Germany s efforts to curb international tax evasion Hannes Fauser (FU Berlin), Jakob Miethe (DIW Berlin) FairTax special session ECE Conference, Brno March 9, 2017
Outline 1 Overview 2 Identification 3 Data 4 Estimation 5 Outlook
Basic idea: tax evasion is hidden but leaves traces indirect identification possible: reactions of financial statistics to policy measures aimed at illegal behaviour (Hanlon et al., 2015; Johannesen & Zucman, 2014) 1 TIEAs (tax information exchange agreements), DTCs (double taxation conventions) 2 domestic tax law changes 3 leaks of identity of evaders ( tax-cds )
Motivation Do deposits from tax havens in German banks react? if yes, evidence of tax evasion (e.g. roundtripping ) exploit monthly Bundesbank deposits data in case study for Germany Advantages: detailed data available (monthly deposits, tax changes, leaks) largest Euro area economy: economically important effects accumulated capital through long standing export surplus (by wealthy HH and firms) long history of tax evasion
Our approach based on Hanlon et al. (2015) part of deposits of German banks vis-à-vis havens actually owned by German evaders using sham corporations Why? home bias of offshore investment, well-being of evader raised by spending offshore funds at home possibilities: credit cards in the name of the shell, donation to private foundation, transfer of funds to a trust held by fiduciaries of the evader unfortunately, not all of these would show up in our data... What shows up: direct accounts in Germany for consumption & investment purposes (we need more - anecdotal - evidence on the portfolio structure of wealthy evaders!)
Key assumptions: 1 Part of the offshore capital of tax-evading German households flows back to Germany in order to be spent or invested. 2 In the deposits data, changes with respect to TIE should exclusively affect illegal capital evaded by German nationals who worry about detection.
Data Main data source: Bundesbank, External Positions of banks monthly liabilities to non-residents of banks in Germany, and of German banks foreign branches and subsidiaries non-residents: foreign enterprises, households and governments enhanced data coverage since 2003 baseline sample 2003-2016
Liabilities, haven vs. nonhaven countries (2003-2016) 800 600 liabs.banks countrytype nonhaven taxhaven 400 200 2006 2010 2014 as.numeric(time) Source: Own depiction, based on Bundesbank data.
Liabilities, haven countries (2003-2016) 150 country 100 Bahamas Bahrain Bermuda Jordan Lebanon Liberia liabs.banks Cayman Islands Curacao Cyprus Guernsey Liechtenstein Luxembourg Malta Marshall Islands Hong Kong Ireland Panama Singapore 50 Isle of Man Jersey Switzerland Virgin Islands, British 0 2006 2010 2014 as.numeric(time) Source: Own depiction, based on Bundesbank data.
Liabilities, selected countries (2003-2016) 150 bank deposits (bn) 100 50 as.factor(country) Cayman Islands France Guernsey Italy Luxembourg Switzerland 0 2004 2008 2012 2016 Source: Own depiction, based on Bundesbank data.
Liabilities, weighted by goods trade (2003-2016) bank deposits/(imports+exports) 75000 50000 25000 country Bahamas Cayman Islands France Guernsey Italy Liechtenstein Luxembourg Switzerland 0 2004 2008 2012 2016 Source: Own depiction, based on Bundesbank and Destatis data.
Liabilities, weighted by goods trade (2003-2016) bank deposits/(imports+exports) 1500 1000 500 country Bahamas France Guernsey Italy Liechtenstein Luxembourg Switzerland 0 2004 2008 2012 2016 Source: Own depiction, based on Bundesbank and Destatis data.
Data for policy measures: TIEAs and DTCs: OECD, BMF dates of signature and entering into force, OECD peer review whether properly drafted and enforced used to construct dataset with dummies for each event, transformed to treatment variables
German treaties with tax havens in our sample period (2003-2016) Counterparty Type Signature In force Andorra TIEA 25 Nov 2010 20 Jan 2012 Anguilla TIEA 19 Mar 2010 11 Apr 2011 Antigua and Barbuda TIEA 19 Oct 2010 30 May 2012 Bahamas, The TIEA 09 Apr 2010 12 Dec 2011 Belgium DTC 21 Jan 2010* not yet* Bermuda TIEA 03 Sep 2009 06 Jun 2012 British Virgin Islands TIEA 05 Oct 2010 04 Dec 2011 Cayman Islands TIEA 27 May 2010 20 Aug 2011 Costa Rica DTC 13 Feb 2014 10 Aug 2016 Cyprus DTC 18 Feb 2011 16 Dec 2011 Dominica TIEA 21 Sep 2010 24 Nov 2014 Gibraltar TIEA 13 Aug 2009 04 Nov 2010 Grenada TIEA 03 Feb 2011 22 Nov 2013 Guernsey TIEA 26 Mar 2009 22 Dec 2010 Ireland DTC 30 May 2011 28 Nov 2012 Isle of Man TIEA 02 Mar 2009 05 Nov 2010 Jersey TIEA 04 Jul 2008 28 Aug 2009 Luxembourg DTC 23 Apr 2012 30 Sep 2013 Malaysia DTC 23 Feb 2010 21 Dec 2010 Malta TIEA 17 Jun 2010 19 May 2011 Mauritius DTC 07 Oct 2011 07 Dec 2012 Monaco TIEA 27 Jul 2010 09 Dec 2011 Montserrat TIEA 28 Oct 2011 31 Mar 2012 Saint Kitts and Nevis TIEA 13 Oct 2010 19 Sep 2016 Saint Lucia TIEA 07 Jun 2010 28 Feb 2013 Saint Vincent and the Grenadines TIEA 29 Mar 2010 07 Jun 2011 San Marino TIEA 21 Jun 2010 23 Dec 2011 Singapore DTC 28 Jun 2004 12 Dec 2006 Switzerland DTC 27 Oct 2010* 21 Dec 2011* Turks and Caicos Islands TIEA 04 Jun 2010 25 Nov 2011 Uruguay DTC 09 Mar 2010 01 Jan 2012 2003/48/EC EUSTD 03 Jun 2003 01 Jul 2005 Note: * indicates signature/enforcement dates of the latest TIE amendment to the DTC, rather than of the original DTC. Sources of data: OECD Exchange of Tax Information Portal and BMF (2017).
tax changes: major and minor increases and decreases, measures increasing domestic threat of detection taken from overview about tax law changes provided by the BMF build dataset for 1968-2016, so far only used since 2003 alternative for tax changes: top statutory personal income tax on dividends and interest further policy var: public scandals involving tax-cds
Key tax policy changes in Germany over our sample period (2003-2016) Policy type Enacted In force Effect income tax 14 Jul 2000 01 Jan 2003 minimum rate -2.9 %pts, top rate -1.5 %pts income tax 14 Jul 2000 01 Jan 2005 (revised) reduction across all income brackets, top rate -5.0 %pts 23 Dec 2003 01 Jan 2004 temporary tax amnesty at reduced rates threat of detection 23 Dec 2003 01 Apr 2005 better access to domestic account information for fiscal authorities income tax 29 Dec 2003 01 Jan 2004 early implementation of tax rates planned for 1 Jan 2005 income tax 24 Jul 2006 01 Jan 2007 top rate +3 %pts income tax and others local/federal corporate tax, income tax 06 Jul 2007 01 Jan 2008 diverse, in total decrease (e.g. corporate tax rate by 10 %pts) income tax 06 Jul 2007 01 Jan 2009 new 25 % flat rate withholding tax on capital income income tax 02 Mar 2009 01 Jan 2009, 01 Jan 2010 rightward shift of tax brackets, i.e. reduction threat of detection 22 Dec 2014 01 Jan 2015 tightening of rules for self-disclosure of evasion Note: Own compilation, based on BMF (2014, 2015, 2016).
Key public scandals involving tax-cds (2003-2016) Event Publication Effects CD with account information from Liechtenstein uncovers tax evasion of wealthy individuals, prominent: Klaus Zumwinkel, former CEO of Deutsche Post fiscal authorities buy CD with Credit Suisse account data, followed by several purchases the same year Federal Constitutional Court decides that acquiring account information is legal purchase of another three CDs with Swiss account data (Coutts, UBS) Uli Hoeneß evasion is publicised after faulty self-disclosure, another CD is bought, Offshore Leaks (with relatively minor impact on German customers, though) Feb 2008 Feb 2010 Nov 2010 Jul 2012 Apr 2013 increased no. of self-disclosures, investigations, fiscal surplus proceeds of around EUR 0.8 bn investigations and self-disclosures authorities can continue paying informants for account data investigations and self-disclosures Hoeneß resigns and is sentenced for prison, major public discussion about tax evasion of the wealthy, high no. of self-disclosures Swiss Leaks, HSBC branch in Geneva Feb 2015 more than 2,000 accounts with German bearing Panama Papers Apr 2016 another major public discussion about offshore financial centers Note: The date refers to when the general public was informed about the event, assuming that potential evaders thinking about self-disclosure did not know earlier. Source: Own compilation.
Further controls: monthly goods trade data (imports and exports) from Destatis not available for some countries (e.g. British crown dependencies simple replacement estimation using population figures) more controls would be desirable, problem: monthly frequency often hardly available example: trade in services data of the Bundesbank only quarterly, and for few countries
We estimate an equation standard to the literature (Johannesen/Zucman 2014, Hanlon et al. 2015): y it = α i + γ t + β 1 Threat Haven + β 2 x it + u it (1) y it : international bank deposits in German banks Year-month (γ t ) and country (α i ) fixed effects x it : control variables such as bilateral trade Threat Haven: treatment variable capturing threat of detection of evasion strategies using tax havens, due to new tax information exchange arrangements (TIEAs, DTCs, the EUSTD)
Results Dependent variable: log(bank deposits tax-havens tax-havens non-havens non-havens (1) (2) (3) (4) Treaty signed 0.404 0.360 0.102 0.080 (0.198) (0.193) (0.138) (0.114) log(imports) 0.088 0.091 (0.051) (0.051) log(exports) 0.181 0.184 (0.061) (0.059) country fixed effect Yes Yes Yes Yes year-month fixed effects Yes Yes Yes Yes Observations 17,034 16,748 17,034 16,748 R 2 0.056 0.107 0.046 0.098 Adjusted R 2 0.041 0.092 0.030 0.083 Note: p<0.1; p<0.05; p<0.01
Robustness: Time Dependent variable: log(bank deposits) tax-havens tax-havens non-havens non-havens (nocris) (postcris) (nocris) (postcris) (1) (2) (3) (4) Treaty signed 0.380 0.201 0.089 0.060 (0.210) (0.091) (0.117) (0.107) log(imports) 0.093 0.072 0.097 0.074 (0.055) (0.033) (0.055) (0.032) log(exports) 0.191 0.139 0.195 0.139 (0.062) (0.062) (0.060) (0.061) country fixed effect Yes Yes Yes Yes year-month fixed effects Yes Yes Yes Yes Observations 14,944 10,636 14,944 10,636 R 2 0.110 0.064 0.101 0.061 Adjusted R 2 0.095 0.045 0.086 0.042 Note: p<0.1; p<0.05; p<0.01
Dependent variable: log(bank deposits) tax-havens tax-havens (1) (2) Bahamas 0.916 0.643 (0.074) (0.189) Bermuda 0.504 0.422 (0.075) (0.083) Guernsey 1.546 1.493 (0.076) (0.073) Ireland 0.696 0.601 (0.072) (0.079) Isle of Man 0.436 0.480 (0.076) (0.072) Jersey 1.071 1.025 (0.079) (0.074) Liechtenstein 0.422 0.401 (0.071) (0.065) Luxembourg 0.144 0.110 (0.070) (0.064) Malta 0.762 0.767 (0.073) (0.066) Singapore 0.771 0.743 (0.078) (0.071) Switzerland 0.254 0.273 (0.073) (0.067) British Virgin Islands 0.049 0.029 (0.073) (0.074) log(imports) 0.083 (0.053) log(exports) 0.179 (0.060) country fixed effect Yes Yes year-month fixed effects Yes Yes Observations 17,034 16,748 R 2 0.089 0.137 Adjusted R 2 0.073 0.122 Note: p<0.1; p<0.05; p<0.01
Test of tax policy changes: y it = α i + γ t + TaxPol Haven + β 2 x it + u it (2) where TaxPol is the tax policy treatment variable (statutory tax rate, domestic threat of detection, public scandals).
Dependent variable: log(bank deposits) tax havens tax havens (1) (2) Top PIT rate on capital income 0.720 (0.787) Better access to domestic account information 0.081 (0.146) EU Directive Administrative assistance 0.107 (0.079) Domestic adaptation to FATCA and CRS 0.015 (0.050) Tightening of self-disclosure 0.129 (0.100) log(imports) 0.087 0.087 (0.051) (0.051) log(exports 0.185 0.188 (0.060) (0.060) country fixed effect Yes Yes year-month fixed effects Yes Yes Observations 16,748 16,748 R 2 0.100 0.103 Adjusted R 2 0.085 0.088 Note: p<0.1; p<0.05; p<0.01
Dependent variable: log(bank deposits) tax havens Zumwinkel 2008 0.061 (0.134) 1st CD wave 2010 0.030 (0.093) Fed. Const. Court verdict 2010 0.069 (0.115) 2nd CD wave 2012 0.003 (0.080) Hoeneß 2013 0.104 (0.058) Swiss Leaks 2015 0.092 (0.080) PanamaPapers 2016 0.132 (0.103) log(imports) log(exports) 0.087 (0.051) 0.188 (0.060) country fixed effect Yes year-month fixed effects Yes Observations 16,748 R 2 0.104 Adjusted R 2 0.089 Note: p<0.1; p<0.05; p<0.01
Further estimations/robustness already done no robustness of baseline results when testing different tax haven lists, e.g. when including Caymans or Cyprus no significant lags or leads (anticipation effects) for our preferred or the full tax haven sample
Further work expand range of controls, in case of necessity quarterly or yearly selection of tax haven sample based on relative importance for German tax evasion customers? Weighting? restrict duration of treatment dummies, in case of TIEAS and DTCs test the period between signature and effective date study earlier sample (dating back until 1968) cost: loss of many havens, partly also reduced time dimension benefit: look at specific 1980s and 90s events (interest income withholding tax 1989, 1993)
Conclusion Information exchange decreases tax evasion via the bilateral haven (deposit shifting possible!) very heterogenous effects over different jurisdictions tax changes have no discernible effect leaked data CDs have the expected sign but are not significant In general, results confirm some tax evasion going on but it is too early in the process to evaluate an overall success or failure of German policy measures
Literature Allingham, M. G. Sandmo, A. (1972), Income tax evasion : a theoretical analysis, Journal of Public Economics 1(3-4), 323-338. Hanlon, M., Maydew, E. L. Thornock, J. R. (2015), Taking the Long Way Home: U.S. Tax Evasion and Offshore Investments in U.S. Equity and Debt Markets, Journal of Finance 70(1), 257-287. Johannesen, N. (2014), Tax evasion and Swiss bank deposits, Journal of Public Economics 111(C), 46-62. Johannesen, N. Zucman, G. (2014), The End of Bank Secrecy? An Evaluation of the G20 Tax Haven Crackdown, American Economic Journal: Economic Policy 6(1), 65-91.