Global Sourcing and Domestic Production Networks

Transcription

1 Global Sourcing and Domestic Production Networks Taiji Furusawa - Hitotsubashi U Tomohiko Inui - RIETI Keiko Ito - Senshu U Heiwai Tang - Johns Hopkins SAIS October 14, 2017

2 Remaining questions about global value chains (GVC) Production has been increasing fragmented within and across countries. An extensive literature on the causes and consequences of global production fragmentation. Research on the evolution of the domestic segment of global value chains has been sparse. Who is trading with whom in the domestic economy? How foreign sourcing complements or substitutes for domestic sourcing? Broad implications: propagation of shocks; knowledge spillover; the aggregate effects of misallocation of resources; welfare gains from trade.

3 What we do in this paper? Use Japanese firms production network data (4.5 million buyer-supplier links) to study: the spatial and industrial patterns of firms global and domestic sourcing; how firms offshoring decisions affect their choices of domestic suppliers. Build a model based on Antràs, Fort, and Tintelnot (2017) (Eaton-Kortum at the firm level): Heterogeneous buyers and sellers; Fixed and variable costs for both domestic and foreign trade; Multiple input industries with varying degrees of product differentiation; Firms endogenous trade costs that depend on the intensity of face-to-face communication.

4 Main Empirical Findings Firms are less likely to source inputs from distant suppliers and foreign suppliers (countries), especially for differentiated inputs. Based on a firm-level instrument for offshoring (using world export supply shocks): 1. Offshoring (for exogenous reasons) triggers firms to add and drop domestic suppliers; the net effect is positive. 2. After offshoring, firms are less likely to drop domestic suppliers, but more likely to drop distant and larger suppliers (relative to the existing sellers). 3. More likely to add suppliers that are larger, more proximate, and from differentiated input industries (relative to the existing sellers). These choices of suppliers reduce the average distance of domestic sourcing (i.e., localization of domestic production networks). Literature Review

5 Literature Review Domestic production networks Acemoglu et al. (12); Oberfield (13); Carvalho and Gabaix (13); Carvalho, Nirei, and Saito (14); Bernard, Moxnes and Saito (16); Boehm, Flaaen, Pandalai-Nayar (15); Baqaee (16); Lim (17); Kikkawa, et al. (17). Firms global sourcing and endogenous firms performance Ramanarayanan (14); Blaum, Lelarge, and Peters (16); Kee and Tang (16); Antràs, Fort, and Tintelnot (17). Network and trade Rauch (99); Rauch and Trindade (02); Chaney (14); Eaton et al. (14); Carballo, Ottaviano, and Volpe Martincus (16); Bernard, Moxnes and Ulltveit-Moe (17); Sugita, Teshima, Seira (17). Non-efficiency aspect of firm performance Jensen and Kletzer (05); Holmes and Stevens (15). Economic Georgraphy Davis and Weinstein (02); Duranton and Overman (05); Redding and Turner (15); Davis and Dingel (16), etc.

6 Data Data from the Tokyo Shoko Research, Ltd. (TSR) 800,000 firms in Japan, for 2005 and Info on between-firm relationships: the names of a firm s top domestic suppliers (up to 24) and buyers (up to 24). Use a two-way matching method to construct the domestic production network in Japan. The top seller (an intermediary) in our constructed production network has over 11,000 buyers in 2010; the top buyer (construction company) has close to 8,000 suppliers. Basic firm-level balance sheet info: employment, sales, location, up to three main industries (4-digit), establishment year, number of factories.

7 Data Basic Survey on Business Structure and Activities (BSBSA), from Japan s Ministry of Economy, Trade and Industry (METI). All firms with at least 50 employees or 30 million yen of paid-in capital in the Japanese manufacturing, mining, wholesale and retail, and several other service sectors. 22,939 and 24,892 firms in 2005 and 2010, respectively. Detailed information on firms business activities: main industry code (3 digit), employment, sales, purchases, exports, and imports (by 5 continents and 12 broad sectors). Firm-Size Rank

8 Number of Suppliers by Buyer Type Average number of suppliers per buyer Continuous Importers Import Starters Non-Importers Number of Sellers

9 Newly Offshoring Electronics Producers Dropped Suppliers

10 Newly Offshoring Electronics Producers Added Suppliers

11 Productivity and the Scope of Sourcing nb of sellers by buyer sales number of sellers (2005) buyer's sales (Million Yen, 2005) 95% CI Kernel-weighted local polynomial kernel = epan2, degree = 0, bandwidth = , pwidth = regressions map

12 Distance and the Number of Sellers Figure 1. Distance and Number of Links (2005 & 2010) number of connections distance km Source: Tokyo Shoko Research

13 Post-offshoring Firm Performance y i = α + β imp i + γ ln TFP i + [FE s + FE r ] + ε i, Table 3: Buyer's Offshoring and Changes in the Pattern of Domestic Outsourcing dddd 10 dddd 05 dddd aaa ddddd rrr Dep. Var.: ln(sales) ln(nb. Sellers) ln(nb. Input Industries) ln(nb. Source Regions) 1 2 (dddd 10 + dddd 05 ) ln(dist) 1 2 (ddddaaa + ddddd rrr ) (1) (2) (3) (4) (5) (6) (7) Imp Starter *** *** *** ** * * ** (0.017) (0.016) (0.015) (0.016) (0.017) (0.023) (0.035) ln(tfp) t ** ** (0.011) (0.011) (0.009) (0.009) (0.011) (0.015) (0.027) Buyer (4-digit) Industry FE Buyer Region FE R-sq Nb Obs Note: The regression sample includes manufacturing buyers only and domestic suppliers from both manufacturing and non-manufacturing industries. Each observation is a buyer. When constructing the buyer-specific measures of domestic sourcing, parent-child relationships and sellers with fewer than 5 employees are dropped. The number of observations in column 7 is significantly smaller because not all buyers added or dropped sellers during the sample period. A buyer's TFP is estimated using the Olley-Pakes method with the buyer's value added as the dependent variable. Robust standard errors, clustered by the buyer's region, are used. All existing importers in 2005 are excluded in the sample, so only import starters and non-importers are included. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. s.e. clustered by buyer s region.

14 Model

15 Primitives Antràs, Fort and Tintelnot (2017; AFT) + Bernard, Moxnes and Saito (2017; BMS) + multiple input industries. Dixit-Stiglitz preferences with σ > 1; monopolistic competition in the final goods market. Production of final goods requires intermediates (S different types), which can be in-sourced and outsourced (to domestic or foreign suppliers). M domestic regions + M foreign regions. Each region has an exogenous number n sr of input suppliers.

16 Final-good Producers (Buyers) First, for each input industry s, aggregates input varieties (indexed by j) into composites: [ 1 x is = x is(j) ρs 1 ρs dj 0 ] ρs ρs 1, where ρ s is the elasticity of substitution between different intermediate varieties. Then assemble the composite inputs into final goods: ( xis ) βs, y i = ϕ i S s=1 β s where ϕ is the buyer s core productivity.

17 Buyer s Problem 1. Buyer i and each potential supplier draw its productivity for input production (z s) from an industry-specific Fréchet distribution, before making sourcing decisions. 2. Pay f to outsource in each industry; and pay additional f s to look in a new region for an input supplier. Supplier set: Ω is. Trade Costs 3. For each input variety j [0, 1] from industry s that it has chosen to outsource, choose the lowest-cost (inclusive of trade costs) supplier in Ω is + itself. 4. For each region r Ω is, choose the sector-specific optimal intensity of communication with the sellers. 5. Buyer i optimally sets its final-good price (= constant mark-up over marginal cost) to sell the goods to consumers.

18 Input Quality and Endogenous Communication An input supplier j s will produce high-quality input with probability q (q = 1 for insourcing). With probability 1 q, the supplier produces low quality inputs, which are useless for the buyer. Firms can engage in (face-to-face) communication with the supplier to increase (q). Communication is costly (assumption: more so for inputs sourced from a more distant location): The iceberg trade cost is multiplied by e m(d)q, where m is an increasing function of distance.

19 Buyer Unit Cost of Production and Endogenous Communication Intensity For input composite s, conditional on the set of sourcing regions chosen, the marginal cost is [ c is = µ(i is0 ) 0 p 1 ρ s dg is0 (p) + r Ω is µ(i isr ) ( ρs 1 ρs q 0 isr p) 1 ρs dg isr (p) where p denotes the lowest cost the buyer pays for each unit of input variety j. The optimal communication intensity: q isr = q isr is decreasing in ρ s and d ir. ρ s (ρ s 1)m(d ir ). ] 1 1 ρs.

20 Firms Equilibrium Sourcing Patterns Thanks to Fréchet and Eaton and Kortum (2002), the share of inputs s sourced from region r: s isr = Φ isr Φ is Φ is0 + r Ωis Φ isr where sourcing capability: T s0 (w 0 c s ) θ s if r = 0 Φ isr = [ ] ρs θs n sr T sr (τ s (d ir )w r c s ) θ s ρs 1 ρs θs e ρs 1 if r > 0, ρ s (ρ s 1)m(d ir )

21 Buyer s Profit Buyer i s profits: where π i (ϕ i ) = Bψ 1 σ i S s=1 ] δ is [f + f s r Ω is ψ i ϕ 1 i S s=1 γ β s s Φ βs is θs. and δ is is a dummy equal to 1 if sourcing in industry s. Hypothesis The share of inputs insourced and the share of inputs sourced to closer regions are both greater for the more differentiated inputs.

22 Effects of Firms Offshoring Direct Replacement Effect: When triggered by foreign cost shocks, firms start offshoring inputs from foreign suppliers, which replace their less productive domestic suppliers in the same industry. Productivity Effect: The resulting decline in the firms marginal costs induces the firm to expand domestic sourcing to the more productive suppliers located farther away. Industry Composition Effect: Outsourcing in new input industries (tend to be more differentiated).

23 Testable Predictions Restructuring of Production Networks Hypothesis 1. Relative to non-importers, import starters drop the less productive suppliers in the same industry-region. The replacement effect is more profound in the newly-offshored industries. Since such industries tend to be less differentiated, the dropped sellers tend to larger and more distantly-located. 2. Relative to non-importers, import starters add sellers that are larger and more distantly-located within industries. They may start sourcing in new input industries, which tend to be more differentiated than the industries that have been already outsourced. Because of endogenous communication, the newly added sellers tend to be more closely-located than sellers in other industries.

24 The Pattern of Domestic Sourcing log Φ isr Φ isrs (i) = log n srs (i) log T sr s (i) + θ s log w rs (i) + ρ s θ s ρ s 1 log m(d ir s (i) ) }{{} input-industry base-region-specific + log n sr + log T sr θ s log w r }{{} input-industry source-region-specific Suppose ρ s θ s ρ s 1 log m(d ir ) θ s log t s (d ir ) m(d ir ) = d β ir t s (d ir ) = d γφ s ir where φ s stands for the time sensitivity of the input delivery. Empirical counterpart: log Ns irs N isr(i) s [ ] ρs θ ] s = β ρ s 1 log d ir γ [φ s θ s log d ir ] + [FE sr(i) + FE sr + ε irs

25 Estimated Parameters ρ s : estimated elasticity of substitution between imported varieties in each industry s in the U.S. from Soderbery (2015), improved the original estimates by Broda and Weinstein (2006). θ s : Caliendo and Parro (2015), who estimate θ using data on bilateral trade between 16 economies for 20 ISIC sectors. Our own firm-based estimates: use the empirical distribution of firms revenue to estimate. φ s : the share of air freight costs in U.S. imports in industry s to proxy for the importance of timely delivery.

26 Distance, Product Differentiation, and Domestic Sourcing Dependent Variable: Estimated θ: Table 4: Distance, Scope of Domestic Outsourcing, and Product Differentiation of Inputs (1) (2) (3) (4) (5) (6) ln(n i,source pref /N i,nearest pref ) input ind Caliendo-Parro Estimates using Firm Data ln(dist) i,source pref θ *** *** (0.001) (0.002) (0.001) (0.002) ln(dist) i, source pref θ ρ/(ρ-1) *** *** *** *** (0.002) (0.002) (0.001) (0.001) ln(dist) i,source pref θ air ** (0.000) (0.000) Input Ind Closest Region FE Input Ind Source Region FE Input Ind Buyer Region FE R-sq Nb of Obs relationships are removed from the sample. Data for 2005 are used while the results based on 2010 data are reported in Table A5 in the appendix. Observations are at the buyer input-industry source-region level. All regressions include input-industry-closest-region and input-industry-source-region fixed effects, where the closest region is the closest prefecture from which firm i sources intermediate inputs in a particular industry. Standard errors, clustered at the input-industry-source-region level, are reported in parentheses. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. s.e. clustered by input-industry-source-region. Parent-child pairs were removed (5%). Results are robust to clustering by buyer; restricting to single-plant buyers or single-plant sellers.

27 Back of the Envelope Calculation Relative to the nearest region, a 10% increase in the distance lowers the number of sellers by 0.5% for an industry with a mean value of θ s (9.82) %= *0.1*9.82. A one standard-deviation increase in ρ s /(ρ s 1) (0.262) from the sectoral mean is associated with an additional 0.13% decline in the relative number of sellers.

28 Extensive Margin of Domestic Sourcing Table 5: The Incidence of Domestic Sourcing and Product Differentiation of Inputs Panel A (1) (2) (3) (4) (5) (6) Dependent Variable: Estimated θ: Outsource source pref, input industry Caliendo-Parro Estimates ln(dist+1) i, source pref θ *** *** *** *** (0.000) (0.000) (0.000) (0.000) ln(dist+1) i,source pref θ ρ/(ρ-1) *** *** *** *** (0.000) (0.000) (0.000) (0.000) ln(dist + 1) i,source pref θ air *** *** (0.000) (0.000) Buyer FE Input Sector (12) Source Region FE R-sq Nb of Obs Panel B (7) (8) (9) (10) (11) (12) s.e. Dependent clustered Variable: by buyer. Estimated θ: Outsource source pref, input industry Estimates using Firm Data

29 Extensive Margin of Offshoring Dependent Variable: Estimated θ: Table 6: The Incidence of Offshoring and Product Differentiation of Inputs (1) (2) (3) (4) Offshore input industry Caliendo-Parro Estimates Estimates using Firm Data Domestic sourcing dummy *** *** *** *** (0.002) (0.002) (0.002) (0.002) TFP buyer, *** *** (0.001) (0.001) TFP buyer,2005 θ ρ/(ρ-1) *** *** *** *** (0.000) (0.000) (0.000) (0.000) Input Sector (12) FE Buyer FE - - R-sq Nb of Obs Note: The regression sample includes manufacturing buyers only and domestic suppliers from both manufacturing and nonmanufacturing sectors. Data for 2005 are used. The unit of observation in all columns is at the buyer-source-region-sector level. Parent-child relationships are removed from the sample. Columns 1 and 2 use the estimated θ from Caliendo and Parro (2014), while columns 3 and 4 use our own firm-based estimates of θ. All columns include input-sector fixed effects, while columns 2 and 4 include buyer fixed effects as well. Standard errors, clustered at the buyer level, are reported in parentheses. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. s.e. clustered by buyer.

30 Offshoring and Restructuring of Production Networks Does a buyer s offshoring decision affect its choices of domestic suppliers? What kind of domestic suppliers are most affected? ( ) xij I ij = β imp i + γ imp i log + δ log sales i x i ] + [FE s(i) + FE r(i) + FE s(j) + FE r(j) + ε ij, i and j are buyer, domestic seller. I ij = Drop ij = 1 if i and j are linked in 2005, but not anymore in I ij = Add ij = 1 if a link between i and j was formed since (x ij /x i ) is a measure of seller characteristics relative to the i s 2005 mean. imp i, is the i s importing dummy (since 2005).

31 Instrument for Firms Offshoring Decision ( imp i ) Following Hummels et al. (2014) shock i = φ is WES s WES s = ln(exp) s,2010 ln(exp) s,2005. Japan is excluded from the set of destination countries. φ is = 1 if firm i outsources inputs in industry s in year Instrument for imp i log x ij x i : shock i log x ij x i (Wooldridge, 2010).

32 Supplier Dropping Dependent Variable Table 7: Offshoring and Supplier Dropping (1) (2) (3) (4) (5) (6) (7) (8) Drop ij OLS 2SLS Seller's Characteristics (x j ) - dist sales emp - dist sales emp Imp Starter i *** *** *** *** (0.004) (0.004) (0.004) (0.004) (0.222) (0.238) (0.225) (0.226) Imp Starter i log(x j /x i05 ) *** 0.129** (0.002) (0.002) (0.003) (0.059) (0.030) (0.052) log(x j /x i05 ) *** *** *** * (0.001) (0.001) (0.001) (0.008) (0.004) (0.007) Input Industry FE Buyer Industry FE Source Region FE Buyer Home Region FE Buyer's ln(sales) 2005 Nb of Buyers Nb of Buyers that Offshore Nb of Obs R-squared Kleibergen-Paap F statistic The sample includes only manufacturing buyers that did not import in Newly added sellers are removed from the sample. The unit of observation is a buyer-seller pair. Parentchild relationships are removed from the sample. In column 5, the dependent variable of the first stage is a buyer's import starting dummy. For each regression in columns 6 to 8, there are two first stages, with the dependent variable being a buyer's import starter dummy or its interaction with a seller characteristic. The instrument for a buyer's import starter dummy Firstis the Stage variable shock i constructed based on Equation (18) in the text, while the instrument for the import starter interaction is shock i interacted with the corresponding seller characteristic. See Table A6 for the regression results of the first stage. Robust standard errors are reported in parentheses. ***, **, * indicate significance at the 1%, 5%, and 10%

33 Supplier Adding Table 8: Offshoring and Supplier Adding (1) (2) (3) (4) (5) (6) (7) (8) Dependent Variable Add ij OLS 2SLS Seller's Characteristics (x j ) - dist sales emp - dist sales emp Imp Starter i *** *** *** *** (0.005) (0.005) (0.005) (0.005) (0.178) (0.182) (0.180) (0.181) Imp Starter i log(x j /x i10 ) ** *** 0.167*** 0.252*** (0.002) (0.002) (0.003) (0.037) (0.025) (0.039) log(x j /x i10 ) *** *** *** *** *** *** (0.001) (0.001) (0.001) (0.006) (0.004) (0.006) Input Industry FE Buyer Industry FE Source Region FE Buyer Home Region FE Buyer's ln(sales) 2005 Nb of Buyers Nb of Buyers that Offshore Nb of Obs R-squared Kleibergen-Paap F statistic The sample includes only manufacturing buyers that did not import in Dropped sellers are removed from the sample, so that the comparison is between new suppliers and continuing suppliers. The unit of observation is a buyer-seller pair. Parent-child relationships are removed from the sample. In column 5, the dependent variable of the first stage is a buyer's import starting dummy. For each regression in columns 6 to 8, there are twofirst stages, with the dependent variable being a buyer's import starter dummyor its interaction with First a seller Stage characteristic. The instrument for a buyer's import starter dummy is the variable shock i constructed based on Equation (18) in the text, while the instrument for the import starter interaction is shock interacted with the corresponding seller characteristic. See Table A7 for the regression results of the first stage. Robust standard errors are

34 Supplier Adding and Dropping (across input industries) Dependent Variable Table 9: Offshoring and Industry Adding and Dropping (1) (2) (3) (4) (5) (6) (7) (8) Drop is Add is Drop is Add is OLS 2SLS Imp Starter i *** ** (0.019) (0.041) (0.001) (0.003) (0.466) (0.823) (0.046) (0.056) Imp Starter i x Rauch *** *** 0.168*** (0.020) (0.001) (0.198) (0.018) Imp Starter i x ρ/(ρ-1) ** ** (0.029) (0.002) (0.339) (0.026) Input Industry FE Buyer Industry FE Buyer Home Region FE Buyer's ln(sales) 2005 Number of Obs R-squared Kleibergen-Paap F statistic The sample includes only manufacturing firms that did not import in The unit of observation is a buyer-input-industry pair. The dependent variable of the regressions in columns (1), (2), (5) and (6) is a dummy variable equal to 1 if a 3-digit industry was dropped by a buyer between 2005 and 2010, 0 otherwise. The dependent variable of the regressions in columns (3), (4), (7) and (8) is a dummy variable equal to 1 if a 3-digit industry was added by a buyer between 2005 and 2010, 0 otherwise. For each regression in columns 5 to 8, there are two first stages, with the dependent variable being a buyer's import starter dummy or its interaction with the input industry's characteristic. The instrument for a buyer's import starter dummy is the variable shock i constructed based on Equation (18) in the text, while the instrument for the import starter interaction is shock i interacted with the corresponding input industry's characteristic. See Table A8 for the regression results of the first stages. Robust standard errors are reported in parentheses. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively.

35 Concluding Remarks How offshoring shapes firms domestic production networks? We show that differentiated inputs are less likely to be sourced from distant regions or abroad. Upon firms offshoring, the resulting reduction in variable cost of production expands the geographic scope of domestic outsourcing within each industry; but the increased need to communicate with suppliers in the newly added (differentiated) industries encourage the offshoring firms to source more locally from smaller suppliers. Firms global sourcing can be a possible source of localization of global value chains, and industry agglomeration.

36 The Spatial Pattern of Domestic Sourcing Table A4: Firm Productivity, Distance, and the Scope of Domestic Sourcing (2010) (1) (2) (3) (4) (5) (6) (7) Dependent Variable ln(# sellers' prefectures) buyer ln(# sellers) buyer ln(# jsic 4-digit outsourced) buyer ln(sales/emp) seller Measure of Buyer's Productivity TFP (OP) VA/Emp TFP (OP) VA/Emp TFP (OP) VA/Emp - Productivity buyer 0.104*** 0.344*** 0.141*** 0.553*** 0.110*** 0.485*** (0.021) (0.016) (0.027) (0.025) (0.023) (0.021) ln(distance) *** (0.001) Buyers' (4-digit) Industry FE Buyer's Prefecture FE Buyer FE Sellers' (4-digit) Industry FE Sellers' Prefecture FE Parent-subsidiary dummy Distance b/w buyer-seller SE clustering Buyers' (4-digit) Industry Buyer R-sq Nb of Obs Note: The regression sample includes manufacturing buyers only and domestic suppliers that are either manufacturing or non-manufacturing. The unit of observation is at the buyer level from columns (1) to (6), and at the buyer-seller level in columns (7). All regressions include the most exhaustive set of fixed effects possible. Standard errors, clustered at the buyer's industry level, are reported in parentheses. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Back

37 First Stage of the Supplier Dropping Regressions Table A7: First-Stage of the FE-IV Regressions Reported in Table 8 (1) (2) (3) (4) (5) (6) (7) Dependent Variable Imp Starter i Imp Starter i log(dist j /dist 10 ) Imp Starter i ln(sales j /sales 10 ) Imp Starter i ln(emp j /emp 10 ) Imp Starter i Imp Starter i Imp Starter i Seller's Characteristics (x j ) - dist dist sales sales emp emp shock i 0.082*** 0.079*** *** *** (0.008) (0.008) (0.013) (0.008) (0.018) (0.008) (0.013) shock i (x j -x i05 ) *** *** *** (0.014) (0.037) (0.011) (0.054) (0.014) (0.051) x j -x i *** (0.003) (0.005) (0.002) (0.010) (0.003) (0.010) Input Industry FE Buyer Industry FE Buyer Home Region FE Buyer's ln(sales) 2005 Nb of Obs R-squared The sample includes only manufacturing buyers that did not import in Newly added sellers are removed from the sample. The unit of observation is a buyer-seller pair. Parent-child relationships are removed from the sample. Robust standard errors are reported in parentheses. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Back

38 First Stage of the Supplier Adding Regressions Table A7: First-Stage of the FE-IV Regressions Reported in Table 8 (1) (2) (3) (4) (5) (6) (7) Dependent Variable Imp Starter i Imp Starter i log(dist j /dist 10 ) Imp Starter i ln(sales j /sales 10 ) Imp Starter i ln(emp j /emp 10 ) Imp Starter i Imp Starter i Imp Starter i Seller's Characteristics (x j ) - dist dist sales sales emp emp shock i 0.082*** 0.079*** *** *** (0.008) (0.008) (0.013) (0.008) (0.018) (0.008) (0.013) shock i (x j -x i05 ) *** *** *** (0.014) (0.037) (0.011) (0.054) (0.014) (0.051) x j -x i *** (0.003) (0.005) (0.002) (0.010) (0.003) (0.010) Input Industry FE Buyer Industry FE Buyer Home Region FE Buyer's ln(sales) 2005 Nb of Obs R-squared The sample includes only manufacturing buyers that did not import in Newly added sellers are removed from the sample. The unit of observation is a buyer-seller pair. Parent-child relationships are removed from the sample. Robust standard errors are reported in parentheses. ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively. Back

39 Firm-size Rank Figure 2: Distribution of Buyers with Different Nb of Suppliers number of connections fractions of firms with at least y connections nb sellers per buyer (reg) nb sellers per buyer nb buyers per seller (reg) nb buyers per seller

40 Number of Sellers Sample: All mfg. buyers Summary Statistics (Number of Buyers and Sellers) Continuing importers Import starters between Continuing Nonimporters A. Number of buyers in 2005 Count 13,784 1,807 1,024 10,135 Share (1.00) (0.13) (0.07) (0.74) B. Number of sellers per buyer in 2005 Mean Median Max. 4,724 4,026 1,471 4,724 C. Number of sellers' prefectures per buyer in 2005 Mean Median Max

41 Productivity and the Scope of Outsourcing nb of prefectures outsourced by buyer sales number of prefectures outsourced to (2005) buyer's sales (Million Yen, 2005) 95% CI Kernel-weighted local polynomial kernel = epan2, degree = 0, bandwidth = , pwidth = Back

42 Trade Costs For each input type outsourced, the buyer pays a fixed cost, f, and an additional f s for each source region. No fixed cost for in-house production of inputs. Shipping intermediates entails iceberg transport cost τ s (d) = e t s (d) 1, where t s is an industry-specific increasing function of the distance d between a pair of buyer and seller. Expected outcomes: The combination of firm productivity and incremental fixed costs gives rise to the standard scope-productivity relationship. Firms will always insource part of the input production in each input type. Back