It seems that each political party has to have some group in society to flog. Perhaps because they got to pick first, the Democrats chose tobacco companies. It may seem risky to attack an industry that provides ample political funds and, in some sense, represents the 25 percent of adults who smoke. Yet the target works wonderfully well in practice, perhaps because most smokers feel guilty about their unhealthy habit.
In a stunning part of his convention speech, Bob Dole seemed to announce that the Republicans had selected teachers' unions as their target. This choice was politically perilous, because education is a lot more appealing than cigarettes. Mr. Dole's key argument, therefore, must have been that the growth of teachers' unions over the past thirty years has contributed to the decline in the quality of education.
How do unions affect educational expenditures and quality? In the view of the unions themselves, they use their knowledge and power to extract more resources for education, which they use to advance student achievement. If this view is correct, then more union power raises educational spending and improves the quality of schooling.
But their critics maintain that teachers' unions seek to enhance the welfare of teachers and have no particular concern for students, parents, administrators, or voters. If so, more union power increases school spending but worsens the quality of schooling.
Who's right? In "How Teachers' Unions Affect Education Production," an article in the August 1996 Quarterly Journal of Economics, Caroline Hoxby of Harvard University tests the two models. Her study used census data, by public school district, from around 1970, 1980, and 1990 on school expenditures, teacher employment and pay, and student enrollment. Teachers were classified as unionized if three criteria were met: a contractual agreement existed between the teachers' organization and the school administration, negotiations took the form of collective bargaining, and at least 50 percent of teachers were members of the teachers' organization. According to the stringent definition, the fraction of students enrolled in unionized public school districts rose from 1 percent in 1963 to 43 percent in 1992.
The most difficult part of the research was separating the effect of teachers' unions on school spending and outcomes from what social scientists call "reverse effects"-for example, a tendency of teachers' unions to arise in districts that are performing poorly. Ms. Hoxby made this separation by using information about the presence of laws that facilitate the creation of teachers' unions. In 1960, collective bargaining by teachers was illegal in many states. By 1970, twenty-three states had granted teachers the right to engage some form of collective bargaining. Nine more states had done so by 1980 and an additional five by 1990. Agency shops (in which unions collect dues from all teachers in a bargaining unit regardless of union membership) or union shops (where all teachers must become union members) were explicitly permitted in two states by 1970, twelve more by 1980, and an additional seven by 1990.
So what are the main findings? First, unionization is estimated to raise per-pupil spending by about 12 percent. Roughly three-quarters of this greater spending takes the form of higher teachers' salaries (estimated to rise by 5 percent with unionization) and lower student-teacher ratios (which fall with unionization by around two students per teacher). These results are consistent with the unions' argument that the higher teachers' salaries and lower student-teacher ratios are needed to improve student performance.
But how does unionization of teachers affect educational outcomes? Here the unions' claims fall apart. Ms. Hoxby measures school quality by the high school dropout rate, the only performance variable that can be derived from the census data for each school district. The key result: Unionization raises the dropout rate by an estimated 2.3 percentage points.
Unionization's ill effects operate partly by lowering the efficiency of school inputs. In nonunionized schools, lower student-teacher ratios and higher teacher salaries lead to better outcomes in the sense of reduced dropout rates. However, in unionized schools, neither student-teacher ratios nor teacher salaries have a detectable effect on dropout rates. It seems that the unions manage to divert extra school inputs into the things they care about-higher teacher pay and the reduced effort required to deal with fewer students-rather than into better student performance.
You may have thought that I was unfair at the beginning to draw a parallel between teachers' unions and tobacco companies, and Ms. Hoxby's results suggest that I was indeed unfair. For consider what would happen if each of these groups were prohibited. If cigarettes were illegal, society would suffer from the maladies that characterized alcohol prohibition in the 1920s and that are associated with the drug-enforcement policy today. Tobacco use would likely decline, but the increases in criminal activity and taxpayer expense would probably overwhelm any benefits from less use. Thus, society would almost surely be worse without tobacco companies.
In contrast, if teachers' unions were prohibited, school expenses would fall, and school quality would rise. The only losers would be the unions and some teachers. Teachers would surely gain, however, in their roles as parents and taxpayers.
One policy implication is that it would be a good idea to reverse the trend of state legislation that favors teachers' unions. But probably more effective would be a universal school choice program allowing competition to the union-dominated public education sector. Teachers' unions know what they're doing when they pour money into campaigns against school choice proposals. Education won't improve until voters become equally informed and make school choice their top priority.