Dividend “tax leakage” is a concern for any investor in collective investment vehicles (CIVs) that own shares in overseas equity markets. Tax leakage occurs when investors in a fund are forced to suffer withholding taxes on dividends from the underlying shares at a higher rate than would have applied if they had purchased those shares directly.
For a pension fund investor, for example, double tax treaties often allow for the rate of withholding tax on overseas equities to be reduced to zero. Accessing foreign equities via a pooled fund may make great practical sense for an investor, as this enables portfolio diversification via a single purchase. However, to the extent that a fund suffers withholding taxes that cannot be reclaimed on its overseas equity income, there is an automatic drag on the returns received by the end-investor.
Collective investment vehicles may be able to gain access to some tax relief under double taxation treaties (in other words, dividend withholding tax rates may be reduced, if not eliminated completely). However, according to Paul Radcliffe, senior vice president in Citi’s Tax Products and Transactions group, “claiming double taxation treaty (‘treaty’) benefits by collective investment vehicles can be a vexed subject. Treaty interpretation can differ from source country (the country in which the income arises) to source country; the treaty claim process can be laborious, unclear and paper-intensive, with treaty forms being required to validate a claim; and the layout and information required by such forms may differ, based on each source country’s requirements.” 1
For passive, benchmark-tracking vehicles such as exchange-traded funds (ETFs) and index funds, dividend tax leakage is a particular problem. Tracker funds aim to do what their name suggests—track their underlying indices. However, any lost dividend income leads directly to an unrecoverable tracking error between the fund and its index benchmark (assuming that the benchmark is based on the receipt of dividends before any deduction for withholding taxes). This tracking error is incurred over and above the more familiar underperformance that results from portfolio management expenses.
According to David Blitz, Joop Huij, and Laurens Swinkels2, “European index funds and exchange-traded funds underperform their benchmarks by 50 to 150 basis points per annum. The explanatory power of dividend withholding taxes as a determinant of this underperformance is at least on par with [that of] fund expenses.”
In practice, index providers do factor in the effect of dividend withholding taxes into their return calculations, by publishing both “gross total return” (i.e., pre-withholding tax) and “net total return” (post-withholding tax) versions of international equity indices.
In turn, index fund and ETF managers usually track the post-tax, net total return, rather than the zero tax (gross total return) versions of their benchmarks.
In this article we seek to quantify the dividend tax leakage that has occurred over the last ten years in seven popular index benchmarks. Dividend tax leakage is defined as the proportion of the gross dividend income forgone by an investor receiving only the net, post-withholding tax index return. We look at the trends in tax leakage and the absolute levels of leakage from the indices surveyed. To get a feel for the importance of dividend tax leakage as a cost faced by the end-investor, we compare average tax leakage with funds’ expense ratios. Finally, we examine the withholding tax rate assumptions used by index providers and question their appropriateness.
Tax Leakage Quantified In Popular Benchmarks
We examined the dividend tax leakage, defined as the difference in return between the net total return and gross total return versions, for seven selected index benchmarks. Return data were collected for the ten calendar year period 2001-2010. Indices were chosen from the ranges offered by the two largest index providers in the European ETF market: MSCI and STOXX (according to BlackRock’s “ETF Landscape: Industry Highlights” publication, these two firms have a combined 41% market share in Europe at end-February 2011, as measured by the assets under management in related ETFs).
The seven indices surveyed included two with exposure to global equities (MSCI World and MSCI Emerging Markets) and five with intra-European cross-border exposure (five supersector indices from the STOXX Europe 600 range, covering banks, basic resources, oil and gas, telecoms and utilities).
For the purposes of comparison, a price index return series (i.e. the index return without any dividend reinvestment) was also calculated in each case.
For investors in funds tracking the MSCI World index (see Figure 1a), practically all the returns over the last ten years have come from dividend income (the price index rose by a mere 0.47% per year for the ten year period 2001-2010). Investors receiving full (gross) dividends from the MSCI World index’s components achieved a return of 2.82% a year, resulting in a total period return of 32%. Investors receiving net (post-tax) dividends achieved a return of 2.31% per annum, resulting in a total period return of 25.63%.
Over ten years, the annual average tax leakage on the MSCI World index was 22%. The degree of tax leakage reflects (a) the country composition of the index over time and (b) the differing withholding tax rates4 applied by the index provider to equities from those countries.
The MSCI Emerging Markets index (see Figure 1b), which underlies some of the world’s largest ETFs, provided handsome capital gains over the last decade, meaning that dividend income was perhaps of a lesser concern to end-investors than price appreciation. Nevertheless, gross dividends from the emerging markets index’s constituents added 3.05% a year to the price index over the ten year period (exceeding in percentage terms the income stream from developed market stocks, as measured by the MSCI World index).
Tax leakage in the net version of the MSCI Emerging Markets index was half that experienced by investors in the MSCI World index (11% per annum over ten years in MSCI Emerging Markets, compared with 22% for the MSCI World).
Why was dividend tax leakage less of a concern for emerging markets investors than for investors in global developed equity markets?
The withholding tax rates applied by MSCI to dividends from equities in the seven countries with the largest weightings in its Emerging Markets index are as follows: China (0% or 10%, depending on the type of share); Brazil (0%); Korea (22%); Taiwan (20%); India (0%); South Africa (0%); Russia (15%).
By contrast, the MSCI World index has one very large component, the US market, with a 49% weighting at the end of 2010. MSCI applies the maximum, 30% withholding tax rate to all US equity dividends when compiling its net index version.
In other words, developed markets tax dividends more heavily, on average, than emerging markets.
The absolute annual dividend tax leakage in MSCI’s two indices is represented graphically in Figure 1c. For the MSCI World index in particular, the tax leakage has grown over the decade. The jump in leakage for both indices in 2008/09 is explainable by the fall in equity market values: as share prices fell, dividend yields rose, and dividend taxation therefore consumed a greater absolute proportion of an investor’s total return.
Investors in STOXX’s Europe 600 banks index, a widely used benchmark, suffered a price drop of over 50% during the last decade (see Figure 2a). However, recipients of net dividends clawed back part of this loss, achieving a return of -33%, while for those receiving full (gross) dividends, losses dropped to -29% for the whole ten year period. Overall, gross dividend income from Europe’s banks added 3.59% a year to an investor’s return during the period.
The tax leakage represented by a move from gross to net dividends was 19% per year for the decade, the largest drop in income for any of the five STOXX Europe 600 supersector indices we surveyed. As shown in Figure 2b, the average annual dividend tax leakage in recent years has been around 50 basis points, although there was a much larger gap between gross and net index returns in 2001.
Although gross dividends added 3.3% a year to the price return for an investors in the STOXX Europe 600 Basic Resources index (see Figure 3a), a similar level of yield to that received by investors in the bank index for the same period, dividend tax leakage for basic resources investors was substantially lower than for investors in the STOXX Europe 600 banks index.
The difference is explained by variations in the withholding tax rates applied by STOXX for companies incorporated in different European countries. Dividends from UK-domiciled entities have a zero withholding tax rate, compared with 26.38% and 25% for dividends from German and French companies, respectively.
While there are many German and French companies in the STOXX Europe 600 banks index, the top four companies in the basic resources index (Rio Tinto, Anglo American, BHP Billiton and Xstrata) are all UK companies, meaning that their dividends attract no taxation in STOXX net return index version.
The absolute annual level of dividend tax leakage in the STOXX Europe 600 basic resources sector has fallen over the last decade as a result of the growth in share prices and the resulting fall in dividend yields (see Figure 3b).
Dividend income represented all the positive return achieved by investors in the STOXX Europe 600 Oil and Gas index over the ten-year period 2001-2010, since the index’s price level ended the decade slightly lower than where it began (see Figure 4a). Gross average annual dividend income added 3.61% a year to the price return over the ten years, while net income added 3.12%.
Absolute annual tax leakage (see Figure 4b) was on a rising trend for the decade, a reflection of the steady rise in yields from shares in the sector.
Telecoms stocks have become a staple of yield-seeking investors’ portfolios and, over the ten years under review, gross dividend income reduced a 46% loss in share prices from the STOXX Europe 600 telecoms index’s components to a more bearable 20% loss (see Figure 5a). With the receipt of net dividends, the loss for the decade was just under 25%.
The percentage tax leakage for the decade for investors in the net total return index was 16%. On an absolute basis, the leakage has been increasing steadily (see Figure 5b) and in 2009 exceeded 1%. The zero difference between gross and net telecoms index returns in 2003 in the chart reflects the almost universal cuts in dividends from sector constituents following the collapse of the technology bubble in 2000-2002.
Finally, for another income-producing equity sector, utilities, dividend leakage has also become a major concern.
The STOXX Europe 600 utilities index produced only a 3% price return for the whole 2001-2010 decade, but for an investor receiving gross dividends, the return figure jumped to a much more respectable 57% (see Figure 6a). Net of taxes, investors gained nearly ten percentage points less.
Over the ten years, the dividend leakage suffered by an investor in the net total return index averaged 15%. However, on an absolute basis (see Figure 6b), the return forgone by an investor receiving the post-tax dividend rate has exceeded 100 basis points a year in both of the last two years: a function both of the recent rise in dividend yields and of the index make-up, with the largest three companies (Germany’s E.ON, France’s GDF Suez and Italy’s ENEL, which collectively represent 36% of the benchmark in March 2011) all suffering a dividend withholding tax rate of 25% or more.
Tax Leakage Compared With Fund Expense Ratios
How important a cost is dividend tax leakage when compared with the stated headline cost of ETFs? In table 7, below, we calculate for all seven indices surveyed above, a simple average of annual absolute return differences between gross and net index versions over the ten year period 2001-2010. We then compare this annual average tax leakage with the average total expense ratio of all European exchange-traded funds currently tracking the relevant index.
Only in the case of ETFs tracking the MSCI Emerging Markets index does the average exchange-traded fund expense ratio exceed the average annual “cost” of dividend tax leakage over the last ten years. For all the other indices surveyed, the cost of tax leakage exceeded the average fund expense ratio. For two of the STOXX Europe 600 supersector indices surveyed—banks and utilities—the ten year average cost of tax leakage was more than double the current average ETF expense ratio.
Dividend tax leakage is a highly significant potential cost to investors, in other words.
Are The Index Dividend Tax Rates Appropriate?
A rule of thumb is that index providers choose to assume the worst possible dividend tax outcome when calculating their net total return indices. For example, MSCI explains in its index calculation methodology publication that “this [net total return] series approximates the minimum possible dividend reinvestment. The dividend is reinvested after deduction of withholding tax, applying the rate to non-resident individuals who do not benefit from double taxation treaties. MSCI uses withholding tax rates applicable to Luxembourg holding companies, as Luxembourg applies the highest rates.”
Is this assumption appropriate for European index and exchange-traded funds?
In several cases, it is possible for funds to achieve a better tax outcome than the index providers’ net total return index calculation implies. ETFs domiciled in France, for example, receive dividends from French companies without any deduction of withholding tax, and the same is true for German-domiciled ETFs receiving dividend income from German companies.
Notwithstanding Paul Radcliffe’s earlier comments about the difficulties faced by collective investment vehicles when reclaiming dividend taxes, in certain cases it is possible for them to do so. In several European countries, for example, the 30% withholding tax rate on dividend distributions from US companies can be reduced to 15% by applying the double tax treaty between the US and the relevant jurisdiction.
The ability to gain treaty access may be dependent on the fund operator being able to demonstrate to the US authorities that a minimum percentage of investors in the collective investment vehicle is resident in the country concerned (51%, for example, in the case of the US double tax treaty with Ireland). For many Irish-domiciled funds sold to investors from across the region, this percentage of local ownership is unlikely to be reached and so treaty access is not guaranteed.
Given that the dividend yield on the S&P 500 index is currently around 2%, funds with access to the double tax treaty can receive a 1.7% post-tax rate of income rather than the 1.4% rate that the worst case tax outcome would imply—a 30 basis points return difference a year and a differential that would build up to significant extra performance over the longer term for investors with treaty access.
Securities lending is often used to “optimise” post-tax dividend rates, particularly within Europe. For example, according to the “Introduction to Securities Lending” published by ISLA, “an offshore lender that would normally receive 75% of a German dividend and incur 25% withholding tax could lend the security to a borrower that, in turn, could sell it to a German investor who was able to obtain a tax credit rather than incur withholding tax. If the offshore lender claimed 95% of the dividend, it would be making a significant pick-up (20% of the dividend yield).”5
ETF and index providers may make use of such tax arbitrage “earnings” to improve fund performance, but they are under no contractual obligation to credit any such earnings to their funds if they track an index version that assumes a worse tax outcome.
In summary, to the extent that an ETF or index fund is tracking a net total return index that assumes the worst possible dividend tax outcome (i.e. no relief under double tax treaties, no gross receipt of “domestic” dividends), that index return represents an extra cost to fund investors if the fund operator is in practice able to receive dividends at a better post-tax rate.
The impact of dividend tax leakage on index investors’ long-term returns can be significant. The extent of tax leakage varies widely by index type, according to the representation of different countries within the index’s constituent list, and depending on the levels of pre-tax dividend yield. In a survey of seven popular international equity indices, tax leakage represented a greater cost to investors than fund expense ratios in all but one case. Index providers’ assumptions of dividend withholding tax rates typically imply the worst possible outcome for investors. In practice, funds may achieve a better post-tax return, although fund managers are under no obligation to pass on any improved tax rate.
3 Defined as 1-((net index return-price index return)/(gross index return-price index return))
4 The index withholding tax rates used by MSCI are specified in http://www.mscibarra.com/, pp 51-52. For STOXX, see http://www.stoxx.com/indices/taxes.html
5 See http://www.isla.co.uk/uploadedFiles/Publications/intro-to-securities-lending-v1-chapter3.pdf, pp 28-29