Export Dynamics of Multiproduct Firms with (Non-)Differentiated Products
Canonical models with heterogeneous firms and sunk export costs predict more profitable firms export (“export sorting”). This paper shows export sorting pattern varies systematically with product differentiation. Using Slovenian administrative data, I document that the gap in value added per worker between exporters and non-exporters is smaller in more differentiated sectors, particularly among multiproduct firms. I rationalise this pattern with a model in which multiproduct firms choose product scope and export participation based on cost efficiency and market-specific product appeal, featuring a “star-product” mechanism. When export costs are shared across products, high-appeal products pull marginal product lines into export bundles, diluting measured productivity among exporters. This composition effect is especially pronounced in high-differentiation sectors, where high markups allow less appealing products to be exported. The model reproduces observed sorting patterns when estimated separately for high- and low-differentiation sectors. I then evaluate a 50 percentage point tariff increase on a single product and find large contractions with substantial spillovers in the low-differentiation sector but minimal spillovers and little aggregate response in the high-differentiation sector.
Collateral Constraints and Asset Composition
Presented at BOFIT Seminar (2024/11/26), Bank of Finland; European Doctoral Programme Jamboree (2024/06/07), Barcelona; Marco Working Group (2024/04/18), European University Institute
This paper studies how collateral constraints shape firms’ asset composition between real estate and non-real-estate capital. I build a dynamic two-asset model with asset-specific convex and non-convex adjustment costs, as well as asset-specific pledgeability. I estimate the model using data on Chinese listed manufacturing firms. The estimates indicate greater pledgeability of real estate but substantially higher fixed costs of adjusting it. With adjustment and financial frictions, the model matches the observed right-skewed distribution of the real-estate-to-non-real-estate ratio and the declining share of real estate in total capital as capital size increases. Removing financial frictions eliminates precautionary investments and reduces aggregate capital and revenue; removing both financial frictions and fixed costs raises aggregate capital and revenue. In both counterfactuals, aggregate TFPR increases. In a real-estate-crisis scenario in which real estate becomes non-pledgeable, investment tilts towards non-real estate due to its higher collateral value and aggregate capital rises; however, misallocation worsens: aggregate TFPR falls by 3.6% and revenue by 0.3%.
Revisiting the Investment Regressions: State-owned Firms vs. Private Firms in China
Presented at Finance Seminar (2023/05/23), Universitat Pompeu Fabra
The investment regression on a sample of Chinese listed firms shows that there is a significant correlation between investment rate and cash flow ratio after conditioning on average Q. This is true for both state-owned firms and private firms, which are conventionally considered to have different financial conditions. As noted by Cooper and Ejarque (2003), the violation of the constant-returns-to-scale assumption creates a wedge between average Q and marginal Q and can lead to a spurious cash effect in the investment regression even when there are no financial frictions. Following their argument, this paper examines the investment regression results of the Chinese firms by estimating a standard Hayashi’s (1982) model which allows for decreasing returns to scale. The estimation results suggest that despite the decreasing returns to scale of capital, the presence of financial constraints brings the standard capital adjustment model closer to the data moments of both state-owned and private firms in China.
Product Differentiation and Export Dynamics of Multi-destination Multi-product Exporter (with Marius Gruenewald)