Episode 98: Rapid Fire Listener Questions, Wealthsimple's Victory Lap, and the Historic State of Value Investing

May 14, 2020 Benjamin Felix

We spend the bulk of today’s episode considering whether Wealthsimple’s use of long bonds and low volatility stocks is really protecting their clients’ downside, and summing up recent arguments by Cliff Asness and AQR levelled against critiques on value investing. Before that, we kick things off with thoughts on why Elon Musk aims to have no possessions, before looking at the links between empathy and the theory of relativity as well as some productivity secrets in recent books by Charles Duhigg and Shane Parrish. Next up, we briefly address a bunch of listener questions on factor tilting, and ETFs concerning COVID-19, the Smith Maneuver, and more! A final listener question about Wealthsimple’s claim mentioned above leads our hosts to wonder whether volatility and drawdown are good measures of risk. Ben made a few models to help answer this question which tested consumption models as another possible measure and brings up an interesting point about the significance of considering long bonds from an expected return or a risk parity perspective. From there, we move to the investment topic of the week – the historic state of value investing. This is a contentious topic with recent papers by Cliff Asness and AQR both weighing in and you’ll hear Ben and Cameron distill the main points from both. We hear about medium-term odds being on the side of value, and some great arguments showing common critiques levelled at value investing to be premature. Finally, Cameron takes us through the psychometric profiling side of measuring risk tolerance before telling listeners why they shouldn't make investment decisions based on reckless critiques. Tune in to get it all!

See this content in the original post

Key Points From This Episode:

A reminder to comment on the new comments section on the RRP website. [0:00:44.2]
Why Elon Musk ways he intends throw away his possessions. [0:04:36.1]
New books about productivity and the links between science and empathy. [0:07:08.2]
Factor tilting: being aggressive versus non-aggressive. [0:12:43.6]
Is there a benefit in capturing size premium using a combination of ETFs? [0:16:54.2]
How to adjust RESP asset allocation as kids get closer to school age. [0:18:46.2]
What ETFs are best to use while implementing the Smith Maneuver. [0:22:36.2]
Has the role of bonds ETFs changed in light of COVID-19? [0:24:12.2]
Thoughts on Wealthsimple’s claim to have protected their clients in this downturn. [0:28:34.2]
Critiquing long term bonds: is volatility/drawdown a good measure of risk? [0:33:28.2]
Ben’s model testing consumption objectives as a measure of risk. [0:36:28.2]
Portfolio topic of the week: the historic state of value investing. [0:42:22.2]
Considering Cliff Asness’s paper about whether value investing is dead. [0:46:05.2]
Considering AQR’s paper addressing critiques levelled at value investing. [0:54:04.2]
Planning topic: the psychometric approach to measuring risk tolerance. [1:05:50.2]

Rapid Fire Listener Questions

In your paper, factor investing with ETFs, you outline a sample portfolio that includes a 22% equity tilt towards US large value and US small value ETFs. Would you consider this a more conservative or aggressive value tilt, and what would be your upper limit for a long term investor?
- First thing to note is that we are revamping the model portfolios.
- I’d call the model portfolio in that paper conservative in its factor tilts.
- An aggressive value tilt is what Alpha Architect is doing - that is a concentrated portfolio of the cheapest stocks in the market.
- Basically to get more aggressive than that you are looking at leverage.
- IUSV and IJS are diversified portfolios of cheap stocks. I’d say they’re pretty conservative as far as factor tilted products go.
- Even if you went 100% IJS you’d be pretty diversified with 488 small cheap stocks. QVAL (Alpha Architect U.S. Quantitative Value ETF) has 41 holdings.
- In either case you’re missing many of the stocks in the market which could lead to big alphas, positive or negative, but probably negative if markets are efficient.
- The level of tilt is highly subjective, like the stock/bond decision.
- It is important to note that adding in factor exposure is not always going to increase risk in absolute terms.
- A small value tilted portfolio might be less risky over the long-term due to risk factor diversification.
- The upper limit is your willingness to endure tracking error. YTD US small value is tracking about -20% relative to the US market.
Is there any noticeable benefit in trying to capture some size premium in Canada by using a combination of ETFs such as a mixture of XIU or XIC with XMD or XCS?
- XIU is short small cap, so that’s not the right start point
- XMD and XCS are both small cap universe including small growth low profitability which hurts the risk-adjusted returns a lot.
- Cliff Asness in episode 93 made a really good point that small cap universe has a higher beta than the market, so if you want more beta you can add small cap universe instead of using leverage to increase expected returns.
- In that case you are not getting an independent risk premium - just more beta.
- Based on that I would only add XMD or XCS to XIC if you wanted more market exposure, meaning you should already be at 100% equity and for whatever reason don’t want to use leverage.
How do you adjust RESP asset allocation as kids get closer to school age? And what do you do with leftover money in RESP (hopefully)?
- Many people consider their RESP savings to be sacred. They are in it for the grant, so are happy with lower expected returns.
- Cameron started his kids’ RESP as all equity for the first 8 years, then 60/40 until they were 14 or so, then flipped to 40/60 from there.
- It depends on whether or not you need the money to fund education costs.
- If you can pay for school out of cash flow you might leave everything in the RESP until the later years of the child’s education.
- In that case the bigger consideration might be tax planning rather than asset allocation. IS the child working?
- Some also use for TFSA contributions
- If you are paying for school out of cash flow you can keep the RESP invested and then make a tax efficient withdrawal at the end of the child’s education, but keep everything invested. You only have to pay back the grants.
- Lots of people get more conservative leading up to education. Some people go to a GIC ladder for the last few years.
What ETFs would be best to use while implementing the Smith manoeuvre? Should it be treated just like any other non registered account? A vast majority of people online are saying that investing in dividend paying individual stocks is the best way to go. What would be your take on this?
- Cameron’s bias is against leverage. He would prefer to build a plan that does not require leverage. To use a term that Wade Pfau used a few episodes ago, aim for safety first planning.
- People say to use dividend stocks because it allows you to take your tax efficient dividends earned in the leveraged non-registered account and pay down more of the non-deductible mortgage without affecting the deductibility of the HELOC.
- This all sounds pretty good. You get the full interest deduction on the loan, earn only eligible Canadian dividends in your taxable account, and make your non-deductible mortgage deductible faster.
- The downside is that you own a concentrated portfolio of Canadian stocks.
- I’d stick with whatever portfolio you have deemed to be optimal.
Has the role of bonds (bond ETFs) changed in light of covid 19 and the resulting downstream effect on bonds (govt or corporate)?
- The big concern with bonds right now is what governments are doing with fiscal policy.
- We are seeing tons of stimulus coming from government deficits.
- This works as long as there is demand for bonds, but if the supply of safe government debt grows too large bond prices could fall.
- For now there is still a ton of demand for bonds as shown by the low yields.
- Japan’s massive fiscal stimulus has not caused problems for them yet.
- Their government debt to GDP is above 200%.
- The US is a little over 100%, Canada is at about 50%.
- Since Japan started their aggressive fiscal policy in the early 1990s their debt has steadily grown.
- Japanese government bond returns hedged to USD since 1990 have been 5.83% while US government bond returns were 5.77% and global government bonds were 6.02% all annualized.

Data source: Dimensional Returns Web

*Annualized number is presented as an approximation by multiplying the monthly number by the square root of the number of periods in a year. Please note that the number computed from annual data may differ materially from this estimate.

As long as there is demand for debt backed by developed economies like Japan and the US, I’d lump Canada in there too, there shouldn’t be an issue.
In this scenario we would expect yields to remain low, but the bond allocation in a portfolio should never really be about the yield anyway.
BMO’s ZAG for example is up 5% YTD which helps to dampen the volatility of stocks.

What do you think about Wealthsimple’s claim that their well-designed portfolios protected their clients in this downturn?
- We talked about Wealthsimple’s portfolio update in an earlier episode, basically saying that we didn’t agree with some of the main philosophical and structural decisions.
- They posted an article at the end of April which was basically a victory lap follow up on their portfolio changes.
- They explain in the article that their addition of long-term bonds and low vol stocks allowed their portfolios to withstand the current situation without too much of a decline.
- Low vol stocks, looking at ACWV vs. ACWI from iShares have dropped about 3% less than the market.
- The big difference here came from their exposure to long-term government bonds. 15-20% of all of their model portfolios consist of long-term government bonds.
- ZFL, the BMO Long Federal Bond ETF, is up over 12% YTD in CAD.
- There is no denying the data. Compare their allocation to a small value-tilted portfolio, which has taken a beating this year, and they look really smart.
- Bonds tend to do better when stocks do poorly. Longer term bonds tend to swing more both up and down, but in a time of crisis this can be good.
- Another way to think about it is that in terms of volatility stocks dominate bonds in a portfolio, but adding in longer bonds helps to give the bond allocation a more equal risk contribution.
- Our criticism of long bonds is that their risk-adjusted returns have been poor over time.
- Wealthsimple finished their commentary by saying “How investments perform in downturns may be more important than how they perform in rallies.”
- That’s not true.
- First let’s look at the historical risk-adjusted returns of a 70/30 US equity portfolio with 5-year treasury notes vs. long-term government bonds as fixed income.
- The data start in July 1926 and end in March 2020.
- The portfolio with long-term bonds out-performed by an annualized 20 bps, but the risk-adjusted performance was pretty close. Keep in mind that these data include the last 40 years which have been the longest strongest bond bull market in history.
- Just for fun, if we look at the data ending July 1981, before interest rates began their steady decline, the long-bonds portfolio under-performed by 24 bps annualized and it was more volatile.
- So maybe it’s true that an allocation to long bonds helps cushion the blow when things drop suddenly.
- Now here’s the interesting part. Is downside actually risk? What about the risk of not meeting your financial objectives?
- I ran 764 historical 30-year simulations with 4% withdrawals from a starting $1m 70/30 portfolio.
- With 5-year treasuries as fixed income, there is one instance of failure for the 30-year period starting in December 1968. That is a 0.13% historical failure rate.
- With long-term bonds as fixed income, the historical failure rate increases to 4.6%.
- A lot of those failures occurred as interest rates were rising, which shouldn’t come as a surprise considering the interest rate sensitivity of long-term bonds.
- In both cases the average ending assets were fairly similar, and the average maximum drawdown was also similar, which was interesting.
- So if risk is drawdown, then long bonds are great. If risk is consumption, then in the historical data long bonds have not been so great.
- I wanted to push this one step further.
- What if we embrace drawdown and volatility instead of trying to hide from it? Try taking on the compensated risks with good risk-adjusted expected returns as opposed to bad.
- I tested a portfolio of 70% US small value stocks and 30 % 5-year treasuries.
- The failure rate goes to 0%.
- The maximum drawdown goes from 69% to 77%. The average maximum drawdown across the 30-year periods goes from 35% to 37%.
- Average ending assets increase by a multiple of 5.
- People might feel better about the long bonds portfolio and there’s something to be said for that.
- But in choosing the right risks to take I’m still hesitant to agree that long bonds make sense, even if they do help in times like these.

Portfolio Topic: The Historic State of Value Investing

We have talked about this before so hopefully it’s not tiresome, but in the world of academic investing this is a big topic right now.
The relative valuation of value stocks is exceptionally cheap right now compared to history.
We acknowledge that we have a bias to find confirming research for an expected value premium.
Also that all of these papers are from other people with a vested interest in confirming a value premium.
Regardless, the data are fascinating.
It’s also worth mentioning what would happen in the event that the expected value premium did disappear - value wouldn’t trail growth forever. It would just be a random outcome going forward.
There have been a bunch of interesting posts and papers and we will try to capture the best points.
Below are charts showing the relative value of cheap stocks vs. expensive stocks over time.

Cliff Asness: Is (systematic) value investing dead?

Following the Fama French price-to-book (academic style) expensive stocks are sometimes less than 4x as expensive as cheap stocks, the media is that they are 5.4x more expensive, and today they are nearly 12x more expensive.
That valuation spread is currently at the 100th percentile vs the 50+ years of data going back to 1967.
Some people make the winner-take-all argument and claim that value measures don’t apply anymore.
This implies that the trend is just starting and we should expect the spread to widen to infinity.
If we extend to other valuation metrics things look pretty much the same.
If we throw out technology, media, and telecoms, value is still cheap relative to history.
If we throw out the largest 5% of stocks for the entire data set, it’s still cheap.
If we go industry neutral, that is ranking stocks within their industries, still cheap.
Maybe cheap companies are just bad businesses?
Looking at gross profitability, return on assets, and leverage, cheap stocks are currently in line with their historical profitability compared to expensive stocks, their return on assets are higher, and their leverage is lower than average.
Measuring the value spread based on intra industry HML sorts while equal weighting the largest 1000 stocks, which Cliff says is the most realistic proxy for live portfolio implementation, the value spread has never before in history been this high.
The cheapness of value today is not coming from a “broken” metric (P/B), nor do the winner take all companies. It is not concentrated in tech, mega caps, or the most expensive stocks.
“We think the medium-term odds are now, rather dramatically, on the side of value, with no “this time is different” explanation we can find (and we’ve tested a lot of them!) holding a drop of water and no other period in the 50+ year history matching today.”

AQR paper: Is (systematic) value investing dead?

A value investor may be harvesting returns that compensate for

(i) errors in expectations with respect to fundamentals,
(ii) a risk premium for exposure to stocks that share exposure to a nondiversifiable source of risk that is reflected in their current cheapness, and/or
(iii) a premium for investors who are willing to overpay for growth or avoid value (i.e., non-risk based preferences).

Value has been strong for decades across stock markets, time periods, and other asset classes, but the recent evidence, particularly in the U.S., has been poor.

The underperformance has resulted in ex post critiques to try and rationalize the under-performance.

These criticisms include:

(i) B/P has not really worked for large stocks for a long time, if ever, (at least if the value strategy is not applied within industries or sectors)
(ii) the explosion in share repurchase activity of firms has changed the nature of book equity rendering B/P measures less useful,
(iii) the growing importance of intangibles and the failure of the accounting system to record such value on the financial statements renders value measures anchored to current financial statements useless,
(iv) central bank interventions and the low interest rate environment over the last decade have distorted asset prices via lowering discount rates that negates the efficacy of value strategies, and
(v) systematic value strategies are just too naïve to work as everyone knows about them.

(i) B/P has not really worked for large stocks for a long time, if ever, (at least if the value strategy is not applied within industries or sectors)

They confirm this to be true, but remind us that building a portfolio using a single metric does not make sense.
Adding other valuation metrics might be helpful, and while not directly related to value, using other systematic sources of returns make value work better.
Dimensional would say that controlling for the other factors in the Fama French five factor model makes other valuation metrics useless, which relates to the next point.
Things like momentum and quality can improve a value strategy.

(ii) the explosion in share repurchase activity of firms has changed the nature of book equity rendering B/P measures less useful,

They sort US stocks by size, relative price, and repurchase activity.
If repurchases affect the efficacy of value measures (B/P especially) we would expect value to be less effective in the high repurchase group and in more recent years where repurchase activity has increased.
They find mixed evidence of value working less well for the ‘High’ share repurchase sub-sample.
For B/P there is some evidence of lower returns for the ‘High’ group relative to the ‘Zero’ or ‘Low’ group in the small cap universe, but not in the large cap universe.
Across other value measures, and the combined valuation metric portfolio, the evidence is muted.
For both small cap and large cap they find no systematic evidence that B/P performs worse for firms that repurchase the most.
“Even though share repurchase intensity has increased over our sample period, it is not the case that B/P has performed worse more recently for share repurchase intensive firms.”

(iii) the growing importance of intangibles and the failure of the accounting system to record such value on the financial statements renders value measures anchored to current financial statements useless,

If the issue is a systematic accounting error that affects an asset (for example, research and development) then comparing similar firms within an industry should help to mitigate it.
For example, within tech the cheapest tech firms should still have a value premium even if they don’t look cheap relative to other industries.
Data vendors are now selling data that has been “corrected” for the flaws in the accounting system.
The corrections require lots of decisions (similar to the accounting metrics they are trying to “correct”).
To test the criticism that fundamental measures of value have become less useful the authors assess the performance of valuation metrics using adjusted data from one of these vendors, Credit Suisse HOLT.
The conclude:
“A key inference to be drawn here is that the recent under-performance of value strategies extends to value measures that attempt to correct for deficiencies in the financial reporting system. It appears unlikely that the growing importance of intangibles or changes in business models is explaining the underperformance of value strategies.”

(iv) central bank interventions and the low interest rate environment over the last decade have distorted asset prices via lowering discount rates that negates the efficacy of value strategies, and

Equity valuation frameworks are based on discounted future cash flows.
Growth stock valuations consist of a larger “speculative” component of future cash flows.
Based on this and leaning on the intuition of the duration concept from fixed income value stocks are effectively short duration assets, and as such their prices will move inversely with interest rates.
“But now the arguments are either unreasonable, or tenuous at best.”
Which interest rate are we talking about? The equity valuation discount rate consists of a risk-free rate and a risky rate. Which risk-free rate are we referring to? Is it the absolute level of rates or changes in rates?
Does the duration concept carry over to stocks? Equity expected cash flows are not fixed which makes the comparison to bonds less obvious.
Long-short, industry-neutral value portfolios exhibit little sensitivity with the level of interest rates.
“What looks like an appealing casual explanation for the troubles of value over the last decade (i.e., low rates benefitting assets with longer dated claims) is only minimally supported by the data, and then only contemporaneously and not predictively.”

(v) systematic value strategies are just too naïve to work as everyone knows about them.

They agree that it is always reasonable to consistently ask whether or not a characteristic will be associated with future returns.
It is also useful to remember why we held the prior belief of higher expected returns.
In the case of value, we have hard to diversify risk and expectation errors, both of which should be persistent.
If it became the case that value was being crowded out by too many investors, we might expect a compression in value spreads. In reality we have had the opposite.

Planning Topic: Risk Profile Revisited

We talked about risk profiling 4 weeks ago, and the difference between the 2 broad approaches to measure an investor’s risk tolerance:
- psychometric (answers to a series of questions that will describe one’s risk-taking mindset)
- Choice of gambles, or prospect theory.
Last time we talked about a paper written by 4 academics, that was commissioned by one of the product leaders in this space, Riskalyze. The paper made a very strong case for the benefit of focusing the short-term (6 month) downside risk, via a series of gamble questions. They argued this is what really mattered, and that other psychometric questions are not as valuable in the decision. The presentation was compelling.
Since then we have had the chance to look at the arguments on the psychometric side:
Psychometric responses are more stable than near-term downside tolerance:
- How easily do you adapt when things go wrong financially?
- When you think of the word "risk" in a financial context, which of the following words comes to mind first? 1. Danger. 2. Uncertainty. 3. Opportunity. 4. Thrill.
- When faced with a major financial decision, are you more concerned about the possible losses or the possible gains?
These responses can then be compared to a population, to help an advisor determine where you might be different from people with a similar score
An argument against the prospect theory, or gamble, is that it is gamified, and not at all like real life. Not to mention, a large proportion of the population are not able, or do not want to figure out, a math type question. Psychometric questions are easy to answer.
Also, many people know how the gamble questions work, and their answers to the gamble question may not reflect their true investing risk profile
In the end, there is merit in both approaches, and it is the advisor’s job to use a tool like this, to determine an investor’s risk tolerance. Other factors go into the their final asset allocation:
- Their ability (capacity) to take on risk (cash flows, time horizon …)
- Their need. How solid is their long-term financial plan?