1 9. Jahrgang (1 993), Heft 4 Wirtschaft und Gesellschaft foreseen increase in demand has the opposite effect of increasing union membership and raising the unions' sensitivity to unemployment. The distinction between insiders and outsiders is less clear at the na­ tional level. According to Hersoug, Kjrer and R121dseth (1986), the central Norwegian trade union confederation (LO) has no statistics on unem­ ployed members and no way of deriving such statistics from official sources. Thus it is the national unemployment rate that enters in the LO's calculations of the employment consequences of its wage demands. More generally, we can distinguish between the current union mem­ bership, older workers who have not been employed in the recent periods and new workers who have just entered the work force. Union leaders at the local level are elected, in most countries, by employed union mem­ bers. Electoral considerations pressure local leaders to represent the cur­ rent union membership alone. In the Nordic countries, in contrast, the national union confederations care equally about unemployment among members and non-members, if only because the central confederations cannot distinguish between the two groups. The central confederations also seem to care, to a lesser extent, about new entrants in the labor mar­ ket, perhaps because of the political ties between the leadership of the blue-collar union confederations and the social democratic party. Thus centralized bargainers appear to have a broader definition of insiders than local bargainers. A broader definition of insiders, in turn, leads di­ rectly to a greater willingness to reduce wage demands for greater em­ ployment. 2.4. Other Externalities There are a number of other externalities in the wage setting process that might induce a central wage setter to choose differently than decen­ tralized wage setters. For example, union members may care about rela­ tive wages. It is standard practice to assume that union members only care about their consumption possibilities as determined by their real consumer wage and their security of employment. Yet observers of indu­ strial relations have lang claimed that workers care about wage differ­ entials as weH as wage levels. Workers may strive for status as well as in­ come, and status may depend on relative income (14). Or workers may be concerned with notions of fairness that are derived from camparisans with what others are paid (15) . Suppose, for whatever reason, that union members care about how much they are paid relative to other workers in addition to the standard concerns with wage levels and employment security. If all unians try to increase their wage relative to the wage of others, none will change posi­ tion provided their relative bargaining strength has not changed. Wages will increase, however, and unemployment will rise. According to this reasoning, centralized wage setting reduces wages by inhibiting the fruitless struggle of each group to raise its wage more than the others. 433