One broad class of theories about why American society has become a lot more civilized since the early-'90s peak in the rates of violent crime, property crime, teen promiscuity, child abuse, etc., is that law enforcement started cracking down harder. Raise the costs of homicide by raising the chance of it leading to life in prison (or whatever), and you'll see fewer homicides committed. I don't doubt that can play a role, but I question how widely that applies and what the magnitude is where it does apply.
As I've mentioned before, the list of risky behaviors that have declined since the early '90s is so vast that it makes less sense to propose ad hoc explanations for each decline -- for homicide, for teen pregnancy, etc. For example, locking up more of the risk-takers is a plausible account of why homicides declined, but not teen promiscuity or teen pregnancy because the risk-takers in those cases weren't given harsher punishments -- or any at all -- by law enforcement. Rather, it's better to talk about a fall in a generalized taste for risk in the minds of individuals. I prefer some kind of frequency-dependent selection model that has no stable equilibrium but instead oscillations.
As an example, imagine that almost everyone played "rock" -- then the frequency of "paper" would shoot up and that of "rock" would plummet. But then the frequency of "scissors" would shoot up and "paper" down. Finally, with "scissors" now the most common, the frequency of "rock" would shoot up and "scissors" down. And the cycle would repeat. What these categories would be in the case of falling levels of thrill-seeking, who knows. The point is that such models can account for very general changes and don't need lots of ad hoc explanations. By its nature the system will oscillate; we don't need to invoke lots of outside influences just before each swing of the pendulum, to mix metaphors.
Still, evidence would be good to look at. You might object to the case of teen promiscuity and pregnancy by saying that they could have gotten tougher on age-of-consent laws or something, so it really was like the "tough on crime" view would predict. (We'd still need evidence that law enforcement ramped up its efforts to enforce such laws, though.) What other risks are hardly ever punished? How about not wearing your seatbelt or not wearing a helmet while biking. The CDC's Youth Risk Behavior Survey polls a national representative sample of high schoolers every other year about all kinds of risky behavior, starting in 1991. Here is the percent of students who rarely or never wore a helmet (among those who rode a bike in the past 12 months), and the percent who rarely or never wear a seatbelt when riding in a car driven by someone else:
Both pictures fit the larger pattern of a general fall in risk-taking since the early '90s. It's dramatic for not wearing a seatbelt, and still down 10 percentage points for not wearing a bike helmet. Taking these risks is virtually never punished -- a conclusion we reach immediately from the fact that even now 85% of bike riders flaut whatever legislation there may be about wearing a helmet. So the fall in risky behavior is not due to a response to changes in particular policy incentives pertaining to homicide, teen pregnancy, etc., but rather to some very broad incentive, not embodied in any one policy, about risk-taking in general. I think that reflects the fall in trust of others -- that would cause you to be much more cautious. (That just pushes the question back to "what causes trust to cycle," but I prefer the same sort of model.)
This suggests that the increasingly "tough on crime" policies of law enforcement were reflecting a deeper and broader change already afoot in the population's taste for risk, rather than causing a thrill-seeking population to adjust its taste for risk downward. Everyone went into "better safe than sorry" mode, and because law enforcement can only be as harsh as the public will allow, the police could then level much stiffer penalties. Most people don't appreciate how much our public institutions reflect the public's desires in a competitive political system where violence is outlawed in the competition. Politicians don't brainwash a recalcitrant public; they give the public what it wants. * It is the same with the agencies of government that enforce the law on the ground.
* See Bryan Caplan's Myth of the Rational Voter.