Working Papers

  • How Does Trade Respond to Anticipated Tariff Changes? Evidence from NAFTA (with Armen Khederlarian) (under review)

(Job Market Paper) [Draft] [Slides]

Abstract: We study how anticipation to policy changes affects the estimate of the elasticity of substitution, the most important parameter in international trade. Standard identification of this parameter uses the tariff variation from the Free Trade Agreements (FTA) and assumes that the trade flows equal their consumption. However, FTAs eliminate initial tariffs through announced phaseouts over multiple years. Firms respond to these future policy changes by delaying their purchases until the tariff cut is effective while consuming past purchases held in inventories. These anticipatory dynamics bias the elasticity of substitution as imports diverge from their consumption. We document that around NAFTA's staged tariff reductions, imports experienced sizable anticipatory slumps followed by liberalization bumps. A trade model with inventories replicates these dynamics and illustrates that consumption of imports provides unbiased estimates of the elasticity of substitution. To overcome the lack of consumption data, we propose an empirical measure of consumed imports, which eliminates the bias in the model simulations. Application to the data shows that using import flows instead of consumed imports overestimates the annual elasticity by 68% and the average elasticity of substitution by 16%. The use of consumed imports increases the ratio of long-run to short-run response from 2 to 3.5. This demonstrates a larger role of the gradual response to trade liberalizations.

Presented at; Lisbon Meetings in Game Theory and Applications 2019, Economics Graduate Student Conference 2019, Midwest International Trade Meetings 2019, Midwest Macro Meetings 2019, European Trade Study Group 2018.

(under review)

[Latest Draft] [Slides] [NBER version]

Press: VoxChina

Abstract: We study the effects on trade from the annual tariff uncertainty about China’s MFN status renewal prior to joining the WTO. We have three main findings. First, counter to the evidence elsewhere, trade increases strongly in anticipation of uncertain future increases in tariffs. Second, even though the trade response can be quite large, the probability of a tariff increase was perceived to be relatively small, with an average annual probability of non-renewal of about 6 percent. And third, what matters more is the expected future tariff rather than the uncertainty around it. We identify these effects using within-year variation in the risk of trade policy changes around the renewal vote and trade flows. We show that an (s,S) inventory model generates this behavior and that variation in the strength of the stockpiling in advance of the vote is increasing in the storability of goods. The model is also consistent with a sizeable fraction of the cross-industry variation in annual trade flows documented elsewhere. Our results explain why trade may hold up well in advance of a prospective policy change such as Brexit or the US-China escalating tariff war of 2018-19, but may fall off sharply even if expected tariff increases do not materialize.

Presented at; F.R.E.I.T. EIIT 2019, Bank of Canada 2019, Midwest International Trade Meetings 2019, Midwest Macro Meetings 2019.

  • Mismeasured Firm-level Productivity Gains from Trade Liberalizations: Evidence from India (with Armen Khederlarian)

[Draft] [Slides]

Abstract: Input tariff reductions shift the composition of firms' inputs towards imported goods. Foreign inputs entail higher ordering costs and delivery lags relative to domestic inputs. We show that neglecting increased inventory costs leads to an upward bias in the productivity gains from trade liberalizations. We build a model of different inventory intensity of domestic and foreign inputs and show that under standard accounting practices the effect of input tariffs on productivity is biased. When the mix of inputs changes, the value of previous stock holdings contains lower holding and ordering costs relative to the value with increased foreign inputs. This understates the input usage. The use of estimated inventory deflators based on observables eliminates the measurement bias. We study the relevance of this potential bias during India's trade liberalization in the early 1990s. We find that inventory holdings increased significantly with reduced inputs tariffs and that not accounting for increased inventory costs overestimates firm-level productivity gains by around 35%. Moreover, the bias is larger early in the reform than later.

Work in Progress


Abstract: We augment a standard general equilibrium model of trade with inventory dynamics and evaluate its implications on welfare and aggregate productivity gains. In the quantitative model, importers hold more inventories than the domestic goods seller. As trade barriers fall, revenue expansion justifies more frequent payment of fixed ordering costs. Efficiency emanates from lower inventory holding costs due to a faster turnover. To assess the quantitative relevance of this channel we calibrate the model to match Colombia's producer-level data prior to the trade liberalization of the early 1990s. We compare gains from the observed trade liberalization in a standard trade model with those of the inventory-augmented one. There are two main results. First, the welfare gain from trade is 18% larger than in the standard no-inventory model. Second, the decomposition of the gain shows that more efficient inventory management accounts for 34% of the aggregate productivity gain.