Originally published by PokfuLam Investments
You can view part one of this series here.
The dividend discount model (DDM) was recognised as the way to value stocks, originating from when stock markets were formed in the 17th century. Then, dividends were the returns expected of stocks. However, things began to change around the turn of the 20th century as investors gained access to prices being achieved on the trading floor via tickers, rather than receiving what brokers agreed to pay for them. The result was that capital gains came into the equation.
By the early 1930’s the issue over those gains and their impact on prices began the academic study of markets, as investors and brokers searched for reasons why prices went so high, and then so low, over the previous 15 years. Initially, the major reason given was false profit statements and forecasts by listed companies desperate for capital, particularly those involved in the Florida Land Boom. Since then, there has been a focus on cashflows as against earnings: much to the chagrin of the accounting profession who have worked since the Great Crash to introduce accounting standards to standardise profitability and asset values, to minimise such issues.
WE HAVE NOT PROGRESSED FAR...
Amazingly, we have not progressed far from the 1930’s in explaining the way that markets price themselves, and particularly, in relation to capital gains. It took until the 1980’s (Shiller (1981); Campbell and Shiller (1988) to prove conclusively that share prices could not be explained by any rational forecasts of dividends alone. Given that there are only two sources of returns to listed stocks (capital gains and/or dividends), the difference MUST be explained by changes in capital gain expectations and/or changes in the discount rate applicable to the NPV calculation.
What this suggests, along with the behaviour of markets before and after 1929, 1987, 2000 and 2008, is that there must be capital gains expectations as part of market prices. If that is so, the question becomes, what determines those expectations, and what discount rate is applicable?
MAJOR ASSUMPTIONS
One of the major assumptions applied when trying to model markets has been that listed and unlisted stocks should have the same worth. This has been convenient given corporate valuation methodologies have effectively been constant since 1900, including DCF which is generally considered the most sophisticated methodology. Security Analysis (Graham and Dodd (1934)) suggested that DCF was the most relevant methodology in relation to stock prices BUT did not specifically explain market pricing. Instead they argued that, like any other methodology, all that could be garnered from corporate valuations was a price below which a stock was cheap, and above which it was too expensive. The justifiable price between those extremes was anyone’s guess, and yet, this has been the dominant approach taken since then.
Despite this, DCF does make a limited, if incomplete, attempt to incorporate capital gain expectations. The key assumption is that total returns equal cash generated. For a company, at the end of the financial year, there are only two uses for cash: pay dividends, or retain cash and add to the net book value of the company. If we look at that in terms of expected returns to stocks, it would mean that stocks would have to trade at or near book value.
As early as 1934, it was obvious that markets do not trade at or near book (Graham and Dodd (1934)). As examples, the Australian market has traded between 1.3x and 3x, averaging 2x since 1995; the S&P500, 1.2x to 6x, average. 3x since 1990; the FTSE, 1.2x and 6x, average. 2.6x since 1993.
Quant methodology has tried to be more specific in terms of forecasting markets, as well as defining and quantifying risk in terms of (and requiring) efficient markets, but does not generally focus on the determinants of specific return components.
EFFICIENT MARKET HYPOTHESIS
Unfortunately, Fama (1991), the author of the Efficient Market Hypothesis in 1965, stated that the strong form EMH was false, and stated a new theory: The Joint Hypothesis Problem. This states that neither rationality nor efficiency can be proven or tested without first developing a market pricing model that explains both absolute price levels and a high proportion of volatility of prices. This is necessary to prove what the “right” price should be: a key necessity of both rationality and efficiency, but which has largely been ignored in academic research. Without such a model, it is impossible to determine any quantum of error inherent in return-based results.
While we would like to explain further how we go about estimating and pricing the capital gain component of share prices specifically, that would be “throwing the baby out with the bath water”. However, a model that works effectively must get this aspect of price modelling correct.
A NEW MODEL
We have developed a model that does work in terms of the S&P/ASX 200 and other major global markets. This was largely detailed in a previous article. That model incorporates both dividends and capital gains, and enables us to measure (or at least estimate) the market equity risk premium implied by prices at a point-in-time. Importantly, it explains approximately 86% of monthly market price volatility and, at the same time, absolute levels of price. We will be discussing the risk premium estimates further in a subsequent article.
One of the key features of the modelling has been valuation of the constituent stocks of the market index. To do this, we value stock-specific return streams, and use cross-sectional analysis to fundamentally quantify market-relative risk without having to assume market efficiency.
The data used in the measurement of market-relative stock risk is detailed in the published article and on the company website, while the PowerPoint presentation on the website shows examples of the actual outputs of the stock-specific modelling.
AN EFFECTIVE MODEL
This work shows just what advancements can be achieved once an effective model has been established. This particularly applies to the measurement of risk, which has always been problematic for analysts both in terms of a definition, and then measurement.
As with the other aspects of the modelling, we have defined risk as the risk premium implied by the price after setting the indicated return inputs.
The only alternative method of estimating the risk premium at the stock level has been beta, which relies upon strong-form market efficiency for its validity. As mentioned above, this must be questioned per Fama (1991).
SO WHAT IS THE ALTERNATIVE?
If not our model, and not beta, what is the alternative for investors to effectively measure investment risk when valuing stocks?
You can view part three of this series here.