This is the fourth post in a mini-series on how factor returns from academia can be different to those from smart beta investing. For the previous post on how trading costs can impact factor returns, click here.
I recently read a fascinating story featuring a blind doctor. It’s not often you hear about blind doctors, so this one caught my attention.
The doctor had two adopted children, one of whom was married to a broker. The doctor was living on the savings that he’d accumulated from a long and difficult career, having moved from a career in pharmacy to medicine. The doctor placed his life’s savings with his son-in-law, the broker, thinking him to be a trustworthy steward of his capital. The broker, incentivised by the commission structure of placing trades for client accounts, proceeded to churn his father-in-law’s account so heavily that the doctor eventually ended up completely broke.
The story comes from a recent Wall Street Journal interview with Charlie Munger, Warren Buffett’s business partner, who often provides unique insights into human behaviour – particularly in the financial system. The full interview is fantastic, and I highly recommend reading the interview in its entirety. One particular quote from the article that drew my attention was:
“What a salesman will do is just awful, if you give a man a family to support, in the commission system.”
How true that is. Once a strong financial incentive is coupled with the existence of a family to support, any “ethically grey” but financially enticing behaviour can be easily rationalised away by telling yourself that it’s “just to pay my daughter’s school fees” or “just so my family can live in a nicer home”. The blame is shifted from the person in question to his family, making the behaviour not only easier, but justified, in his eyes.
There are countless examples in the financial world of human behaviour being the cause of terrible decision making, which usually end with average people losing a lot of money. Human behaviour is just a prevalent in the factor world as it is elsewhere, and influences the construction, AUM, and investor behaviour when looking at smart beta funds. We can’t forget that these smart beta funds are being both run and invested in by human beings, who are subject to the same behavioural biases and flaws as everyone else.
I’ll focus on three aspects of behaviour in particular for this post – career risk, incentives, and performance chasing.
We’ve already seen that not all multifactor smart beta ETFs are created equal. In this post on factor dilution, we saw that ETF providers can err on the side of lower factor exposure in their smart beta funds, meaning lower tracking error versus the benchmark, to give a lower chance of clients abandoning their funds. Everybody wants to remain employed, and from a career-risk perspective, it’s easier to sell clients on the benefits of factor investing and provide index-like returns, than it is to provide them with actual factor returns.
The graph below shows the performance of the MSCI USA multifactor index (used by UBS’ ‘MSCI USA Select Factor Mix’ ETF), the Goldman Sachs multifactor index (used for GSLC), and the JP Morgan multifactor index (used for JP Morgan’s ‘Diversified Return US Equity’ ETF):
Investors into these index tracking funds can be equally to blame. Investment managers and advisers are often tempted to buy these index-hugging smart beta products because it gives the appearance of remaining “actively” invested. Many investment firms use this as a selling-point to justify their active management approach, and ultimately justify their fees. It’s their belief that nobody would invest with a manager who just invests in passives, as clients believe they can do that themselves, so they must restrict their investments to predominantly active funds. These index-tracking products allow the manager to espouse the benefits of smart beta investing to clients (which enables them to charge a higher fee and keep them employed), whilst generating the performance of the S&P 500 – a notoriously difficult benchmark for active managers to beat.
So both ETF providers and their clients (those of us in the investment industry) have career-risk related incentives to producing/buying market-tracking smart beta funds. ETF providers don’t want investors redeeming due to tracking error, and investment managers need to give the appearance of remaining active investors.
Combining these issues means that there are millions of pounds worth of AUM being invested in multifactor funds whose performances will provide nothing like the factor returns advertised in academia.
One of my favourite quotes in investing, which applies to much of life outside of investing too, comes again from Mr Munger. He said,
“Show me the incentive, and I’ll show you the outcome.”
It’s both an incredibly powerful, and incredibly true statement. Incentives are an amazingly powerful force when it comes to shaping people’s behaviour, and have been the cause of almost every financial scandal in history. Financial incentives, especially when in the form of a commission system, have the power to warp peoples’ belief systems, and cause them to behave in ways that are completely contradictory to their normal behaviour – especially when a family is involved.
Whilst incentives play a more minor and less consequential role in smart beta investing, their effects can still explain the construction of almost every smart beta fund.
Research by Verdad Capital has shown that almost no value funds actually use the academic version of value, despite the plethora of evidence confirming its efficacy. Why? Because it caps the fund manager’s remuneration.
Below is a chart from Verdad’s research, showing US stock market into deciles based on price-to-book ratio. There is a strong linear relationship between valuation, market capitalization, and traded volume. The cheapest two deciles of the stock market have median market capitalizations of less than $400m and median daily trading volume less than $1.5m (in contrast, the most expensive two deciles have median market capitalization of over $1.9B and average daily volume of over $15M).
Source: Verdad Capital
The chart shows that the cheapest stocks are disproportionately small in terms of size and volume. This means that an active manager looking to construct a value portfolio in the cheapest 2 deciles of value stocks (similar to the academic definitions of value) would be unable to manage more than $200m or so. The cheapest two deciles of the market are almost entirely composed of micro-cap stocks that are hard for any fund with over $200m in assets under management to trade.
This would mean that in order for a fund to effectively capture academic value exposure, they would have to manage less than $200m. This is a pretty unlikely decision for a fund manager to make, as it effectively puts a cap on the revenue generated by the fund, and therefore sets a cap on the fund manager’s remuneration.
When faced with a decision to effectively capture the academic definition of value, or to make more money, it’s unsurprising that most fund managers choose to opt for the latter.
A further reason why factor funds underperform their academic factors is down to investor behaviour. A recent study by the University of Rotterdam found that factor funds underperform their academic factors by about 3% per year. And investors, because they buy and sell these funds, underperform a buy-and-hold factor fund by about an additional 1%.
The table below shows the difference in returns between a buy-and-hold approach and what investors actually received.
Investors are losing, on average, 1.7% per year from attempting to time factor funds, and those investing in the size and momentum factors are losing over 2.5% per year. All of this is caused by nothing more than investor behaviour.
The authors conclude:
“Although factor funds have attracted significant fund flows over our sample period, it appears that investor fund flows have been driven by factor funds earning high past returns and not by the funds providing factor exposures. We argue that rather than timing factors and factor managers, investors would be better off by using a buy-and-hold strategy and selecting a multi-factor manager.”
Throughout this series, we’ve seen a number of different reasons why smart beta might not provide the same returns as found in academia. Most have been due to the nature of factors themselves, and would likely still occur if those who created funds to capture them, and those who invested in them, were completely rational investing robots.
But behaviour always plays a role in investing, and, as ever, our behavioural biases are our undoing when it comes to smart beta. The desire for fund providers and asset allocators to remain employed, the desire for fund managers to maximise their remuneration, and the performance chasing tendencies of investors all combine to ensure that the odds are stacked against us when it comes to factor investing.
If we do decide to invest in smart beta, we need to take care in ensuring that the funds we’re choosing aren’t mere index-trackers, that the size of the fund won’t meaningfully reduce our factor exposure, and that we’re able to stick with performance through the inevitable periods of underperformance. It’s by no means an easy feat, but then again, nobody ever said outperforming the market was easy.
The next post in the series looks at the results of what we’ve looked at so far in this ‘Academia vs real life’ section. We’ve seen many reasons why academic factor returns should be different to real life smart beta returns, and so the next post looks at some examples of how these differences have fed through into the real world.