In this paper, I provide the first nonparametric decomposition for the variations in the labor share for a production network economy with distortions. This decomposition uncovers the firm-level granular sources of variations for the labor share and segments them in supply and demand shocks. The firms’ share of revenue that reaches directly or indirectly labor compensation, or payment centralities, act as sufficient statistics that measure the strength of the response of the aggregate labor share to specific supply and demand shocks at the firm level. Using the input-output tables for the United States, I estimate sectoral payment centralities, showing that they are positively correlated with sectoral markdowns, negatively correlated with sectoral intermediate input intensity, and have fallen over the last two decades. Variations in sectoral payment centralities explain 99.1% of the labor share volatility, and stronger sectoral distortions in a handful of industries (e.g., credit intermediation and computers and electronics) drove the reductions in the labor share in the United States over the XXIst century.

In this paper, I develop an aggregation theory for distorted production network economies with heterogeneous households. I provide general decompositions for how the aggregate and distributional effects of shocks are sensitive to underlying consumer and firm heterogeneity. The workers’ value-added over labor income ratios (distortion centralities) gauge the importance of workers in the production of heavily distorted firms and are sufficient statistics for the effect of income distribution variations on TFP. The average distortion centrality faced by a household’s expenditure (expenditure centrality) and a firm’s revenue (revenue centrality) are sufficient statistics for the effect of expenditure variations on TFP. Labor misallocation rises and TFP falls as labor income shifts toward high distortion centrality workers, consumption shifts toward high expenditure centrality households, or demand shifts toward high revenue centrality firms. The reason is that when aggregate expenditure on relatively undistorted firms rises, their labor demand increases, reallocating workers from distorted firms with high marginal productivity to relatively undistorted firms with low marginal productivity. These second-best results show how distributional variations affect aggregate output by changing the aggregate allocation efficiency of workers. I estimate the first production network model with household heterogeneity for the United States. I show that variations in the income distribution have been responsible for 20% of the TFP volatility. Additionally, income distribution variations reduced misallocation between 2001 and 2009, and accentuated misallocation after the Great Recession. Heterogeneities in the production network are essential in explaining income and real consumption inequalities.

• 2023
• North American Summer Meeting of the Econometric Society
• Asian Meeting of the Econometric Society
• Australasia Meeting of the Econometric Society
• Canadian Economic Association Conference
• Lacea Lames 2023
• Universidad del Rosario
• Labor, Firms, and Macro Reading Group
• Banco de La Republica
• 2024
• ASSA 2024
• University of Hawai’i at Manoa
• Universidad Diego Portales
• Bank of Canada
• Teconlogico de Monterrey
• Fundacao Getulio Vargas - Sao Paulo
• Tilburg University

In this paper, I develop a general aggregation theory that explains the role of production networks in country-level TFP. This theory applies to a distorted production network open economy with endogenous factor supply. My main contribution is to provide decompositions for the country-level TFP variation that account for the possibility that factors of production and dividends cross national boundaries. The country-level TFP depends on sufficient statistics that characterize the effect on domestic real GDP from (i) firm-level productivity and markdown shocks in domestic and foreign firms and (ii) variations in the global income distribution. These decompositions do not require quantity measures of variations, facilitating their empirical implementation, as price data is no longer necessary. Additionally, for an efficient economy, a Hulten type of decomposition exists for each country, and the global sales distribution is a sufficient statistic to characterize the first-order propagation of global shocks on country-level TFP. These results support a theory of economic spillovers and contagion through industrial networks, corroborating the essential role of global value chains in creating strong complementarities and commonalities in business cycles across countries.

In this paper, we estimate how changes in the structure of global input-output networks have influenced growth. Using an open economy production network model, we identify sufficient statistics that characterize how productivity shocks across domestic and foreign firms influence country-level TFP. We estimate these sufficient statistics using data on input-output networks and sectoral productivity shocks. Structural changes in global input-output networks between 1965 and 2000 were advantageous for developing countries and unfavorable for advanced economies. Holding the global input-output network fixed, TFP growth in China and India would have been 26.6% and 9.7% lower between 1965 and 2000. Whereas for the US and Australia, TFP growth would have been 4.0% and 16.8% higher. Finally, we show that the dynamics of the domestic intermediate input cost share capture the importance that the structure of the global input-output network has on the amplification of shocks on TFP. Our analysis illustrates the importance of industrial linkages and robust domestic intermediate input markets for economic growth.

We offer the first non-linear analysis of how microeconomic disruptions in an inefficient production network economy impact aggregate TFP. Our decompositions, applicable to any general equilibrium economy, provide non-parametric insights. We identify essential general equilibrium metrics for capturing the non-linear repercussions of microeconomic fluctuations through network connections. Our findings encompass firm-level productivity shocks, wasted distortions, and rebated distortions. We reveal that substantial shocks or high production process complementarity/substitution introduce substantial bias in the linear approximation of distortions and productivity shocks found in the literature. Our non-linear second-order effects substantially mitigate this bias.