78 Technological Revolutions By FRANCESCO CASELLI * In skill-biased ( de-skilling ) technological revolutions learning investments re- quired by new machines are greater ( smaller ) than those required by preexisting machines. Skill-biased ( de-skilling ) revolutions trigger reallocations of capital from slow- ( fast- ) to fast- ( slow- ) learning workers, thereby reducing the relative and absolute wages of the former. The model of skill-biased (de-skilling) revo- lutions provides insight into developments since the mid-1970’s ( in the 1910’s ) . The empirical work documents a large increase in the interindustry dispersion of capital-labor ratios since 1975. Changes in industry capital intensity are re- lated to the skill composition of the labor force. ( JEL E23 J31 O33) Income and wage inequality is rising in the United States, as well as in several other coun- tries. Started in the 1970’s, this large increase in earnings dispersion results from both ab- solute gains at the top, and absolute losses at the bottom, of the wage distribution. In fact, even median real earnings have declined. Var- ious empirical explanations have been pro- posed for these trends, and the jury is still out on the quantitative importance of the different hypotheses. However, a growing consensus at- tributes a significant role to skill-biased tech- nical change. The Information-Technology Revolution is the obvious suspect. 1 Despite the intensity of the empirical debate, theoretical work on skill-biased technological change has—with few exceptions—lagged behind. In this paper I present a simple model of tech- nological revolutions. I explore the interaction between skills and technology, and offer an interpretation for the changing nature of this interaction. In doing so, I propose an expla- * Graduate School of Business, University of Chicago, Chicago, IL 60637. I thank Alberto Alesina, Gadi Barlevy, Robert Barro, Eli Berman, Steve Davis, Sven Feldmann, Oded Galor, Zvi Griliches, Larry Katz, Peter Klenow, David Laibson, John Leahy, Greg Mankiw, Jonathan Morduch, Steve Tadelis, Mehmet Yorukoglu, Alwyn Young, Jaume Ventura, two referees, and the many others who have provided comments and/or data. 1 Alternative hypotheses include the globalization of the world economy, changes in labor-market institutions, and the appearance of winner-take-all markets. For recent surveys of this literature see the symposium in the Spring 1997 issue of the Journal of Economic Perspectives. nation for the recent changes in the wage struc- ture, including the absolute decline in the wage of unskilled workers. Further, the model sheds light on the possible future behavior of the wage structure: will the trend towards greater inequality continue or will it reverse itself ? Fi- nally, I generate predictions for other macro- economic developments that can be expected to accompany the current changes in the wage structure. Discussions of skill-biased technological change usually focus on issues of substitut- ability between skilled and unskilled labor. In- stead, I propose to focus on substitutability among technologies. A technology is a com- bination of machines of a certain type and workers who have the skills necessary to use them. A technological revolution is the intro- duction of a new type of machines. Machines of the new type are more productive than ma- chines of preexisting types, but they can only be operated by workers who have developed a set of machine-specific skills. The acquisi- tion of such skills is costly, and the labor force is heterogeneous in the cost of acquisition. The revolution is skill biased if the new skills are more costly to acquire than the skills required by preexisting types of equipment. The revo- lution is de-skilling if the new skills can be acquired at a lower cost than the skills asso- ciated with preexisting technologies. I will ar- gue in this paper that possible examples of skill-biased technological revolutions are the steam engine, the dynamo, and—of course— information technology. One possible exam-