1 Dr. Eric Pichet Professor of Economics Director at IMPI, Postgraduate Degree in Wealth Management Bordeaux Business School THE ECONOMIC CONSEQUENCES OF THE FRENCH WEALTH TAX (ISF) Published in La Revue de Droit Fiscal, n° 14- 5 avril 2007 ABSTRACT Despite attempts to “un-wind” the Impôt de Solidarité sur la Fortune (“Solidarity Wealth Tax”, the French wealth tax) during the last legislature (2002-2007), ISF yields had soared by 2006, jumping from €2.5 billion in 2002 to €3.6 billion. Analysis of the economic consequences of this ISF wealth tax has raised the following conclusions: - Tax collection costs remain low (around 1.6% of proceeds). - Not raising the threshold in line with inflation between 1998 and 2004 created windfall revenues for the French State of €400 million in FY 2004 alone. - ISF fraud mainly involving an under-assessment of property assets has stabilised over time at around 28% of total revenues, equivalent (had the legal framework remained unchanged) to a shortfall for the State of €700 million in 2004. - Capital flight since the ISF wealth tax’s creation in 1988 amounts to ca. €200 billion. - The ISF causes an annual fiscal shortfall of €7 billion, or about twice what it yields. - The ISF wealth tax has probably reduced GDP growth by 0.2% per annum, or around 3.5 billion (roughly the same as it yields). - In an open world, the ISF wealth tax impoverishes France, shifting the tax burden from wealthy taxpayers leaving the country onto other taxpayers. Key words Wealth tax, Economic consequences of taxation, French taxation, tax competitiveness, fiscal flight. Electronic copy available at: http://ssrn.com/abstract=1268381 2 INTRODUCTION “Income tax acts as the carrot, wealth tax as the stick” Emile de Girardin, Le Socialisme et l’impôt. One paradox of France’s last legislature was that despite attempts to “un-wind” the Impôt de Solidarité sur la Fortune (“Solidarity Wealth Tax”, the French wealth tax, hereafter referred to as ISF wealth tax, or ISF) - 24 legal changes mostly aimed at reducing the levy’s proceeds - ISF yields had soared by 2006, jumping from €2.5 billion in 2002 to €3.6 billion. Analysis of the economic consequences of this ISF wealth tax has raised the following conclusions: - Tax collection costs remain low (around 1.6% of proceeds). - Not raising the threshold in line with inflation between 1998 and 2004 created windfall revenues for the French State of €400 million in FY 2004 alone. - ISF fraud mainly involving an under-assessment of property assets has stabilised over time at around 28% of total revenues, equivalent (had the legal framework remained unchanged) to a shortfall for the State of €700 million in 2004. - Capital flight since the ISF wealth tax’s creation in 1988 amounts to ca. €200 billion. - The ISF causes an annual fiscal shortfall of €7 billion, or about twice what it yields. - The ISF wealth tax has probably reduced GDP growth by 0.2% per annum, or around 3.5 billion (roughly the same as it yields). - In an open world, the ISF wealth tax impoverishes France, shifting the tax burden from wealthy taxpayers leaving the country onto other taxpayers. Implemented by the Rocard government in 1988 1 and built on the smoking ruins of an older Impôt sur les Grandes Fortunes, (IGF, Tax on Great Wealth) 2 , the ISF wealth tax is both marginal in the French fiscal system (its 2006 yield of €3.6 billion represents ca. 1.5% of total 1 1989 Provisional Budget n° 88-1149 dated 23 December 1988. Reflections on wealth tax go back centuries, one example being the writings of Diomède Carafa (1406-1487), notably his De Regis and bonis principis officia, or contemporary texts by Matteo Palmieri (1405-1475) who wrote Della Vita civile plus Molina (15351600), Lessius (1554-1623) and Lugo (1583-1600). Holland was the first country to implement a wealth tax in 1892, imitated by Germany the following year. Over the past 30 years, there has been a strong trend towards the outright elimination of any taxation of accumulated wealth. This latter wave, which began in 1977 in Ireland (followed by Austria in 1994), finally hit Denmark and Germany in 1997 (the latter having committed euthanasia on its Vermögensteuer via a 22 June 1995 Karlsrûhe Court decree giving the government until 1 January 1997 to enact a reform); Italy in 1998; and Luxemburg and Finland in 2006. At the time of writing (2007), a wealth tax survives in only three European countries: France, Sweden and Spain. 2 The IGF was created in France’s 1982 Provisionial Budget and eliminated by the Chirac governement in 1986. Electronic copy available at: http://ssrn.com/abstract=1268381 3 tax revenues) and highly symbolic (hence eminently political). 3 The intense media interest in French rock star Johnny Hallyday’s tax exile Switzerland is merely the latest uproar surrounding a tax whose scope covers no less than the way each of us relates to money. 4 Over the past few years, however, there have been some interesting changes in the debate surrounding the ISF wealth tax. Some age-old ideological and moral constraints (upon which most of the controversies stirring up France’s political classes are still dependent) seem to have been finally lifted, to be replaced by more scientific approaches that have emerged in the wake of the highly publicised expatriation of a few large French estates, or the wealth taxdriven sale of some of the jewels in France’s industrial crown 5 . Concerns about fiscal efficiency have raised awareness as to the tax’s negative side-effects on the French economy. Since Adam Smith, we have known that four criteria define a good tax in a market economy. Two relate to the principles of equity and legal security, which stipulate that each taxpayer’s levy be proportionate both in terms of the amount involved and the individual’s ability to pay, and also that the tax must be certain, i.e. non-arbitrary. The ideal tax would also satisfy the principles of economy (in virtue of which collection costs should be as low as possible) and convenience, which stipulates that, “All taxes must be collected in a time and way that is most convenient for the taxpayer”.6 In a modern economy, another element should be added to Adam Smith’s magic square, to wit neutrality, the idea being that a tax may not damage a country’s economic development. 3 It is impossible to exaggerate the importance of semantics in the political arena in general and specifically in the field of taxation. Take for example the IGF Tax on Great Wealth created in 1981 (for the 1982 Provisional Budget) by a freshly elected Socialist-Communist government waging war against the privileges of the mythical “200 families”, in a commendable search to “change life”. When the defunct IGF was reincarnated in 1988 as the ISF wealth tax, it had a far more modest goal of “taking a little from the wealthy to give to the poor”, by reintroducing a now more or less ideologically neutral levy to finance the creation of a RMI (Revenu Minimum d’Insertion, Minimum Income Provision). For further details, see France’s 1989 Provisional Budget. 4 Galbraith noted, not without reason, that unlike many goods in this world (“cars, mistresses, etc.”), money is one of the rare things that interest those who have it and those who don’t. For further details on Galbraith’s thinking, re-read Money: Whence it Came, Where it Went, Houghton Miffin, 1975. Note the preface’s affirmation, perfectly transposable to taxation, that the history of money is one of mismanagement. 5 The latest example being the Taittinger affair. One of the protagonists of this takeover bid, Albert Frère, mulling over the moral of the tale, has disarmingly confessed that, “The ISF wealth tax is a great tax that has helped the Americans to take ownership of the Taittinger group, and me to increase my personal fortune.” 6 The Wealth of Nations, Vol II, Book V, Chap II. For a summary of Adam Smith’s views on taxation; see Adam Smith, Le Père de l’économie, E. Pichet, Les Editions du Siècle, 2003, p. 150-154. 4 This highly desirable fiscal virtue is the goal to which we dedicate the present paper, one that updates and supplements a 2000 research report called The Real Cost of the ISF Wealth Tax. 7 Throughout the present text, we will situate ourselves in a world of economic efficiency, trying to determine the extent to which the ISF wealth tax remain economically neutral, i.e. whether it is profitable both for the Treasury and for the country. The recently completed legislature enacted many exemptions (starting to “un-wind” the ISF wealth tax in the flowery but appropriate parlance) that should have led to a fall in yields yet these continued to rise. To understand this paradox, we will identify the various factors that affect the ISF wealth tax’s net yield – some of which relate to the business climate (like the not insignificant effect of the ISF threshold’s not having been index-linked from 1998 to 2004) whereas others are structural (like collection expenses or the cost of fraud). We will use this empirical data to try and model a theoretical ISF wealth tax by calculating the variance with actual ISF proceeds. Lastly, we will ascertain how much tax flight the ISF wealth tax has caused since 1988, and what it costs the French economy. I. Analysis of ISF wealth tax proceeds “I’d rather be full of paradoxes than prejudice.” Jean-Jacques Rousseau, Emile or De l’Education, Book II. In 1988, the founders of France’s ISF wealth tax argued quite sensibly that its resurrection would increase State revenues to help fund some of the country’s new anti-poverty RMI Minimum Income Provisions. 8 15 years later, analysis indicates that despite the multiple exemptions granted to taxpayers over the course of the last legislature, ISF yields have risen constantly. This requires explanation. 1.1 The paradoxical rise in yields 7 CEREFI Research Report, n° 25-2000, The Real Cost of the ISF Wealth Tax, Bordeaux Business School. According to the latest statistics, this hasn’t happened. In 2006, for example, the RMI cost twice as much as the ISF yielded (€6 billion vs. €3.6 billion). 8 5 Without ever questioning the sacred cow that the ISF wealth tax has become over the years, the legislative stipulations of the past five years have all surreptitiously tended towards its subtle un-winding. Amongst the 24 modifications contained in the French Tax Code (article 885 and subsequent), only nine have been enacted within a Provisional Budget. In our opinion, this translates a desire to dissimulate, in highly technical texts, cuts in this tax. - A subtle and purposeful un-winding Amongst the better known measures, ranked in decreasing order, note first and foremost the famous “tax shield” 9 that is very symbolically codified in article 1 of the French Tax Code: “Direct taxes may not exceed 60% of a taxpayer’s income”. This measure, whose initial effects will only be felt in 2007, should cost the French Treasury €400 million. 10 The SNUI Tax Workers Union strongly criticised this “gift to the wealthy” since it felt that out of the 100,000 taxpayers affected, the 10,000 to 15,000 who are liable to pay the ISF wealth tax would themselves recover as much as €200 million. 11 The second series of abatements came from the Dutreil Economic Initiative Law n° 2003-721 dated 1 August 2003, with its three measures benefiting persons liable to pay the ISF wealth tax: the launch of a collective commitment to keep shares that, if held for six years and where the shares involved constitute at least 20% of a firm’s financial and voting rights, would benefit from an abatement of 75% 12 ; a total exemption on securities received by participants in a SME rights issue 13 ; and a total exemption of shares owned by taxpayers who fulfill certain business asset conditions but remain below the family group threshold (as long as the stake in question exceeds 50% of their taxable goods’ gross value) 14 . Lastly, Law n° 2004-1485 dated 30 December 2004, which became the Revised 2004 Budget, renewed with the tradition of index-linking ISF thresholds to France’s ex-tobacco price index, a system last used in 1998. - A clear rise in gross proceeds over the course of the legislature. 9 Article 74 I of Law n°2005-1719 dated 30 December 2005 becoming the 2006 Provisional Budget. 2006 Provisional Budget 11 See the SNUI’s Regards sur le projet de loi de finances 2006, (A look at the 2006 Provisional Budget) 12 French General Tax Code Article 885 I bis 13 French General Tax Code Article 885 I ter 14 French General Tax Code Article 885 O bis 10 6 The ensuing losses for the State, due both to persistent fraud and to the recent adoption of measures benefiting taxpayers, should logically have led to a fall in ISF proceeds. The numbers seem, however, to substantiate a statement by France’s Ministry of Finance describing the ISF as a “tax that is relatively impermeable to the business cycle”. 15 This is because preceding measures’ combined effects have been unable to prevent the inexorable rise in the absolute value of gross proceeds since 1989, as the table below shows 16 : Year 15 16 ISF wealth tax Variation, ISF Nr. of taxpayers Variation, number proceeds (€ bio) wealth tax of taxpayers 1989 0.692 - 126,300 - 1990 0.924 +33.3% 140,461 +11.2% 1991 0.980 +6.2% 150,177 +6.9% 1992 1.069 +8.9% 157,666 +5.0% 1993 1.098 +2.8% 163,125 +3.5% 1994 1.268 +15.4% 171,706 +5.3% 1995 1.299 +2.5% 175,926 +2.5% 1996 1.349 +4.6% 174,726 -0.7% 1997 1.540 +12.8% 178,899 +2.4% 1998 1.677 +10.7% 192,734 +7.7% 1999 1.936 +14.5% 212,008 +10.0% 2000 2.424 +24.9% 244,656 +15.4% 2001 2.668 +9.5% 268,448 +9.7% 2002 2.461 -7.4% 281,434 +4.8% 2003 2.335 - 5.1% 296,795 +5.5% 2004 2.646 +13.3% 333,493 +12.4% 2005 3.076 +16.3% 394,518 +18.3% 2006 3.64 +18.3% 450,000 +14.1% French Tax Directorate, 26 May 1999 statement Response to Dolez written question, Journal Officiel, AN, n° 33, 17 August 2004, p. 6431. 7 Add to this uninterrupted rise that ISF proceeds are much higher in France than they are on equivalent taxes in Spain (€1.2 billion) or Sweden 17 . To what factors should we attribute this outcome that has been so beneficial for France’s State - if not its taxpayers? Gross ISF proceeds may have risen constantly, but many observers criticise the levy’s collection cost, saying that it costs more than it yields, given the recovery expenses incurred by the French State. Of course, this could also be a myth – so we need to explore this question in further detail. - Low collection costs leading to high net direct yields Despite the lack of any detailed study on the ISF’s direct cost, France’s tax authorities possess a great deal of knowledge on this topic. The Angels report specifies that in 1997 the ratio of administration costs to net proceeds 18 was 1.6% in France, versus an average of 1.1% 19 in ten other countries analysed, as the following chart shows: 17 CDI Tax Directorate Report, 2004, page 52. Bernard Angels Senate Report n° 205 (1999-2000). Most data here came from a Lepin report entitled “Comparative analysis of two tax administrations”, General Finance Inspectorate Report, 1999. 19 Angels report, page 165. 18 8 Revenues counted included taxes and payroll charges, except in Germany, Spain, France, Italy and UK (taxes only) / France: Tax Directorate + CP (ACOSS dole payments excluded) Source: IGF – French Ministry of Economics, Finance and Industry. ISF collection costs lato sensu (i.e. total costs including tax base and auditing costs plus recovery costs) were calculated in the Lepine report for fiscal year 1997, totalling €18.5 million for tax base/auditing and €3.5 million for recovery, or €22 million all in all. Finance Inspector Lepine also specified that 593 people worked to administer the ISF, and that its administration cost, as a percentage of its total yield of €1.356 billion, amounted to 1.6% (or 1.36% for tax base/auditing and 0.26% for recovery), in line with average French tax collection costs of ca. 1.4% (to wit, 0.87% for tax base/auditing and 0.54% for recovery). As we can see and unlike what we hear in speeches by politicians who accuse the ISF wealth tax of all the evils of the world, the levy’s collection costs are reasonable, averaging about the same as any other tax in France. 20 Clearly Professor Christian Sainte-Etienne is being unrealistic when he affirms so overtly that, “The ISF yields €2.4 billion gross and €2.1 billion, due to its considerable administrative and verification costs”. 21 In 2006, charges probably came in at ca. €50 million, equivalent at most to 1.4% of the amount collected. The ISF’s relatively low collection costs reflect its declarative nature, as well as the type of population liable to pay this tax. 1.2 Factors explaining the rise in ISF proceeds during France’s last legislature Whereas various legislative measures have worked to (marginally) erode ISF yields, its altogether higher proceeds seem to be driven by two factors. The obvious one is the rise in property values since 1998. The other is less obvious but just as significant - the failure to raise ISF thresholds for nearly seven years has given the State a rent that was relatively painless for taxpayers. An analysis of ISF fraud will complete this analysis of possible explanatory factors. 20 It is even probable that the cost has fallen since 1997, since in the mean time ISF yields rose from €1.36 billion in 1996 to €3.6 billion in 2006. Of course, we can still question whether French collection costs are in the norm. The Table below says that they are 50% higher than in comparable countries. 21 Croissance équitable et concurrence fiscale, Council for Economic Analysis, Christian Saint-Etienne and Jacques Le Caheux, 2005, page 48. 9 - Methodology used in calculations To model the theoretical changes in ISF proceeds from 1997 to 2004, as for all the calculations found in the present text, we will be using a relationship delineated in the French Tax Council’s 16th report indicating the number of households with holdings exceeding a certain threshold. The breakdown of taxpayers by tax bracket can be expressed by the following equation: 16 500 –1/0,53 ®: N = 17 100 X P and represented by the following curve: According to the French Tax Council, this relationship ® applies to 1996, our model’s base year. Its gives a number N of households liable to pay ISF wealth tax for estates at least equal to P. Its aim is to "offer an order of magnitude of the number of households who would become liable to pay ISF wealth tax in case the threshold were lowered". 22 22 16th French Tax Council Report, Journal Officiel, 12 August 1998. 10 An Estate Distribution Law will be used as a starting point for the present text’s calculations of what the State gained from the failure to increase the threshold between 1998 and 2004. It will also help us to understand how the rise in property values from 1998 to 2005 (up 100%) forced many taxpayers with holdings below the filing threshold to lodge ISF tax form 2725, since (and this is a key element to which we will return below) holdings do not break down in a linear manner. In other words, a 10% rise in property values (the asset type that accounts for ca. 50% of the holdings of individuals found in the first ISF bracket) increases the number of persons liable to pay the tax by more than 10%. This is an exponential phenomenon that accelerates with the rise in property investments. - The impact of freezing the threshold between 1998 and 2004. Every year between 1988 and 1996 and as is logical, the ISF threshold was inflation-linked, much in the way that income tax thresholds are (the aim of such revaluations being to avoid taxing people for inflation-driven increases in nominal wealth).23 On 1 January 1997, however, the brackets were more or less crystallised, with the lowest threshold staying at €720,000 until 2003 and the 1998 to 2004 budgets failing to raise any of the other thresholds. The impact is akin to a hidden and painless extra levy (the dream for all tax authorities). Higher proceeds were automatically generated, since inflation over 1997-2004, despite remaining moderate, did reach a cumulative total of 10%. The effects of not raising the threshold did not go unnoticed by one Senator Marini, who has a well-known interest in the taxation of French estates. In his estimation, the aforementioned situation generated an extra €200 million. 24 We agree entirely with the honourable parliamentarian’s affirmation that not raising thresholds has automatically increased the number of persons liable to pay the ISF wealth tax, although his way of calculating the gain that this represents for the State (based on a rule of three including an implicit postulate about estates’ linearity) should be adjusted for methodological reasons. Like the Senator, we start with the INSEE index, using 1997 as base 100. 25 Changes in this index will allow us to readjust the threshold brackets every year to incorporate real inflation. All we have to do is 23 We will not be broaching here the intellectually interesting question of the legitimacy of re-setting the ISF threshold annually, nor the less straightforward but probably more relevant question of how rational it is to use a CPI index to assess a taxable base largely comprised of financial and property assets. 24 Marini Senate Information Report n° 351, 16 June 2004, page 21. 25 1997 is the “right” year since the failure to raise thresholds hit taxpayers after 1998. Thresholds for the 1998 ISF wealth tax calculated using estates’ 31 December 1997 values should have been up by 1.2%. 11 deduce (via linear regression) the theoretical number of taxpayers that should have been found in each bracket had the threshold been raised by the actual inflation rate 26 . 26 This calculation was made using data from 1999, 2000, 2001 and 2002 Senate reports, followed by Finance Ministry figures obtained through the French Senate in 2003 and 2004. 12 Act ual t axpayers 0.72 - > 1.16 1.16 - > 2.30 2.30 - > 3.60 3.60 - > 6.90 6.90 - > 15 > 15 t ot al 1997 92072 64511 12927 6562 2827 1998 96357 71424 14557 7265 3131 178899 192734 Cont ribut ions in € m io 0.72 - > 1.16 1.16 - > 2.30 2.30 - > 3.60 3.60 - > 6.90 6.90 - > 15 > 15 TOTAL I SF 1997 101 347 206 247 536 1998 108 385 233 275 587 1437 1588 Avg paym ent s in € m io 0.72 - > 1.16 1.16 - > 2.30 2.30 - > 3.60 3.60 - > 6.90 6.90 - > 15 > 15 1997 0.00109697 0.00537893 0.01593564 0.03764096 0.18960028 Cum ulat ive nr. t axpayers 1997 178899 86827 22316 9389 2827 0 0.72 - > 1.16 - > 2.30 - > 3.60 - > 6.90 - > > 15 1.16 2.30 3.60 6.90 15 Ln( est at e) 1999 103055 80586 16460 8394 2548 965 212008 2000 113209 96121 20204 10629 3261 1232 244656 2001 124594 105474 22148 11401 3530 1301 268448 2002 134083 109547 22485 10756 3361 1202 281434 2003 148018 112759 21858 10042 3044 1074 296795 2004 160349 131442 25445 11710 3370 1177 333493 1999 118 430 260 312 252 415 1787 2000 134 518 321 399 331 534 2237 2001 147 567 351 426 353 551 2395 2002 153 579 355 397 317 468 2269 2003 186 591 343 370 264 402 2156 2004 186 691 399 431 313 420 2440 1998 0.00112083 0.00539034 0.01600605 0.03785272 0.18748004 1999 0.00114502 0.00533591 0.01579587 0.03716941 0.0989011 0.43005181 2000 0.001183651 0.005389041 0.015887943 0.037538809 0.101502607 0.433441558 2001 0.001179832 0.005375732 0.015847932 0.037365143 0.1 0.423520369 2002 0.001141084 0.005285403 0.015788303 0.036909632 0.094317168 0.389351082 2003 0.0012566 0.00524127 0.0156922 0.03684525 0.08672799 0.37430168 2004 0.00115997 0.00525707 0.01568088 0.03680615 0.09287834 0.35683942 1998 192734 96377 24953 10396 3131 0 1999 212008 108953 28367 11907 3513 965 2000 244656 131447 35326 15122 4493 1232 2001 268448 143854 38380 16232 4831 1301 2002 281434 147351 37804 15319 4563 1202 2003 296795 148777 36018 14160 4118 1074 2004 333493 173144 41702 16257 4547 1177 Ln( Cum ulat ive N) Ln( Cum ulat ive N) Ln( Cum ulat ive N) Ln( Cum ulat ive N) Ln( Cum ulat ive N) Ln( Cum ulat ive N) Ln( Cum ulat ive N) Ln( Cum ulat ive N) 13 Act ual t axpayers 0.72 - > 1.16 1.16 - > 2.30 2.30 - > 3.60 3.60 - > 6.90 6.90 - > 15 > 15 t ot al 1997 92072 64511 12927 6562 2827 1998 96357 71424 14557 7265 3131 178899 192734 1997 101 347 206 247 536 1998 108 385 233 275 587 1437 1588 Avg paym ent in € m io 0.72 - > 1.16 1.16 - > 2.30 2.30 - > 3.60 3.60 - > 6.90 6.90 - > 15 > 15 1997 0.00109697 0.00537893 0.01593564 0.03764096 0.18960028 Cum ulat ive t axpayers 0.72 - > 1.16 1.16 - > 2.30 2.30 - > 3.60 1997 178899 86827 22316 Paym ent s in €m io 0.72 - > 1.16 1.16 - > 2.30 2.30 - > 3.60 3.60 - > 6.90 6.90 - > 15 > 15 TOTAL I SF 1999 103055 80586 16460 8394 2548 965 212008 2000 113209 96121 20204 10629 3261 1232 244656 2001 124594 105474 22148 11401 3530 1301 268448 2002 134083 109547 22485 10756 3361 1202 281434 2003 148018 112759 21858 10042 3044 1074 296795 2004 160349 131442 25445 11710 3370 1177 333493 1999 118 430 260 312 252 415 1787 2000 134 518 321 399 331 534 2237 2001 147 567 351 426 353 551 2395 2002 153 579 355 397 317 468 2269 2003 186 591 343 370 264 402 2156 2004 186 691 399 431 313 420 2440 1998 0.00112083 0.00539034 0.01600605 0.03785272 0.18748004 1999 0.00114502 0.00533591 0.01579587 0.03716941 0.0989011 0.43005181 2000 0.001183651 0.005389041 0.015887943 0.037538809 0.101502607 0.433441558 2001 0.001179832 0.005375732 0.015847932 0.037365143 0.1 0.423520369 2002 0.001141084 0.005285403 0.015788303 0.036909632 0.094317168 0.389351082 2003 0.0012566 0.00524127 0.0156922 0.03684525 0.08672799 0.37430168 2004 0.00115997 0.00525707 0.01568088 0.03680615 0.09287834 0.35683942 1998 192734 96377 24953 1999 212008 108953 28367 2000 244656 131447 35326 2001 268448 143854 38380 2002 281434 147351 37804 2003 296795 148777 36018 2004 333493 173144 41702 14 Act ual t axpayers 0.72 - > 1.16 1.16 - > 2.30 2.30 - > 3.60 3.60 - > 6.90 6.90 - > 15 > 15 t ot al 1997 92072 64511 12927 6562 2827 1998 96357 71424 14557 7265 3131 1999 103055 80586 16460 8394 2548 965 212008 2000 113209 96121 20204 10629 3261 1232 244656 2001 124594 105474 22148 11401 3530 1301 268448 2002 134083 109547 22485 10756 3361 1202 281434 2003 148018 112759 21858 10042 3044 1074 296795 2004 160349 131442 25445 11710 3370 1177 333493 178899 192734 Cont ribut ions in €m io 0.72 - > 1.16 1.16 - > 2.30 2.30 - > 3.60 3.60 - > 6.90 6.90 - > 15 > 15 TOTAL I SF 1997 101 347 206 247 536 1998 108 385 233 275 587 1588 1999 118 430 260 312 252 415 1787 2000 134 518 321 399 331 534 2237 2001 147 567 351 426 353 551 2395 2002 153 579 355 397 317 468 2269 2003 186 591 343 370 264 402 2156 2004 186 691 399 431 313 420 2440 1437 Avg. paym ent in €m io 1997 1998 1999 2000 2001 2002 2003 2004 15 As we can see, the net effect is clear: in FY 2004 alone, not raising the threshold was worth €400 million to the Treasury (15% of total ISF proceeds), a sum that by itself explains much of the rise in yields. - The not insignificant cost of fraud. ISF fraud can mean anything ranging from potential taxpayers failing to file returns, attempts to dissimulate property held abroad or people under-assessing their holdings, in particular property or unlisted securities. To exculpate some culprits, note how hard it can be to come up with accurate estimates. Conscious of this, French lawmakers have implemented a 10% legal tolerance, specifying that, “No penalty will be levied when the shortfall of sums declared does not exceed 10% of the actual tax base.” 27 To ascertain how fraud affects ISF yields, we have, using the aforementioned model, calculated the amount due after adjusting for under-assessment bias and integrating three hypotheses: an under-assessment of 10% by all taxpayers; an under-assessment of 20%; and (most realistically) varying levels of under-assessment depending on the bracket. In line with the Estates Financialisation Law, 28 the larger an estate, the greater the share of securities held. What we can reasonably infer from this is that taxpayers belonging to the lower ISF brackets, whose estates feature a higher proportion of property assets, are more likely to file fraudulent returns than upper bracket taxpayers, whose holdings are massively invested in listed securities they are forced to declare at their legal value29 , something that clearly restricts the potential for fraud. In the latter model, we assume an under-assessment of 20% for the first two brackets; 15% for the third; 10% for the fourth; and 5% for the last two. For each of these scenarios, we have carried out simulations covering 1999 to 2004 using the above model. The outcomes were remarkably stable over this period: in the first hypothesis (10% under-assessment), the shortfall was slightly below 20% of proceeds, or ca. €486 million in 2004. 30 In the second hypothesis (down 20%), the loss of tax revenues equalled 27 French Tax Code Article 1733. See further details below in Section II.1. 29 Article 885 T bis of the French Tax Code, derived from article 26 V in Law n° 88-1149 dated 23 December 1998 becoming the 1999 Budget, stipulates “Listed securities are to be assessed using the last known price or the average of the 30 latest closing prices preceding the day of taxation”. 30 By the by, and asides from any under-assessment of goods, these results give us a sense of the volatility of a tax that is highly sensitive to estate variations caused by fluctuations in the property and equity markets. 28 16 48% of total proceeds, or €1 billion in 2004. In the 3rd hypothesis, fraud cost the State €680 million in tax revenues in 2004, or ca. 27% of all ISF proceeds. In this latter instance, the model estimates that nearly 220,000 households did not declare the ISF wealth tax they theoretically owed. This came on top of major fraud, i.e. people aware that they were supposed to pay the ISF wealth tax but did not file a return. All in all, we can conservatively estimate total ISF fraud at a minimum of €700 million in 2004, or 28% of the total intake. As such, ISF fraud has much more than a symbolic cost, since it clearly reduces the revenue that the State might otherwise expect, at least in an ideal world where every taxpayer behaved like a model citizen. This is why net ISF proceeds are so far below their theoretical value – especially when we include the shortfalls that the State suffers due to the expatriation of so many of its (larger) taxpayers. II The economic cost of the ISF wealth tax “I am sustained in my action by the possibly false hope that truth will ultimately emerge, even if causes a few problems at first.” Paraphrasing of John Maynard Keynes Capital flight caused by the ISF wealth tax is a reality that goes a lot further than stories about France’s celebrity tax refugees. We may not be able to measure this precisely, but we can try to ascertain the real effects of this human flight, and above all the consequences for the country’s tax revenues and general economic situation. 2.1 An unquestionable capital flight To analyse and comprehend how ISF wealth tax proceeds have evolved over the course of the past legislature, our model will use the R relationship together with the breakdown in the proportion of financial assets owned by households liable to pay this tax, expressed as a percentage of total 1996 holdings. The findings will substantiate the validity of the aforementioned (Large) Estate Financialisation Law. - The (Large) Estate Financialisation Law 17 This law, which is key to understanding the ISF wealth tax’s economic implications, can be summarised as: “The bigger the estate, the higher the proportion of financial assets”. This is perfectly illustrated by the Table below: Threshold for first bracket (€0.72 mio to €1.16 mio) 50% For 2nd bracket (€1.16 mio to €2.31 mio) 55% For 3rd and 4th brackets (€2.31 mio to €6.95 mio) 75% For 5th and 6th brackets (above €6.95 mio) 85% Source: 16° Report, French Tax Council We can hypothesise that the proportion of property to financial assets has been relatively stable over the period in question (1996-2004), although there has been a rise in property holdings in the first bracket (due to main residences’ increased proportion as part of taxpayers’ overall holdings, up from 11.6% in 2002 to 15% in 2005), spurred by the rise in property values and by the fact that the estate of some of the new persons liable to pay the ISF wealth tax is mainly comprised of their main residence. Breakdown of taxable bases comprising the first bracket of the 2005 ISF wealth tax 11.7% 23.1% 14.0% 21.2% 28.4% 1.6% Main residence Other property Securities Cash Benefits Other chattels Source: Ministry of Economics, Finance and Industry. 18 - ISF proceeds model since 1997: the tax should have yielded more and affected more people To study the ISF’s theoretical changes between 1996 and 2004, we have tracked the Paris Notaries Index (over this one period, since Paris property values have not always mirrored overall French residential property prices) as well as the CAC-40 stock index (see below). Prices changes in Paris property and CAC-40 stock index between 1997 and 2005 50 40 43 35 36 30 23 (%) 20 13 10 12 9 11 4 16 16 14 11 7.4 Changes in CAC-40 0 -10 Annual variation in Paris housing prices -2 -5 -20 -22 -30 -34 -40 1997 1998 1999 2000 2001 2002 2003 2004 2005 Source: Association of Paris Notaries To assess the amount of ISF wealth tax that should have been be paid from 1997 to 2004, we have used a conservative hypothesis to represent higher financial asset prices, equal to only one-half the rise in the CAC-40 (a reasonable figure, since some asset portfolios also include bonds) and only one-half the rise in the Notaries Index for property assets (here, on the other hand, we used the lowest available figures). All in all, our model applies a +10% change in financial assets in 1996 (instead of +20%), +10% in 1997 (vs. +30%) followed by +12% in 1998, +20% in 1999, +0% in 2000 (to incorporate dividends and fixed income returns), before a 10% fall in 2001 and another drop of 15% in 2002, followed by a rise of 8% in 2003 and 5% in 2004. As for taxpayers’ property assets, they were valued as being down 5% in 1997, stable in 1998, followed by an annual rise of 5% in 1999, 2000, 2001, 2002 and 2003 and +10% in 2004 31 . 31 As per Paris Notaries Index. 19 The model tells us that the difference between the theoretical and actual ISF yield oscillated between €500 million and €1 billion, depending on the year in question. As for the number of persons liable to pay the tax, given the rise in property values and the distribution of estates in France, there should have been twice as many individuals in the first bracket but “only” 20% to 30% more in the upper brackets (c.f. last Table) - the exception being the very highest brackets, where there was a clear tax evasion effect. 32 The explanation for this variance lies in the fraud committed in the lower ISF brackets, with many taxpayers omitting to file a return or strongly under-assessing their property. In the higher brackets, the culprit was tax expatriation. -Estimated number of taxpayers and volume of capital to have left France since the ISF’s enactment in 1988. Ministry of Finance statistics on this subject are only useful insofar as they indicate a trend. From 1997 to 2003, official figures mention the delocalisation of one taxpayer a day, followed by a sharp increase after 2003, reaching 650 departures daily in 2005, or a twofold rise in expatriations in three years. Officially, around two taxpayers leave France daily due to the ISF wealth tax 32 Even if we offer a very conservative estimate of the rise in estates over the period in question, it is noteworthy that even though taxpayers are mainly focused in the first ISF bracket, this category only accounts for a small share of total ISF revenues. According to the 2004 French Tax Council Report (page 146), more than 50% of all persons liable to pay the ISF account for less than 7% of its total yield. One key element is the extreme concentration of holdings, with the wealthy (to use class-laden semantics) constituting anything but a homogeneous class: 1.6% of all taxpayers pay 35% of all ISF revenues; 0.5% pay 23%. Inversely, the 47.6% constituting the first ISF bracket only pay 7% of total ISF proceeds. 20 700 650 600 550 500 450 400 350 300 250 200 649 568 370 363 350 363 384 383 1997 1998 1999 2000 2001 2002 368 2003 2004 2005 Source: Ministry of Economics, Finance and Industry More importantly, Ministry figures 33 suffer from methodological shortcomings that irremediably disqualify them due to their unjustified minimisation of the number of departures, and above all, the amount of capital exported. This is because these calculations show little understanding of participants’ profiles. Many tax refugees leave France to avoid both capital gains taxes and the ISF wealth tax. As we know, a company director who is not liable to pay the ISF if his/her estate ex-business assets is worth less than the tax threshold, but who could sell the firm for a good price, has an objective interest in first moving to Belgium (in which case, s/he will not show up in the tax refugee statistics) before selling the company, thereby avoiding France’s 27% capital gains tax. Moreover, if this person is the company’s founder, as is often the case, the capital gain achieved is usually fairly close to the firm’s actual valuation. The director can then move definitively to Belgium, which has no wealth tax. This situation, amplified since the elimination of the “exit tax” 34 (which in our opinion has acted as a powerful driver for fiscal expatriation), is in no way incorporated in official national statistics that would otherwise soar by a factor of five or even ten. 35 According to the French Tax Directorate, the most popular destinations for individuals leaving France because they are liable to pay the ISF wealth tax are Belgium for 18% of all 33 The Ministry of Finance estimates that 35,000 persons leave France every year, mostly for professional reasons. The Tax Directorate has recently developed an internal observatory to garner information on ISF refugees. A Tax Centre was built in 1999 to track individuals leaving France for fiscal reasons, before informing the aforementioned observatory. 34 See State Council decree dated 10 November 2004: “Lasteyrie du Saillant”. 35 A simple survey of the main wealth management actors serving expatriate company directors who are about to sell their firm indicates that these figures are far too low. 21 fiscal expatriates (with average taxable holdings of €3.6 million, including 83% in property) Switzerland for 16% (with average taxable holdings of €9 million, including 91% in securities); and then the US and the UK (each accounting for 12%, with average taxable holdings of €2.5 million, including 70% in securities). Depending on the Swiss bankers we asked, 1,200 wealthy French persons move to Geneva Canton every year (or ten times official Ministry of Finance figures) and around 1,000 to Vaud Canton (Lausanne). The total number of French persons who have moved to Switzerland for tax reasons is around 12,000 in Geneva and 10,000 in the Valais region. Based on an average estate of €5 million - or half the figure given by the French Tax Directorate (excluding business assets) when valuing the taxable holdings of ISF liable individuals who have moved to Switzerland 36 - we find 20,000 households in Switzerland alone, amounting to total assets of €100 billion. A reasonable number would therefore be twice this amount, or the flight of a total €200 billion since the ISF was born. 37 2.2 Assessing expatriation’s fiscal and economic costs These expatriation waves have well-known causes. In our opinion, they are threefold: the Juppé government’s 1995 decision to cap the levy’s cap; the creation of a 6th bracket with a rate of 1.8% (superseding net returns on money market investments at the time, after deducting 27% in withholding taxes); and paradoxically the exit tax 38 , whose abrogation we view as a powerful driver of tax expatriation, given the new wave of departures triggered after the exit tax was eliminated to achieve European legal harmonisation (Ministry of Finance statistics attest to a doubling of “official” departures between 2003 and 2002). Expatriation has two costs: fiscal, being the shortfall in yields caused by taxpayers’ departure (reduction of the tax base); and economic, being the induced fall in GDP due to the fact that capital is being invested abroad, where it creates jobs, etc. All of which raises questions as to how we should analyse this “double punishment” that the country is inflicting on itself? 36 2001 Ministry of Finance data reproduced in the French National Assembly’s 2006 Dassault report n° 3135 on “France’s impoverishment due to the expatriation of estates and taxpayers”, page 17. 37 c.f. Luxemburg’s Wealth Tax Elimination Bill (November 2005), 63,000 French tax refugees live in Belgium. 38 2005 Provisional Budget, article 19. 22 - The shortfall in tax revenues The exercise that consists of estimating the fiscal shortfall caused by expatriation is a highly perilous one. It is illusory to hope to obtain from the Finance Ministry any precise information on this topic, especially since it is an area where a distinction must be made between the direct shortfall (the ISF wealth tax that should have been collected had the taxpayers stayed in France) and the indirect shortfall caused by the fact that these expatriates no longer pay income, residential or inheritance tax in the country. Since they have to reside for more than 183 days per annum outside of France’s national borders to enjoy non-resident status, they no longer consume as much in France, creating shortfalls in the country’s income tax yields (starting with the capital gains that originally led to their departure), VAT, business tax, etc. These losses are all the greater because the taxpayers involved have high incomes and therefore spend more, given their “far above-average payment capabilities”. On the face of things, we would be reaching if we pretended we could estimate the magnitude of the tax shortfall caused by the expatriation of large estates. We can, however, attempt an initial approximation. We know that the relationship between the State’s tax revenues in 2003 (around €220 billion) and total household holdings in France, assessed by the INSEE that same year at €7,103 billion (€4,381 billion in property assets and €2,722 billion in financial assets), produces a ratio of 3%. Applying this to the total €200 billion in wealth expatriated, we get a loss of proceeds for the State that is somewhere around the €6 billion mark. These estimates of the State’s fiscal shortfall for all taxes combined are lower than the ones calculated by Patrick Artus, Research Director at IXIS-CIB, who recently affirmed that, “The ISF wealth tax costs twice as much in VAT gone missing as it actually yields”. 39 - Evaluation of how the ISF wealth tax affects the country’s economic activity Capital flight is far from being anecdotic in France. In economics, many empirical studies have indicated a strong correlation between savings and investment (itself a wealth-creating mechanism), and jobs and growth. Economists often use a Cobb-Douglas function (1928) to model a country’s output depending on the production factors like capital and labour. 39 15 February 2007 interview in Le Monde newspaper. 23 where Y represents GDP; K the amount of capital; L the amount of labour; and c, α and β constants that reflect the state of technology. It is not easy, with this kind of simple framework, to assess with any precision the economic impact on a country like France of the expatriation of €200 billion. 40 What is certain is that lasting capital flight implies investment behaviour whose effects will be felt in the vicinity of the saver’s geographic location. There is also the converse implication that expatriates who make direct or indirect investments abroad will invest less domestically. As noted by Senator Michel Charzat 41 , capital expatriation is “one of the key elements likely to shift a firm’s centre of gravity abroad” – especially in light of the competitive fiscal environment that developed countries face today. 42 Over a longer period, it is likely that the €200 billion that non-resident French persons have sent abroad because of the ISF wealth tax will deprive France of tens of thousands of jobs, explaining some of the growth differential between France, England and Germany. Note that a shortfall equivalent to 0.2% of GDP is basically tantamount to cutting €3.5 billion from France’s national wealth – all of which gives us an idea of the ISF wealth tax’s real cost. Faced with this reality, and unable to kill off this tax (the only way to prevent human and capital flight), some French politicians have been trying to soften its blow. One proposal that keeps arising is a total or partial exemption of taxpayers’ main residence.43 Our final calculation will involve ascertaining the fiscal impact of such a measure. - The effects of a possible exemption of main residences We know that “small” ISF taxpayers, whose assets are mainly held in property and not in financial instruments (and who would therefore have problems moving abroad) have been hit by the recent rise in property values. The share of main residences in these individuals’ estates rose from 11.6% to 15% between 2002 and 2005. The impact of the ISF wealth tax should be 40 Research by economists like Bagehot, Obstfeld, Harrod, Domar and Kaldor, highlighting the importance of savings in economic growth phenomena, could be applied to initiatives of this sort. 41 Charzat Report on Territorial Attractiveness, at the behest of French Prime Minister, July 2001. 42 “Entrepreneurs are people who have shown that they know how to create wealth and jobs. Because of the ISF wealth tax, they will now be creating jobs elsewhere. Britain and Switzerland try to attract wealthy people from other countries. Our challenge is to do the same”. 29 May Economic and Social Council statement by Mr. Payre. 43 Main residences benefit from a 20% abatement. This stipulation is Praetorian in origin and became law in France through the 30 December 1998 Provisional Budget n° 98-1266. It is codified in article 885 S of the French Tax Code. 24 relativised for this group, however - in the most cases, it acts as a sort of second property tax that costs around €1,000 to €2,000 per annum. According to estimates we made using the aforementioned R relationship, if we hypothesised that the main residence represents around 45% of all holdings of individuals in the 1st ISF bracket, 30% in the 2nd, 15% in the 3rd, 10% in the 4th, 7.5% in the 5th and 5% in the 6th and final, this decision would lead to a threefold reduction in the number of persons liable to pay the ISF, auguring a fall from 333,000 individuals (using 2004 figures) to 117, 0000. At the same time, ISF proceeds would only fall by €880 million. Naturally, a decision of this sort would further reinforce the ISF’s concentration, whilst leaving the causes of large estates’ expatriation unaffected. CONCLUSION “The shortest mistakes are always the best.” MOLIERE, L’Etourdi, IV, 3. Following this analysis, we will have learnt a few noteworthy lessons about the ISF wealth tax’ main faults. The ISF’s much higher proceeds over the course of the past French legislature mainly translate rising property values since 1998 and are a consequence of the tax’s threshold not being index-linked over the period 1998-2004. One result is that a further 100,000 taxpayers now find themselves in the lower two brackets. Another is that the average tax levied has dropped. We may have to reject politicians’ occasional argument that the ISF wealth tax suffers from high collection costs (in reality, these are very reasonable at below 2%) but a detailed analysis of the ISF’s structure confirms that the strong concentration of taxpayers in the upper two brackets constitutes a strong fiscal incentive to leave France when one’s estate exceeds €10 million (a point at which the annual ISF wealth tax bill is around €100,000 euros, tantamount to a gross annual salary of €200,000 euros) and especially €15 million (with taxable holdings of €15 million, the annual levy is ca. €200,000, equivalent to gross annual wages of €400,000). 25 According to our estimates, the Ministry of Economics has very largely under-estimated the capital flight caused by this (I)ncentive to (S)cram from (F)rance. In all likelihood, something like €200 billion has already left the country. This means that the Treasury is losing ca. €7 billion annually or around twice total ISF proceeds. In addition, in terms of investment, job creation and consumption, the French economy is suffering losses that are hard to estimate but very certainly exceed the direct fiscal shortfall, substantiating ex-Prime Minister Raymond Barre’s statement that the ISF is like an economic leukaemia for France. This symbolic, mythical and expiatory ISF wealth tax – accepted by 98% of the population, basically by people who do not pay it, complicates the life of small taxpayers because it creates the flight of large estates. The fact that it costs more than it yields engenders a paradoxical situation in which all of France’s other taxpayers, including its least wealthy citizens, must bear the brunt of its overall tax burden, notably by paying more in VAT. The outcome of the next presidential and legislative elections is likely to have no effect on the alchemy of this tax and the strange potlatch in which our fellow citizens engage every 15 June will continue to produce its inevitable side effects. The tax exodus will continue, adding year in year out to the burden of France’s least affluent taxpayers, who will have to compensate for the absence of wealthier expatriates - the sad reality of an (I)diotic and (S)tupid (F)rench tax that other countries view with a hint of pity and a great deal of incomprehension.