Agglomeration and Trade: State-Level Evidence from U.S. Industries
One sentence summary: Industry- and state-level international exports and intermediate input trade within the U.S. are systematically connected to production agglomeration and specialization effects.
The corresponding paper by Hakan Yilmazkuday has been published at Journal of Regional Science.
Abstract
This paper investigates the connection between economic agglomeration and trade patterns within the U.S. at the industry level. On the consumption side, industry- and state-specific international imports and elasticities of substitution are shown to be systematically connected to consumption agglomeration effects, while on the production side, industry- and state-specific international exports and intermediate input trade are shown to be systematically connected to production agglomeration and specialization effects. Industry structures play an important role in the determination and magnitude of these effects.
Non-technical Summary
According to the United States (U.S.) trade data, intranational trade volume is more than 6 times international trade volume, on average, between 1993 and 2007. This paper sheds light on this difference by introducing a regional trade model that connects international and intranational trade patterns to the distributions of production and consumption within the U.S.. The investigation can be categorized under two topics:
- On the production side, what portion of the state- and industry-level production is consumed as a final good within the U.S., and what portion of it is used as an intermediate input within the U.S. or exported to other countries? How are these portions connected to economic agglomeration and specialization effects?
- On the consumption side, what are the elasticities of substitution across products of different states at the state and industry levels? Are these elasticities systematically connected to economic agglomeration and specialization effects? Do these elasticities have further implications for the shares of consumption (at the state and industry levels) imported from other countries; are these shares systematically connected to economic agglomeration and specialization effects?
The model consists of individuals and firms in a discrete framework where there are finite number of goods and regions. There are two types of goods, namely traded and non-traded. Each region produces non-traded goods together with a variety of each traded good. Traded goods can be traded up to a transportation cost, and each region may consume varieties of each traded good besides non-traded goods. Production of traded goods is achieved by only labor, while the production of non-traded goods requires traded goods. Thus, traded goods can be used either as a final good or an intermediate input in the model. Individuals in each region have different elasticities of substitution across varieties of each traded good. This, in turn, leads optimization of each monopolistically competitive firm resulting in prices equal to marginal costs with region/good specific mark-ups. According to the model, the main motivation behind trade is found to be the heterogeneity across regions/goods in terms of factor costs, production technologies, transportation technologies, locations, and taste parameters.
Non-traded goods are consumed only locally by definition. So, only the traded goods are modelled in this paper although the existence of non-traded goods, through their intermediate input usage, is considered explicitly. After carefully controlling for intermediate input trade and international trade, the remaining part, the final good trade, is analyzed extensively. In particular, the model is numerically solved using the available data to figure out the region/good specific elasticities of substitution and portions of production that are used as final goods within the country. After that, possible economic connections between international imports, elasticities of substitution, and consumption patterns, as well as connections between international exports, intermediate input trade, and production patterns, are investigated through agglomeration and specialization of the industries at the state level.
Instead of using trade flow data, which do not have sufficient information about the exact distribution (e.g., agglomeration, specialization) and structure (e.g., technology, marginal costs) of production and consumption across regions, the consumption, production, and trade implications of a partial equilibrium model are tested using industry-specific production and consumption data at the state level. In particular, four state-level industry data are considered within the U.S.: 1) Food and beverage and tobacco products, 2) Apparel and leather and allied products, 3) Computer and electronic products, and 4) Furniture and related products.
On the consumption side, it is shown that the industry- and state-level elasticities of substitution can be significantly explained by consumption agglomerations; the elasticities are positively (respectively, negatively) affected by agglomeration of consumption for food and furniture (respectively, for apparel and electronics). The differences across these industries are connected to the homogeneity of the products, where homogeneity is further supported by numerically calculated median elasticities of substitution across states/industries. Consumption agglomerations are also connected to international imports at the industry and state levels.
On the production side, it is shown that the industry- and state-level portion of production that is used as a final good within the country can be significantly explained by both agglomeration and specialization of the industries; these portions are negatively related to both effects. In other words, the industry- and state-level portion of production that is used as an intermediate input or exported abroad is significantly and positively related to agglomeration and specialization of the industries across states. Thus, agglomeration and specialization of industries play an important role in determining the patterns of trade, both intranationally and internationally. Finally, comparisons across industries suggest that the spillover effects are much higher for electronics compared to food, apparel, or furniture, in terms of both consumption and production. High explanatory powers in the regression analyses further support the model.