- Email: [email protected]
Tariff and exchange rate pass-through for Chinese exports: A firm-level analysis across customs regimes
China Economic Review 46 (2017) 87–96
Contents lists available at ScienceDirect
China Economic Review journal homepage: www.elsevier.com/locate/chieco
Tariff and exchange rate pass-through for Chinese exports: A firmlevel analysis across customs regimes
MARK
Florence Bouveta, Alyson C. Mab,⁎, Ari Van Asschec a b c
Department of Economics, Sonoma State University, 1801 E. Cotati Avenue, Rohnert Park, CA 94928, USA School of Business Administration, University of San Diego, 5998 Alcala Park, San Diego, CA 92110, USA Department of International Business, HEC Montréal, 3000 chemin de la Côte-Sainte-Catherine, Montréal, QC H3T2A7, Canada
AR TI CLE I NF O
AB S T R A CT
JEL classification: F31 F41 F14
We examine whether a firm's import content share differentially affects the degree of tariff and exchange rate pass-through into its export prices. Our pricing-to-market model suggests that a firm's import content share negatively affects the degree of exchange rate pass-through but does not affect the degree of tariff pass-through. Using firm-level data for Chinese exporting firms during the period 2000–2006, we find evidence of an almost complete exchange rate passthrough. As expected, when we distinguish firms by their trade regime, processing-trade firms, especially pure-assembly firms which tend to have higher import-content share, have a lower exchange rate pass-through than ordinary trade firms. We find no evidence that the tariff passthrough differs across the various trade regimes.
Keywords: Exchange-rate pass-through Tariff pass-through China Firm type Custom regime
1. Introduction A central concern among policymakers is how large fluctuations in tariffs and exchange rates affect the prices of internationally traded goods. This has generated a vast literature in the field of international economics, yet few studies have taken into account one of the most notable features of international trade in recent years, that is that firms increasingly rely on imported inputs to produce their exports. Johnson (2014) estimates that the import content embedded in gross world exports rose by about 10 percentage points between 1970 and 2008. This growth accelerated over time, with the import content of exports increasing roughly three times faster after 1990 than in the 1970s and 1980s. In this paper, we seek to analyze both theoretically and empirically how the import content share of exports affects both the tariff and exchange rate pass-through. In a first step, we set up a simple pricing-to-market model that allows us to investigate how the import content share of exports alters the effectiveness of tariffs and exchange rate. Similar to Feenstra (1989), our model shows that the tariff and exchange rate pass-through is symmetric as long as the entire value chain of export products is concentrated in the home country. This symmetry breaks down, however, when GVCs emerge. Once a portion of the exporter's intermediate inputs are imported, and these costs are not denominated in the exporter's domestic currency, then an exporter's marginal cost of production will only be partly exposed to exchange rate fluctuations, while remaining fully exposed to tariff changes. As a result, a firm's import content share negatively affects the degree of exchange rate pass-through (Amiti, Itskhoki, & Konings, 2014; Goldberg & Knetter, 1997), but does not affect the degree of tariff pass-through. Our model therefore predicts that the exchange rate pass-through is smaller in customs regimes with a higher import content share of exports. At the same time, the tariff pass-through should not be statistically different across customs regimes with varying import contents.
⁎
Corresponding author. E-mail addresses: [email protected] (F. Bouvet), [email protected] (A.C. Ma), [email protected] (A. Van Assche).
http://dx.doi.org/10.1016/j.chieco.2017.08.013 Received 16 November 2016; Received in revised form 28 April 2017; Accepted 28 August 2017 Available online 01 September 2017 1043-951X/ © 2017 Published by Elsevier Inc.
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Guided by the theory, we in a second step empirically evaluate the prediction that the exchange rate pass-through is smaller in customs regimes with a higher import content share, whereas the tariff pass-through is identical. We do so by using a highly detailed dataset on the universe of trade transactions by Chinese firms during the period 2000–2006. The dataset categorizes trade transactions into two broad customs regimes which have been documented to have significant differences in the import content share of their exports: ordinary trade (OT) regime and processing trade (PT). Specifically, firms under PT enjoy duty-free imports of intermediate inputs, and as a result the import content share of exports are on average 50 percentage points higher than in OT (Kee & Tang, 2016; Koopman, Wang, & Wei, 2012). Our estimation results show that, in line with our theoretical predictions, the exchange rate pass-through is significantly lower for firms in PT than in OT, but there is no significant difference in tariff pass-through (measure by anti-dumping) for firms in the two regimes. Digging deeper, we further evaluate if our results hold when we separate PT into the two sub-regimes pure assembly (PA) and import and assembly (IA). In a recent paper, Manova and Yu (2016) estimate that the import content in PA is significantly higher than in IA. In line with our theoretical predictions, we find that the exchange rate pass-through is lower in PA than in IA. At the same time, there is no difference in tariff pass-through for the three regimes. Our paper relates to a number of literatures. First, it contributes to a vast field of study that has analyzed the incomplete passthrough of exchange rate and trade policy shocks into export prices (Blonigen & Haynes, 2002; Olivei, 2002; Campa & Goldberg, 2005; Marazzi et al., 2005). In the past decade, substantial progress has been made in the study of both phenomena, albeit separately. Theoretically, the literature has focused on the various channels that lead to incomplete pass-through (Devereux & Engel, 2002): short-run nominal rigidities with prices sticky in the local currency (Gopinath & Rigobon, 2008); pricing-to-market (Atkeson & Burstein, 2008; Goldberg & Knetter, 1997; Krugman, 1987; Vigfusson, Sheets, & Gagnon, 2009; Yoshida, 2010); and import content of exports (Aksoy & Riyanto, 2000; Amiti et al., 2014; Goldberg & Campa, 2010). We add to this literature by identifying under which circumstances the symmetry between exchange rate pass-through and tariff pass-through breaks down. Empirically, a growing number of studies have started using firm-level data to circumvent methodological issues (notably endogeneity and estimation bias) related to estimating exchange rate pass-through with aggregate price data (Amiti et al., 2014; Berman, Martin, & Mayer, 2012). We follow this literature by using firm level data which allows us to treat exchange rate fluctuations as exogenous (Feinberg, 1996). Our paper is also related to recent studies that have investigated whether GVC trade behaves differently from regular trade. Yi (2003) proposes that GVC trade should be more sensitive to a world-wide decline in trade costs than regular trade since the same component crosses borders multiple times. Ma and Van Assche (2014) argue that GVCs allow firms to more easily shirk tariffs and use Chinese data to provide evidence of this. Cheung, Chinn, and Fujii (2010) and Thorbecke and Smith (2010) estimate that Chinese aggregate processing exports are more sensitive to foreign income changes than ordinary trade, but Gangnes, Ma, and Van Assche (2014) demonstrate that this is mostly due to a composition effect. We add to this literature by demonstrating that exchange rate pass-through is different between GVC trade and regular trade, but not tariff pass-through. Lastly, our study adds to the debate on trade policies' stance against China (Hu, Li, Yang, & Chao, 2016). The growing trade deficit of the United States versus China has fueled the widespread concern among U.S. policymakers that the increased exposure to Chinese imports is reducing U.S. welfare by leading to higher unemployment, lower labor force participation, and reduced wages (Autor, Dorn, & Hanson, 2016). Numerous U.S. policymakers and commentators have swiftly reacted by putting forward a myriad of trade policies aimed to limit the growth of Chinese exports, ranging from pressuring the country to appreciate their currency to treating China's alleged currency manipulation as a source of dumping that would permit the imposition of antidumping on Chinese imports (Staiger and Sykes, 2010). Our paper provides new insights to policymakers by illustrating that policies targeting China's exchange rate may not be as effective due to China's heavy reliance on imported inputs. The remainder of the paper is organized as follows. We present in Section 2 a pricing-to-market model to understand how changes in exchange rate and ad valorem tariffs affect a firm's pricing decisions. In Section 3, we describe the dataset on Chinese exporting firms, as well as the macroeconomic data used in this paper and some summary statistics. Section 4 reports our model specification. Section 4 provides estimations of exchange rate pass-through and antidumping pass-through for Chinese export prices. Section 5 concludes and discusses directions for future research. 2. Model Our model is a variant of Burstein and Gopinath (2014). Exporting firm i in industry k sets an optimal price, piknt, as the markup over its marginal costs when selling to destination country n in period t:
piknt = μiknt + mciknt + τknt ,
(1)
where lower case characters denote variables expressed in natural logarithm. The destination-country-specific mark-up μiknt depends on the price charged by the exporting firm i relative to the aggregate industry price level in the destination country n. That is, μiknt = μiknt[piknt − pknt]. We assume that μiknt is a decreasing function of the firm's price relative to the aggregate industry level price in ∂μ [.] the destination country. In other words, there is decreasing returns to scale so that Γiknt = − ∂ (p iknt− p ) > 0 . iknt knt A firm faces ad valorem trade cost τknt that is both industry and destination-country specific. The dollar marginal cost is given by mciknt[qiknt, wikt, miknt(ent)], where qiknt is the quantity sold by firm i in destination country n in period t, wikt summarizes those variables that impact the local production costs incurred by the firm such as wages and total factor productivity, and miknt(ent) captures the cost ∂m [.] of imported inputs which are denominated in foreign currency, where we assume ∂iknt = 1. We specify marginal cost in log so that: e nt
mciknt = Aiknt + κknt qiknt + αiknt miknt (ent ) + (1 − αiknt ) wikt
(2) 88
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
where Aiknt denotes a constant term, κknt is the elasticity of the marginal cost with respect to output (which we assume is common across firms) and αiknt < 1 is the cost share of the imported inputs in total marginal costs, respectively. Inserting Eqs. (2) into (1) and log-differentiating, we obtain that the log change in price Δpiknt can be approximated as:1
Δpiknt = −Γiknt Δpiknt + κknt Δqiknt + αiknt Δent + (1 − αiknt )Δwikt + Δτknt
(3)
Assume that log demand is given by qikn = q[pikn − pkn] + qkn where qkn denotes the log of aggregate demand in industry k in country n. Log differentiating,
Δqiknt = −εiknt Δpiknt where εiknt =
∂q [.] − ∂p iknt
Δpiknt =
(4)
> 0 is the price elasticity of demand. Combining Eqs. (3) and (4) and collecting terms, we obtain:
1 (αiknt miknt Δent + (1 − αiknt ) Δwikt + Δτknt ) 1 + Γiknt + Φiknt
(5)
where Φiknt = κikntεiknt > 0 is the partial elasticity of marginal cost with respect to the relative price. We can use Eq. (5) to investigate the elasticity of domestic export prices with respect to ad valorem tariffs:
Δpiknt 1 = ≤ 1. Δτknt 1 + Γiknt + Φiknt
(6)
Intuitively, an ad valorem tariff increase leads to an imperfect pass-through to domestic prices for two reasons. First, scale economies, captured by Γiknt. If there are decreasing returns to scale, Γiknt > 0, then a reduction in the quantity a firm produces will reduce the average cost of the firm and dampen its desire to raise prices. Second, if the elasticity of demand is high so that Φiknt > 0, then the demand decrease will decrease a firm's marginal cost and dampen its desire to raise prices. Note, however, that it is not a function of the import content share of exports. We state this as our first hypothesis: Hypothesis 1. All else equal, import content (αin) does not affect a firm's tariff pass-through. In a similar fashion, we can use Eq. (5) to investigate the elasticity of domestic export prices with respect to an exchange rate movement:
Δpiknt αiknt = Δent 1 + Γiknt + Φiknt
(7)
Intuitively, when the currency in the exporting country appreciates relative to the destination country, this raises the foreignmarket marginal costs of the exporting firm, all else equal. The sensitivity of foreign-market marginal costs to the exchange rate movement depends on αikn. Suppose the exporting firm sources all of its inputs domestically so that αiknt = 1, then there is a symmetry between ad valorem tariff pass-through and exchange rate pass-through. If αiknt < 1, however, some of its inputs are sourced globally and these inputs are priced in the foreign currency, whose foreign price is unaffected by the exchange rate movement. In this case, only fraction 1 − αiknt of the firm's marginal cost is affected and the exchange rate pass-through will be a declining function of the import content share. Hypothesis 2. All else equal, firms with a higher import content (αiknt) will have a smaller exchange rate pass-through. In the subsequent sections, we will empirically test the symmetry between tariff and exchange rate pass-through where we control for import content by taking into account variations across customs regimes. 3. Data and methods To empirically test our predictions, we take advantage of a firm-level dataset on trade for the period 2000–2006 that was compiled by the General Administration of Customs of the People's Republic of China, or China Customs Statistics in short. The database contains detailed information on the universe of trade transactions at the firm-industry-destination country-year level. It reports f.o.b. values (in US$) and quantities at the HS 8-digit level, thus allowing us to calculate unit values. We limit the products to those in the manufacturing sectors by omitting agriculture and specialty items such as rare coins and painting. Firm characteristics such as size and number of employees are not available in the data set and will be controlled for with firm fixed effects. The key goal of this paper is to investigate whether import content share affects the degree of exchange rate and tariff passthrough. Estimating a firm's import content share of exports is notoriously difficult since it requires information about a firm's domestic value added (Amiti et al., 2014). We circumvent this problem by taking advantage of the distinction between two broad customs regimes: processing trade (PT) and ordinary trade (OT). These two trade forms differ in terms of import tariff treatment and the ability of firms to sell on the domestic market:
• Under PT, firms enjoy the right of duty-free imports of intermediate goods and capital equipment that are used in their export 1 To simplify the exposition, we assume here that a tariff and exchange rate movement does not affect the aggregate industry level price in the destination country. Our key results remain unaffected if we relax this assumption.
89
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Fig. 1. Share of trade by custom regime in total export, 2000–2006. Source: Authors' calculation using the trade data from the China Custom Statistics.
processing activity, but face restrictions in selling to the domestic market.
• Under OT, firms need to pay import duties on imported inputs but can sell their output locally. Several seminal studies have demonstrated that these distinct customs characteristics ensure that firms in PT tend to have a far higher foreign content embodied in their exports than firms in OT. Using input-output data, Koopman et al. (2012) estimated that the foreign content share in 2002 amounted to 89.9% of processing exports compared to 10.7% in ordinary exports. In 2007, the foreign content share of processing exports was 89.5%, while it was 16.1% in ordinary exports. Using firm-level data, Kee and Tang (2016) estimated that the foreign content share hovered around 50% of Chinese processing exports, whereas it hovered around 10% for Chinese ordinary exports. In addition to the well-known distinction between OT and PT, a key distinction in the data is between two types of processing trade in China's Customs Statistics: pure assembly (PA) and import and assembly (IA). The main difference between the two customs sub-regimes, which result from different legal regimes to organize processing trade activities, lies in the allocation of control rights of the imported inputs (Feenstra & Hanson, 2005).
• In the PA regime, the manufacturer in China receives materials from a foreign principal free of charge. It then performs its •
manufacturing functions as per the requirements of the foreign principal and then exports the finished goods to the foreign principal. The manufacturer only gets compensated by the overseas party with a processing fee. In the IA regime, the manufacturer in China purchases and pays foreign exchange for imported materials free of import duty and VAT. It then performs manufacturing functions and exports the finished products to an overseas party (not necessarily the same as the input supplier) with the compensation of foreign exchange collected.
Manufacturers in IA and PA share the common feature that they perform manufacturing services to third-party companies which are located overseas. IA firms have a number of extra responsibilities over PA firms however. First, IA firms have the responsibility of searching and obtaining the imported materials themselves prior to conducting the manufacturing activity, whereas PA firms passively receive orders and materials from its foreign client and export all the processed goods to this foreign principal. Second, IA firms need to pay for the imported raw materials and parts, whereas PA firms receive them from their foreign trading partners free of charge. Third, to comply with the more complex regulations and approval processes of IA compared to PA, IA firms are required to make greater investments in inventory storages and management than PA firms (Feenstra & Hanson, 2005). These extra responsibilities imply that the foreign content share for PA trade is significantly higher than for IA trade (Manova & Yu, 2016). Fig. 1 presents the importance of PT in total exports. From 2000 to 2006 the share of PT in total exports exceeded 50%. Fig. 1 also shows that the share of PA in total exports decreased from 17.2% in 2000 to 9.8% in 2006. During the same period, the share of IA increased from 39.1% to 43.0%. We conduct our empirical analysis in three steps. In a first step, we investigate the determinants of exchange rate and tariff passthrough by estimating a generalized model of Eq. (5) for the period 2000–2006:
piknt = α 0 + α1 pikn, t − 1 + α2 ent + α3 ADknt + Xiknt β + δn + δk + δi + εiknt ,
(8)
where lower case characters denote variables expressed in natural logarithm.2 piknt is the f.o.b. export price of firm i in the importer's 2 Similar to Blonigen and Haynes (2002) and Marchand (2012) in estimating exchange-rate and tariff pass-through our specification is in log-level rather than first difference. Moreover our dataset has a large N and very small T (annual data from 2000 to 2006). With a large N (> 2 million observations) and small T (10 year period), Berman et al. (2012) also use a log-level estimation to analyze the exchange rate pass-through among French firms.
90
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
currency,3 in industry k, selling in destination market n, in year t; ent denotes the nominal spot exchange rate between the renminbi and the currency of destination country n expressed as units of country n's currency for one renminbi; ADknt is a dummy variable that takes on the value of one if there is an antidumping measure imposed by country n in industry k in year t; Xiknt is a vector of demandshifting control variables; δn, δk and δi are destination country, industry and firm fixed effects; and εiknt is a stochastic error. In a second step, we investigate variations in exchange rate and tariff pass-through across customs regimes (ordinary trade and processing trade). Finally, within processing trade, we distinguish the exchange rate pass-through between pure assembly and import and assembly. 3.1. Key variables of interest We have created two measures to capture exchange rate fluctuations: a nominal and a real exchange rate measure. 3.1.1. Nominal exchange rate Exchange rate series are nominal spot exchange rate expressed as units of the domestic currency for one renminbi. The data are obtained mostly from the International Monetary Fund's International Financial Statistics database, which we supplemented with data from several central banks. Because the database only provides data on the exchange rate between the domestic currencies and the US dollar, we constructed the exchange rate series by combining the aforementioned series with the spot exchange rate between the US dollar and the renminbi. For Eurozone countries, the exchange rate is expressed in terms of euros per renminbi. If some countries joined the Eurozone later than 2000 (such as Estonia and Slovenia), we use the fixed rate between these countries' national currencies and the euro to expand the series prior to their adoption of the euro. 3.1.2. Real exchange rate The real exchange rate is calculated as the nominal spot exchange rate multiple by the ratio of the Chinese PPI to the destination PPI. Our analysis requires data on time-varying country-product level tariffs which are difficult to obtain for China. To operationalize this, we have opted to rely antidumping measures which are country-product specific and exhibit sufficient variation over time. 3.1.3. Antidumping As our measure of a country-product-specific trade policy shock, we use antidumping cases against China as identified in the World Bank's Global Antidumping Database (GAD) (Bown, 2009).4 The benefit of using antidumping as a measure for a countryspecific trade policy shock is that it is generally imposed by a country on firms of a specific country in a specific industry. The GAD database has detailed information on each antidumping case, such as product information (6-digit HS codes), the investigating country, the target country, the preliminary determination date and the year it was revoked. For our analysis, we collect information on all antidumping cases against China during the period 2000–2006. We match the GAD data with the Chinese Customs data at the HS-6 digit level, the most disaggregated level at which the two datasets are comparable (Ma & Van Assche, 2014). Finally, we create a dummy ADknt which takes on the value of one if there is an antidumping measure imposed by country n in industry k in year t. It is important to point out that while the use of the dummy variable ADknt makes it impossible to identify the size of tariff passthrough, given that it is country-product-specific it allows us to test our hypothesis that the tariff pass-through for a specific product does not vary across customs regimes with varying import content. An incomplete exchange rate and tariff pass-through implies that the exporting firm reacts to a currency appreciation or to an antidumping measure by reducing its f.o.b. price expressed in the domestic currency. In our estimation equation, we would thus find evidence of incomplete exchange rate and tariff pass-through if both α2 < 0 and α3 < 0. The size of the coefficient also matters for exchange rate pass-through: a coefficient closer to zero would indicate almost complete exchange rate. It is difficult to interpret the size of the coefficient α3 due to our use of the dummy variable ADknt. 3.2. Control variables We include a lag of the export price to account for serial correlation in pricing decisions (Bussière, Chiaie, & Peltonen, 2014; Zhou & Kim, 2011). We further include a set of control variables that prior literature has identified as important drivers of exchange rate and tariff pass-through. We proxy trends in the production costs in China and the destination country with the natural log of the Chinese and destination market producer price index: ln(Chinese PPI) and ln(destination PPI). We also include the price of foreign competitors in the destination market, ln(competing price), given by the unit value of a country's import at the 6-digit HS level from all countries excluding China. To capture the degree of import competition, we include another control variable, Ln(Chinese import share), which measures the share of Chinese imports in total imports in industry k in destination country n. As a robustness check, we 3 While the Chinese Customs data include all countries in which China exports to, due to data limitation in the control variables such as Producer Price Index, the following countries are included in the estimation: Australia, Austria, Belgium, Bulgaria, Canada, Croatia, Denmark, Estonia, France, Germany, India, Indonesia, Ireland, Italy, Japan, Republic of Korea, Luxembourg, Malaysia, Mexico, Netherlands, Philippines, Poland, Portugal, Singapore, Spain, Sweden, Taiwan, Thailand, United Kingdom, United States, and Vietnam. 4 Antidumping captures country-product-specific shocks that are imposed onto China's exports during the period 2000–2006 relative to the lack of variation in tariff levels.
91
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Table 1 Data description. Series
Definition
Source
Export price
Firms' export unit value price in destination market expressed in currency of the destination market Nominal bilateral spot exchange rate, expressed as units of destination market's currency per renminbi Equals one if antidumping duties are imposed on Chinese firms from a specific industry Producer price index, 2010 = 100 Producer price index, 2010 = 100
Chinese Custom Statistics
Exchange rate Anti-dumping dummy variable Chinese PPI Destination market PPI Export price of foreign competitors Share of Chinese imports in total imports in destination market Herfindahl-Hirschman Index for imports Real GDP of destination market Real GDP per capita of the destination markets
Non-Chinese firms' export unit value price in destination market expressed in currency of the destination market Share of Chinese imports in total imports in destination market Measure of concentration among importing firms. It is computed as the sum of the squared import shares Real GDP expressed in currency of the destination market Real GDP per capita expressed in currency of the destination market
IFS, European Central Bank, Bank of Canada, Forex.com, Bank of Estonia Chinese Custom Statistics OECD, OECD, Singapore Department of Statistics, Malaysia Department of Statistics Chinese Custom Statistics Chinese Custom Statistics Authors' computation based on Chinese Custom Statistics OECD OECD
substitute a Herfindahl-Hirschman index using import share for each country in destination country n at the industry level to capture import competition at the 6-digit HS level. Ln(destination real GDP) is included as our measure of market size and business cycle, and per capita income respectively (Li, Juanyi Jenny, & Zhao, 2014, Li, Ma, & Yuan, 2015, Manova and Zhang, 2012). Finally, we include fixed effects for firms, countries, and industries. As noted by Manova and Zhang (2012) the fixed effects control for systematic differences in unit values across firms, product characteristics, and destination countries, including industry and producer costs that are associated with changes in production technology or input costs. In other words, our analysis investigates whether within a firm-product-destination combination the tariff and exchange rate pass-through varies across customs regimes. The series and their sources are described in Table 1, while Table 2 provides data summary statistics. 3.3. Variations across customs regimes To investigate variations in exchange rate and tariff pass-through across customs regimes, we create the dummy variable PTr which equals 1 if trade occurs under PT, and 0 otherwise. We then interact PTr with our two variables of interest to investigate if tariff and exchange rate pass-through significantly differs between PT and OT:
piknrt = α 0 + α1 piknr , t − 1 + α2 ent + α3 ADknt + α4 PTr ∗ent + α5 PTr ∗ADknt + α6 PTr + Xiknt β + δk + δn + δt + εiknrt ,
(9)
There will be evidence of Hypothesis 1 if α5 = 0. In that case, the tariff pass-through is not significantly different between the PT Table 2 Summary statistics. Variable By firm-product-country-year Local price Competing price Chinese import share Herfilndahl-Hirschman Index Anti-dumping dummy variable Ordinary Trade Dummy Variable Pure-Assembly Dummy Variable Import-and-Assembly Dummy Variable Chinese PPI by Industry By country-year Nominal exchange rate Real exchange rate Destination PPI Chinese PPI Destination real GDP/Capita in US $ Destination real GDP in US $
Obs
Mean
Std. dev
Min
Max
12,931,467 503,613 455,802 1,071,351 12,931,467 12,931,467 12,931,467 12,931,467 28,545
2,904,153 3,045,247 17.76 4293 0.02 0.84 0.10 0.06 102.11
1.92E + 09 1.26E + 08 23.43 2515 0.11 0.37 0.30 0.24 4.81
2.39E-11 3.00E-08 5.00E-07 350 0 0 0 0 93.08
1.95E + 12 4.91E + 10 100 10,000 1 1 1 1 122.50
211 211 217 7 217 217
103.55 202.09 103.04 82.80 23,730 1.10E + 12
382.01 748.65 4.22 5.09 18,067 2.26E + 12
0.07 0.08 93.40 77.50 433 3.85E + 09
2005.48 3727.64 118.90 90.90 83,576 1.34E + 13
92
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Table 3 Benchmark specification.
ln(local pricet − 1) ln(ex_rate)
(1)
(2)
(3)
0.809*** [0.001] − 0.028*** [0.003]
0.809*** [0.001] − 0.027*** [0.003] − 0.002*** [0.000]
0.809*** [0.001] − 0.028*** [0.003]
ln(ex_rate) ∗ PT ln(ex_rate) ∗ PA ln(ex_rate) ∗ IA Antidumping
− 0.002*** [0.000] 0.0004 [0.001] − 0.012*** [0.003]
− 0.012*** [0.003] − 0.005 [0.005]
antidumping ∗ PT antidumping ∗ PA antidumping ∗ IA ln(Chinese PPI) ln(destination PPI) ln(competing price) ln(destination real GDP) ln(Chinese import share)
− 0.006 [0.005] − 0.004 [0.008] 0.525*** [0.013] − 0.000 [0.006] 0.073*** [0.001] 0.174*** [0.003] − 0.006*** [0.000]
0.526*** [0.013] 0.0001 [0.981] 0.074*** [0.001] 0.174*** [0.003] − 0.006*** [0.000] 0.036*** [0.002]
PT PA IA Firm FE Country FE Industry FE Observations R-squared
− 0.002*** [0.0003] 0.0004 [0.000] Yes Yes Yes 4,875,525 0.955
Yes Yes Yes 4,864,456 0.955
− 0.002*** [0.000] 0.0004 [0.001] − 0.012*** [0.003]
− 0.006 [0.005] − 0.004 [0.008] 0.525*** [0.013] − 0.000 [0.006] 0.073*** [0.001] 0.174*** [0.003] − 0.006*** [0.000]
− 0.002*** [0.0003] 0.0004 [0.000] Yes Yes Yes 4,875,525 0.955
than OT. There will be evidence of Hypothesis 2 if α4 < 0. This would indicate that the exchange rate pass-through is significantly smaller in PT than OT. In a final step, we further divide PT into the two sub-regimes PA and IA. Specifically, we estimate:
piknrt = α 0 + α1 piknr , t − 1 + α2 ent + α3 ADknt + α4 PAr ∗ent + α5 PAr ∗ADknt + α6 PAr + α 7 IAr ∗ent + α8 IAr ∗ADknt + α 9 IAr + Xiknt β (10)
+ δk + δn + δt + εiknrt ,
Hypothesis 1 suggests that α5 = α8 = 0. In that case, the tariff pass-through is not affected by the declining import content under PA-type processing trade, IA-type processing trade and ordinary trade. There will be evidence of Hypothesis 2 if α4 < α7 < 0. This would indicate that the exchange rate pass-through is significantly smaller in PA-type processing trade than in IA-type processing trade and ordinary trade in that order. 4. Results In Table 3 we present our results with industry, firm, and country fixed effects. In column (1), we present our benchmark model which estimates Eq. (8) pooled across customs regimes. In column (2), we provide the regression results for Eq. (9). The results of column (1) show that the ERPT is nearly complete. Firms react to an appreciation of the renminbi by passing 97% of the rise in the exchange rate, leading to only a minor decrease in the export price. The negative and statistically significant coefficient on the antidumping dummy suggests that exporters try to mitigate the negative demand impact of the new duties by reducing their export price. The coefficients on the control variables are almost all statistically significant and with the expected sign with the exception of the PPI of the destination markets. An increase in the export price is also expected following a rise in import price of nonChinese foreign competitors in the destination country and following an increase in the destination market's income through a higher 93
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Table 4 Robustness check.
ln(local pricet − 1) ln(ex_rate) ln(ex_rate) ∗ PA ln(ex_rate) ∗ IA
(1)
(2)
(3)
0.810*** [0.001] −0.028*** [0.003] −0.002*** [0.000] 0.001 [0.001]
0.810*** [0.001] − 0.024*** [0.003] − 0.002*** [0.001] 0.001 [0.001]
0.810*** [0.001]
ln(real ex_rate) ln(real ex_rate) ∗ PA ln(real ex_rate) ∗ IA Antidumping antidumping ∗ PA antidumping ∗ IA ln(Chinese PPI) ln(destination PPI) ln(competing price) ln(destination real GDP) ln(HHI) PA IA
−0.014*** [0.003] −0.005 [0.005] −0.003 [0.008] 0.524*** [0.013] −0.005 [0.006] 0.074*** [0.001] 0.174*** [0.003] −0.006*** [0.000] 0.038*** [0.002] 0.026*** [0.002]
− 0.014*** [0.003] − 0.006 [0.005] − 0.003 [0.008] 0.528*** [0.013] − 0.003 [0.006] 0.074*** [0.001] 0.169*** [0.003] − 0.011*** [0.001] 0.038*** [0.002] 0.026*** [0.002] 0.007*** [0.001] Yes Yes Yes 4,898,352 0.956
WTO Firm FE Country FE Industry FE Observations R-squared
Yes Yes Yes 4,898,352 0.956
− 0.022*** [0.002] − 0.002*** [0.000] 0.001 [0.001] − 0.012*** [0.003] − 0.005 [0.005] − 0.002 [0.008]
0.074*** [0.001] 0.167*** [0.002] − 0.011*** [0.001] 0.038*** [0.002] 0.026*** [0.002]
Yes Yes Yes 4,898,352 0.956
demand. Owing to increased competition among Chinese exporting firms, a greater share of Chinese imports in total imports in industry j in destination country k decreases the Chinese export price. The empirical test of Hypotheses 1 and 2 are presented in column (2). To begin with, the statistically insignificant coefficient on the interaction between a tariff or antidumping duties and processing trade substantiates Hypothesis 1. Antidumping duties do not affect the pricing strategy of processing trade firms differently from ordinary trade firms within the same sector and towards the same destination country. Moreover, the negative and statistically significant coefficient on the interaction term ln(ex_rate) ∗ PT provides evidence in support of Hypothesis 2. In particular, the result suggests that PT firms absorb the same renminbi appreciation by further decreasing the export price than OT firms, albeit by a marginal amount. Taken together, the results suggest that exchange rate passthrough is smaller in customs regimes with higher import content share, whereas tariff pass-through is invariant across customs regimes with different import content shares. We further disaggregate the processing trade regime to account for the allocation of control rights of the imported inputs. In column (3), we test for differences in the exchange rate and antidumping pass-through of both types of processing trade firms, namely PA firms and IA firms.5 In support of Hypothesis 1, the coefficients on the interaction terms antidumping ∗ PA and antidumping ∗ IA are statistically insignificant, suggesting that the antidumping pass-through is similar across the three customs regimes despite large differences in import content share. However, the coefficients on the interaction between these two types of processing trade with the exchange rate term indicates that the Hypothesis 2 holds only for pure-assembly firms.
5 The larger number of observations in column 2 stem from the fact that some firms operating both as input-control and pure assembly would have been counted only once in column 1 but twice in column 2.
94
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
We have investigated a number of alternative specifications to check the robustness of our results. First, we have re-estimated Eq. (10) with the inclusion of the Herfindahl-Hirschman Index to capture import competition at the 6-digit HS level.6 The results, presented in column (1) of Table 4 suggest that our results are consistent with the benchmark specification. Second, we have in column (2) accounted for China's accession to the WTO by including a dummy variable that equals to 1 from 2002 forward.7 Again, our key results are unchanged. Finally, we have in column (3) used real exchange rates instead of nominal exchange rates and found that our results remain consistent with our predictions. 5. Conclusion In this paper we examine whether a firm's import content share differentially affects the degree of tariff and exchange rate passthrough into its export prices. Our pricing-to-market model suggests that a firm's import content share negatively affects the degree of exchange rate pass-through but does not affect the degree of tariff pass-through. Using firm-level data for Chinese exporting firms, we find evidence of an almost complete exchange rate pass-through (around 97%), which is in line with the baseline estimation in the literature (Kiliç, 2016; Li et al., 2015). As expected, when we distinguish firms by their trade regime, processing-trade firms, especially pure-assembly firms which tend to have higher import-content share, have a slightly lower ERPT than ordinary trade firms. This almost-complete ERPT suggests that an appreciation of the renminbi would translate into higher export prices for Chinese products. Such an appreciation might not necessarily lead to a reduction in China's exports and thus a decrease in its trade surplus with other countries for several reasons. First, large ERPT for export prices might not translate into similar ERPT for import prices, which Campa and Goldberg (2005) refer to as the “exchange rate disconnect” puzzle. They suggest that this disconnect might be related to distribution costs in the destination markets, as well as to other local value-added components. Second, as noted by Xing (2012), most of China's trade surplus originates from processing trade firms. Whether a renminbi appreciation affects Chinese exports depends on the inputs' origin. One would expect the renminbi appreciation to have a more significant effect on the exports of firms whose value added comes mostly from China than of the exports of firms whose manufacturing costs comprise mostly components made in high-wage countries, such as Japan and South Korea. The growing share of domestic content in Chinese exports (Thorbecke, 2013) implies that the renminbi will have an increasing role to play in addressing China's trade imbalances. References Aksoy, Y., & Riyanto, Y. E. (2000). Exchange rate pass-through in vertically related markets. Review of International Economics, 8(2), 235–251. Amiti, M., Itskhoki, O., & Konings, J. (2014). Importers, exporters, and exchange rate disconnect. American Economic Review, 104(7), 1942–1978. Atkeson, A., & Burstein, A. (2008). Pricing-to-market, trade costs, and international relative prices. American Economic Review, 98(5), 1998–2031. Autor, D. H., Dorn, D., & Hanson, G. H. (2016). The China shock: Learning from labor market adjustment to large changes in trade. Annual Review of Economics, 8(1). Berman, N., Martin, P., & Mayer, T. (2012). How do different exporters react to exchange rate changes? The Quarterly Journal of Economics, 127(1), 437–492. Blonigen, B., & Haynes, S. E. (2002). Antidumping investigations and the pass-through of antidumping duties and exchange rates. American Economic Review, 92(4), 1044–1061. Bown, C. P. (2009). Global antidumping database (current version 5.0) Brandeis University mimeo. Burstein, A., & Gopinath, G. (2014). “International prices and exchange rates.” 4:391–451. Handbook of international economics(4th ed.). 4. Handbook of international economics (pp. 391–451). Elsevier. Bussière, M., Chiaie, S. D., & Peltonen, T. (2014). Exchange rate pass-through in the global economy: The role of emerging market economies. IMF Economic Review, 62, 146–178. Campa, J. M., & Goldberg, L. S. (2005). Exchange rate pass-through into import prices. The Review of Economics and Statistics, 87, 679–690. Cheung, Y.-W., Chinn, M. D., & Fujii, E. (2010). China's current account and exchange rate. China's growing role in world trade (pp. 231–271). University of Chicago Press. Devereux, M. B., & Engel, C. (2002). Exchange rate pass-through, exchange rate volatility, and exchange rate disconnect. Journal of Monetary Economics, 49(5), 913–940. Feenstra, R. C. (1989). Symmetric pass-through of tariffs and exchange rates under imperfect competition: An empirical test. Journal of International Economics, 27(1), 25–45. Feenstra, R. C., & Hanson, G. H. (2005). Ownership and control in outsourcing to China: Estimating the property-rights theory of the firm. Quarterly Journal of Economics, 120, 729–761 (2,May). Feinberg, R. M. (1996). A simultaneous analysis of exchange rate pass-through into prices of imperfectly substitutable domestic and import goods. International Review of Economics and Finance, 5(4), 407–416. Gangnes, B. S., Ma, A. C., & Van Assche, A. (2014). Global value chains and trade elasticities. Economics Letters, 124(3), 482–486. Goldberg, L. S., & Campa, J. M. (2010). The sensitivity of the CPI to exchange rates: Distribution margins, imported inputs, and trade exposure. The Review of Economics and Statistics, 92(2), 392–407. Goldberg, P. K., & Knetter, M. M. (1997). Goods prices and exchange rates: What have we learned? Journal of Economic Literature, 35, 1243–1272. Gopinath, G., & Rigobon, R. (2008). Sticky borders. Quarterly Journal of Economics, 123(2), 531–575. Hu, M., Li, Y., Yang, J., & Chao, C. C. (2016). Actual intervention and verbal intervention in the Chinese RMB exchange rate. International Review of Economics and Finance, 43, 499–508. Johnson, R. C. (2014). Five facts about value-added exports and implications for macroeconomics and trade research. The Journal of Economic Perspectives, 28(2), 119–142. Kee, H. L., & Tang, H. (2016). Domestic value added in exports: Theory and firm evidence from China. American Economic Review, 106(6), 1402–1436. Kiliç, R. (2016). Regime-dependent exchange-rate pass-through to import prices. International Review of Economics and Finance, 41, 295–308. Koopman, R., Wang, Z., & Wei, S.-J. (2012). Estimating domestic content in exports when processing trade is pervasive. Journal of Development Economics, 99(1), 178–189. Krugman, P. (1987). Pricing to market when the exchange rate changes. In S. Arndt, & J. D. Richardson (Eds.). Real-financial linkages among open economies (pp. 49–70). Cambridge: MIT Press.
6 7
We are grateful to an anonymous reviewer for the suggestion of the inclusion of the HHI and the WTO dummy. China entered the WTO in December 2001.
95
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Li, H., Ma, H., & Yuan, X. (2015). How do exchange rate movements affect Chinese exports ? A firm-level investigation. Journal of International Economics, 97(1), 148–161. Li, Y. A., Juanyi Jenny, X., & Zhao, C. (2014). Import response to exchange rate fluctuation: Evidence from Chinese firms. Forum for research in empirical international trade working papers series, number 623. Ma, A. C., & Van Assche, A. (2014). Vertical specialization, tariff shirking, and trade. In B. Ferrarini, & D. Hummels (Vol. Eds.), Asia and global production networks. 2014. Asia and global production networks (pp. 148–179). Massachusetts: Edward Elgar Publishing, Inc (Nov.). Manova, K., & Yu, Z. (2016). How firms export: Processing vs. ordinary trade with financial frictions. Journal of International Economics, 100, 120–137. Manova, K., & Zhang, Z. (2012). Export prices across firms and destinations. The Quarterly Journal of Economics, 127(1), 379–436. Marazzi, M., Sheets, N., Vigfusson, R. J., Faust, J., Gagnon, J., Marquez, J., ... Rogers, J. (2005). Exchange rate pass-through to U.S. import prices: Some new evidence. Board of governors of the federal reserve system international finance discussion paper 2005-833. Marchand, B. U. (2012). Tariff pass-through and the distributional effects of trade liberalization. Journal of Development Economics, 99(2), 265–281. Staiger, R. W., & Sykes, A. O. (2010). ‘Currency manipulation’and world trade. World Trade Review, 9(4), 583–627. Thorbecke, W. (2013). Investigating China's disaggregated processed exports: Evidence that both the RMB and exchange rates in supply chain countries matter. The World Economy, 6(10), 1245–1260. Thorbecke, W., & Smith, G. (2010). How would an appreciation of the renminbi and other east Asian currencies affect China's exports? Review of International Economics, 18(1), 95–108. Vigfusson, R. J., Sheets, N., & Gagnon, J. (2009). Exchange rate pass-through to export prices: Assessing cross-country evidence. Review of International Economics, 17(1), 17–33. Xing, Y. (2012). Processing trade, exchange rates and China's bilateral trade balances. Journal of Asian Economics, 23(5), 540–547. Yi, K.-M. (2003). Can vertical specialization explain the growth of world trade? Journal of Political Economy, 111(1), 52–102. Yoshida, Y. (2010). New evidence for exchange rate pass-through: Disaggregated trade data from local ports. International Review of Economics and Finance, 19(1), 3–12. Zhou, J., & Kim, D. (2011). China's share of U.S. imports and exchange rate pass-through to U.S. import prices. China & World Economy, 19(3), 87–104.
96
Contents lists available at ScienceDirect
China Economic Review journal homepage: www.elsevier.com/locate/chieco
Tariff and exchange rate pass-through for Chinese exports: A firmlevel analysis across customs regimes
MARK
Florence Bouveta, Alyson C. Mab,⁎, Ari Van Asschec a b c
Department of Economics, Sonoma State University, 1801 E. Cotati Avenue, Rohnert Park, CA 94928, USA School of Business Administration, University of San Diego, 5998 Alcala Park, San Diego, CA 92110, USA Department of International Business, HEC Montréal, 3000 chemin de la Côte-Sainte-Catherine, Montréal, QC H3T2A7, Canada
AR TI CLE I NF O
AB S T R A CT
JEL classification: F31 F41 F14
We examine whether a firm's import content share differentially affects the degree of tariff and exchange rate pass-through into its export prices. Our pricing-to-market model suggests that a firm's import content share negatively affects the degree of exchange rate pass-through but does not affect the degree of tariff pass-through. Using firm-level data for Chinese exporting firms during the period 2000–2006, we find evidence of an almost complete exchange rate passthrough. As expected, when we distinguish firms by their trade regime, processing-trade firms, especially pure-assembly firms which tend to have higher import-content share, have a lower exchange rate pass-through than ordinary trade firms. We find no evidence that the tariff passthrough differs across the various trade regimes.
Keywords: Exchange-rate pass-through Tariff pass-through China Firm type Custom regime
1. Introduction A central concern among policymakers is how large fluctuations in tariffs and exchange rates affect the prices of internationally traded goods. This has generated a vast literature in the field of international economics, yet few studies have taken into account one of the most notable features of international trade in recent years, that is that firms increasingly rely on imported inputs to produce their exports. Johnson (2014) estimates that the import content embedded in gross world exports rose by about 10 percentage points between 1970 and 2008. This growth accelerated over time, with the import content of exports increasing roughly three times faster after 1990 than in the 1970s and 1980s. In this paper, we seek to analyze both theoretically and empirically how the import content share of exports affects both the tariff and exchange rate pass-through. In a first step, we set up a simple pricing-to-market model that allows us to investigate how the import content share of exports alters the effectiveness of tariffs and exchange rate. Similar to Feenstra (1989), our model shows that the tariff and exchange rate pass-through is symmetric as long as the entire value chain of export products is concentrated in the home country. This symmetry breaks down, however, when GVCs emerge. Once a portion of the exporter's intermediate inputs are imported, and these costs are not denominated in the exporter's domestic currency, then an exporter's marginal cost of production will only be partly exposed to exchange rate fluctuations, while remaining fully exposed to tariff changes. As a result, a firm's import content share negatively affects the degree of exchange rate pass-through (Amiti, Itskhoki, & Konings, 2014; Goldberg & Knetter, 1997), but does not affect the degree of tariff pass-through. Our model therefore predicts that the exchange rate pass-through is smaller in customs regimes with a higher import content share of exports. At the same time, the tariff pass-through should not be statistically different across customs regimes with varying import contents.
⁎
Corresponding author. E-mail addresses: [email protected] (F. Bouvet), [email protected] (A.C. Ma), [email protected] (A. Van Assche).
http://dx.doi.org/10.1016/j.chieco.2017.08.013 Received 16 November 2016; Received in revised form 28 April 2017; Accepted 28 August 2017 Available online 01 September 2017 1043-951X/ © 2017 Published by Elsevier Inc.
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Guided by the theory, we in a second step empirically evaluate the prediction that the exchange rate pass-through is smaller in customs regimes with a higher import content share, whereas the tariff pass-through is identical. We do so by using a highly detailed dataset on the universe of trade transactions by Chinese firms during the period 2000–2006. The dataset categorizes trade transactions into two broad customs regimes which have been documented to have significant differences in the import content share of their exports: ordinary trade (OT) regime and processing trade (PT). Specifically, firms under PT enjoy duty-free imports of intermediate inputs, and as a result the import content share of exports are on average 50 percentage points higher than in OT (Kee & Tang, 2016; Koopman, Wang, & Wei, 2012). Our estimation results show that, in line with our theoretical predictions, the exchange rate pass-through is significantly lower for firms in PT than in OT, but there is no significant difference in tariff pass-through (measure by anti-dumping) for firms in the two regimes. Digging deeper, we further evaluate if our results hold when we separate PT into the two sub-regimes pure assembly (PA) and import and assembly (IA). In a recent paper, Manova and Yu (2016) estimate that the import content in PA is significantly higher than in IA. In line with our theoretical predictions, we find that the exchange rate pass-through is lower in PA than in IA. At the same time, there is no difference in tariff pass-through for the three regimes. Our paper relates to a number of literatures. First, it contributes to a vast field of study that has analyzed the incomplete passthrough of exchange rate and trade policy shocks into export prices (Blonigen & Haynes, 2002; Olivei, 2002; Campa & Goldberg, 2005; Marazzi et al., 2005). In the past decade, substantial progress has been made in the study of both phenomena, albeit separately. Theoretically, the literature has focused on the various channels that lead to incomplete pass-through (Devereux & Engel, 2002): short-run nominal rigidities with prices sticky in the local currency (Gopinath & Rigobon, 2008); pricing-to-market (Atkeson & Burstein, 2008; Goldberg & Knetter, 1997; Krugman, 1987; Vigfusson, Sheets, & Gagnon, 2009; Yoshida, 2010); and import content of exports (Aksoy & Riyanto, 2000; Amiti et al., 2014; Goldberg & Campa, 2010). We add to this literature by identifying under which circumstances the symmetry between exchange rate pass-through and tariff pass-through breaks down. Empirically, a growing number of studies have started using firm-level data to circumvent methodological issues (notably endogeneity and estimation bias) related to estimating exchange rate pass-through with aggregate price data (Amiti et al., 2014; Berman, Martin, & Mayer, 2012). We follow this literature by using firm level data which allows us to treat exchange rate fluctuations as exogenous (Feinberg, 1996). Our paper is also related to recent studies that have investigated whether GVC trade behaves differently from regular trade. Yi (2003) proposes that GVC trade should be more sensitive to a world-wide decline in trade costs than regular trade since the same component crosses borders multiple times. Ma and Van Assche (2014) argue that GVCs allow firms to more easily shirk tariffs and use Chinese data to provide evidence of this. Cheung, Chinn, and Fujii (2010) and Thorbecke and Smith (2010) estimate that Chinese aggregate processing exports are more sensitive to foreign income changes than ordinary trade, but Gangnes, Ma, and Van Assche (2014) demonstrate that this is mostly due to a composition effect. We add to this literature by demonstrating that exchange rate pass-through is different between GVC trade and regular trade, but not tariff pass-through. Lastly, our study adds to the debate on trade policies' stance against China (Hu, Li, Yang, & Chao, 2016). The growing trade deficit of the United States versus China has fueled the widespread concern among U.S. policymakers that the increased exposure to Chinese imports is reducing U.S. welfare by leading to higher unemployment, lower labor force participation, and reduced wages (Autor, Dorn, & Hanson, 2016). Numerous U.S. policymakers and commentators have swiftly reacted by putting forward a myriad of trade policies aimed to limit the growth of Chinese exports, ranging from pressuring the country to appreciate their currency to treating China's alleged currency manipulation as a source of dumping that would permit the imposition of antidumping on Chinese imports (Staiger and Sykes, 2010). Our paper provides new insights to policymakers by illustrating that policies targeting China's exchange rate may not be as effective due to China's heavy reliance on imported inputs. The remainder of the paper is organized as follows. We present in Section 2 a pricing-to-market model to understand how changes in exchange rate and ad valorem tariffs affect a firm's pricing decisions. In Section 3, we describe the dataset on Chinese exporting firms, as well as the macroeconomic data used in this paper and some summary statistics. Section 4 reports our model specification. Section 4 provides estimations of exchange rate pass-through and antidumping pass-through for Chinese export prices. Section 5 concludes and discusses directions for future research. 2. Model Our model is a variant of Burstein and Gopinath (2014). Exporting firm i in industry k sets an optimal price, piknt, as the markup over its marginal costs when selling to destination country n in period t:
piknt = μiknt + mciknt + τknt ,
(1)
where lower case characters denote variables expressed in natural logarithm. The destination-country-specific mark-up μiknt depends on the price charged by the exporting firm i relative to the aggregate industry price level in the destination country n. That is, μiknt = μiknt[piknt − pknt]. We assume that μiknt is a decreasing function of the firm's price relative to the aggregate industry level price in ∂μ [.] the destination country. In other words, there is decreasing returns to scale so that Γiknt = − ∂ (p iknt− p ) > 0 . iknt knt A firm faces ad valorem trade cost τknt that is both industry and destination-country specific. The dollar marginal cost is given by mciknt[qiknt, wikt, miknt(ent)], where qiknt is the quantity sold by firm i in destination country n in period t, wikt summarizes those variables that impact the local production costs incurred by the firm such as wages and total factor productivity, and miknt(ent) captures the cost ∂m [.] of imported inputs which are denominated in foreign currency, where we assume ∂iknt = 1. We specify marginal cost in log so that: e nt
mciknt = Aiknt + κknt qiknt + αiknt miknt (ent ) + (1 − αiknt ) wikt
(2) 88
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
where Aiknt denotes a constant term, κknt is the elasticity of the marginal cost with respect to output (which we assume is common across firms) and αiknt < 1 is the cost share of the imported inputs in total marginal costs, respectively. Inserting Eqs. (2) into (1) and log-differentiating, we obtain that the log change in price Δpiknt can be approximated as:1
Δpiknt = −Γiknt Δpiknt + κknt Δqiknt + αiknt Δent + (1 − αiknt )Δwikt + Δτknt
(3)
Assume that log demand is given by qikn = q[pikn − pkn] + qkn where qkn denotes the log of aggregate demand in industry k in country n. Log differentiating,
Δqiknt = −εiknt Δpiknt where εiknt =
∂q [.] − ∂p iknt
Δpiknt =
(4)
> 0 is the price elasticity of demand. Combining Eqs. (3) and (4) and collecting terms, we obtain:
1 (αiknt miknt Δent + (1 − αiknt ) Δwikt + Δτknt ) 1 + Γiknt + Φiknt
(5)
where Φiknt = κikntεiknt > 0 is the partial elasticity of marginal cost with respect to the relative price. We can use Eq. (5) to investigate the elasticity of domestic export prices with respect to ad valorem tariffs:
Δpiknt 1 = ≤ 1. Δτknt 1 + Γiknt + Φiknt
(6)
Intuitively, an ad valorem tariff increase leads to an imperfect pass-through to domestic prices for two reasons. First, scale economies, captured by Γiknt. If there are decreasing returns to scale, Γiknt > 0, then a reduction in the quantity a firm produces will reduce the average cost of the firm and dampen its desire to raise prices. Second, if the elasticity of demand is high so that Φiknt > 0, then the demand decrease will decrease a firm's marginal cost and dampen its desire to raise prices. Note, however, that it is not a function of the import content share of exports. We state this as our first hypothesis: Hypothesis 1. All else equal, import content (αin) does not affect a firm's tariff pass-through. In a similar fashion, we can use Eq. (5) to investigate the elasticity of domestic export prices with respect to an exchange rate movement:
Δpiknt αiknt = Δent 1 + Γiknt + Φiknt
(7)
Intuitively, when the currency in the exporting country appreciates relative to the destination country, this raises the foreignmarket marginal costs of the exporting firm, all else equal. The sensitivity of foreign-market marginal costs to the exchange rate movement depends on αikn. Suppose the exporting firm sources all of its inputs domestically so that αiknt = 1, then there is a symmetry between ad valorem tariff pass-through and exchange rate pass-through. If αiknt < 1, however, some of its inputs are sourced globally and these inputs are priced in the foreign currency, whose foreign price is unaffected by the exchange rate movement. In this case, only fraction 1 − αiknt of the firm's marginal cost is affected and the exchange rate pass-through will be a declining function of the import content share. Hypothesis 2. All else equal, firms with a higher import content (αiknt) will have a smaller exchange rate pass-through. In the subsequent sections, we will empirically test the symmetry between tariff and exchange rate pass-through where we control for import content by taking into account variations across customs regimes. 3. Data and methods To empirically test our predictions, we take advantage of a firm-level dataset on trade for the period 2000–2006 that was compiled by the General Administration of Customs of the People's Republic of China, or China Customs Statistics in short. The database contains detailed information on the universe of trade transactions at the firm-industry-destination country-year level. It reports f.o.b. values (in US$) and quantities at the HS 8-digit level, thus allowing us to calculate unit values. We limit the products to those in the manufacturing sectors by omitting agriculture and specialty items such as rare coins and painting. Firm characteristics such as size and number of employees are not available in the data set and will be controlled for with firm fixed effects. The key goal of this paper is to investigate whether import content share affects the degree of exchange rate and tariff passthrough. Estimating a firm's import content share of exports is notoriously difficult since it requires information about a firm's domestic value added (Amiti et al., 2014). We circumvent this problem by taking advantage of the distinction between two broad customs regimes: processing trade (PT) and ordinary trade (OT). These two trade forms differ in terms of import tariff treatment and the ability of firms to sell on the domestic market:
• Under PT, firms enjoy the right of duty-free imports of intermediate goods and capital equipment that are used in their export 1 To simplify the exposition, we assume here that a tariff and exchange rate movement does not affect the aggregate industry level price in the destination country. Our key results remain unaffected if we relax this assumption.
89
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Fig. 1. Share of trade by custom regime in total export, 2000–2006. Source: Authors' calculation using the trade data from the China Custom Statistics.
processing activity, but face restrictions in selling to the domestic market.
• Under OT, firms need to pay import duties on imported inputs but can sell their output locally. Several seminal studies have demonstrated that these distinct customs characteristics ensure that firms in PT tend to have a far higher foreign content embodied in their exports than firms in OT. Using input-output data, Koopman et al. (2012) estimated that the foreign content share in 2002 amounted to 89.9% of processing exports compared to 10.7% in ordinary exports. In 2007, the foreign content share of processing exports was 89.5%, while it was 16.1% in ordinary exports. Using firm-level data, Kee and Tang (2016) estimated that the foreign content share hovered around 50% of Chinese processing exports, whereas it hovered around 10% for Chinese ordinary exports. In addition to the well-known distinction between OT and PT, a key distinction in the data is between two types of processing trade in China's Customs Statistics: pure assembly (PA) and import and assembly (IA). The main difference between the two customs sub-regimes, which result from different legal regimes to organize processing trade activities, lies in the allocation of control rights of the imported inputs (Feenstra & Hanson, 2005).
• In the PA regime, the manufacturer in China receives materials from a foreign principal free of charge. It then performs its •
manufacturing functions as per the requirements of the foreign principal and then exports the finished goods to the foreign principal. The manufacturer only gets compensated by the overseas party with a processing fee. In the IA regime, the manufacturer in China purchases and pays foreign exchange for imported materials free of import duty and VAT. It then performs manufacturing functions and exports the finished products to an overseas party (not necessarily the same as the input supplier) with the compensation of foreign exchange collected.
Manufacturers in IA and PA share the common feature that they perform manufacturing services to third-party companies which are located overseas. IA firms have a number of extra responsibilities over PA firms however. First, IA firms have the responsibility of searching and obtaining the imported materials themselves prior to conducting the manufacturing activity, whereas PA firms passively receive orders and materials from its foreign client and export all the processed goods to this foreign principal. Second, IA firms need to pay for the imported raw materials and parts, whereas PA firms receive them from their foreign trading partners free of charge. Third, to comply with the more complex regulations and approval processes of IA compared to PA, IA firms are required to make greater investments in inventory storages and management than PA firms (Feenstra & Hanson, 2005). These extra responsibilities imply that the foreign content share for PA trade is significantly higher than for IA trade (Manova & Yu, 2016). Fig. 1 presents the importance of PT in total exports. From 2000 to 2006 the share of PT in total exports exceeded 50%. Fig. 1 also shows that the share of PA in total exports decreased from 17.2% in 2000 to 9.8% in 2006. During the same period, the share of IA increased from 39.1% to 43.0%. We conduct our empirical analysis in three steps. In a first step, we investigate the determinants of exchange rate and tariff passthrough by estimating a generalized model of Eq. (5) for the period 2000–2006:
piknt = α 0 + α1 pikn, t − 1 + α2 ent + α3 ADknt + Xiknt β + δn + δk + δi + εiknt ,
(8)
where lower case characters denote variables expressed in natural logarithm.2 piknt is the f.o.b. export price of firm i in the importer's 2 Similar to Blonigen and Haynes (2002) and Marchand (2012) in estimating exchange-rate and tariff pass-through our specification is in log-level rather than first difference. Moreover our dataset has a large N and very small T (annual data from 2000 to 2006). With a large N (> 2 million observations) and small T (10 year period), Berman et al. (2012) also use a log-level estimation to analyze the exchange rate pass-through among French firms.
90
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
currency,3 in industry k, selling in destination market n, in year t; ent denotes the nominal spot exchange rate between the renminbi and the currency of destination country n expressed as units of country n's currency for one renminbi; ADknt is a dummy variable that takes on the value of one if there is an antidumping measure imposed by country n in industry k in year t; Xiknt is a vector of demandshifting control variables; δn, δk and δi are destination country, industry and firm fixed effects; and εiknt is a stochastic error. In a second step, we investigate variations in exchange rate and tariff pass-through across customs regimes (ordinary trade and processing trade). Finally, within processing trade, we distinguish the exchange rate pass-through between pure assembly and import and assembly. 3.1. Key variables of interest We have created two measures to capture exchange rate fluctuations: a nominal and a real exchange rate measure. 3.1.1. Nominal exchange rate Exchange rate series are nominal spot exchange rate expressed as units of the domestic currency for one renminbi. The data are obtained mostly from the International Monetary Fund's International Financial Statistics database, which we supplemented with data from several central banks. Because the database only provides data on the exchange rate between the domestic currencies and the US dollar, we constructed the exchange rate series by combining the aforementioned series with the spot exchange rate between the US dollar and the renminbi. For Eurozone countries, the exchange rate is expressed in terms of euros per renminbi. If some countries joined the Eurozone later than 2000 (such as Estonia and Slovenia), we use the fixed rate between these countries' national currencies and the euro to expand the series prior to their adoption of the euro. 3.1.2. Real exchange rate The real exchange rate is calculated as the nominal spot exchange rate multiple by the ratio of the Chinese PPI to the destination PPI. Our analysis requires data on time-varying country-product level tariffs which are difficult to obtain for China. To operationalize this, we have opted to rely antidumping measures which are country-product specific and exhibit sufficient variation over time. 3.1.3. Antidumping As our measure of a country-product-specific trade policy shock, we use antidumping cases against China as identified in the World Bank's Global Antidumping Database (GAD) (Bown, 2009).4 The benefit of using antidumping as a measure for a countryspecific trade policy shock is that it is generally imposed by a country on firms of a specific country in a specific industry. The GAD database has detailed information on each antidumping case, such as product information (6-digit HS codes), the investigating country, the target country, the preliminary determination date and the year it was revoked. For our analysis, we collect information on all antidumping cases against China during the period 2000–2006. We match the GAD data with the Chinese Customs data at the HS-6 digit level, the most disaggregated level at which the two datasets are comparable (Ma & Van Assche, 2014). Finally, we create a dummy ADknt which takes on the value of one if there is an antidumping measure imposed by country n in industry k in year t. It is important to point out that while the use of the dummy variable ADknt makes it impossible to identify the size of tariff passthrough, given that it is country-product-specific it allows us to test our hypothesis that the tariff pass-through for a specific product does not vary across customs regimes with varying import content. An incomplete exchange rate and tariff pass-through implies that the exporting firm reacts to a currency appreciation or to an antidumping measure by reducing its f.o.b. price expressed in the domestic currency. In our estimation equation, we would thus find evidence of incomplete exchange rate and tariff pass-through if both α2 < 0 and α3 < 0. The size of the coefficient also matters for exchange rate pass-through: a coefficient closer to zero would indicate almost complete exchange rate. It is difficult to interpret the size of the coefficient α3 due to our use of the dummy variable ADknt. 3.2. Control variables We include a lag of the export price to account for serial correlation in pricing decisions (Bussière, Chiaie, & Peltonen, 2014; Zhou & Kim, 2011). We further include a set of control variables that prior literature has identified as important drivers of exchange rate and tariff pass-through. We proxy trends in the production costs in China and the destination country with the natural log of the Chinese and destination market producer price index: ln(Chinese PPI) and ln(destination PPI). We also include the price of foreign competitors in the destination market, ln(competing price), given by the unit value of a country's import at the 6-digit HS level from all countries excluding China. To capture the degree of import competition, we include another control variable, Ln(Chinese import share), which measures the share of Chinese imports in total imports in industry k in destination country n. As a robustness check, we 3 While the Chinese Customs data include all countries in which China exports to, due to data limitation in the control variables such as Producer Price Index, the following countries are included in the estimation: Australia, Austria, Belgium, Bulgaria, Canada, Croatia, Denmark, Estonia, France, Germany, India, Indonesia, Ireland, Italy, Japan, Republic of Korea, Luxembourg, Malaysia, Mexico, Netherlands, Philippines, Poland, Portugal, Singapore, Spain, Sweden, Taiwan, Thailand, United Kingdom, United States, and Vietnam. 4 Antidumping captures country-product-specific shocks that are imposed onto China's exports during the period 2000–2006 relative to the lack of variation in tariff levels.
91
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Table 1 Data description. Series
Definition
Source
Export price
Firms' export unit value price in destination market expressed in currency of the destination market Nominal bilateral spot exchange rate, expressed as units of destination market's currency per renminbi Equals one if antidumping duties are imposed on Chinese firms from a specific industry Producer price index, 2010 = 100 Producer price index, 2010 = 100
Chinese Custom Statistics
Exchange rate Anti-dumping dummy variable Chinese PPI Destination market PPI Export price of foreign competitors Share of Chinese imports in total imports in destination market Herfindahl-Hirschman Index for imports Real GDP of destination market Real GDP per capita of the destination markets
Non-Chinese firms' export unit value price in destination market expressed in currency of the destination market Share of Chinese imports in total imports in destination market Measure of concentration among importing firms. It is computed as the sum of the squared import shares Real GDP expressed in currency of the destination market Real GDP per capita expressed in currency of the destination market
IFS, European Central Bank, Bank of Canada, Forex.com, Bank of Estonia Chinese Custom Statistics OECD, OECD, Singapore Department of Statistics, Malaysia Department of Statistics Chinese Custom Statistics Chinese Custom Statistics Authors' computation based on Chinese Custom Statistics OECD OECD
substitute a Herfindahl-Hirschman index using import share for each country in destination country n at the industry level to capture import competition at the 6-digit HS level. Ln(destination real GDP) is included as our measure of market size and business cycle, and per capita income respectively (Li, Juanyi Jenny, & Zhao, 2014, Li, Ma, & Yuan, 2015, Manova and Zhang, 2012). Finally, we include fixed effects for firms, countries, and industries. As noted by Manova and Zhang (2012) the fixed effects control for systematic differences in unit values across firms, product characteristics, and destination countries, including industry and producer costs that are associated with changes in production technology or input costs. In other words, our analysis investigates whether within a firm-product-destination combination the tariff and exchange rate pass-through varies across customs regimes. The series and their sources are described in Table 1, while Table 2 provides data summary statistics. 3.3. Variations across customs regimes To investigate variations in exchange rate and tariff pass-through across customs regimes, we create the dummy variable PTr which equals 1 if trade occurs under PT, and 0 otherwise. We then interact PTr with our two variables of interest to investigate if tariff and exchange rate pass-through significantly differs between PT and OT:
piknrt = α 0 + α1 piknr , t − 1 + α2 ent + α3 ADknt + α4 PTr ∗ent + α5 PTr ∗ADknt + α6 PTr + Xiknt β + δk + δn + δt + εiknrt ,
(9)
There will be evidence of Hypothesis 1 if α5 = 0. In that case, the tariff pass-through is not significantly different between the PT Table 2 Summary statistics. Variable By firm-product-country-year Local price Competing price Chinese import share Herfilndahl-Hirschman Index Anti-dumping dummy variable Ordinary Trade Dummy Variable Pure-Assembly Dummy Variable Import-and-Assembly Dummy Variable Chinese PPI by Industry By country-year Nominal exchange rate Real exchange rate Destination PPI Chinese PPI Destination real GDP/Capita in US $ Destination real GDP in US $
Obs
Mean
Std. dev
Min
Max
12,931,467 503,613 455,802 1,071,351 12,931,467 12,931,467 12,931,467 12,931,467 28,545
2,904,153 3,045,247 17.76 4293 0.02 0.84 0.10 0.06 102.11
1.92E + 09 1.26E + 08 23.43 2515 0.11 0.37 0.30 0.24 4.81
2.39E-11 3.00E-08 5.00E-07 350 0 0 0 0 93.08
1.95E + 12 4.91E + 10 100 10,000 1 1 1 1 122.50
211 211 217 7 217 217
103.55 202.09 103.04 82.80 23,730 1.10E + 12
382.01 748.65 4.22 5.09 18,067 2.26E + 12
0.07 0.08 93.40 77.50 433 3.85E + 09
2005.48 3727.64 118.90 90.90 83,576 1.34E + 13
92
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Table 3 Benchmark specification.
ln(local pricet − 1) ln(ex_rate)
(1)
(2)
(3)
0.809*** [0.001] − 0.028*** [0.003]
0.809*** [0.001] − 0.027*** [0.003] − 0.002*** [0.000]
0.809*** [0.001] − 0.028*** [0.003]
ln(ex_rate) ∗ PT ln(ex_rate) ∗ PA ln(ex_rate) ∗ IA Antidumping
− 0.002*** [0.000] 0.0004 [0.001] − 0.012*** [0.003]
− 0.012*** [0.003] − 0.005 [0.005]
antidumping ∗ PT antidumping ∗ PA antidumping ∗ IA ln(Chinese PPI) ln(destination PPI) ln(competing price) ln(destination real GDP) ln(Chinese import share)
− 0.006 [0.005] − 0.004 [0.008] 0.525*** [0.013] − 0.000 [0.006] 0.073*** [0.001] 0.174*** [0.003] − 0.006*** [0.000]
0.526*** [0.013] 0.0001 [0.981] 0.074*** [0.001] 0.174*** [0.003] − 0.006*** [0.000] 0.036*** [0.002]
PT PA IA Firm FE Country FE Industry FE Observations R-squared
− 0.002*** [0.0003] 0.0004 [0.000] Yes Yes Yes 4,875,525 0.955
Yes Yes Yes 4,864,456 0.955
− 0.002*** [0.000] 0.0004 [0.001] − 0.012*** [0.003]
− 0.006 [0.005] − 0.004 [0.008] 0.525*** [0.013] − 0.000 [0.006] 0.073*** [0.001] 0.174*** [0.003] − 0.006*** [0.000]
− 0.002*** [0.0003] 0.0004 [0.000] Yes Yes Yes 4,875,525 0.955
than OT. There will be evidence of Hypothesis 2 if α4 < 0. This would indicate that the exchange rate pass-through is significantly smaller in PT than OT. In a final step, we further divide PT into the two sub-regimes PA and IA. Specifically, we estimate:
piknrt = α 0 + α1 piknr , t − 1 + α2 ent + α3 ADknt + α4 PAr ∗ent + α5 PAr ∗ADknt + α6 PAr + α 7 IAr ∗ent + α8 IAr ∗ADknt + α 9 IAr + Xiknt β (10)
+ δk + δn + δt + εiknrt ,
Hypothesis 1 suggests that α5 = α8 = 0. In that case, the tariff pass-through is not affected by the declining import content under PA-type processing trade, IA-type processing trade and ordinary trade. There will be evidence of Hypothesis 2 if α4 < α7 < 0. This would indicate that the exchange rate pass-through is significantly smaller in PA-type processing trade than in IA-type processing trade and ordinary trade in that order. 4. Results In Table 3 we present our results with industry, firm, and country fixed effects. In column (1), we present our benchmark model which estimates Eq. (8) pooled across customs regimes. In column (2), we provide the regression results for Eq. (9). The results of column (1) show that the ERPT is nearly complete. Firms react to an appreciation of the renminbi by passing 97% of the rise in the exchange rate, leading to only a minor decrease in the export price. The negative and statistically significant coefficient on the antidumping dummy suggests that exporters try to mitigate the negative demand impact of the new duties by reducing their export price. The coefficients on the control variables are almost all statistically significant and with the expected sign with the exception of the PPI of the destination markets. An increase in the export price is also expected following a rise in import price of nonChinese foreign competitors in the destination country and following an increase in the destination market's income through a higher 93
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Table 4 Robustness check.
ln(local pricet − 1) ln(ex_rate) ln(ex_rate) ∗ PA ln(ex_rate) ∗ IA
(1)
(2)
(3)
0.810*** [0.001] −0.028*** [0.003] −0.002*** [0.000] 0.001 [0.001]
0.810*** [0.001] − 0.024*** [0.003] − 0.002*** [0.001] 0.001 [0.001]
0.810*** [0.001]
ln(real ex_rate) ln(real ex_rate) ∗ PA ln(real ex_rate) ∗ IA Antidumping antidumping ∗ PA antidumping ∗ IA ln(Chinese PPI) ln(destination PPI) ln(competing price) ln(destination real GDP) ln(HHI) PA IA
−0.014*** [0.003] −0.005 [0.005] −0.003 [0.008] 0.524*** [0.013] −0.005 [0.006] 0.074*** [0.001] 0.174*** [0.003] −0.006*** [0.000] 0.038*** [0.002] 0.026*** [0.002]
− 0.014*** [0.003] − 0.006 [0.005] − 0.003 [0.008] 0.528*** [0.013] − 0.003 [0.006] 0.074*** [0.001] 0.169*** [0.003] − 0.011*** [0.001] 0.038*** [0.002] 0.026*** [0.002] 0.007*** [0.001] Yes Yes Yes 4,898,352 0.956
WTO Firm FE Country FE Industry FE Observations R-squared
Yes Yes Yes 4,898,352 0.956
− 0.022*** [0.002] − 0.002*** [0.000] 0.001 [0.001] − 0.012*** [0.003] − 0.005 [0.005] − 0.002 [0.008]
0.074*** [0.001] 0.167*** [0.002] − 0.011*** [0.001] 0.038*** [0.002] 0.026*** [0.002]
Yes Yes Yes 4,898,352 0.956
demand. Owing to increased competition among Chinese exporting firms, a greater share of Chinese imports in total imports in industry j in destination country k decreases the Chinese export price. The empirical test of Hypotheses 1 and 2 are presented in column (2). To begin with, the statistically insignificant coefficient on the interaction between a tariff or antidumping duties and processing trade substantiates Hypothesis 1. Antidumping duties do not affect the pricing strategy of processing trade firms differently from ordinary trade firms within the same sector and towards the same destination country. Moreover, the negative and statistically significant coefficient on the interaction term ln(ex_rate) ∗ PT provides evidence in support of Hypothesis 2. In particular, the result suggests that PT firms absorb the same renminbi appreciation by further decreasing the export price than OT firms, albeit by a marginal amount. Taken together, the results suggest that exchange rate passthrough is smaller in customs regimes with higher import content share, whereas tariff pass-through is invariant across customs regimes with different import content shares. We further disaggregate the processing trade regime to account for the allocation of control rights of the imported inputs. In column (3), we test for differences in the exchange rate and antidumping pass-through of both types of processing trade firms, namely PA firms and IA firms.5 In support of Hypothesis 1, the coefficients on the interaction terms antidumping ∗ PA and antidumping ∗ IA are statistically insignificant, suggesting that the antidumping pass-through is similar across the three customs regimes despite large differences in import content share. However, the coefficients on the interaction between these two types of processing trade with the exchange rate term indicates that the Hypothesis 2 holds only for pure-assembly firms.
5 The larger number of observations in column 2 stem from the fact that some firms operating both as input-control and pure assembly would have been counted only once in column 1 but twice in column 2.
94
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
We have investigated a number of alternative specifications to check the robustness of our results. First, we have re-estimated Eq. (10) with the inclusion of the Herfindahl-Hirschman Index to capture import competition at the 6-digit HS level.6 The results, presented in column (1) of Table 4 suggest that our results are consistent with the benchmark specification. Second, we have in column (2) accounted for China's accession to the WTO by including a dummy variable that equals to 1 from 2002 forward.7 Again, our key results are unchanged. Finally, we have in column (3) used real exchange rates instead of nominal exchange rates and found that our results remain consistent with our predictions. 5. Conclusion In this paper we examine whether a firm's import content share differentially affects the degree of tariff and exchange rate passthrough into its export prices. Our pricing-to-market model suggests that a firm's import content share negatively affects the degree of exchange rate pass-through but does not affect the degree of tariff pass-through. Using firm-level data for Chinese exporting firms, we find evidence of an almost complete exchange rate pass-through (around 97%), which is in line with the baseline estimation in the literature (Kiliç, 2016; Li et al., 2015). As expected, when we distinguish firms by their trade regime, processing-trade firms, especially pure-assembly firms which tend to have higher import-content share, have a slightly lower ERPT than ordinary trade firms. This almost-complete ERPT suggests that an appreciation of the renminbi would translate into higher export prices for Chinese products. Such an appreciation might not necessarily lead to a reduction in China's exports and thus a decrease in its trade surplus with other countries for several reasons. First, large ERPT for export prices might not translate into similar ERPT for import prices, which Campa and Goldberg (2005) refer to as the “exchange rate disconnect” puzzle. They suggest that this disconnect might be related to distribution costs in the destination markets, as well as to other local value-added components. Second, as noted by Xing (2012), most of China's trade surplus originates from processing trade firms. Whether a renminbi appreciation affects Chinese exports depends on the inputs' origin. One would expect the renminbi appreciation to have a more significant effect on the exports of firms whose value added comes mostly from China than of the exports of firms whose manufacturing costs comprise mostly components made in high-wage countries, such as Japan and South Korea. The growing share of domestic content in Chinese exports (Thorbecke, 2013) implies that the renminbi will have an increasing role to play in addressing China's trade imbalances. References Aksoy, Y., & Riyanto, Y. E. (2000). Exchange rate pass-through in vertically related markets. Review of International Economics, 8(2), 235–251. Amiti, M., Itskhoki, O., & Konings, J. (2014). Importers, exporters, and exchange rate disconnect. American Economic Review, 104(7), 1942–1978. Atkeson, A., & Burstein, A. (2008). Pricing-to-market, trade costs, and international relative prices. American Economic Review, 98(5), 1998–2031. Autor, D. H., Dorn, D., & Hanson, G. H. (2016). The China shock: Learning from labor market adjustment to large changes in trade. Annual Review of Economics, 8(1). Berman, N., Martin, P., & Mayer, T. (2012). How do different exporters react to exchange rate changes? The Quarterly Journal of Economics, 127(1), 437–492. Blonigen, B., & Haynes, S. E. (2002). Antidumping investigations and the pass-through of antidumping duties and exchange rates. American Economic Review, 92(4), 1044–1061. Bown, C. P. (2009). Global antidumping database (current version 5.0) Brandeis University mimeo. Burstein, A., & Gopinath, G. (2014). “International prices and exchange rates.” 4:391–451. Handbook of international economics(4th ed.). 4. Handbook of international economics (pp. 391–451). Elsevier. Bussière, M., Chiaie, S. D., & Peltonen, T. (2014). Exchange rate pass-through in the global economy: The role of emerging market economies. IMF Economic Review, 62, 146–178. Campa, J. M., & Goldberg, L. S. (2005). Exchange rate pass-through into import prices. The Review of Economics and Statistics, 87, 679–690. Cheung, Y.-W., Chinn, M. D., & Fujii, E. (2010). China's current account and exchange rate. China's growing role in world trade (pp. 231–271). University of Chicago Press. Devereux, M. B., & Engel, C. (2002). Exchange rate pass-through, exchange rate volatility, and exchange rate disconnect. Journal of Monetary Economics, 49(5), 913–940. Feenstra, R. C. (1989). Symmetric pass-through of tariffs and exchange rates under imperfect competition: An empirical test. Journal of International Economics, 27(1), 25–45. Feenstra, R. C., & Hanson, G. H. (2005). Ownership and control in outsourcing to China: Estimating the property-rights theory of the firm. Quarterly Journal of Economics, 120, 729–761 (2,May). Feinberg, R. M. (1996). A simultaneous analysis of exchange rate pass-through into prices of imperfectly substitutable domestic and import goods. International Review of Economics and Finance, 5(4), 407–416. Gangnes, B. S., Ma, A. C., & Van Assche, A. (2014). Global value chains and trade elasticities. Economics Letters, 124(3), 482–486. Goldberg, L. S., & Campa, J. M. (2010). The sensitivity of the CPI to exchange rates: Distribution margins, imported inputs, and trade exposure. The Review of Economics and Statistics, 92(2), 392–407. Goldberg, P. K., & Knetter, M. M. (1997). Goods prices and exchange rates: What have we learned? Journal of Economic Literature, 35, 1243–1272. Gopinath, G., & Rigobon, R. (2008). Sticky borders. Quarterly Journal of Economics, 123(2), 531–575. Hu, M., Li, Y., Yang, J., & Chao, C. C. (2016). Actual intervention and verbal intervention in the Chinese RMB exchange rate. International Review of Economics and Finance, 43, 499–508. Johnson, R. C. (2014). Five facts about value-added exports and implications for macroeconomics and trade research. The Journal of Economic Perspectives, 28(2), 119–142. Kee, H. L., & Tang, H. (2016). Domestic value added in exports: Theory and firm evidence from China. American Economic Review, 106(6), 1402–1436. Kiliç, R. (2016). Regime-dependent exchange-rate pass-through to import prices. International Review of Economics and Finance, 41, 295–308. Koopman, R., Wang, Z., & Wei, S.-J. (2012). Estimating domestic content in exports when processing trade is pervasive. Journal of Development Economics, 99(1), 178–189. Krugman, P. (1987). Pricing to market when the exchange rate changes. In S. Arndt, & J. D. Richardson (Eds.). Real-financial linkages among open economies (pp. 49–70). Cambridge: MIT Press.
6 7
We are grateful to an anonymous reviewer for the suggestion of the inclusion of the HHI and the WTO dummy. China entered the WTO in December 2001.
95
China Economic Review 46 (2017) 87–96
F. Bouvet et al.
Li, H., Ma, H., & Yuan, X. (2015). How do exchange rate movements affect Chinese exports ? A firm-level investigation. Journal of International Economics, 97(1), 148–161. Li, Y. A., Juanyi Jenny, X., & Zhao, C. (2014). Import response to exchange rate fluctuation: Evidence from Chinese firms. Forum for research in empirical international trade working papers series, number 623. Ma, A. C., & Van Assche, A. (2014). Vertical specialization, tariff shirking, and trade. In B. Ferrarini, & D. Hummels (Vol. Eds.), Asia and global production networks. 2014. Asia and global production networks (pp. 148–179). Massachusetts: Edward Elgar Publishing, Inc (Nov.). Manova, K., & Yu, Z. (2016). How firms export: Processing vs. ordinary trade with financial frictions. Journal of International Economics, 100, 120–137. Manova, K., & Zhang, Z. (2012). Export prices across firms and destinations. The Quarterly Journal of Economics, 127(1), 379–436. Marazzi, M., Sheets, N., Vigfusson, R. J., Faust, J., Gagnon, J., Marquez, J., ... Rogers, J. (2005). Exchange rate pass-through to U.S. import prices: Some new evidence. Board of governors of the federal reserve system international finance discussion paper 2005-833. Marchand, B. U. (2012). Tariff pass-through and the distributional effects of trade liberalization. Journal of Development Economics, 99(2), 265–281. Staiger, R. W., & Sykes, A. O. (2010). ‘Currency manipulation’and world trade. World Trade Review, 9(4), 583–627. Thorbecke, W. (2013). Investigating China's disaggregated processed exports: Evidence that both the RMB and exchange rates in supply chain countries matter. The World Economy, 6(10), 1245–1260. Thorbecke, W., & Smith, G. (2010). How would an appreciation of the renminbi and other east Asian currencies affect China's exports? Review of International Economics, 18(1), 95–108. Vigfusson, R. J., Sheets, N., & Gagnon, J. (2009). Exchange rate pass-through to export prices: Assessing cross-country evidence. Review of International Economics, 17(1), 17–33. Xing, Y. (2012). Processing trade, exchange rates and China's bilateral trade balances. Journal of Asian Economics, 23(5), 540–547. Yi, K.-M. (2003). Can vertical specialization explain the growth of world trade? Journal of Political Economy, 111(1), 52–102. Yoshida, Y. (2010). New evidence for exchange rate pass-through: Disaggregated trade data from local ports. International Review of Economics and Finance, 19(1), 3–12. Zhou, J., & Kim, D. (2011). China's share of U.S. imports and exchange rate pass-through to U.S. import prices. China & World Economy, 19(3), 87–104.
96