Market timing part 3: Graham Classic Central Value Model
Over the last few weeks, we’ve been discussing some technical methods by which to time stock market cycles. Our ultimate objective has been to enter the market (i.e. Standard & Poor's 500 in the U.S. and the S&P/TSX in Canada) at near market troughs and to exit the market at near-market peaks, without staying too long out of the market or missing any major market moves. There are various techniques or “formula timing” indicators that we follow.
The first indicator discussed was our Triple Crossover Moving Average Model. The second indicator discussed was the well-known Fed Model. This week we are going to discuss the once well-known, but long forgotten, Classic Benjamin Graham Central Value Model. This model involves determining an equilibrium or fair-value estimate for the market, as well as an upper and lower valuation range based on historical interest rate and fundamental data.
Methodology
Graham estimated the central value or fair value of the market by capitalizing the average EPS over the previous 10 years by an “equilibrium” multiplier equal to 1 divided by k-times the yield on AAA-rated corporate bonds. Graham then established lower and upper price bounds equal to 80% and 120% of the central value estimate, respectively.
In Graham’s original model k=2. That is, he capitalized earnings at 1 divided by twice the AAA-rated corporate bond yield. While that might have worked in the 1940s and 1950s, capitalizing at k=2 in today's fast-moving market environment does not provide satisfactory results. To determine the value of "k" for our model results below, we rely on an optimization procedure that varies the parameter value one-by-one until it converges on a “best-fitting” central value.
A summary of the valuation formula is provided below:
Central value = (10-year average EPS) x (1/(k x AAA-Corporate Bond Yield))
Graham explained that when the market price line trends near or exceeds the upper price bound (that is, 120% of its central value), then investors can expect the market to fall or move sideways until it is back within the central value range. At these times, investors should consider shrinking their positions or taking some profits. Similarly, Graham explained that when the market price line trends near or falls below the lower price bound (that is, 80% of its central value), then investors can expect the market to rise until it is back within the central value range. At these times, value investors should consider growing their positions.
U.S. S&P 500 findings
Figure 1: S&P500, Central Value Estimate, Upper and Lower Valuation Bands, Jan 1990 - Jan 2016
The results presented above are quite satisfactory. The model accurately predicted a sideways market through the early 90s and then signalled significant overvaluation leading up to the bursting of the tech bubble in 2000 to 2001, which would have helped investors sell-out before the market top. It does not, unfortunately, provide very strong buy signals near the market bottom of 2002. This type of failure in the model was actually flagged by Graham in the 1940s in which he stated that "...while it may be sufficiently dependable to be worth using for its overall results [over long periods of time], like all others of the kind, it is not fully dependable."
It happened that the model signaled fair valuation from 2003 and 2006, but trended at the upper-end of the valuation range with occasional breaches. The model successfully signaled overvaluation in 2007. Had an investor been guided by the formula, he/she might have sold out in 2007 and avoided incurring massive losses in 2008. Also, in 2009, the model signalled a strong contrarian buy signal. The model signaled undervaluation again in 2011. The model signalled fair valuation in 2013 and 2014 and, at current rates and based on the most up-to-date earnings data, the market appears to remain fairly valued but at the low end of the valuation range.
Canada S&P/TSX findings
Figure 2: S&P/TSX, Central Value Estimate, Upper and Lower Valuation Bands, Jan 1990 - Mar 2016
The Canadian results presented above are also quite satisfactory. The model accurately signalled a buy market in the early 90s and then signalled significant overvaluation leading up to the bursting of the tech bubble in 2000 to 2001, which would have helped investors sell-out before the market top. It does not provide a very strong buy signal near the market bottom of 2002, but investors purchasing when the price line returned to the equilibrium valuation range would have been well rewarded. It happened that the model signaled slight-to-moderate overvaluation from 2004 to 2006, with the price line trending above the upper-end of the valuation range. The model continued to signal overvaluation through 2007 and peaked in 2008.
Had an investor been guided by the formula, he/she would have been wise to sell when the spread between the price line and upper valuation band grew to abnormal levels, indicating bubble behaviour. This could have helped them avoid massive losses in 2008 and early 2009. The model also successfully signalled a strong contrarian buy signal in 2009. An investor buying in 2009 and holding through the end of 2015 as the market continued to signal fair-valuation would have earned a nice 64% rate of return. The model gave another strong buy signal in 2011. At current rates and based on the most up-to-date earnings data, the market appears to be undervalued and can be expected to appreciate in value until it is back within the equilibrium valuation range.
Stay tuned
So far we have discussed three market timing indicators that we find particularly useful for gauging long-term market trends. We will build on this discussion further next week when we discuss Graham's Modern Central Value Model.
