Just finished going through “Value Investing: Tools and Techniques for Intelligent Investment” by James Montier. Took me a long time and lots of sporadic reading to finish this book.
This book, as its title suggests, is all about value investing. There are some books out there where the title is not really reflective of the substance, for this one, it really truly is. The book basically captures all the arguments of “Why Value Investing”, quoting tons and tons of studies showing that value investing works, how it works, why it works, why academic finance / efficient market theory is wrong, and also has articles on the psychology of investing.
This is THE book to convince someone of the value investing approach. If any one doubts the approach, throw this book at them 🙂
Montier is heavily influenced by Graham & Dodd. He uses Graham’s deep value screen a couple of times, and also G&D’s Cyclically-Adjusted P/E (CAPE) to assess whether how overbought/oversold the market is compared its historical mean.
Some of the key themes that was highlighted throughout the book includes:
- Humans can’t forecast, hence DCF is too tough. Use liquidation value or earnings power value.
- Mean reversion is the rule. Dogs outperform Stars.
Overall, this is pretty good book to go through. The results of many of his studies are pretty interesting and useful for further refinement or incorporation into your investing style.
A more detailed summary of key points follow.
CAPM is Flawed
- Fama and French’s 2004 review of the CAPM (using data from 1923 to 2003) showed that CAPM woefully underpredicts the returns to low beta stocks, and massively overestimates the returns to high beta stocks. Over the long run, there is no relationship between beta and return. Using low P/B stocks, a long low beta, short high beta strategy creates additional 8.3% p.a. returns.
- A cap-weighted market index is not efficient. Rob Arnott (at Research Affiliates) showed that fundamentally weighted indices (based on earnings and dividends, etc.) can generate higher return with lower risk.
Discounted Cash Flow (DCF) is Too Tough: Use Other Alternatives
- Forecasting is too tough. The average 24-month forecast error of analysts is ~94%, and average 12-month forecast error is ~45%.
- The terminal value is too sensitive to the perpetual growth rate and cost of capital assumptions.
- Cash flow from the 1st 3 years makes up 10% of the DCF value.
- Cash flow from next 5 years makes up 15% of the DCF value.
- Cash flow from year 9 onwards makes up the other 75% of the DCF value.
- Three alternatives
- Reverse-engineer to back out the implied growth rate based on the current price. Then assess whether the implied growth rate is reasonable. Historically, the annual growth rate of operating income (before depreciation) over 10 years is at 0% (10th percentile), 5% (25th percentile), 10% (50th percentile), 12.5% (75th percentile), 25% (100th percentile).
- Use assets’ liquidation value. Graham had the following weights (cash and marketable securities: 100%, receivables (less reserves): 80%, inventories: 66.67%, fixed assets and intangibles: 15%).
- Calculate earnings power value (EPV), i.e. value of what the company might be expected to earn year after year if the business conditions prevailing during the period were to continue unchanged. First calculate a normalized EBIT by multiplying average sales (over 5 years) by the average EBIT margin (over 5-10 years). Then subtract interest payments and taxes to get the earnings power, which can then be capitalized to get the value.
[My note: While EPV does not take into account growth, it is a conservative figure based on already achieved historical performance, so it is a good gauge to value a company that has a temporary set-back. Using EPV will make the assumption that in a few years time the company will be making what it already made in the past, so if it is cheap by this measure it is a good bet. Most of the value in a company’s stock is locked in the future, but humans can’t forecast the future well, so the best gauge is still what the company has historically achieved.]
Value Depends on Where You Are in a Cycle
- There is a good quote from Howard Marks of Oaktree Capital on how the value of something depends on the cycle, and that it may sometimes be better to wait for a better buying opportunity later even if the price is attractive now:
“In my opinion, there are two key concepts that investors must master: value and cycles. For each asset you’re considering, you must have a strongly held view of its intrinsic value. When its price is below that value, it’s generally a buy. When its price is higher, it’s a sell. In a nutshell, that’s value investing.
But values aren’t fixed; they move in response to changes in the economic environment. Thus, cyclical considerations influence an asset’s current value. Value depends on earnings, for example, and earnings are shaped by the economic cycle and the price being charged for liquidity.
Further, security prices are greatly affected by investor behavior; thus we can be aided in investing safely by understanding where we stand in terms of the market cycle. What’s going on in terms of investor psychology, and how does it tell us to act in the short run? We want to buy when prices seem attractive. But if investors are giddy and optimism is rampant, we have to consider whether a better buying opportunity mightn’t come along later.”
Dogs Outperform Stars: The Story of Mean Reversion
- Star stocks (those with past and forecast high sales growth) underperform dog stocks (those with low past and forecast sales growth) by ~6% p.a.
- The most despised companies outperform the most admired companies by ~2.5% p.a. (Anginer, Fisher and Statman). The companies are rated based on a survey conducted by Fortune magazine on 8 attributes such as quality of management, HR skills, use of corporate assets, long-term investment value, etc.
- The stocks of the “visionary companies” in Collins and Porras’ Built to Last: Successful Habits of Visionary Companies book, performed no better than the S&P500 subsequently, the “comparator companies” performed better after 17 years.
- For U.S. stocks, there is a 0.98 correlation between past growth and forecast growth, and a -0.9 correlation between forecast growth and actual delivered future growth.
- For European stocks, there is a 0.88 correlation between past growth and forecast growth, and a -0.77 correlation between forecast growth and actual delivered future growth.
- Profitability (earnings/total assets) is mean reverting at the rate of ~40% p.a (in both directions).
- Buy-side analysts are worse than sell-side analysts in forecasting earnings. From 1997 to 2004, the mean absolute forecast error of buy-side analysts vs. sell-side analysts is 27% vs. 6% (short-term forecast: 0-3 months), and 68% vs. 38% (long-term forecast: 18+ months).
Sell on Profit Warnings: Importance of Precommitment
- Precommitment refers to making a decision with a rational frame of mind, and not during the heat of the moment.
- From 1998 to 2000, the average UK stock dropped 17% on the day of the profit warning announcement. The average stock then drifts further down for the next ~3.5 months (to ~23% loss) before slowly recovering to ~18% loss at ~6.5 months (resistance), drifting down again to ~20% loss at ~9 months, before rebounding all the way up.
- You need to follow the automatic rule to sell on profit warnings, and not make excuses to hold on to the stock.
- Buying stocks 6 to 9 months after a profit warning delivers abnormal returns of 7.7% over the following 12 months (though with standard deviation more than 100%).
Emotions over Cognition
- Humans react immediately based on a quick and dirty decision maker (the X-system), rather than using their more logical but slower system (the C-system).
- Fear and greed causes our X-system to make incorrect decisions to buy and sell at the wrong time.
- 60% of fund managers fail to get all 3 questions right due to our X-system
- A bat and a ball together cost $1.10. The bat costs a dollar more than the ball. How much does the ball cost?
- It takes 5 machines 5 minutes to make 5 widgets. how long will it take 100 machines to make 100 widgets?
- In a lake there is a patch of lily pads. The patch doubles in size every day. If it takes 48 days for the patch to cover the entire lake, how long would it take for teh patch to cover half the lake?
- Different parts of the brain are activated when assessing immediate and delayed rewards. The X-system causes people to want $10 today instead of $11 tomorrow, but the C-system causes people to forego $10 one year later for $11 one year plus one day later.
Risks to Permanent Loss of Capital
Montier highlights that the risks to permanent loss of capital consists of 3 interrelated sets of risks. It is interesting to note that Buffett’s methodology of a good business, good management, good price, addresses each of the 3 risks nicely.
- Valuation risk – need to get a good price with margin of safety
- Business/earnings risk – need to get a business with sustainable competitive advantage and strong management
- Balance sheet/financial risk – need to get a business with a strong balance sheet (quantitative evaluation of a good business)
Deep Value Opportunity Criteria Screen
Designed by Ben Graham in the few years before his death, supplemented and published by James B. Rea in the Journal of Portfolio Management (1977).
- Get decent returns
- Trailing earnings yield > 2*AAA bond yield
- Dividend yield >= 2/3*AAA bond yield
- From a decent company
- Earnings CAGR over 10 years >= 7%
- Number of annual earnings decline of 5% or more in the last 10 years <= 2
- At a low price
- P/E < 40% of peak P/E (based on 5-year moving average earnings)
- With low risk
- P < 2/3*Tangible book value
- P < 2/3*Net current assets (i.e. net working capital – all debt)
- Total debt < 2/3*Tangible book value
- Current ratio > 2
- Total debt <= 2*Net current assets
Rea had used the criteria to run a fund – American Diversified Global Value Fund – but the fund did not do well. Saxo Bank has a Global Value Equity Portfolio based on the criteria above with some modifications:
- 60-day average daily trading volume > EUR 1M (for liquidity)
- dividend yield >= 2*AAA corporate bond yield
- interest-bearing debt < 2/3*Equity
- CAPE 5-year (cyclically adjusted P/E using 5 years average earnings) <= 50
They then took the stocks and ranked them based on a weighted average of separate rankings for Piostroski score, CAPE, and P/B, then choosing the top 15 stocks. The portfolio is adjusted monthly, and on average 3 stocks are swapped in/out each month. The portfolio had 29.6% annualized gross returns (before transaction cost and taxes) from 2003-2011.
Montier has an article on buying Graham net-nets, highlighting the buying a basket of global net-nets would have generated a return of over 35% p.a. from 1985-2007 (interestingly while the article appeared on Sep 2008, the study stopped at 2007 just before the crash). An interesting note is that 5% of the portfolio constituents drop more than 90% in a year, so a diversified and nerves of steel are required.
Graham and Dodd P/Es
On the use of average earnings for P/Es to smooth out effects of business cycles, Montier reported that using P/Es constructed using 10 years of average earnings will result in a 5% p.a. outperformance compared with using P/E with one-year trailing earnings. Graham argued that investors should never pay more than 16x average earnings for any stock. This is both to be conservative and so that the earnings yield is not less than 6%.
Statistics on G&D P/Es (10-year moving average of earnings)
- Average since 1881 is 18x for S&P500 if including bubble years, 16x if excluding bubble years.
- Bargain basement level is around 10x.
- Absolute minimum over last 130 years is 5x.
Montier also compares between 2 strategies, one where the strategy buys in when the S&P500 G&D P/E (10-year) = 10x, and another when the S&P500 G&D P/E (10-year) = 13x
- For the 1st strategy (i.e. 10x), on average the strategy would be buying in 4 months before the the bottom, which is a further 20% drop, and would need to wait 12 months (from the point of buy-in) to break even.
- For the 2nd strategy (i.e. 13x), on average the strategy would be buying in 9 months before the bottom, which is a further 17% drop, and would need to 17 months to break even.
It is difficult to adjust the buy-in P/E threshold to maximize the returns because if you lower the P/E threshold, you might find yourself not being in the market for decades! Montier listed the month/year in which the 2 strategies would be buying in.
- 10x: Sep 1917, Oct 1931, Mar 1942, Mar 1982.
- 13x: Apr 1884, Jan 1907, May 1913, Jan 1917, Aug 1931, Nov 1937, Feb 1941, Feb 1949, Aug 1974, Jan 1977, May 1981, Feb 1984.
Finally on the robustness of the G&D P/E measure, Montier tested another method to a cyclically-adjusted P/E ratio (CAPE). He did it by calculating the average ROE over 10 years, and multiplying that with the current Book Value to get the cyclically-adjusted EPS. The P/E ratio calculated using this method performed similar to the G&D P/E ratio.
Shorting is Tough!
Form a portfolio of stocks passing the following screen
- P/S > 1
- The lowest P/S quintile of Europe stocks outperformed the highest quintile by 9% p.a. from 1985-2007.
- Piostroski F score <= 3
- From 1972 to 2001, average returns for U.S. firms with low F scores (0-3) is 7.3% p.a., medium F scores (4-6) is 15.5% p.a., and highest F scores (7-9) is 21% p.a. This compares with the market return of ~12.8% p.a.
- From 1985 to 2007, average returns for European firms with low F scores (0-3) is 4.4% p.a., medium F scores (4-6) is 13.1% p.a., and highest F scores (7-9) is 15% p.a. This compares with the market return of 12.5% p.a.
- Total asset growth >= 10%
- High asset growth is a sign of lack of capital discipline.
- 17% of capital invested are underperforming and should be terminated, 16% should not have been started, 40% of managers hide or misrepresent information when submitting capital investment proposals.
- From 1968 to 2003, firms with low asset growth outperformed firms with high asset growth by 13% p.a. after controlling for market, size and style.
- A research later showed that the market tends to respond favorably to announcements of capital investment increases. However there is diminishing marginal returns from investment growth which negatively impacts future profitability. Negative returns result later when the market realizes the negative implications. It also showed that the effect can be explained by growth of Net Operating Assets (NOA) instead of the full Total Assets.
- NOA = Operating assets – operating liabilities
- Operating assets = almost all the common balance sheet asset items except excess cash and other assets that can be liquidated without affecting operations
- Operating liabilities = accounts payable, accrued liabilities, deferred income tax liabilities, pension and other post-employment obligations (does not include liabilities arising from financing of assets such as long-term lease liabilities, debt, or equity).
- Change in NOA = Operating Income – Free Cash Flow
A portfolio with European stocks meeting the above characteristics, rebalanced annually, would have
- Declined 6%+ p.a. between 1985 and 2007.
- Have absolute negative returns 45% of the time (10 out of 23 years).
- Underperformed the index 81% of the time (18 out of 23 years).
- 60% of the stocks with absolute negative returns.
- A few stocks that do exceptionally well on the long side.
- Outperformed the market by 6% in 2003.
My Note: Clearly the above criteria chosen to form a short portfolio wasn’t good. You want the portfolio to have a high probability of making absolute negative returns, if its just underperforming the market, it is troublesome and more risky to have to long the market, short the portfolio, leverage it up for decent returns. P/S > 1 is too lax, its better to choose small companies with much higher P/S (see my earlier post on Kenneth Fisher’s Super Stocks book). F Score should be applied to low P/B stocks not just any stocks, a lower F Score merely means that it does not perform as well as higher F Score low P/B stocks, which is not good enough. Using other measures for financial stability position (e.g. the other F score, Altman Z, or A score) should be better. Again for the total asset growth criteria, relative performance is not good enough.
David Einhorn from Greenlight Capital I think does the best shorting I have seen, i.e. he shorts companies engaging in some form of fraud (e.g. accounting). Shorting because you think something will underperform the market, or because of overvaluation (e.g. Whitney Tilson vs Netflix) is not a good way to go, the probability of it working out is lower, the high valuation can go ever higher, the company might chance on a breakthrough, etc. Fraud on the other hand is a slippery slope that those companies will continue to slide down.
Montier’s C-Score (for cooking the books) captures how many of 6 common earnings manipulations a firm is engaging in
- Increasing difference between net income and cash flow from operations
- May indicate more aggressive capitalization of costs
- Increasing day sales outstanding (DSO)
- Measures how long it takes to collect receivables
- (Accounts receivable / total credit sales) * number of days in period (e.g. 365)
- Indicates accounts receivable are growing faster than sales
- Increasing day sales of inventory (DSI)
- Measures how long it takes to turn inventory into sales
- (Inventory / Cost of Sales) * number of days in period
- Increasing inventory indicates slowing sales
- Increasing “Other Current Assets” to Revenues.
- Really cunning CFOs make “re-allocate” receivables and/or inventory into “Other Current Assets” to make DSO and DSI look healthy.
- Decreasing depreciation relative to gross property plant and equipment
- May indicates useful life being altered to massage earnings
- High total asset growth
- May indicate a serial acquirer that is using acquisitions to distort earnings
From 1993-2007, high C-Score U.S. stocks underperform the market by ~8% p.a. (~5% for European stocks). The lower the C Score, the better the performance. A portfolio with stocks of C-Score 5 and P/S > 2, resulted in negative absolute return of 4% p.a.
Last but not least, Montier had a good quote from Jim Chanos from 2005 on his 4 categories of short selling targets:
- The first and most lucrative are the booms that go bust. We’ve had our most success with debt-financed asset bubbles, as opposed to just plain asset bubbles where there are ticking time bombs in terms of debt needing to be repaid and where there are people ahead of the shareholders in the bankruptcy or workout process. The debt-financed distinction is important. It kept us from shorting the internet in the 90s that was a valuation bubble more than anything else.
- The second group of opportunities are those created by technological obsolescence. Economists talk quite rightly about the benefits of creative destruction, where new technologies and innovations advance mankind and grow GDP. But such changes also render whole industries obsolete – what is playing out now is the transformation from analogue to a digital world.
- The next group of categories revolves around poor accounting. This can run the gamut from simple overstatement of earnings to outright fraud. We’re trying to find cases where the economic reality is significantly divorced from the accounting presentation of the business.
- The final category are “consumer fads” – where investors become way too optimistic: Cabbage Patch Kids in 1980s, NordicTrack in the early 1990s and Salton with the George Forman grills.
How to Identify Firms with Competitive Advantage
Bruce Greenwald and Judd Kahn highlighted in their Competition Demystified book that the main factor is the barrier to entry. Some key points
- If there are no barriers to entry, the firm will ultimately earn returns equal to the cost of capital. The total emphasis must be on operational efficiency to be the lowest cost producer.
- Three step process to analyse competitive advantage
- Identify the competitive landscape in which the firm operates. Which markets is it really in? Who are the competitors in each one?
- Test for the existence of competitive advantages in each market (i.e. are there even firms with competitive advantage in this particular market? answer can be No). Do incumbent firms maintain stable market shares? Are they exceptionally profitable over a substantial period?
- Identify the likely nature of any competitive advantage that may exist. Do the incumbents have proprietary technologies or captive customers? Are there economies of scale or regulatory hurdles from which they benefit?
Successful M&A Deals are Hard to Come By
A 2006 KPMG survey on M&A deals (found an older survey here) showed that
- 70% failed to add shareholder value. Of these, only 35% achieved their expected synergies.
- 30% add value but only 61% of these met their synergy targets.
- Typically 43% of the expected synergy value was included in the purchase.
Christofferson et al. (2004) on delivering synergies
- 70% fail to achieve expected revenue synergies.
- 60% manage to deliver at least 90% of expected cost-cutting synergies.
Another 2010 survey by KPMG has more interesting tidbits on the determinants of M&A success (not in the Montier’s book)
- Cash deals are significantly more successful than stock deals or stock-and-cash deals
- Acquisitions made by acquirers with low P/Es are significantly more successful than those made by acquirers with high P/Es. This is likely due to 2 reasons: i) low P/E acquirers value the target more conservatively, ii) high P/E acquirers’ stock dropped when their P/E reverts to the industry mean
- Acquirers that do 3-5 deals in a year are more successful than those that do 6-10 deals a year.
- Acquirers that do the deal for geographic expansion, increasing financial strength, or hard asset buys, do well. Deals for other reasons do not do well: increase revenue, distribution, product expansion, earnings accretion, cost savings, acquire intellectual property.
On a related note, Bain & Co. showed that 75% of companies’ moves into adjacent markets end in failure. Mergers which increase focus deliver small but positive abnormal returns. Focus-decreasing mergers end up destroying value with an abnormal return of negative ~20% over 3 years.
How to Look at Bonds
Bond valuation has 3 components
- Real yield. Roughly equal to the long-term real growth rate (in the long-run, marginal cost of capital = marginal benefit of capital; marginal benefit corresponds to the growth rate, marginal cost corresponds to the real yield). The real yield can be obtained from the yield in the TIPS market.
- Expected inflation. Obtained from surveys of professional forecasters, or find the implied inflation from the inflation swap market.
- Inflation risk premium to account for the uncertainty of inflation. Academic work estimates it at 25 to 50 bps.
Evaluate bonds by taking the nominal bond yields, breaking the yield down into the 3 components above, and see whether you agree with the market’s view. For e.g. the 10-year U.S. Government bond nominal yield of 2% is implying 0% inflation since the real yield from TIPS is 2%.
Things To Dwell Further
- Wright model for probability of recession
- Different P/E measures: Hussman P/E, cyclically adjusted P/Es, Graham and Dodd P/E (i.e. average over at least 5 years, preferably 7 to 10), Graham’s formula P/E.