Currency Wars Unconventional Monetary Policy Does Not Stimulate

Currency Wars? Unconventional Monetary Policy Does Not Stimulate Exports Andrew K. Rose Berkeley-Haas, ABFER, CEPR, NBER

Motivation: What are the Export Effects of Unconventional Monetary Policy (UMP)? • “We’re in the midst of an international currency war, a general weakening of currency. This threatens us because it takes away our competitiveness…” - Brazilian Finance Minister Guido Mantega, Mon Sep 27, 2010, reported by the Financial Times and Reuters • “A ‘currency war’ … occurs when a country eases monetary policy specifically to depreciate its currency, with the ultimate objective of cheapening its exports and gaining unfair competitive advantage in international trade…” - Ben Bernanke (2016, p 2) Rose: Export Effects of Currency Wars 2

Strategy • Use plain-vanilla gravity model of exports • Requires neither UMP exogeneity nor instrumental variables • Straightforward methodology, data • Three assumptions 1. Identify “currency wars” with unconventional monetary policy • UMP: Quantitative easing, negative nominal interest rates, … 2. Currency wars are bilateral events between those engaging in UMP and those not • Estimate effect of UMP by aggressors on potential victims 3. Focus on export effects of UMP Rose: Export Effects of Currency Wars 3

Preview of Findings • Countries that engage in UMP experience drop in exports vis-à-vis countries that did not engage in UMP, ceteris paribus 1. Quantitative easing associated with drop in exports ≈10% • Highly statistically significant 2. Negative nominal interest rates have similar effect • No effect from conditional forward guidance policies • Any currency wars have been lost by aggressors! Rose: Export Effects of Currency Wars 4

A Naïve Look Rose: Export Effects of Currency Wars 5

But that’s … Simplistic • No obvious pattern: • Both exports and imports continue to drop following QE 1, trends unaffected by QE 2, stagnation unaffected by QE 3 • But … • Only US considered • Multilateral blurs American exports to importers engaged in their own UMP (EMU, Japan, UK, Switzerland, …) with exports to importers not engaged in UMP • No controls at all Rose: Export Effects of Currency Wars 6

Standard Gravity Model • Follow Head-Mayer (2014) survey, use Least Squares with Time. Varying Country Dummy Variables: ln(Xijt) = UMPijt + 1 CUijt + 2 FTAijt + {λit} + {ψjt} + {φij} + ijt • is coefficient of interest • UMPijt is 1 if i engages in UMP at t and j does not, 0 ow Rose: Export Effects of Currency Wars 7

Estimation • LS with many fixed effects on positive export flow observations • 11, 773 exporter-quarter • 12, 997 importer-quarter • 26, 096 exporter-importer • FE hold constant any overall effect UMP (or anything else!) had on exports/imports at a point in time • estimates pair-specific time-varying export effect of UMP • Plausibly exogenous since main UMP export effects through common export effect, {λit} • is additional effect of belligerent on defender Rose: Export Effects of Currency Wars 8

Data Set • IMF Do. TS trade: >200 “countries” 2000 Q 1 -2016 Q 4 (with gaps) • Exports are US$ average of FOB exports and CIF imports • RTAs: WTO • Currency Unions: Glick-Rose: Export Effects of Currency Wars 9

Countries Afghanistan Bermuda Colombia Ethiopia Haiti Kuwait Moldova Panama Albania Bhutan Comoros Falkland Islands Honduras Kyrgyz Republic Mongolia Algeria Bolivia Congo, DR Faroe Islands Hungary Laos American Samoa Bosnia & Herzegovina Congo, Rep Fiji Iceland Angola Botswana Costa Rica Finland Cote d'Ivoire Antigua & Barbuda Brazil Sierra Leone Tajikistan Vietnam Papua New Guinea Singapore Tanzania West Bank & Gaza Montenegro Paraguay Slovak Republic Thailand Yemen Latvia Morocco Peru Slovenia Timor-Leste Zambia India Lebanon Mozambique Philippines Solomon Islands Togo Zimbabwe France Indonesia Lesotho Myanmar Poland Somalia Tonga Argentina Brunei Darussalam Croatia Gabon Iran Liberia Namibia Portugal South Africa Trinidad & Tobago Armenia Bulgaria Cuba Gambia Iraq Libya Nauru Qatar South Sudan Tunisia Aruba Burkina Faso Cyprus Georgia Ireland Lithuania Nepal Romania Spain Turkey Australia Burundi Czech Republic Germany Israel Luxembourg Netherlands Russian Federation Sri Lanka Austria Cambodia Denmark Ghana Italy Macedonia Netherlands Antilles Rwanda St. Kitts & Nevis Tuvalu Azerbaijan Cameroon Djibouti Gibraltar Jamaica Madagascar New Caledonia Saint Helena St. Lucia Uganda Bahamas Canada Dominica Greece Japan Malawi New Zealand Saint Pierre & Miquelon St. Vincent & Grenadines Ukraine Bahrain Cape Verde Dominican Republic Greenland Jordan Malaysia Nicaragua Samoa Sudan United Arab Emirates Bangladesh Central African Republic Ecuador Grenada Kazakhstan Maldives Niger Sao Tome & Principe Suriname United Kingdom Barbados Chad Egypt Guam Kenya Mali Nigeria Saudi Arabia Swaziland United States Belarus Chile El Salvador Guatemala Kiribati Malta Norway Senegal Sweden Uruguay Belgium China, Hong Kong Equatorial Guinea Korea, North Mauritania Oman Serbia Switzerland Uzbekistan Belize China, Macao Eritrea Guinea-Bissau Korea, South Mauritius Pakistan Serbia & Montenegro Syria Vanuatu Benin China, Mainland Estonia Guyana Kosovo Mexico Palau Seychelles Taiwan Venezuela Rose: Export Effects of Currency Wars Turkmenistan 10

Measures of Unconventional Monetary Policy 1. QE: 1 for quarters when central bank actively acquiring securities through Quantitative Easing program, 0 ow • 2. 4% of sample • Variant with unusual stocks of assets held (not flow purchases) 2. NNIR: 1 for quarters with Negative Nominal market Interest Rates, 0 ow • 2. 5% of sample • Variant with official policy rates 3. Check with state-contingent forward guidance • 0. 4% of sample Rose: Export Effects of Currency Wars 11

Measures of Unconventional Monetary Policy Quantitative Easing USA, QE 1 2008 Q 4 -2010 Q 1 USA, QE 2 2010 Q 4 -2011 Q 2 USA, QE 3 2012 Q 3 -2014 Q 4 UK, QE 1 2009 Q 1 -2010 Q 1 UK, QE 2 2011 Q 4 -2012 Q 2 UK, QE 3 2012 Q 3 -2012 Q 4 Negative Nominal Interest Rates Switzerland 2011 Q 3 - Denmark 2012 Q 3 - Sweden 2015 Q 1 - Japan 2001 Q 1 -2006 Q 1 Japan 2010 Q 4 - 2016 Q 1 - EMU 2015 Q 1 - 2014 Q 2 - Rose: Export Effects of Currency Wars 12

Main Results • Nuisance Effects • Positive, statistically significant of both controls on (log) exports • Good fit • Coefficient of Interest ( ) • Negative effect of QE on exports ≈ -11% ceteris paribus • |t-statistic| > 5 • Similar effect of NNIR • Cannot reject equality of QE, UMP effects • No significant effect of state-contingent forward guidance • Few obs? • No indication of successful currency wars! Rose: Export Effects of Currency Wars 13

Unconventional Monetary Policy and Exports 1 Exp (not Imp) QE (2. 4%) Exp (not Imp) NNIR (2. 5%) Exp (not Imp) FG (0. 4%) Currency Union RTA 2 3 4 5 6 -. 07** (. 02) -. 05* (. 02). 00 (. 07). 32** (. 02). 04** (. 01). 89 -. 07** (. 02) -. 05* (. 02) 1. 37 -. 11** (. 02) -. 09** (. 02) R 2 . 35** (. 02). 04** (. 01). 89 . 33** (. 02). 04** (. 01). 89 . 32** (. 02). 04** (. 01). 89 -. 02 (. 07). 35** (. 02). 04** (. 01). 89 RMSE 1. 37 Rose: Export Effects of Currency Wars . 32** (. 02). 04** (. 01). 89 14

Results seem robust • Variants of UMP (key regressor) • Stocks (not flows) of QE • Using official (not market) interest rates • Using 1 st/4 th lag or 1 st lead • Dropping early or late data • Drop exporters engaging in UMP, one by one • Drop large sets of importers • Drop observations with large outliers Rose: Export Effects of Currency Wars 15

Sensitivity Analysis, 1 Default UMP Variants (Stock QE, Official NNIR) First lag of UMP Fourth lag of UMP First lead of UMP After 2011 Before 2016 Drop 3σ outliers Quantitative Easing by Exporter, not Importer -. 11** (. 02) -. 14** (. 02) -. 10** (. 02) -. 12** (. 03) -. 10** (. 02) -. 06** (. 02) -. 08** (. 02) -. 07** (. 02) Neg. Nom. Int. Rate in Exporter, not Importer -. 09** (. 02) -. 10** (. 02) -. 08** (. 02) -. 07** (. 02) -. 09** (. 02) -. 04* (. 02) -. 06** (. 01) Rose: Export Effects of Currency Wars Test for Equality (p-value). 66. 09. 59. 24. 66. 21. 71 >. 99 16

Sensitivity Analysis, 2 Quantitative Easing by Exporter, not Importer Drop US as exporter Drop UK as exporter Drop Japan as exporter Drop Denmark as exporter Drop Sweden as exporter Drop Switzerland as exporter Drop Germany as exporter Drop China, HK as exporter Drop Advanced Countries as Importers Drop Asians DCs as importers Drop Africans as importers Drop Latin/Caribbean as importers Neg. Nom. Int. Rate in Exporter, not Importer -. 11** -. 08** (. 02) -. 11** -. 09** (. 02) -. 09** -. 07** (. 02) -. 11** -. 08** (. 02) -. 10** -. 08** (. 02) -. 08* -. 12** (. 04) -. 12** -. 09** (. 02) -. 09** -. 07** (. 02) -. 12** -. 10** Rose: Export Effects of Currency Wars (. 02) Test for Equality (p-value). 57. 70. 75. 50. 84. 16. 67. 58. 06. 51. 34. 87 17

Monadic Fixed Effect Event Studies • Currency war effect buried within exporter-quarter FE, {λit}? • Can examine {λit} directly • Very little movement when country begins to engage in UMP, either QE or NNIR Rose: Export Effects of Currency Wars 18

Exporter-Date Fixed Effects Rose: Export Effects of Currency Wars 19

Conclusion • Q: Did use of Unconventional Monetary Policy (UMP), deliberately or inadvertently, raise exports to countries that did not use UMP? • A: No! • Countries using quantitative easing and/or negative nominal interest rates did *not* experience export booms • If anything, their exports fell, ceteris paribus Rose: Export Effects of Currency Wars 20