how do exporters respond to antidumping investigations? · a⁄ected foreign exporters.3...
Post on 13-Mar-2020
6 Views
Preview:
TRANSCRIPT
How Do Exporters Respond to AntidumpingInvestigations?
Yi Lu,a Zhigang Tao,b and Yan Zhangba National University of Singapore
b University of Hong Kong
Revised: March 2013
Abstract
Using monthly transaction data covering all Chinese exporters,we investigate how Chinese exporters respond to U.S. antidumpinginvestigations during the 2000-2006 period. Speci�cally, we uncoverdi¤erent responses across di¤erent products (i.e., extensive margine¤ect, intensive margin e¤ect, price e¤ect and trade de�ection e¤ect)and di¤erent responses across �rms within the same products (i.e., �rmheterogeneity, trade intermediaries versus direct exporters, and single-product versus multi-product direct exporters). Our study helps pieceup a complete picture of the e¤ectiveness of antidumping measures,and points to the importance of incorporating market structures andpolicy uncertainty into the �rm heterogeneity literature.
Keywords: Antidumping investigations; Di¤erence-in-di¤erences es-timation; Extensive and intensive margins; Trade intermediaries; Single-versus multi-product exportersJEL Codes: F13; D22; F14; L25
1
1 Introduction
Despite the increasing trend for international trade due to rounds of tari¤reductions and advancements in telecommunications and logistics, we havewitnessed persistent and even increasing use of contingent trade protectionpolicies (e.g., Prusa, 2001; Zanardi, 2006; Bown, 2011). In particular, gov-ernments around the world have resorted to antidumping measures, whichare permissible under the World Trade Organization (WTO) rules and regu-lations, to protect their �rms and industries, especially in times of economicdi¢ culties. The widespread use of antidumping measures has spurred econo-mists to study its impacts on �rm behavior, which has signi�cant implicationson long-run economic growth and national competitiveness.1
While signi�cant insights have been gained from the existing literatureregarding the impacts of antidumping measures on protected domestic �rmsand industries,2 much less is known about the corresponding impacts ona¤ected foreign exporters.3 Understanding how a¤ected foreign exportersrespond to antidumping measures is, however, an essential component inpiecing up a picture of market competition between domestic �rms and for-eign exporters in both the short-run (i.e., right after antidumping measures)and the long-run (i.e., after the expiration of antidumping measures) and itsimplications for industry dynamics and national economy. Moreover, under-standing whether foreign a¤ected exporters should continue their exportingbehavior in response to negative shocks brought by antidumping investi-gations compliments the existing �rm heterogeneity literature that focusesprimarily on the entry decision into export market.This paper provides the �rst empirical analysis using a monthly �rm-
product level data to examine responses of a¤ected foreign exporters to an-tidumping investigations across di¤erent narrowly-de�ned product categories(i.e., HS-6 digit) and across �rms within the same product categories.Speci�cally, our study utilizes antidumping cases �led by the United
States (the U.S.) against Chinese exporters over the 2000-2006 period. Thechoice of this research setting is motivated by the fact that China has becomethe world�s largest recipient of antidumping measures along with its rise froman insigni�cant player in the world trade system in 1978 to the world�s largest
1For surveys of existing studies on antidumping, see Blonigen and Prusa (2003) andFalvey and Nelson (2006).
2For recent studies, see, for example, Gallaway, Blonigen and Flynn (1999), Koningsand Vandenbussche (2008), and Pierce (2011).
3A few papers look at how antidumping duties a¤ect foreign exporters�pricing behavior(Blonigen and Park, 2004), export-destination diversi�cation (Bown and Crowley, 2006,2007) and FDI strategies for serving foreign markets (Blonigen, 2002).
2
exporter. Meanwhile, the U.S. is the world�s second largest initiator of an-tidumping cases against China because of its rising trade de�cit with Chinaand the apparently related loss of manufacturing jobs in the U.S. (see, forexample, Autor, Dorn, and Hanson, forthcoming; Pierce and Schott, 2012).To carry out the empirical investigation, we draw on data from two sources:China Customs data (2000-2006) and the World Bank global antidumpingdatabase. From the �rst data set, we obtain information on monthly ex-port transactions at the Chinese HS-8 digit product category by all Chineseexporters to the U.S., including export volume, export value and exporteridentity. From the second data set, we compile all the antidumping investi-gations carried out by the U.S. against Chinese exporters at the U.S. HS-10digit product category over the 2000-2006 period, including information suchas initiation date, preliminary determination dates, and �nal determinationdates. The two data sets are then combined at the HS-6 digit product cate-gory level, which is common to China and the U.S.Our identi�cation strategy relies on the comparison of outcome variables
(such as export volume, number of exporters, export price, and trade de�ec-tion) for exporters in the a¤ected product category (the treatment group)with the same variables for those in the una¤ected product category (thecontrol group) before and after the various important stages of the antidump-ing investigation process, i.e., the di¤erence-in-di¤erences (or DID) method.Speci�cally, we use two alternative control groups: �rst, for an HS-6 digitproduct subject to antidumping investigations, we use as the control groupall other una¤ected HS-6 digit products within the same HS-4 digit category.Second, we follow Blonigen and Park (2004) in constructing a matched con-trol group based on the likelihood of products being subject to antidumpinginvestigations.Our main �ndings are summarized as follows:
� We �nd a substantial trade-dampening e¤ect of antidumping investiga-tions at the HS-6 digit product level upon both a¢ rmative preliminaryand �nal International Trade Commission (ITC) determinations.
� Using a binary treatment status variable (which implicitly assumes thesame antidumping duties across products), we �nd signi�cant extensivemargin e¤ects (i.e., a decrease in the number of exporters), no intensivemargin e¤ects (i.e., a decrease in the export volume per exporter),little freight on board (F.O.B.) export price adjustments, and no tradede�ection e¤ect (i.e., exports of the concerned products to marketsother than the U.S.). However, modest intensive margin e¤ects andexport price adjustment are detected when antidumping duties are usedinstead of the binary treatment status variable.
3
� When we divide products into three quantiles corresponding to low (i.e.,< 50%), medium (i.e., 50~100%) and high (i.e., > 100%) antidumpingduties, we �nd di¤erent extensive margin e¤ects, intensive margin ef-fects and price e¤ects across these three quantiles, but consistently notrade de�ection e¤ects. Speci�cally, for products in the low quantile,there is no extensive margin e¤ect, signi�cant intensive margin e¤ect,and signi�cant price increase. For products in the high quantile, thereis a substantial extensive margin e¤ect, no intensive margin e¤ect andmodest price increase. For products in the medium quantile, there is amarginally signi�cant extensive margin e¤ect, no signi�cant intensivemargin e¤ect and no signi�cant price increase.
� Within the same product categories, we �nd that less productive ex-porters are more likely to exit the U.S. market upon both a¢ rmativepreliminary and �nal ITC determinations. Meanwhile, direct exportersare found to be more likely than trade intermediaries to exit the U.S.market upon both a¢ rmative preliminary and �nal ITC determina-tions. Moreover, we �nd that multi-product direct exporters are morelikely than are single-product direct exporters to exit the U.S. marketupon the a¢ rmative preliminary ITC determination, but the oppositeholds upon the a¢ rmative �nal ITC determination.
These results are found to be robust in a series of checks on variouspotential data and estimation issues, such as validity checks on the DIDestimation, quarterly data (instead of monthly data), exclusion of outly-ing observations, inclusion of unsuccessful and withdrawn cases, exclusionof antidumping cases under investigations by other countries, exclusion ofprocessing traders and foreign �rms, check on the aggregation bias, controlfor other trade shocks like U.S. safeguard investigations and China�s WTOaccession, di¤erential impacts across products with di¤erent import demandelasticities, and alternative de�nition of single-product direct exporters (seeSection 5.6 for details).Our results suggest that U.S. antidumping investigations wipe out weaker
Chinese exporters and leave behind more productive ones often with multi-market and multi-product coverage. And in many product categories (espe-cially those facing higher margins of antidumping duties), the wipeout resultsin a substantial consolidation of Chinese exporters, under which the surviv-ing, stronger Chinese exporters can even raise their F.O.B. export pricesand at the same time grab the market share left by the weaker ones. Mean-while, existing studies (e.g., Pierce, 2011) on the impacts of U.S. antidumpingmeasures on its domestic, protected �rms have shown that while protected
4
�rms are able to increase their prices, their physical productivity actuallyfalls. And the protection through temporary imposition of antidumping du-ties is more tilted toward the weaker domestic producers, thereby slowingdown the resource reallocation towards more productive ones. Taken to-gether, U.S. antidumping investigations de�nitely bring temporary bene�tsto domestic producers, who expand their market share, as Chinese importssubstantially fall and numerous Chinese exporters exit the market. In thelong-run (especially when the antidumping duties are lifted), however, an-tidumping investigations may spell more troubles for U.S. domestic producersin their competition with the Chinese exporters, as the former becomes lessproductive on average whereas the latter experiences just the opposite.Our study also uncovers several patterns of exiting behavior by incumbent
exporters across di¤erent products with di¤erent market structures (i.e., de-gree of �rm heterogeneity, import demand elasticity, etc). Meanwhile, withinproducts, we also detect di¤erent exiting behavior by exporters at di¤erentstages of antidumping investigations, with di¤erent productivity levels, andwith varying degrees of diversi�cation across products and markets. Whilethe current �rm heterogeneity literature focuses primarily on the decisioninto exporting market under substantial �xed costs of exporting, our studycalls for a more general theory that accounts for both the entry and exit de-cisions, and addresses interactions between �rm heterogeneity and complexmarket structures and policy uncertainty.The rest of the paper is organized as follows. Section 2 describes the
institutional background of antidumping investigations in the U.S. The esti-mation strategy is discussed in Section 3 and data are reported in Section 4.Section 5 presents empirical �ndings, while Section 6 devotes to the discus-sions of these results. The paper concludes with Section 7.
2 Institutional Background of AntidumpingInvestigations in the U.S.
In this section, we brie�y describe the institutional context of antidumpinginvestigations in the U.S. and its relevance to our identi�cation strategy(Staiger and Wolak, 1994).In the U.S., there are two government bodies involved in antidumping
investigations: the Department of Commerce (DoC) and the InternationalTrade Commission (ITC). The DoC determines whether an imported productunder investigation is sold in the U.S. at less than its �fair value�, whilethe ITC determines whether the imported product has materially injured
5
the relevant U.S. domestic industries. Each of these two bodies makes twodeterminations, i.e., the preliminary and �nal determinations.Once an antidumping petition against an imported product is �led and
then considered, the ITC �rst makes a preliminary determination within45 days. If the determination is negative, the investigation is terminated.Otherwise (i.e., where the preliminary ITC determination is a¢ rmative),the DoC conducts its investigation and makes a preliminary determinationin the next 115 days. Regardless of the DoC�s preliminary determination(a¢ rmative or negative), the investigation process continues. However, ifthe DoC�s preliminary determination is a¢ rmative, importers of the a¤ectedimported product have to post a cash deposit or bond to cover dumpingduties the DoC estimates to be payable.After the DoC�s preliminary determination but before the ITC�s �nal de-
termination, the antidumping investigation can be terminated due to with-drawal by the petitioner(s), or can be suspended due to agreements reachedbetween a¤ected foreign exporters and the DoC. If an antidumping investiga-tion is neither terminated nor suspended, the investigation moves on to thenext stage, in which the DoC makes a �nal determination within 75 days ofits preliminary decision. If the DoC�s �nal determination is negative, the in-vestigation is terminated. Otherwise, the ITC has 45 (or 75) days to conducta second round of investigation and make a �nal determination, depending onwhether the DoC�s preliminary determination was a¢ rmative (or negative).Once both the DoC and the ITC reach a¢ rmative �nal determinations, theDoC must issue an antidumping order to levy antidumping duties within 7days.In summary, there are �ve important points in time during an antidump-
ing investigation: initiation, the preliminary ITC determination, the prelim-inary DoC determination, the �nal DoC determination, and the �nal ITCdetermination.
3 Estimation Strategy
3.1 Estimation Speci�cation
In contrast to the yearly data used in most of the literature, our monthlyexport transaction data allow us to investigate whether exporters responddi¤erently to di¤erent stages of the antidumping investigation process. Asnoted in the Section 2, there are �ve stages in an antidumping investigation:initiation of the case, the preliminary ITC determination, the preliminaryDoC determination, the �nal DoC determination, and the �nal ITC deter-
6
mination. Given that the DoC makes a¢ rmative determinations in mostantidumping cases, we focus on the three remaining dates in the antidump-ing investigation, i.e., the initiation date, the date of the preliminary ITCdetermination, and the date of the �nal ITC determination. The a¢ rmative�nal ITC determination leads to the imposition of dumping duties, whichconsequently increase the costs of the U.S. importers for the export productsconcerned. The a¢ rmative preliminary ITC determination, in combinationwith (almost certainly an) a¢ rmative preliminary DoC determination, re-quires U.S. importers to pay a deposit as a bond for the expected dumpingduties. Even the initiation of an antidumping investigation might have ane¤ect on U.S. importers, as it brings uncertainty to their businesses. Wetherefore expect exporters to have progressively negative responses to thethree stages of an antidumping investigation (initiation, preliminary ITC de-termination, and �nal ITC determination). Moreover, di¤erent exporters(i.e., with di¤erent productivity levels; trade intermediaries versus directexporters; single-product direct exporters versus multi-product direct ex-porters) may respond di¤erently during di¤erent stages of the antidumpinginvestigation process.To identify the possible e¤ects of antidumping investigations, we employ
the di¤erence-in-di¤erences estimation strategy at both the product (de�nedat the HS-6 digit level) and �rm-product levels. Speci�cally, we exploit twosources of variations: time variation (before and after a critical date in the an-tidumping investigation process) and cross-sectional variation (a¤ected prod-ucts/�rms or the treatment group, and una¤ected products/�rms or the con-trol group). The identi�cation relies on comparison of outcome variables forthe treatment group with those for the control group both before and afterthe relevant stages of the antidumping investigation process.The estimation speci�cation at the product level takes the following form
ypt = �1Treatmentp � Post1pt + �2Treatmentp � Post2pt+�3Treatmentp � Post3pt + �p + �t + "pt; (1)
where ypt is the outcome variable (i.e., the logarithm of export volume, thelogarithm of the number of exporters, the logarithm of export price, andthe logarithm of total export volume to countries other than the U.S.) forproduct p in month t; Treatmentp is a dummy variable taking the valueof 1 if product p belongs to the treatment group (i.e., is being investigatedfor dumping) and 0 otherwise; �p is the product dummy capturing all time-invariant product characteristics; �t is the month dummy capturing e¤ectscommon to all products in the same month; and "pt is an error term. The
7
three time variables corresponding to the three dates of interest in the an-tidumping investigation process are constructed as follows.
Post1pt =
�1 if t 2 [tp0; tp1)0 otherwise
; (2)
Post2pt =
�1 if t 2 [tp1; tp2)0 otherwise
; (3)
and
Post3pt =
�1 if t � tp20 otherwise
; (4)
where tp0 is the date of initiation (speci�cally, the month in which the case isinitiated) for product p; tp1 is the date of the preliminary ITC determinationfor product p; and tp2 is the date of the �nal ITC determination for productp. The coe¢ cients of interest in this study are �1, �2; and �3. To deal withthe potential heteroskedasticity and serial correlation, we cluster standarderrors at the product level (see Bertrand, Du�o, and Mullainathan, 2004).Note that in the above speci�cation (1), we use a binary variable to
classify the treatment status (i.e., Treatmentp), implicitly assuming thatantidumping duties are the same across products. In reality, however, dif-ferent products are charged with di¤erent antidumping duties. To explorethe e¤ect of the magnitude of antidumping duties, we also use an alternativeestimation speci�cation as follows
ypt = �1Treatmentp � Post1pt + �2Preliminary Dutiespt � Post2pt+�3Final Dutiespt � Post3pt + �p + �t + "pt; (5)
where Preliminary Dutiespt and Final Dutiespt are the antidumping dutiesimposed upon a¢ rmative preliminary and �nal determinations, respectively.4
The estimation speci�cations for the �rm-product level analysis are sim-ilar to speci�cations (1) and (5), with the only change being replacement ofthe outcome variable ypt and Final Dutiespt at the product level with thatat the �rm-product level.5
While the above speci�cations allow us to estimate average e¤ects ofantidumping investigations, we, in a robustness check, explore potential het-erogenous e¤ects across products. Speci�cally, we interact our regressors ofinterests with the elasticity of substitution at the HS-6 product level (dataobtained from Broda and Weinstein, 2006; see also Nizovtsev and Skiba,2010), and investigate whether the e¤ects are stronger for products withmore elastic demand.
4We thanks anonymous referees for suggesting this alternative estimation speci�cation.5Note that there are no within-product, across-�rms variations in the preliminary an-
tidumping duties.
8
3.2 Estimation Issues
We discuss a few estimation issues in this sub-section before proceeding todescribe our data in the section that follows.First, we construct two alternative control groups. The �rst control group
encompasses all una¤ected products/�rms within the HS-4 digit product cat-egory to which the a¤ected products/�rms belong (referred to as ControlGroup 1 ). The second control group is a matched group (referred to as Con-trol Group 2 ) constructed using the method employed by Blonigen and Park(2004). Speci�cally, we �rst estimate the probability of a product being sub-ject to antidumping investigations (see Table A.1 of the Appendix for theLogit regression results). The variables used to predict the probability ofbeing investigated for dumping include the import value of the product, thereal GDP growth rate in the U.S., an exchange rate index, a dummy vari-able indicating whether the product was previously subject to antidumpinginvestigations, and an HS 4-digit product dummy, similar to those used byBlonigen and Park (2004). The matched control group comprises una¤ectedproducts with predicted probabilities equal to at least the 75th percentileof the predicted probability of the treatment group (see also Konings andVandenbussche, 2008; Pierce, 2011).Second, the consistent estimation of f�1; �2; �3g hinges upon the assump-
tion that the di¤erence in the error term of the pre- and post-antidumpinginvestigation period for the treatment group is the same as the correspondingone for the control group, i.e.,
E [4"ptjTreatmentp = 1] = E [4"ptjTreatmentp = 0] : (6)
With panel data for multiple periods and multiple groups, we conduct two va-lidity checks following Angrist and Pischke (2009) and Imbens andWooldridge(2009). Firstly, as a speci�c check on whether there is any di¤erence in timetrends between the treatment and control groups before the initiation of anantidumping investigation, we add an additional regressor, Treatmentp �Prept, where
Prept =
�1 if t 2 [tp0 � 12; tp0)0 otherwise
; (7)
and the corresponding regression equation becomes
ypt = �0Treatmentp � Prept + �1Treatmentp � Post1pt+�2Treatmentp � Post2pt + �3Treatmentp � Post3pt+�p + �t + "pt: (8)
9
Any statistical signi�cance of �0 would indicate di¤erences in time trends be-tween the treatment and control groups before initiation of the antidumpinginvestigation, thereby invalidating the DID estimation.Secondly, to allow for the possibility that di¤erent HS-6 digit products
have di¤erent time trends, we further include product-speci�c linear timetrends and estimate the following equation
ypt = �1Treatmentp � Post1pt + �2Treatmentp � Post2pt+�3Treatmentp � Post3pt + �p + �p � t+ �t + "pt: (9)
A valid DID estimation requires that f�1; �2; �3g remain robust to the in-clusion of product-speci�c time trends (�p � t).Third, a potential concern is that antidumping investigations may coin-
cide with some other shocks to Chinese exports, therefore compounding thee¤ects of antidumping investigations. Speci�cally, we consider two importanttrade events that happened during our sample period. The �rst one is thesafeguard investigations by the U.S. government against Chinese exports.During the sample period (i.e., 2000-2006), there were totally 5 safeguardinvestigations against Chinese exports, but only two involved products in ei-ther our treatment or control groups, both of which ended up with negative�nal determinations. Nonetheless, to isolate the e¤ects of antidumping inves-tigations, we include, as a control, a dummy variable indicating the periodof safeguard investigations.The second one is China�s accession into the WTO by the end of 2001,
which led to a reduction of China�s import tari¤s and more competitivedomestic market, thereby possibly a¤ecting the exporting behavior of Chinese�rms.6 If the timing of China�s progressive tari¤ reduction coincides withthat of U.S. antidumping investigations, it would compound the e¤ect ofantidumping investigations. To address this concern, we include, as a control,China�s import tari¤s, in one of our robustness checks.
4 Data
Our study draws on data from two sources. The �rst is China Customs datafor the 2000-2006 period, which is generously provided by China Data Center
6Note that before China joined the WTO by the end of 2001, it already enjoyed themost-favored-nation (MFN) status from the U.S. (as well as its other major trading part-ners). Hence, Chinese exporters did not get much improvement in access to the U.S.market, though the elimination of uncertainty in the annual review of the MFN status didcontribute to the rise of China�s export to the U.S. (Pierce and Schott, 2012).
10
at Tsinghua University, Beijing.7 This data set covers the monthly importand export transactions of every Chinese exporter and importer, speci�callyincluding product information (classi�ed at the Chinese HS-8 digit level),trade volume, trade value, identity of Chinese exporter or importer, andexport destinations or importing countries. As our analysis focuses on an-tidumping cases taken by the U.S. against Chinese exporters, we extractinformation about monthly export transactions by Chinese exporters to theU.S.The second data source is the Global Antidumping Database (GAD) of
the World Bank, covering all antidumping cases around the world from 1980to 2010 (Bown, 2010). The GAD has detailed information on each antidump-ing case, such as product information (classi�ed at the U.S. HS-10 digit level),initiation date, preliminary determination dates and duties, and �nal deter-mination dates and duties. For our analysis, we collect information on allU.S. antidumping cases against China during our sample period (i.e., 2000-2006).We match the two data sets (i.e., the China Customs data and the GAD
data) at the HS-6 digit level, the most disaggregated level at which thetwo data sets are comparable. By doing so, we essentially aggregate exportinformation in the China Customs data from the Chinese HS-8 digit level tothe HS 6-digit level, and aggregate U.S. antidumping cases (against China)from the U.S. HS-10 digit level to the HS 6-digit level.There are a total of 47 U.S. antidumping cases against Chinese exporters
during the period of 2000-2006. Two cases (one in early 2000 and the other inlate 2006) are dropped as pre- or post-antidumping period is not long enoughfor us to carry out DID estimation. Three further cases are also dropped be-cause they overlap with earlier antidumping cases in the same HS-6 productcategories (see also Konings and Vandenbussche, 2008). Twenty-eight casesout of the remaining 42 ended with a¢ rmative �nal ITC determinations(referred to as successful cases); 5 out of the 6 cases that had a¢ rmativepreliminary ITC determinations received negative �nal ITC determinations(referred to as unsuccessful cases) and 1 was withdrawn before the �nal ITCdetermination (referred to as withdrawn cases); �nally, 8 cases were eitherwithdrawn before preliminary ITC determinations or given negative prelim-inary ITC determinations (referred to as terminated cases). As our analysislooks into the e¤ects of antidumping at the three di¤erent stages of the an-tidumping investigation (i.e., initiation, preliminary ITC determination, and
7The year 2000 is the earliest year when China Customs released this monthly tradetransaction data set; whereas the year 2006 is the latest year for which the data set isavailable to the authors.
11
�nal ITC determination), we focus on the sample of 28 successful cases in themain analysis. For a robustness check, we include the unsuccessful and with-drawn cases, and �nd our results remain qualitatively the same.8 See TableA.2 of the Appendix for a list of all the U.S. antidumping cases investigatedagainst Chinese exporters over the 2000-2006 period.The matched panel data from 2000 to 2006 contain 16,302 product-
month level observations and 800,079 �rm-product-month level observations.Among the 346 HS-6 digit product categories included in the matched data,81 product categories were successfully subject to antidumping duties.9 How-ever, as antidumping investigations take place at the U.S. HS-10 digit level(similar to the Chinese HS-8 digit level), one may be concerned about a po-tential aggregation bias, that is, some adjustments taking place at the HS-10digit level could not be detected at the HS-6 digit level. To address thisconcern, we conduct a robustness check by examining whether there are dif-ferential responses for HS-6 digit products with di¤erent numbers of HS-10digit products. The premise is that adjustments at the HS-10 digit levelshould be relatively easier for those HS-6 digit products with more HS-10digit products. Hence, a �nding of insigni�cant di¤erential responses wouldindicate that the concern of the aggregation bias is not a serious one in oursetting.One of the focuses of this paper is to investigate the possible heteroge-
neous response to antidumping investigations in light of the recent literatureon �rm heterogeneity and trade. We �rst follow the method developed byAhn, Khandelwal, and Wei (2011) for the same data by dividing �rms in oursample into trade intermediaries and direct exporters. Speci�cally, trade in-termediaries are identi�ed as �rms whose names contain Chinese characters( i.e., "Jinchukou", "Jingmao", and "Maoyi") with the English-equivalentmeaning of importer, exporter, and/or trading. The validity of this identi�-cation approach comes from the legacy of China�s centrally-planned systemand the reform strategy adopted after 1978. Speci�cally, to insulate the Chi-nese domestic market from international competition, the Chinese centralgovernment only authorized 12 state-owned enterprises to conduct exportand import in the pre-reform era (i.e., 1949-1978) and used these aforemen-tioned Chinese characters for easy identi�cation and regulation. Since 1978,China has adopted a gradualism approach in liberalizing its economy with
8We also experiment with other possible robustness checks involving changes in thesample of cases, such as combining the 28 successful cases with the only withdrawn case(as withdrawn cases are generally cases that end with a¢ rmative �nal ITC determinations)and combining the 28 successful cases with the 5 unsuccessful cases (as they all have datesfor preliminary and �nal ITC determinations), and �nd qualitatively similar results.
9Note that one antidumping case may involve several HS-6 digit product categories.
12
an increasing number of �rms allowed to conduct foreign trade. However,its tradition in using self-revealing names for trading corporations has con-tinued in the post-reform era. Ahn, Khandelwal, and Wei (2011) �nd that�rms identi�ed as trade intermediaries by this method are indeed very di¤er-ent from direct exporters in terms of the trading volume, product categories,and export destinations.Furthermore, we divide the sample of direct exporters into two types:
single-product exporters and multi-product exporters. Speci�cally, an ex-porter is identi�ed as a single-product exporter if it exports only one HS-6digit product to the U.S. before the initiation of an antidumping investiga-tion (referred to as Single-product exporters to the U.S.). However, thereis a concern that some of these single-product exporters may export someother products to countries other than the U.S. To relieve this concern, weconduct a robustness check by excluding those �rms that export other prod-ucts to countries other than the U.S. (the resulting subset of Single-productexporters to the U.S. is referred to as Single-product exporters to the U.S.and worldwide).For products subject to antidumping investigations during our sample
period, there were 9,356 exporters before the initiation of antidumping in-vestigations. Of these �rms, 3,465 were trade intermediaries. Among theremaining 5,891 direct exporters, 627 were single-product direct exportersto the U.S., and 265 were single-product direct exporters to the U.S. andworldwide.As the monthly data are quite noisy, we conduct a robustness check us-
ing quarterly instead of monthly data. Meanwhile, to further alleviate theconcern over outlying observations, we experiment by excluding the obser-vations at the top and bottom 1% of the corresponding outcome variables.Furthermore, as other countries across the world may conduct antidumpinginvestigations into the same products as those investigated by the U.S. inthe same period, this may confound our results. To alleviate this concern,we experiment by excluding cases (i.e., 4 in total) also being investigatedfor dumping in other countries. Finally, as some of China�s exporters con-duct processing trade with U.S. companies and a signi�cant percentage ofChina�s exporters are foreign-owned enterprises operating in China, we con-duct robustness checks by excluding processing trade from our sample andby focusing on the sub-sample of China�s indigenous exporters.
13
5 Empirical Findings
In this section, we �rst provide �ve baseline empirical �ndings regardinghow exporters respond to antidumping investigations in sub-sections 5.1-5.5.We then present a series of robustness checks on the validity of our DIDestimation and other econometric concerns in sub-section 5.6. We defer tothe next section for providing an explanation for our empirical �ndings.
5.1 Product-Level Quantity Response
We begin by examining the possible trade-dampening e¤ect of antidumpinginvestigations at the product level. Before presenting regression results re-garding equation (1), we plot time trends of export volume for the treatmentand control groups over the pre- and post-antidumping investigation periodsin Figures 1a-1d. The upper panel reports the results obtained using ControlGroup 1 (i.e., all una¤ected HS-6 digit products within the HS-4 digit prod-uct category to which the a¤ected product belongs), while the lower panelreports the results obtained using Control Group 2 (i.e., the matched controlgroup following Blonigen and Park, 2004). The left panel reports time trendsof export volume separately for the treatment group and the correspondingcontrol group, while the right panel reports the time trend of the di¤erencein export volume between the treatment group and the corresponding con-trol group. Each �gure contains three vertical dotted lines, from left to right,marking respectively the dates of initiation of the antidumping investigation,the preliminary ITC determination, and the �nal ITC determination.A few results emerge from these �gures. First, there is clearly an upward
trend in the export volume of both the treatment and control groups beforeinitiation of the antidumping investigation, consistent with the general trendof increasing Chinese exports to the U.S. in recent decades. Second, more im-portantly, before initiation of the antidumping investigation, the treatmentand control groups do not exhibit any di¤erential time trends, implying thatthere is no selection on the outcome variable and hence alleviating concernsabout the validity of our DID estimation. Third, there is a clear dampeninge¤ect of antidumping investigations on the export volume of the treatmentgroup, consistent with �ndings in the literature (e.g., Prusa, 2001; Vanden-bussche and Zanardi, 2010; Egger and Nelson, 2011). Fourth, regardingthe three di¤erent stages of antidumping investigations, we observe signif-icant e¤ects exerted by both a¢ rmative preliminary and a¢ rmative �nalITC determinations, but not by initiation of the investigation.10 Note that
10Di¤erent from ours, Staiger and Wolak (1994) �nd signi�cant e¤ects at the initition
14
the decline in export volume does not take place immediately after the a¢ r-mative determinations. One possible reason is that some existing contractsbetween U.S. importers and Chinese exporters need to be ful�lled despitethe issuance of the a¢ rmative determinations.Regression results corresponding to equation (1) are reported in Columns
1 and 2 of Table 1, where Control Group 1 and Control Group 2 are usedrespectively. It is found that both �2 and �3 are negative and statisticallysigni�cant at the 1% level, while �1 is negative but statistically insigni�cant,all of which are consistent with the �ndings revealed in Figures 1a-1d. Themarginal impact of an a¢ rmative preliminary ITC determination can be cal-culated as �̂2� �̂1 = �0:322 (from Column 1) or �0:315 (from Column 2),both signi�cant at the 1% level. Meanwhile, the marginal impact of an a¢ r-mative �nal ITC determination can be calculated as �̂3� �̂2 = �0:651 (fromColumn 1) or �0:665 (from Column 2), both signi�cant at the 1% level. Interms of economic magnitude, an a¢ rmative preliminary ITC determinationcauses the growth of export volume for the treatment group to lag behindthat for the control group by around 32% during the period between thepreliminary and �nal ITC determinations. An a¢ rmative �nal ITC deter-mination causes the growth of export volume for the treatment group to lagfurther behind that for the control group by 65 � 67% from the date of the�nal ITC determination to the end of our sample period.11
In Columns 3 and 4 of Table 1, we replace the binary variables for thetreatment status with the continuous variables of antidumping duties to in-corporate the fact that the magnitude of antidumping duties varies acrossproducts, i.e., the estimation speci�cation (5). Consistently, we �nd thatboth the preliminary duties and the �nal duties have negative and statisti-cally signi�cant impacts on export volume at the product level. In terms ofthe magnitude, an one-standard-deviation increase in the preliminary (�nal)duties leads to a decrease in export volume by around 23% (25%) during theperiod between the preliminary and �nal ITC determinations (from the dateof the �nal ITC determination to the end of our sample period).
5.2 Extensive Versus Intensive Margins
Now that we have documented a substantial dampening e¤ect of antidumpinginvestigations on export volume, we next anatomize this e¤ect by investigat-ing its underlying mechanism. Speci�cally, we look at the e¤ect of antidump-
of antidumping investigations for 450 U.S. manufacturing industries from 1958 to 1985.11Prusa (2001) shows that antidumping duties cause the value of imports to fall by an
average of 30-50%, while Egger and Nelson (2011) �nd a modest impact of antidumpingduties using structural estimation of the gravity model.
15
ing investigations on both the number of exporters to the U.S. (the extensivemargin e¤ect) and export volume per exporter (the intensive margin e¤ect).Figures 2a-2d plot time trends of the number of exporters for the treat-
ment and control groups over the pre- and post-antidumping investigationperiods. Clearly, antidumping investigations cause a signi�cant decrease inthe number of exporters. Speci�cally, between the initiation of an antidump-ing investigation and the preliminary ITC determination, there is barely anychange in the number of exporters. However, after an a¢ rmative prelimi-nary ITC determination, the number of exporters decreases sharply, followedby another substantial decrease upon the release of an a¢ rmative �nal ITCdetermination.Figures 3a-3d present time trends of export volume per exporter for the
treatment and control groups over the pre- and post-antidumping investiga-tion periods. There is a decrease in export volume per exporter upon therelease of an a¢ rmative preliminary ITC determination, followed by a muchbigger decrease upon the a¢ rmative �nal ITC determination.Note that in the above analysis on the intensive margin e¤ect, we di-
vide total export volume by total number of exporters. That is, for thepre-investigation period, we include exporters that eventually exit the U.S.market due to antidumping investigations, and for the post-investigation pe-riod, we include new entrants into the U.S. export market. As it is a conven-tion in the literature to de�ne the intensive margin e¤ect as the growing orshrinking of exporters that survive throughout the entire period (referred toas surviving exporters), the aforementioned results (Figures 3a-3d) are com-pounded by the entry and exit of exporters, and hence may not fully capturethe intensive margin e¤ect. Accordingly, we follow the literature by focusingon surviving exporters to examine the intensive margin e¤ect of antidumpinginvestigations. Figures 4a-4d present time trends of export volume for sur-viving exporters and their control groups over the pre- and post-antidumpinginvestigation periods. Interestingly, in contrast to the aforementioned inten-sive margin results for all exporters, there is no clear di¤erential time trendbetween the export volumes of surviving exporters and their control groupsboth before and after the antidumping investigation. In other words, we �ndno evidence supporting the intensive margin e¤ect, and the patterns shownin Figures 3a-3d are mainly caused by the entry and exit of exporters.Regression results regarding the extensive margin e¤ects of antidumping
investigations are reported in Columns 1-4 of Table 2. In Columns 1 and 2where a binary variable of treatment status is used (i.e., estimation speci�ca-tion (1)), we �nd both �2 and �3 to be negative and statistically signi�cantat the 1% level. Similar results are found when we use antidumping dutiesinstead of the binary variable in Columns 3-4. These results are consistent
16
with the �ndings revealed in Figures 2a-2d, implying that antidumping in-vestigations exert a strong extensive margin e¤ect. In terms of economicmagnitude, an a¢ rmative preliminary ITC determination (an one-standard-deviation increase in the preliminary duties) leads to a decrease in the numberof exporters by around 16% (10%) during the period between the preliminaryand �nal ITC determinations. And an a¢ rmative �nal ITC determination(an one-standard-deviation increase in the �nal duties) leads to a decreasein the number of exporters by around 17% � 18% (7%) from the �nal ITCdetermination until the end of our sample period.In Columns 5-12 of Table 2, we report regression results regarding the in-
tensive margin e¤ects of antidumping investigations. When using the sampleof all exporters subject to antidumping investigations, we �nd in Columns5-8 that both a¢ rmative determinations (i.e., �2 and �3) and antidumpingduties (i.e., �4 and �5) have negative and statistically signi�cant impactson the export volume per exporter. These results are consistent with the�ndings revealed in Figures 3a-3d.To alleviate the concern that these results of intensive margin e¤ects
are compounded by the entry and exit of exporters, we, in the remainingcolumns of Table 2, focus on the sample of surviving exporters. Interestingly,when we do not di¤erentiate products by the magnitude of their antidumpingduties, we �nd that, on average, products being investigated do not exhibitsigni�cant intensive margin e¤ects throughout all the stages of antidumpinginvestigations (i.e., Columns 9-10). These results are consistent with the�ndings revealed in Figures 4a-4d. However, once we use antidumping duties(varying across products) instead of the binary treatment status variable, we�nd negative and statistically signi�cant (at the 1% level) intensive margine¤ects of both preliminary and �nal antidumping duties (i.e., Columns 11-12). In terms of economic magnitude, an one-standard-deviation increasein the preliminary (�nal) duties leads to a decrease in export volume perexporter by around 7% (7%) during the period between the preliminary and�nal ITC determinations (from the �nal ITC determination until the end ofour sample period).
5.3 Heterogeneous Responses
In the previous section, we document that much of the trade-dampening ef-fect of antidumping investigations is attributed to the sharp decrease in thenumber of exporters upon both the a¢ rmative preliminary and a¢ rmative�nal ITC determinations. We are interested in knowing what kinds of ex-porters are more likely to exit the export market at these two importantdates in the antidumping investigation process. The recent development in
17
trade literature centers on how �rm heterogeneity, in particular �rm produc-tivity, a¤ects exporting behavior. Hence, we start with looking at whethermore productivity exporters are less likely to exit after the a¢ rmative an-tidumping determinations. Meanwhile, more recent studies in internationaltrade have gone beyond �rm productivity by looking at di¤erent types of ex-porters, i.e., trade intermediaries versus direct exporters, and single-productversus multi-product direct exporters. Following these lines of studies, wealso look at the possible di¤erences in exiting likelihood among these typesof exporters.
5.3.1 Firm Productivity
Unfortunately, due to data limitation, we do not have the information fromthe China Customs data to measure �rm productivity directly.12 Instead,we use export volume as a proxy for �rm productivity.13 Indeed, by mergingthe China Customs data with China�s annual surveys of manufacturing �rms(that contain the information about �rm productivity), we �nd a positiveand signi�cant correlation between export volume and �rm productivity.An exporter is classi�ed as exiting the U.S. market following an a¢ r-
mative preliminary (�nal) ITC determination if it does not export any ofthe a¤ected HS-6 digit products after the a¢ rmative preliminary (�nal) ITCdetermination (denoted as Exit). The regression speci�cation is as follows:
Exitfp = � Export V olumefp + �p + "fp; (10)
where the control of product dummy (�p) allows us to compare �rms of exit-ing likelihood within a narrowly-de�ned product category (i.e., HS-6 productlevel). Speci�cation (10) is estimated using the Probit model.As shown in Columns 1 and 3 of Table 3, among all exporters, exporters
with larger export volumes are less likely to exit after both a¢ rmative pre-liminary and �nal determinations, albeit the latter is imprecisely estimated.Such results also hold for the sample of direct exporters (i.e., Columns 1 and3 of Table 4) and the control for di¤erent types of exporters (Columns 2 and
12The China Customs data only have the information about the output (i.e., exportvolume and export value) but not the information about the inputs (i.e., labor, capitaland materials), which prohibits us from calculating �rm productivity.13Export price is not a good proxy for �rm productivity due to a number of reasons.
Firstly, more productive exporters may charge lower prices due to their lower productioncosts; but they could also charge higher prices because of the higher quality of theirgoods. Secondly, higher export prices may diminish the likelihood of a �rm being imposedby antidumping duties, and hence have a direct in�uence on its exit likelihood, whichcompounds the results using export price as a proxy for �rm productivity.
18
4 of Tables 3-4) as well as �rm-speci�c antidumping duties (Columns 5 ofTables 3-4). Given that more productive �rms export more to the U.S., theseresults imply that more productive exporters are more likely to survive thenegative shocks (brought out by antidumping investigations).
5.3.2 Trade Intermediaries versus Direct Exporters
Table 3 also reports regression results regarding the di¤erences in the likeli-hood of exiting the U.S. market between trade intermediaries and direct ex-porters, with the regressor of interest being Trade Intermediary that takesthe value of 1 if the exporter is a trade intermediary and 0 otherwise.As shown in Columns 2 and 4 of Table 3, Trade Intermediary has nega-
tive and statistically signi�cant estimated coe¢ cients, suggesting that tradeintermediaries are less likely to exit the U.S. market for the a¤ected productsthan are direct exporters after both an a¢ rmative preliminary ITC determi-nation and an a¢ rmative �nal ITC determination. These results are robustto the control of �rm productivity (proxied by export volume), and also thecontrol of �rm-speci�c �nal antidumping duties.14
5.3.3 Single-Product versus Multi-Product Direct Exporters
In Table 4, we examine the relative likelihood of exit from the U.S. marketfor the a¤ected products between single-product and multi-product directexporters following both an a¢ rmative preliminary ITC determination andan a¢ rmative �nal ITC determination, where the key regressor is SingleProduct taking the value of 1 if the direct exporter is a single-product directexporter to the U.S. and 0 otherwise.Column 2 of Table 4 reports results regarding the likelihood of exit af-
ter an a¢ rmative preliminary ITC determination. It is found that SingleProduct has a negative and statistically signi�cant estimated coe¢ cient. In-terestingly, however, the estimated coe¢ cients of Single Product becomepositive and statistically signi�cant when we examine the likelihood of exitfollowing an a¢ rmative �nal ITC determination (Columns 4-5 of Table 4).Together, our results suggest that single-product direct exporters are lesslikely to exit the U.S. market for the a¤ected products than are multi-productdirect exporters following an a¢ rmative preliminary ITC determination, butthe opposite holds after an a¢ rmative �nal ITC determination.15
14The lack of signi�cance of the antidumping duties variable could be due to the cor-relation between export volume and duties (�rms with greater export volumes are givenlower duties) and the small variations in the magnitude of duties across �rms within thesame product categories.15In an unreported regression (available upon request), by combining exiters at both
19
5.4 Price Response
We now proceed to analyze possible e¤ects of antidumping investigations onexport prices (i.e., F.O.B. prices charged by Chinese exporters), �rst at theproduct level and then at the �rm-product level.Figures 5a-5d present time trends of export prices of a¤ected HS-6 digit
products and their control groups over the pre- and post-antidumping in-vestigation periods. Interestingly, there is no clear di¤erential time trendof export prices between the treatment and control groups either before orafter an antidumping investigation. Figures 5b and 5d clearly show that thedi¤erence in export prices between the treatment and control groups is quite�at throughout the whole sample period, despite of some �uctuations.Figures 6a-6d present time trends of export prices of a¤ected products
among surviving �rms and those of their control groups over the pre- andpost-antidumping investigation periods. We still �nd no clear di¤erentialtime trend of export prices between the treatment and control groups eitherbefore or after an antidumping investigation. Moreover, Figures 6b and 6dclearly demonstrate there is little di¤erence between export prices of a¤ectedproducts of surviving �rms and those of their control groups throughout thewhole sample period, despite of some �uctuations.Regression results regarding the e¤ects of antidumping investigations on
export prices at the product level are reported in Columns 1-4 of Table 5,where a binary treatment status variable is used Columns 1-2 and antidump-ing duties are used in Columns 3-4. There is no signi�cant price change uponthe a¢ rmative preliminary determination (preliminary duties), but modestand statistically signi�cant price increase upon the a¢ rmative �nal determi-nation (�nal duties).The product level analysis may, however, be contaminated by the entry
and exit of exporters upon a¢ rmative antidumping determinations. To cap-ture the price e¤ect of antidumping investigations more precisely, we focuson the sample of surviving exporters in Columns 5-8 of Table 5. Consistentwith the �ndings revealed in Figures 6a-6d, no signi�cant price changes aredetected when we pool all products together without controlling for the dif-ference in the magnitude of antidumping duties (i.e., Columns 5-6). However,when using antidumping duties as the regressor of interest in Columns 7-8,we �nd that surviving exporters have statistically signi�cant, albeit smallin magnitude (around 2%), price increases upon the imposition of prelimi-nary antidumping duties. And there is no further price increase after theimposition of �nal antidumping duties.
a¢ rmative preliminary and �nal ITC determinations, we �nd that a larger percentage ofsingle-product direct exporters exit the U.S. market than their multi-product counterparts.
20
5.5 Trade De�ection Response
In this subsection, we examine whether Chinese exporters respond to U.S.antidumping investigations by diverting their exports to countries other thanthe U.S., namely the trade de�ection response (e.g., Bown and Crowley,2007).Figures 7a-7d present time trends of total export volume to other coun-
tries of a¤ected HS-6 digit products and their control groups over the pre-and post-antidumping investigation periods. Clearly, there is no clear dif-ferential time trend of total export volume to other countries between thetreatment and control groups either before or after an antidumping investi-gation. Figures 7b and 7d clearly show that the di¤erence in total exportvolume to other countries between the treatment and control groups is quite�at throughout the whole sample period, despite of some �uctuations. Re-gression results reported in Table 6 rea¢ rm the �ndings revealed in Figures7a-7d.In unreported tables (available upon request), we investigate possible
trade de�ection to Canada or OECD countries (that may share similar eco-nomic structures as the U.S.), and among di¤erent types of Chinese exporters(i.e., trade intermediaries, single-product direct exporters, and multi-productdirect exporters). None of these exercises yields any signi�cant e¤ects of an-tidumping investigations on trade de�ection.
5.6 Robustness Checks
In this section, we conduct a series of robustness checks on the aforementionedDID estimation results for all the relevant outcome variables examined insub-sections 5.1-5.5 (i.e., quantity response, extensive and intensive margine¤ects, price response, and trade de�ection response).First, the validity of our DID estimation hinges upon the assumption that
the treatment and control groups are comparable before the treatment occurs.To check speci�cally whether there is any di¤erence in time trends betweenthe treatment and control groups before the initiation of an antidumpinginvestigation, we conduct a robustness check according to speci�cation (8).The estimation results are summarized in Tables A.3.1-A.3.2 of the Appen-dix. Clearly, there is no evidence of any di¤erential time trends betweenthe treatment and control groups before the initiation of an antidumpinginvestigation, thus lending support to the validity of our DID estimations.Our main �ndings on the e¤ects of antidumping investigations also remainrobust.Second, one may be concerned that products in the treatment group
21
and their counterparts in the control group may follow di¤erent time trends.To address this concern, we allow for product-speci�c time trends in ourestimation, i.e., speci�cation (9). The estimation results are reported inTables A.4.1-A.4.2 of the Appendix. Clearly, our main �ndings on the e¤ectsof antidumping investigations remain robust to the inclusion of product-speci�c time trends, again implying that our DID estimations are valid.Third, to alleviate the concern that our monthly data could be noisy
as not all exporters export to the U.S. every month, we conduct a robust-ness check by using quarterly instead of monthly data (i.e., aggregation ofmonthly export transactions to the quarterly level). The regression resultsare reported in Tables A.5.1-A.5.2 of the Appendix. It is found that in ad-dition to the statistically signi�cant impacts of antidumping investigationsreported earlier, a �nal ITC determination has a negative and signi�cant(at the 5% level) impact on the export volume of surviving exporters (i.e.,providing limited evidence supporting the intensive margin e¤ect). In addi-tion, the magnitudes of the e¤ects for the sample of quarterly data are muchbigger.Fourth, to further address the concern that our results may be a¤ected
by some outlying observations, we focus on a sub-sample excluding the ob-servations at the top and bottom 1% of the corresponding outcome variables.Regression results reported in Tables A.6.1-A.6.2 of the Appendix show therobustness of our earlier �ndings and o¤er limited evidence supporting theintensive margin e¤ect.Fifth, note that in sub-sections 5.1-5.5, we include only successful an-
tidumping cases (i.e., 28 cases with a¢ rmative preliminary and a¢ rmative�nal ITC determinations out of 42 antidumping cases), partly because weseek to investigate the di¤erential impacts of a¢ rmative preliminary ITC de-terminations and a¢ rmative �nal ITC determinations. To check whether ourmain results are sensitive to the selection of antidumping cases, we conducta robustness check by including the 5 unsuccessful cases and 1 withdrawncase. The regression results are reported in Tables A.7.1-A.7.2 of the Appen-dix. It is found that our main results regarding the e¤ects of antidumpinginvestigations remain qualitatively the same as those reported earlier.16
Sixth, it is possible that other countries conduct antidumping investiga-
16It is noted that the impacts of both a¢ rmative preliminary and �nal ITC determi-nations are smaller than those obtained using the original sample of 28 successful cases.Intuitively, as the �nal ITC determinations are negative for the 5 unsuccessful cases, theinclusion of these cases dilutes the e¤ects of the �nal ITC determinations. Meanwhile,the smaller impacts of the preliminary ITC determinations with the inclusion of the 5unsuccessful cases suggest that the evidence for these cases is less convincing and hencethe limited impacts.
22
tions into the same products as those examined by the U.S. during the sameperiod, thereby confounding the e¤ects of the U.S. antidumping investiga-tions on Chinese exporters and complicating the interpretation of our results.To address this concern, we conduct a robustness check by excluding suchoverlapping antidumping cases (i.e., 4 cases). The regression results are re-ported in Tables A.8.1-A.8.2 of the Appendix. Clearly, our main �ndingsremain robust to this sub-sample.Seventh, as some of Chinese exporters conduct processing trade with
U.S. companies, one may concern whether antidumping investigations mayhave di¤erent impacts on Chinese processing traders than those ordinarytraders, and hence compounding our �ndings. To alleviate this concern,we conduct a robustness check by excluding processing traders from oursample,17 and �nd our results remain robust (see Tables A.9.1-A.9.2 in theAppendix). In addition, as a signi�cant percentage of China�s exporters areforeign-owned enterprises operating in China rather than China�s indigenous�rms, one may wonder if foreign-owned exporters respond di¤erently fromChina�s indigenous exporters. To investigate this possibility, we conducta robustness check using the sub-sample of China�s indigenous exporters,and again �nd our results remain robust (see Tables A.10.1-A.10.2 in theAppendix).Eighth, to address the concern of a potential aggregation bias, we conduct
a robustness check by including interaction terms between our key explana-tory variables with the number of HS-10 digit products within each HS-6digit product. The regression results are reported in Tables A.11.1-A.11.2of the Appendix. It is found that none of these interaction terms has anystatistical signi�cance. Meanwhile, our main �ndings remain robust to theinclusion of these interaction terms. These results imply that our �ndingsare not a¤ected by the potential aggregation bias.Ninth, another potential concern is that the timing of antidumping in-
vestigations may coincide with other shocks to trade environment, therebycontaminating the e¤ects of antidumping investigations. To alleviate suchconcerns, we consider two important trade shocks that happened during oursample period, speci�cally, U.S. safeguard investigations against Chinese ex-porters and China�s accession into the WTO. Regression results includingan additional control for U.S. safeguard investigations are reported in TablesA.12.1-A.12.2, while regression results including an additional control forChina�s import tari¤ rates are reported in Tables A.13.1-A.13.2. It is found
17In the Customs data, there is information regarding the nature of trade, such asordinary trade and di¤erent types of processing trade (including processing exports withassembly, processing exports with imported materials, foreign aid, compensation trade,etc). In this robustness check, we only include ordinary trade.
23
that our main �ndings regarding the e¤ects of antidumping investigationsremain robust to the control of these two important trade shocks.Tenth, the aforementioned exercises give us the average e¤ects of an-
tidumping investigations. To explore potential heterogenous e¤ects acrossproducts, we consider a key di¤erence among products, namely, the elasticityof import substitution. Estimation results reported in Tables A.14.1-A.14.2.reveal little di¤erential e¤ects of antidumping investigations across productswith di¤erent elasticity of import substitution. This can be explained by thelimited variations in the elasticity of import substitutions of Chinese exports.Eleventh, in the investigation of the di¤erential exit likelihood between
single-product direct exporters and multi-product direct exporters, we de�nethe former as those selling only the concerned product to the U.S. market.It is possible, however, that these single-product direct exporters may sellsome other products to countries other than the U.S. In other words, thesesingle-product direct exporters are arguably multi-product direct exportersin a broader sense. To better delineate the di¤erence between single-productand multi-product direct exporters, we adopt a stricter de�nition of the for-mer, i.e., Single-product exporters to the U.S. and worldwide, and carry outa robustness check on Table 4. Regression results are reported in Table A.15in the Appendix. Evidently, we �nd similar results to those in Table 4, i.e.,single-product direct exporters are less likely to exit the U.S. market upon thea¢ rmative preliminary ITC determination than their multi-product counter-parts, but the opposite holds upon the a¢ rmative �nal ITC determination.
6 Discussion
In this section, we discuss our aforementioned empirical �ndings along theline of di¤erences across products (i.e., extensive and intensive margin e¤ects,price response and trade de�ection response), and that of di¤erences across�rms within products (i.e., exit likelihood).
6.1 Di¤erences across Products
Our empirical investigations reveal the following e¤ects of U.S. antidumpinginvestigations on Chinese exporters across products:
� there is signi�cant extensive margin e¤ect, i.e., a sharp decrease in thenumber of exporters;
� intensive margin e¤ect is not found when a binary variable of treatmentstatus is used, but uncovered when the antidumping duties are used;
24
� there is little adjustment in F.O.B. export prices when a binary variableof treatment status is used, but a modest increase in prices when theantidumping duties are used;
� there is no evidence of trade de�ection to other countries.
The contrasting �ndings obtained when di¤erent measures of antidump-ing investigations are used (i.e., a binary variable of treatment status versusantidumping duties) imply that products facing di¤erent levels of antidump-ing duties may behave di¤erently.18 To further investigate and hence enhanceour understanding of how exporters respond to antidumping investigations,we examine the possibly di¤erential e¤ects of antidumping investigations inthree quantiles corresponding to low (i.e., < 50%), medium (i.e., 50 � 100%)and high (i.e., > 100%) antidumping duties.The regression results for these three quantiles are summarized in Columns
1-4 of Table 7.19 It is found that there is no signi�cant and di¤erential tradede�ection response across the three quantiles (Column 4 of Table 7). Onepossible explanation for the consistent �ndings of no trade de�ection is thatthe �xed costs of exporting are country-speci�c (e.g., Chaney, 2008; Arko-lakis, 2010), as a result of which the decision to enter each foreign market isindependent. Indeed, we �nd in our data that Chinese exporters to the U.S.are heavily weighted in the U.S. market, i.e., about 63% of these exporters�world export revenues comes from the U.S. market.More interestingly, we �nd di¤erent extensive margin e¤ect, intensive
margin e¤ect and price e¤ect (Columns 1-3 of Table 7). Speci�cally, forproducts in the low quantile, there is no extensive margin e¤ect, signi�cantintensive margin e¤ect, and signi�cant price increase. For products in thehigh quantile, there is a substantial extensive margin e¤ect, no intensivemargin e¤ect and modest price increase. For products in the medium quan-tile, there is a marginally signi�cant extensive margin e¤ect, no signi�cantintensive margin e¤ect and no signi�cant price increase.Low Quantile. Note that the margins of antidumping duties in the
low quantile are quite small, i.e., average about 21% (Column 5 of Table7). Given such a small magnitude of duties, the a¢ rmative determinationsshould be viewed as surprises to the producers in these product categories.According to Blonigen and Park (2004), exporters under such a scenario will
18The exercises using the binary treatment status variable estimate the e¤ects upona¢ rmative determinations without taking into account the magnitude of antidumpingduties. To the extent that the e¤ects are stronger in the case of higher duties, theseestimates are overly in�uenced by products facing higher duties.19For detailed regression results, see Table A.16 in the Appendix.
25
raise prices over time. Meanwhile, the small magnitude of duties makes iteasy for producers to raise prices, and by doing so, they can get rid of theduties (and nuisances) in the future years through administrative reviews.Moreover, raising prices is achievable given that the demand for these prod-ucts are relatively inelastic (i.e., the average elasticity of import substitutionis 3:1; see Column 6 of Table 7). Interestingly, the estimation results suggestthat producers in these product categories increase prices by around 24%,which is similar in magnitude to the duties imposed.Given the imposition of antidumping duties and the rise of F.O.B. export
prices, it is expected that the �nal sales prices of the concerned products inthe U.S. market will generally increase, which consequently leads to a declinein demand for the concerned products. At the status quo, shrinking marketdemand is likely to lead to a decrease in �rm export volume across the board.Indeed, it is found that the average export volume per exporter decreases by20%, or a signi�cant intensive margin e¤ect. However, these exporters couldcompensate the loss in demand by the increase in their F.O.B. export prices,i.e., 24% increase in export prices versus 20% decrease in export volume.As a result, there is limited loss in the total revenue of each exporter, whichexplains why there is no signi�cant exit of exporters, i.e., no extensive margine¤ect.High Quantile. Note that the margins of antidumping duties in the
high quantile are extremely large, i.e., average about 185% (Column 5 ofTable 7). Given such huge negative shocks (i.e., the substantial increasein the �nal sales prices), the demand for the concerned products in the U.S.market declines substantially, making the life of weak exporters really di¢ cultand pushing them to exit the market. Meanwhile, there is a large degree ofexporter heterogeneity among these product categories (the highest of thethree quantiles), i.e., the average coe¢ cient of variations (measured as thestandard deviation of export volume divided by its mean) is 0:32 (in Column7 of Table 7), implying that a large number of weak �rms may exit the U.S.market after the a¢ rmative antidumping determinations. Indeed, we �ndthat the number of exporters fall by 52% (a substantial extensive margine¤ect).Given the wipeout of so many weak exporters, the concerned product
markets should have gone through a great consolidation, under which moreproductive exporters are able to not only survive but even grow after theimposition of antidumping duties. Indeed, we �nd that surviving exportersmodestly increase their F.O.B. prices and maintain their export volume bygrabbing the market share left by the exiting �rms.Medium Quantile. The margins of antidumping duties in the medium
quantile are quite signi�cant, averaging about 87% (Column 5 of Table 7).
26
The negative shocks to the concerned products are certainly larger than thoseto the products of the low quantile, but not as devastating as those in thehigh quantile. Meanwhile, there is a limited degree of exporter heterogeneityamong these product categories, i.e., being the lowest of the three quantiles.Combined, a signi�cant percentage of exporters is expected to exit the U.S.market, though not to the same extent as that we observe in the high quantile.Indeed, we �nd that the number of exporters fall by less than one third (andmarginally statistically signi�cant).Given that the elasticity of import substitution in this quantile is the
largest of the three, the markets may remain competitive even after experi-encing a modest shake-out. As a result, the surviving exporters may havedi¢ culties in raising their F.O.B. export prices. In fact, we observe modest,albeit statistically insigni�cant, decreases in export prices by 5%. Meanwhile,as the �nal sale prices of the concerned products increase on one hand andthe limited exit of weak exporters from the U.S. market on the other hand,surviving exporters may have di¢ culties in maintaining their ex ante exportvolume. Indeed, we observe a decrease in the export volume of survivingexporters by 14%, albeit statistically insigni�cant.
6.2 Di¤erences across Firms within Products
Our empirical investigations reveal the following e¤ects of U.S. antidumpinginvestigations on Chinese exporters across �rms within products:
� Productivity e¤ect: less productive exporters are more likely to exitthe U.S. market;
� Trade intermediaries versus direct exporters: direct exporters are morelikely to exit the U.S. market than trade intermediaries;
� Single-product direct exporters versus multi-product direct exporters:multi-product direct exporters are more likely to exit the U.S. marketthan single-product direct exporters upon issuance of an a¢ rmativepreliminary ITC determination, but the opposite holds following ana¢ rmative �nal ITC determination.
Productivity E¤ect. The observed relation between productivity andexit likelihood is in line with the �rm heterogeneity literature. Speci�cally, inthe case of a per-period �xed cost of exporting, the Melitz (2003) model showsthat when facing negative shocks induced by antidumping investigations,exporters experience a fall in their revenue, making some less productiveones unable to recover the per-period �xed cost of exporting and thereby
27
forcing them to exit from the U.S. market. In the world without �xed cost ofexporting, the Melitz and Ottaviano (2008) model suggests that the negativeshock causes a decrease in exporters�markups, as a result of which some lessproductive ones incur losses and hence exit the U.S. market.Trade Intermediaries vs. Direct Exporters. The observed di¤eren-
tial likelihood of exiting between trade intermediaries and direct exporterssuggests that these two types of exporters are rather di¤erent in their ex-porting behavior. Instead of arbitrarily picking theories to explain theirdi¤erences, we strive to o¤er an explanation that is grounded on observeddi¤erences between trade intermediaries and direct exporters in the data.It is found that trade intermediaries are more multi-market and more
multi-product oriented than direct exporters. Speci�cally, on average (acrossall a¤ected products), 91% of trade intermediaries sell products other thanthe a¤ected products, whereas the corresponding number for direct exportersis 81%. Meanwhile, 68% of trade intermediaries sell the a¤ected products tocountries other than the U.S., whereas the corresponding number for directexporters is 64%.The multi-market and multi-product nature of trade intermediaries equips
them with more capabilities to tap into their reserves in other products andother markets and hence to cross-subsidize their a¤ected products in the U.S.,which allows them to better weather the storms brought by antidumpinginvestigations than direct exporters.Single-product Direct Exporters vs. Multi-product Ones. We
observe that overall single-product direct exporters are more likely to exit theU.S. market in response to antidumping investigations than multi-productdirect exporters (see footnote 15). This can be explained by the greatercapabilities of multi-product direct exporters to cross-subsidize the a¤ectedproducts than their single-product counterparts, which is in line with ouraforementioned explanation on the di¤erential likelihood of exiting betweentrade intermediaries and direct exporters.What is more intriguing is the observed reversal in the di¤erential like-
lihood of exiting between the two types of direct exporters across di¤erentstages of antidumping investigations: higher likelihood of exiting by multi-product direct exporters upon a¢ rmative preliminary ITC determinationsbut by single-product direct exporters upon a¢ rmative �nal ITC determina-tions.Such reversal suggests that there should be di¤erences between prelim-
inary and �nal ITC determinations. According to the regulations of U.S.antidumping investigations, upon an a¢ rmative preliminary ITC determina-tion, exporters need to pay deposits (similar in magnitude to the estimatedmargins of �nal antidumping duties), which will then be used to cover the
28
antidumping duties incurred between the preliminary and �nal ITC determi-nations in the case of an a¢ rmative �nal ITC determination. However, thereare many cases with a¢ rmative preliminary yet negative �nal ITC determi-nations, in which case the deposits are refunded. In fact, during the period of2000-2006, 6 out of 34 U.S. antidumping investigations against Chinese ex-porters that had a¢ rmative preliminary ITC determinations ended up withnegative �nal ITC determinations.Such uncertainty on the outcome of the �nal ITC determination can gen-
erate di¤erent exiting behavior across di¤erent exporters. In the Appendix,we build a simple model to examine the exiting behavior of exporters upondi¤erent stages of antidumping investigations. It is found that the weakestexporters exit immediately upon the a¢ rmative preliminary ITC determina-tions, whereas the strongest exporters stay throughout the whole antidump-ing investigation process and even after the a¢ rmative �nal ITC determina-tions. Interestingly, those in the middle choose to stay after the a¢ rmativepreliminary ITC determinations but decide to exit upon the a¢ rmative �nalITC determinations.The di¤erential exit likelihood between single-product direct exporters
and multi-product direct exporters upon di¤erent stages of antidumping in-vestigations implies that single-product direct exporters are relatively moreconcentrated in the middle range of export volume, while multi-product di-rect exporters are scattered more in the lowest and highest ranges of exportvolume; in other words, multi-product direct exporters are more heteroge-nous than their single-product counterparts. Indeed, we �nd that the averagecoe¢ cient of variations for single-product direct exporters is 0:23, while thecorresponding number for multi-product direct exporters is 0:32. Intuitively,multi-product direct exporters in the highest range of export volume couldrepresent exporters producing the a¤ected products as their core productsand also selling other peripheral products; whereas those in the lowest rangeof export volume are just the opposite.
7 Conclusion
Antidumping measures have become a popular tool enabling governments toprotect their domestic �rms and industries. Much insight has been gainedfrom a large and growing literature on how e¤ective antidumping measuresare in trade protection. An equally important but overlooked issue is howantidumping measures a¤ect the behavior of foreign exporters, the under-standing of which should help us gain a complete picture of the e¤ectivenessof such measures.
29
In this paper, we use China Customs data to investigate how Chineseexporters respond to U.S. antidumping investigations during the 2000-2006period. To identify the e¤ects of antidumping investigations, we use thedi¤erence-in-di¤erences estimation strategy involving the comparison of out-come variables of exporters in the a¤ected product categories with those ofexporters in una¤ected product categories before and after the various im-portant stages of the antidumping investigation process.We �nd that much of the trade-dampening e¤ect of antidumping inves-
tigations at the product level operates through the extensive margin (i.e., adecrease in the number of exporters) rather than the intensive margin (i.e.,a decrease in export volume per exporter), especially among product cat-egories facing high margins of antidumping duties. We also �nd that thebulk of the decrease in the number of exporters is exerted by less productiveexporters, by direct exporters as opposed to trade intermediaries (who aremore multi-market and multi-product oriented), and by single-product directexporters as opposed to their multi-product counterparts. Combined withthe existing studies on protected �rms, our results imply that antidumpinginvestigations may spell more troubles for U.S. domestic producers in theircompetition with the Chinese exporters in the long-run, especially when theantidumping duties are lifted.This paper also documents patterns of exiting behavior across di¤erent
types of exporters, and at di¤erent stages of antidumping investigations.These new stylized facts call for a more general theory that accounts for boththe entry and exit decisions of exporting, and addresses interactions between�rm heterogeneity and complex market structures and policy uncertainty.
30
References
[1] Ahn, JaeBin, Amit Khandelwal, and Shang-jin Wei. 2011. "The Roleof Intermediaries in Facilitating Trade", Journal of International Eco-nomics 84, 73-85
[2] Angrist, Joshua D. and Jörn-Ste¤en Pischke. 2009. Mostly HarmlessEconometrics, Princeton: Princeton University Press
[3] Arkolakis, Costas. 2010. "Market Penetration Costs and the New Con-sumers Margin in International Trade", Journal of Political Economy118, 1151-1199
[4] Autor, David, David Dorn, and Gordon Hanson. Forthcoming. "TheChina Syndrome: Local Labor Market E¤ects of Import Competition inthe United States", American Economic Review
[5] Bertrand, Marianne, Esther Du�o, and Sendhil Mullainathan. 2004."How Much Should We Trust Di¤erences-in-Di¤erences Estimates?",Quarterly Journal of Economics 119, 249-275
[6] Blonigen, Bruce A. 2002. "Tari¤-jumping Antidumping Duties", Journalof International Economics 57, 31-49
[7] Blonigen, Bruce A. and Jee-Hyeong Park. 2004. "Dynamic Pricing inthe Presence of Antidumping Policy: Theory and Evidence", AmericanEconomic Review 94, 134-154
[8] Blonigen, Bruce A. and Thomas Prusa. 2003. "Antidumping", in JamesHarrigan and E. Kwan Choi (eds), Handbook of International Trade,Oxford: Blackwell
[9] Bown, Chad P. (2010) "Global Antidumping Database,� The WorldBank, available at http://econ.worldbank.org/ttbd/gad/
[10] Bown, Chad P. 2011. "Taking Stock of Antidumping, Safeguards andCountervailing Duties, 1990-2009", World Economy 34, 1955-1998
[11] Bown, Chad P. and Meredith A. Crowley. 2006. "Policy Externalities:How US Antidumping A¤ects Japanese Exports to the EU", EuropeanJournal of Political Economy 22, 696-714
[12] Bown, Chad P. and Meredith A. Crowley. 2007. "Trade De�ection andTrade Depression", Journal of International Economics 72, 176-201
31
[13] Broda, Christian and David Weinstein. 2006. "Globalization and theGains from Variety", Quarterly Journal of Economics 121, 541-585
[14] Chaney, Thomas. 2008. "Distorted Gravity: The Intensive and Exten-sive Margins of International Trade", American Economic Review 98,1707-1721
[15] Egger, Peter and Douglas Nelson. 2011. "How Bad is Antidumping?Evidence from Panel Data", Review of Economics and Statistics 93,1374-1390
[16] Falvey, Rodney E. and Douglas R. Nelson. 2006. "100 Years of An-tidumping", European Journal of Political Economy 22, 545-553
[17] Gallaway, Michael P., Bruce A. Blonigen, and Joseph E. Flynn. 1999."Welfare Costs of the U.S. Antidumping and Countervailing DutyLaws", Journal of International Economics 49, 211-244
[18] Imbens, Guido W. and Je¤rey M. Wooldridge. 2009. "Recent Devel-opments in the Econometrics of Program Evaluation", Journal of Eco-nomic Literature 47, 5-86
[19] Konings, Jozef and Hylke Vandenbussche. 2008. "Heterogeneous Re-sponses of Firms to Trade Protection", Journal of International Eco-nomics 76, 371-383
[20] Melitz, Marc. 2003. "The Impact of Trade on Intra-Industry Realloca-tions and Aggregate Industry Productivity", Econometrica 71, 1695-1725
[21] Melitz, Marc and Gianmarco Ottaviano. 2008. "Market Size, Trade, andProductivity", Review of Economic Studies 75, 295-316
[22] Nizovtsev, Dmitri and Alexandre Skiba. 2010. "Import Demand Elas-ticity and Exporter Response to Anti-Dumping Duties", working paper
[23] Pierce, Justin R. 2011. "Plant-level Responses to Antidumping Duties:Evidence from U.S. Manufacturers", Journal of International Economics85, 222-233
[24] Pierce, Justin R. and Peter Schott. 2012. "The Surprisingly Swift Declineof U.S. Manufacturing Employment", working paper
[25] Prusa, Thomas. 2001. "On the Spread and Impact of Antidumping",Canadian Journal of Economics 34, 591-611
32
[26] Staiger, Robert W. and Frand A. Wolak. 1994. "Measuring Industry-Speci�c Protection: Antidumping in the United States", Brookings pa-pers on Economic Activity: Microeconomics 1994, 51-103
[27] Vandenbussche, Hylke and Maurizio Zanardi. 2010. "The chilling tradee¤ects of antidumping proliferation", European Economic Review 54,760-777
[28] Zanardi, Maurizio. 2006. "Antidumping: A Problem in InternationalTrade", European Journal of Political Economy 22, 591-617
33
Figure 1: Time trends of export volume, product level Figure 1a (Control group 1) Figure 1b (Control group 1)
Figure 1c (Control group 2) Figure 1d (Control group 2)
Note: The upper panel reports the results obtained using control group 1, whereas the lower panel reports the results obtained using control group 2. The left panel reports time trends of the treatment and control groups separately, whereas the right panel reports the time trend of the difference between the treatment and control groups. The three vertical lines mark respectively the date points of initiation of the antidumping investigation, the preliminary and final ITC determinations. The vertical axis is the logarithm of export volume.
-.50
.51
1.5
2
-10 0 10 20month
treatment group control group
-2-1
.5-1
-.50
-10 0 10 20month
-.50
.51
1.5
2
-10 0 10 20month
treatment group control group-2
-1.5
-1-.5
0-10 0 10 20
month
Figure 2: Time trends of the number of exporters Figure 2a (Control group 1) Figure 2b (Control group 1)
Figure 2c (Control group 2) Figure 2d (Control group 2)
Note: The upper panel reports the results obtained using control group 1, whereas the lower panel reports the results obtained using control group 2. The left panel reports time trends of the treatment and control groups separately, whereas the right panel reports the time trend of the difference between the treatment and control groups. The three vertical lines mark respectively the date points of initiation of the antidumping investigation, the preliminary and final ITC determinations. The vertical axis is the logarithm of the number of exporters.
0.5
1
-10 0 10 20month
treatment group control group
-.6-.4
-.20
-10 0 10 20month
0.5
1
-10 0 10 20month
treatment group control group-.6
-.4-.2
0-10 0 10 20
month
Figure 3: Time trends of export volume per exporter, all exporters Figure 3a (Control group 1) Figure 3b (Control group 1)
Figure 3c (Control group 2) Figure 3d (Control group 2)
Note: The upper panel reports the results obtained using control group 1, whereas the lower panel reports the results obtained using control group 2. The left panel reports time trends of the treatment and control groups separately, whereas the right panel reports the time trend of the difference between the treatment and control groups. . The three vertical lines mark respectively the date points of initiation of the antidumping investigation, the preliminary and final ITC determinations. The vertical axis is the logarithm of export volume per exporter.
-1-.5
0.5
1
-10 0 10 20month
treatment group control group
-1.5
-1-.5
0
-10 0 10 20month
-1-.5
0.5
1
-10 0 10 20month
treatment group control group-1
.5-1
-.50
-10 0 10 20month
Figure 4: Time trends of export volume, surviving exporters Figure 4a (Control group 1) Figure 4b (Control group 1)
Figure 4c (Control group 2) Figure 4d (Control group 2)
Note: The upper panel reports the results obtained using control group 1, whereas the lower panel reports the results obtained using control group 2. The left panel reports time trends of the treatment and control groups separately, whereas the right panel reports the time trend of the difference between the treatment and control groups. The three vertical lines mark respectively the date points of initiation of the antidumping investigation, the preliminary and final ITC determinations. The vertical axis is the logarithm of export volume.
-1-.5
0.5
1
-10 0 10 20month
treatment group control group
-1-.5
0.5
1
-10 0 10 20month
-1-.5
0.5
1
-10 0 10 20month
treatment group control group-1
-.50
.51
-10 0 10 20month
Figure 5: Time trends of export prices, product level Figure 5a (Control group 1) Figure 5b (Control group 1)
Figure 5c (Control group 2) Figure 5d (Control group 2)
Note: The upper panel reports the results obtained using control group 1, whereas the lower panel reports the results obtained using control group 2. The left panel reports time trends of the treatment and control groups separately, whereas the right panel reports the time trend of the difference between the treatment and control groups. The three vertical lines mark respectively the date points of initiation of the antidumping investigation, the preliminary and final ITC determinations. The vertical axis is the logarithm of exporter prices.
-1-.5
0.5
1
-10 0 10 20month
treatment group control group
-1-.5
0.5
1
-10 0 10 20month
-1-.5
0.5
1
-10 0 10 20month
treatment group control group
-1-.5
0.5
1
-10 0 10 20month
Figure 6: Time trends of export prices, surviving exporters Figure 6a (Control group 1) Figure 6b (Control group 1)
Figure 6c (Control group 2) Figure 6d (Control group 2)
Note: The upper panel reports the results obtained using control group 1, whereas the lower panel reports the results obtained using control group 2. The left panel reports time trends of the treatment and control groups separately, whereas the right panel reports the time trend of the difference between the treatment and control groups. . The three vertical lines mark respectively the date points of initiation of the antidumping investigation, the preliminary and final ITC determinations. The vertical axis is the logarithm of export prices.
-1-.5
0.5
1
-10 0 10 20month
treatment group control group
-1-.5
0.5
1
-10 0 10 20month
-1-.5
0.5
1
-10 0 10 20month
treatment group control group
-1-.5
0.5
1
-10 0 10 20month
Figure 7: Time trends of export volume to other countries, product level Figure 7a (Control group 1) Figure 7b (Control group 1)
Figure 7c (Control group 2) Figure 7d (Control group2)
Note: The upper panel reports the results obtained using control group 1, whereas the lower panel reports the results obtained using control group 2. The left panel reports time trends of the treatment and control groups separately, whereas the right panel reports the time trend of the difference between the treatment and control groups. . The three vertical lines mark respectively the date points of initiation of the antidumping investigation, the preliminary and final ITC determinations. The vertical axis is the logarithm of export volume.
0.5
11.
52
-10 0 10 20month
treatment group control group
-1-.5
0.5
1
-10 0 10 20month
0.5
11.
52
-10 0 10 20month
treatment group control group
-1-.5
0.5
1
-10 0 10 20month
Table 1: The effect of antidumping investigation on export volume, product level
(1) (2) (3) (4) Dependent variable Log (export volume) Control group 1 2 1 2 Initiation (β1) ‐0.108 ‐0.135 ‐0.004 ‐0.021
(0.164) (0.163) (0.158) (0.158) Preliminary ITC determination (β2) ‐0.430** ‐0.450**
(0.112) (0.112) Final ITC determination (β3) ‐1.081** ‐1.115**
(0.193) (0.195) Preliminary duties (β4) ‐0.0027** ‐0.0028**
(0.0007) (0.0007) Final duties (β5) ‐0.0060** ‐0.0061**
(0.0013) (0.0013) Month fixed effects yes yes yes yes Product fixed effects yes yes yes yes Number of observations 16,294 14,993 16,294 14,993 R‐squared 0.759 0.744 0.760 0.762
Note: Standard errors, clustered at the product level, are reported in the bracket. ** represents statistical significance at the 1% level.
Table 2: The effect of antidumping investigation, extensive versus intensive margins
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Specification Extensive margin Intensive margin Dependent Variable Log (number of exporters) Log (export volume per exporter) Sample Whole sample Whole sample Surviving firms Control Group 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.067 ‐0.076 ‐0.016 ‐0.021 ‐0.041 ‐0.060 ‐0.013 ‐0.001 0.026 0.025 ‐0.012 0.004
(0.042) (0.042) (0.037) (0.037) (0.146) (0.145) (0.144) (0.143) (0.050) (0.050) (0.040) (0.047)
Preliminary ITC determination (β2) ‐0.228** ‐0.235** ‐0.198* ‐0.212* ‐0.034 ‐0.036
(0.056) (0.057) (0.090) (0.090) (0.035) (0.035)
Final ITC determination (β3) ‐0.402** ‐0.417** ‐0.671** ‐0.693** ‐0.097 ‐0.101
(0.090) (0.091) (0.149) (0.149) (0.052) (0.051)
Preliminary duties (β4) ‐0.0012** ‐0.0012** ‐0.0015** ‐0.0016** ‐0.0008** ‐0.0008**
(0.0003) (0.0003) (0.0005) (0.0005) (0.0002) (0.0002)
Final duties (β5) ‐0.0022** ‐0.0022** ‐0.0037** ‐0.0038** ‐0.0015** ‐0.0016**
(0.0005) (0.0005) (0.0009) (0.0009) (0.0003) (0.0003)
Month fixed effects yes yes yes yes yes yes yes yes yes yes yes yes
Product fixed effects yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 16,302 14,997 16,302 14,997 16,302 14,997 16,302 14,997 547,007 538,113 547,007 538,113
R‐squared 0.932 0.936 0.932 0.936 0.659 0.665 0.660 0.664 0.227 0.226 0.227 0.227
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table 3: The effect of antidumping investigation on the likelihood of exit, trade intermediaries versus direct exporters
(1) (2) (3) (4) (5)
Dependent Variable Exit
Cutoff point Preliminary ITC determination Final ITC determination
Log (export volume) ‐0.048** ‐0.048** ‐0.012 ‐0.013 ‐0.009
(0.005) (0.005) (0.007) (0.007) (0.008)
Trade intermediaries ‐0.184** ‐0.100** ‐0.084**
(0.022) (0.031) (0.036)
Final duties ‐0.0009
(0.001)
Product fixed effects yes yes yes yes yes
Number of observations 16,580 16,580 10,446 10,446 7,290
Pseudo R2 0.028 0.032 0.027 0.028 0.029 Note: Standard errors, clustered at the product level, are reported in the bracket. ** represents statistical significance at the 1% level.
Table 4: The effect of antidumping investigation on the likelihood of exit, single‐product direct exporters versus multiple‐product direct exporters
(1) (2) (3) (4) (5)
Dependent Variable Exit
Cutoff point Preliminary ITC determination Final ITC determination
Log (export volume) ‐0.023** ‐0.059** ‐0.005* ‐0.033** ‐0.029*
(0.002) (0.006) (0.002) (0.009) (0.014)
Single‐product firms ‐0.852** 0.478** 0.585**
(0.074) (0.069) (0.086)
Final duties ‐0.0004
(0.001)
Product fixed effects yes yes yes yes yes
Number of observations 9,063 9,055 5,411 5,410 2,615
Pseudo R2 0.045 0.046 0.046 0.049 0.084 Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table 5: The effect of antidumping investigation on export prices
(1) (2) (3) (4) (5) (6) (7) (8) Specification Product Level Surviving firms Dependent Variable Log (export price) Control Group 1 2 1 2 1 2 1 2 Initiation (β1) 0.006 0.021 0.008 0.019 ‐0.017 ‐0.017 ‐0.024 ‐0.024
(0.034) (0.033) (0.029) (0.028) (0.019) (0.019) (0.016) (0.016) Preliminary ITC determination (β2) ‐0.011 0.001 0.011 0.012
(0.035) (0.036) (0.015) (0.015) Final ITC determination (β3) 0.105 0.119* 0.047 0.049
(0.055) (0.057) (0.034) (0.034) Preliminary duties (β4) 0.0002 0.0003 0.0002* 0.0002*
(0.0002) (0.0002) (0.0001) (0.0001) Final duties (β5) 0.0006** 0.0007** 0.0002 0.0002
(0.0002) (0.0003) (0.0001) (0.0001) Month fixed effects yes yes yes yes yes yes yes yes Product fixed effects yes yes yes yes yes yes yes yes Number of observations 16,294 14,993 16,294 14,993 547,007 538,113 547,007 538,113 R‐squared 0.839 0.847 0.839 0.847 0.612 0.613 0.612 0.613
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table 6: The effect of antidumping investigation on trade deflection
(1) (2) (3) (4) Dependent variable Log (export volume) Control group 1 2 1 2 Initiation (β1) ‐0.156 ‐0.191 ‐0.173 ‐0.211
(0.137) (0.140) (0.128) (0.132) Preliminary ITC determination (β2) ‐0.019 ‐0.010
(0.117) (0.116) Final ITC determination (β3) ‐0.044 ‐0.003
(0.160) (0.160) Preliminary Duties (β4) 0.0003 0.0004 (0.0007) (0.0007) Final duties (β5) ‐0.0007 ‐0.0005 (0.001) (0.001) Month fixed effects yes yes yes yes Product fixed effects yes yes yes yes Number of observations 12,484 11,561 12,484 11,561 R‐squared 0.850 0.857 0.850 0.857
Note: Standard errors, clustered at the product level, are reported in the bracket.
Table 7: The effect of antidumping investigation, quantile regression
(1) (2) (3) (4) (5) (6) (7) Specification Price response Extensive margin Intensive margin Trade deflection Final duty Import elasticity DispersionQuantile group1 0.240** 0.223 ‐0.203* 0.632 20.85 3.10 0.27 Quantile group2 ‐0.048 ‐0.303+ ‐0.138 ‐0.200 86.54 8.98 0.24 Quantile group3 0.077** ‐0.524* 0.020 ‐0.026 185.18 4.09 0.32
Note: Standard errors, clustered at the product level. +,*and ** represent statistical significance at the10%, 5% and 1% levels, respectively.
Table A.1: Logit regression on the likelihood of being investigated for dumping
(1)Dependent Variable Probability of being investigated for dumpingImport value 0.473** (0.036) Real GDP growth rate 0.031 (0.047) Exchange rate index (1989=100) 1.598** (0.602) Previously investigated 0.749** (0.205) Industry fixed effect yes Number of observations 2,243 Pseudo R2 0.230 Note: This estimation method is the logit regression; the unit of observation is at the HS6‐digit product level. Standard errors, clustered at the product level, are reported in the bracket. ** represents statistical significance at the 1% level.
Table A.2: US antidumping cases against China over 2000‐2006
CASE_ID PRODUCT Initiation date
Preliminary ITC Final ITC
Date Decision Date Decision USA‐AD‐868 Steel Wire Rope 03/2000 04/2000 A 04/2001 N USA‐AD‐874 Steel Concrete Rebar 07/2000 08/2000 A 07/2001 A USA‐AD‐885 Desktop Note Counters and Scanners 07/2000 09/2000 N . . USA‐AD‐891 Foundry Coke 09/2000 11/2000 A 09/2001 A USA‐AD‐893 Honey 10/2000 11/2000 A 11/2001 A USA‐AD‐895 Pure Magnesium 10/2000 12/2000 A 11/2001 A USA‐AD‐899 Hot‐Rolled Carbon Steel Products 11/2000 01/2001 A 11/2001 A USA‐AD‐921 Folding Gift Boxes 03/2001 04/2001 A 12/2001 A USA‐AD‐922 Automotive Replacement Glass Windshields 03/2001 04/2001 A 04/2002 A USA‐AD‐932 Folding Metal Tables and Chairs 05/2001 06/2001 A 06/2002 A USA‐AD‐935 Structural Steel Beams 06/2001 07/2001 A 06/2002 N USA‐AD‐943 Circular‐Welded Non‐Alloy Steel Pipe 06/2001 07/2001 A 07/2002 N USA‐AD‐951 Blast Furnace Coke 07/2001 08/2001 N . . USA‐AD‐968 Cold‐Rolled Steel Products 10/2001 10/2001 A 11/2002 N USA‐AD‐986 Ferrovanadium 11/2001 01/2002 A 01/2003 A USA‐AD‐989 Ball Bearings 02/2002 05/2002 A 04/2003 N USA‐AD‐990 Non‐Malleable Cast Iron Pipe Fittings 02/2002 04/2002 A 04/2003 A USA‐AD‐994 Oil Country Tubular Goods 04/2002 05/2002 N . . USA‐AD‐1010 Lawn and Garden Steel Fence Posts 05/2002 06/2002 A 06/2003 A USA‐AD‐1013 Saccharin 07/2002 08/2002 A 06/2003 A USA‐AD‐1014 Polyvinyl Alcohol 09/2002 10/2002 A 10/2003 A USA‐AD‐1020 Barium Carbonate 10/2002 11/2002 A 10/2003 A USA‐AD‐1021 Malleable Iron Pipe Fittings 11/2002 12/2002 A 12/2003 A USA‐AD‐1022 Refined Brown Aluminum Oxide 11/2002 01/2003 A 11/2003 A USA‐AD‐1030 44'‐Diamino‐22'‐Stilbenedisulfonic Acid and Stilbenic
Fluorescent Whitening Agents 04/2003 . W . .
USA‐AD‐1034 Color Television Receivers 05/2003 06/2003 A 06/2004 A USA‐AD‐1036 44'‐Diamino‐22'‐Stilbenedisulfonic Acid Chemistry 05/2003 07/2003 N . . USA‐AD‐1043 Polyethylene Retail Carrier Bags 06/2003 08/2003 A 08/2004 A USA‐AD‐1046 Tetrahydrofurfuryl Alcohol 06/2003 08/2003 A 08/2004 A USA‐AD‐1047 Ironing Tables and Certain Parts Thereof 07/2003 08/2003 A 08/2004 A USA‐AD‐1049 Electrolytic Manganese Dioxide 08/2003 . T . . USA‐AD‐1058 Wooden Bedroom Furniture 11/2003 01/2004 A 12/2004 A USA‐AD‐1059 Hand Trucks 11/2003 01/2004 A 12/2004 A USA‐AD‐1060 Carbazole Violet Pigment 23 11/2003 01/2004 A 12/2004 A USA‐AD‐1064 Certain Frozen and Canned Warmwater Shrimp and Prawns 01/2004 03/2004 A 01/2005 P USA‐AD‐1070a Crepe Paper Products 02/2004 04/2004 A 01/2005 A USA‐AD‐1070b Certain Tissue Paper Products 02/2004 04/2004 A 03/2005 A USA‐AD‐1071 Magnesium 03/2004 05/2004 A 04/2005 A USA‐AD‐1073 Certain Circular Welded Carbon Quality Line Pipe 03/2004 05/2004 A . T USA‐AD‐1082 Chlorinated Isocyanurates 05/2004 07/2004 A 06/2005 A USA‐AD‐1091 Artists' Canvas 04/2005 05/2005 A 05/2006 A USA‐AD‐1092 Diamond Sawblades and Parts Thereof 05/2005 07/2005 A 07/006 N USA‐AD‐1095 Certain Lined Paper Products 09/2005 10/2005 A 09/2006 A USA‐AD‐1099 Carbon and Certain Alloy Steel Wire Rod 11/2005 01/2006 N . . USA‐AD‐1102 Activated Carbon 02/2006 . W . . USA‐AD‐1103 Certain Activated Carbon 03/2006 05/2006 A 04/2007 A USA‐AD‐1104 Certain Polyester Staple Fiber 06/2006 08/2006 A 05/2007 A USA‐AD‐1107 Coated Free Sheet Paper 11/2006 12/2006 A 12/2007 N
Note: A: affirmative; N: negative; W: withdrawal; T: terminated.
Table A.3.1: Robustness check, differential time trends before the antidumping investigation
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2 1 2 Previous 12 months (β0)
‐0.005 ‐0.023 ‐0.008 ‐0.015 0.011 ‐0.002 0.041 0.039 0.008 0.022 0.002 0.004 ‐0.283* ‐0.241*
(0.135) (0.134) (0.047) (0.047) (0.104) (0.103) (0.039) (0.039) (0.035) (0.036) (0.018) (0.018) (0.137) (0.135)
Initiation (β1) ‐0.111 ‐0.146 ‐0.071 ‐0.084 ‐0.036 ‐0.061 0.050 0.047 0.010 0.031 ‐0.016 ‐0.014 ‐0.285 ‐0.301
(0.193) (0.192) (0.055) (0.055) (0.167) (0.165) (0.055) (0.055) (0.045) (0.046) (0.024) (0.025) (0.168) (0.171)
Preliminary ITC determination (β2)
‐0.432** ‐0.461** ‐0.232** ‐0.243** ‐0.193 ‐0.212 ‐0.009 ‐0.013 ‐0.008 0.012 0.013 0.015 ‐0.152 ‐0.123
(0.148) (0.147) (0.064) (0.064) (0.120) (0.119) (0.047) (0.047) (0.045) (0.047) (0.021) (0.022) (0.151) (0.151)
Final ITC determination (β3)
‐1.083** ‐1.126** ‐0.406** ‐0.424** ‐0.665** ‐0.694** ‐0.070 ‐0.076 0.108 0.130 0.048 0.051 ‐0.188 ‐0.126
(0.217) (0.218) (0.096) (0.097) (0.169) (0.169) (0.061) (0.061) (0.065) (0.068) (0.037) (0.038) (0.178) (0.178)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations
16,294 14,993 16,302 14,993 16,302 14,997 547,007 538,113 16,294 14,993 547,007 538,113 12,484 11,561
R‐squared 0.759 0.762 0.932 0.936 0.659 0.665 0.227 0.226 0.839 0.847 0.612 0.613 0.854 0.857
Note: Standard errors, clustered at the product level, are reported in the bracket. *and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.3.2: Robustness check, differential time trends before the antidumping investigation
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection Dependent variable Log (export volume) Log (number of
exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control group 1 2 1 2 1 2 1 2 1 2 1 2 1 2 Previous 12 months (β0)
0.154 0.153 0.068 0.068 0.093 0.090 ‐0.010 ‐0.010 0.007 0.012 ‐0.009 ‐0.009 ‐0.248* ‐0.219*
(0.111) (0.108) (0.042) (0.041) (0.085) (0.084) (0.042) (0.043) (0.027) (0.027) (0.018) (0.018) (0.109) (0.107)
Initiation (β1) 0.051 0.033 0.008 0.003 0.046 0.031 ‐0.015 ‐0.015 0.011 0.023 ‐0.027 ‐0.027 ‐0.252 ‐0.282
(0.172) (0.171) (0.044) (0.043) (0.155) (0.154) (0.042) (0.042) (0.033) (0.033) (0.019) (0.019) (0.138) (0.143)
Preliminary duties (β4)
‐0.0025** ‐0.0025** ‐0.0011** ‐0.0011** ‐0.0014* ‐0.0014* ‐0.0008** ‐0.0008** 0.0002 0.0003 0.0002 0.0002 ‐0.0001 ‐0.0001
(0.0008) (0.0008) (0.0003) (0.0003) (0.0006) (0.0006) (0.0003) (0.0003) (0.0002) (0.0002) (0.0001) (0.0001) (0.001) (0.001)
Final duties (β5) ‐0.0057** ‐0.0058** ‐0.0021** ‐0.0021** ‐0.0035** ‐0.0036** ‐0.0016** ‐0.0016** 0.0007** 0.0007** 0.0002 0.0002 ‐0.001 ‐0.001
(0.0014) (0.0014) (0.0005) (0.0005) (0.0009) (0.0009) (0.0003) (0.0003) (0.0002) (0.0003) (0.0001) (0.0001) (0.001) (0.001)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations
16,294 14,993 16,302 14,997 16,302 14,997 547,007 538,113 16,294 14,993 547,007 538,113 12,484 11,561
R‐squared 0.759 0.762 0.932 0.936 0.659 0.664 0.227 0.227 0.839 0.847 0.612 0.613 0.850 0.857
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.4.1: Robustness check, inclusion of product‐specific time trends
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export
volume) Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.129 ‐0.134 ‐0.071 ‐0.075 ‐0.063 ‐0.064 0.002 0.003 0.015 0.022 ‐0.034* ‐0.034* 0.079 0.065
(0.149) (0.148) (0.042) (0.042) (0.126) (0.126) (0.052) (0.052) (0.030) (0.030) (0.017) (0.017) (0.127) (0.126))
Preliminary ITC determination (β2)
‐0.363** ‐0.357** ‐0.215** ‐0.216** ‐0.148 ‐0.142 ‐0.067 ‐0.067 ‐0.011 ‐0.008 ‐0.011 ‐0.010 0.203* 0.190
(0.131) (0.131) (0.067) (0.066) (0.100) (0.100) (0.049) (0.049) (0.038) (0.039) (0.015) (0.015) (0.099) (0.097)
Final ITC determination (β3)
‐0.903** ‐0.888** ‐0.357** ‐0.354** ‐0.547** ‐0.535** ‐0.124 ‐0.125 0.058 0.059 0.002 0.003 0.260 0.272
(0.221) (0.220) (0.110) (0.109) (0.158) (0.158) (0.070) (0.070) (0.067) (0.067) (0.024) (0.024) (0.159) (0.158)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product time trends yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations
16,294 14,993 16,302 14,997 16,302 14,997 547,007 538,113 16,294 14,993 547,007 538,113 12,484 11,561
R‐squared 0.796 0.762 0.944 0.947 0.706 0.710 0.229 0.228 0.856 0.862 0.613 0.615 0.883 0.888
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.4.2: Robustness check, inclusion of product‐specific time trends
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection Dependent variable Log (export volume) Log (number
of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.001 ‐0.008 0.003 ‐0.001 ‐0.008 ‐0.011 ‐0.044 ‐0.044 0.034 0.040 ‐0.025 ‐0.025 0.029 0.017
(0.141) (0.140) (0.039) (0.039) (0.125) (0.124) (0.040) (0.040) (0.030) (0.028) (0.016) (0.016) (0.117) (0.117)
Preliminary duties (β4)
‐0.0018* ‐0.0018* ‐0.0009* ‐0.0008* ‐0.0010 ‐0.0010 ‐0.0011** ‐0.0011** 0.0003 0.0003 0.0001 0.0001 0.0014* 0.001
(0.0008) (0.0007) (0.0004) (0.0004) (0.0006) (0.0006) (0.0002) (0.0002) (0.0002) (0.0002) (0.0001) (0.0001) (0.0007) (0.001)
Final duties (β5) ‐0.0038** ‐0.0038** ‐0.0013* ‐0.0013* ‐0.0025** ‐0.0025** ‐0.0020** ‐0.0020** 0.0006* 0.0006* 0.0001 0.0001 0.001 0.001
(0.0014) (0.0014) (0.0006) (0.0006) (0.0009) (0.0009) (0.0004) (0.0004) (0.0003) (0.0003) (0.0001) (0.0001) (0.001) (0.001)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product time trends
yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations
16,294 14,993 16,302 14,997 16,302 14,997 547,007 568,113 16,294 14,993 547,007 538,113 12,484 11,561
R‐squared 0.796 0.798 0.944 0.947 0.706 0.710 0.229 0.229 0.856 0.862 0.613 0.615 0.883 0.888
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.5.1: Robustness check, quarterly data
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export
volume) Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) 0.113 0.086 ‐0.056 ‐0.064 0.221 0.204 0.065 0.064 ‐0.035 ‐0.022 ‐0.020 ‐0.019 ‐0.446* ‐0.463*
(0.273) (0.272) (0.043) (0.043) (0.266) (0.264) (0.055) (0.056) (0.073) (0.073) (0.021) (0.021) (0.194) (0.194)
Preliminary ITC determination (β2)
‐0.735** ‐0.751** ‐0.177** ‐0.184** ‐0.490* ‐0.499* ‐0.047 ‐0.049 0.026 0.036 0.007 0.008 ‐0.065 ‐0.035
(0.155) (0.156) (0.054) (0.054) (0.135) (0.135) (0.045) (0.045) (0.035) (0.036) (0.012) (0.013) (0.132) (0.132)
Final ITC determination (β3)
‐1.409** ‐1.442** ‐0.279** ‐0.295** ‐1.018* ‐1.035* ‐0.128* ‐0.132* 0.068 0.081 0.020 0.021 ‐0.127 ‐0.127
(0.229) (0.232) (0.076) (0.076) (0.191) (0.193) (0.056) (0.057) (0.044) (0.046) (0.031) (0.031) (0.177) (0.177)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 6,314 5,736 6,314 5,736 6,314 5,736 312,311 306,445 6,310 5,736 311,702 305,858 4,458 4,087
R‐squared 0.781 0.785 0.957 0.959 0.672 0.677 0.202 0.202 0.854 0.862 0.591 0.593 0.856 0.865
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.5.2: Robustness check, quarterly data
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection Dependent variable Log (export volume) Log (number
of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) 0.355 0.339 ‐0.009 ‐0.013 0.400 0.389 0.029 0.030 ‐0.043 ‐0.034 ‐0.019 ‐0.019 ‐0.482* ‐0.506**
(0.287) (0.286) (0.037) (0.037) (0.279) (0.278) (0.047) (0.048) (0.070) (0.069) (0.017) (0.017) (0.190) (0.191)
Preliminary duties (β4)
‐0.0028** ‐0.0027** ‐0.0007* ‐0.0007* ‐0.0017* ‐0.0017* ‐0.0005* ‐0.0005* 0.0003 0.0003 0.0001* 0.0001* ‐0.001 ‐0.0004
(0.0009) (0.0009) (0.0003) (0.0003) (0.0007) (0.0007) (0.0002) (0.0002) (0.0002) (0.0002) (0.00006) (0.00006) (0.001) (0.001)
Final duties (β5) ‐0.0070** ‐0.0070** ‐0.0015** ‐0.0015** ‐0.0049** ‐0.0049** ‐0.0013** ‐0.0013** 0.0003 0.0004 0.0001 0.0001 ‐0.002 ‐0.001
(0.0015) (0.0015) (0.0004) (0.0004) (0.0011) (0.0011) (0.0003) (0.0003) (0.0002) (0.0002) (0.0001) (0.0001) (0.001) (0.001)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations
6,314 5,736 6,314 5,736 6,314 5,736 312,311 306,445 6,310 5,736 311,702 305,858 4,458 4,087
R‐squared 0.780 0.784 0.957 0.959 0.670 0.675 0.202 0.202 0.854 0.862 0.591 0.593 0.857 0.865
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.6.1: Robustness check, exclusion of outliers
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export
volume) Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.071 ‐0.093 ‐0.060 ‐0.070 ‐0.012 ‐0.024 0.020 0.020 0.011 0.026 ‐0.014 ‐0.014 ‐0.149 ‐0.179
(0.169) (0.168) (0.043) (0.043) (0.151) (0.150) (0.043) (0.043) (0.033) (0.033) (0.018) (0.019) (0.138) (0.140)
Preliminary ITC determination (β2)
‐0.438** ‐0.459** ‐0.230** ‐0.237** ‐0.205* ‐0.218* ‐0.035 ‐0.036 ‐0.005 0.008 0.014 0.015 ‐0.031 ‐0.019
(0.112) (0.112) (0.057) (0.057) (0.089) (0.088) (0.031) (0.031) (0.034) (0.035) (0.013) (0.013) (0.115) (0.114)
Final ITC determination (β3)
‐1.074** ‐1.111** ‐0.397** ‐0.412** ‐0.671** ‐0.693** ‐0.101* ‐0.104* 0.106 0.123* 0.043 0.044 ‐0.011 0.000
(0.192) (0.193) (0.091) (0.091) (0.146) (0.147) (0.047) (0.047) (0.054) (0.057) (0.031) (0.032) (0.159) (0.160)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations
15,827 14,595 15,834 14,599 15,834 14,599 531,400 522,784 15,827 14,595 531,400 522,784 12,173 11,290
R‐squared 0.774 0.776 0.934 0.938 0.676 0.681 0.257 0.256 0.842 0.85 0.623 0.625 0.855 0.962
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.6.2: Robustness check, exclusion of outliers
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection Dependent variable Log (export volume) Log (number
of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) 0.272 0.258 0.095* 0.087* 0.171 0.165 ‐0.005 ‐0.004 0.010 0.018 ‐0.022 ‐0.022 ‐0.162 ‐0.197
(0.164) (0.163) (0.040) (0.040) (0.153) (0.152) (0.035) (0.036) (0.032) (0.032) (0.015) (0.016) (0.128) (0.131)
Preliminary duties (β4)
‐0.0012* ‐0.0013* ‐0.0005* ‐0.0005* ‐0.0008 ‐0.0008 ‐0.0006** ‐0.0006** 0.0003 0.0003 0.0002* 0.0002* 0.000 0.000
(0.0006) (0.0006) (0.0002) (0.0002) (0.0005) (0.0005) (0.0002) (0.0002) (0.0002) (0.0002) (0.0001) (0.0001) (0.001) (0.001)
Final duties (β5) ‐0.0019** ‐0.0020** ‐0.0003 0.0004 ‐0.0017** ‐0.0017** ‐0.0013** ‐0.0014** 0.0007* 0.0008* 0.0002 0.0002 ‐0.001 ‐0.000
(0.0006) (0.0007) (0.0002) (0.0002) (0.0005) (0.0005) (0.0003) (0.0003) (0.0003) (0.0003) (0.0001) (0.0001) (0.001) (0.001)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations
15,827 14,595 15,834 14,599 15,834 14,599 531,400 522,784 15,827 14,595 531,400 522,784 12,173 11,290
R‐squared 0.771 0.773 0.933 0.936 0.675 0.679 0.257 0.256 0.843 0.850 0.623 0.625 0.855 0.862
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.7.1: Robustness check, inclusion of unsuccessful and withdrawn cases
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export
volume) Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.155 ‐0.179 ‐0.037 ‐0.045 ‐0.082 ‐0.097 0.010 0.011 ‐0.021 ‐0.022 ‐0.021 ‐0.022 ‐0.231* ‐0.225*
(0.115) (0.114) (0.026) (0.026) (0.100) (0.100) (0.044) (0.044) (0.044) (0.047) (0.015) (0.015) (0.093) (0.094)
Preliminary ITC determination (β2)
‐0.373** ‐0.411** ‐0.132** ‐0.142** ‐0.205** ‐0.218** 0.022 0.022 0.002 0.003 ‐0.005 ‐0.006 ‐0.085 ‐0.055
(0.098) (0.100) (0.036) (0.037) (0.077) (0.077) (0.032) (0.032) (0.035) (0.034) (0.017) (0.017) (0.081) (0.080)
Final ITC determination (β3)
‐0.755** ‐0.798** ‐0.213** ‐0.238** ‐0.447** ‐0.473** 0.007 0.004 0.049 0.060 0.004 0.004 ‐0.023 ‐0.020
(0.151) (0.151) (0.058) (0.058) (0.116) (0.117) (0.056) (0.057) (0.046) (0.048) (0.034) (0.034) (0.126) (0.127)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations
22,821 20,373 22,823 20,373 22,823 20,373 909,293 884,678 22,813 20,369 906,737 882,169 19,491 17,317
R‐squared 0.781 0.769 0.950 0.953 0.660 0.670 0.205 0.205 0.843 0.846 0.531 0.532 0.855 0.858
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.7.2: Robustness check, inclusion of unsuccessful and withdrawn cases
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection Dependent variable Log (export volume) Log (number
of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control group 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.003 ‐0.010 0.008 0.009 0.005 ‐0.002 ‐0.008 ‐0.006 ‐0.025 ‐0.021 ‐0.031* ‐0.033* ‐0.224* ‐0.220*
(0.110) (0.110) (0.024) (0.024) (0.098) (0.097) (0.044) (0.044) (0.044) (0.044) (0.013) (0.015) (0.090) (0.091)
Preliminary duties (β4)
‐0.0027** ‐0.0027** ‐0.0008** ‐0.0008** ‐0.0016** ‐0.0016** ‐0.0002 ‐0.0002 0.0004* 0.0004* ‐0.0001 ‐0.0001 0.000 0.00
(0.0007) (0.0007) (0.0002) (0.0002) (0.0005) (0.0005) (0.0002) (0.0002) (0.0002) (0.0002) (0.0001) (0.0001) (0.001) (0.001)
Final duties (β5) ‐0.0055** ‐0.0055** ‐0.0016** ‐0.0016** ‐0.0033** ‐0.0033** ‐0.0002 ‐0.0002 0.0006** 0.0006** ‐0.0002 ‐0.0002 ‐0.000 ‐0.000
(0.0012) (0.0012) (0.0004) (0.0004) (0.0009) (0.0009) (0.0004) (0.0004) (0.0002) (0.0002) (0.0002) (0.0002) (0.001) (0.001)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations
22,823 20,373 22,823 20,373 22,823 20,373 909,293 884,678 22,815 20,369 906,737 882,169 19,491 17,317
R‐squared 0.765 0.769 0.950 0.953 0.66 0.670 0.205 0.205 0.838 0.846 0.531 0.532 0.855 0.858
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.8.1: Robustness check, exclusion of antidumping cases investigated by other countries
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export
volume) Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.227 ‐0.263 ‐0.048 ‐0.059 ‐0.180 ‐0.204 0.025 0.024 0.034 0.046 ‐0.017 ‐0.016 ‐0.154 ‐0.185
(0.151) (0.150) (0.047) (0.047) (0.126) (0.125) (0.050) (0.050) (0.035) (0.035) (0.019) (0.020) (0.123) (0.128)
Preliminary ITC determination (β2)
‐0.308** ‐0.325** ‐0.203** ‐0.209** ‐0.103 ‐0.114 ‐0.031 ‐0.033 ‐0.001 0.007 0.011 0.012 ‐0.014 ‐0.006
(0.106) (0.107) (0.059) (0.059) (0.086) (0.086) (0.035) (0.035) (0.035) (0.036) (0.015) (0.015) (0.117) (0.117)
Final ITC determination (β3)
‐0.826** ‐0.863** ‐0.363** ‐0.376** ‐0.457** ‐0.482** ‐0.092 ‐0.096 0.079 0.093 0.046 0.047 ‐0.078 ‐0.034
(0.177) (0.179) (0.095) (0.095) (0.136) (0.138) (0.052) (0.052) (0.056) (0.059) (0.034) (0.035) (0.160) (0.161)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations
14,425 13,277 14,431 13,280 14,431 13,280 543,567 534,818 14,425 13,277 543,567 534,818 11,608 10,717
R‐squared 0.781 0.785 0.933 0.937 0.675 0.680 0.223 0.223 0.848 0.856 0.611 0.613 0.856 0.864
Note: Standard errors, clustered at the product level, are reported in the bracket. ** represent statistical significance at the 1% level, respectively.
Table A.8.2: Robustness check, exclusion of antidumping cases investigated by other countries
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export
volume) Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.167 ‐0.193 ‐0.003 ‐0.009 ‐0.164 ‐0.184 ‐0.014 ‐0.014 0.041 0.051 ‐0.024 ‐0.024 ‐0.165 ‐0.200
(0.142) (0.140) (0.042) (0.041) (0.121) (0.120) (0.040) (0.040) (0.031) (0.031) (0.016) (0.016) (0.113) (0.119)
Preliminary duties (β4) ‐0.0021** ‐0.0021** ‐0.0010** ‐0.0010** ‐0.0011* ‐0.0010* ‐0.0008** ‐0.0008** 0.0002 0.0003 0.0002* 0.0002* 0.000 0.000
(0.0007) (0.0007) (0.0003) (0.0003) (0.0005) (0.0005) (0.0002) (0.0002) (0.0002) (0.0002) (0.0001) (0.0001) (0.001) (0.001)
Final duties (β5) ‐0.0046** ‐0.0047** ‐0.0019** ‐0.0019** ‐0.0027** ‐0.0028** ‐0.0015** ‐0.0015** 0.0006* 0.0006* 0.0002 0.0002 ‐0.001 ‐0.001
(0.0011) (0.0011) (0.0005) (0.0005) (0.0007) (0.0007) (0.0003) (0.0003) (0.0003) (0.0003) (0.0001) (0.0001) (0.001) (0.001)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 14,425 13,277 14,431 13,280 14,431 13,280 543,567 534,818 14,425 13,277 543,567 534,818 11,608 10,717
R‐squared 0.781 0.785 0.935 0.937 0.675 0.680 0.224 0.223 0.848 0.856 0.611 0.613 0.856 0.864
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.9.1: Robustness check, exclusion of processing trade
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.036 ‐0.050 ‐0.011 ‐0.019 0.006 0.002 0.021 0.022 ‐0.016 ‐0.007 ‐0.016 ‐0.016 ‐0.080 ‐0.099
(0.186) (0.186) (0.038) (0.038) (0.164) (0.163) (0.048) (0.049) (0.028) (0.028) (0.019) (0.019) (0.204) (0.206)
Preliminary ITC determination (β2) ‐0.407** ‐0.418** ‐0.183** ‐0.192** ‐0.153 ‐0.155 ‐0.055 ‐0.053 ‐0.015 ‐0.009 0.025* 0.026* 0.047 0.071
(0.141) (0.140) (0.054) (0.054) (0.121) (0.120) (0.033) (0.033) (0.030) (0.030) (0.012) (0.012) (0.157) (0.157)
Final ITC determination (β3) ‐0.835** ‐0.854** ‐0.227** ‐0.244** ‐0.498** ‐0.501** ‐0.091* ‐0.088* 0.033 0.038 0.014 0.013 ‐0.221 ‐0.152
(0.185) (0.185) (0.080) (0.080) (0.142) (0.141) (0.041) (0.041) (0.042) (0.044) (0.030) (0.023) (0.222) (0.226)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 14,818 13720 14,818 13,720 14,818 13,720 359,944 354,771 14,811 13,720 359,611 354,441 11,101 10,476
R‐squared 0.711 0.778 0.953 0.856 0.684 0.689 0.281 0.283 0.795 0.803 0.612 0.615 0.833 0.841
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.9.2: Robustness check, exclusion of processing trade
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) 0.046 0.037 0.023 0.019 0.040 0.036 0.010 0.010 ‐0.010 ‐0.002 ‐0.021 ‐0.021 ‐0.084 ‐0.111
(0.174) (0.174) (0.032) (0.032) (0.155) (0.155) (0.042) (0.042) (0.022) (0.022) (0.015) (0.015) (0.182) (0.186)
Preliminary duties (β4) ‐0.003** ‐0.003** ‐0.001** ‐0.001** ‐0.002* ‐0.002* ‐0.001* ‐0.000* 0.000 0.000 0.000* 0.000* ‐0.000 ‐0.000
(0.001) (0.001) (0.000) (0.000) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) (0.001)
Final duties (β5) ‐0.005** ‐0.005** ‐0.001** ‐0.001** ‐0.003** ‐0.003** ‐0.001** ‐0.001** 0.000 0.000 0.000 0.000 ‐0.001 ‐0.001
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) (0.001)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 14,818 13,720 14,818 13,720 14,818 13,720 359,944 354,771 14,811 13,714 359,611 354,441 11,101 10,476
R‐squared 0.777 0.778 0.953 0.956 0.684 0.689 0.281 0.281 0.795 0.803 0.612 0.613 0.833 0.841
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.10.1: Robustness check, exclusion of foreign firms
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.194 ‐0.189 ‐0.027 ‐0.031 ‐0.123 ‐0.115 0.023 0.022 ‐0.023 ‐0.019 ‐0.001 0.001 ‐0.270 ‐0.322
(0.182) (0.180) (0.041) (0.041) (0.159) (0.157) (0.064) (0.064) (0.030) (0.030) (0.023) (0.024) (0.147) (0.151)
Preliminary ITC determination (β2) ‐0.677** ‐0.664** ‐0.180** ‐0.182** ‐0.422** ‐0.408** ‐0.035 ‐0.037 0.003 0.004 0.020 0.022 ‐0.122 ‐0.140
(0.172) (0.170) (0.059) (0.060) (0.132) (0.131) (0.041) (0.040) (0.038) (0.038) (0.017) (0.017) (0.123) (0.122)
Final ITC determination (β3) ‐1.076** ‐1.063** ‐0.308** ‐0.313** ‐0.651** ‐0.635** ‐0.105* ‐0.109* 0.028 0.029 0.061 0.063 ‐0.021 ‐0.021
(0.219) (0.217) (0.090) (0.091) (0.155) (0.153) (0.052) (0.051) (0.038) (0.039) (0.036) (0.036) (0.181) (0.182)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 14,996 13,812 14,996 13,812 14,996 13,812 319,881 313,922 14,990 13,810 319,881 316,922 11,197 10,483
R‐squared 0.747 0.749 0.944 0.947 0.673 0.679 0.247 0.246 0.796 0.796 0.563 0.563 0.863 0.871
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.10.2: Robustness check, exclusion of foreign firms
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.023 ‐0.019 0.017 0.015 ‐0.013 ‐0.009 0.014 0.015 ‐0.025 ‐0.020 ‐0.017 ‐0.017 ‐0.262* ‐0.304*
(0.174) (0.173) (0.033) (0.033) (0.154) (0.153) (0.052) (0.052) (0.029) (0.029) (0.019) (0.019) (0.136) (0.141)
Preliminary duties (β4) ‐0.003** ‐0.003** ‐0.001** ‐0.001** ‐0.002* ‐0.002* ‐0.000 ‐0.000 ‐0.000 ‐0.000 0.000 0.000 ‐0.001 ‐0.001
(0.001) (0.001) (0.000) (0.000) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) (0.001)
Final duties (β5) ‐0.006** ‐0.006** ‐0.002** ‐0.002** ‐0.003** ‐0.003** ‐0.001** ‐0.001** 0.000 0.000 0.000 0.000 ‐0.000 ‐0.000
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) (0.001)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 14,996 13,812 14,996 13,812 14,996 13,812 319,881 313,922 14,990 13,810 319,881 313,922 11,197 10,483
R‐squared 0.746 0.748 0.944 0.947 0.672 0.678 0.247 0.246 0.796 0.796 0.563 0.563 0.863 0.871
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.11.1: Robustness check, inclusion of interaction with number of products in the same HS6‐digit
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.248 ‐0.280 ‐0.126* ‐0.135* ‐0.124 ‐0.147 0.046 0.045 0.004 0.019 ‐0.086** ‐0.087** ‐0.086 ‐0.171
(0.227) (0.226) (0.060) (0.060) (0.200) (0.198) (0.074) (0.074) (0.045) (0.044) (0.022) (0.022) (0.231) (0.235) Preliminary ITC determination (β2) ‐0.480** ‐0.506** ‐0.281** ‐0.288** ‐0.196 ‐0.216 ‐0.079 ‐0.081 ‐0.037 ‐0.023 ‐0.014 ‐0.015 0.125 0.103
(0.157) (0.156) (0.067) (0.067) (0.120) (0.120) (0.049) (0.049) (0.044) (0.045) (0.019) (0.019) (0.198) (0.197) Final ITC determination (β3) ‐1.171** ‐1.212** ‐0.498** ‐0.514** ‐0.667** ‐0.695** ‐0.135 ‐0.139 0.083 0.100 ‐0.013 ‐0.015 0.660* 0.674*
(0.274) (0.275) (0.109) (0.110) (0.204) (0.205) (0.075) (0.075) (0.071) (0.073) (0.045) (0.046) (0.274) (0.281)
Initiation (β1)*hs 0.083 0.085 0.032 0.032 0.052 0.054 ‐0.013 ‐0.014 0.000 0.000 0.044** 0.045** ‐0.027 0.006
(0.068) (0.068) (0.021) (0.021) (0.054) (0.054) (0.031) (0.031) (0.015) (0.015) (0.012) (0.013) (0.074) (0.073) Preliminary ITC determination (β2)*hs 0.026 0.029 0.027 0.027 ‐0.001 0.002 0.028 0.028 0.014 0.013 0.016 0.017 ‐0.081 ‐0.061
(0.057) (0.057) (0.027) (0.027) (0.039) (0.038) (0.021) (0.021) (0.013) (0.013) (0.009) (0.009) (0.067) (0.061) Final ITC determination (β3)*hs 0.047 0.051 0.050 0.050 ‐0.001 0.002 0.024 0.024 0.011 0.010 0.038 0.040 ‐0.426** ‐0.408**
(0.098) (0.098) (0.044) (0.044) (0.066) (0.066) (0.034) (0.034) (0.022) (0.022) (0.022) (0.023) (0.109) (0.113)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes Number of observations 16,253 14,968 16,261 14,972 16,261 14,972 546,959 538,091 16,253 14,968 546,959 538,091 12,426 11,549
R‐squared 0.759 0.762 0.932 0.936 0.659 0.665 0.227 0.226 0.839 0.847 0.612 0.613 0.852 0.858
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively. ‘hs’ in the interaction terms means the number of products in the same HS6‐digit.
Table A.11.2: Robustness check, inclusion of interaction with number of products in the same HS6‐digit
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.120 ‐0.136 ‐0.053 ‐0.056 ‐0.067 ‐0.081 0.034 0.035 0.004 0.013 ‐0.078** ‐0.080** ‐0.177 ‐0.257
(0.217) (0.216) (0.049) (0.049) (0.197) (0.196) (0.056) (0.056) (0.038) (0.038) (0.021) (0.021) (0.219) (0.224)
Preliminary duties (β4) ‐0.002 ‐0.002 ‐0.001* ‐0.001* ‐0.001 ‐0.001 ‐0.001* ‐0.001* ‐0.000 ‐0.000 0.000 0.000 0.002 0.002
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) (0.001)
Final duties (β5) ‐0.006** ‐0.006** ‐0.003** ‐0.003** ‐0.004* ‐0.004* ‐0.001** ‐0.001** 0.000 0.000 ‐0.000 ‐0.000 0.003 0.003
(0.002) (0.002) (0.001) (0.001) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.002) (0.002)
Initiation (β1)*hs 0.071 0.070 0.022 0.021 0.050 0.050 ‐0.030 ‐0.031 0.002 0.003 0.035** 0.036** 0.003 0.031
(0.062) (0.061) (0.019) (0.019) (0.051) (0.051) (0.021) (0.021) (0.012) (0.012) (0.012) (0.012) (0.067) (0.066) Preliminary duties (β4)*hs ‐0.0004 ‐0.0004 0.0000 0.0000 ‐0.0004 ‐0.0004 ‐0.0002 ‐0.0002 0.0002 0.0002 0.0001 0.0001 ‐0.001* ‐0.001*
(0.0005) (0.0005) (0.0003) (0.0003) (0.0004) (0.0004) (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) (0.0001) (0.000) (0.000)
Final duties (β5)*hs 0.0001 0.0000 0.0002 0.0002 ‐0.0001 ‐0.0002 ‐0.0003 ‐0.0003 0.0002 0.0002 0.0003 0.0003 ‐0.003** ‐0.002**
(0.0008) (0.0008) (0.0004) (0.0004) (0.0006) (0.0006) (0.0002) (0.0002) (0.0002) (0.0002) (0.0002) (0.0002) (0.001) (0.001)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 16,253 14,968 16,261 14,972 16,261 14,972 546,959 538,091 16,253 14,968 546,959 538,091 12,426 11,549
R‐squared 0.759 0.762 0.932 0.936 0.659 0.664 0.227 0.227 0.839 0.847 0.612 0.613 0.852 0.857
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively. ‘hs’ in the interaction terms means the number of products in the same HS6‐digit.
Table A.12.1: Robustness check, safeguard
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Specification Quantity response Extensive margin Intensive margin Price response
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms
Control Group 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.108 ‐0.135 ‐0.067 ‐0.076 ‐0.041 ‐0.060 0.026 0.024 0.006 0.021 ‐0.017 ‐0.017
(0.164) (0.163) (0.042) (0.042) (0.146) (0.145) (0.050) (0.050) (0.034) (0.033) (0.019) (0.019)
Preliminary ITC determination (β2) ‐0.430** ‐0.450** ‐0.228** ‐0.235** ‐0.198* ‐0.212* ‐0.034 ‐0.036 ‐0.011 0.001 0.011 0.012
(0.112) (0.112) (0.056) (0.057) (0.090) (0.096) (0.035) (0.035) (0.035) (0.036) (0.015) (0.015)
Final ITC determination (β3) ‐1.082** ‐1.117** ‐0.403** ‐0.407** ‐0.672** ‐0.694** ‐0.098 ‐0.102* 0.104 0.119* 0.047 0.048
(0.193) (0.195) (0.090) (0.091) (0.149) (0.149) (0.051) (0.051) (0.055) (0.057) (0.034) (0.034)
Safeguard 0.248 0.260 0.058 0.051 0.188 0.210 0.170** 0.170** 0.010 0.007 0.035 0.034
(0.160) (0.172) (0.034) (0.036) (0.172) (0.187) (0.045) (0.045) (0.069) (0.071) (0.030) (0.030)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 16,294 14,993 16,302 14,997 16,302 14,997 547,007 538,113 16,294 14,993 547,007 538,113
R‐squared 0.759 0.762 0.932 0.936 0.659 0.665 0.227 0.226 0.839 0.847 0.612 0.613 Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.12.2: Robustness check, safeguard
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Specification Quantity response Extensive margin Intensive margin Price response
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms
Control Group 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.004 ‐0.021 ‐0.016 ‐0.021 0.013 0.0004 ‐0.012 ‐0.012 0.008 0.019 ‐0.024 ‐0.024
(0.158) (0.158) (0.037) (0.037) (0.144) (0.143) (0.040) (0.040) (0.029) (0.028) (0.016) (0.016)
Preliminary duties(β4) ‐0.0027** ‐0.0028** ‐0.0012** ‐0.0012** ‐0.0015** ‐0.0016** ‐0.0008** ‐0.0008** 0.0002 0.0003 0.0002* 0.0002*
(0.0007) (0.0007) (0.0003) (0.0003) (0.0005) (0.0005) (0.0002) (0.0002) (0.0002) (0.0002) (0.0001) (0.0001)
Final duties(β5) ‐0.0060** ‐0.0060** ‐0.0022** ‐0.0022** ‐0.0037** ‐0.0038** ‐0.0015** ‐0.0016** 0.0006** 0.0007** 0.0002 0.0002
(0.0013) (0.0013) (0.0005) (0.0005) (0.0009) (0.0009) (0.0003) (0.0003) (0.0002) (0.0003) (0.0001) (0.0001)
Safeguard 0.103 0.111 0.009 0.000 0.093 0.112 0.171** 0.171** 0.027 0.026 0.038 0.038
(0.089) (0.095) (0.052) (0.056) (0.122) (0.137) (0.044) (0.044) (0.060) (0.062) (0.030) (0.030)
Month dummy yes yes yes yes yes yes yes yes yes 0.0002 yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 16,294 14,993 16,302 14,997 16,302 14,997 547,007 538,113 16,294 14,993 547,007 538,113
R‐squared 0.759 0.762 0.932 0.936 0.659 0.664 0.227 0.227 0.839 0.847 0.612 0.613 Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.13.1: Robustness check, inclusion of China's import tariff rates
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export
volume) Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.074 ‐0.094 ‐0.073 ‐0.080 0.003 ‐0.012 0.044 0.040 ‐0.005 0.008 ‐0.020 ‐0.019 ‐0.133 ‐0.128
(0.179) (0.178) (0.048) (0.048) (0.156) (0.155) (0.052) (0.053) (0.037) (0.037) (0.019) (0.020) (0.148) (0.147)
Preliminary ITC determination (β2) ‐0.461** ‐0.485** ‐0.251** ‐0.255** ‐0.203 ‐0.223* ‐0.008 ‐0.013 ‐0.050 ‐0.037 0.009 0.011 0.001 0.012
(0.139) (0.139) (0.068) (0.068) (0.110) (0.110) (0.031) (0.032) (0.039) (0.040) (0.014) (0.014) (0.146) (0.144)
Final ITC determination (β3) ‐1.160** ‐1.203** ‐0.432** ‐0.444** ‐0.716** ‐0.751** ‐0.053 ‐0.059 0.060 0.075 0.042 0.044 ‐0.012 0.031
(0.218) (0.219) (0.097) (0.097) (0.171) (0.171) (0.051) (0.051) (0.056) (0.059) (0.031) (0.031) (0.172) (0.172)
Import tariff rates 0.266 0.042 0.004 0.006 0.021 0.035 0.006 0.009 0.002 ‐0.001 ‐0.005 ‐0.006 0.048* 0.042*
(0.024) (0.024) (0.008) (0.009) (0.019) (0.019) (0.006) (0.006) (0.006) (0.006) (0.003) (0.003) (0.020) (0.021)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 14,151 12,996 14,158 12,999 14,158 12,999 498,968 490,818 14,151 12,996 498,968 490,818 10,846 10,018
R‐squared 0.761 0.764 0.935 0.938 0.654 0.658 0.222 0.221 0.843 0.850 0.612 0.614 0.852 0.858
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.13.2: Robustness check, inclusion of China's import tariff rates
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export
volume) Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) 0.093 0.088 0.017 0.0175 0.078 0.074 ‐0.018 ‐0.019 0.018 0.025 ‐0.026 ‐0.026 ‐0.147 ‐0.150
(0.167) (0.165) (0.042) (0.041) (0.149) (0.148) (0.042) (0.042) (0.032) (0.032) (0.017) (0.017) (0.126) (0.125)
Preliminary duties (β4) ‐0.0030** ‐0.0030** ‐0.0012** ‐0.0012** ‐0.0018** ‐0.0018** ‐0.0008** ‐0.0008** 0.0002 0.0002 0.0002* 0.0002* 0.001 0.001
(0.0008) (0.0008) (0.0003) (0.0003) (0.0006) (0.0006) (0.0002) (0.0002) (0.0002) (0.0002) (0.0001) (0.0001) (0.001) (0.001)
Final duties (β5) ‐0.0066** ‐0.0067** ‐0.0022** ‐0.0022** ‐0.0043** ‐0.0043** ‐0.0016** ‐0.0016** 0.0007** 0.0007** 0.0002 0.0002 ‐0.001 ‐0.000
(0.0014) (0.0014) (0.0005) (0.0005) (0.001) (0.001) (0.0003) (0.0004) (0.0002) (0.0002) (0.0001) (0.0001) (0.001) (0.001)
Import tariff rates 0.024 0.039 0.004 0.005 0.019 0.032 0.007 0.011 0.003 ‐0.001 ‐0.006 ‐0.007 0.047* 0.042*
(0.023) (0.023) (0.008) (0.008) (0.018) (0.018) (0.007) (0.006) (0.006) (0.006) (0.004) (0.004) (0.020) (0.021)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 14,151 12,996 14,158 12,999 14,158 12,999 498,968 490,818 14,151 12,996 498,968 490,818 10,846 10018
R‐squared 0.761 0.764 0.935 0.938 0.654 0.658 0.223 0.246 0.843 0.850 0.612 0.614 0.852 0.858
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A.14.1: Robustness check, inclusion of interaction with the import demand elasticity
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection
Dependent Variable Log (export volume) Log (number of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample
Control Group 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.399 ‐0.427 ‐0.107* ‐0.117* ‐0.295 ‐0.314 0.040 0.037 0.064 0.079 ‐0.020 ‐0.019 ‐0.404** ‐0.435**
(0.204) (0.202) (0.050) (0.051) (0.181) (0.178) (0.063) (0.063) (0.047) (0.047) (0.023) (0.023) (0.144) (0.147)
Preliminary ITC determination (β2) ‐0.576** ‐0.598** ‐0.264** ‐0.274** ‐0.313* ‐0.327** ‐0.052 ‐0.057 0.080 0.091 0.019 0.021 ‐0.131 ‐0.119
(0.155) (0.154) (0.071) (0.071) (0.126) (0.123) (0.043) (0.043) (0.046) (0.047) (0.017) (0.017) (0.109) (0.116)
Final ITC determination (β3) ‐1.102** ‐1.164** ‐0.455** ‐0.476** ‐0.631** ‐0.674** ‐0.097 ‐0.105 0.109 0.138 0.083* 0.086* ‐0.377* ‐0.334
(0.250) (0.252) (0.111) (0.111) (0.193) (0.194) (0.070) (0.070) (0.068) (0.070) (0.037) (0.037) (0.174) (0.175)
Initiation (β1)*elas 0.054 0.054 0.004 0.005 0.050 0.050 ‐0.005 ‐0.005 ‐0.013 ‐0.013 0.004** 0.004** 0.023 0.022
(0.033) (0.032) (0.006) (0.006) (0.030) (0.029) (0.004) (0.004) (0.008) (0.008) (0.001) (0.001) (0.013) (0.013)
Preliminary ITC determination (β2)*elas 0.026 0.026 0.007 0.008 0.020 0.019 0.005* 0.005* ‐0.020* ‐0.020* 0.000 0.000 0.015 0.014
(0.023) (0.023) (0.006) (0.006) (0.020) (0.019) (0.002) (0.002) (0.008) (0.008) (0.001) (0.001) (0.007) (0.007)
Final ITC determination (β3)*elas 0.005 0.010 0.011 0.011 ‐0.007 ‐0.004 0.009 0.010* 0.001 0.002 ‐0.003 ‐0.003 0.053 0.051
(0.026) (0.027) (0.008) (0.008) (0.021) (0.021) (0.005) (0.005) (0.007) (0.007) (0.003) (0.003) (0.014) (0.014)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 14,256 13,159 14,264 13,163 14,264 13,163 433,413 426,250 14,256 13,159 433,250 426,974 10,577 9,777
R‐squared 0.758 0.762 0.927 0.933 0.665 0.670 0.209 0.207 0.840 0.846 0.505 0.620 0.860 0.864
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively. ‘elas’ in the interaction terms means the import demand elasticity.
Table A.14.2: Robustness check, inclusion of interaction with the import demand elasticity
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Specification Quantity response Extensive margin Intensive margin Price response Trade deflection Dependent variable Log (export volume) Log (number
of exporters) Log (export volume per exporter) Log(export price) Log (export volume)
Sample Whole sample Whole sample Whole sample Surviving firms Whole sample Surviving firms Whole sample Control group 1 2 1 2 1 2 1 2 1 2 1 2 1 2
Initiation (β1) ‐0.025 ‐0.260 ‐0.048 ‐0.052 ‐0.197 ‐0.209 0.003 0.002 0.020 0.030 ‐0.040* ‐0.040* ‐0.403** ‐0.438**
(0.172) (0.170) (0.041) (0.041) (0.157) (0.155) (0.048) (0.048) (0.034) (0.034) (0.017) (0.018) (0.136) (0.152) Preliminary duties (β4) ‐0.002 ‐0.002 ‐0.001** ‐0.001** ‐0.001 ‐0.001 ‐0.001** ‐0.001** 0.000 0.000 0.000* 0.000* ‐0.000 ‐0.000
(0.001) (0.001) (0.000) (0.000) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) (0.001)
Final duties (β5) ‐0.005** ‐0.006** ‐0.002** ‐0.002** ‐0.003* ‐0.003* ‐0.002** ‐0.002** 0.000 0.000 0.000 0.000 ‐0.003** ‐0.003**
(0.002) (0.002) (0.001) (0.001) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.001) (0.001) Initiation (β1)*elas 0.041* 0.041* 0.002 0.002 0.039* 0.039* ‐0.005 ‐0.005 ‐0.002 ‐0.002 0.005** 0.005** 0.028 0.027
(0.019) (0.019) (0.003) (0.003) (0.018) (0.017) (0.003) (0.003) (0.004) (0.004) (0.001) (0.001) (0.015) (0.014) Preliminary duties (β4)*elas ‐0.000 ‐0.000 ‐0.000 ‐0.000 ‐0.000 ‐0.000 0.000* 0.000* ‐0.000 ‐0.000 0.000 0.000 0.000 0.000
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Final duties(β5)*elas ‐0.000 ‐0.000 0.000 0.000 ‐0.000 ‐0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.001** 0.001**
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Month dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Product dummy yes yes yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 14,256 13,159 14,264 13,163 14,264 13,163 433,413 426,250 14,256 13,159 433,413 426,250 10,577 9,777
R‐squared 0.758 0.761 0.927 0.931 0.665 0.670 0.209 0.207 0.840 0.846 0.505 0.505 0.861 0.865
Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively. ‘elas’ in the interaction terms means the import demand elasticity.
Table A.15: The effect of antidumping investigation on the likelihood of exit, single‐product direct exporters to
the U.S. and worldwide versus multiple‐product direct exporters
(1) (2) (3) (4) (5)
Dependent Variable Exit
Cutoff point Preliminary ITC determination Final ITC determination
Log (export volume) ‐0.063** ‐0.056** ‐0.015 ‐0.029** ‐0.035*
(0.006) (0.006) (0.010) (0.010) (0.014)
Single‐product firms ‐1.090** 0.715** 0.920**
(0.127) (0.101) (0.123)
Final duties ‐0.0005
(0.001)
Product fixed effects yes yes yes yes yes
Number of observations 8,513 8,505 4,669 4,668 2,350
Pseudo R2 0.038 0.043 0.044 0.045 0.079 Note: Standard errors, clustered at the product level, are reported in the bracket. * and ** represent statistical significance at the 5% and 1% levels, respectively.
Table A. 16: The effect of antidumping investigation, quantile regression (detail)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)Specification Price response Extensive margin Intensive margin Trade deflectionDependent Variable Log(export price) Log(number of exporters) Log(export volume per exporter) Log(export volume)Sample Surviving firms Whole sample Surviving firms Whole sampleQuantile group 1 2 3 1 2 3 1 2 3 1 2 3Initiation(β1) ‐0.024 ‐0.026 ‐0.025 0.177 ‐0.129+ ‐0.078 0.010 0.097 0.174 0.674 ‐0.240 ‐0.139
(0.031) (0.035) (0.026) (0.148) (0.076) (0.054) (0.115) (0.068) (0.115) (0.825) (0.432) 0.154 Preliminary ITC determination (β2) 0.043 ‐0.012 0.027 0.071 ‐0.190+ ‐0.281** ‐0.176* ‐0.078 0.008 0.913 ‐0.215 0.021
(0.033) (0.019) (0.021) (0.127) (0.113) (0.070) (0.070) (0.067) (0.047) (0.914) (0.182) (0.143) Final ITC determination (β3) 0.240** ‐0.048 0.077** 0.223 ‐0.303+ ‐0.524** ‐0.203* ‐0.138 0.020 0.632 ‐0.200 ‐0.026
(0.032) (0.047) (0.024) (0.205) (0.155) (0.116) (0.075) (0.085) (0.057) (0.796) (0.204) (0.207) Month dummy yes yes yes yes yes yes yes yes yes yes yes yes Product dummy yes yes yes yes yes yes yes yes yes yes yes yes
Number of observations 89,968 261,646 195,393 1,121 5,457 9724 89,968 261,646 195,393 368 3,897 8,219 Quartile group3 0.075 0.674 0.452 0.958 0.957 0.897 0.037 0.234 0.235 0.725 0.856 0.850
Note: Standard errors, clustered at the product level, are reported in the bracket. +,* and ** represent statistical significance at the10%, 5% and 1% levels, respectively.
Appendix B: A Simple Model of How Di¤erentExporters Respond to Antidumping InvestigationsDi¤erently
Here we use a simpe model to illustrate how di¤erent exporters respond toantidumping investigations (i.e., preliminary and �nal ITC determinations) dif-ferently. There are two stages, stage 1 (corresponding to the preliminary ITCdetermination) and stage 2 (corresponding to the �nal ITC determination).Firm pro�ts and antidumping duties at both stages are assumed to be thesame, i.e., �, and d, respectively. We focus on the case where the preliminaryITC determination is a¢ rmative, but at the stage 1 exporters do not know theoutcome of the �nal ITC determination. Exporters hold the common belief atstage 1 that the probability of an a¢ rmative determination at stage 2 is p.Consider the exit decision at stage 2 �rst, and to make the investigation
non-trivial, focus on the case of an a¢ rmative �nal ITC determination. Giventhe model setup, the optimal decision at stage 2 is: exporters choose to exit if� < d; and to stay if � � d.Next, consider the exit decision at stage 1. Given the a¢ rmative preliminary
ITC determination, if exporters choose to stay, then in the case of an a¢ rmative�nal ITC determination, their net payo¤ is ��d+�(��d), where � is the discountrate; whereas in the case of negative �nal ITC determination, their net payo¤ is� + ��. Hence, the expected payo¤ of staying after the a¢ rmative preliminaryITC is p [� � d+ �(� � d)] + (1 � p) [� + ��] = (1 + �) (� � pd). Hence, theoptimal decision at stage 1 is: exporters choose to stay if � � pd and to exit if� < pd.Combined, we have the following optimal sequential decisions:
� exporters stay after both a¢ rmative preliminary and �nal ITC determi-nations if � � d;
� exporters stay after the a¢ rmative preliminary ITC determination butexit after the a¢ rmative �nal ITC determination if d > � � pd;
� and exporters exit immediately after the a¢ rmative preliminary ITC de-termination if pd > �
1
top related