Printer Friendly


  I.  INTRODUCTION                                               398
 II.  UNDERSTANDING TAX OVERHANG                                 401
      A. Conceptual Example                                      402
      B. Numerical Example                                       402
      C. Countervailing Benefits                                 404
      D. Overhang Versus Rebalancing                             405
      E. Overhang is Unique to Mutual Funds                      406
III.  TAX OVERHANG IS BAD                                        408
      A. Overview                                                408
      B. Inefficiency                                            409
      C. Mutual Funds Versus Separately Managed Accounts Versus
         Hedge Funds                                             411
 IV.  TAX OVERHANG IS SIGNIFICANT                                414
      A. Expressed as a Ratio                                    414
      B. Expressed in Dollars                                    417
      C. Expressed as a Drag on Yield                            419
  V.  TAX OVERHANG AVOIDANCE STRATEGIES                          419
      A. ETFs                                                    420
      B. Vanguard                                                424
      C. NextShares ETFs                                         428
      D. Ret-low                                                 432
 VI.  REFORM OPTIONS                                             436
      A. Introdu ction                                           436
      B. Pure Pass-th rough Taxation                             437
      C. Shareholder-centric Ref orm                             439
      D. Fund-cent ric Reform                                    441
      E. Evaluation                                              442
VII.  CONCLUSION                                                 442


Mutual funds are taxed as pass-through entities. Fund shareholders are taxed on income earned by the fund in the year the income is earned. The fund itself pays negligible tax. This system is implemented by compelling the fund to pay out most of its profits as dividends, and allowing the fund a dividends-paid deduction. The deduction offsets most or all of the fund's taxable income. The dividends allocate the income among the shareholders. (1) The overarching policy goal of these rules is to facilitate pooled investment. This is done by taxing each member of the pool as she would be taxed if she made investments for her own account.-' This system usually works reasonably well. One significant caveat is the treatment of tax overhang.

Tax overhang is the accrued, unrealized gain on the fund's portfolio. In other words, it is the difference between the value and tax basis of a fund's investments. (3) Tax overhang and the pass-through rules applied to mutual funds interact to accelerate the tax burden imposed on mutual fund shareholders. In effect the shareholder is taxed on the earlier of (a) the fund's realization of any gain that exists in its portfolio, assuming the shareholder owns shares when the gain is distributed (typically close to year end), (4) or (b) the shareholder's disposition of her fund shares. (5)

A fund that has both tax overhang and turnover in its portfolio will tend to impose a tax burden on the shareholders earlier than they would like, and in many instances earlier than would result from a comparable investment in a separately managed brokerage account (6) The greater the turnover and the tax overhang, the more the tax burden accelerates. Sometimes, the rules result in taxable income passing through to fund shareholders who have no economic income, or a loss, from their mutual fund investment.

Tax overhang distorts the behavior of fund managers, current fund shareholders, and those considering investing in fund shares. These distortions represent dead weight loss. Accelerating the shareholders' tax is unfair. It is also incongruent with the policy goal of simulating the results that fund shareholders would achieve if they made direct investments themselves, rather than pooled investment through the fund.

Until the early 2000s one could chalk this issue up as an unfortunate aspect of a system that, for the most part, worked well. True, tax overhang was a problem, but it was a pervasive problem that applied uniformly across the industry. During the past fifteen years, things have changed. Some investment advisors have developed expertise in avoiding tax overhang. Yet the strategies these advisors use are not universally available. Only certain types of funds, and certain fund complexes, have mastered the art of dodging this burden. As I detail below, most exchange traded funds (ETFs) are able to plan around tax overhang; and Vanguard and Eaton Vance both have taxstrategy patents on avoidance structures that are available to their funds, but not to other funds--unless they are willing to pay royalties to license the technology.

Planning to avoid tax overhang centers on section 852(b)(6) of the Internal Revenue Code. (7) This rule permits funds to avoid paying tax on accrued, unrealized gain on securities held in the fund's portfolio if the securities are distributed to shareholders in redemption of fund shares. (8) In contrast, if a fund liquidates securities with accrued, unrealized gains to raise funds to make cash dividend distributions or cash payments to redeem a fund shareholder, the resulting gain will be income to the fund. This income will then be passed through to shareholders who remain invested in the fund following the cash distribution or redemption. Present law thus favors in-kind redemptions over cash redemptions and dividends.

A difficulty with using section 852(b)(6) to avoid triggering fund-level tax is that fund shareholders expect to be paid in cash when the redeem their shares. The ability to convert fund shares to cash based on the closing value of the fund's portfolio is a definitive feature of an open-end mutual fund. The planning strategies that have sprung up involve financial intermediaries taking the role of shareholders and agreeing to accept in-kind payment in redemption of their shares. The architects of these strategies have been completely forthright that a principal goal is tax avoidance.

Legislative history to section 852(b)(6) is nonexistent. The best explanation for its existence is that tax policy should conform to policies underlying cognate provisions in the securities laws. The Investment Company Act, the primary statute governing mutual funds, permits funds to redeem shareholders in kind." This provision protects funds from having to make a fire sale of securities if, in the judgment of the fund's investment manager, the market is temporarily dysfunctional." If funds were required to recognize taxable income when they made in-kind redemptions, then tax consequences would stifle transactions that the securities laws were designed to facilitate.

If coordination between tax and securities law is the reason why section 852(b)(6) exists, present exploitation of the rule has decoupled the rule from its underlying justification. Funds are now being structured to make in-kind redemptions standard practice, not a market-driven substitute for cash redemptions. They are doing this for the explicit purpose of mitigating tax overhang. There is no temporarily dysfunction market and no fire sale--just wily tax planning.

In Section II, I define mutual fund tax overhang and illustrate its effect using simple examples. I also distinguish overhang from the tax burden imposed on fund shareholder when funds trade in the ordinary course of business (e.g., to rebalance the fund's portfolio) and explain why tax overhang is an issue for open-end mutual funds, but not for other entities (such as closed end funds or garden variety corporations). In Section III, I elaborate on the reasons tax overhang is bad--essentially because it is inefficient and unfair. In Section IV, I quantify the magnitude of the issue. To preview, it is a big problem. In Section V, I describe the avoidance strategies that funds have used to avoid the burden of tax overhang. In Section VI, I describe and evaluate the options for reform. Finally, in Section VII, I offer some concluding observations.


Suppose that on December 30, 2016, an investor buys $1000 in mutual fund shares. On January 15, 2017, when the net asset value (NAV) of the fund remains exactly the same as it was on the date of the investor's purchase, the fund makes its 2016 capital gains distribution. The investor's share of the distribution is $100. The investor now has $900 of mutual fund shares (diminished from $1000 by the distribution), $100 cash, and a tax liability of as much as $23.80, assuming the investor is in the top rate bracket. The investor has not enjoyed any increase in her wealth. Nonetheless, she is taxed because of trading by the fund that pre-dated her investment and the way that fundand shareholder-level tax rules interrelate.

Below, I give conceptual and numerical examples to illustrate the phenomenon in greater depth. A deep understanding is useful when considering the avoidance strategies and potential reform options I discuss below. It is possible, though, to get lost in the details. The key point to take away from all of the examples is that, in certain cases, fund shareholders are taxed on fund income that accrued prior to the shareholder's investment. This result, though conceptually inappropriate, is a natural consequence of present law.

A. Conceptual Example

Begin with Shareholder #1, who buys an arbitrary number of fund shares. The fund invests the proceeds in additional portfolio investments. Later, after a 50% rise in the value of fund shares, Shareholder #1 redeems her fund shares. Shareholder #l's gain is 50%, just what it would be if she had sold her share of the fund's portfolio herself. Shareholder #1 is taxed on her gain even if the fund continues to hold the underlying investments.

Shareholder #2 enters the fund just as Shareholder #1 exits. Shareholder #2's purchase funds Shareholder #l's redemption. (This explains how the fund was able to hold its investments and, at the same time, pay redemption proceeds to Shareholder #1.) The 50% gain that accrued on the fund's portfolio holdings during Shareholder #l's ownership will be taxed through to Shareholder #2, assuming that the fund sells out the positions on which the gain accrued while Shareholder #2 is invested in the fund. This result follows even if Shareholder #2 does not sell any of her fund shares, and even if the NAV of the fund has not increased during the period of Shareholder #2's ownership.

This result for Shareholder #2 is harsh, but not as harsh at it first seems. If Shareholder #2 is taxed on this gain--from her perspective phantom income--Shareholder #2 will increase her tax basis in her fund shares. This basis increase will allow Shareholder #2 to avoid tax on subsequent gain (or she might benefit from a cognizable tax loss). The net effect is to accelerate income, given that any savings from increased tax basis will only benefit her later, when she exits the fund. Nominal (undiscounted) income for Shareholder #2 is the same as would be true if she had invested on her own.

B. Numerical Example

The following example gives a more detailed illustration, showing how the fund might be forced to trade by net outflows. At time 1, investors A, B, and C invest $1 each in shares of Fund, a newly organized open-end fund. Shares are sold for $0.10, so each shareholder acquires 10 shares. Fund uses the $3 proceeds to make a portfolio investment in 3 shares of X Co stock for $1 per share, total cost: $3.

At time 2, the value of X Co stock has increased to $3 per share, so the NAV of Fund shares risen from $0.10 to $0.30 per share, a 200% return. Still at time 2, a new investor, D, joins the fund with an investment of $3. Fund uses the proceeds of the sale of its shares to D to purchase an additional share of X Co stock, bringing its holdings in X Co to four shares total. The four X Co shares have tax bases of $ 1, $ 1, $ 1, and $3 (the $3 basis is attached to the most recently purchased share).

At time 3, when the value of X Co is still $3 per share, implying a pershare NAV for fund shares of $0.30, investors A and B redeem their fund shares. To raise cash to pay A and B for their shares, Fund must sell off some of its portfolio investments in X Co shares. It chooses to sell the most recently purchased X Co share (basis of $3) and one of the older, seasoned shares (basis of $1), so its gain is $6 - $3 - $1 = $2.

A and B are taxed on the redemption transaction. Each receives proceeds of $3 in exchange for shares with a total basis of $1, resulting in gain of $2. Fund must distribute its $2 of gain on its sale of X Co shares to the remaining shareholders, C and D, in proportion to their holdings of Fund shares, so each receives distributed income of $1. C and D both pay tax on the distribution.

The distribution to C and D reduces the assets of Fund from $6 to $4, so the NAV of fund shares drops from $0.30 to $0.20. C and D, like most fund shareholders," participate in Fund's automatic dividend reinvestment plan, and so each makes an investment in 5 additional fund shares. C and D's automatic reinvestment increases Fund's NAV from $4 back to $6, but now there are 30 fund shares outstanding rather than 20, so per-share NAV remains $0.20.

At this point, C and D hold 15 shares each. C's shares have a total tax basis of $2 (equal to the sum of C's original investment plus her reinvested dividend) and D's shares have a total tax basis of 4 (equal to the sum of D's original investment plus her reinvested dividend).

At time 4, the value of X Co is $4 per share, implying that Fund's portfolio is worth $8 total. Fund shares held by each of C and D have a value of $4. D redeems her shares for $4. The resulting gain to D is nil, given that her basis offsets the sales proceeds exactly.

We see from this example that the effect of mutual fund tax overhang is to skew the timing of D's tax from her investment in Fund. Figure II.A recaps the example.
Figure II.A. Example of mutual fund tax overhang

Time 1            Time 2            Time 3

A, B, & C invest  Fund portfolio    Fund NAV remains stable
$1 each in new    gains 200%        NAV = $12 total
fund              NAV / $3to$9      -
-                 -                 Fund redeems A&B for $3 ea.
Fund invests $3   D invests $3      in
X Co shares       for 1/4 stake in  To raise cash, fund sells X Co
                  fund              shares, realizes $2 gain (*)
                  -                 -
                  Fund buys $3      $2 gain divided between C&D,
                  more X Co         DRIP 1* C&D's basis in fund,
                                    $1 each

Time 1            Time 4

A, B, & C invest  Fund NAV / $6 to $8
$1 each in new    fund
                  Fund redeems D for $4,
-                 no gain to D
Fund invests $3   (D's basis = redemption
in X Co shares    proceeds)
                  To raise cash, fund sells
                  X Co shares, $3 gain
                  $3 gain allocated to C

The value of D's investment in the fund did not increase between time 2 (purchase) and time 3 (redemption of A and B), yet D was taxed on fund portfolio gain that predated her investment in Fund. Then the value of D's investment did increase between time 2 and time 3. D is not taxed on this increase in the value of her shares given the basis adjustment allowed for the tax at time 2. Thus, D pays no more tax than she would pay under a system that didn't treat tax overhang this way, but her tax liability is accelerated. This increases D's effective tax rate by stripping away some of the benefit usually enjoyed by buy and hold investors--specifically, the ability to defer tax on gains until disposition, referred to as the realization rule.

C. Countervailing Benefits

As described thus far, it appears that mutual funds are not tax efficient compared to direct investment via a personal brokerage account. There are, however, two features of mutual fund taxation that supply a benefit to mutual fund shareholders compared to their tax treatment if they made direct investments in portfolio securities through separate accounts. These features partially offset the bad tax effects of mutual fund tax overhang.

The first benefit requires the concatenation of two factors: (1) the fund grows its assets through new net investment (as distinct from portfolio appreciation) after a given shareholder invests, and (2) the fund has positive investment performance. When these factors coalesce, then, from the perspective of a shareholder who owned through the period of growth and investment success, the built-in gain in the fund's portfolio will be less, on average, than would be true under a counterfactual with the same portfolio composition but without any new net investment. If things set up this way, and the fund triggers gain on its portfolio, the tax on our illustrative shareholder's fund investment would be smaller than the gain she would realize if she made the same trades herself in a separate account. (12)

The second beneficial feature of fund taxation is that the fund's deduction of its investment expenses is, in effect, passed through to the shareholders. If they bore their own investment expenses (as would be the case if they held investments directly in a separate account) the expenses would likely be disallowed."

These two benefits are independent of one another. Both are likely to be smaller in magnitude than the bad effects of tax overhang, but they mitigate the tax drag mutual funds impose on their taxable shareholders. (14)

D. Overhang Versus Rebalancing

As detailed in the preceding examples, present law sometimes subjects fund shareholders to worse tax treatment compared to the results that they would encounter if they made a comparable investment on their own. As discussed below, funds use elaborate strategies to avoid the bad effects. Policymakers and academics have proposed rule changes to eliminate these bad effects. The avoidance strategies and proposed policy changes work, very generally, by immunizing fund shareholders from the tax consequences of fund-level portfolio turnover. When focus is placed on tax overhang by itself, the planning seems benign or even laudable (avoiding costly tax effects that that are conceptually unjustified), and the policy changes seem like a clear improvement (aligning results in practice with those that make sense in concept).

Unfortunately, however, there is a difficulty: failure distinguish between (a) the bad effects of trading by a fund with tax overhang and (b) the natural consequences of trading to implement a shifting investment thesis, rebalancing trades to correct for portfolio skew, or trades for any other reason.

Suppose a rich investor has sufficient wealth to achieve diversification herself and to hire a professional manager. She doesn't need to invest via a fund. If she invests on her own, the composition of her portfolio is likely to shift over time. As she sells some holdings and buys others, she will realize taxable gains and losses. Now suppose she invests though a fund. Assume the fund has zero overhang when she buys shares. Later, after her purchase, the fund's portfolio turns over. Normally, the resulting gains would pass through to the shareholders in the fund, including our investor. Assume though that as a consequence of tax planning (along the lines discussed in Part VI) or rule changes (as discussed in Part V) the tax consequences of fund trading do not pass through the fund to its shareholders.

In this example, rebalancing of the fund's holdings should, in concept, result in tax to the investor, assuming the goal is symmetry between investments in funds and separate accounts. The difficulty is that planning strategies and some of the reform proposals designed to shield investors from the bad effects of tax overhang work by immunizing fund shareholders from tax on fund-level gains. But not all fund-level gains are attributable to overhang, as just illustrated.

As described below in greater detail, it is difficult or impossible to permit planning but to restrict the scope to mitigating tax overhang. It is similarly difficult or impossible to make surgical amendments to the rules to cut out the bad effects of overhang without, at the same time, shielding fund shareholders from fund-level gains that don't result from overhang. The overinclusiveness of the planning and reforms directed at tax overhang complicates the question whether permitting much of the mutual fund tax planning observed presently is good or bad policy, and whether reforms aimed at mitigating tax overhang are worthwhile. As discussed below, such reforms are likely to hit their intended target (mitigating tax overhang) but also to generate unintended benefits (by facilitating avoidance of gains tax on garden variety rebalancing trades).

E. Overhang is Unique to Mutual Funds

Tax overhang is unique to mutual funds. The phenomenon occurs because (a) there are two taxpayers responsible for accounting for the same income--the fund and the shareholder; (b) the economic burden of tax on income earned by one taxpayer (the fund) is visited on the other taxpayer (the shareholder), even if the shareholder has no income (or less income than the fund); and (c) the price the shareholder pays for her shares does not take into account the fund's accrued, unrealized gain--or, more to the point, the tax that will be owed when the gain is recognized.

Why, you might ask, don't these factors exist for garden variety corporations taxed under the normal rules of corporate tax (Subchapter C)? The answer is that factors (a) and (b) exist, but factor (c) is a unique issue for open-end mutual funds. Factor (c) arises for mutual funds because of a securities law rule found in the Investment Company Act, which requires trading in shares of open-end funds to be priced at per share NAV, an estimate determined without regard to a fund's accrued, unrealized tax liability.

Take a simple example. Shareholder A is the sole owner of X corporation, which is taxed under Subchapter C (i.e. X is not a mutual fund). X begins the year with $100 of net assets and during the first 11 months of the taxable year earns taxable income of $10, increasing its net assets to $110 (before tax). If shareholder A sells her stake in X to B, B will not pay $110 for A's X stock. A likely price in the A-to-B sale would instead be $106.50. This is the net assets of X corporation after taking into account X's obligation to pay corporate tax (assuming a 35% corporate tax rate). Even though X hasn't yet paid or even reported the $10 income begetting the $3.50 tax, the burden of the inchoate tax liability is borne by A in the form of a reduction in the price B (or any thoughtful buyer) would be willing to pay for the X shares. (15)

The example demonstrates that in the Subchapter C context, the first two conditions necessary for tax overhang are present but the third is absent. There are two tax reporting entities and the economic burden of the tax on one taxpayer (the corporation) is borne by the other (the shareholder). The difference is that for a garden variety Subchapter C corporation the particular shareholder bearing (indirectly) the burden of the corporate tax should, in theory, be the shareholder owning the stock when the taxable income accrued, even if the income has not been realized. The burden is borne by the selling shareholder not as a tax remittance per se, but as a tax induced reduction in the price the buyer is willing to pay. (16)

For mutual funds, unlike Subchapter C corporations, the price a buyer pays for shares in the fund is set equal to the per share NAV of the fund, without regard to the accrued gain on the fund's portfolio. One way to the think about overhang is as the discount to NAV that a buyer of fund shares would demand if she were permitted to buy fund shares at a discount to NAV calibrated to account for her share of the fund's inchoate tax. In this circumstance, the buyer is powerless to negotiate for a discount to NAV to account for the funds tax position. Section 11 of the Investment Company Act prohibits trading in mutual fund shares at any price other than one tied to the per share NAV of the fund, a figure that determined without considering taxes. (17)

In sum, tax overhang results from the confluence of tax rules interrelating fund and shareholder taxation and securities law rules fixing the trading price of open-end fund shares to a price based on pre-tax value to the shareholder. Some of these rules apply in contexts other than open-end mutual funds, but their overlap of all rules necessary to trigger overhang occurs only for open-end funds.


A. Overview

Mutual fund tax overhang is bad for three primary reasons. First, it is inefficient. It skews the incentives of fund sponsors, fund managers, and investors, altering their behavior compared to how they would conduct business absent such overhang.

Second, tax overhang imposes a burden on mutual fund investors that is not borne by taxpayers investing on their own account or through private funds (such as hedge funds and private equity funds). (18) Taxpayers who have sufficient wealth to achieve investment diversification within their own portfolios, and those who qualify as "accredited investors" eligible for private fund investment, will tend to be wealthier, on average, than taxpayers who invest exclusively through mutual funds. This is not an instance where all taxpayers have the same options and some choose wisely and others foolishly. Rather, the wealthy taxpayers have better choices than do other taxpayers. Consequently, the incremental burden imposed on mutual fund investors is regressive. (19)

Third, tax overhang allocates taxable income to the wrong taxpayers. This is seen most obviously in the case of a mutual fund shareholder who unwittingly buys into a fund just before it declares a dividend and bears tax on income that accrued prior to her investment in the fund. The tax assigned to the new shareholder should have been assigned to the shareholders who invested in the fund when the gain accrued.

I already explained and illustrated this third problem above. The shareholder treated unfairly in the conceptual example is Shareholder #2, and in the numerical example is shareholder D. That is all I will say on this point. Below, I explain the first two problems with present law in more detail.

B. Inefficiency

Tax overhang causes a significant amount of wasteful tax planning. Although the tax planning is effective in deflecting the negative effects of tax overhang, the planning is costly to implement (in terms of advisor fees, time, and attention), and saps resources from the central objective of the fund manager--namely, maximizing risk-adjusted return. The wasteful planning that occurs in response to present law falls into three categories: creating fund structures and contracts designed to purge overhang, portfolio churning, and tax induced opening of new funds.

Many investors are taxpayers, and so organize their investments to minimize taxes (e.g., by tax loss harvesting, among other strategies). One might argue that in this way, mutual funds are no different from other taxpayers. Although this is true, mutual fund complexes have the money, expertise, and resources to engage in tax planning on a massive scale. The magnitude of the resulting dead weight loss from planning to avoid tax overhang points to a problem worth solving if, as I argue in Section VI, a solution is feasible.

Turn now to the three categories of wasteful planning. The first of the three categories of wasteful planning is when funds create structures and enter into contracts designed to purge overhang. Most of these structures and transactions rely, at their core, on section 852(b)(6). Section 852(b)(6) is a mutual-fund-specific exception to section 311(b), the rule that corporations are taxed when they distribute built-in gain property to their shareholders. The predecessor to section 852(b)(6) was added to the Code in 1969, at the same time as the original version of the rule now codified at section 311(b) was added to repeal the General Utilities doctrine.' Section 852(b)(6) laid dormant for several decades until funds recognized how it could be used to purge tax overhang.

In Section V, I discuss the specific structures and transactions designed to make use of the rule to purge overhang. The upshot is that there are strategies that are available to some funds that facilitate a total purging of tax overhang. For a variety of reasons, other funds are unable to make use of these strategies. The result is a hodge-podge of tax-efficient and tax-inefficient funds. This distinction among funds increases complexity for fund managers, investors, and regulators (both the IRS and the SEC), in furtherance of no apparent policy objective. If policymakers are serious about collecting the tax liability inherent in tax overhang from mutual fund investors, they should repeal section 856(b)(6) to negate the benefit of these structures; or, if not, they should implement reform so the resulting tax drag doesn't continue to fall unevenly among funds and fund shareholders.

The second type of wasteful planning is portfolio churning. A large capital gains tax overhang will tend to reduce net inflows into a fund. This is a consequence of two partially offsetting effects: there is a large negative effect on gross inflows, partially offset by a smaller negative effect on gross outflows/ (1) In other words, when there is significant overhang, existing investors are locked into their positions. Potential new investors, on the other hand, are repelled by the potential for adverse tax consequences from buying into a fund with large accrued, unrealized gains in its portfolio. The data that exists suggest that managers respond to this state of affairs by churning the fund's portfolio to trigger built-in gains and thereby reducing tax overhang."

Managers do this, it is argued, out of greed. It is harmful to the shareholders presently invested in their funds, but helpful to potential new shareholders. Manager compensation increases with the fund's net assets. If the manager knows or expects that particular conduct will result in gross inflows from new investors large enough to offset the gross outflows from existing investors plus the resulting tax cost, then the fund manager has a direct financial incentive to engage in that conduct. This reality, and the well-accepted notion that fund governance structures are weak, explains why portfolio churning occurs even though fund managers owes a fiduciary duty to the current shareholders (harmed by churning) and no similar duty to potential new investors (the beneficiaries of churning). (23)

The third and final type of wasteful planning is tax-motivated launching of new funds in areas of the market dominated by large funds with significant capital gains tax overhang. A new fund cannot have tax overhang at inception, making a new fund relatively attractive seasoned funds burdened by tax overhang. A comprehensive study of funds launched between 1979 and 1992 shows that for equity funds a one standard deviation increase in tax overhang increases the likelihood of a new fund opening in the same sector by 39%." Opening a new fund will tend to impose costs on fund sponsors and therefore indirectly on investors. The costs include the overhead of establishing and marketing a new fund, and the lost economies of scale that would have been captured if, counterfactually, investors in the new fund had instead invested in one of the pre-existing funds.

To summarize, tax overhang is bad because it generates a significant amount of wasteful planning. This planning helps to maximize taxable investors after-tax yield from fund investments, but not by improving investment performance; instead the gains reaped by shareholders from planning of this sort is, at best, a zero-sum transfer from the fisc (equal to the avoided tax), or, in some cases (such as tax motivated opening of new funds) sheer waste.

C. Mutual Funds Versus Separately Managed Accounts Versus Hedge


If a burden analogous to the one resulting from mutual fund overhang were borne by investors purchasing marketable securities in separately managed accounts, or investing through hedge funds, then one might reasonably conclude that the problem is sufficiently pervasive that there is nothing special to complain about in the context of mutual funds. The implication would be that an overarching solution such as eliminating capital gains taxation altogether or adopting mark-to-market might improve the tax system for all investors, but no special mutual-fund-specific intervention is warranted. This turns out, however, not to be the case: tax overhang is unique to the mutual fund context.

To see why, start with separately managed accounts. Many taxpayers have brokerage accounts in which they hold stocks and bonds directly, rather than investing thought a fund. This approach tends to make more sense for wealthy taxpayers who have reached a critical mass enabling them to assemble a diversified portfolio of direct holdings of securities. For many, this process involves paying a professional manager to select the securities for their account, and to manage the process of rebalancing the portfolio, reinvesting dividends and interest, and the other never-ending tasks associated with direct ownership of stocks and bonds. Taxpayers choosing this path need not concern themselves with tax overhang. Tax overhang is a consequence of pooling capital in a commonly-managed vehicle; if there is no pooling, there is no tax overhang. (25)

Many wealthy taxpayers also invest in hedge funds. Like mutual funds, hedge funds involve pooled investments in in marketable securities for the benefit of passive owners of fund shares. Intuition therefore suggests that hedge funds would subject their investors to tax overhang just like mutual funds. It turns out, however, that this is not the case.

Usually, hedge funds are partnerships for federal tax purposes.' Hedge funds have figured out a method of manipulating partnership tax accounting to alleviate the burden of tax overhang on hedge fund investors." The technique, known as "stuffing allocations," is controversial.-" Some commentators believe the strategy "works" in the sense that it would be sustained by the courts if challenged on audit.'' Others do not think stuffing allocations work." Regardless of how one comes out in the debate regarding the legal basis for stuffing allocations, it is clear that the strategy is effective, practically speaking. The use of stuffing allocation is widely reported to be universal in the industry, and there are no reported cases of challenge, let alone successful challenge, by the IRS. Successful implementation of the strategy thus far is the result of anemic partnership audit rates and IRS inattention, not validation of the strategy on its merits.

To illustrate the mechanics of stuffing allocations, imagine there is a three-person partnership--a hedge fund--formed by X, Y, and Z. Each contributes $1 and the fund invests the $3 in a portfolio of stocks. The portfolio increases in value by 150% to $7.50, at which point X decides to withdraw. To redeem X, the fund must sell one-third of its portfolio at a gain of $7.50/3 -$1.00 = $1.50. (31)

In the absence of a stuffing allocation, the fund would divide this $1.50 gain three ways among the partners. X's allocation would drive X's outside basis in her partnership interest from $1 to $1.50. The redemption of her interest would thus result in gain of $2.50 - $1.50 = $1. Altogether, the resulting income to X would be $1.50. X's withdrawal triggered an increase in taxable income of $0.50 for both Y and Z, even though they sat idle while X withdrew.

If, on the other hand, the fund has agreed to stuffing allocations, the full gain on the sale to fund redemption of X's interest is allocated ("stuffed," in the jargon) to X's capital account. This will increase the fund's allocation of gain to X from $0.50 (without stuffing) to $1.50 (with stuffing). After the $1.50 allocation but before the redemption of X's partnership interest, X's outside basis is $2.50, equal to her original investment plus the stuffing allocation. Ergo, X will have no gain on the redemption (redemption proceeds of $2.50 will be offset in full by her tax basis). As that the full $1.50 gain on the sale of portfolio securities to fund X's withdrawal is allocated to X, none is left to allocate to other partners in the fund, and they pay no tax as a consequence of X's exit.

The key point is that the stuffing allocation has no net effect on X's tax position--under both scenarios she will have $1.50 of taxable income as a consequence of her withdrawal from the fund--but the stuffing allocation relieves the tax burden that would fall on the other two fund investors in the absence of stuffing. Stuffing allocations are designed to redirect the tax cost that withdrawing hedge fund investors would impose on investors remaining in the fund.

To summarize, then, tax overhang is a byproduct of pooled investment. Gains triggered in the pool are sometimes assigned to the wrong member of the pool. This phenomenon, which has a significant effect on after-tax investment performance of mutual funds and generates significant inefficiencies, is not of concern for well-heeled investors rich enough to set up well-diversified separate accounts or to invest through hedge funds. Though hedge funds are pooled vehicles and thus in concept expose fund investors to tax overhang, hedge fund advisors have thus far successfully planned around the issue.


By every measure, mutual fund tax overhang is an economically significant issue, both to taxable mutual fund investors and to the fisc. There are various ways to express the magnitude of tax overhang: built-in gains in fund portfolios (or the resulting inchoate tax), which can be described in dollars or as the ratio of built-in gains to portfolio value; the drag that tax overhang imposes on net-of-tax fund yield; and the relationship between tax-overhangrelated drag on performance and other fund expenses borne by investors. I summarize estimates of the magnitude below.

A. Expressed as a Ratio

Financial economists have estimated the magnitude of mutual fund overhang. Typically, their studies summarize overhang as a percentage of mutual fund assets held in equity funds.'' The most recent study relies on data from the CRSP Survivorship Bias Free Mutual Fund database for the 1990-2012 period, focusing on funds investing primarily in common stocks. The average tax overhang across all fund-years was 12.07% of net asset value. This composite figure breaks down into a share attributable to long-term capital gains (10.27%) and short-term capital gains (1.80%).33 A more complete set of summary statistics is in Table IV.A.

For long-term gains, the mean is significantly greater than the median, implying that measures of central tendency mask a composite of many relatively low overhang funds with a smaller number of funds that have very significant overhang on a relative basis. Note also that the standard deviation is very large, approximately 80% greater than the mean for long-term gains. Heterogeneity among funds is considerable.

A similar estimate is made in another paper relying on the similar data (again using a CRSP mutual funds database), in this case for the years 1962-2005. The authors focused on overhang in stock funds. They find the overhang varied significantly by year. Their estimates for the most recent ten years in their study are set forth in Table IV.B.

The estimate of average overhang for the last three years in the study (2003-2005) is consistent with the more recent data in the first study described and summarized in Table VI.A, above.

Finally, an older study summarizes fund level data for 1999. Specifically, the authors estimate the tax overhang for the 20 largest equity mutual funds that were open to new investors on December 31, 1999, based on data collected from the Morningstar Principia database. The data in this study paints a bleaker picture for tax-sensitive investors, as summarized in Table IV.C.

The data shows that there is very substantial variation among funds' tax overhang. For some, the overhang is more than half of the fund's assets; for others, the fraction is as low as one-tenth. This snap shot no doubt reflects the impact of the stock market bubble that was inflating at the time and reminds us that the economic significance of the tax treatment of mutual fund tax overhang rises and falls with the equity markets.

B. Expressed in Dollars

A crude estimate of the financial significance of mutual fund tax overhang, expressed in dollars, can be made by multiplying the following three amounts: (a) the aggregate market value of equity held by U.S. mutual funds; (b) the share of such assets held in taxable accounts; and (c) an estimate of the average overhang for all funds, expressed as a fraction of net assets. Three illustrative computations, using a range of overhang fractions, consistent with the ratios discussed in the prior discussion, are set forth in Table IV.D.

Thus, as a first approximation, there is usually an overhang of several hundred billion dollars, sometimes up to a trillion dollars or more, attributable to taxable mutual fund shareholders. The cash taxes due if the overhang were realized and passed through to fund shareholders, reflected in the bottom row of the table, depends on the effective tax rate applied to this figure. In Table IV.D I assume the effective tax rate (ETR) equals one-half of the top capital gains tax rate (presently 20%). Economists have used this rate to ballpark the burden of deferred capital gains, arguing that it "is broadly consistent with effective tax rate calculations using a range of plausible values for realization rates and discount rates.""

To put this in context, the most conservative (lowest) estimated tax liability reflected in Table IV.D--$44 billion--would rank as the eleventh largest tax expenditure. The highest estimate--$133 billion--would be the second largest tax expenditure, exceeding items such as the deferral of income from CFCs, the home mortgage interest deduction, and the capital gains preference."

C. Expressed as a Drag on Yield

Another way to conceptualize the magnitude of mutual fund tax overhang is to consider the drag that capital gains realizations impose on investor yields on their fund investments. It turns out that the reduction in yield attributable to tax overhang is, on average, probably of the same order of magnitude as the drag from advisor fees, even though fees get more attention from the financial press and investors.

One estimate (based on data from 1990-2012) indicates that the tax burden imposed on equity funds was, on average 1.12% per year, compared with an average expense ratio for the funds in the dataset of 1.21 % per year.'" The tax burden in this estimate includes (both long- and short-term) capital gains, plus dividends. Thus, not all of it is attributable to tax overhang. But most of it is.

Another estimate of the tax cost imposed on mutual fund investors by persistent tax overhang was developed using a simulation model. The model was constructed using simulated portfolios of equity securities drawn from the 50 largest companies in terms of market capitalization in 1983, and tracking the return on the portfolios over the next 15 years. The after-tax returns on the portfolios were calculated for an investor facing the top tax rates presently imposed on ordinary income and capital gains." The simulation considered the impact of the various factors on after-tax returns, including accounting practices used by the fund,'" the investment strategy implemented by the fund, and whether the fund was growing or shrinking. Under a middle of the road version of the simulation--not the most or least tax efficient along any of the dimensions of planning available to the fund manager--a fund would experience a 2.5% per year reduction in its after-tax yield, assuming the fund is neither growing or shrinking." This is an enormous burden.


This section explains how mutual funds purge tax overhang from their portfolios. They do so by structuring themselves in a way that allows them to exploit section 852(b)(6). As described above, this is the present-law version of the RIC-specific exception to General Utilities repeal inserted back in 1969. (40) Mechanically, the funds exploit section 852(b)(6) by creating circumstances where it is practical for the fund to redeem an institutional shareholder in kind, rather than paying cash.

I begin with the first, and oldest, strategy--the exchange traded fund (ETF). Then I turn to two patented fund structuring techniques, one owned by Vanguard and the other by Eaton Vance. I conclude with the program patented by the boutique investment firm ReFlow.


The ETF label describes a variety of exchange-traded instruments that facilitate indirect investment in a portfolio of assets invested according to a specific strategy or program. (41) The portfolio might be held by an entity organized as a corporation qualifying as a registered investment company, a unit investment trust, or a partnership. '' Most ETFs are registered investment companies (i.e., mutual funds). Given the focus of this article, unless I am explicit, assume that when I mention an ETF, I mean an investment company ETF.

ETFs operate for the most part just like conventional open-end funds. This means there is an entity (for state law purposes, typically a business trust or a corporation) taxable as a corporation that holds title to the portfolio investments. The entity contracts with a registered investment advisor to manage the fund's investments for the benefit of the fund's shareholders. (4) '

The big difference between ETFs and open-end funds is that most ETF shareholders do not trade with the fund directly. Individual shareholders cannot redeem their shares from the fund. Neither does the fund stand ready to sell shares to would-be individual investors. Instead, fund shares trade in the secondary market, on public stock exchanges, like common stocks and other listed securities.

The fund, however, does trade its shares with a special class of institutional investors referred to as authorized participants. The fund enters into contracts with these "APs," usually a market maker or other institutional investors in its shares. (41) APs, in contrast to ordinary fund shareholders, trade fund shares directly with the fund. The proviso is that any such trades--be it the redemption of outstanding ETF shares by the fund or the issuance of new shares--are in huge bundles, typically aggregates of 50,000 to 100,000 shares, called "creation units." (45)

ETFs disseminate data continuously on the value of the fund's holdings.'" The price quotes are referred to as an "indicative NAV" or "iNAV." Quotes typically are released at fifteen-second intervals. (47) This facilitates price arbitrage by the APs. If a fund trades at a sufficient discount to iNAV for an AP to profit by tendering creation unit bundles of shares in exchange for redemption proceeds, APs will do so to earn arbitrage profits. Conversely, if the fund trades at enough of a premium to iNAV, APs can earn arbitrage profits by purchasing creation unit bundles of shares from the fund at iNAV and to selling them in the market at the premium price. Aps, thus, compete away significant gaps between iNAV and ETF share price.

ETFs are more tax efficient than conventional open-end mutual funds. There are two reasons for this. First, ETFs have less flow than do otherwise comparable funds organized as conventional open-end funds. When there is a balance between buyers and sellers in ETF shares, and the ETF share price is tracking closely with the funds iNAV, the fund need not concern itself with keeping cash on hand to redeem shareholders. Instead, departing shareholders simply sell shares on the market to new shareholders. Decrease in gross cashflow, both in and out, means less frequent need to manage liquidity by trading portfolio holdings. A decrease in flow-related portfolio churn implies that less of the fund's tax overhang is realized. This improves tax efficiency. (48)

Second, ETFs often process redemption trades with an AP in-kind, rather than using cash. (4) * By doing so, the ETF can purge the built-in gain on the fraction of its portfolio distributed in kind. The built-in gain does not shift to the distributee AP in the trade; the AP's basis in the distributed property equals the fair market value of the property at the time of the trade.-" Conventional mutual funds could, in concept, process large share redemptions from institutional shareholders paying for the redeemed shares in kind. (As discussed below, funds are adopting new structures that move in this direction.) For conventional open-end funds, however, the common understanding of fund shareholders and the overarching securities law regulatory architecture are both premised on the understanding that shareholders wishing to exit their investment in a conventional open-end mutual fund will be paid for their shares in cash based on closing NAV.

In circumstances where the fund is facing a large tax liability from an existing position, as occurs when a large holding is eliminated from the index the fund tracks (and thus must be removed from the fund's portfolio), there might be more need for in-kind redemptions--to purge built-in gains--than there are AP's wishing to be redeemed. Recently, funds have been accepting huge inflows on the eve of the date when the built-in gain position must be removed from the fund's portfolio, and then reversing the inflows with inkind redemptions of the built-in gain positions in the days that follow, free of tax."

As a consequence of these strategies, ETF shareholders are exposed to little or no capital gains tax liability on their fund shares. In a 2012 study, Morningstar compared ETFs and conventional open-end funds that follow the same benchmarks." The comparison was performed over five, ten, and fifteen- year periods ending on December 31, 2011. The results are set forth in Table V.A.

The top two rows in Table V.A show fractions. The denominator is the number of indices for which there was a matched pair of funds to compareone ETF and one open-end fund--both following the same index. The numerator is the number of funds with non-zero capital gains distributions. Thus, over the most recent five-year period 23 of 26 (64%) of the open-end funds exposed shareholders to capital gains distributions, but only 10 of 36 (28%) of ETFs did.

The bottom two rows are the (unweighted) average of the distributions expressed as a share of NAV in percent. Thus, the number 1.1561 in the bottom left corner means that over the five-year period in the study 1.1561% of the funds' average NAV was paid out in capital gains distributions. This number is more than 40-fold (4000%) greater than the corresponding figure for ETFs.

These results are broadly consistent with another study conducted using data from the period 1990-2012, which compared ETFs and matched set of mutual funds tracking the same index. The results, set forth in Table V.B, confirm that ETFs are substantially more tax efficient than open-end mutual funds.

Focus on the comparison of short- and long- term capital gains for the ETFs and matched index funds, the bolded rows in the middle of Table V.B. As a fraction of the average open-end fund distribution, the ETF average was 0.05/0.33 (15%) and 0.02/1.33 (1.5%) for short- and long-term gains, respectively. Put differently, the open-end fund long-term gains distributions were more than 65-fold greater than for ETFs tracking the same benchmark.

B. Vanguard

Vanguard has patented the idea of a single regulated investment company that issues both ETF shares and conventional open-end mutual fund shares. (53) Individual investors and institutions both may invest in either share class. (54)

Like shareholders in stand-alone ETFs, shareholders owning ETFs in Vanguard funds using its patented structure generally must exit their positions in ETF shares by selling the shares on the exchange, just as they would sell shares of stock in a listed company. Vanguard funds have Authorized Participant contracts in place with market makers and other institutional investors, similar to the AP contracts described above in the context of a standalone ETF. As is usually true for ETFs, APs contracting with Vanguard funds are allowed to aggregate large blocks of ETF shares--creation units--and redeem the ETF shares with the fund, or to purchase newly minted creation units from the fund based on the portfolio NAV. As for stand-alone ETFs, this is the process that keeps the market value of ETF shares close to the iNAV of the fund's portfolio, protecting investors from tracking error. (55)

It is in derogation of Section 18 of the Investment Company Act for a fund to issue two share classes exhibiting the differences that exist between the ETF shares and the conventional shares in the Vanguard funds structured in accordance with its business method patent. Section 18 generally prohibits an investment company from issuing more than one class of shares representing a stake in the same investment portfolio; (56) An SEC rule interpreting section 18 permits certain differences among share classes with respect to distribution and customer service without violating the single-shareclass restriction. (57) The SEC rule does not, however, permit differences in ability to redeem and exchange trading such as those that distinguish the conventional and ETE classes issued by Vanguard funds. Accordingly, Vanguard sought and has been granted exceptive relief by the SEC from complying with section 18, on the basis that granting exemption from the strictures of section 18 "is appropriate in the public interest and consistent with the protection of investors and the purposes fairly intended by the policy and provisions" of the ICA. (58)

The fund is invested in the portfolio of securities at the bottom of Figure V.A. Investors Al through An and BI through Bn are the holders of garden variety investor-class and institutional-class shares in the funds. (59) The investors in both classes acquired their fund shares from the fund directly in a cash purchase; they will be redeemed by the fund for cash when they wish to sell. The ETF shareholders toward the top right part of Figure V.A hold their fund shares in brokerage accounts at their brokers (named X and Y). The brokers, with the help of market makers and the clearinghouse, sell shares to new ETF shareholders, and make a market into which existing ETF shareholders can sell shares without the involvement of the fund itself. The brokers might trade with one or more APs (bottom right of Figure V.A). APs in turn trade with the fund itself, in creation units.

The advantages of Vanguard's structure are manifold. They include the following:

(1) Fund overhead will be lower than forming two funds, one for the ETF share class and a second fund for the open-end shares. (60)

(2) Combining share classes in a single fund increases the size of the fund and reduces tracking error. (61)

(3) Flow of funds into and out of the fund will be minimized. Active traders (e.g., market timers) will gravitate to the ETF shares and will therefore be trading in the secondary market, rather than contributing or withdrawing money to or from the fund. (62) Less flow means fewer portfolio transaction by the fund are necessary to manage liquidity, which will tend to reduce the fund's realization of capital gains. (63)

(4) Finally, and most significantly from the vantage point of this article, when the fund does trade with its shareholders--in particular, when it redeems creation baskets of shares from an AP--the fund will pay in kind, and will select the lowest cost lots of securities in the funds inventory for each security distributed. By doing so the fund will purge its tax overhang to the greatest extent possible. (64)

Compare the advantages gained through the Vanguard structure with those gained by a garden variety ETF. Advantages (1) and (2) are related to fund size, and thus are unique to the Vanguard strategy assuming fund sponsors aside from Vanguard must sponsor two funds per investment strategy for every fund offered by Vanguard if they wish to market both open-end and ETF shares. Advantage (3) is common to all ETFs; Vanguard's structure derives no benefit on this point aside from the one enjoyed by all ETFs.

Advantage (4) will help Vanguard funds in the same way it helps all ETFs. An important caveat to this last point, however, is that in a conventional ETF the benefit is available to ETF shareholders alone, given that the fund doesn't have any other share classes. In a Vanguard structure, by contrast, the benefit of the overhang-purging in-kind distribution redounds to the benefit of all classes of shares. Thus, the significant difference for Vanguard funds is that the ETF redemption trades wipe out gain that would be taxed not only to ETF shareholders but also to fund shareholders owning open-end shares. This could be good or bad, depending on one's point of view. The benefit of the in-kind distribution from the perspective of the ETF shareholders might be diluted compared to the benefit they would get if they and the open-end fund shareholders weren't accounting for capital gains tax realizations together; but if they weren't in the same fund, the economies of scale (advantages (1) and (2)) would not materialize." (5)

C. Next Shares ETFs

Exchange traded mutual funds (ETMFs), patented by Eaton Vance, (66) are a hybrid of conventional open-end mutual fund shares and ETFs. The Vanguard structure discussed above is also a hybrid of conventional open-end fund and ETF, but the hybridity is along a different dimension. For Vanguard funds, there is one fund with two types of share classes. Each share class in a Vanguard fund behaves like its prototype. ETMFs, on the other hand, issue a single share class (NextShares) that exhibits some features of an open-end fund share and some features of an ETF share.

Like ETFs, ETMF shares trade throughout the day on public exchanges, subject to the important caveat that the price agreed on between the seller and buyer in any trade is the closing NAV of the fund, plus or minus an agreedon spread expressed as an adjustment to the per share price. In other words, when parties trade mid-day, they don't yet know (and thus cannot specify) the price that the buyer will pay the seller at closing; instead, they agree on a spread to the value fixed at the market close. (67)

For example, suppose a buyer agrees to pay a seller per share NAV plus $0.05 for a 100-share block of ETMF shares at 1:00 pm. The fund quotes an intra-day value for its portfolio. Say the intra-day quote is $98 per share. Shortly after 4:00 pm (when the market closes), the fund determines that its closing NAV per share is $100 per share. The buyer is obligated to pay the seller $ 10,005. This is the product of the number of shares (100), the per share NAV at closing ($100), and the agreed-on premium to NAV the buyer agreed to pay (+$0.05). If the agreed on spread to NAV had been -$0.05 cents, then the payment would be $9,995. The quoted intra-day value of the fund at the time of the trade, $98, does not factor into the pricing of the trade. (68)

If specialists in ETMF shares are able to match buyers and sellers and clear trades at agreed on spreads to NAV at closing, then trading would not involve any new money (or securities) flowing into or out of the fund. In this circumstance, the ETMF would function like an ETF, and unlike an openend fund. Suppose, though, that would-be buyers of ETMF shares outnumber sellers, and that this causes the spread (i.e., the premium) to NAV to increase. At some spread to NAV, an arbitrage opportunity will ripen, meaning that it will make economic sense to create new ETMF shares until the spread narrows.

New shares are created in transactions between the fund (technically, the fund's distributor) and an AP. The process of ETMF share creation involves the following steps:

(1) The fund announces the "basket." The basket is the package of securities that the AP must tender in the swap for the creation units. The basket is not a pro rata slice of the fund's portfolio, but rather a slice of most of the fund's portfolio, with adjustments. Skewing between the basket and the fund's holdings is done to obscure the exact holdings of the fund (to hide its precise strategy), and for other reasons such as securities law constraints and because some securities held by the fund might be difficult to procure. (69)

(2) The AP compiles the basket.

(3) The AP notifies the fund of its intention to acquire "creation units."

(4) After the market closes, the fund transfers the creation units to the AP in exchange for the basket, plus a cash balancing adjustment. The cash is necessary because the basket will not equal the value of the creation units, owing to the skew, described above. In addition, the AP would pay the fund a transaction fee calibrated to cover the costs to the fund of processing the transaction. These costs would include the costs of clearing and settling the securities transfers, and of converting the cash paid by the AP into the securities the fund intends to hold in its portfolio (in other words, the cost of reversing the skew).

(5) The AP unbundles the creation units, and sells them on the market. The AP will have institutional expertise to estimate (and also to control and minimize) the administrative cost and risk involved in executing this choreographed series of steps. If the spread to NAV increases to a level where it will be profitable for the AP to engage in the creation process, the AP will have an economic incentive to do so.

Redemptions follow a similar pattern, in reverse. Significantly, the redemption process involves an in-kind payment by the fund for creation unit shares being redeemed. This is important because it is the transaction that purges the funds tax overhang, given that the transaction will fall within section 852(b)(6). Eaton Vance has been quite explicit that this is by design. In its application for exemptive relief with the SEC, it explains that "[t]hrough in kind redemption... ETMFs would seek to achieve tax efficiencies for its shareholders by avoiding the tax consequences of selling portfolio positions to meet redemption requests in cash[.]" (70) Similar statements are contained in Eaton Vance's patent application; (1)

The circumstances that would result in a redemption of ETMF shares by the fund arise when the discount to NAV on trades in the secondary market increases to a level at which it is in the economic interest of an AP to facilitate share redemptions. Like share creation, share redemption may only be done by APs. To redeem shares, an AP acquires ETMF shares in sufficient quantity to form a creation unit. They then deliver the creation unit to the fund in exchange for the basket, plus cash. The cash serves the same role in the redemption process as in the creation process, i.e., to obscure the fund's exact holdings, and to equalize the total consideration exchanged between the AP and the fund.

The AP's profit on the trade would be the spread between (a) the discount to NAV it paid to purchase the ETMF shares packaged into the creation unit and (b) any administrative costs in purchasing those shares, or in unwinding positions in the basket securities received from the fund as compensation in the redemption transaction.

With this understanding of the ETMF share creation and redemption process it is clear why ETMFs are a hybrid of open-end fund shares and ETF shares. Like open-end fund shares, new shares are issued and outstanding shares are redeemed by the fund itself. And, like ETF shares, intra-day trading is permitted. The important differences between ETMFs and open-end funds are (a) the circumstances under which ETMFs create or redeem shares are more limited than for conventional open-end funds, and (b) the identity of the parties eligible to undertake creation or redemption trades with the fund are also more limited. The important difference between ETMFs and ETFs is how trades are priced. Pricing of trades in ETF shares are negotiated just like stock trades (to the penny), whereas ETMF shares trade based on a spread to the fund's NAV at market close.

The structure is based on the idea that current and future investors in the fund do not need the fund to participate in the transactions that generate shareholder liquidity when inflows and outflows are balanced, implying no significant premium or discount to NAV. If the fund does not need to manage its flow, it will be less costly for the fund to operate, and the fund will pass the cost savings on to investors in the form of enhanced net yield. On the other hand, when spreads to NAV become significant, implying either a market surplus or shortage of ETMF shares, the fund can be called on to reduce or increase float to minimize spreads to NAV. When the fund is called on to participate directly in the market for its shares by buying or selling creation units, the cost to the fund is borne by the AP (and indirectly by the parties to trades with the AP), and not by the fund's shareholders more generally. It will thus be largely a matter of indifference to fund shareholders adopting a long-term buy-and-hold strategy that other investors in ETMF shares are frequent traders.

The benefits of this approach are similar to the benefits of the ETF structures (both the garden variety structure and Vanguard's embellishment): (1) it minimizes the need for the fund to manage its daily flow, and (2) the fund is able to redeem creation units in kind from APs, thereby purging fund overhang.

D. Re Flow

The final tax-advantaged structure worth discussing was the brainchild of Gordon Getty, son of the late oil tycoon J. Paul Getty. (72) Getty patented the ingenious idea, and the patent is now held by ReFlow (a firm Getty controls), which implements the strategy." In essence, ReFlow provides liquidity to mutual funds, for a fee, when they have net outflows from other investors. Obtaining liquidity from ReFlow means that the fund does not have to realize (taxable) gain on its portfolio to generate liquidity internally.

Recall that conventional mutual funds, in contrast to ETFs and ETMFs, stand ready to trade with individual shareholders (not just APs). In addition to being a regulatory requirement, this is essential to their existence given that fund shares do not trade on any stock exchange. (74) Trades between investors and the fund are priced based on the fund's NAV at market close.

On days when a fund subscribing to ReFlow's service has net outflows, (75) ReFlow offers to buy shares from the fund for its own account, making ReFlow a fund shareholder. On a later day or days when the fund has net inflows, the fund redeems ReFlow's shares, based on NAV. After 28 days, the fund is obligated to repurchase any shares ReFlow still holds (i.e., shares that were not already redeemed on days when the fund had net inflows). When the fund redeems ReFlow, it might pay cash for ReFlow's shares, or ReFlow might agree to take an in-kind distribution of portfolio holdings from the fund. (74)

To see the genius of this strategy, consider a contrived example. Assume that over a given month a fund has net inflows of $10 million on every even trading day and net outflows of $10 million on every odd trading day. If there are an equal number of even and odd trading days over the month, the net flows for the month are zero (inflows and outflows are in balance). Without the service offered by ReFlow, the fund would face a choice between bad options. It could keep cash on hand (uninvested capital) to redeem shareholders on odd days when there are net outflows. The pool of uninvested capital would be replenished on the following day when cash flows in from new investors. This uninvested capital would be a drag on investment performance.

Alternatively, the fund could fully invest all of its capital in portfolio securities (no cash hoard), but then the fund would have to liquidate portfolio holdings as needed to fund shareholder redemptions on days when there are net outflows. Liquidating holdings is costly in terms of transaction costs. It also imposes tax costs--specifically, it causes realization of tax overhang.

ReFlow gives funds a third option. Whether ReFlow's service is preferable to the alternatives depends on the fee ReFlow charges compared with the performance drag from the other methods of generating liquidity. ReFlow determines its fee using a Dutch auction, referred to as the "Auction Program," which is implemented by computer over the internet.

The Auction Program is described in depth in ReFlow's successful request for an SEC No-Action letter, and to a lesser extent in its patent application." The program is open to any open-end fund that elects to participate, assuming its regulatory affairs are in good order. (78) At the start of each business day, ReFlow announces the terms of the day's auction on its website. The announcement includes a description of the total funds ReFlow has available to invest on that day (the "Initial Auction Amount"), the minimum dollar amount for which any fund participating in the program may bid, and the minimum fee that ReFlow will accept from a fund bidding in the auction.

Prior to the close of the trading day, participating funds that wish to take part in that day's auction will place bids. (79) Each bid is an offer to pay ReFlow a fee calculated as a percentage of the value shares purchased by ReFlow on the date those shares are purchased. The bid has two parts--liquidity need and the maximum fee the fund is willing to pay. A fund can express its liquidity need as a percentage of the fund's daily net redemptions or as a specific dollar amount. Funds may not obtain more liquidity from ReFlow than is required to bring their net redemptions to zero.

ReFlow's computer system then accepts information from participating funds, including each fund's net purchases or redemptions for that day, the number of outstanding fund shares, and the fund's NAV. (80) As soon as ReFlow's computer system collects and processes all of the information, ReFlow announces the results on its web site. The Auction Program is automated; neither ReFlow nor any fund can influence the outcome of a given day's auction after the market close. The ReFlow software operates by undertaking the following steps, in order:

(1) ReFlow collects reports from funds, as described above.

(2) Funds in which ReFlow owns shares from prior days' trading are redeemed by those funds that are in a net inflow position on a given day. The redemption amount equals the fund's net inflows of new capital or ReFlow's holdings, whichever is smaller. (81)

(3) ReFlow adds the cash proceeds of such redemptions to the Initial Auction Amount; the sum of these two items is the "Final Auction Amount."

(4) Once ReFlow calculates the Final Auction Amount, the software conducts the auction. The participants in the auction on a given day are those funds that (a) submit bids or have standing bids to participate and (b) experience net outflows. The auction is a Dutch auction, meaning that the fee charged to all successful bidders is the lowest fee bid at which the entire Final Auction Amount is invested.

To illustrate, assume that fund-level flow information, NAVs, and bids are received from the funds. (82) Assume further that ReFlow has sold shares back to all of the funds in which ReFlow is a shareholder that experience net inflows, which generally triggers automatic redemption of ReFlows shares. After adding the proceeds of the redemption trades to the Initial Auction Amount, Final Auction Amount is, say, $1 million. Funds A through E bid in the auction, as follows:
Fund/Bidder  Dollar Amount of Bid  Fee Offer (Basis

A              $200,000            90
B              $300,000            80
C               $75,000            75
D              $125,000            45
E              $600,000            39
Sum          $1,300,000

The outcome of the auction is that Funds A through D receive all of the cash that they bid for at a fee of 39 basis points (the fee offered by E, the lowest qualifying bidder). Funds A through D collectively bid for $700,000, which leaves $300,000 of the Final Auction Amount for E, the lowest qualifying bidder. Fund E's trade with ReFlow is limited to $300,000 of the $600,000 bid by E, the limit set by the final auction amount.

There are other minor details regarding the implementation of ReFlow's auction, but this description captures the essence of the program. (83)

Consider the fee that funds should be willing to pay to ReFlow. Any fund that is eligible to purchase liquidity from ReFlow at auction necessarily has net outflows--otherwise it wouldn't need (or qualify for) ReFlow's service. A fund with net outflows must get cash from somewhere. The basic choice is between (a) supplying its own liquidity internally, for instance by selling securities from its portfolio or by keeping a cash hoard (i.e., not fully investing the funds' capital in the first place), or (b) contracting with ReFlow.

A fund participating in Relfow's program should bid an amount approaching the cost to the fund of supplying liquidity internally. Funds gain no advantage by bidding less than this amount based on information (or speculation) regarding the fee other funds would be willing to pay and hoping to obtain liquidity at a bargain. This would be a suboptimal auction strategy. The fee paid by all winning bidders is the lowest fee that clears the market for the Final Auction Amount, not the actual bids funds make to participate in the auction. This market-cleaning fee will result in a cost savings to all winning bidders, assuming that they have accurately gauged the cost of selfsupplied liquidity, taking into account tax on portfolio overhang, drag on portfolio performance, administrative frictions, etc.

Participating funds are offering ReFlow a fee which is, from the funds' perspective, in part a surrogate for the tax the fund shareholders would owe if the fund were required to manage its liquidity without help from ReFlow. Transacting with ReFlow might make sense for the fund and its shareholders, but is it wise policy to permit this? If policymakers are serious about collecting the tax that funds would owe--but for ReFlow's service--then they ought to view ReFlow with jaundiced eye. On the other hand, if policymakers are sanguine about planning of this type, then why subject fund shareholders to pass-through taxation of capital gains in the first place? If the government is not going to collect the tax, is it not better for the savings to benefit fund shareholders than for it to accrue as rents in favor of Getty, the inventor?


A. Introduction

Congress should revise the rules to mitigate the bad effects of mutual fund tax overhang. Present law is notable for its inefficiency, unfairness, and inconsistent application. Significant effort is spent by fund managers controlling the magnitude of their funds' capital gains overhang, effort that would be better spent on sharpening their investment thesis and its execution. Under the present system, shareholders in taxable accounts can be expected to focus at least some attention on the tax overhang issue prior to choosing among funds; if they do, it is time not spent selecting funds based on factors that should be more important to the project, and if they don't they are being punished for their inattention.

What would a better policy look like? Some superordinate tax reforms would cure the problem of mutual fund tax overhang. Repealing the corporate income tax, integrating the corporate and shareholder tax bases for all business corporations (not just for RICs and other specialized corporations), or implementing mark-to-market accounting for investment securities would all likely sweep away the issues described above. This article on mutual fund tax overhang is not the place for a discussion of these reforms.

The question I will focus on in this part is narrower. I will assume no fundamental change to the background structure of current law. Assuming overarching policy stasis, I will first consider the targeted reforms that have been proposed to address the problem of mutual fund tax overhang and, second, will offer an evaluation and recommendation.

B. Pure Pass-through Taxation

Some commentators have proposed pure pass-through taxation for mutual funds and their shareholders. (84) Apparently, the notion is that this would, or at least conceivably could, improve things compared with the status quo. If one considers this option in a bit more depth than has been attempted by prior commentary, it does not seem promising.

Many would reject the idea that Subchapter K--or some similar regime--ought to be unleashed on the hoi polloi. I expect that the concern would be that neither taxpayers nor the IRS is equipped to deal with the stifling complexity that would follow. To me, pure pass-through taxation of funds would be a mistake, both because complexity would be intolerably high (though possibly for different reasons than one might suspect) and because it would not address the capital gains overhang problem.

A significant share of the complexity of Subchapter K is a byproduct of section 704(b)'s substantial economic effect test. As noted above, section 18 of Investment Company Act provides that funds are limited to a plain vanilla capital structure; complex sharing arrangements of the type that provoke complexity under section 704(b) are prohibited. (85) Thus, the most feared aspect of partnership taxation would not be difficult to apply to mutual funds.

The difficulties in conforming Subchapter K or some similar pass through regime to mutual funds would be how to allocate fund income among the shareholders when there is frequent turnover in ownership of the fund. (86) It is very common for shareholders to enter and leave funds mid-year, either by buying or redeeming funds directly with the fund itself (as in an open-end fund) or by trading on an exchange (as with an ETF). Conceptually, if mutual funds were taxed as partnerships then the fund should divvy up income earned by the fund throughout the year among the shareholders based on time-weighted percentage ownership, gauged when the income being allocated was earned by the fund. There are at least two difficulties implementing such a system: (1) it is not clear that the fund has the data to make this allocation, and (2) even if it does, it is not apparent when certain items of income are "earned by the fund" in the relevant sense.

Take two simple examples of the second issue: is dividend income earned by the fund on the date the dividend is declared, on the dividend record date, on the date the dividend is paid, or over the interval between the dividend in question and the prior dividend? Is interest income accruing in favor of a fund that owns fixed income securities earned when the interest is paid, or as it accrues over the period specified in the indenture? There are better and worse answers to these questions, but there is no simple way to divvy up the income among the fund shareholders in a fashion that is both easy to administer and would result in comprehensive accounting (and flow through) of fund income to shareholders. Again, the basic difficulty is identifying who owns what fraction of the fund's outstanding shares for how long during the (ambiguous) interval the income was earned by the fund.

Handling income allocation issues in a satisfactory way is necessary, but not sufficient, to conclude that pass through taxation is a good move. In addition, pass through taxation would have to cure (or at least improve) the problem of mutual fund tax overhang. As partnership tax is currently structured in the U.S., it would not result in an improvement. If funds were taxed as pass-throughs they would still accrue unrealized built-in gain in their portfolios, just the same as under the present system. When new shareholders join the fund, the result would be what partnership tax lawyers refer to as a reverse section 704(c) allocation problem.

To translate the jargon, when a newcomer joins the fund, the fund would allocate the built-in gain to the capital accounts of the preexisting shareholders (on whose watch the gain accrued), not the newcomer. Even with a simple capital structure, accounting for section 704(c) issues is likely intolerably complex if implemented in the mutual fund context. (87) To get a rough sense of the issue without exploring all of the details, consider that when the fund sells out a position at a gain, different fund shareholders would have different amounts of taxable gain assigned to them depending on when they bought into the fund. Fund shares would no longer be fungible; they would differ in their tax attributes. This is both a conceptually and administratively challenging problem, one that is felt acutely be publicly traded partnerships under present law. (88) Other fiendish details would arise--but for present purposes it is enough to observe that an issue akin to capital gains overhang looms large in partnership tax, and the method of correcting it is stiflingly complex. In sum, seeking to cure the problem of capital gains tax overhang in Subchapter M by transitioning to a flow through model is not promising.

C. Shareholder-centric Reform

In 2001, Representative Jim Saxton (R-NJ), then the Chairman of the Joint Economic Committee, proposed allowing shareholders to exclude up to $3000 (single taxpayers) or $6000 (married, joint filing) of capital gains distributions per year. (89) The proposal, which was reupped by Saxton in 2004 (90) and again in 2007 by Representative Jim Ryan (R-WIS), (91) is elegant in its simplicity.

Distributee shareholders would have a maximum exclusion of $3000 or $6000 to allocate among all distributions received in a given year. (92) If distributions for the taxpayer were equal to or less than the maximum, all would be excluded, assuming the shareholder participated in automatic dividend reinvestment. The shareholder's basis in shares purchased with the reinvested dividends would be zero. Alternatively, if the shareholder had elected to use average cost basis, the shareholder would decrease her average basis in her mass of mutual fund shares by the quotient of (a) the number of shares purchased with the excluded dividend divided by (b) the total number of shares they held following the dividend reinvestment. (93)

In a policy paper accompanying his proposal, Rep. Saxton indicated that he chose the exclusion threshold ($3000 or $6000, depending on filing status) to moot the tax significance of mutual fund overhang for the vast majority of individual investors."' The report estimates that approximately 80 to 85% of taxpayers reporting capital gains tax distributions from mutual funds would be able to shield all of their gains under the proposal. (95)

The proposal is underspecified. To make it viable, policymakers would need to resolve several second-order implementation details. Open issues include the following:

Would the taxpayer have flexibility to allocate their exclusion among between short- and long-term gains, or are they compelled to allocate the exclusion in a particular order?

Does section 1014 apply to the gain excluded under the proposal or is there a special exception to section 1014?

(1) Who is eligible for the exclusion (e.g., should all taxpayers be eligible, or should there be a "kiddie tax"-type provision to prevent income splitting)?

(2) What is the shareholder's holding period for shares purchased with the (exempted) reinvested dividends (can the shareholder tack)?

(3) For taxpayers who have capital gains distributions in excess of the threshold, and who own more than one fund, how does the taxpayer allocate the exclusion among distributions received in a given year?

Various versions of the proposal have either ignored or taken positions on these particulars." For these details, good policy is to select a clear rule that can be administered straightforwardly.

Academic commentators have proposed reforms that are, in effect, variations on the proposal first floated by Saxton. The main differences between these variations and Saxton's proposal are how to calibrate the threshold for exclusion. Some have suggested setting it by reference to AGI, rather than as a dollar cap on dividends received." Others have suggested using a constant fraction of dividends received by each shareholder, so that taxpayers with larger fund holdings (or who owned funds with more gains realizations, or both in combination) would get a larger exclusion. (98)

An important change in tax administration that took place after most of these proposals were floated: broker basis reporting. Under section 6045(g), brokers are now required to track and report to taxpayers and to the IRS their customers' tax basis in "covered securities," (99) a category that includes shares in regulated investment companies." (100) This new reporting protocol would make the proposals easier to implement. Concerns voiced at that time included the additional paperwork burden on taxpayers who would have difficulty keeping things straight, or compliance problems if shareholder would be inclined to "forget" to reduce their per share basis in shares following dividend reinvestment. As things stand now, this seems like a less significant issue than it was before broker basis reporting.

D. Fund-centric Reform

An organizing principle underlying all of the shareholder-centric reforms is that, from the point of view of the fund, nothing has changed in terms of the nuts-and-bolts operation of Subchapter M. The legislative text that would implement these proposals would simply add new code section 1045, alongside other nonrecognition provisions aimed at individuals such as those for like kind exchanges, divorce, and casualty losses." (101)

It is also possible to center reform on the fund itself. One idea is to remove the obligation for funds to distribute gains. A second possibility is to repeal or reform section 852(b)(6) with an eye towards imposing the burden of tax overhang more uniformly among funds without regard to how they are structured.

Begin with the first idea. Professor John Coates has proposed that funds be relieved of the obligation that they distribute capital gains (possibly restricted to those held long-term). (103) He suggested that exempting the shareholders from tax--as under the shareholder-centric approaches described above--and relieving the fund from making capital gains distributions in the first place are "economically equivalent.""" This is true from the vantage point of a fund shareholder who fits within the threshold for exempting fund gains; however, there are important differences between fund- and shareholder-centric reforms.

First, if the elimination of capital gains tax is made at the fund level, it is necessarily a one-size-fits-all reform. All mutual fund shareholders, without regard to their income level or the total amount of mutual fund dividends they receive, would enjoy the benefit of the change. Second, it would simplify record keeping for the fund, which would no longer have to manage the process of reporting capital gains distributions and tracking basis adjustments triggered by dividend reinvestment. Third, fund-level exclusion of gains has the potential to open the US mutual fund market up to foreign investors. (104) Certain amendments were made to Subchapter M in 2004 in an attempt to make US-based mutual funds attractive to foreign investors, (105) but these changes did not result in an influx of foreign capital.

Turn now to the second idea: amending section 852(b)(6). The goal of such a reform would be to plug the loophole that is being exploited by funds using the structures described above. Mechanically, this could be accomplished by simply repealing section 852(b)(6) and imposing tax on the fund when it redeems its shares in kind. This is the default background rule for corporations other than regulated investment companies.

Another, more complex, way to accomplish this goal would be to force the fund to reduce the basis in the securities in its portfolio that are not distributed in-kind to preserve the built-in gain that is eliminated by any in-kind distribution. (106) This would be less radical in that in-kind distributions would continue to be nonrecognition transactions, but they would merely defer tax on the funds tax overhang, rather than eliminating it. It would also be possible to pass the gain on to the shareholder receiving the distribution implementing a carryover basis rule in place of the rule that applies now. (107) This would be unwise, however, as the fund could arrange to have one or more tax indifferent (e.g., foreign) APs who would be indifferent to their basis in the distributed securities.

E. Evaluation

Any of the reform proposals would an improvement to the status quo. The rank order among them depends on the goal of reform. The overarching goal of Subchapter M is to integrate the corporate and shareholder tax bases for mutual funds (and real estate investment trusts). This is desirable, the thinking goes, because small fry investors should be able to band together to hire professional advisors and diversify their market exposure. Doing so would be cost prohibitive if pooling capital subjected them to the corporate double tax. To further the overarching goal, reform should aim towards generating for mutual fund shareholders the tax treatment they would have attained if they instead invested in separately managed accounts.

If one accepts this as the objective of reform, the last of the reform options discussed above--repealing [section]852(b)(6) or reducing basis on retained securities to preserve the built-in gain leaving the portfolio in an in-kind distribution--is misguided. It is a minor step forward in that it imposes the overhang tax on funds of all types, rather than imposing the overhang tax unevenly among funds depending on their structure. But the basic problem remains, i.e., fund shareholders often will be taxed on what is, from their perspective, phantom income owing to realizations at the fund level.

The other fund-centric reform--eliminating the requirement that funds distribute gains--would be easier to implement than shareholder-centric reform from an administrative and recordkeeping point of view, but it is a blunt approach. It is impossible under such a scheme to adjust the rules based on individual shareholders' particular circumstances. The desirability of this option (compared with shareholder-centric reform) depends on how one reconciles administrative considerations and the ability to fine tune tax burdens of individual shareholders based on their personal circumstances.

In the age of computer-automated accounting, electronic records, and broker basis reporting, the record keeping benefit achieved by fund-centric reform probably does not matter much. (108) If annual fund-shareholder paperwork could be eliminated altogether it would be a significant advance, but this seems infeasible. No one has proposed to adjust the treatment, by funds or shareholders, of fund income other than capital gains. The preexisting structure would continue to apply to ordinary income items (such as interest) and other items (such as foreign tax credits), meaning they would pass through from fund to shareholder. This process requires a paper trail. Therefore, eliminating the requirement that funds distribute gains would, at best, reduce--but would not eliminate--recordkeeping and reporting.

An issue that is raised by fund-centric reform is the possibility that rich taxpayers would form funds to exploit the new scheme as a de facto exception to the realization rule." (1) " The following devious scheme would have to be ruled out: one or more wealthy taxpayers form a RIC and enter into rebalancing trades within the portfolio in ways that would ordinarily draw tax if the investments had been made via separate account, but which are exempt from tax because of the interposition of a fund between the shareholder and the trades that trigger realized gain. This strategy seeks to exploit the inability to distinguish between the bad effects of tax overhang and the natural tax consequences of portfolio rebalancing, discussed above. (110) There are solutions to this problem targeting egregious cases, but they would add complexity."

This leaves shareholder-centric reform. Although not perfect, it seems to be the best option. It is broadly consistent with the overarching policy of Subchapter M. It would eliminate most of the wasteful planning by funds and shareholders that goes into avoiding the burden of the overhang tax. It would not be ripe for abuse at least in the obvious way that fund-centric reform would be. On the downside, the recordkeeping and reporting burden of this option would be more significant than for some of the other options. Yet, as noted, with automation and third party reporting this drawback shouldn't doom the approach.


Mutual fund tax overhang is an important issue for individuals investing in mutual funds, a group that includes most of us. Even if tax overhang does not affect you directly (for instance, because you fund investments are in your 401(k) plan or IRA), the funds in which you invest are affected. Consequences include wasteful planning and living with the general distraction created by the rules that generate tax overhang.

Moreover, under present practice, the burden resulting from tax overhang is lumpy. Some funds are subject to a significant burden and others to no burden at all. This pattern is a result of tax savvy and idiosyncratic opportunity, not a deliberate policy.

As a technical matter, the problem would be relatively easy to fix. It is time for a change.

Ethan Yale

(1) This is an oversimplification. For all of the details, see SUSAN A. JOHNSTON & JAMES R.BROWN. JR., TAXATION OF REGULATED INVESTMENT COMPANIES AND THEIR SHAREHOLDERS (2015); RICHARD M. HERVEY, TAXATION OI- REGULATED INVESTMENT COMPANIES AND THEIR SHAREHOLDERS, TM PORTFOLIOS 740-3D (2015). The fund is only required to distribute 90 % of its taxable income. I.R.C [section] 852(a)(1)(A). In practice, most Regulated Investment Companies (RICs) distribute most, or all, of their net capital gains. See supra HERVEY, at A-l 19.

(2) Mark J. Roe, Political Elements in the Creation of a Mutual Fund Industry, 139 U. PA. L. REV. 1469,1483 (1991); JOHNSTON & BRowN. supra note 1, at 53.02 ("Thedistributionbased system of integration adopted in Subchapter M was originally intended to eliminate corporate-level taxation on the portion of a RICs income that is taxed at the shareholder level on a current and nonpreferential basis....").

(3) Michael J. Barclay et al., Open-End Mutual Funds and Capital-Gains Taxes, 49 J. FIN.ECON.3,11 (1998).

(4) I.R.C. [section] 852(b)(3).

(5) I.R.C. 8 1001(a).

(6) On average, equity mutual funds tend to have high portfolio turnover. The annual, asset-weighted average turnover rate lor equity mutual funds during the period 1980-2015 was 60 %. See INVESTMENT COMPANY INSTITUTE, 2016 INVESTMENT COMPANY FACT BOOK 37 fig.2.6 [hereinafter ICIFACTBOOK (2016)].

(7) This rule was initially codified at I.R.C. section 311(d)(2). In 1986, it was shifted from Subchapter C to Subchapter M, and is now codified at I.R.C. section 852(b)(6). For commentary on the rule from the time of its enactment, see Clifford L. Porter, Redemptions of Stock with Appreciated Property: Section 311(d), 24 TAX LAW. 63 (1970).

(8) The gain inherent in the distributed property is eliminated in the distribution. It does not pass on to the shareholder participating in the in-kind redemption. The shareholder has gain (or loss) equal to the difference between the value of the property distributed and the shareholder's basis in the shares being redeemed. The shareholder's basis in the distributed property is set equal to its value on the distribution date. I.R.C. [section] 302(a), (b)(5); l.R.C. [section] 1012.

(9) Investment Company Act of 1940. 15 U.S.C.A. [section] 2(a)(32) (2012).

(10) JOHNSTON & BROWN, supra note 1, at 5J3.06[2][c] (the ICA rule permitting in-kind redemption "is intended to relieve open-end investment companies from having to make forced sales of their securities that might otherwise occur if redemptions have to be satisfied with cash distributions. This purpose of the 1940 Act rule would be undermined if RICs were required to recognize gain on distributions of property in respect of redemption requests since, to avoid entity-level taxation on that gain, RICs would be forced to make either additional distributions of property (potentially triggering the recognition of more gain) or distributions of cash (also potentially requiring additional securities sales, and thus recognition of more gain).").

(11) In 2015, shareholders in equity funds reinvested over 93 % of their dividends. The reinvestment rate has generally remained stable through time. 1CI FACTBOOK (2016), supra note 6, at 200.

(12) Joel M. Dickson et al., Tax Externalities of Equity Mutual Funds, 53 NAT'LTAX J. 607, 609-10, 625 (2000) ("We have demonstrated that the existence of positive net cash flow can provide a significant benefit to existing mutual fund shareholders.").

(13) Id. at 609 ("Effectively, fund expenses arc fully deductible for all taxpayers because they lower the taxable income received by shareholders. Generally, investment fees assessed [for separate accounts]... are itemized deductions that can be used only to the extent they exceed 2 % of adjusted gross income."); see also l.R.C. [section] 67.

(14) Dickson et al., supra note 12, at 609. Dickson, Shoven, and Sialm demonstrate using simulations that the negative effects of fund taxation generally dominate the positive effects, but not always. Which effect dominates depends on the pattern of net cash flows for the fund, details regarding the accounting method the fund uses to compute its gain, and other factors.

(15) B might further reduce the price paid by the dividend tax that B would owe on any distribution of corporate profits.

(16) Long ago the Supreme Court recognized that inchoate corporate taxes would diminish the price buyer would be willing to pay for corporate stock. United States v. Phellis, 257 U.S. 156, 172 (1921) (the buyer of corporate stock "simply stepped into the shoes... of the stockholder whose shares he acquired" with respect to accrued, unpaid taxes).

(17) THOMAS LEE HAZEN, TREATISE ON THE LAW OF SECURITIES REGULATION [section] 20:24 (7th ed. 2016). Underwriters are permitted to adjust per share NAV to account for sales loads and fees, but not taxes. Id. The description in the text is accurate for open-end mutual funds. For closed end funds, there is no similar limitation; rather, prices are set by the buyers and sellers in the market, just like the prices of common stock. Economists have found that some fraction of the discount to NAV usually observed for closed-end fund shares can be explained by tax overhang. See Charles M. C. Lee et al., Investor Sentiment and the Closed-End Fund Puzzle, 46 J. FIN. 75. 79-80 (1991); Burton G. Malkiel, The Valuation of Closed-End Investment Company Shares, 32 J. FIN. 847-48 (1977) ("'A commonly accepted rational for the existence of discounts on closed-end fund shares is that investors face a built-in capital gains tax liability when the buy into" funds that have significant tax overhang).

(18) Hedge funds and private equity funds are taxed as partnerships. Partnership taxation can sometimes generate tax overhang for new investors in a seasoned fund; the partnership tax bar has, however, found a way to avoid the bad effects of tax overhang. 1 describe this in more detail below.

(19) The claim that the burden on mutual fund tax overhang is regressive is subject to a significant caveat. Mutual fund tax overhang only imposes a (direct) burden on taxpayers with mutual fund holdings outside of a retirement plan. Taxpayers in this category, if they plan thoughtfully, have eclipsed the annual savings limits for tax deferred accounts (401(k) plans, IRAs. etc.), and are thus relatively wealthy (or at least relatively high-income) compared to the average taxpayer. Thus, the issue here is not "rich versus poor" but instead the proper graduation in tax burdens for taxpayers sliding up the scale from rich to ultra-rich.

(20) General Utilities & Operating Co. v. Helvering, 296 U.S. 200 (1935). In General Utilities, the Supreme Court held that corporations could distribute built-in gain property without triggering tax on the built-in gain. Beginning in 1969. Congress began the process of legislatively overturning the rule of General Utilities, subject to a number of exceptions, including the exception now codified at section 852(b)(6).

(21) Barclay et al., supra note 3, at 7.

(22) Id. at 19,33-34.

(23) Id. See generally Paul G. Mahoney, Manager-Investor Conflicts in Mutual Funds, 18 J.ECON.PERSP. 161 (2004).

(24) Ajay Khorana & Henri Servaes, The Determinants of Mutual Fund Starts, 12 RHV. FIN. STUD. 1043, 1065-66 (1999) ("[M]orc equity funds are started when asset inflows into existing funds expose potential shareholders to high eapital gains realizations.").

(25) The level of wealth required to diversify without pooling, and to hire professional management, is being driven down by technology. The rise of so-called "robo-advisors" will manage customer accounts using computer algorithms. Although this is an attractive option for some, the fees are generally higher, sometimes an order of magnitude higher, than for many low-cost index mutual funds.

(26) David S. Miller & Jean Bertrand, Federal Income Tax Treatment of Hedge Funds, Their Investors, and Their Managers, 65 TAX LAW. 309 (2012).

(27) Id. at 333.

(28) MATTHEW LAY ET AI... NEW YORK STATE BAR ASSOCIATION TAX SECTION RETORT ON AGGREGATION ISSUES FACING SECURITIES PARTNERSHIPS UNDER SUBCHAPTER K 35-36 (2010) (stating that stuffing allocations are controversial and citing to commentators taking differing positions on the permissibility of stuffing allocations under current law).

(29) Id.

(30) Andrew W. Ncedham, The Problem with Stuffing Allocations, 141 TAX NOTES 737 (Nov. 18,2013).

(31) This example is similar to the one used by Needham. See id. at 737-38.

(32) Bond funds arc less likely to have overhang, given that bond returns arc predominately comprised of interest income, not capital appreciation.

(33) Clemens Sialm & Hanjiang Zhang, Tax Efficient Asset Management: Evidence from Equity Mutual Funds 33 (Nat'l Bureau of Econ. Research, Working Paper No. 21060, 2015).

(34) Daniel Bergstrcsser & James Poterba, Do After-Tax Returns Affect Mutual Fund Inflows'?, 63 J. FIN. ECON. 381,387 (2002).

(35) See U.S. Dep't of the Treasury, Office of Tax Analysis. Tax Expenditures FY2020 (Released October 2018) at 33, available at One might object that the consequence of mutual fund tax overhang results in a "mere" timing difference; it is not an outright exclusion rule (like the nontaxation of employer provided healthcare, the largest tax expenditure). See id. This is true, but note that the tax expenditure budget is constructed on a cash basis and includes many rules that shift income through time, such as deferral of CFC income, among many others. Id. at 2.

(36) Sialm & Zhang, supra note 33, at 2.

(37) Dickson et al., supra note 12, at 615.

(38) Id.

(39) Id.

(40) See General Utilities Operating Co. v. Helvering, 296 U.S. 200.


(42) Id. at 309.



(45) STRAUSS, supra note 43, at 17:2.2.

(46) Id.

(47) See ETF vs. Mutual Funds: A Comparison, VANGUARD, (last visited Mar. 31, 2019).

(48) Roger M. Edelen. Investor Flows and the Assessed Performance of Open-End Mutual Funds, 53 J. FIN. ECON. 439 (1999); see also Dickson et al., supra note 12.

(49) STRAUSS, supra note 43, at 17:2.4 ("purchases and redemptions of ETFs generally are effected in kind").

(50) I.R.C. [section] 302(a), (b)(5); I.R.C. [section] 1012.

(51) See Zachary R. Mider et al., The ETF Tax Dodge Is Wall Street's 'Dirty Little Secret,' BLOOMBERG BUSINESSWEEK (Mar. 29,2019),; see also Elisabeth Kashncr. The Heartbeat of ETF Tax Efficiency, FACTSET (Dec. 18. 2017).


(53) U.S. Patent No. 6,879,964 B2 (filed Apr. 12. 2005) (entitled "investment company that issues a class of conventional shares and a class of exchange-traded shares in the same fund"). The validity of Vanguard's patent is open to doubt in the light of Alice Corp. v. CLS Bank International, 573 U.S. 208 (2014). The same is true for the Eaton Vance and ReFlow patents discussed below. The Alice Court invalidated patent claims on the ground that an abstract idea could not be patented just because it was implemented using a computer. In the wake of Alice, business method patents have been routinely invalidated. See Loyalty Conversion Sys. Corp. v. American Airlines, Inc., 66 F. Supp.3d 829 (E.D.Tex. 2014). Nevertheless, news reports indicate that various fund complexes have licensed or are in active talks to license the structures used to avoid tax overhang, including Vanguard's patent and the others discussed below, implying the patent claims have some merit. E.g., Jackie Noblett, Vanguard Looking to License its Cost Cutting ETF Patent, FIN. TIMES, Apr. 26,2015. It is also worth noting that patents of the sort discussed in this article were ruled out by LeahySmith America Invents Act on September 16, 2011, P.L. 112-29 (Sept. 16, 2011), [section]14, as discussed in an online symposium issue of the Columbia Journal of Tax Law. See Online Symposium, Tax Matters, 3 COI.UM. J. OF TAX L. 1. Some or all of the patents discussed in this article would be likely invalid under the law as revised if it applied to them, but all are grandfathered (the rule making tax strategies patentable has prospective effect only).

(54) Institutions that elect to invest using conventional shares might be offered an institutional version of the open-end share class with a lower fee. This is permitted under the ICA. even though multiple share classes are generally not allowed. 15 U.S.C. [section] 80a-18(0(1)); see 17C.F.R. [section]270.18f-3.

(55) Vanguard Index Funds, et al.. Investment Company Act Release No. 24789, 65 Fed. Reg. 79439 (Dec. 19, 2000); Vanguard Index Funds, et al.; Notice of Application, Investment Company Act Release No. 24680, 65 Fed. Reg. 61005 (Oct. 13, 2000).

(56) 15 U.S.C. [section] 80a-[section] 18(0- The SEC staff has taken the position that section 18 restricts any material difference in the rights of an investment company's shareholders. See, e.g., ReFlow Funds, LLC, SEC No-Action Letter, 2002 WL 1493234, at *6. *14 (July 15. 2002) ("Section 18(f) is intended to limit a fund's ability to leverage its portfolio through issuance of senior securities and through borrowing.... [W]e have taken the position that any action by an issuer that results in material differences among the rights of its shareholders may create a senior security in violation of Section 18.").

(57) 17C.F.R.[section]270.18f-3.

(58) Vanguard Index Funds et al.. Investment Company Act Release No. 24789 , 65 Fed. Reg. 79439 (Dec. 19,2000).

(59) In practice, there might be a single class of open-end fund shares, or more than two classes.

(60) U.S. Patent No. 6,879,964 B2 col. 3 (issued Apr. 12, 2005).

(61) Id.

(62) Id. ("The intra-day trading feature of the [ETF] class will draw market timers out of the conventional shares, where they cause problems, into the [ETFjs, where they do not").

(63) Id. See also Edelen, supra note 48.

(64) The patent is perfectly explicit on this point, explaining that "by selecting the lowest cost lots of each stock distributed" in creation unit in-kind trades the fund "reduces the unrealized capital gains that currently exist in the fund, thereby benefitting existing shareholders." U.S. Patent No. 6,879,964 B2 col. 3 (issued Apr. 12, 2005).

This approach (a form of cherry picking), which maximizes the benefits to the shareholders, contravenes the IRS private letter ruling policy. In the context of rulings under section 852(b)(6), the IRS has required, as a condition to issuing a favorable ruling, representations from the fund that in-kind distribution is comprised of property with a tax basis proportional to the tax basis of the fund's portfolio. I.R.S. Priv. Ltr. Rul. 200536002 (June 7, 2005), I.R.S. Priv. Ltr. Rul. 200509013 (Nov. 17. 2004), I.R.S. Priv. Ltr. Rul. 200414043 (Dec. 31, 2003), I.R.S. Priv. Ltr. Rul. 200341014 (Jul. 1, 2003). There is no statutory or regulatory support for this requirement, and my understanding based on conversations with knowledgeable industry insiders and their advisors is that, at present, few funds seek rulings on the application of section 852(b)(6).

(65) A research report by Morningstar reports that the two largest ETF issuers--Vanguard and BlackRock--have been arguing publicly about whether Vanguard's dual-share-class structures results in inefficiency for ETF shareholders. The report concludes that it does to some extent, but notes that this might be a result of the torrid growth of the fund (huge net inflows) creating an environment where authorized participants were buying far more creation units than they were selling, so the fund could not purge its gain. See JUSTICE & LEE, supra note 52, at 6-7.

(66) NextShares regulatory documents--SEC filings, prospectuses, and the like--are available online. Regulatory and Technical Documents, NEXTSHARES.COM (Mar. 25, 2019),

(67) U.S. Patent No. 7,444,300 Bl (issued Oct. 28, 2008). See also Eaton Vance Management, et al., Investment Company Act Release No. 31333 (Nov. 6,2014).

(68) Eaton Vance Management, et al., Investment Company Act Release No. 31333 (Nov. 6,2014).

(69) U.S. Patent No. 7.444,300 BI, col. 6 (issued Oct. 28, 2008).

(70) Eaton Vance Management, et al.. Investment Company Act Release No. 31333, at 4 (Nov. 6,2014).

(71) U.S. Patent No. 7,444,300 Bl col. 13 (issued Oct. 28, 2008) ("The general requirement for in-kind creation and redemption not only protects fund shareholders from the cost of providing liquidity to traders by creating a clear audit trail for the order entry process, redemption in-kind (or partly in cash at the option of the fund) offers substantial advantages for taxable shareholders through deferral of capital gains realizations until a shareholder decides to sell fund shares.").

(72) Justin Hibbard, How Gordon Getty Got To "Aha!". BLOOMBERG NEWS (Oct. 10. 2005). This article describes an interview with Getty in which her relates that he had the idea while on safari in Africa. "It just sort of came to me." Id.

(73) Amazingly, "[w]hen he was setting up ReFlow, Getty couldn't find a computer programmer to write a program he wanted in Java. So he taught himself the language over the weekend and showed up on Monday with a rudimentary version of the program." Id.

(74) 15U.S.C. [section]80a-22(e).

(75) When a fund is selling more new shares than it is redeeming, the fund is said to be experiencing "net inflows." When redemptions outstrip new sales, the fund is experiencing "net outflows."

(76) See ReFlow Funds, LLC, SEC No-Action Letter, 2002 WL 1493234 (July 15, 2002), ; see also U.S. Patent No. 7.444.300 (issued Oct. 28, 2008).

(77) See ReFlow Funds, LLC, SEC No-Action Letter, 2002 WL 1493234 (July 15, 2002),

(78) The fund participating in the program must make certain assurances to ReFlow. such that the Fund's board of directors has approved participation in the Auction Program, and that it is in compliance with all legal and regulatory requirements applicable to open-end RICs.

(79) Funds may modify or cancel their bids up until market close, when the bids are fixed.

(80) For some funds, this information is not available until early in the morning the next day.

(81) ReFlow can suspend redemptions but stated in its request for a SEC No-Action Letter that it exercise this right rarely, if ever. See ReFlow Funds, LLC, SEC No-Action Letter, 2002 WL 1493234 (July 15, 2002),

(82) This example is taken from ReFlow's No-Action Letter. See ReFlow Funds, LLC, SEC No-Action Letter, 2002 WL 1493234 (July 15,2002),

(83) Details include an error correction mechanism and a rule that prevents ReFlow from violating the anti-pyramiding rules under section 12 of the Investment Company Act. See ReFlow Funds, LLC, SEC No-Action Letter, 2002 WL 1493234 (July 15, 2002),

(84) John C. Coates IV, Reforming the Taxation and Regulation of Mutual Funds, 1 J. LEGAL ANALYSIS 591,614 (2009) (concluding that "[F]or a large number of middle-class mutual fund investors-particularly those in lower tax brackets who invest in large company index funds, which tend to have the lowest net capital gains-it is not clear that the benefits of reduced capital gains taxes would be worth the annual additional tax compliance and record-keeping costs that partnership tax would create."); John Morley, Collective Branding and The Origins of Investment Fund Regulation, 6 VA. L. & Bus. REV. 341, 398 (2012) (concluding that "passthrough partnership taxation [might be] impractical because accounting for notional income and losses would generate very large amounts of paperwork." but that the present system is "at least partly a historical accident" and that "alternatives ought to be taken seriously").

(85) See 15 U.S.C. [section] 80a-[section] 18(f); see, e.g.. ReFlow Funds, LLC, SEC No-Action Letter, 2002 WL 1493234, at *6, * 14 (July 15,2002).

(86) In partnership tax jargon, the question would be how to deal with issues akin to those that arise under section 706(d).

(87) See Mark P. Gergen, The End of The Revolution in Partnership Tax. 56 S.M.U. L. REV. 343 (2003).

(88) See, e.g., Deborah Fields et. al., Triangles in a World of Squares: A Primer on Significant U.S. Federal Income Tax Issues for Natural Resources Publicly Traded Partnerships, TAXES, Dec. 2009, at 30 ("[T]he need for fungibility does not intersect well with some of the partnership tax rules. Many of these rules were drafted well before the growth of PTPs and arc focused on a world in which buyers and sellers of partnership interests know who each other are. Under some of those rules, the tax characteristics of a purchased interest in the hands of the buyer can be dependent, in part, on the tax characteristics of such unit in the hands of the seller.").



(91) H.R. 387, 110th Cong. (2007).

(92) Id. at [section] 1. By the time of Saxton's 2007 proposal the exclusion amounts had grown to $5,000 and $10,000.

(93) For instance, suppose that a shareholder owns 100 shares in a fund with a per share basis of $200, or $2 per share. This is the shareholder's only fund. The fund makes a capital gains distribution of $0.75 per share this year. The taxpayer's distribution is $0.75 x 100 shares = $75, all of which is excluded. Suppose that the funds NAV is $5 per share on the date of the distribution. If the shareholder participates in automatic dividend reinvestment, her distribution would fund the purchase of $75/5=25 new shares. If the shareholder tracks basis by share block, her basis in those 25 shares would be zero. If she uses average cost basis her basis in her 125 post reinvestment shares would be reduced from an average per share basis of $2 (the pre-distribution average) to $200/125=$ 1.60 per share. This is a reduction in per share basis of 20 %, equal to the number of shares purchased with the reinvested dividend ($25) divided by total post-dividend holdings ($125).

(94) SAXTON, supra note 89, at 18-19.

(95) Id.

(96) See H.R. 387, 110th Cong. (2007); supra notes 89-90.

(97) Coates, supra note 84, at 616.

(98) Samuel D. Brunson, Mutual Funds, Fairness, and the Income Gap, 65 ALA. L. REV. 139, 160-61 (2013).

(99) I.R.C. [section] 6045(g)(1).

(100) I.R.C. [section] 6045(g)(3).

(101) See H.R. 387, 110th Cong. (2007).

(102) Coates, supra note 84. at 615.

(103) Id. at 616.

(104) Id. at 617.

(105) American Jobs Creation Act of 2004, Pub. L. No. 108-357. [section]411. 118 Stat. 1418 (2004); HERVEY, supra note 1. at A-197.

(106) This approach is proposed by a recent article. See Steven Z. Hodaszy, Tax-Efficient Structure or Tax Shelter? Curbing ETFs' Use of Section 852(b)(6) for Tax Avoidance, 70 TAX LAW. 537 (2017).

(107) See I.R.C. [section] 302(a), (b)(5); I.R.C. [section] 1012.

(108) See supra Part V.3.

(109) Coates, supra note 84, at 615.

(110) See supra Part 11.4.

(111) Id.
Table IV.A Estimate of Equity Mutual Fund Tax Overhang, 1990-2012

                                   mean   median  std. dev.

Tax Burden (% per year)             1.12  0.43     1.58
STCG (in% per year)                 0.81  -        2.40
LTCG (in% per year)                 2.78  0.09     4.67
Short Term CG Overhang (% of NAV)   1.80  1.11     5.57
Long Term CG Overhang (% of NAV)   10.27  7.79    18.10

Source: Clemens Sialm & Hanjiang Zhang, Tax Efficient Asset
Management: Evidence from Equity Mutual Funds 33 tbl. 1 (Nat'l Bureau
of Econ. Research, Working Paper, Dec. 27, 2014).

Table IV.B Estimate of Mutual Fund Tax Overhang and Realized Capital

Year  Realized CG   Unrealized   Observations
      Yield         CG Overhang
      (% per year)  (% of NAV)

1996   4.9           7.81        3562
1997   6.92          9.28        4033
1998   5.81         10.91        3490
1999   7.22         17.27        4062
2000  12.89          1.29        4133
2001   3.26         -3.58        2217
2002   2.23          3.46        1120
2003   4.98         10.92        1737
2004   3.66         12.85        3498
2005   6.3          11.83        4149

Source: Feng Chen, Arthur Kraft, & Ira Weiss, Tax Planning by Mutual
Funds: Evidence from Changes in the Capital Gains Tax Rate, 64 Nat'lTax
J. 105, 119 (2011).

Table IV.C Estimate of Mutual Fund Tax Overhang for 20 Largest Equity
Funds on Dec. 31. 1999, Summary Statistics (*)

              Unrealized CG x NAV

Average (**)  39.14%
Max           56.00
Min           10.00
Std. Dev.     12.23

(*) Funds closed to new investors excluded
(**) Simple (unweighted) average
Source: Daniel Bergstresser & James Poterba. Do After-Tax Returns
Affect Mutual Fund Inflows, 63 J. FIN. ECON. 381,392 (2002).

Table IV.D Estimate of Equity Mutual Fund Tax Overhang, and Resulting
Implied Tax Liability

(billions of dollars)
Estimate                                    A         B         C

Total Value of Stock Holdings of U.S.
Open-End                                    $10,817   $10,817   $10,817
Funds (*)
Taxable Investor Share (**)                      41%       41%       41%
Overhang Ratio                                   10%       20%       30%
Overhang Attributable to Taxable Investors     $443      $887    $1,330
Effective Tax Tate                               10%       10%       10%
Implied Tax Liability                           $44       $89      $133

(*) Author's calculations as described in the text
(**) Investment Company Institute, 2018 Investment Company Fact Book
234 (Table 27), 287 (Figure A3)

TABLE V.A Comparing Tax Efficiency of Open-end Funds and ETFs

                         5-year   10-year  15-year

Number of ETFs           10/36    10/26     1/2
Number of Mutual Funds   23/36    17/36    13/36
Average of ETFs           0.0269   0.0513   0.1957
Average of Mutual Funds   1.1561   0.4630   0.5650

TABLE V.B Tax Burden by Fund Type

                                  ETFs      Matched Index
                                            Mutual Funds

Tax Burden (% per year)               0.34      0.65
Before-tax Teturn (% per month)       0.50      0.51
After-tax Teturn (% per month)        0.47      0.46
Total Distributions (% per year)      1.74      3.11
Dividend Yield (% per year)           1.67      1.46
Short Term Gains (% per year)         0.05       033
Long Term Gains (% per year)          0.02       133
Expense Ratio (% per year)            0.30      0.31
Turnover (% per year)                31.39     28.73
Total Net Assets ($ millions)     3,257.74  2,649.15
Age (years)                           5.14      7.94
COPYRIGHT 2019 Virginia Tax Review
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2019 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Yale, Ethan
Publication:Virginia Tax Review
Date:Mar 22, 2019

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters