The risk premium is the extra return investors expect for holding shares rather than a risk-free asset. With respect to the latter, government bonds are typically used as proxies. The size of this risk premium is one of the most controversial figures in economics and finance, because it plays an important role in a broad range of financial topics. The risk premium is key for asset allocation, the valuation of stocks, the investment decisions of individual firms and the determination of fair returns for regulated monopolies.
Unfortunately, the current risk premium cannot be directly observed from the market. There are two general ways of estimating the equity risk premium – using historical data or using forward-looking market expectations. Historical risk premia have several flaws. A survival bias will lead to too-high estimates, and even with decades of data the standard error will remain quite high. Typically, very long time series are used, so that one new data point does not affect the average by much. As a consequence, the risk premium estimate would basically remain unchanged and time-variations in the risk appetite of the average investor, which would lead to time-varying risk premia, cannot be modelled. A forward-looking methodology is not subject to these problems, and it seems the more natural approach since the value of an equity reflects the future prospects of a company, not the past. We therefore use a model that results in a forward-looking measure of what the market is pricing in for the risk premium – that is, the level of return the market can expect in the long term for equities in excess of government bonds.
To calculate this figure, it is necessary to know what the expected return is. For this, we use a present value model that follows a modern approach in valuation by using a residual income model (RIM) that focuses on the creation of shareholder value. A RIM links the fair value of a company with the rate of return a company generates and the return investors demand. The difference between the return a company generates and the investor’s required return is called abnormal earnings or residual income. The fair value of a company is then the present value of all future abnormal earnings discounted at a rate that reflects the return a shareholder expects, plus the current book value of the shareholders’ equity. Hence, the abnormal earnings can be interpreted as that part of the earnings that adds or destroys value to the shareholders’ equity. We assume that the current share price is the best estimate for the fair value, and then determine the appropriate required rate of return. Finally, the yield on 10-year Bunds is subtracted to arrive at the risk premium currently priced in the market.
Our analysis is sector-focused. Using a bottom-up approach, we calculate the risk premia on individual stocks. The risk premium of each of the 18 DJ Stoxx sectors is then calculated as a weighted average of the risk premia on the individual stocks. Based on data from 15 February 2002, figure 1 shows the changes in the sector risk premia, both since the beginning of the year and over the past year. So far in 2002, risk premia went up in most sectors and in the broad market. This effect was particularly strong in the telecom (+116bp) and energy (+51bp) sectors.
Figure 2 shows as a yellow square the current ex ante risk premia for all individual DJ Stoxx sectors. By contrast, the red square depicts the long-term average for the respective sectors. These averages are based on monthly data since December 1993. For 15 of the 18 sectors, the risk premium is currently higher than on average in the past. The interpretation of this finding sometimes causes confusion. Since the risk premium is a return measure, the higher it is, everything else being equal, the more attractive the respective sector looks. Consequently, in our case 15 of the 18 sectors look more lowly rated than on average in the past.
The band around the historical average reflects plus (minus) one standard deviation and indicates how far valuations tend to differ from historic average valuations. At the moment there are eight sectors that look quite lowly rated by the market. Those sectors are: autos, banks, energy, financial services, healthcare, insurance, non-cyclical goods and services and utility suppliers. By contrast, in none of the three sectors basic resources, media and telecom, which are more highly rated than on average in the past, is this higher valuation substantial.
Aggregating the individual sector risk premia, we derive a value of 4.67% for the broad DJ Stoxx. This is 25bp higher than at the end of December 2001. As a consequence, the European stock market as a whole looks quite lowly rated again. This is illustrated in figure 3, where the red line depicts the evolution of the risk premium on the DJ Stoxx over time, starting in December 1993. The yellow line shows the long-term average, while the upper (lower) grey line reflects the long-term average plus (minus) one standard deviation.
Some interesting extreme valuations can be noticed from this chart as well. At the height of the emerging markets crisis during late summer 1998, prices dropped significantly, while analysts did not revise their earnings forecasts by much. As a consequence, the risk premium shot up to reach the maximum of 5.67% for the whole period that we investigated, that is, the stock market then reached its most attractive valuation compared to bonds. With prices recovering again, the expected risk premium declined in the following months. Another interesting phase is the bubble of TMT stocks during the last quarter of 1999 and the beginning of 2000. The rapid price increases of technology, media and telecom stocks were not accompanied by appropriate increases in earnings and dividend forecasts, resulting in a substantial decline of the expected risk premium on the DJ Stoxx. Hence, the overall market looked remarkably higher valued than on average in the past. With the bubble bursting, the risk premium went up again.
Following the terrorist attacks in the US, stock prices came down substantially, resulting in a strong jump in the expected risk premium. As a consequence the broad market looked extremely lowly rated. The price recovery that followed in the fourth quarter of 2001 did, therefore, not come as a surprise to us, as it was consistent with our model.
To summarise, we are again in a situation were the risk premium on the DJ Stoxx is not only higher than the past average, but also higher than the average plus one standard deviation – in other words, the broad market does look quite lowly rated at the moment. The same thing can be said for eight sectors, namely autos, banks, energy, financial services, healthcare, insurance, non-cyclical goods and services and utility suppliers. None of the sectors currently looks quite highly rated by the market.
Klaus Reimer, Alexander Zanker and Andreas Nawroth are in the alpha generation team at Commerzbank Securities in Frankfurt