Chong Xiang

The Home Market Effect and Bilateral Trade Patterns

We test for home-market effects using a difference-in-difference gravity specification. The home-market effect is the tendency for large countries to be net exporters of goods with high transport costs and strong scale economies. It is predicted by models of trade based on increasing returns to scale but not by models of trade based on comparative advantage. In our estimation approach, we select pairs of exporting countries that belong to a common preferential trade area and examine their exports of goods with high transport costs and strong scale economies relative to their exports of goods with low transport costs and weak scale economies. We find that home-market effects exist and that the nature of these effects depends on industry transport costs. For industries with very high transport costs, it is national market size that determines national exports. For industries with moderately high transport costs, it is neighborhood market size that matters. In this case, national market size plus market size in nearby countries determine national exports.

Diversification cones, trade costs and factor market linkages

This paper finds non-uniform differences in the distribution functions of factor usage intensities among 10 rich OECD countries: countries form 3 distinct groups (e.g. UK, France and US in different groups) such that the differences are more pronounced across groups than within the same group, and capital abundant countries are in capital abundant groups. The estimation works even if the same industry codes represent different goods across countries in the data. This pattern is consistent with the multiple-cone factor proportions theory with zero trade costs. An alternative interpretation involves high trade costs across the 3 groups of countries: 40% ~ 70% on ad valorem basis to invalidate the multiple-cone interpretation and 60% ~ 100% to account for all the observed differences in factor usage intensities. These high trade costs illustrate how badly factor price equalization is violated for the 10 OECD countries. Both interpretations share one implication: factor market linkages are weak between the countries in different groups.

The sufficiency of the 'lens condition' for factor price equalization in the case of two factors

Factor price equalization (FPE) is a central theme in trade theory, for which Dixit and Norman (1980) establish the necessary and sufficient condition (the FPE condition). Deardorff (1994) provides a more intuitive condition (the lens condition) and establishes its necessity in general, as well as its sufficiency for the case of 2 countries. In this paper, I prove that the lens condition is sufficient for FPE in the case of 2 factors. This theorem has implications for empirical work.

New goods and the skill premium

In a two-cone Heckscher-Ohlin model with CES preferences and a continuum of goods, adding new goods to the Norths technology necessarily increases the Northern skill premium if the new goods are skilled-labor intensive, but may even increase the premium if they are unskilled-labor intensive. Thus, the introduction of new goods into US technology could have done more to increase the US skill premium than a closed economy model would predict. I also explore how new Northern goods affect the Southern skill premium and what happens if new goods generate preference reversals across existing goods. I develop a two-step solution method that simplifies the analysis of many other comparative static exercises in a two-cone Heckscher-Ohlin model with a continuum of goods.

Factor Uses and the Pattern of Specialization

We show that in the setting of multiple goods and factors, the factor proportions theory has the following prediction: across industries, the impacts of the endowment of a given factor on industry outputs have positive co-variance with the relative uses of this factor. The intuition is that, on average, the industries that use a given factor heavily have positive output responses, following an increase in the endowment of this factor. This co-variation condition is robust to Hicks neutral- and factor-augmenting productivity differences, and constitutes a direct test of the production side of the factor proportions theory. We also show that the co-variation condition finds empirical support. This is evidence that is consistent with the factor proportions theory.

Product Cycles in U.S. Imports Data

In this paper, I construct product-level U.S.-manufacturing-imports data for new products. I show that consistent with product cycles, the Norths new-products exports to the United States, relative to its old-products exports, grow faster than the Souths for over a decade; then the South catches up with the North, and this pattern is reversed. This finding holds up in parametric, nonparametric, and semiparametric estimations, and only when new products are properly identified and old products within the same industries are used as controls. There is also evidence that product cycles become shorter over time and they are technology related.

The Production of Cognitive and Non-cognitive Human Capital in the Global Economy

A country’s welfare depends on its ability to accumulate cognitive and non-cognitive human capital. In this paper, we model the productions of cognitive and non-cognitive human capital in general equilibrium. We use revealed comparative advantage to infer countries’ non-cognitive and cognitive productivities without a direct measure for the non-cognitive dimension. Our model also delivers analytical expressions for how non- cognitive and cognitive productivities can be aggregated into a single human-capital quality index, or HCQI, and how HCQI relates to output per worker. Our model allows us to obtain the values of non-cognitive and cognitive productivities and HCQI, using publically available data for a sample of mostly high-income countries. We find that: 1. many countries with low test scores have high non-cognitive productivities; 2. the hard-to-measure non-cognitive human capital is important for HCQI, and HCQI is important for output per worker; 3. the trade-o¤ between cognitive- and non-cognitive productivities can be visualized and analyzed using an iso-HCQI curve: e.g. uneven cognitive and non-cognitive productivities tend to lower HCQI; 4. this trade-o¤ can be quantified, and has policy implications: e.g. excessive attention to test scores may decrease aggregate output; 5. international trade matters, theoretically, for HCQI, because the gains from trade help to compensate a country for uneven productivity across human capital types: e.g. the iso-HCQI curve would have a very different shape under free trade.

Specialization, Factor Uses and the Rybczynski Theorem

We show that in the case of multiple goods and factors, the factor proportions theory has the following prediction: across industries, the impacts of the endowment of a given factor on industry outputs have positive co-variance with the relative uses of this factor. The intuition is that on average, the industries that use a given factor heavily have positive output responses following an increase in the endowment of this factor. The co-variation condition accommodates Hicks-neutral factor-augmenting productivity differences across countries. We also show that the predicted positive co-variance finds empirical support. This is evidence that is consistent with the Rybczynski mechanism of the factor proportions theory.