¹ For examples and further discussion, see Pontusson, , “From Comparative Public Policy to Political Economy: Putting Political Institutions in Their Place, and Taking Interests Seriously,” Comparative Political Studies 28 (Spring 1995)CrossRefGoogle Scholar.

² Kitschelt, , The Transformation of European Social Democracy (New York: Cambridge University Press, 1994)CrossRefGoogle Scholar.

³ Piven, , “The Decline of Labor Parties: An Overview,” in Piven, , ed., Labor Parties in Postindustrial Societies (New York: Oxford University Press, 1991), 1Google Scholar.

⁴ Cf. also Ivor Crewe, “Labor Force Changes, Working Class Decline, and Labour Vote,” in Piven (fn. 3).

⁵ The three exercises explore different observable implications of the same theory. Taken together, they are meant to compensate for the absence of a large data set against which to test the theory. Cf. King, Gary, Keohane, Robert, and Verba, Sidney, Designing Social Inquiry (Princeton: Princeton University Press, 1994)Google Scholar.

⁶ These are the only two cases for which I have been able to locate adequate time-series data on the distribution of manufacturing employment by plant size. Unless otherwise noted, the plant data employed below actually refer to “establishments,” sometimes also called “manufacturing units.” An establishment, defined as a single location of production facilities, may consist of several plants, factories, or workplaces.

⁷ I do not use policy measures in the analysis of Swedish and British time-series data because such measures (e.g., employment rates) are not readily available for the period prior to 1960.

⁸ Cf. Rueschemeyer, Dietrich, Stephens, Evelyne Huber, and Stephens, John, Capitalist Development and Democracy (Chicago: University of Chicago Press, 1992)Google Scholar, chap. 2.

⁹ Martin, Andrew, “Is Democratic Control of Capitalist Economies Possible?” in Lindberg, Leon, ed., Stress and Contradiction in Modern Capitalism (Lexington, Mass.: D. C. Heath, 1975)Google Scholar.

¹⁰ For references to Marx and Lenin, see Ingham, Geoffrey, Size of Industrial Organization and Worker Behavior (Cambridge: Cambridge University Press, 1970), 15Google Scholar; and Goldfield, Michael, The Decline of Organized Labor in the United States (Chicago: University of Chicago Press, 1987), 150Google Scholar.

¹¹ In each of these formulations, the term wage earners may be substituted for workers. For the most part, the following analysis is oblivious to the distinction between blue-collar (manual) and white-collar (nonmanual) employment.

¹² Auer, and Fehr-Duda, , “Industrial Relations in Small and Medium-Sized Enterprises: Evidence from Six Countries,” in Auer, and Fehr-Duda, , eds., Industrial Relations in Small and Medium-Sized Enterprises (Luxembourg: Office for Official Publications of the European Communities, 1989)Google Scholar.

¹³ For further discussion, see Ingham (fn. 10).

¹⁴ Wallerstein, , “Union Organization in Advanced Industrial Democracies,” American Political Science Review 83 (June 1989)CrossRefGoogle Scholar.

¹⁵ For a model of how to estimate plant concentration rates and for empirical results for Canada, France, Italy, Japan, Sweden, the U.S., and the U.K. in the late 1950s, see Bain, Joe, International Differences in Industrial Structure (New Haven: Yale University Press, 1966)Google Scholar, chap. 3. Regressing 1960 union density data reported by Goldfield (fn. 10), 16, on Bain's concentration rates for these seven countries yields a positive and statistically significant coefficient, but the association between these variables is not very consistent (R²=18.3%). Since Canada represents a major outlier in this regression (without it, R²=58.4%), it may be that a larger sample would yield results more consistent with the fact that union density correlates strongly with plant size on a national basis. For evidence on the association of union density and plant size within countries, see Goldfield (fn. 10), 149–52; Auer and Fehr-Duda (fn. 12), 30; and Loveman, Gary and Sengenberger, Werner, “Introduction: Economic and Social Reorganization in the Small and Medium-Sized Enterprise Sector,” in Sengenberger, , Love-man, , and Piore, Michael, eds., The Re-emergence of Small Enterprises (Geneva: International Institute for Labour Studies, 1990), 39Google Scholar.

¹⁶ Stephens, , “Industrial Concentration, Country Size and Trade Union Membership,” American Political Science Review 85 (September 1991)CrossRefGoogle Scholar.

¹⁷ Crewe (fn. 4), 32–33.

¹⁸ Marsden, “United Kingdom,” in Sengenberger, Loveman, and Piore (fn. 15), 233.

¹⁹ Auer and Fehr-Duda (fn. 12), 20–21.

²⁰ Loveman and Sengenberger (fn. 15), 33–37.

²¹ Przeworski, and Sprague, , Paper Stones (Chicago: University of Chicago Press, 1986)Google Scholar.

²² For most countries, data for establishment size for services simply do not exist, but the German data in Table 5 below support this argument.

²³ Cf. Clegg, Hugh, Trade Unionism under Collective Bargaining (London: Social Science Research Council, 1976)Google Scholar, chap. 2.

²⁴ The Swedish data also include all plants, whereas the British data exclude plants with fewer than eleven employees.

²⁵ In both cases, the evolution of average plant size follows the same general pattern as the evolution of employment share of large plants. Regressing average plant size on the employment share of large plants yields an R² of 57.4% using Swedish (1913–74) data, and 83.9% using British data.

²⁶ Kjellberg, Anders, Facklig organisering i tolv länder (Lund: Arkiv, 1983), 269–79Google Scholar.

²⁷ Cf. Janoski, Thomas and Isaac, Larry, “Introduction to Time-Series Analysis,” in Janoski, and Hicks, Alexander, eds., The Comparative Political Economy of the Welfare State (New York: Cambridge University Press, 1994), 33–34CrossRefGoogle Scholar.

²⁸ The R² of the regression of union density on the employment share of large plants is 95.0%. Readers not too familiar with statistics might appreciate a few words about the meaning of regression results at this point. The important thing is that the coefficient is a measure of the strength of the relationship between two variables, and R² is a measure of the consistency of the relationship. The coefficient refers to the unit increase of the y-variable associated with a one-unit increase of the x-variable, and thus describes the slope of the regression line. R² represents the amount of the variation on the y-variable that is explained by the regression of the y-variable on the x-variable(s), and tells us about the spread of observations around the regression line. A weak relationship (a coefficient of low value) may be very consistent (a high R² value). Following conventional usage, a relationship between two variables is here described as statistically insignificant if the value of the coefficient is lower than the value of the standard error of the coefficient (i.e., at-ratio of less than 1).

²⁹ Cf. Larry Isaac, Susan Carlson, and Mary Mathis, “Quality of Quantity in Comparative/Historical Analysis,” in Janoski and Hicks (fn. 27).

³⁰ The vote shares used in these regressions were taken from Korpi, Walter, The Democratic Class Struggle (London: Routledge and Kegan Paul, 1983), 237Google Scholar; and Krieger, Joel, “Britain,” in Kesselman, Mark and Krieger, , eds., European Politics in Transition, 2d ed. (Lexington, Mass.: D. C. Heath, 1992), 71Google Scholar.

³¹ Korpi (fn. 30), chap 1.

³² For further discussion of Sweden as a case of reformist labor hegemony, and the contrast with the U.K., see Pontusson, , Swedish Social Democracy and British Labour, Occasional Paper no. 19 (Ithaca, N.Y.: Western Societies Program, Cornell University, 1988)Google Scholar.

³³ The time periods used to calculate average vote shares (1965–71, 1974–81, and 1985–92) were selected so as to include at least two elections for each country. Average vote share over several elections is used because the outcome of any given election is influenced by a host of particularistic factors. Union density is a far less volatile statistic. The last column of Table 1 (and Table 3 below) should be ignored for the time being. See Appendix 1 for the data on vote share and union density used to calculate the mobilization index.

³⁴ Esping-Andersen, , Three Worlds of Welfare Capitalism (Princeton: Princeton University Press, 1990)Google Scholar.

³⁵ Regressing policy dominance on mobilization yields an R² of 46.2% for 1970, 50.1% for 1980, and 72.2% for 1990. The high R² value for 1990 is attributable to the decline of the mobilizational capacity of the Austrian labor movement in the 1980s (see below).

³⁴ Cf. Castles, Francis, The Social Democratic Image of Society (London: Routledge and Kegan 1978)Google Scholar; and Esping-Andersen, , Politics against Markets (Princeton: Princeton University Press, 1985)Google Scholar.

³⁷ Bain (fn. 15), chap. 3.

³⁸ Rowthorn, , “Corporatism and Labour Market Performance,” in Pekkarinen, Jukka, Pohjola, Matti, and Rowthorn, , eds., Social Corporatism (Oxford: Clarendon Press, 1992)Google Scholar. For further discussion, figures, and references, see Pontusson, “Trade Unions and the Representation of Worker Interests in Corporatist Political Economies” (Paper presented at the annual meeting of the American Political Science Association, New York, September 1–4,1994).

³⁹ OECD, Economic Outlook: Historical Statistics (Paris: OECD, various years).

⁴⁰ Traxler, , “Austria: Still the Country of Corporatism,” in Ferner, Anthony and Hyman, Richard, eds., Industrial Relations in the New Europe (Oxford: Blackwell, 1992), 271Google Scholar.

⁴¹ Kitschelt (fn. 2).

⁴² As noted above (fn. 25), there is fairly close correlation between average plant size and the employment share of large plants in the Swedish and British time-series data.

⁴³ In each of the regressions reported below, the association between social democratic rollback and decline of average plant size is stronger and more consistent if we use the raw plant-size change figures rather than residuals.

⁴⁴ Elements of the kind of analysis that would be necessary to explain cross-national differences in industrial reorganization are developed by Mjoset, Lars, “Nordic Economic Policies,” International Organization 47 (Summer 1987)Google Scholar; and Robert Boyer, “New Directions in Management Practices and Work Organization” (Revised draft of paper presented at the OECD conference on Technical Change as a Social Process, Helsinki, December 11–13, 1989).

⁴⁵ As with my other variables, change in export dependence is here measured as the ratio of exports as percentage of GDP in 1990 to exports as percentage of GDP in 1970, and change in growth rates is measured as the ratio of the average annual growth rate in 1979–90 to the corresponding figure for 1968–79; all data from OECD (fn. 39).