Production Networks and Aggregate Fluctuations: Evidence, Theory, and Policy

In this dissertation we study the sources of aggregate fluctuations with emphasis on the role of firms’ linkages and non-linearities in production. In Chapters 1 and 2, we present a new set of empirical facts on i) the relationship between the structure of inter-sectoral linkages and aggregate fluctuations, and ii) the relationship between firms’ elasticity of substitution in production and the spreads on corporate bonds during periods of high corporate debt and in recessions. We then construct a multisector model that is able to deliver the observed empirical patterns. In Chapter 3 we test the predictions of different dynamic multisector models regarding how well they predict the observed co-movement of sectoral output growth in the U.S. industrial sectors. Finally, in Chapter 4 we study the normative implications of combining the models in Chapters 1-3.
In Chapter 1, we study what features of the countries’ domestic structure of sectoral linkages can help account for the observed differences in cross-country volatility of growth and the size of macroeconomic downturns. The empirical cross-country evidence, cross- sectional and panel data, suggests that unlike previous multisector models predict, the density of sectoral connections – number of sectoral connections in countries’ input-output tables – is strongly correlated with aggregate volatility and the size of downturns. Therefore, we develop a multisector model that is able to deliver a role for the density of connections. The key implications of the model are that: i) the density of sectoral connections amplifies shocks – as observed for manufacturing oriented countries – if firms have low elasticity of substitution between labor and intermediates, while ii) density mitigates shocks – as ob- served for service oriented countries – if firms have high elasticity. In addition, iii) the density of connections amplifies the size of downturns – as observed for manufacturing oriented countries – if low elasticity firms face frictions in the use of intermediates, while iv) density mitigates downturns – as observed for service oriented countries – if high elasticity firms face frictions in the use of labor.
In Chapter 2, we study what are the sectors of the economy that have more troubles in financing inputs during recessions in the U.S. Using firm level and sector level data, we find that during times of high corporate debt i) manufacturing sectors with lower estimated elasticities of substitution between labor and intermediates pay higher spread on corporate bonds, while service sectors with larger elasticities pay higher spreads on corporate bonds. We also find that during recessions firms with low elasticity between labor and intermediates pay higher spreads. The model economy in Chapter 1, with occasionally binding working capital constraints on labor or intermediate inputs, can deliver these observed empirical regularities. Intermediates become more expensive in recessions, therefore inflexible and intermediate-intensive manufacturing firms need to increase borrowing and eventually become (more) financially constrained. On the other hand, flexible and labor-intensive ser- vice firms reduce their relative demand for (more expensive) intermediates and eventually become constrained in financing labor input. The model also implies that during recessions service sector firms face much tighter credit conditions.
In Chapter 3, we contrast the predictions of different multisector DSGE models with the observed correlations of sectoral output growth in the U.S. industrial sectors. We study models that differ in the timing of the investment decision and in the technology used to build physical capital. To focus on the role of capital accumulation, we assume unitary elasticity of substitution in production and never-binding working capital constraints. We find that our model with a single type of physical capital that firms rent from households outperforms alternative models with sector specific capital and investment goods’ linkages. Unlike models with sectoral specific capital accumulation, our model has a reduced state space, therefore it can solved using global methods and occasionally-binding working capital sectoral constraints.
Finally, in Chapter 4 we study the policy implications of the models in Chapter 1 and 2, extended to have physical capital as the model in Chapter 3. We characterize a network pecuniary externality that arises when working capital constraints bind or are likely to bind. There is an intra-temporal externality when firms are constrained in the use of inputs. Firms in one sector do not internalize how their production decisions affect equilibrium prices and then affect the constraints of firms’ in other sectors. There are inter-temporal externalities in which households do not internalize that i) by increasing investment in a sector today, can relax that sector constraint tomorrow, and ii) that by increasing aggregate investment today can reduce tomorrow’s cost of capital and relax firms constraints.