A closed-end fund is a fund run by a portfolio manager that trades on a market. Fund holders own shares in the fund, but the fund doesn’t allow redemptions nor new issuances of those shares. In contrast, an open-ended fund allows holders to give their shares to the manager of the fund, who pays them out the “net asset value” (NAV) of their shares. The NAV is simply the current market price of the underlying assets that the fund holds.
The share price of a closed-end fund can trade at a discount or a premium to the NAV of the underlying securities.
The very existence of a discount or premium calls into question the efficient market hypothesis. If the efficient market hypothesis is true, then buying a basket of closed-end funds will provide a market return (minus fees). Because no stock is known ex ante to outperform, buying a basket of closed-end funds simply gives you a diversified market portfolio. If this were true though, there shouldn’t be such a high discount on some of these funds. In New Zealand, the Marlin Global fund trades at a (as of the end of last week) 17.6% discount.
Now, to tell you a secret, I really do want the efficient market hypothesis to be false. If it is, all this blogging, reading, and investing might have a chance of succeeding and I won’t have been wasting all this time. If you read Brooklyn Investor’s post on Phillip Goldstein, you’ll see one person who seems to be earning above market returns by investing in closed-end funds at a discount to their NAV.
Reviewing the finance literature of closed-end funds, there are effectively seven reasons why the discount might be rational. If the discount is a rational feature, the market might be efficient and the discount might not a source of subsequent outperformance.
The seven potential reasons are:
- Built-in potential gain tax.
- Distribution policies
- Investments in restricted stock
- Holding of foreign stock
- Past performance
- Portfolio turnover
- Management fees
Built-in potential gain tax
This doesn’t apply. For a New Zealander investor (like me) the fund is subject to the Portfolio Investment Entity (PIE) regime. This taxes the fund by deeming five percent of the net assets to be income (on which tax is then charged). As long as investors hold more than $50 000 of foreign assets (bit more complicated than this) in their personal portfolios, there is no tax advantage from holding assets directly as opposed through the fund. There is no capital gains tax in New Zealand, so the explanations in the USA about investors no longer having an “option” to bring forward capital losses and delay capital gains cannot apply.
I’ll come back to this at the end, but Marlin Global has a particularly attractive distribution policy that should reduce its discount. Basically, Marlin Global pays out, quarterly, 2% of the net asset value of the fund. Annually that’s 8%. The pay-out ratio is almost 10% when you look at the current market price as compared with the NAV and the distribution policy. One thing to keep in mind is that this is not a true “yield” because you’re not getting paid the income from the fund (or the dividends from the companies it owns). You’re being paid 8% of the net asset value of the fund yearly. You’re getting capital amounts back, not income amounts.
To me, that’s a good news story as it should minimise the difference between the NAV and the share price. If the manager cannot grow NAV by 8% a year, reduces the assets under management. In turn, this could force the fund to close and redeem at NAV as assets shrink. Of course, if the manager can grow NAV at over 8% per year, the discount should shrink on its own as investors start to believe that the manager has some ability. I conclude that distribution policies cannot explain the discount.
Investments in restricted stock
This is the idea that the NAV might be overstated because the fund owns “restricted stock” that it cannot sell. From a quick review of Marlin Global’s holdings, that doesn’t seem to be the case. However, it does hold positions in some very small companies. If the fund tried to sell those quickly it would likely not receive the current market price. I conclude that this might explain a very small discount.
Holding of foreign stock
This goes both ways. Some posit (and this is borne out in plenty of evidence) that the holding of foreign stock should create a premium, as it is likely to be a much more cost-effective way to get exposure to foreign stocks than holding the stocks directly. Inevitably a small investor will have larger trade and foreign exchange fees than a fund manager. Marlin Global holds non-New Zealand or Australian investments. This is one way to remove the “home bias” – invest in a foreign closed-end fund at a discount. I conclude this doesn’t explain any of the discount.
Everyone knows that past performance doesn’t predict future results, don’t they? This explanation for the discount assumes that past performance does predict future results, and therefore poor past performance requires a discount. This raises the inherent contradiction in looking at closed-end funds. If we assume the market is efficient, then past performance doesn’t predict future results, and there should be no discount. But we do see a discount. But if we assume the market is inefficient, then past performance does predict future results, and there should be a discount. But is the discount evidence of market efficiency? How can it be when one of the inputs explaining the discount is that the market is inefficient? This is tough stuff.
I’ve come to my conclusion that the market is inefficient on the basis of evidence that there do seem to be a category of investors and stocks that do outperform. In short, value investors and “value stocks” do seem to persistently outperform.
Let’s take the situation where past performance does predict future results. How is the past performance of the Marlin Global fund?
Let’s be conservative and say that the manager has no edge, and is throwing darts at a dart board and simply buying a less than optimally diversified random group of stocks that should, over time, approximate the market return. In this case, the past performance cannot explain the discount.
This is the idea that the expected portfolio turnover will be suboptimal (in most cases, too high) and fees will eat up the NAV. The first bullet point of Marlin’s “investment philosophy” is:
· The Company seeks to buy and hold shares in companies for the medium to long term.
There’s no evidence that the manager is buying and selling too often, and the manager’s philosophy is the antithesis of that. So portfolio turnover can’t explain the discount.
This is the hardest one to analyse. The discount is justified if the net present value of the fees expected over the life of the fund equal the discount. The Marlin Global Fund has a bit of a weird fee structure. Basically, the management fee is 1.25%, but reduced by 0.1 percentage points for every one percentage point of underperformance against a benchmark, to a minimum of 0.75%.
There is also a performance fee.
The performance fee is the 15% of the lesser of:
- The "excess return" of the fund.
- The difference between the NAV at the end of the year, and the highest NAV during the previous year.
That’s a pretty odd comparator.
One the one hand, it’s an absolute comparator, which has the advantage of only charging fees in years where shareholders get a positive return. On the other hand, the manager gets a performance fee for a rising stock market.
I’m sceptical the general increase in stock prices over the last century will continue at the same rate. Because of this, I’m more comfortable with an absolute fee – the manager will have to work for it.
15% is high though.
Keep in mind that the fee is the lesser of the two items above. This effectively means there’s a “high-water mark” that the NAV must reach before a fee can be charged. For the year ended 30 June 2012, the NAV high-water mark was $0.97. The current NAV is $0.8281. That’s a fair way to go before a fee is charged (+17% increase in NAV). If the discount was rational, it would be the net present value of the expected value of fees.
That’s the sort of maths that I can’t do.
One way to test whether it is rational is to look at the discount over time and see if it increases as the NAV increases (and approaches the high-water mark), and decreases as the NAV decreases (and falls away from the high-water mark. As the NAV increases, the expected value of the fees go up, and the discount should widen. However, the opposite occurs.
The graph shows that the discount was widest at market lows (March 2009), in direct contradiction to any rational expected value of fees calculation. The purpose of the graph is to show how the discount has reduced since the “distribution” policy. The “distribution” is really a return of capital. I think it naturally keeps the discount within a band. The other thing you’ll note from the graph is how volatile the discount is. And that brings me to my conclusion. After reviewing all these reasons, I see some support for a discount, but I can’t believe that the discount that regularly occurs is a product of rational pricing. Therefore I expect to purchase shares at a substantial discount the next time I see it.
As I mentioned a few times above, the distribution of 8% of the NAV is a form of a catalyst. If the manager can't keep capital growth up with the distributions, funds under management start to dwindle, the NAV drops, and there's more and more pressure to wind the fund up (and less incentive to keep it going, given that fees are a function of the funds under management).
The other option is that the manager stops the distribution policy and the discount widens. While this is possible, the manager has also instituted a small stock buy back. Given this, there seems evidence that the manager is likely to try other tactics to lower the discount, and there's investor pressure to do so.
I won’t purchase at anything smaller than a 20% discount. That may take some time, but I'm not in a hurry.