First of all, I absolutely love this blog and particularly enjoy reading about your perspective on the looming "bond bubble" and why investors should not fear it. I am curious if you have any thoughts on the growing popularity of "low volatility" funds such as SPLV. I am intrigued by the idea of fund that would potentially slightly underperform the S&P 500 in good years while outperforming in the down years. I think many investors are still scarred by the 2008-2009 crash and would like to take a more defensive approach when it comes to investing. What if someone were to replace a portion of their holdings in VTSAX with SPLV. Do you think there is any value in this approach if they wanted to minimize the down years? Thanks
My Reply
Thank you for your kind words about my blog -- and my apologies for not posting much lately. Low volatility funds like ETFs SPLV and USMV are a very interesting development. First a couple of words about the composition of these funds. Although they may resemble large-cap value funds (most notably, by having higher-than-average dividend yield) they are instead composed of all sorts of large-cap stocks that have exhibited low volatility in the recent past. Many of these stocks happen to be value stocks, but as we'll see shortly the entirety of these funds is in no way value-oriented (at least not by traditional price/book measure of value).
SPLV uses Standard & Poors low volatility index and USMV uses MSCI minimum volatility index. The details of how the indices are constructed are involved. MSCI methodology calls for all industry sectors to be represented in roughly similar proportions as in the entire stock market. S&P methodology does not have such strict requirements resulting in a much less diversified set of stocks that are very heavy on utilities and consumer staples (aka "defensive" consumer stocks). There are a couple of other strikes against low volatility funds. They have (slightly) higher management costs and are also likely result in higher turnover and associated hidden frictional costs.
Let's first confirm that these low volatility funds do in fact provide stock-like returns at lower volatility:
Both SPLV and USMV have short histories, but their story checks out for the year and a half or so that they've been around. The blue line on the chart above is Vanguard's Total Stock Market fund (VTSMX) and the orange and green lines are SPLV and USMV respectively (with all dividends re-invested, which is the only way to compare investments, of course). SPLV and USMV have in fact provided returns roughly on par with the entirety of the stock market but at significantly lower volatility. So far so good.
The question then becomes whether it's realistic to expect low volatility funds to continue to provide market-level returns at much lower volatility? As with any question about the future there is no way to be absolutely sure, but looking at valuations I am strongly inclined to say that no, it's not realistic.
Before we jump to the numbers note that it is very important for an apples-to-apples comparison to use valuations from a single authority (e.g. Morningstar) when comparing different fund families funds because one company's valuations methodology can vary wildly from another. For example, iShares.com shows its S&P 500 fund, IVV, to have P/E of nearly 20 and P/B of 4.0 while Vanguard shows its S&P 500 fund, VOO, to have P/E of about 16 and P/B of 2.2. Since the two funds are all but identical, there is a clear disconnect in the methodologies the two companies use to come up with those numbers. But when we compare the same two funds on Morningstar, we find P/Es of 14.0 and 13.3 for IVV and VOO respectively and P/Bs of 2.0 for both -- all but identical, just as we would expect. We don't have to dig into the details of which one of these widely varying sets of numbers is the right one (and we probably wouldn't have enough information to decide anyway). But we do want to make sure we compare apples to apples, which is why we use Morningstar's figures and not iShares' or Vanguard's.
With that in mind, let's look at these funds' valuations using numbers found at Morningstar.com:
The above chart compares the valuations of SPLV, USMV with more traditional value and growth halves and the entirety of S&P 500 stocks (represented by three iShares ETFs). The top half shows actual numbers for 3 value-oriented and 3 growth-oriented measures. The bottom half ranks the 5 ETFs on those 6 measures and finds average ranking for each ETFs. The fund whose column is colored blue is most desirable -- it is either cheapest or fastest-growing. Then come green, yellow, orange, and finally red, representing the most expensive or slowest-growing fund.
What we find is that our low volatility funds are less value-y than value stocks and less growth-y than growth stocks. In other words low volaility funds are relatively expensive and relatively slow-growing compared to the entirety of the stock market. That's the price you pay for that low volatility.
Although SPLV and USMV seem likely to retain their low volatility going forward (though at least in case of SPLV I would be worried that the lack of diversification could eventually backfire when the specific industries -- especially utilities -- over-represented in that fund are hit unusually hard), it is in my opinion not realistic to expect them to match the returns of the broader stock market which is both priced cheaper and has historically grown faster.
That does not necessarily mean that these funds are obvious duds. Low volatility is, of course, highly desirable. Just understand that you are paying a price for that low volatility. If you do decide to invest in low volatility funds, based on what I saw preparing this post I would definitely go with USMV over SPLV. USMV has much better diversification, lower expense ratio, and its holdings appear to be less overpriced on valuations relative to the broader market.
"What we find is that our low volatility funds are less value-y than value stocks and less growth-y than growth stocks. In other words low volaility funds are relatively expensive and relatively slow-growing compared to the entirety of the stock market. That's the price you pay for that low volatility."
ReplyDeletePlease could you expand on that? It occurs to me that one interpretation is that low volatility funds actually 'capture the best' of value and growth investing rather than the worst as you suggest. There is at least a theoretical argument to be made that the extreme ends of the value and growth sector may be more volatile and produce lower returns through mean-reversion effects if nothing else. I guess there may be research and data to test this. Your conclusion just didn't seem certain to me, though I may have misunderstood.
On the main substance my inclinations based on only a superficial understanding still are very close to your analysis.
I thought you might be interested in this review below. I haven't read it in depth yet and I am still very much getting to grips with low volatility.
http://us.spindices.com/documents/research/low-volatility-effect-comprehensive-look-201208.pdf
A few early thoughts though:
Even if low volatility ETFs do deliver higher 'undliuted’ excess (to cap-weighted index) returns going forward it seems unlikely that there will be any excess returns after costs and taxes.
I can't yet work out how much back-testing effects will diminish forward excess returns.
Market efficiency (even if not perfect) will likely erode future excess returns
Volatlity may not be relevant for a self-directed private investor who doesn't have to answer to clients or the marketing department of an investment fund company. As a long term steady investor in the accumulation phase modest levels of volatility don't bother me. Volatility may even improve returns via dollar-cost averaging and rebalancing 'boosts'.
I am concerned about big bear markets and the occasional inevitable 30-50% fall in equities. This obviously is why I hold bonds and some cash! I think there is a possible danger that investors will overestimate the added security of low volatility ETFs and reduce fixed income and thereby inadvertently ADD risk to their portfolios.
Nearly all the problems in the investment industry arise from a non-alignment of interests between the professionals and their clients. I strongly suspect that low volatility ETFs are an example of this phenomenon. For the professionals they may produce flattering short-term results, they certainly increase complexity and they increase charges. All fantastic for the industry. But for private investors these attributes of low-volatility ETFs are either irrelevant (short-term results) or are harmful.
Interested as ever in your thoughts though understand how many comments you have to consider.
As I mentioned I am new to low volatility and need to understand more before I am 100% confident in the above comments
I didn't mean to suggest that low volatility funds capture the worst of stocks -- just that they are valued more expensively than average. That pricing is quite logical since it buys you low volatility. But it also means that it is not rational to expect their returns to match broader market returns going forward.
DeleteAs to backtested results, I see no reason not to trust the MSCI's and S&P's numbers. But equally, I see little to no reason to take those numbers as indicative of expected returns going forward. Partly because of omitted frictional costs -- the favorite digression of all backtesters. Partly because low volatility backtesting I've seen is a very shallow one or two decades -- of which 10-15 years were at bubble valuations with necessarily low returns. Partly because the markets will become more efficient over time -- and certainly for strategies that are out in the public such as low volatility investing. And partly simply because any and all backtesting will result in successful "strategies" if the backtester is sufficiently motivated -- and MSCI and S&P have plenty of motivation to come with shiny new indices to license out.
Again, I am not saying that low volatility stocks are irrationally priced or that investing in a low volatility fund (provided the expenses are low and the fund is reasonable diversified -- and both are true for USMV) is a bad idea. I am saying that expecting their future returns to match or beat those of the broad market while providing lower volatility (as the backtested research might lead you to believe) is not realistic in my opinion.
One other idea I had was to see if USMV's holding might by chance match those of Berkshire Hathaway. Reason being is that Warren Buffett is on record for preferring the steady growth model as opposed to investing in either severely distressed value businesses or pie-in-the-sky-expectations growth ones.
DeleteHere are top 25 holdings of USMV: http://portfolios.morningstar.com/fund/holdings?t=USMV®ion=USA&culture=en-us
Here are top 40 holdings of BRK: http://www.dataroma.com/m/holdings.php?m=brk
Of top 25 USMV holdings, only JNJ, PG, IBM, and WMT are also found in top 40 Berkshire Hathaway holdings. That is within what would be expected by pure chance, assuming the entire population of both BRK and USMV holdings is composed of, say, 500 largest stocks.
So there is little reason to believe that low volatility stocks (at least as defined by MSCI's index) are what Warren Buffett's likes to invest in.
Would the analysis or conclusion be any different for emerging markets (say EEMV or DEM vs VWO)? I know that DEM is not directly marketed as low volatility but the higher dividend focus is often touted as the reason behind it's lower beta.
ReplyDeleteI think that any type of sector or valuation strategy with emerging markets is a complete crapshoot. The data there -- both historical and even current -- is very sparse and unreliable. The oldest emerging markets index is only about 25 years old and the countries thought of as emerging have undergone dramatic shifts, both politically and economically, in that time period.
DeleteBut yes, looking at Morningstar portfolio data for EEMV and EEM, EEM's valuations are lower across the board (and growth measures show what appear to be nonsensical numbers, only underscoring the low quality of emerging markets data).
Putting specific markets aside, the most fundamental law of investing is that risk and reward go hand in hand. It would be an inefficiency of the highest order if less risky, low volatility stocks consistently delivered returns on par with those of more risky, high volatility ones. It is nearly the equivalent of bonds beating stocks in returns, as well as in safety. It certainly can (and did, and will) happen from time to time but it is not the sort of thing one can rationally expect to continue indefinitely into the future.
Again, this doesn't mean that following a low-volatility-stocks-only strategy is irrational. What is irrational is to expect such a strategy to match or beat a total market one.
Hi LTR
ReplyDeleteHope I'm not spreading bogus info but my understanding is that low vol theory claims that the risk/return relationship is upside down, at least for certain strata of the market, making it quite a significant anomaly. It's not just watering down equity and forgoing upside, but a discovery that low vol stocks outperform. Check out falkenblog for more info
I understand the claims being made in relation to this strategy. What I am saying is that inferring significant ongoing alpha from it (after all expenses) is wishful thinking of fairly high order -- even if the backtested numbers are correct and low volatility strategy really did deliver 3% or 5% or 10% alphas in the past.
DeleteI skimmed through http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2210003 and I just don't understand how a paper can in all seriousness imply (or cause the reader to infer) alpha of 6-9% (at lower risk, no less!) and then go on to say something along the lines of too many people are too stuck in their ways to take advantage of this golden opportunity. How about a simpler explanation of (a) the backtested returns bear no resemblance to future ones and/or (b) people will arbitrage this into oblivion now that it's public knowledge.
There is also no mention of expenses and no mention of changes in relative valuations (which, if exist and responsible for alpha, cannot continue forever). To their credit they do backtest the US market starting from late 1960s and not just the shallow 20-year backtesting I've seen elsewhere (to appreciate how meaningless the 20-year backtesting is consider that long treasuries were the best performing asset class in that time span).
In any case the only way this debate will be settled is through actual market returns over the coming years and decades. I would love to be proven wrong here since I could then realize my mistake and take full advantage of high returns at low volatility. But I am not holding my breath.
If backtested returns bear no resemblance to future ones then what is 'expected return'? Where do FF risk factors come from? Why do you invest in stocks at all? There's no way we'd invest in the stock market if it had a history of depreciation.
ReplyDeleteI meant that as applying to this specific strategy not stocks as a whole. In general, the more specific/complex/sensitive-to-various-conditions a strategy is, the less meaningful backtested results are. And you are right -- I am skeptical of FF risk factors for the same reason (and, yes, I know I'm in a minority there; I don't try to talk out anybody out of titling to SV if they so choose)
DeleteAlso, we don't need to have any backtested data to invest in stocks. Stocks can be expected to have return equal to the sum of:
1. "dividend" percentage (in the wider meaning of the term "dividend" -- buybacks are theoretically dividends too)
2. dividend growth (which is not equal to but trends in the same directions as GDP growth, which in turns breaks into population growth + productivity growth, both of which can be expected to be positive or at least non-negative)
3. change in valuations (which should zero out over long enough time)
The above is true regardless of past stock returns. In fact it explains past stock returns (again, over sufficiently long periods of time that valuation changes become irrelevant).