Globalization and Income Distribution: A Specific Factors Continuum Approach∗

Does globalization widen inequality or increase income risk? Globalization amplifies the effect of idiosyncratic relative productivity shocks. But wider markets reduce the effect of economy-wide supply shocks on world prices. Both forces are at work in the specific factors continuum model of this paper. Ex post equilibrium exhibits positive (negative) premia for export (import-competing) sector specific factors. Globalization widens inequality in North and South. Globalization increases personal income risk from idiosyncratic productivity shocks, but reduces aggregate shock risk acting on the factoral terms of trade. Both forces have their greatest impact on the poorest and least impact for the richest trading sectors, while the distribution in nontraded sectors is unaffected. JEL Classification: F10. Contact information: James E. Anderson, Department of Economics, Boston College, Chestnut Hill, MA 02467, USA. ∗I would like to thank, without implicating, Karim Chalak, Arnaud Costinot, Fabio Ghironi, Philippe Martin and participants in seminars at Boston College and Paris School of Economics. Globalization is often thought to have increased inequality. The model of this paper delivers the stark result that globalization raises inequality in both North and South. Globalization is also commonly thought to have increased personal income risk. New markets open to trade while in old markets like wine, French vintners are more exposed to supply innovations from Argentine and Australian vintners and vice versa. A contrasting economic intuition suggests that wider markets reduce the effect of national supply shocks on the variance of world prices and thus factor incomes. The two opposing forces are sorted out in this paper using a formal model that isolates the key elements while abstracting from inessential details. The model embeds the specific factors model in the infinitely many goods continuum setup pioneered by Dornbusch, Fischer and Samuelson, complementing their analyses (1977, 1980) of the Ricardian and Heckscher-Ohlin models. For thinking about income inequality and insecurity, the specific factors model has several advantages over the Ricardian and Heckscher-Ohlin models. First, specific factors rationalize the tremendous heterogeneity of rents to otherwise observationally identical factors, especially the premium for export sector employment. Second, specific factors are essential to the theoretical and empirical success of the now-standard political economy of trade model (Grossman and Helpman, 1994; Goldberg and Maggi, 1999). Third, resource reallocation between firms within sectors in this paper is driven by a Darwinian gale blowing through the sectoral factor market. When paired with wage bargaining, the model can explain why larger and more productive firms pay higher wages for skilled labor. Finally, the specific factors model reverses the excessive specialization imposed by trade equilibrium in the Ricardian and Heckscher-Ohlin continuum models. Diversified production of traded goods occurs with measure zero in those models, while here diversification occurs with measure one. The latter is arguably a more useful metaphor. The model is stripped down to focus on the distributional consequences of combining specificity, random productivity and globalization. After the endowment of potentially skilled labor is allocated across sectors, specific skills are acquired and the skilled labor combines with intersectorally mobile unskilled labor to produce output as efficiently as possible given the This regularity was given prominence by Katz and Summers (1989). The phenomenon is well documented in the US and other developed countries. Sparser available evidence finds the same pattern in poorer countries as well — see Milner and Tandrayen (2006) on sub-Saharan Africa and Tsou, Liu and Huang (2006) on Taiwan. realizations of productivity shocks. All sectors have identical ex ante potential production functions, an assumption that shuts off Heckscher-Ohlin and more importantly Stolper-Samuelson distributional properties for simplicity. The pure Ricardian continuum model is contained within the model as a special case, when the production function assigns zero marginal product to the specific factor or when the specific factor allocation is perfectly efficient. As with the original Dornbusch-Fischer-Samuelson Ricardian model, sharp implications are obtained that appear likely to obtain in more general cases. The tractability of the model suggests that it it is a good platform on which to build extensions. The paper first characterizes the ex post equilibrium production and trade patterns. Familiar comparative static results with respect to growth, transfers and trade costs are reviewed, echoing Dornbusch, Fischer and Samuelson. A more novel result shows that in the presence of aggregate productivity shocks, globalization reduces the variance of the factoral terms of trade and hence the dispersion of personal incomes. The allocation of the specific factors is given ex post, but a choice variable ex ante. In the face of productivity shocks, the best that can be done through efficient capital markets is to equalize ex ante expected returns. The ex post returns differ from their expectation due to realized productivity shocks and the efficient reallocation of the mobile factor to accommodate them. The equilibrium internal income distribution given the efficient allocation of skills exhibits higher earnings for export sectors (those receiving high productivity realizations) than for import competing sectors (those receiving low productivity realizations). Globalization widens income inequality in each country, raising the top, lowering the bottom and narrowing the middle. Viewed ex ante, the model formalizes the popular sense that personal incomes are more risky in globalizing world with purely idiosyncratic productivity shocks. When productivity shocks include an economy-wide component that shifts relative productivity between countries, there is a countervailing effect on income risk. Globalization reduces the factoral terms of trade variance, reducing the ex ante expected dispersion of personal incomes. The model implies that both effects of globalization are biggest for the poorest specific factors. The effect of globalization on income distribution has previously been studied, but in models for which theory does not fit well with empirics. For example, the factor proportions model applications surveyed in Feenstra (2004) have income distributions of low dimension, in contrast to empirical