Resolving the Exposure Puzzle The Many Facets of

Resolving the Exposure Puzzle: The Many Facets of Exchange Rate Exposure FDIC October 27, 2006 Söhnke M. Bartram Lancaster University Gregory W. Brown University of North Carolina Bernadette A. Minton Ohio State University

Motivation • FX risk is a major financial risk to nonfinancial firms • “FX Exposure Puzzle” – Theoretical models predict exposure • Bodnar, Dumas and Marston 2002; Marston 2001; Adler and Dumas 1984; Shapiro 1975 – Empirical studies find weak evidence • Allayannis and Ihrig 2001; Dominguez and Tesar 2001 a, b; Griffin and Stulz 2001; Williamson 2001 2

How to Resolve the Puzzle? • General idea/hypothesis of this paper – Firms do have potentially large FX risk – Pricing and financial policies reduce exposures • Analysis – Use structural models to analyze different facets of FX exposure and hedging • Gross (or pre-hedging) exposure • Net (or post-hedging) exposure 3

What We Do • Motivating Example: Global Automotive Industry – Bodnar and Marston (2002) model – Detailed firm-level data • Enhanced Bodnar, Dumas and Marston (BDM, 2002) model – Firm selling and producing in local and foreign market – Exposure is a function of: • market share • product substitutability, • sales and cost in foreign currency – Use model to analyze a large sample of global manufacturing firms around the world 4

Contribution • Resolving the Exposure Puzzle – Model predicts that firms should have large gross FX exposures – Firms reduce these exposures via three channels: • Pass-through (10%-15%) • Operational hedging (10%-15%) • Financial hedging (45%-50%) – Residual FX exposures (as estimated in the prior literature) are economically and statistically small 5

Automotive Industry • Industry study to motivate main analysis – – 16 auto manufacturers from 6 countries Mature and competitive industry Important FX risk (Williamson 2001) Bodnar and Marston (2002) h 1: percent foreign sales h 2: percent foreign cost r : profit margin 6

Automotive Companies Volkswagen Foreign Sales 0. 52 Margin 0. 16 Gross Exposure (h 2=0) 3. 29 7

Automotive Companies Volkswagen Foreign Sales 0. 52 Margin 0. 16 Gross Exposure (h 2=0) 3. 29 Foreign Production 0. 42 Model Exposure 1. 02 8

Automotive Companies Volkswagen Foreign Sales 0. 52 Margin 0. 16 Gross Exposure (h 2=0) 3. 29 Foreign Production 0. 42 Model Exposure 1. 02 FX-Derivatives 0. 46 Foreign Currency Debt 0. 18 Residual Exposure 0. 38 9

Automotive Companies Volkswagen Foreign Sales 0. 52 Margin 0. 16 Gross Exposure (h 2=0) 3. 29 Foreign Production 0. 42 Model Exposure 1. 02 FX-Derivatives 0. 46 Foreign Currency Debt 0. 18 Residual Exposure 0. 38 Actual Exposure 0. 16 p-value [0. 270] 10

Automotive Companies Volkswagen Mean Foreign Sales 0. 52 0. 55 Margin 0. 16 0. 22 Gross Exposure (h 2=0) 3. 29 2. 70 Foreign Production 0. 42 0. 38 Model Exposure 1. 02 1. 26 FX-Derivatives 0. 46 0. 41 Foreign Currency Debt 0. 18 0. 09 Residual Exposure 0. 38 0. 76 Actual Exposure 0. 16 0. 18 [0. 270] [0. 291] p-value 11

BDM Model • FX exposure of exporter • Exposure depends on – Product market competition => pass-through r : degree of product substitutability l : market share of exporting firm in foreign market – Operational Hedging g : fraction of marginal cost due to foreign currency-based inputs 12

Enhanced BDM Model • Two extensions of BDM model – Both firms can have cost in local and foreign currency – Firm sells globally • Global firm is sales-weighted average of foreign and domestic operations (f is percent of foreign sales). • Model is more broadly applicable – Captures global firms in global markets – Allows for broader set of exposures and pass-through – BDM model is a special case of enhanced model 13

Enhanced BDM Model • Foreign Exchange Rate Exposure • Foreign Exchange Rate Pass-Through 14

Parameters f (1 -f) Description Empirical Counterpart Percent foreign sales Percent domestic sales 1 -Percent foreign sales Exporting firm γf 1 Firm’s fraction of marginal costs in foreign currency Percentage of foreign assets of firm gf 2 Import competing firms’ fraction of marginal costs in the foreign market, also foreign currency costs of other exporting firms Weighted average of the percentage of domestic assets of foreign firms and percentage of foreign assets of other domestic firms lf Market share of firm in foreign markets Rest-of-world GDP-weighted average of import penetration ratio 15

Parameters Description Empirical Counterpart Import-competing firm γd 1 Exporters’ fraction of marginal costs in the domestic market, also domestic currency costs of other domestic, import-competing firms Weighted average of the percentage of foreign assets of foreign firms and percentage of domestic assets of other domestic firms gd 2 Firm’s fraction of marginal costs in the domestic currency 1 -Percentage of foreign assets of firm Market share of import competing firm in domestic market 1 -Domestic market import penetration ratio Degree of product substitutability in the foreign (f) and domestic (d) markets Specified exogenously (=1 - γf 1) ld rf, rd 16

BDM vs. Enhanced BDM 17

Automakers Again… 18

Sample and Data • 1, 161 manufacturing firms from 16 countries • Accounting data (USD) (Thomson) • Market data (LC) (Datastream) • Import penetration (UNIDO, SSIS) • Herfindahl indices to measure industry competition (complete Worldscope universe) • Collect data from annual reports for FX derivatives and foreign currency debt 19

Sample Statistics Actual FX Exposure Mean Stdev 25 th Perc. 75 th Perc. 0. 071 1. 945 -0. 806 1. 058 Foreign Sales 34. 5% 27. 2% 11. 6% 53. 5% Foreign Assets 19. 1% 21. 7% 2. 2% 27. 7% Foreign Debt Dummy 0. 871 FX Derivatives Dummy 0. 659 Import Penetration 24. 1% 17. 3% 11. 9% 32. 8% 20

Model Exposures & Pass-Through Mean Stdev rf =rd=0. 5 0. 428 0. 296 rf =rd=0. 7 0. 477 0. 316 rf =rd=0. 9 0. 549 0. 355 Model Exposure 21

Model Exposures & Pass-Through Mean Stdev rf =rd=0. 5 0. 428 0. 296 rf =rd=0. 7 0. 477 0. 316 rf =rd=0. 9 0. 549 0. 355 rf =rd=0. 5 -0. 088 0. 215 rf =rd=0. 7 -0. 061 0. 218 rf =rd=0. 9 -0. 035 0. 223 Model Exposure Model Pass-Through 22

Level of Model Exposure Actual FX Exposure High Low p-value 0. 128 -0. 091 0. 033 23

Level of Model Exposure High Low p-value Actual FX Exposure 0. 128 -0. 091 0. 033 Foreign Debt 0. 978 0. 764 <0. 001 FX Derivatives 0. 799 0. 519 <0. 001 24

Level of Model Exposure High Low p-value Actual FX Exposure 0. 128 -0. 091 0. 033 Foreign Debt 0. 978 0. 764 <0. 001 FX Derivatives 0. 799 0. 519 <0. 001 Import Penetration 0. 289 0. 193 <0. 001 Foreign Import Penetration 0. 295 0. 293 0. 474 Industry Herfindahl 0. 152 0. 140 0. 151 Country-Industry Herfindahl 0. 478 0. 355 <0. 001 25

Level of Model Exposure High Low p-value Actual FX Exposure 0. 128 -0. 091 0. 033 Foreign Debt 0. 978 0. 764 <0. 001 FX Derivatives 0. 799 0. 519 <0. 001 Import Penetration 0. 289 0. 193 <0. 001 Foreign Import Penetration 0. 295 0. 293 0. 474 Industry Herfindahl 0. 152 0. 140 0. 151 Country-Industry Herfindahl 0. 478 0. 355 <0. 001 Model Pass-through -0. 211 0. 088 <0. 001 26

Hedging Effects (1) Intercept -0. 37*** FC Debt FX-Derivatives 27

Hedging Effects (1) Intercept FC Debt -0. 37*** (2) -0. 10 -0. 31*** FX-Derivatives 28

Hedging Effects (1) Intercept FC Debt FX-Derivatives -0. 37*** (2) -0. 10 (3) -0. 30*** -0. 31*** -0. 10 29

Hedging Effects (1) Intercept FC Debt FX-Derivatives -0. 37*** (2) -0. 10 (3) -0. 30*** -0. 31*** (4) -0. 10 -0. 30*** -0. 10 -0. 01 30

Hedging Effects Coef. p-value Intercept -0. 15 [0. 156] FC Debt only -0. 22 [0. 086] 0. 42 [0. 156] -0. 27 [0. 019] FX Derivatives only Both 31

Importance of Hedging Channels • Gross FX exposure = 0. 674 – no market share, no foreign assets, no financial hedging – values at sample means, rf=rd =0. 7 • Firms reduce FX exposure via 3 channels (1) Pass-through (10%-16%) - Market shares at sample average (2) Operational hedging (9%-16%) - Foreign assets at sample average (3) Financial hedging (46% - 50%) - FC Debt (45%) - FX Derivatives (1%) • Consistent with Guay and Kothari (2003), derivatives have limited impact on risk profile of the firm 32

Industry Portfolios 33

Summary • Comprehensive analysis of FX exposure • Resolving the “Exposure Puzzle” – Firms have large gross FX exposures – Firms reduce these exposures via pass-through, operating and financial hedging – Residual FX exposures are economically and statistically small – FX risk management is effective, and companies can stop whining about FX risk 34