With high-fee active managers and hedge funds in decline, and technology lowering much of the cost of day-to-day investment management, how can a fund manager justify a higher-margin product? Judging by the past, such a product would likely play on consumers’ desire to beat the market, without trading off liquidity, concentration, or opacity risks.
Enter ‘smart’ beta funds, the newest in a long line of investment funds offering the possibility (not the guarantee) of higher risk-adjusted returns compared to the market. The investment decisions of these new funds aren’t based on gut feel and the networking ability of a human advisor; they use quantitatively designed algorithms. They are often marketed based on hypothetical historical data regarding various factor risk premia. They cost up to 600% of similar normal beta ETFs. And they are growing quickly.
But, are they good for investors?
So far, the answer is no.
Market Capitalization: Still the Anchor
The starting point for any allocation model is a market capitalization-weighted portfolio. By anchoring to market capitalizations, you free ride on the collective wisdom of millions of investors and traders globally. You also know how much you are diverging from market allocation; that divergence is the foundation for generating outperformance.
The same logic can also inform portfolio construction for clients who want to take on more or less risk. While the market holding of U.S. small caps, for instance, reflects the risk that investors on average are willing to bear, individual investors may be comfortable with more or less risk in their portfolio. It is up to the individual advisor to determine how to take on more or less risk while still maintaining a diversified portfolio.
The cleanest way to do this is to use the global market as a benchmark for determining asset allocation. Allocations should consider the market value of available assets; the implied expected returns from those assets should guide a proportional allocation to the different markets. This was the insight behind the pioneering work of Fischer Black and Bob Litterman in creating the Black-Litterman model for creating diversified portfolios at every risk level.
Factors Are Explainers of Variance, Not Predictors of Value
Researchers seeking to explain fund performance have identified up to 300 factors that can independently explain historical investment risks of a portfolio. In effect, each of these is a dimension that an individual investment is assessed on, like price-to-earnings ratio. A fund is then assessed in terms of its exposure to these factors, and its performance deviation from a passively managed equal factor portfolio.
While factors help explain why a given fund performed a certain way, factors are not necessarily good predictors of future performance. Nor are they necessarily a compensated risk premium, a key difference discussed in a recent Vanguard white paper. There are many risks you could take on that don’t have a positive expected risk premia; lotteries, casinos, and currency risk are good examples. Just because they explain variance doesn’t mean they’re attractive.
So it’s important to understand that a market cap portfolio is a factor portfolio. It just takes on the market allocations of each factor. Investing in ‘smart’ beta portfolios means the manager overweights some factors and underweights others. Some of these may be static overweights, while others may be dynamic depending on the market cycle.
However, while the jury is still out on whether actual implementations successfully produce alpha, most evidence implies they’re don’t.
A real investment fund is different from the theoretical ideal of an index. For investments where the precision of managing to the index matters, and the expected alpha of the fund is small, implementation costs can swamp alpha.
Proponents of ‘smart’ beta claim they are simply applying the teachings of financial economists who have found return predictors other than market beta. While we don’t quibble with their findings, it’s another issue entirely whether those predictors continue to hold in such a way that they can be used to generate outperformance in a fund after fees, taxes, and implementation.
Issue 1: Is ‘smart’ beta just an expensive way of getting high beta?
Consider the below graph that compares a long-running ‘smart’ beta fund RSP, which equal weights stocks in the S&P 500 versus an S&P 500 index tracker.
Higher Risk, Higher Return
Is it fair to say RSP has outperformed? Its cumulative returns are definitely higher. But it’s declines are also steeper. By equal weighting stocks, the fund has over-weighted (in a very simple manner) to smaller-cap stocks relative to market cap. This is a often a simple way of taking on more risk, with a similar outcome to using leverage.
Once You Control for Risk….
We’ll take two of the ‘smart’ beta ETFs with the longest track records: RSP and PRF, a PowerShares ETF that follows the RAFI U.S. 1000 index. In order to assess performance of these actual investments, we’ll perform the simplest risk-adjusted test possible—testing for risk-adjusted outperformance after accounting for the risk exposure from a market-weighted investment.
If the ‘smart’ beta ETFs are successful at delivering better risk-adjusted returns, they will have a positive alpha coefficient in a regression. If they are just more (or less) volatile than the market cap benchmark, they’ll have a beta greater than (or less than) 1.
The results depicted below show zero improvement in volatility-adjusted returns, but a beta coefficient (risk-taking) of greater than 1.
Does RSP Add Alpha?
Does PRF Add Alpha?
Taking on more risk, on average, leads to higher returns. This is hardly outperformance. A consistent finding with ‘smart’ beta ETFs is that they take on more risk, not just different kinds of risk. Their volatilities tend to be greater than their market cap benchmarks, which must be controlled for when assessing performance. If the consumer could have achieved a similar result without a ‘smart’ beta fund by simply increasing risk (and saving higher costs), that would have been a preferable strategy.
But two examples might not convince you. Instead, take research conducted by Denys Glushkov, Research Director at Wharton Research Data Services (WRDS), covering 164 ‘smart’ beta ETFs from 2003 to 2014. According to Denys:
“I find no evidence that SB ETFs significantly outperform their risk-adjusted passive benchmarks. Positive returns from intended factor bets are offset by negative returns from unintended factor bets resulting in an overall performance wash.”
Risk-adjusted performance of SB funds is also insignificant when compared with the performance of the blended benchmark that provides passive cap-weighted exposure to market, size, and value factors. After decomposing benchmark-adjusted performance of SB funds into selection, static, and dynamic allocation effects, I find that their factor timing ability is neutral at best.”
Issue 2: Higher (and Hidden) Costs to the Consumer
While the higher expense ratios of these active index funds is clear, there are other hidden costs associated with investing in these funds. The first two derive from the higher transaction volume caused by a non-market cap index.
Non-market cap based indices have substantially higher transaction volumes because, to generate a chance of outperformance, they need to periodically re-deviate from their previous holdings, both by selling out of some assets and buying into others. This exposes you to transaction-based costs.
Liquidity Transaction Costs
Every market transaction exposes the fund (and its customers) to transaction costs such as the bid-ask spread. The more active a fund is, the more often it’s exposed to this type of transaction cost.
Market cap index funds have built-in, transaction-minimizing features. As prices change, the funds’ holdings automatically match the market cap weighted allocations. The turnover required in a market cap weighted portfolio derives only from index inclusions and corporate actions, which are both less frequent and less expensive.
Many active index funds have higher turnover due to the need to track the index in a way that often moves against market cap weighting.
Tax Transaction Costs
These transactions (hopefully) will generate taxes in taxable accounts. Like other 40-Act funds, ETFs pass through realized tax gains to the end investors. As a result, while the pre-tax returns may beat a passive index, the after-tax returns may be significantly less than a more passively managed strategy in a taxable account. In the case that the returns do not keep up with the passive strategy, the turnover can continue to reduce the end investor’s returns.
Tax Drag – Reduction in Returns Due to Taxes
|U.S. Large Cap||U.S. Total Market|
Source: Morningstar.com, using same funds as above.
A Negative Risk Premium?
The core principle of all investing is earning a risk premium. If you are bearing uncertainty, you should be able to demand to be paid more, especially net of fees and taxes. ‘Smart’ beta funds appear to be based on the opposite notion: that the investor should bear known higher costs—management fees, transaction costs, and tax costs—in exchange for the very uncertain chance that active indices will outperform on average and over a long period of time.
And it’s far from certain that documented historical performance by ‘smart’ beta factors will persist. A recent paper by McLean and Pontiff in the Journal of Finance reviewed the persistence of such premiums out-of-sample (not based on backtests) and after publication of the premiums. They find that portfolio returns are nearly 30% lower out-of-sample, and 60% lower post-publication. That’s market efficiency reducing the effectiveness of known risk premia.
So, as it was before the growth of ‘smart’ beta funds, market cap weighted indices are probably still the worst way to invest, except for all those others.
This article was first published on ETF.com.
More from Betterment: