Working Papers

Growth and the Fragmentation of Production

(with Ezra Oberfield)

How much do changes in the fragmentation of production contribute to growth? Using detailed plant-level data on the manufacturing sector in India between 1990 and 2014, we study a version of Smithian growth, the link between greater fragmentation of supply chains and productivity. We propose a measure of a plant's vertical span, which corresponds roughly to the number of stages in a supply chain that the plant performs in-house; when plants have smaller vertical spans, production is more fragmented. We find that fragmentation increases with development in both the cross-section and time series. Further, within locations at a point in time, larger plants tend to have smaller vertical spans, and those that increase sales tend to decrease vertical span. Using changes in demand during the tariff liberalization in the 1990s, we provide evidence that increased demand causes specialization. We find evidence from economies of scale in specialization. We construct a general equilibrium model to rationalize these findings and estimate the sources and magnitude of scale economies. Goods are produced in a succession of steps, each combining labor and a set of intermediate inputs, giving rise to a tree-like structure. Firms exert effort to find suppliers for inputs, and choose the set of production stages (and thereby inputs) to produce the output at lowest cost. The structure implies that the returns to searching are more strongly diminishing for inputs that are further upstream. Firms with high productivity draws are therefore more likely to choose to be more vertically specialized. (Updated 2022-11-03)

Five Facts about MPCs: Evidence from a Randomized Experiment

(with Xavier Jaravel and Etienne Fize)

Forthcoming, American Economic Review

We conduct a randomized controlled trial to study the consumption response of French households to unanticipated one-time money transfers of 300 Euros. Using prepaid debit cards, we consider three implementation designs: (i) a transfer without restrictions; (ii) a transfer where any unspent value expires after three weeks; (iii) a transfer subject to a 10% negative interest rate every week. We observe the participants’ main bank accounts, such that we can compute the impact of the transfer on their overall spending. We establish five facts about MPCs in this setting. First, we find that participants in the baseline treatment group have an average marginal propensity to consume (MPC) of 23% over one month. Second, we find that implementation design matters: the one-month MPC is substantially higher for treatment groups where any remaining balance becomes unusable after three weeks (61%) or where remaining balances are subject to the 10% negative interest rate every week (35%). Third, we document that the cumulative consumption responses are concentrated in the first weeks following the transfer and are flat thereafter. Fourth, we find that there is significant MPC heterogeneity by observed household characteristics, including by liquid wealth, current income, proxies for permanent income, and gender; the MPC remains high even for agents with high liquid wealth. Fifth, we estimate the unconditional distribution of MPCs across households and find that a large fraction of households have high MPCs. These facts are difficult to reconcile with the consumption response in standard Heterogeneous Agent New Keynesian models, which is long-lived and driven by households with little liquid wealth. Furthermore, we observe that households in the treatment groups with a short expiry date or a negative interest rate frequently use other means of payment while still having a sufficient balance on the prepaid card to cover their expenses, indicating that participants see money as non-fungible. Our finding that households consume more when presented with an urgent spending need lends support to theories where the salience of treatments affects economic choices. We conclude that implementation design and the targeting of transfers can greatly alter the effectiveness of stimulus policies. (Updated 2024-06-01)

Trade and the End of Antiquity

(with Thomas Chaney)

What caused the end of antiquity, the shift of economic activity away from the Mediterranean towards northern Europe? We assemble a large database of coin flows between the 4th and 10th century and use it to document the shifting patterns of exchange during this time period. We build a dynamic model of trade and money where coins gradually diffuse along trade routes. We estimate the parameters of this model and recover time-varying bilateral trade flows and real consumption from data on the spatial and temporal distribution of coins. Our estimates suggest that technical progress, increased minting, and to a lesser degree the fall in trade flows over the newly formed border between Islam and Christianity contributed to the relative growth of Muslim Spain and the Frankish lands of northern Europe and the decline of the Roman-Byzantine world. Our estimates are consistent with the increased urbanization of western and northern Europe relative to the eastern Mediterranean from the 8th to the 10th century. (Updated 2024-08-30)

Firm Adaptation in Production Networks: Evidence from Extreme Weather Events in Pakistan

(with Clare Balboni and Mazhar Waseem)

This paper considers how far private adaptation may reduce future vulnerability to climate risks. Using data on monthly firm-to-firm transactions from Pakistan, we find that flood-affected firms are more likely to relocate to safer ground, and shift purchases towards suppliers in less flood-prone regions and reached via less flood-prone roads. The results indicate that firms are imperfectly informed about flood risk, and update their beliefs following floods. We quantify aggregate impacts using a spatial model of endogenous production network formation. The findings suggest that climate chage impacts will be mediated as firms learn from the experience of increasingly frequent disasters. (Updated 2024-07-18)

The Network Origins of Firm Dynamics: Contracting Frictions and Dynamism with Long-Term Relationships

(with Ezra Oberfield, Ruairidh South, and Mazhar Waseem)

Published and Forthcoming

The Comparative Advantage of Firms

(with Swati Dhingra and John Morrow)

Journal of Political Economy 130(12), p. 3025-3100, December 2022

Resource-based theories propose that firms grow by diversifying into products that use common capabilities. We provide evidence for common input capabilities, using a policy that removed entry barriers in input markets to show that the similarity of a firm's and an industry's input mix determines firm production choices. We model industry choice and economies of scope from input capabilities. When the model is estimated for Indian manufacturing, input complementarities make firms 5% more likely to produce in an industry and are quantitatively as important as time-invariant drivers of coproduction rates. Upstream entry barriers were equivalent to a 9.5% tariff on inputs.

[Download paper (.pdf)] [Article page at journal] [Online Appendix]

Vertical Integration and Foreclosure: Evidence from Production Network Data

(with Jan Sonntag)

Management Science 68(1), 141-161, January 2023

This paper studies the prevalence of potential anticompetitive effects of vertical mergers using a novel data set on U.S. and international buyer-seller relationships and across a large range of industries. We find that relationships are more likely to break when suppliers vertically integrate with one of the buyers' competitors than when they vertically integrate with an unrelated firm. This relationship holds for both domestic and cross-border mergers and for domestic and international relationships. It also holds when instrumenting mergers using exogenous downward pressure on the supplier's stock prices, suggesting that reverse causality is unlikely to explain the result. In contrast, the relationship vanishes when using rumored or announced but not completed integration events. Firms experience a substantial drop in sales when one of their suppliers integrates with one of their competitors. This sales drop is mitigated if the firm has alternative suppliers in place. These findings are consistent with anticompetitive effects of vertical mergers, such as vertical foreclosure, rising input costs for rivals, or self-foreclosure.

[Last working paper version (.pdf)] [Article page at journal]

Misallocation in the Market for Inputs: Enforcement and the Organization of Production

(with Ezra Oberfield)

Quarterly Journal of Economics 135(4), p. 2007-2058, November 2020

The strength of contract enforcement determines how firms source inputs and organize production. Using microdata on Indian manufacturing plants, we show that production and sourcing decisions appear systematically distorted in states with weaker enforcement. Specifically, we document that in industries that tend to rely more heavily on relationship-specific intermediate inputs, plants in states with more congested courts shift their expenditures away from intermediate inputs and appear to be more vertically integrated. To quantify the impact of these distortions on aggregate productivity, we construct a model in which plants have several ways of producing, each with different bundles of inputs. Weak enforcement exacerbates a holdup problem that arises when using inputs that require customization, distorting both the intensive and extensive margins of input use. The equilibrium organization of production and the network structure of input-output linkages arise endogenously from the producers' simultaneous cost minimization decisions. We identify the structural parameters that govern enforcement frictions from cross-state variation in the first moments of producers' cost shares. A set of counterfactuals show that enforcement frictions lower aggregate productivity to an extent that is relevant on the macro scale.

[Download paper (.pdf)] [Article page at journal] [Online Appendix]

The Impact of Contract Enforcement Costs on Value Chains and Aggregate Productivity

Review of Economics and Statistics 104(1), p. 34-50, January 2022

I study how supplier contracting frictions shape the patterns of intermediate input use and quantify the impact of these distortions on aggregate productivity. Using the frequency of litigation between US firms as a novel measure to capture the need for formal enforcement, I find a robust relationship between countries' input-output structure and their quality of legal institutions. In countries with high enforcement costs, firms have lower expenditure shares on intermediate inputs in sector pairs where US firms litigate frequently for breach of contract. A quantitative model shows that improvement of contract enforcement institutions would lead to sizable welfare gains.

[Download paper (.pdf)] [Article page at journal] [Download enforcement-intensity measures (.zip)]

Teaching

PhD International Trade (IHEID, Fall 2024): Syllabus

Macro A (IHEID, Fall 2024): see Moodle

Microeconomics (Sciences Po, Fall 2018-2023): see Moodle

International Trade & Finance (Sciences Po, Fall 2018-2023): see Moodle

Macroeconomics and Finance (Insead, Spring 2017)

KEPP 2385: Macroeconomics 2 (for Master Economics, Sciences Po, Spring 2015, 2016, 2017)

BECO 1750: Money and Banking (Sciences Po, Fall 2014, 2015, 2016)

EC220: Introduction to Econometrics (LSE, 2009/10)

EC400: Introductory Course in Maths and Statistics (LSE, 2010/2011)

EC321: Monetary Economics (LSE, 2010/2011)

EC321: Money and Banking (LSE, 2010/2011)

EC210: Macroeconomic Principles (LSE, 2011/2012)

EC442: Macroeconomics for PhD (LSE, 2013/2014)

Software

I wrote some packages to help with doing econometrics in Julia. In particular:

• GLFixedEffectModels is a package for estimating Generalized Linerar Models with high-dimensional fixed effects
• Douglass is a package that allows you to do data manipulation tasks using a Stata-like syntax
• RegressionTables creates journal-style regression tables similar to 'estout' in Stata or 'stargazer' in R
• KeplerGL to make beautiful maps

All these packages, and some others, are on Github.