The Role of Regulations and Networks in Firms Offshoring Decisions Work in Progress Simone Moriconi (IESEG School of Management) Giovanni Peri (University of California Davis) Dario Pozzoli (Copenhagen Business School) MWIEG 2018, University of Drexel
Motivation The offshoring of production has expanded dramatically in the recent decades increasing the potential for economic growth and technological transfer (Fenstra and Hanson, 2003; Hummels et al. 2001). Offshoring is often motivated by the firm s desire to reduce costs, to move production closer to foreign consumers, or to utilize a foreign workforce (Bernard et al. 2006). Do firms choose specific countries because of their institutions or because they know them better thanks to their networks?
Previous literature The literature on heterogenous firms show that firms actually engage in offshoring activities only if their productivity levels are high enough to cover the entry costs (Bernard and Jensen, 1999; Melitz, 2003). However, there is no study on which factors determine the entry costs of externalizing production abroad. Partial exceptions can be traced out within the literature focusing on FDI (Olney, 2013; Antras et al. 2009; Okada, K. and Samreth, S., 2014).
Main hypothesis Regulations at destinations shape the firms entry costs: destinations tend to be over-regulated in some areas, e.g., excess bureaucracy for the registration of new companies, or weakly regulated in other areas, e.g., getting or recovering existing credits (Djankov et al., 2002). To press Complementarities can exist across regulations at destinations (Bassanini and Duval, 2009). Firms can exploit networks with the country of destination to discount firms entry costs (Gould, 1994; Head and Ries, 1998; Rauch and Trindade, 2002; Peri and Silvente, 2010). To aims
Press Back
Aims In this paper we use Danish employer-employee matched dataset combined with the Doing Business Database (WB) for the period 2006-2012 in order: 1. to estimate the effects of regulations on firms extensive margin of offshoring and to distinguish those that increase entry costs from those that create an offshoring-friendly business environment. 2. to explore whether there are complementarities across regulations. 3. to check whether firms networks affect their decisions to offshore and whether there is a significant interplay between networks and regulations.
Talk outline Theoretical framework Data and descriptive evidence Empirical models and identification strategy Main Results Refinements Conclusions
Theoretical framework
The Setup 1. We assume a multi-country economy, with a continuum of countries i, j [0, 1]. There are two sectors: i) homogeneous good (perfect competition); ii) differentiated good (monopolistic competition). 2. The world is populated by a unit measure of consumers with identical preferences. Utility is a function of the homogeneous good x o of quantity q(x) of each variety x of the differentiated good (Dixit and Stiglit, 1977). 3. In each country there is a continuum of firms z, heterogeneous in their productivities θ z [0, 1], which produce a single variety of the differentiated product. 4. The production technology of firms combines head-quarter tasks, h, and manufacturing tasks, m.
Offshoring and supply Let s consider a firm in country i that offshores production to a foreign country j. Its Cobb-Douglas production function is: x ij (z) = θ z (h) 1/2 ( λ j m ) 1/2 (1) The output x ij (z) depends on the firm s productivity, θ z, the local headquarters inputs, h, and the manufacturing inputs, m, which are offshored to country j. We assume λ i = 1, so λ j > 1 implies that firms will access more efficient manufacturing inputs by offshoring production. A positive fixed cost f > 1 is born by all firms when they start production. An additional positive fixed cost, r j > 0, is born by the firm when offshoring production to country j. These costs are attenuated with the firm s network in country j, 0 < φ zj 1.
Offshoring and supply The profit function of an offshoring firm is π ij (z) = px ij (z) (h + m) (f + r j φ zj ) (2) If we substitute the demand and production function (1) into (2), we obtain: π ij (z) = A (θ z (h) 1/2 ( λ j m ) ) 1/2 1+σ σ (h + m) (f + r j φ zj ) The firm chooses h and m to maximize (3). (3)
Offshoring and supply From the condition πij (z) > 0, we derive the following productivity threshold: θ > ˆθ zj, where ˆθ zj = 2σ ( ( )) 1 σ f + (σ 1)A σ rj φ 1+σ zj 1+σ γj 1/2 Only firms characterized by a high enough productivity level θ > ˆθ zj will find it profitable to offshore production to country j. The number of offshoring firms depends, among other factors, on fixed costs, r j, and the network to the destination country, φ zj.
Offshoring and supply We then show that the number of offshoring firms relative to non-offshoring firms depends on the following ratio: Θ off = θ ( ) zi f 1 1/(σ 1) = λ 1/2 ˆθ zj f + r j φ j zj Proposition 1: The number of offshoring firms decreases when the institutional fixed costs increase, i.e., dθ off /dr j < 0. Proposition 2: The number of offshoring firms increases when the size of the network in the destination increases, i.e., dθ off /dφ zj > 0.
Data and descriptive evidence
Administrative data sources 1. Integrated Database for Labor Market Research IDA. 2. Firms business accounts FIRM. Large and representative sample of private sector Danish firms (E.g. value added, sales, capital stock). 3. Foreign Trade Statistics Register Intrastat and Extrastat. Custom Data covering the universe of firms trading. Firm s imports (in DKK) disaggregated by Product (8 digit level) and Destination. We drop firms <10 employees and non manufacturing firms = 2,000 firms over the period 2006-2012.
Offshoring measures We follow the well-established method of measuring offshoring as imports which was first constructed by (Feenstra, 1999) at the industry level and then applied to the firm-level Danish data by (Hummels 2014). We construct a narrow offshoring measure that is defined as the summation of imports in the same HS4 category as those sold by the firm. Two margins: extensive and intensive. Alternative measures of offshoring in the robustness checks (broad offshoring and FDI-based measures from Esperian).
Sectors with the largest share of offshoring firms (2006-2012) Share of offshoring firms within industry 0.02.04.06 Computer and elect. Leather Motor vehicles Machinery and equipment Other manuf Basic metals Electrical machinery Pharma Textile Basic chemicals Ships and transport eq. Plastic and rubber
Doing Business Database (2006-2012) We measure 3 groups of regulations at the country of destination: 1. Labor regulations: i) whether fixed term contracts are prohibited; ii) maximum number of working days per week; iii) whether the employer must notify or consult a third party before a collective dismissal of employees; iv) minimum wage. 2. Business regulations: i) Time required to start a business (days); ii) Time required to register property (days); iii) Time to prepare and pay taxes (hours); iv) Time to export (days). 3. Credit regulations: i) Lack of private credit bureau coverage; ii) Lack of investors protection index; iii) Enforcing contracts, cost (percent of claim) and viii) Rate of insolvency.
Regulations by destination country in 2006 and 2012 Index of labor market rigidity, 2012-2 -1 0 1 2 GE TGTD HN NE NP TZ ZA LV UA KE ES DZ MDSN PY PA IQ FI SC PT EC PECG FJ TJ GQ GRMA AO IR BJ CM AL AR BOCR CV DJGH EE GT DO AT GM GW MG ML PH ZM SY LA ST YE MR NI IENL NO LT AM BFGD JO BZ BT CL BW DE KR MN LK MU RS SK TN TH GA HU UZ UY KG FR EG KZ LC ET GN LS TR SVE BE AGIL CO BY CA DM IS CZ BG LB GB IT NA PL VN JM SZ SE BI AZ KN AE BN CH CI HK KW NZ OM VC SA SG TT UG MY AU -2-1 0 1 2 Index of labor market rigidity, 2006 Index of business regulations, 2012-2 -1 0 1 2 BO GQUZ VN TJ KZ GNE BI BJ CM PL MR SNLS LA EC BT ET GA KGPY KE ZM BF TT IL BZ AL AR CI DZ GM LK NP NA ST FJ NI MG CZ EG BG AM BWCR CL GT IR KN DJ AZ AG IT DE BECV DO GH GR JM MN TZ UG UY MA ML KW PH SZ PA LB SV ZA SC HN AT DMSK SY YE TH FR GES GD MD HU PE VC TN CO ISIE JOLV MY CA CH LT GB LC PTRS FI KR SA TR MU NO AUNZOM HK AENL SE SG EE -2-1 0 1 2 Index of business regulations risk, 2006 Index of credit risk, 2012-4 -2 0 2 AO BF CV LA ST GD CM CG BJ BTBI AG BZ CH BNBO CI DJ DM GM GA GN ET GT GQ GW IQ HN JO KE KN TD VCVE PH MG ML MR LS MN NE UA TZ DO LC EC SA TG EG DZ JM GH KG HU AE IR LB MD LV AL BG KW AZ MA SNUZ UG NI SC SY NPPA SZ ZM YE VN PYTTN OM CLFJ CRBY FR MU PE TJ NA AM EE LK GE ES ZA SV SKGR TR KZ RS AR AT PTPL BE BW CZ FI TH UY LT IL NL CO IT MY DE SG SE AUHK IECA GB KR NO NZIS -4-2 0 2 Index of credit risk, 2006
Extensive margin of offshoring, regulations and network 0.1.2.3.4.5.6 Extensive margin of offshoring -3-2 -1 0 1 2 Index of labor market rigidity Extensive margin of offshoring 0.1.2.3.4.5.6-2 -1 0 1 2 Index of business regulations Extensive margin of offshoring 0.1.2.3.4.5.6-4 -2 0 2 Index of credit risk Extensive margin of offshoring 0.05.1.15.2 0.01.02.03.04 Network
Intensive margin of offshoring, regulations and network Intensive margin of offshoring 0 5 10 15 20 Intensive margin of offshoring 0 5 10 15 20-3 -2-1 0 1 2 Index of labor market rigidity -2-1 0 1 2 Index of business regulations Intensive margin of offshoring 0 5 10 15 20 Intensive margin of offshoring 0 5 10 15 20-4 -2 0 2 Index of credit risk 0.01.02.03.04 Network
Empirical model and identification strategy
Empirical model We estimate the following bilateral regression model: Off ijmct = α + r jt 1 β + γφ jct 1 + X it 1 ζ + θ i + θ j + θ m + θ c + θ t + ɛ ijt The dependent variable, Off ijmct is firm i s offshoring (extensive and intensive margins) to destination j at time t. The vector X it 1 comprises a set of firm characteristics (such as productivity, capital intensity, number of destinations, foreign ownership plus detailed workforce characteristics). We also include firm fixed effects, θ i, destination fixed effects, θ j, industry fixed effects, θ m, municipality fixed effects, θ c and time fixed effects, θ t.
Identification strategy: Regulations To estimate the coefficients β on regulations (r jt 1 ): we exploit changes in the national regulations which mimic changes to the bilateral costs of offshoring, which are exogenous to Denmark and vary across destination countries. The effects for Danish firms are fairly comparable to a tariff reduction (increase), as Danish firms have very limited influence on the outcome of these reform processes carried outside Denmark. The vector r jt 1 is lagged one period to account for the fact that companies cannot immediately adjust offshoring activities in response to changes in regulations.
Identification strategy: Network φ jct 1 proxies for the strength of the firm i s networks to the country of destination j. This variable is computed as the share of foreign workers from country j in the municipality c in which the firm is localized at time t 1. Unobserved municipality-specific shocks could influence both immigration and offshoring. We instrument φ jct 1 with its shift-share prediction (Card, 2005): φ IV jct 1 = F jt 1 (F jc96 /F j96 ) P c96
Main hypothesis Hypothesis 1: β < 0 for the extensive margin of offshoring. The coefficient β measures the bilateral impact of an increase in the fixed costs associated with regulation r on the decision of firm i to offshore in country j. Hypothesis 2: γ > 0, for the extensive margin of offshoring. The coefficient γ measures the impact of a network of relation between firm i and destination j at the base year, that can help firm i to decrease the total fixed costs of offshoring to the same destination. Hypothesis 3: β = 0 and γ = 0 for the intensive margin of offshoring, conditional on offshoring.
Main Results
Regulations, network and offshoring (1) Extensive Margin Intensive Margin Labor regulations Limits on fixed term contractst 1 0.001982 0.002126 0.002519 0.059972 0.057303 0.055331 (0.003768) (0.003893) (0.003854) (0.037883) (0.038522) (0.038639) Limits working days per weekt 1 0.000095 0.000212 0.000384 0.022606 0.024306 0.025662 (0.004173) (0.004318) (0.004313) (0.021384) (0.022046) (0.022806) Employment protection measurest 1 0.010065 0.010435 0.010425 0.013253 0.011061 0.009268 (0.003795) (0.003913) (0.003786) (0.017696) (0.022478) (0.021987) Minimum waget 1 0.003026 0.003156 0.002490 0.042080 0.037917 0.042466 (0.002754) (0.002852) (0.002780) (0.041490) (0.042418) (0.041427) Business regulations Time to open a businesst 1 0.001487 0.001962 0.002945 0.008002 0.001684 0.000050 (0.002761) (0.002912) (0.003052) (0.080203) (0.078492) (0.079155) Time to register propertyt 1 0.003040 0.002912 0.003028 0.021449 0.023402 0.023510 (0.003764) (0.003883) (0.003827) (0.036025) (0.029861) (0.029921) Time to pay taxest 1 0.001481 0.001758 0.001848 0.004664 0.005053 0.004673 (0.002637) (0.002727) (0.002643) (0.031913) (0.034325) (0.033540) Time to exportt 1 0.004146 0.004347 0.002766 0.055328 0.056919 0.060322 (0.002674) (0.002768) (0.002703) (0.051635) (0.050651) (0.051202) Credit regulations 100-credit coveraget 1 0.014083 0.014580 0.014058 0.038518 0.044338 0.043336 (0.006627) (0.006841) (0.006784) (0.026804) (0.023480) (0.023775) 10-investors protectiont 1 0.004643 0.004831 0.004660 0.009180 0.008470 0.009404 (0.004354) (0.004497) (0.004441) (0.033030) (0.029279) (0.029699) Enforcing contractst 1 0.005105 0.005419 0.004456 0.018960 0.012949 0.010622 (0.003641) (0.003812) (0.003833) (0.033957) (0.027537) (0.027496) 100-resolving insolvencyt 1 0.026081 0.026917 0.026000 0.039105 0.035770 0.040871 (0.005748) (0.005925) (0.005804) (0.038549) (0.032192) (0.035170) Network 0.011507 0.011646 0.011646 0.064152 0.073399 0.073504 (0.000520) (0.000557) (0.000557) (0.080092) (0.069885) (0.069845) First Stage-Network IV Coeff. 0.643 (0.201) 0.642 (0.201) 0.659 (0.200) 0.701 (0.221) 0.704 (0.227) 0.699 (0.217) R-sq 0.122 0.122 0.125 0.287 0.287 0.287 N 1,403,850 1,403,850 1,403,850 46,282 46,282 46,282
Regulations, network and offshoring (2) Extensive Intensive [1] [2] [3] [4] [5] [6] Index of labor market rigidityt 1 0.006879 0.006864 0.006878 0.002333 0.001438 0.004303 (0.003366) (0.003348) (0.003365) (0.028773) (0.033421) (0.030408) Index of business regulationst 1 0.001837 0.001911 0.001833 0.021747 0.027755 0.018506 (0.003540) (0.003580) (0.003538) (0.038746) (0.042515) (0.040410) Index of credit riskt 1 0.037939 0.038081 0.037926 0.060807 0.056316 0.064886 (0.007514) (0.007433) (0.007513) (0.038084) (0.033891) (0.038564) Network 0.012646 0.012644 0.012571 0.073572 0.073631 0.066968 (0.000557) (0.000557) (0.000656) (0.068565) (0.064981) (0.072315) Index of labor market rigidityt 1 index of business regulationst 1 0.002166 0.006998 (0.001490) (0.096458) Index of labor market rigidityt 1 index of credit riskt 1 0.001422 0.060976 (0.002366) (0.084148) Index of business regulationst 1 index of credit riskt 1 0.000764 0.081142 (0.001112) (0.133644) Index of labor market rigidityt 1 network 0.000042 0.033467 (0.000489) (0.029501) Index of business regulationst 1 network 0.000245 0.010891 (0.000543) (0.031858) Index of credit riskt 1 network 0.001561 0.019065 (0.000702) (0.022388) Mean Y 0.033 0.033 0.033 10.182 10.182 10.181806 R-sq 0.121 0.126 0.122 0.287 0.287 0.287195 N 1,403,850 1,403,850 1,403,850 46,282 46,283 46,284
Interactions between credit risk and network Marginal effect of network (1) -.01 -.005 0.005.01-2 -1 0 1 2 Index of credit risk (standardized) Marginal effect of network (2) -.01 -.005 0.005.01-2 -1 0 1 2 Index of credit risk (standardized) Marginal effect of network (3) -.01 -.005 0.005.01-2 -1 0 1 2 Index of credit risk (standardized) Source: Marginal effect of network (1) is calculated by interacting our network variable with the index of credit risk and setting the index of labor market rigidity and of business regulations at the 25th percentile of their distributions. Marginal effect of network (2) is calculated by interacting our network variable with the index of credit risk and setting the index of labor market rigidity and of business regulations at the median percentile of their distributions. Marginal effect of network (3) is calculated by interacting our network variable with the index of credit risk and setting the index of labor market rigidity and of business regulations at the 75th percentile of their distributions. indicates significance at the 95% level.
Regulations, network and exporting Extensive Intensive Index of labor market rigidityt 1 0.011589 0.011098 0.011585 0.013869 0.008838 0.014724 (0.008833) (0.008715) (0.008832) (0.122734) (0.125938) (0.122628) Index of business regulationst 1 0.009744 0.009741 0.009734 0.070687 0.074695 0.071581 (0.009978) (0.010009) (0.009980) (0.152021) (0.155339) (0.152145) Index of credit riskt 1 0.097620 0.097272 0.097592 0.668429 0.671539 0.667848 (0.015581) (0.015424) (0.015580) (0.174569) (0.176449) (0.175119) Network 0.010768 0.010769 0.010587 0.000785 0.000727 0.007581 (0.001066) (0.001118) (0.001297) (0.020308) (0.020971) (0.023483) Index of labor market rigidityt 1 index of business regulationst 1 0.004967 0.049488 (0.003901) (0.086969) Index of labor market rigidityt 1 index of credit riskt 1 0.003527 0.033184 (0.006855) (0.092702) Index of business regulationst 1 index of credit riskt 1 0.000052 0.040061 (0.002704) (0.088095) Index of labor market rigidityt 1 network 0.000505 0.004880 (0.001151) (0.017550) Index of business regulationst 1 network 0.001316 0.004457 (0.001358) (0.018224) Index of credit riskt 1 network 0.001160 0.003890 (0.000280) (0.016745) Labor productivityt 1 0.002160 0.002153 0.002164 0.006524 0.006567 0.006557 (0.000618) (0.000603) (0.000621) (0.019243) (0.018444) (0.018912) R-sq 0.290 0.301 0.290 0.272 0.272 0.272 N 1,403,850 1,403,851 1,403,852 144,701 144,701 144,701
Refinements of Main Results
Robustness checks Subsamples: Only exporting firms. Developing versus developed destination countries. Labor intensive versus capital intensive sectors separately. Service industry. Refinements on the offshoring variable: Broad offshoring. FDI-based measure from Esperian. Intensive margin calculated as the share of bilateral imports out of total imports. Alternative interaction specifications and non-linearities.
Preliminary conclusions This paper explores how regulations and network affects the firm s offshoring outcomes by using a representative sample of Danish manufacturing firms (2006-2012). First, we find that regulations that reduces credit risks tend to increase firms propensity to offshore to the this destination. Second, we show that regulations increasing labor market rigidity have a negative impact on firms offshoring decision. Third, our results also suggest that firms networks with the destination country has an independent fixed-cost reducing effect on the extensive margin. However, the positive impact of networks is attenuated in those destination markets with high levels of credit risks.