What's Cooking in the Ivory Tower
As far as summer events go, this August's annual meeting of the American Political Science Association doesn't have quite the same marquee value as, say, the recent opening of the Disney Store in Times Square. Let's face it, the average academic just doesn't look as good in a toga as Hercules. But make no mistake, the professor is probably the braver of the two. As you read this sentence, not only political scientists, but social scientists from all branches of the field, are boldly slaughtering sacred cows of both the right and the left. Much of their work has far-reaching implications. We took an informal survey of some of the nation's top researchers to determine the most important of this year's findings and came up with plenty of bad news for conservatives -- and liberals.All Hail the Men in BlueSocial scientists are the first to admit the recent nationwide drop in crime has them stumped. "It defies all current theories and predictions. We're just not able to explain it," says RAND crime expert Peter Greenwood. Political leaders are happy to claim the credit, pointing to their recent efforts to put more cops on the street. That's just a feel-good cosmetic fix -- and a costly one at that -- retort the criminologists. They argue that external forces, like demographic changes, economic improvements, and shifting patterns of drug use play a much bigger role. But without any concrete evidence, who's a voter to believe?Although he hasn't solved the entire question, University of Chicago economist Steven Levitt has at least hit upon a way to study the police angle. It was a tough nut to crack because of the difficulty of interpreting the relationship between cops and crime. For instance, the presence of a large number of cops in a city with lots of crime could indicate that the cops are ineffective, or it could simply mean that the high crime rate has led the city to hire more cops. Levitt found a clever way around this conundrum by observing that mayors and governors almost always hire more police officers in an election year regardless of whether crime is up or down. So by comparing the level of crime before and after election-motivated police increases, he's been able to measure whether those increases have had an impact. His stunning conclusion: Every police officer hired reduces five violent crimes and seven property crimes a year.But before the "tough-on-crime" politicos start strutting, they should know that not all of their favorite miracle cures are supported by the research. Lock-em-up approaches to drug offenders, for example, are proving highly ineffective. A particularly illuminating study by a RAND research team headed by Jonathan Caulkins analyzed the cost-effectiveness of various crime-fighting strategies and found that, dollar-for-dollar, providing treatment to drug offenders reduces serious crime 10 times more than conventional enforcement techniques (such as arrests, confiscations and short-term jail stays) and 15 times more than long mandatory sentences.The lesson in all this: Treat drug offenders and carry a night stick.Cultural MattersThe Cadillac Queen may have dominated the welfare debate in the 1980s, but the 1990s have seen the rise of a new leading lady: the teenage mom. Everyone from Charles Murray to Bill Clinton is now convinced that the crux of the welfare problem is the increasing number of young girls having children out of wedlock. There are plenty of statistics to back their claim. For instance, although fewer than 10 percent of families receiving Aid to Families with Dependent Children (AFDC) are headed by teenage mothers, about half of AFDC families are led by women who first gave birth as teens. Clearly a sensible national policy would focus on preventing these near-children from having children of their own. But how to do it?Before we can answer that, we've got to figure out why so many poor, unmarried teens are having kids in the first place. What is it with these girls? Is there something about not having money that makes them behave irresponsibly? In a fascinating study whose findings surprised even the researcher, University of Chicago Professor Susan E. Mayer found that the key determinant of whether a child drops out of high school and becomes an unwed mother is not her family's income, but her parents' character as demonstrated by their "skills, diligence, honesty, good health, and reliability." In other words, a child whose parents have these traits is almost as likely to graduate from high school and avoid teen pregnancy if she's poor as if she were middle class.Mayer's findings will frustrate knee-jerk liberals who maintain that the poor are just like the rest of us -- all they need is more money. Au contraire, discovered Mayer, a long-time liberal herself; many teen welfare moms have fallen into this poverty trap because they were raised by parents with the wrong values. As long as children's most basic material needs for food, shelter, and medical care are taken care of, additional cash has little impact on their life prospects. For example, Mayer's calculations show that even if the income of the poorest 20 percent were doubled, the national teen child-bearing rate would only decrease by 2 percent. But Mayer's work also contradicts the conservative assumption that poor people are bound to get their act together if you just press them hard enough. Republicans subscribing to this view have introduced provisions in the new welfare law that prohibit states from giving federal block grant money to unmarried under-age moms unless they are either in high school, enrolled in a state-approved training program, or have already obtained their diploma or GED. Yet Mayer notes that there's little evidence to suggest the threat of these sanctions alone will make teen moms comply with the new rules. What's needed, she says, are training programs that will teach them new values.Getting Down to Brass TacksGiven the obvious importance of training programs for teens, the question that naturally arises is, how well are the existing ones performing? A number of researchers are trying to answer that. And their findings can charitably be described as mixed. Take the four-year evaluation of Ohio's innovative "Learning, Earning, and Parenting" (LEAP) program recently completed by Johannes Bos and Veronica Fellerath of the Manpower Demonstration Research Corporation (MDRC). LEAP is a statewide program that uses a combination of bonuses like cash benefits, counseling, and help with child care and transportation, along with sanctions like welfare grant reductions, to encourage teenage welfare moms to finish school. Sounds pretty nifty. But the study found that while LEAP increased the school enrollment, employment rates and earnings of mothers who were still in school when they started the program, it had almost no impact on teens who had already dropped out of school. The MDRC also found that despite its other successes, LEAP failed to prevent the teens from having more children, or to boost the graduation rate of even in-school teens. Politicians who blithely promise that "training programs" will prevent people from falling through the cracks of our new, harsher, welfare system need to take note.Lest all these dark tidings have you reaching for your Prozac, you'll be glad to learn that there's better news from a home visitation project being studied by researchers at the University of Colorado. The program pairs highly trained nurses with teens who are expecting their first child. The nurses drop by for an hour-and-a-half each week at the start, then gradually reduce the frequency of their visits until the teen's child reaches age two. The nurses train the teens in basic parenting and contraception. But they also focus on helping the teens develop self-confidence, motivation, the ability to set goals and plan ahead, and an overall sense of self-sufficiency. At $6,700 per family, the program isn't cheap, but the results are astonishing: An evaluation published this August by a group led by David Olds found that 15 years after they completed the program, mothers were 31 percent less likely to have given birth to a second child. Program participants had also spent 33 percent less time on AFDC. Now the team is preparing to study the program's long-term impact on the children. If Susan Mayer's research is right, the program's success in changing the teen mothers' character should have significantly improved the life prospects of their kids. We'll keep you posted.You Read It Here FirstOrwellian though it may sound, the phrase "natural rate of unemployment" was actually coined by Milton Friedman in 1968. Friedman defined the term as the level of unemployment at which the rate of inflation does not change. When unemployment dips below this natural rate, Friedman explained, employers must compete for the smaller supply of available workers by offering higher wages. Workers spend the extra cash and drive up inflation. When the unemployment rate rises above the natural rate, employers are suddenly inundated with job applicants, enabling them to lower wages and causing inflation to fall. After a spate of studies in the late '60s and early '70s established the natural rate to be 6.0 percent, that figure was etched in stone -- and has been revered as a divine constant of economics ever since.But between 1995 and 1996 a pivotal event forced economists to reconsider: During this period the unemployment rate slipped to well below 6 percent and inflation dropped. How to explain this? A few months ago one of the early pioneers of the topic, Professor Robert J. Gordon of Northwestern University, laid out his evidence that the natural rate had actually fallen over time. Also promoting this view is Joseph Stiglitz, formerly Chairman of the Council of Economic Advisers and now Chief Economist of the World Bank. In a recent paper, Stiglitz estimated that since its peak in the early 1980s, the natural rate dropped 1.5 percentage points. Why the slide? Stiglitz identifies a number of causes. To start with, increased competition from foreign workers and the declining power of unions have combined to bring down wages. A second reason is that the maturing of baby boomers has increased the proportion of middle-aged people in our population. This is significant because middle-aged folks are generally committed to staying in their current job -- so employers don't need to raise wages to hold onto them in times of low unemployment. And any middle-aged workers who are inclined to push for higher wages are restrained by the knowledge that they could easily be replaced by the few remaining older people still looking for work -- who tend to be aggressive job-hunters. As a result of all these characteristics, middle-aged people are able to sustain much lower levels of unemployment without triggering inflation. The more of them in the labor force, the lower the natural rate of unemployment for the nation as a whole.Stiglitz has also found intriguing evidence for a third possible cause of the drop in the natural rate, namely, that when the unemployment rate fell and then stayed below the existing natural rate, this caused the natural rate itself to move down toward the lower level. If Stiglitz's hypothesis is true, the most likely explanation is that once all those people who had been unemployed were re-hired, they got back into the habit of working. Now when they lose their job, they hunt for the next one more aggressively -- making it easier for employers to attract workers without raising wages even when unemployment is low.Whatever its causes, the decline of the natural rate is a major development. "There's a tendency by some to look at unemployment numbers as just statistics," Stiglitz notes, "but these are real people without jobs experiencing real suffering. If we can have a society where we have 4.5 versus 6 or 7 percent of the population out of jobs, that is a big thing."Long time readers of The Washington Monthly will not be surprised to hear that it's possible to have lower than 6 percent unemployment without overheating the economy. This magazine has been after the Federal Reserve to aim for lower unemployment since the early 1980s. Why was the Fed so unwilling to experiment? In part because the traditional wisdom held that if they pushed too far and allowed unemployment to fall even slightly below the natural rate, inflation would spiral out of control. To compensate, the Fed would then be forced to drastically increase unemployment -- plunging the economy into a recession. But in his second major discovery, Stiglitz found the exact opposite to be true. An error on the side of too little unemployment causes only a slight increase in inflation. What's more, to bring down that inflation you don't have to pull back the reins as much as you had loosened them in the first place. Only a small increase in the unemployment rate will suffice. What do all these findings mean in practical terms? Says Stiglitz, "the evidence very strongly argues for a policy of cautious activism" on the part of the Fed. I hate to say it, butÉWe told you so!A Moment of InspirationGreat discoveries often happen in mundane places. Archimedes had his Eureka moment whilst taking a bath. Isaac Newton's revelation was supposedly brought on by the thud of a falling apple. And Harvard Professor Gary King's insight into one of the longest standing unsolved problems in quantitative social science struck him as he sat in an Ohio court room.The date was November 16, 1994, and King had been listening to expert testimony about whether Ohio's state redistricting plan violated the U.S. Voting Rights Act. To prove their case, the plaintiffs had to show that African Americans in the districts in question consistently voted as a block, did not favor the same candidates as non-black voters, and were being prevented by the majority from electing their picks. The trouble was there were no direct figures about the percentage of blacks in each district who voted for each party -- or even about how blacks voted nationwide. That's because there's no good way of collecting this data: It's illegal to sneak a peek into the voting booths, and expensive to conduct precinct-by-precinct surveys. Even if such surveys were done, when it comes to sensitive racial issues they are notoriously unreliable. Of course, social scientists do have some statistics to work with. For instance, they know the percentage of blacks who live in each area as well as the percentage of voters in each area who cast their ballots for each party. The key challenge, then, is to use this type of "aggregate" or "ecological" data to figure out how the subsets overlap. This so-called "ecological inference problem" has turned out to be very tricky to solve.Since it was first identified in 1919, the ecological inference problem has tormented conservative and liberal researchers alike. It has stymied everyone from epidemiologists trying to determine whether radon exposure causes cancer, to historians wondering which socio-economic classes voted for Hitler. Meanwhile, qualitative political scientists have spent years searching for a way out. Until King came along, the best methods were grossly inaccurate -- at times producing results that were downright ridiculous.Which brings us back to that day in November. Sitting in the court room, King listened in dismay as the expert witness reported that, according to the estimates produced by the most accepted method, in a large number of districts, the proportion of blacks who voted for Democrats was over 100 percent. "That was an absurd moment!" says King. And that's when he had his epiphany. It's pretty technical stuff, but here's a stab at an explanation in English: King realized he should break down the data into the smallest units possible, in this case the precincts composing each district. For each precinct he would determine the range of possible answers -- in other words, the smallest and largest number of blacks that could possibly have voted for each party given the precinct's total voting break-down. Then King would narrow that range further by taking into account the voting patterns he'd observed across all the other precincts. For instance, if in a particular district he knew that between zero and 80 percent of blacks could have voted Democratic, and he knew that precincts with lots of blacks tend to vote overwhelmingly for Democratic candidates, he could reasonably assume that blacks tend to favor Democrats, and that the percentage of blacks who voted Democratic in the particular district he was analyzing was probably a lot closer to 80 than to zero. (King uses a statistical method to determine just how close -- and whether his assumption even applies to the circumstances of that particular precinct -- but I won't get into that here.) Finally, King would tally each of the narrow precinct ranges he'd come up with to produce an estimate for how blacks voted in the district as a whole.With the help of a National Science Foundation grant, King labored "day and night for two years" to refine his new technique. This spring he presented it in a book published by Princeton University Press. King's method has withstood over 17,000 tests and is now being used in numerous court re-districting cases. But the courts were merely the first to appreciate the usefulness of his discovery. After King set up a web page that offers users free software to implement his method, the response confounded his expectations. "I recently came into my office to use my computer and it was so slow I couldn't do anything," he recalls. That's hardly surprising. The last time King checked, his web site was getting 35,000 hits a week.Alien Invasion?For years, academics seemed strangely unmoved by the heated immigration debate raging beyond their ivory walls. For once this otherwise fractious community was in near-total agreement: Immigration, went the prevailing wisdom, was good for the economy. Right-wing raving about "those damn fahrners" stealing American jobs was wrong, period. "It was just widely accepted that there wasn't much immigrant impact on wages," explains UC Berkeley Professor David Card. What little research was done on the subject supported this view. A 1990 study by Card, for instance, found that the massive influx of low-skilled Cubans arriving in Miami on the 1980 Mariel boatlift had very little effect on the wages of even the city's low-skilled natives. At first glance, an excellent -- and extremely comprehensive -- 400-page study of immigration released this past May by the National Academy of Sciences looked like just another confirmation of the general consensus. Most news accounts played up the study's finding that, on balance, immigrants add up to $10 billion a year to the economy (largely by providing cheap labor that lowers the cost of goods and services). That $10 billion figure sounds impressive -- until you compare it to our $8 trillion GNP. What's more, the study found that, at least for the first 23 years after their arrival, immigrants actually cost the government more in public services than they give back in taxes. In other words, native taxpayers are picking up the slack. But the real blow from immigration is delivered to those who can least afford it: poor, low-skilled American workers. This is largely because, since the mid-1960s, the immigrants arriving in the United States have been progressively less educated. By increasing the supply of low-skilled workers in the country, immigrants drive down wages for everyone in this group. But what about that 1990 Card study showing a negligible impact on the wages of low-skilled workers? Three economists have a bold new explanation. Harvard's Lawrence Katz, George Borjas, and Richard Freeman (all of whom were also co-authors of the National Academy of Sciences study) argue that to measure the impact of immigration you have to examine its effects on low-skilled workers not just in the city where immigrants live, but in the nation as a whole. That's because the competition from low-skilled immigrants flooding into cities like Miami forces the city's low-skilled natives to move to cities with fewer immigrants, like Atlanta. But that means Atlanta now has a larger supply of low-skilled workers than it would have had otherwise. As a result, wages in Atlanta are driven down, even if wages in Miami (which has the same number of low-skilled workers as it would have had without immigration) stay the same. Using this approach, Borjas, Freeman, and Katz determined that between 1979 and 1995, immigration has caused the average hourly wage of American high school dropouts to fall by about 4.5 percent.Borjas et al's reasoning is supported by data gathered by University of Michigan demographer William Frey. But David Card says his data doesn't match Frey's, and he's skeptical of Borjas et al's argument. And so the controversy rages on.The jury may still be out on questions ranging from the impact of immigration to the benefits of welfare reform, but one thing is already certain: Neither liberals nor conservatives can rely on today's social science research to provide automatic support of their views.