Plants and Imported Inputs: New Facts and an Interpretation

Beginning with Wilfred J. Ethier (1979, 1982), an important current of research has emphasized gains to trade from the greater availability of intermediate inputs, as opposed to the greater availability of consumption goods emphasized by Paul R. Krugman (1979) and others. It has been standard in this literature to model input varieties as symmetric, differentiated horizontally but not vertically. In contrast, anecdotal accounts, especially from developing countries, often stress the importance of gaining access to high-quality inputs on the import market. In theoretical discussions, the need to distinguish between the number of inputs and the quality of those inputs can be avoided by treating different qualities of a good as distinct varieties (see, e.g., Paul Romer 1994) or by redefining units of measurement. But in empirical work, one inherits the product categories and units in the data, and typically one must specify whether the availability-of-inputs mechanism is expected to operate through an increase in the number of input categories or through an increase in the quality of inputs within categories. Because of data constraints—in particular because of a lack of information on input and output prices in standard plant-level datasets—it has been difficult to investigate the role of input-quality differences, and recent empirical work, notably by Christian Broda, Joshua Greenfield, and David Weinstein (2006) and Pinelopi K. Goldberg et al. (2008), has tended to focus more on changes in the number of input categories than on quality differences within those categories. In this short paper, we draw on rich product level information from the Colombian manufacturing census to present a new set of facts about importing plants and input prices. The dataset is unique in that it contains detailed, representative, consistently measured information on the unit values of all inputs and outputs of plants. For the 1982–1988 period, the dataset also contains unit values separately for domestic and imported purchases of each input. As we discuss in more detail below, we interpret the new facts as suggesting that Colombian plants purchase higher-quality inputs on the import market than on the domestic market, within narrow product categories. Our empirical work has been guided in part by a theoretical framework from a related paper, Kugler and Verhoogen (2008). In that paper, we hypothesize a complementarity between input quality and plant productivity in generating output quality, and extend the model of Marc J. Melitz (2003) to accommodate it. The model predicts that, in equilibrium, more-productive plants are larger, use higher-quality inputs, produce higher-quality outputs, and are more likely to enter the export market than less-productive plants in the same industry. Using the Colombian plant census, we show that the cross-sectional correlations between a number of observable variables—output prices, input prices, plant size, and export status—as well as differences in those correlations across sectors,—are consistent with our theoretical framework and difficult to reconcile with alternative models that impose symmetry of either inputs or outputs. The distinctive aspect of the current paper is the focus on the distinction between imported and domestic inputs.