want safer cars and safer sex have sometimes vehemently opposed the suggestion that we try to develop a safer cigarette. (See Viscusi's Smoke Filled Rooms. Viscusi advocates safer cigarettes.) In the case of smoking, they see that reducing risk will change behavior. Some of these people cannot see that reducing risk with condoms will also change behavior. Hence the outrage directed at the Pope.

At 10:38 AM, April 26, 2009, Blogger Duncan said...

I don’t think anyone’s disputing that reducing risk with condoms will change behaviour. As is said upthread, that’s one of the reasons for opposition to condom use. The outrage at the Pope comes more from the fact that reducing risk with condoms will also reduce the spread of HIV. It’s a bad idea not to acknowledge that.

At 7:36 AM, April 27, 2009, Blogger earth that was said...

All this is relevant to the swine flu crisis. Authorities are, at least somewhat reluctant to recommend mass public adoption of (say N-95) masks as they give a "false sense of security".

(There are other arguments too but this one is the most relevant to the topic.)

At 5:05 PM, May 04, 2009, Blogger Dessert Survivor said...

The comment by Duncan on April 26 makes no sense to me. If one acknowledges that condoms change behavior, then one cannot assert that condom use reduces the spread of AIDS. Logically, it is possible that condom use will increase AIDS. As Friedman states, it is an empirical issues.

Driver's education provides a similar situation. Driver's education makes teenagers better drivers. But it does not therefore follow that driver's education reduces teenage accidents because it changes behavior. In particular, it puts more teenagers on the road. It is possible that driver's education may both make teenagers better drivers and cause more teenage accidents.

One way to express these ideas is in the framework of moral hazard. In this framework, things that reduce risk such as insurance reduce the costs of certain events, and therefore people do not have to be as careful about avoiding those events. Another way is in terms of complements and substitutes. Driver's education and teenage driving are complements, so making driver's education more available will also increase teenage driving. Are condoms and sexual intercourse outside of monogamous marriage substitutes, complements, or unrelated? A lot of advocacy of condom use seems to be based on an implicit assumption that they are unrelated. If they are complements, the push for condoms may have consequences quite different from those that are expected. It may, for example, increase AIDS rather than reduce it. The same argument, by the way, holds for teenage pregnancies.

(I thought I had made a similar comment a while back, but it seems not to have been added.)

At 4:45 AM, May 09, 2009, Blogger Duncan said...

Oops, left this hanging.

"If one acknowledges that condoms change behavior, then one cannot assert that condom use reduces the spread of AIDS." Of course one can.

"Logically, it is possible that condom use will increase AIDS. As Friedman states, it is an empirical issues."

Yes - an empirical issue. No one suggested otherwise.

It's entirely clear that in any number of cases condom use has had a massive effect on the spread of HIV AIDS. The success of Thailand's 100% condom policy, for example. (Check out figure one here [PDF], and tell me that’s not a positive correlation.)

There are also lots of cases where the balance of influence is much less clear - Uganda's 'ABC’ drive was clearly effective, but what proportion of that was down to condom use, and what proportion down to increased abstinence & faithfulness is unclear and hard to measure. (A bit problematic to try to separate out relative influences in an at least partly mutually-reinforcing set of policies, of course, but hey….) Research-community consensus seems to be that condom use was a significant but not the most important influence on falling HIV transmission rates.

What is clear - and this is the rational core of the ‘risk compensation’ argument - is that inconsistent condom use isn’t a whole lot more effective than no condom use at all. If inconsistent condom use is associated with increased risky sexual activity - if people use condoms inconsistently, and believe that this provides a level of risk reduction that in fact it doesn’t - this could, clearly, have serious negative consequences. (Though there are lots of other factors in play here too. For instance - the fact that anyone who is using condoms to limit risk, whether correctly or not, is at least aware of the risk, and since spread of HIV has often been associated with ignorance about how transmission works, you‘d expect to see some correlation between condom use and reduction in risky behaviour, at least for some communities. That reduction could achieved without the use of condoms, of course.)

It’s important to see what the claim is, here. David Friedman, our host, wrote above, with regard to the risks associated with condom use:

“it reduces the incentive to avoid sex entirely, to avoid sexual acts such as anal intercourse that are particularly likely to transmit AIDS, to avoid sex with people likely to give you AIDS, such as prostitutes.”

But these are not in fact the relevant risks. The relevant risk is participating in these risky sexual activities without using condoms. Consistent condom use reduces the risk of transmission to such an extent that, at a population-wide level, it’s simply inconceivable that increased sexual activities of these kinds could balance out the positive effects of condom use. [If you‘ve got any research suggesting otherwise I‘d love to see it…]

Just as with the Green piece I quoted before, the problem is in fact non-use of condoms. If one wants to make the risk-compensation argument, one needs to be clear that this is what one’s arguing - and accept that it doesn’t have the obvious anti-condom implications that it seems to. There’s still an argument to be made, naturally - but it would help if it stated the claim coherently, and was tethered to actual empirical data.

At 4:16 AM, May 13, 2009, Anonymous Anonymous said...

The Church position against contraception and abortions is related to the fact that orphans form the working backbone of the Church. Orphan children are easy to brainwash, indoctrinate and use as lifelong unpaid robot slaves,and they are found free.

At 5:02 PM, September 21, 2009, Anonymous Anonymous said...

The thing you are forgetting with the AIDS/sex thing is that overall utility has gone up since more people are getting laid.


Post a Comment

<< Home