Domestic Value Added in Exports Theory and Firm Evidence from China

Transcription

1 Domestic Value Added in Exports Theory and Firm Evidence from China Hiau Looi Kee (World Bank) Heiwai Tang (Johns Hopkins) IMF Research Department Seminar Dec 16, 2015

2 Designed by Apple in California, Assembled in China Inside an ipad 3 Source: Antras (2012) Lecture

3 Downward Trend in Domestic Value Added in Exports across the World Figure 1: Ratio of Value-Added to Gross Exports for the World with ROW without ROW Source: Johnson and Noguera (2014)

4 China has recently defied the global trend Figure A7: DVAR of Aggregate Exports (Single-industry Firms) DVAR Year mdvar_agg 95 c.i. (lower bound) 95 c.i. (upper bound) Also documented by Koopman, Wang and Wei (2012) for 2002 and 2007.

5 What caused China to defy the global trend? Several possible answers to this question with conflicting implications.

6 What caused China to defy the global trend? Several possible answers to this question with conflicting implications. Changing composition of Chinese exports (towards the industries with high domestic content).

7 What caused China to defy the global trend? Several possible answers to this question with conflicting implications. Changing composition of Chinese exports (towards the industries with high domestic content). Increasing domestic production costs, which would imply that the country has become less competitive.

8 What caused China to defy the global trend? Several possible answers to this question with conflicting implications. Changing composition of Chinese exports (towards the industries with high domestic content). Increasing domestic production costs, which would imply that the country has become less competitive. Gradual substitution of domestic for imported materials by its exporters.

9 What is this paper about? 1. Develop methodologies to use customs transaction-level data merged with firm survey data to measure and analyze a country s ratio of domestic value added in exports to gross exports (DVAR) at the firm, industry, and national level. 2. Provide a detailed description of the trend, the pattern, and the mechanism of the rising DVAR of Chinese exports ( ). 3. Develop a theoretical model to quantitatively assess the determinants of China s rising DVAR.

10 Different from the standard approach Existing literature measures industry and aggregate DVARs using input-output (IO) tables (Hummels, Ishii and Yi (2001); Antras, Chor, Fally, and Hillberry (2012); Johnson and Noguera (2012a, 2012b) and Koopman, Wang, and Wei (2012, 2013)). Advantages: capture IO linkages within and across countries. However, the presence of firm heterogeneity may result in significant aggregation biases in the estimates of the DVAR. Advantages of our ground-up approach: Embrace firm heterogeneity. Permit statistical tests on the rising trend, using bootstrapped standard errors for our aggregate estimates. Examine several micro mechanisms, such as changes in firms export composition, production costs and material shares. Tang, Wang, and Wang (2015): constrained optimization techniques to incorporate firm heterogeneity when using standard IO tables to portray the domestic segment of GVC.

11 Main Findings Over , the DVAR of Chinese processing exporters gradually increased from 65% to 70%. Most of the increase is due to processing exporters (9 percentage-point increase) substitute domestic materials for imported materials. This material substitution is caused by an increasing supply and decreasing prices of domestic input varieties triggered by decreasing input tariffs facing the upstream sectors and increasing FDI in the downstream sectors. The rise in DVAR is NOT due to reallocation of resources across industries nor firm entry and exit.

12 Road Map Data Methodology of using firm and customs transaction data to measure DVAR Facts Reduced-form firm-level evidence Simple model to identify the determinants of firm DVAR Quantitatively assess the contribution of each determinant to the rise in DVAR

13 Data Data set 1 : the universe of Chinese import and export transactions in each month between 2000 and Data on imports and exports (in USD) at the HS 6-digit level from a firm to/from each country. Data set 2 : firm-level manufacturing survey data from China s National Bureau of Statistics (NBS). Covers all state-owned firms and all private firms with sales > 5 million RMB (about 600,000 USD during the sample period). Balanced-sheet variables: firm ownership, output, value added, exports, employment, original value of fixed asset, and intermediate inputs.

14 Methodology Identities A firm s (i) total revenue: PY i π i + wl i + rk i + P D M D i + P I M I i. Domestic materials P D M D i may embody foreign content (δ F i ). P D M D i δ F i + q D i Imported materials P I M I i may embody domestic content (δ D i ). P I M I i δ D i + q F i.

15 Methodology Benefits of focusing on processing trade By law, processing firms need to export all its output. DVA i π i + wl i + rk i + qi D + δi D ( = EXP i IMP i + δi D δi F + δi K For processing exporters, we only need to remove foreign content in domestic materials, δ F i. ). DVAR i DVA i EXP i = 1 PI M I i PY i δf i EXP i For each industry-year, impute the estimates, use the growth rate of the number of non-processing importers in the upstream sectors and the estimated δf i EXP i from Koopman, Wang, and Wei (2012) for

16 Caveat: Indirect Importing Even under the processing regime, some firms import materials and sell them domestically, i.e., Carry-Along Trade (Bernard et al. (2012)). excessive importers and excessive exporters. Solutions: Merge customs data with manufacturing firm survey data to identify excessive importers and excessive exporters. We focus on a subset of single-industry processing exporters that have their EXP IMP bounded between the two cutoffs: ( ) IMP OT IMP EXP (25) EXP PD M D + P I M I, EXP ( ) OT where DVAR(25) OT = 1 IMP EXP is the 25 percentile of the DVAR (25) of ordinary exporters in the same industry. About the merged sample Multi-industry firms

17 DVAR of Processing Exports: 9 ppt increase igure 1: DVAR of Processing Exports ( ), with 95% (Bootstrapped) Con dence ntervals DVAR Year Measured DVAR 95 c.i. (lower bound) 95 c.i. (upper bound) Note: Bootstrapped Sample Comparing our numbers with Koopman, Wang, Wei (2012) Table 1: Issues and Assumptions or Solutions

18 2: DVAR in Trend Chinese ( ) Processing by Industry Exports with 95% by(bootstrapped) Industry Con dence 04:beverages & spirit 06:chemical products 07:plastics & rubber 08:raw hides & skins :wood & articles * 10:pulp of wood 11:textiles 12:footwear & headgear, etc. 13:stone, plaster, cement, etc. 14:precious metals 15:base metals * 16:machinery, mechnical & elec eqmt 17:vehicles & aircrafts 18:optical, photographic, etc. 20:misc manufacturing year Dashed lines = 95% confidence interval. * = industries with average DVAR lower in 2007 than 2000.

19 The rise in DVAR is driven by within-sector increases. ( ) DVAR t = Σ j Iit w jt ( DVAR jt ) + Σ j Iit DVARjt ( wjt ), }{{}}{{} within between Figure 3: Decomposing the DVAR Growth into Within- and Between-industry Grow year within total change between

20 Extension to Non-Processing and Aggregate Exports The methodology developed above is suitable for pure exporters who export all their output (e.g., those engaged in global value chains in the form of processing trade). However, non-processing exporters both export and sell domestically. Extend our methodology to measure the DVAR of the non-processing exporters by making one proportionality assumption at the firm level: the allocation of the firm s inputs to the production for exports is proportional to the share of exports in total sales The DVA and DVAR of a non-processing exporter are: ( DVA O i = EXP i IMP i δi K + δ F i ) ( EXP i PY i ) ; DVARi O = DVA i = 1 IMP i δi K EXP i PY i + δ F i

21 DVAR in Overall Exports Figure 5: DVAR of China s Aggregate (Processing + Ordinary) Exports year dvar_agg Filter + DVAR < 25% DVAR(Ord) *Filter: m>=imp & exp>=imp

22 Reasons for the rising DVAR? The increase in DVAR is not due to the reallocation of resource between industries. Within a sector, high DVAR firms could increase sales more, while low DVAR firms may exit. On the other hand, it could also be a within-firm upgrading phenomenon, due to 1. Rising production costs; 2. Firms substitution imported materials with domestic materials China moved up the global production chain. What drives the substitution?

23 Dependent variable: DVAR of firm exports Table 4: Dependent Variable: DVARThe it = Ratio β i + of βdomestic t + β X Value X it Added + ɛ it, in Exports to Gross Exports (DVAR) (1) (2) (3) (4) (5) (6) Sample All All Dom private Foreign Multiple Ind Unfiltered β *** *** *** *** *** (0.007) (0.006) (0.080) (0.006) (0.005) (0.005) β *** *** *** *** *** (0.004) (0.004) (0.106) (0.004) (0.006) (0.004) β *** *** 0.190** *** *** *** (0.008) (0.008) (0.078) (0.008) (0.005) (0.005) β *** *** *** *** *** (0.008) (0.011) (0.127) (0.011) (0.005) (0.010) β *** *** *** 0.117*** 0.101*** (0.007) (0.009) (0.124) (0.008) (0.005) (0.010) β *** 0.136*** 0.257* 0.136*** 0.146*** 0.133*** (0.010) (0.011) (0.133) (0.012) (0.005) (0.010) β *** 0.147*** 0.300** 0.146*** 0.161*** 0.150*** (0.013) (0.017) (0.140) (0.016) (0.006) (0.014) ( ) P D M D +P I M I P Y it *** *** ** *** *** ( (0.007) (0.008) (0.060) (0.010) (0.006) (0.004) wl ) P Y it (0.016) (0.155) (0.017) (0.009) (0.006) N R-sq Notes: Firm and year fixed effects are always included. Data set: merged NBS-customs data. Columns (1) and (2) use Firm and year fixed effects the whole included. sample; columns Bootstrapped (3) and (4) standard include onlyerrors domestic areprivate in parentheses. and foreign-invested * p < firms, 0.10; respectively. ** p < 0.05; *** Column (5) includes firms that operate in multiple industries as well. Column (6) includes single-industry firms p < that do not satisfy our rules to filter firms that engage in indirect trade. Bootstrapped standard errors, clustered

24 Dependent variable: Imports/ Total Materials Table 5: Dependent Variable: Share of imports in total materials Sample All Dom private Foreign Multiple Ind δ ** ** *** (0.010) (0.047) (0.011) (0.006) δ *** 0.137** *** *** (0.006) (0.062) (0.007) (0.007) δ *** *** *** (0.007) (0.067) (0.007) (0.007) δ *** *** *** (0.008) (0.061) (0.008) (0.006) δ *** *** *** (0.010) (0.066) (0.009) (0.006) δ *** *** *** (0.011) (0.081) (0.009) (0.007) δ *** *** *** ( (0.017) (0.086) (0.013) (0.007) wl ) ** P Y it (0.042) (0.190) (0.039) (0.042) ln (K/L) it (0.003) (0.030) (0.003) (0.003) N R-sq Note: Firm and year fixed effects are always included. Data set: merged NBS and customs data. Column Firm(1) anduses yearthe fixed whole effects sample; included. columns Bootstrapped (2) and (3) standard include only errors domestic are in parentheses. private and foreign-invested * p < 0.10; ** p firms, < 0.05; *** p < respectively. Column (4) includes firms that operate in multiple industries as well. Bootstrapped standard errors, clustered at the industry level, are reported in parentheses. * p<0.10; ** p<0.05; *** p<0.01.

25 Dependent variable: ln(nb. import variety) Table 6: Dependent Variable: ln(number of import varieties) Sample All Dom private Foreign Multiple Ind γ *** * *** *** (0.016) (0.124) (0.018) (0.013) γ *** *** *** (0.016) (0.284) (0.016) (0.016) γ *** *** *** (0.029) (0.419) (0.026) (0.016) γ *** *** *** (0.039) (0.352) (0.035) (0.015) γ *** *** *** (0.046) (0.367) (0.045) (0.016) γ *** *** *** (0.054) (0.336) (0.054) (0.019) γ *** *** *** ( ) (0.090) (0.345) (0.081) (0.020) P D M D +P I M I P Y it ( wl ) P Y it (0.025) (0.332) (0.019) (0.020) (0.038) (1.033) (0.054) (0.059) N R-sq Note: Firm and year fixed effects are always included. Data set: merged NBS and customs data. Column Firm and (1) year usesfixed the whole effects sample; included. columns Bootstrapped (2) and (3) standard include only errors domestic are in private parentheses. and foreign-invested * p < 0.10; ** firms, p < 0.05; *** p < respectively. Column (4) includes firms that operate in multiple industries as well. Bootstrapped standard errors, clustered at the industry level, are reported in parentheses. * p<0.10; ** p<0.05; *** p<0.01.

26 Dependent variable: ln(nb. export variety) Table 7: Dependent Variable: ln(number of export varieties) Sample All Dom private Foreign Multiple Ind θ * (0.022) (0.223) (0.012) (0.021) θ ** *** (0.042) (0.221) (0.029) (0.020) θ ** *** 0.130*** (0.049) (0.318) (0.035) (0.018) θ ** 0.598** 0.126*** 0.161*** (0.056) (0.267) (0.039) (0.016) θ *** 0.821*** 0.210*** 0.236*** (0.040) (0.310) (0.029) (0.019) θ *** 0.945*** 0.283*** 0.316*** (0.050) (0.316) (0.033) (0.017) θ *** 1.086*** 0.267*** 0.306*** ( ) (0.046) (0.338) (0.030) (0.022) P D M D +P I M I P Y it ( wl ) P Y it (0.018) (0.266) (0.017) (0.018) * * (0.059) (0.726) (0.028) (0.031) N R-sq Note: Firm and year fixed effects are always included. Data set: merged NBS and customs data. Column Firm and (1) year uses the fixedwhole effects sample; included. columns Bootstrapped (2) and (3) standard include only errors domestic are in parentheses. private and foreign-invested * p < 0.10; ** p firms, < 0.05; *** p < respectively. Column (4) includes firms that operate in multiple industries as well. Bootstrapped standard errors, clustered at the industry level, are reported in parentheses. * p<0.10; ** p<0.05; *** p<0.01.

27 Theory or Firm DVAR Assume a translog cost function, P M ( Pit I, ) PD it that is symmetric, homogeneous of degree one: ln P M ( P I t, P D t ) = α i + α 0I ln Pt I + α 0D ln Pt D + 1 ( ) 2 ( ) ( 2 α II ln Pt I + αid ln Pt I ln Pt D + 1 ( ) 2 2 α DD ln Pt D. ) It can provide a second-order approximation to any functional form.

28 Theory or Firm DVAR Recall the accounting identity: DVAR it = 1 PI t M I it P it Y it + ϕ it where ϕ it is a classical regression error term, capturing Sherphard s Lemma implies: P I t M I it P M t M it = M it DVAR it = 1 + PM t P it Y it ( ln PM P I it, Pit D ) ln Pit I = α 0I α ID ln PI t, P D t ( α 0I + α ID ln PI t P D t δ F it EXP it. ) + ϕ it, i, t. DVAR depends positively only on PI t P D t Cobb-Douglas Production Function (given that α ID > 0).

29 Another Benefit of Using a Translog Production Function According to Blackorby and Russell (1989), the elasticity of substitution between the two variables equals the cross-price elasticity ( ε ID ) ( ) t minus the own price elasticity ε DD t : We can express both ε ID t σ t = ε ID t ε DD t and ε DD t as functions of α ID and s D t : ε DD t ln MD t ln P D t = α DD s D t + s D t 1 = α ID s D t + s D t 1; ε ID t ln MD t ln P I t = α ID s I t + s D t, We estimate α ID. σ t = s D t α ID ( 1 s D t ) + 1 > 1, Note that σ t could change over time (and across industries) due to changing s D t.

30 Factors Affecting P I t /P D t Exchange Rates: Let E t = foreign currency value of a Chinese yuan. Pt I = Pt I /E t Yuan depreciation higher P I t /E t P D t higher DVAR FDI: Rodriguez-Clare (1996) and Kee (2015): increased FDI in the output industry can raise the supply and/or quality of domestic input variety higher Pt I /Pt D higher DVAR Upstream Input Tariffs: Goldberg, Khandelwal, Pavcnik, and Topalova (2010): Lower import tariffs lead to significant growth of domestic product variety (in India) higher DVAR

31 Exploring the reasons for the rising firm DVAR We first estimate DVAR it = β i + β jt + β X X it + ɛ it. β i = the firm fixed effect; ɛ it = residual. The estimated β jt, ˆβ jt, captures the average within-firm change in DVAR of each industry j in each year relative to We estimate the following system of three equations using 3SLS: ) ln ( ) P I jt P D jt ˆβ jt = ω 1 j + ω 1 p ln ( P I jt P D jt + ι 1 jt, = ω 2 j + ω 2 E ln E jt + ω 2 v ln V D jt + ι2 jt, ln V D jt = ω 3 j + ω 3 T τu kt + ω3 F ln FDI jt + ω 3 E ln E jt + ι 3 jt,

32 Determinants of the Within-firm Increase in DVAR Within-Firm change in DVAR Change in log relative price of imported materials Change in log relative price of imported materials Change in log upstream variety Change in log upstream variety Change in log (average) upstream tariffs Change in log upstream variety Change in log foreign capital stock

33 Determinants of the Within-firm Increase in DVAR Table 8: Determinants of the Within-firm Increase in the DVAR (1) (2) (3) Dep. Var t,00 DV AR jt t,00 ln(p I /P D ) jt t,00 ln ( ) Vjt D t,00 ln ( P I /P D) 0.269*** jt (0.026) t,00 ln(e jt ) (RMB appreciation) 1.479* *** (0.891) (0.031) t,00 ln ( ) Vjt D t,00 ln ( τ ) U jt *** (3.177) * (0.007) t,00 ln (F DI jt ) 0.017*** (0.002) Industry Fixed Effects N R-sq t,00 is the operator that subtracts the variable of interest from its corresponding value in Bootstrapped standard errors (with 500 repetitions) are reported in parentheses. Coefficients are estimated using Bootstrapped standard errors (with 500 repetitions) are reported in parentheses. Coeffi cients are estimated 3SLS. Columns using 3SLS. (1), (2), Columns and (1), (3) (2), are and third, (3) second, are third, and second, first and stages, firstrespectively. stages, respectively. * p < * p<0.10; ** ** p < p<0.05; *** p < *** p<0.01. Economic Significance

34 Quantitative Analysis To understand how much of the change in firm and aggregate DVAR can be explained by our model, we would need to first estimate the translog parameter, α ID. A firm s DVAR depends on the share of materials in total sales, P M t M it P it Y it, and the translog parameter, α ID, as follows: M it DVAR it = 1 + PM t P it Y it The partial impact of a change in ln ( P I t P D t ( ( P I α 0I + α ID ln t Pt D DVAR ( it ) = PM t M it α P I ID. ln t P it Y it Pt D )). ) on firm DVAR is

35 Quantitative Analysis (cont ) With the estimate of α ID and the actual data on PM t P it Y it, we can calculate how much of the change in firm and industry DVAR is due to the change in the relative price as predicted by our model: M it DVAR it = PM t M it P it Y it α ID ln PI t P D t such estimates allow us to assess the time-series variation in σ and examine whether the rise in firm DVAR is driven by an increasing σ or not.

36 Quantitative Analysis (cont ) To estimate α ID, we estimate the following: Pt I Mit I = a i α ID ln PI t M it Pt D P M t + ξ it, where a i is the firm fixed effect that subsumes α 0I and ξ it is the residual. In other words, α ID is estimated from the within-firm variation in the relative price between imported and domestic materials. We bootstrap the standard errors and instrument for ln PI t exchange rate, FDI and upstream input tariffs. P D t using

37 Estimated Sigma based on the Model σ t = s D t α ID ( 1 s D t ) + 1 > 1 Industry s D 2000 s D 2007 α IV ID s.e. σ IV 2000 σ IV 2007 whole sample *** (0.019) beverages & spirit (16-24) *** (0.211) chemical products (28-38) *** (0.072) plastics & rubber (39-40) *** (0.058) raw hides & skins (41-43) *** (0.112) wood & articles (44-46) (0.529) pulp of wood (47-49) *** (0.180) textiles (50-63) *** (0.066) footwear & headgear, etc. (64-67) *** (0.059) stone, plaster, cement, etc. (68-70) (0.121) precious metals (71) (0.155) base metals (72-83) *** (0.091) machinery, mechanical electrical & equipmt (84-85) *** (0.024) vehicles & aircraft (86-89) *** (0.069) optical, photographic, etc. (90-92) *** (0.048) misc manufacturing (94-96) *** (0.057)

38 Quantitative Analysis (cont ) Between , the average change in ln PI t P D t is The estimated average α ID for the whole sample is 0.376; and the mean share of material cost in total sales is The predicted increase in DVAR it is %, which is not statistically different from the sample average within-firm increase in DVAR (14.7%).

39 Conclusions Develop methods to use firm-level and customs transaction-level data to compute DVAR in exports at various levels of aggregation. The DVAR of Chinese exports increased from 0.65 to 0.70 over China s moving up the global value chain is mostly driven by its processing exporters sourcing more domestically. mainly driven by processing firms substituting domestic materials for imported materials, at both the intensive and extensive margins. The expansion of domestic input variety is induced by decreasing input tariffs facing upstream suppliers and increasing FDI. Focusing on relative prices of domestic inputs, our model explains nearly all of the increase in Chinese firm s and aggregate DVAR from 2000 to Contributed to global trade slowdown (Constantinescu, Mattoo, and Ruta, 2015)?

40 Top 10 Destinations for Chinese Exports Table: Top 10 Destinations of China s Processing Exports Rank USD (Bil) USD (Bil) 1 United States United States Hong Kong Hong Kong Japan Japan Germany 5.62 Netherlands Korea, Republic of 5.34 Germany Netherlands 3.90 Korea, Republic of United Kingdom 3.90 Singapore Singapore 3.62 United Kingdom Taiwan 2.92 Taiwan France 2.10 France Source: China s Customs Trade Data. Back

41 Caveat: Multi-industry firms The above framework is helpful to figure out DVAR at the firm level, but not at the product- or industry- level. Not possible to assign the imported materials to each of the product a firm exports without knowing product-level production functions. Focus on the subset of processing exporters that only operate in a single sector (clusters of HS2 based on UN classifications). E.g.: Machinery, Mechanical Electrical & Equipment (HS 84-85), Textiles (HS 50-63), Footwear & Headgear (HS 64-67) All imports are used for exports in the same sector compute DVAR for each sector using the sample of single-sector processing exporters. Back

42 Firm Heterogeneity and Aggregation Bias Large firms tend to have a higher import-to-sales ratio (Amiti, Itskhoki and Konings, 2014 and Blaum, Lelarge and Peters, 2014). Table 3: Decomposition Exercise: Firm Heterogeneity and Aggregation Bias DVAR of Total Exports Number of rms in the sample (1) Census (0.021) 3419 (2) Original Sample (0.023) 2623 (3) KWW (2012) Estimates N/A (4) Large Firms Only (0.034) 123 Notes: With the exception of (3), all numbers are calculated by the authors based on di erent samples. (1) refers to the 2004 Census of Manufacturing Plants; (2) restricts the sample in (1) to the original survey dataset. (3) is the IO table-based estimate from KWW (2012); (4) restricts the sample in (2) to only rms with total exports larger than 300 million RMB. Bootstrapped standard errors are reported in parentheses. Back

43 Import Tariffs by Industry 01:live animals 02:vegetables 03:animal or vegetable oil 04:beverages & spirit 05:mineral products (Weighted) Average Import Tariffs :chemical products 07:plastics & rubber 08:raw hides & skins 09:wood & articles 10:pulp of wood 11:textiles 12:footwear & headgear, etc. 13:stone, plaster, cement, etc. 14:precious metals 15:base metals 16:machinery, mechical & eletrical equipmt 17:vehicles & aircrafts 18:optical, photographic, etc. 19:arms and ammunition 20:misc manufacturing Graphs by di year

44 Economic Significance From 2000 to 2007, the average increase in ln PI t P D t is The coeff implies a 11.3% increase in the within-firm increase in DVAR. The average log change in input tariffs (across industries) is The coeff implies a 0.7% increase in domestic input varieties, about 20% of the increase over The presence of FDI in the same industry has a positive and significant impact on the variety of upstream materials. The average log change in the stock of FDI (across downstream industries) is The coefficient of implies a 2% increase in domestic input varieties. Back

45 Representation of Different Subsamples by Export Value Table A3: Representation of Di erent Subsamples By Export Values 19 Industry Sales (million usd) customs (mil usd) merged % of customs ltered % of customs 04:beverages & spirit (16-24) :chemical products (28-38) :plastics & rubber (39-40) :raw hides & skins (41-43) :wood & articles (44-46) :pulp of wood (47-49) :textiles (50-63) :footwear & headgear, etc. (64-67) :stone, plaster, cement, etc. (68-70) :precious metals (71) :base metals (72-83) :machinery, mech, elect eqmt (84-85) :vehicles & aircraft (86-89) :optical, photographic, etc. (90-92) :misc manufacturing (94-96) Total Source: China s Customs Trade Data and National Bureau of Statistics (NBS) Manufacturing Survey. Sections 1, 2, 3, 5, and 19 are non-manufacturing sectors and are excluded from the analysis. Sample pooled across Back

46 Determinants of firm DVAR Assume production function Optimization: Y it = φ it K α K it ( M it = M D σ 1 σ it L α L it Mα M it, α K + α L + α M = 1 and σ > 1. + M I σ 1 σ it ) σ σ 1, Pt I Mit I = α M Pt I Mit I P it Y it µ it P M t M it DVAR it = 1 PI t M I it P it Y it = 1 α M µ it ( P I t P D t ) σ 1

47 Determinants of firm DVAR (cont ) DVAR it = 1 PI t M I it P it Y it = 1 α M µ it ( P I t P D t ) σ 1 Given mark-up (µ it ) and cost share of materials (α M ), factors that increase the relative price of imported materials ( PI it ) will increase Pit D firm DVAR. If wages and productivity do not affect P I t /P D t, they will have no impact on DVAR. Back