Introduction
French philosopher and mathematician Blaise Pascal once wrote:
“I have only made this letter longer because I have not had the time to make it shorter."
This quote came to mind when we planned this blog. Although it's possible to write a whole book on factor investing, this blog attempts to distil relevant information in a concise format.
Risk and Return
We could devote an entire blog post to risk and return (Noel's book covered this in detail, including spreadsheets), but below, we (hopefully) give enough of an overview to put the subsequent factor-related discussions in context.
In simple terms, by accepting more risk (in this context, risk is short-term volatility, which means the price of your investments moving up and down), you can expect a greater return over the longer term, but this is not guaranteed.
Below is a graph of risk and return, which shows what we expect to see with investments of varying risk levels over the longer term. Equities (shares), for example, are more volatile than bonds and, therefore, tend to generate greater returns.
Capital Asset Pricing Model (CAPM)
The relationship between risk and return was formalised in the 1960s as the CAPM. CAPM is a single-factor model, which means that only one factor of the stock impacts the formula's output: beta. Beta measures an investment's riskiness (volatility) relative to the market, and as seen above, an asset with a higher beta should have a higher expected return.
Fama-French three-factor model
Over time, it became apparent that the CAPM model didn’t accurately reflect the relative expected performance across all stocks. In their 1992 paper, Eugene Fama and Kenneth French summarised the limitations of the CAPM model and proposed a new model for equity pricing, the Fama-French three-factor model. They found that the share prices of smaller firms tended to outperform larger ones, and cheaper (value) companies tended to outperform more expensive (growth) companies.
The three factors in this model were:
Beta (covered above).
SMB: Small companies minus big companies.
HML: High (value) companies minus low (value) companies.
This model explains over 90% of the returns of a diversified portfolio, compared with the average 70% given by the CAPM.
The Fama-French model can be visualised using the style box below. Outperforming (according to their research) equities (small companies with value characteristics) live in the bottom left square.
Factor zoo
There are hundreds of potential factors - a factor zoo! Author Larry Swedroe believes that for a factor to be considered investable, it must have the following characteristics:
Persistent: It holds across long periods of time and different economic regimes.
Pervasive: It holds across countries, regions, sectors and even asset classes.
Robust: It holds for various definitions (for example, there is a value premium, whether it is measured by price-to-book [P/B], earnings, cash flow or sales).
Investable: It holds up not just on paper but also after considering actual implementation issues, such as trading costs.
Intuitive: There are logical risk-based or behavioural-based explanations for its premium and why it should continue to exist.
The most common factors are highlighted below:
Value: Discovered by Basu in 1977 - this is covered above in the Fama French model.
Size: Discovered by Banz in 1981 - this is also covered above.
Low volatility: The low volatility anomaly (sometimes known as betting against beta) identified that low volatility stocks often had higher returns than those with higher volatility, which contradicts the CAPM model examined above.
Momentum: The momentum factor identified that stocks whose prices had recently been rising continued to go up. This factor was incorporated into the Fama-French six-factor model analysis in 2018.
Profitability: Identified by Novy-Marx in 2013, this was incorporated into the Fama-French five-factor model. This factor identified that the most profitable firms generated higher returns.
Investment: Documented by Fama and French in 2006, this identifies the negative relationship between expected investment and expected return. This also features in the five-factor model.
I said I would attempt to keep this blog short and sweet. For those who want to read further, I would recommend:
AQR: Example paper: There is no size effect
Alpha Architect: Example blog: Long only value investing: Size doesn't matter
Example video: Again,discussing why Size doesn't matter
Excess Returns: Example podcast: Challenging the Foundation of Asset Pricing Theory with Andrew Chen and Alejandro Lopez-Lira
Rational Reminder: Example episode: RR #213 - Expected Returns and Factor Investing
Resolve Asset Management: Example episode: Value Investing is BACK - with Tobias Carlisle
As you can see, there is everlasting debate on factor investing, including why the historical factor data has changed over time.
Why should factor investing work?
There are two explanations for why factors exist:
Behavioural
Risk-based
Taking the value premium, Fama argues that the value factor is risk-based. In the same video, Richard Thaler argues that the value factor is behavioural, stating that value firms "look scary". A (very readable) 1994 paper by Lakonishok, Shleifer and Vishny gives a very plausible argument (backed by lots of data) for the behavioural explanation.
Realistically, there may never be a consensus. Fama reminisces about a dinner he had with Andrei Shleifer, during which Andrei's 1994 paper was discussed, and they agreed to disagree!
How have the factors performed in recent years?
There is no escaping that factor investing has faced a challenging period over the last decade. For example, in 2022, AQR identified that the value spreads were at .com bubble highs.
This was partly due to the value strategy crashing from 2018 to 2020. Fama-French wrote about this challenging period in a 2020 paper.
Asset manager Two Centuries found the 2020 value factor drawdown was one of the worst on record in the last 200 years.
How have live factor funds performed?
Kenneth French publishes historical factor data (the same data that was recently questioned) on the Dartmouth College website. There are three ways that live fund data may differ from that expected when looking at French's data:
Some factor funds are "long only." This differs from the Fama French dataset, which, using the size factor as an example, subtracts large company performance from small company performance. Fund manager AQR, which tends to offer long/short funds, argue that long-only offerings are likely to be dominated by market (beta) rather than factor returns.
The factor "loadings" in funds may differ. For example, a factor fund may contain 20% of holdings that fit in the lower left square, while another factor fund may have only 10%.
Live factor funds have costs, and while these tend to be lower than those of traditional active funds, they must still be considered.
Some providers may create factor indices that indicate how a historical factor fund may have performed. However, unlike with truly passive investing, where the index and the fund tend to typically only differ (known as tracking error) by the cost of the fund, factor fund index returns can vary dramatically from live factor fund returns for a couple of reasons:
Fees, as mentioned above.
The factor index may include several factors, even if some of the factors may not have been incorporated into the live fund at inception. This point was highlighted by U.S.-based investment adviser Allan Roth:
"Several years ago, I couldn’t attend an investing conference without the term “smart-beta” being mentioned every five minutes. Yet, I never heard once that this wasn’t a free lunch. Now there are smart-beta conferences where past performance is touted using back-tested data on new funds that didn’t actually exist over that period."
This can lead to differences in factor fund vs. factor index returns of 1-2% per year. For this reason, we consider live fund data the gold standard and will now focus on this.
We have around fifteen years of live factor (global) fund data from a U.K. perspective, and we can see that the factor fund has trailed the global passive benchmark by just over 1% a year (11.12 vs 12.22) while being more volatile (12.25 vs 11.19).
To evaluate factor performance over longer periods, we must look towards the U.S., where we can find global data from April 1998. We have created three portfolios:
Portfolio 1: A passive benchmark.
Portfolio 2: Light factor tilts.
Portfolio 3: More focused factor tilts.
All portfolios are globally diversified and contain 100% equities. Note that we believe 100% equity portfolios are probably not going to be the best option for most retirees, but this blog purely focuses on factor investing rather than optimising asset allocations.
If we analyse the period April 1998 - May 2024, we can see the factor fund portfolios have performed well against the passive benchmark, returning around £8.6m (portfolio two) and £10.3m (portfolio three) versus £6.2m. Admittedly, the factor funds were riskier when measured both in terms of volatility (Stdev) and max drawdown, but they exhibited superior Sharpe ratios (Sharpe ratio measures risk-adjusted returns).
If we look back over the last twenty years (May 2004-May 2024), the picture is less clear. The three portfolios have delivered broadly similar returns, but the factor funds' risk-adjusted returns (Sharpe ratio) are inferior.
Why do we use factor funds?
Given the relative performance of live factor funds versus a passive alternative, one could question why we use factor funds. We don't see convincing evidence to suggest that factor investing is a free lunch (generating persistent greater risk-adjusted returns than the overall market); the market is far too efficient for that. Therefore, we lean more towards Fama's view of factor investing (risk-based) than Thaler's (behavioural-based).
Furthermore, looking at the returns of the last 15 years (see the U.K. example above), some might also question whether factor investing (using live funds) generates greater raw (ignoring risk) returns than the overall market. No one can know for sure, but our guess would be that the last decade has certain similarities to the late 90s, where U.S. large-cap growth equities performed very well (until they didn't) and factor investors endured a torrid time. Outside of the U.S., factor investing has been more successful in recent years. For example, an emerging markets factor fund has outperformed the passive benchmark by around 0.8% a year over the last 20 years at the same level of risk.
Unfortunately, the U.S. makes up such a large percentage of global markets and has dragged down overall factor returns.
So why do we incorporate factor funds into our client portfolios? The answer is relatively straightforward - we care more about minimising downsides than generating the best possible returns. We can use real-world data to give an example.
Imagine our client retiring in January 2000, on the eve of the .com bust, and withdrawing an inflation-adjusted $50,000 a year. The client has a choice of the three portfolios described above.
Portfolio 1: A passive benchmark.
Portfolio 2: Light factor tilts.
Portfolio 3: More focused factor tilts.
Three years later, portfolio one has fallen over 50% to well under $500,000. Portfolio two has fared much better, falling less than 30% to just over $700,000, while portfolio three has fallen less than 10%.
If we fast forward to the end of 2018, the benchmark portfolio is exhausted. The factor fund portfolios have had far better outcomes, with both having final balances far greater than the initial balance.
We can see the benefits of factor investing to a lesser extent during 2022, when inflation was a growing threat, and the markets were impacted. The factor portfolios also suffered but to a lesser degree than the passive benchmark.
Our portfolios
We introduced the style box above, and below, we compare the equity component of Pyrford Financial Planning's portfolios compared to the FTSE Global All Cap, which, along with the MSCI ACWI, is widely considered the global equity benchmark.
The FTSE Global All Cap is concentrated more towards large-capitalisation (73%) and growth investment style (38%) equities. In contrast, our portfolios are more balanced across investment styles but still have a bias towards large capitalisation. While we could tilt more towards smaller capitalisation equities (think more Portfolio 3), we believe our current approach has an acceptable blend of keeping tracking error (versus the global equity benchmark) within reasonable bounds while providing sufficient protection when future large-cap growth bubbles burst. We fully accept inevitable long periods of relative underperformance versus the market. Not everyone seems to have the same patience - below are charts showing the outflows for AQR and Dimensional in recent years.
It's fair to say that short-term performance chasers are probably not a good fit for our investing approach.
Conclusion
So there you have it. The above was a roundabout way of saying that we don't use factor funds to try and generate returns in excess of the market (either in absolute or risk-adjusted terms) but instead to act as diversifiers to give our clients the best chance of a successful outcome in retirement.
Want to find out more?
Please contact us if you want to build a retirement plan that gives you peace of mind and the confidence to spend your money in retirement.
About us
The team at Pyrford Financial Planning are highly qualified Independent Financial Advisers based in Weybridge, Surrey. We specialise in retirement planning and provide financial advice on pensions, investments, and inheritance tax.
Our office telephone number is 01932 645150.
Our office address is No 5, The Heights, Weybridge KT13 0NY.
Please note: This blog is for general information only and does not constitute advice. The information is aimed at retail clients only.
Although best efforts are made to ensure all information is accurate, you should not rely on this blog for your personal situation or planning.
The value of your investment can go down as well as up and you may not get back the full amount you invested. Past performance is not a reliable indicator of future performance.