How does the structure of U.S. teacher unions compare with that of other developed countries? What does this structure imply for the performance and flexibility of our schools? A comparative survey shows U.S. teacher unions just two major ones form an unusually concentrated oligopoly: 

• European countries tend to enjoy four or more teacher unions, representing a variety of ideologies, teaching methods, and religious orientations.

• Some countries restrict political action by public sector unions, or have teacher choice systems for dues, or parental choice systems for schools, that encourage diversity and diffusion among teachers unions.

• Some Asian countries have low membership rates of 40% or less, or outlaw true teacher "unionism" altogether. Examples include Singapore, Taiwan, and Korea.

• Countries with low "union concentration" whether because of diverse union choices, limits on union political activity, or low union membership altogether perform better on standardized reading, math, and science tests.

• Countries with union oligopoly structures where a small number of ideologically and operationally homogenous unions wield great political power tend to have more rigid school systems that resist reform.

• Thus, policies that would actually increase union concentration, like a proposed merger of America's two major teacher unions, would hurt performance and reduce flexibility.

• Policies that would lower the barriers to entry for new teacher associations might provide a healthy degree of competition and diversity to U.S. education. They would tend to make the system more responsive to parents and to institutional reform.

Teachers unions as a dominant player, in a public education sector that is itself dominant, are a very recent thing. 

ONLY since the late 1950s have federal and state spending crept close to local and private spending on education. 

In that period also, teacher union membership moved from a minority in most states, and bare majority in a few, to its present position making unions the dominant bargaining and political entity in education, with membership rates of more than 75 percent of public school teachers, and a bargaining representation rate of 82 percent._* 

This rapid change and growing role in running our children's schools produces some natural confusion, frustration, and alienation by parents. Are teachers unions "to blame" for the decline in education that appears to accompany their emergence as oligopoly institutions? 

The concentration of so much of the teacher labor market into two bodies would constitute an "oligopoly" or "monopsony" in most economics texts. 

Or is this unfair? After all, many countries and U.S. states with very high levels of union membership, staunch collective bargaining rules and rates, and other traditional measurements of union strength out-perform the U.S. (or, within the U.S., out-perform other states) at math, reading, science, and other standardized performance tests. 

RECENT WORK by a number of scholars, most notably Professor Myron Lieberman of Bowling Green University, suggests there is some question whether public sector unionism and collective bargaining are even appropriate for the school environment. Pro-union academics, and even such reformist education leaders as AFT president Al Shanker, have taken up the pro-union cause with equal vigor, if less depth. 

Charles Kerchner and Douglas E. Mitchell, two scholars of teacher unionism, put it this way in their survey: 

"In the space of a quarter-century, teachers unions in the United States have changed public education its governance, school organization, and even the nature of teaching work. The transformation has been as swift as, and more complete than, the massive industrial changes brought about by... the spread of unionism in the 1930s." (The Changing Idea of a Teachers' Union, New York, Falmer Press, 1988, page 2.) 
 

THE FIRST characteristic we compared, both internationally and on a U.S. state-by-state level, concerned variations in collective bargaining rights and practices. But the states which outlaw collective bargaining tend to be those with low per capita income, begging the question whether poor performance reflects lack of resources or bargaining policies. And there is little variation regarding bargaining per se on an international level. (More so on right to strike and teacher pay.) Such variations as exist don't seem to correlate strongly with performance on standardized tests, though if anything, they rather tend to disprove the thesis that bargaining is a plus for educational performance. 

Figure 1 compares various measurements of collective bargaining including collective bargaining on a simple yes-or-no scale and the right to strike. These are expanded from information in scholar Bruce S. Cooper's Labor Relations in Education. Also included is an index of teacher pay as a share of per capita GDP as calculated by the AFT itself in a 1993 study, "How U.S. Teachers Measure Up Internationally." This seems like one reasonable way to measure collective bargaining clout input by the outcome of wages, though as we have noted, it is not necessarily in the realm of teacher pay or narrow bargaining rights that U.S. teacher unions enjoy such distinctive power. 

(It should be noted, however, that such a system will tend to make countries with very low birth rates look as though they are paying more, and those with high birth rates, relatively less. There will be a similar affect for countries with relatively low or high private sector unemployment rates. Both factors tend to overstate how much Europe, with its stagflationary environment since the 1970s, pays teachers, and understate how much the U.S. and others do.) 

THE COUNTRIES are listed in order of their performance on recent standardized reading, math, and science tests, which we have indexed into a composite score. (We re-based reading scores to a base of 100, scaling variation from the midpoint to the comparable variation from midpoint for the 100-based scores; then averaged available scores.) This facilitates a comparison between different bargaining factors. A high concentration of strong or weak bargaining arrangements at the top or bottom of the list would suggest some connection between them and student performance. A more or less random distribution of those factors among the high- and low-test countries weakens the case that collective bargaining as such is a strong factor, either plus or a minus. 
 

Thus, of the five top-scoring countries on standardized tests, four have no or only some collective bargaining for teachers. The same, however, is true of three of the bottom five countries. Only one of the 11 countries in between Switzerland, with its traditional low-regulation labor market has only "some" collective bargaining, and all other ten have full bargaining, most at the national level. 

Grouping by test scores, one finds a similar story of weak or non-existent correlation. Among the top five countries by test score (South Korea, Taiwan, Sweden, Japan, and Singapore) the teacher pay index is 148. (A teacher of medium seniority typically earns 148 percent of the country's per capita Gross Domestic Product.) In the bottom five (Portugal, Spain, Italy, Germany, and Great Britain) teacher pay is 155.4. 

Thus there are some admitted gaps in the case that collective bargaining harms education, at least in terms of the measurable outcome of test scores. There are also problems, however, for the teachers unions and their advocates themselves. These problems become even larger when one examines the claims made by the proponents of collective bargaining on its behalf. 

ALBERT SHANKER, president of the American Federation of Teachers, provided an instructive example in one of his advertised commentaries in The New York Times (July 21, 1996). Although it does not cite Lieberman, Shanker's article appears to be a response to his work, particularly as regards bargaining. 

"The people who makes this claim," (that unions harm schools), Shanker writes, "don't explain why most of the countries with successful school systems like Japan, Germany, and France also have strong teacher unions." 

Shanker doesn't define "strong" unions or "successful school systems." Nor does his assertion appear to fit a reasonable definition on any level. Consider: 
 

• Japan's last teacher union president complained of state "repression" and called Japan's system one of the most "anti-teacher" in the world. Then, about 38 percent of Japan's teachers were union members. Now less than one teacher in three is a union member. The rate has declined for 20 consecutive years. 

• Germany, meanwhile, is hardly a top-performing education system. In fact, the failure of the German economy to innovate, as measured for example by its waning share of industrial patents and stagnant unemployment rates, has been a source of angst for years. (E.g., Amity Shlaes, Foreign Affairs, Spring, 1994.) 

• France has performance scores modestly above average. France, though, employs significant state support for Catholic, Protestant, and other private schools. Both the AFT and the NEA have both been harsh opponents of such parental choice proposals for the United States; if France's schools are indeed "strong," perhaps Shanker would support introduction of the French system here. 

"U.S.-style collective bargaining," Prof. Kerchner writes, "is not a mainstay of many of the rest of the world's teachers' unions." Often, for instance, teacher pay is set as part of an overall civil service scale. (In Cooper, page xvii.) 

Indeed, in Shanker's table (ranked by test performance), the top five countries are Taiwan, Korea, Russia, Switzerland, and Hungary. 
 

• Taiwan and Korea have de facto union membership rate of close to zero. (Korea's weak teachers association has more than 50 percent membership, has only recently taken on some functions of a "teachers union" in the American sense.) 

• Switzerland, Hungary, and Russia have had parental choice features going back to the mid-1980s. Switzerland has strict constitutional provisions limiting public sector union political clout. 

"Does this prove that collective bargaining leads to higher student achievement? I wouldn't go that far...." 
 

Figure 3 compares union membership for primary and secondary teachers, combined, by country. 

There are broader variations vis-a-vis Japan, Singapore, and other Asian tigers. But for these, the comparison of figures itself is more doubtful. For instance, we've listed South Korea twice in the table to reflect the fact that the country has a teachers association with about 65 percent membership. Nominally a union, the Korean Teachers Association has no right to strike; engages in wage "discussions" twice annually that would not be considered collective bargaining by most Western union leaders; is not highly active politically; and is highly proscribed in all its activities by virtue of Korea's tough laws on public sector unions. 

So should Korea's nominal 65 percent union membership lead it with its high performance rate on tests to be ranked among the strong union group? Or should Korea's unfriendly policies towards teachers unions as bargaining and political entities lead us to count the KFTA as a mere association and actual union membership as close to zero? It would certainly be news to Korean teachers to learn that they enjoy twice as much strength as teachers in Japan. 

To avoid making arbitrary judgments either way, we simply left Korea out of our comparison base regarding membership and performance. The impact of including Korea would be small, but if our judgement about where it belongs is fair namely, as an association it would tend to raise the performance of the low-membership countries. 

WITH THESE caveats in mind, the following table nevertheless suggests some interesting comparisons. 
 

A regression analysis of the teacher union membership rate to math scores produced an r-squared factor of 0.681; to reading scores, an r-squared of 0.516; and to overall scores, an r-squared of 0.582. In short, differences in union membership rates correlate with 51 percent to 68 percent or more of the differences in test scores. Much more careful, multi-variable regressions would be required to draw any dramatic conclusions from this fit between performance and unionization rates. The single-variable regression, however, is at least consistent with a hypothesis that membership matters. 

The differences in union policy, and there are substantial ones, are much greater in areas outside the bargaining and membership framework. Mandatory dues, and their extent, vary widely. So do the degree of union choice (or, in the U.S., heavy concentration) offered to teachers, and the presence (or absence) of market structures in the surrounding education system as a whole. 

Furthermore, there is the whole question of union and education structure beneath the visible surface of aggregate strength measurements. The number of unions that make up a given membership or bargaining percentage may be as low as one (China) or two (the United States) or as many as several dozen. There is wide variation in terms of ideology, religion, political activism, and other factors and considerable variation in the degree of variety within a given country for these factors. 

A BROADER view of "union strength," then, might yield greater insight into the education culture than simple bargaining coverage or membership rates as single factors. Perhaps the elusive concept of "union strength" is not the most useful point for comparison. 

A common-sense survey of at the systems suggests why. 
 

UNTIL A 1992 consolidation, Australian educators might belong to any one of more than 100 employee unions, according to the Australian Financial Review. Even after the merger, Australia has several dozen regional unions providing bargaining, clustered together under four major umbrella unions. There are significant Catholic and protestant teachers associations that act as a lobbying force in favor of funds to private schools. Australia also has a school choice system, which, as we will discuss, can sometimes have important implications for the structure and impact of teachers unions. 

Obviously, the political and social combinations resulting from this policy mix are likely to differ substantially from that of Canada, where compulsory fees for a single dominant union are the norm; or Britain, where teachers unions have managed in recent years to limit some school choice reforms, enact tight rules for hiring foreign teachers (who often will work for less), and block international performance testing and one-man one-vote for the union structure itself. Yet Canada's union membership rate for teachers is comparable, at 78 percent, to Australia's 62 percent. And Great Britain's rate, 60.4 percent, suggests its unions are slightly "weaker." 

Denmark, meanwhile, has several major teachers unions and a half-dozen with more than two percent membership. "In Denmark," as one teacher put it, "there is little question whether you are likely to belong to a union, but a question which one." Yet its union membership rate is more than 90 percent, and collective bargaining provisions, if anything, are stronger than those in the United States. Traditional strength measurements might rank Denmark well above the U.S., but its education structure provides greater teacher and parent choice, a less oligopolistic structure. 

Here we see the number of teachers unions, by country, that enjoy approximately 3 percent or more membership of teachers as a whole; and, as a measure of diversity, which countries have significant religious-school unions. 
 

The United States is almost alone in the OECD group in its degree of union concentration. Canada equals us in terms of number of unions, but the considerable degree of organization of and support for Roman Catholic school teachers, especially in Quebec and Ontario, provides a diversity we lack. 

REGARDING union concentration and diversity, the U.S. does resemble some of the highly successful systems of Asia. They, too, have little tradition of union organization among private or religious school teachers, or of public support for those schools, and they generally have two unions or even one. (Countries with significant Moslem or Christian populations are an exception.) 

But note: The one or two unions in Japan, Korea, Taiwan, and Singapore are weak and restricted politically. Three of the four have membership rates of less than 35 percent. As recently as 1993, Korea arrested more than 100 teachers for "illegal union activity," reducing their sentence only after they spent time in jail. When Singapore last had a significant teacher work stoppage, the president of the United States was Gerald R. Ford. The last major strike of any kind was during Ronald Reagan's second term. 

Looking at union diffusion and diversity gives one a picture so much of "union strength" as of union concentration along the lines of the "industrial concentration ratios" used by economist Richard Caves and others to measure competition in industries. (American Industry: Structure, Conduct, Performance, Prentice-Hall, 1977.) They measure the hegemony of a few union elites. 

This notion might also be referred to as "teacher choice," meaning the consumer freedom enjoyed by individual teachers against, if need be, the aggregated influence of teachers as a group, or as represented by their union leadership. And it focuses not just on the nominal bargaining relationship, but on political leverage. 

Union leaders like to point out, and scholars of comparative education have noted, that the U.S. is a distinctly large country with an unusually decentralized education system regarding school budgets: More than 10,000 local union bargaining units and a corresponding number of school boards and other management. All this makes for a considerable, and in some ways admirable, degree of decentralization. 

Still, while a teacher may choose to join a somewhat different bargaining unit if he is willing to change jobs he will nearly always be represented either of the American Federation of Teachers or the National Education Association. That, with the small degree of difference in ideas and operation, is the sum and circumference of his choices. 

Remember, moreover, that we are not only or even primarily concerned with the impact of U.S. teacher union structure on the bargaining process. If we were, we would pay a good deal more attention to teacher salaries, the gap between what (unionized) public school teachers earn and (generally non-union) private and religious school teachers do. 

Rather, our concern is as much or more with the impact of unionism on the culture and politics of schools on the degree of choice for teachers, parents, and others in the system. This means choice between different national institutions, ideologies, religious leanings (or lack of any at all), or operational style. (Some teachers unions surveyed, for instance, forgo the right to strike not because of statutory prohibition but as a matter of their internal constitution and professional mission.) 

The U.S. combination of strong union membership rates and high concentration and low diversitymakes it truly distinctive. 

Figure 6, on the next page, compares the degree of union concentration or "oligopolization" for the countries studied with performance on standardized tests. For each country, we assigned a rating of 0 through 5 to the concentration of union power 5 being a high degree of oligopoly or monopoly power, 0 being none. The ratings were in the final sense a subjective judgment, along the lines of AdTI's previous work extending the Freedom House survey back through selected years to the late 19th century. (See The Democratic Imperative, New Republic Books, 1989.) But it is based on six inputs, some of them quantifiable: 
 

• Union membership, from 0 percent to 100 percent. (0-1 points.) 

• Union density, chiefly the number of teachers unions with membership above the 3 % level. This constitutes an element of teacher choice, and diffusion of power for union elites. (0-1 point.) 

• Union diversity, including ideological-partisan diversity, religious diversity, educational approach or "ethos" differences, and others. We define union diversity in these matters by the union's official activities as such, and not by membership diversity if it is not reflected in policy. That is, although more than 30 percent of the members of the National Education Association are, according to one poll, Republican, the NEA consistently endorses Democratic Party candidates; its PAC directs 99 percent of funding to Democrats; and its platform is substantially in line with that of America's center-left Democratic Party. The same observations hold for America's other major union, the AFT. Hence the United States scores a 0.1 for ideological and religious diversity. (0-1 points.) 

• Teacher pay as a share of GDP and teacher pay in the union sector versus pay in the non-union sector. These are a plausible measurement-by-outcome of union bargaining strength. (0-0.5 points). 

• The degree of compulsion in paying dues, ability to direct them to or away from desired activities, and other matters regarding government policy towards unions, (0-0.5 points.) 

• Degree of freedom in the demand side of the education market is there school choice, and can parents readily choose between schools (including private) with a minimum of economic penalty? (0-1 points.) Whatever the direction of causality, systems that enjoy school choice seem to enjoy a large element of union diversity as well. Systems with low or no choice tend to be those with oligopolistic union structures. 

The point is not to prove that oligopoly teachers unions directly hamper school performance, but to suggest, by comparison to bargaining data for instance, that structure matters. This relationship deserves more careful study by scholars. 
 

Yet the low-scoring, high-oligopoly group's per-pupil expenditure was 174 percent of the high-performance group. 
 
THIS EFFORT perforce involves some arbitrary judgements and close calls. But this sort of analysis is important if analysts are to get beyond good-and-evil for teachers unions and understand how and they operate for good and ill, in ways that can frustrate or promote education efficiency and reform. 

It may be politically vital. The fact is, parents and teachers both have legitimate interests in our schools. An attempt to ride over one or the other completely is ill-conceived, and will face a difficult time politically. Our philosophy of citizenship, like that of Alexis de Tocqueville, is that parents come first. But to take a phrase from the book of the union leadership, teachers "should have a strong role." 
 

Policies that tend to promote or broaden teacher choice are not universal, but there are examples. In most cases as with the more familiar concept of school choice, interestingly they seem to have evolved. Enactment of such reforms was a response to perceived rights of conscience and free association, and not necessarily part of a self-conscious effort to "weaken" or "rein in" teacher unions. Often they were in response to demands by teacher rank-and-file. 

The literature on such policy variations is thin, and it's hard to get union leaders and policy-makers to discuss them at all, let alone across the time zones and language barriers encountered doing international work of this sort. 

Figure 7 notes mechanisms and developments that tend to erode union concentration and promote diversity. It is illustrative, not comprehensive, and includes examples of interesting policy variations in other countries and in various U.S. states. 
 

Many of the most notable policy efforts have aimed at, or had the effect of, promoting a different character of unionism and a different structure of the union "market," with relative indifference to "more power" or less in bargaining for unions as a whole. 
 

At the recent convention of the American Federation of Teachers, AFT President Al Shanker made this notable statement about the proper objectives of a teacher union member: "It is as much your duty to preserve public education as it is to negotiate a good contract." (August 2, 1996.) 

This was hardly a slip of the tongue. It is, in fact, the animating spirit of the leadership of both the AFT and the NEA. Both place great stress on a political defense of the system meaning, public schools as against mere workplace issues. It could be argued that in recent years the NEA and AFT have put proportionately more effort into protecting the public school system from perceived encroachment (from vouchers, parental rights, or other proposals) than they have into improving the wages or work conditions of their members. If this sounds like a harsh judgment, consider it in light of Shanker's instruction to the union's activists that they ought to stress this political effort at least equally. 

SCHOOL CHOICE makes an interesting point of reference for judging union political orientation. For one thing, school choice is now widespread in the developed world, even in pockets of the United States and Canada. For another, it is the education reform of choice by most center-right parties. School choice formed the thrust of education reform in Sweden (1992) and in Great Britain under former minister of education Margaret Thatcher (1981, 1988). Ronald Reagan (1980), George Bush (1988), and Robert Dole (1996) all campaigned on it. 

Finally, school choice is the bete noir of many union officials, at least in the United States. (Interestingly, the reception has been much less hostile among many countries of Europe and Oceana, as a forthcoming paper will discuss.) Much of the interest in restructuring collective bargaining or other aspects of policy towards teacher unions comes from the fact that education reform proponents are frustrated with the ability of union officials to frustrate school choice plans in Congress, the state legislatures, and the courts. 

When Dole took the unusual step of criticizing the teachers unions as such during his acceptance speech to the Republican National Convention, he threw a spotlight on the issue of union clout that virtually ensures the subject will be with us for some time. Many forthcoming debates will be filled by words like "bargaining," the "right to strike," and "relative wages." Often, though, the real bone of contention will be the desire of some to change the school system to a more market-based, parent-as-consumer approach; and the fervent and conflicting desire of others to, as Shanker put it, "protect public education" as we know it. 

To many, the salient impact of teacher unions has not been to win significantly higher wages for teachers. More important, to most, is the impact on education itself. Cooper writes: "[U]nionization of public school teachers across the world is a movement of great importance a revolution of sorts. It has redefined the way school organizations work; the process of decision making; hiring, evaluating, and firing of staff." 

If reformers desire to "fight the unions" in order to achieve some other, systemic objective, they need to examine what sort of union structure seems to promote, or at least be most consistent with, the choice schemes that they desire. "Who wills the ends," as St. Thomas More reportedly said, "yet does not will the essential means to the ends, makes practical war upon the ends themselves, indeed against his very soul." 

THE FIT between different union structures and school choice is thus of topical interest, as well as offering insight into the flexibility of various structures, by testing their adaptability to a common, but sometimes bitterly disputed, education reform. 
 

Instead, most systems fall on the right combining "strong" unions, from a traditional view, with school choice; and "weak" unions, by traditional measures, with little or no school choice. (Russia and Hungary are omitted from the table as systems in transition; the United States and Canada are judgment calls: Boston University's Charles Glenn, a leading expert on choice internationally, would be more likely to classify one or both of these as "major choice" systems. We say the support given to choice is too small financially, institutionally, and geographically for them to qualify.) 
 

Reformers, however, should note the empty column on the lower right. At the least, high union concentration (low "teacher choice") has not yet been combined with significant parental choice. And, at the same time, many countries have managed to combine "strong" unions in the bargaining sense with a large degree of parental rights and choice, provided there is significant union diversity and choice. "Weak" unions, in terms of bargaining and membership, meanwhile, are no guarantee of choice: Korea, Taiwan, Singapore, Japan. In fact, they seem to go with non-choice systems. 

In countries such as France and the Netherlands, which have had school schemes since the mid-19th century, the chicken-egg question of teacher choice and parental choice is most acute. Other countries, however, have enacted choice schemes, or significantly expanded choice options, in recent years. These include Sweden, Great Britain, Australia, and postwar Germany. In each case there was a high degree of union diversity and teacher choice before parental choice was established or significantly expanded. This diversity, of course, at once weakened the monolithic forces opposed to reform, and established some groups in favor of choice, such as financially stressed parents in the un-funded private system and the lower-paid teachers working with their children. 

"We have been for the choice system," says Sven Kinnander, international representative of the Lararnas Riksforbund teachers union in Sweden. "We welcome these schools because it creates the possibility of more than one employer for teachers. Before 1992 we had very few of these free schools." Teachers also benefit, he argues, from the wider choice of workplace environment. "We now have different pedagogical orientations," and more religious schools. 

(Post-war Germany and post-Revolutionary France under the education reforms of 1830 and 1849 did not have significant union diffusion or even organization. Both, though, suffered through something close to a collapse in the viability and legitimacy of national institutions. Similar developments may be at work today in Eastern Europe. Only the most extreme education reformers, of course, would advocate creating a national collapse in order to facilitate parental rights and school choice.) 
 
 

In the piece, Shanker praises such systems as Germany and Japan which, as we note above, are not especially strong havens of union rights. What's notable, though, is that Shanker doesn't mention such countries as Denmark, the Netherlands, or Sweden. All three perform well on some standardized tests. All have been implicitly praised in past AFT studies and press releases for spending lots of money on government schools. They also have strong rights of collective bargaining, and generally well-paid teachers. 

All three, however, have parental choice systems and parental rights constitutional guarantees. And all have significant degrees of union diversity and teacher choice. Apparently, these factors dampened the AFT's enthusiasm for their spending levels and bargaining system. 

A former Jersey City official who has worked with the teachers union says that: "I would often have gladly given them everything they wanted on pay, if we could have had systemic reform.... But I'm not sure that was possible because their first, primary interest was in keeping the system the same." This comment is remarkably symmetrical to Shanker's statement about union priorities. 

Reformers may wonder: If union leaders want to protect the current system of education more than the pay and conditions of their workers, are they more interested in teacher union power as such, or in the power of the teacher-union structure of today? 

One of the more informative and balanced reports on teacher unions internationally is Bruce S. Cooper's Labor Relations in Education. Most of the chapters are written by teacher union leaders, including one on the United States by Shanker. The introduction, by Charles Kerchner, observes: 

"Given the close alliance between the NEA and the Democratic Party in the United States, it is interesting to contemplate what would occur if an avowedly Republican education alternative were to develop." Or, one might add, a union of private-school teachers; a union for teachers in Jewish or Roman Catholic schools; or one for inner-city teachers favoring parental choice scholarships. Or, for that matter, a non-ideological, non-sector-specific union but one that would competewith the NEA and the AFT? (A la PAGE in Georgia.) 
 

Is there any doubt that in their heart of hearts, the leaders of both major unions would favor a pay cut, and "weaker" unions, over a stronger position for unions as a whole but one in which they are less dominant? That they treat promoting teacher rights as less important than denying parental choice? 

The leaders have all but made this syllogism explicit. An NEA resolution that says vouchers would "destroy public education." And the AFT president admonishes union elites to "defend public education" as staunchly as they work for "a good contract." It follows that the animating purpose of the NEA and AFT, in the eyes of its leaders, is fundamentally political and statist. Pay and conditions for its rank and file are no more important, and may be less. 

It is not too much a leap to believe that, as the NEA and AFT hierarchy care more about the school system than about teacher contract provisions, so they care more about protecting their own degree of exemption from competition. The mutual leadership know what they are doing when they place great emphasis, as at both 1996 conventions, on reducing even the small amount of competition that now exists between the NEA and AFT. Their plans to merge the two organizations, so similar in ideology and operation, into one, are not idle or purposeless. 
 

MUCH WORK remains to be done by comparative education scholars to test these and other tentative hypotheses. From the evidence we've examined, however, there are grounds for education reformers to direct their research, analysis, and prescriptions in certain directions. 

Efforts to simply "weaken" or "oppose" teacher unions have potential political drawbacks and unintended consequences. There are some reasons to doubt an "anti-union" policy by U.S. or U.S. state officials in and of itself could significantly improve performance. Such a policy might even prove to divert attention and resources from more systemic battles. 

Such policies might, of course, promote school productivity, if it is deemed that teachers can and should work for lower wages. It might have political benefits, for example by helping reformers to overcome union elite resistance to particular policy improvements such as school choice. 

But even if "union clout" is driven down in some overall way, the data presented here suggest there may be scant or no improvements in education performance, and perhaps no added flexibility regarding policy reforms either. Union bargaining clout could be thwarted, for example, without disrupting at all the quasi-hegemony of major teachers unions in the political culture. Membership rates could be lower as well without serious disruption to politics as they are. 

Suggestive evidence of this is the non-relationship between teacher salaries and student performance. Other things being equal, teacher salaries should in some way reflect the strength of a union in its bargaining relationship. In turn, they should serve as a proxy for union "strength." Insofar as union strength is a good thing, as per the union's elites, or a bad thing, as per neoconservative and liberal critics of the union, there should be measurable impacts, good or bad, for students and schools. 

If union leaders are right that American schools mainly lack resources, then high teacher salaries (and the power relationships they reflect) should lead to superior performance. If critics of "unionism" per se are right, then high salaries (and the strong union hand they are a sign of) should correlate with poor performance. But the performance of students is more or less randomly distributed against teacher salaries. 

UNION power vis-a-vis administrators certainly creates difficulties, and the collective bargaining process could be improved. It may be that such improvements would have a major impact on education performance; but we can find no evidence of this from this admittedly brief survey of international teacher union policies. 

The power of a few union officials to block reforms and, in effect, set policies for large parts of the system is much harder to measure. But when the effort is made, it appears to yield important insights into how our schools are run, and therefore, what they produce. 

As an example, we adapted the industry concentration ratio, as used by Richard Caves and other analysts of oligopoly commercial behavior, to the structure of teachers unions in the United States and other countries. While this involves interpretation of data, it at least can start from a hard data base. 

REFORMS should focus to at least some extent on teacher choice with a careful distinction between the power of teachers as a group of several million diverse individuals, and the particular institutions which happen, now, to represent them. A further distinction must be drawn between unions and the particular elites who lead them. 

Starting from this conceptual base, several reforms would appear to make sense. Most of them, notably, have the effect of strengthening teacher choice in the workplace, in the surrounding politics of how the education workplace should be run, and at the bargaining table.