|
Face - Index - Top/Bottom - Previous/Next "Monetary review - 4th Quarter 2000" |
 |
Stock Prices, Property Prices and Monetary PolicySteen Ejerskov, Economics Introduction and main conclusionsDuring the 1990s the USA and several European countries, including Denmark, saw periods of strongly rising prices for stocks and owner-occupied homes, two important assets in private individuals' portfolios. The development in asset prices is closely monitored by central banks since substantial fluctuations in asset prices significantly influence financial and economic stability, including price stability. Economic and financial crises have more often been related to strong fluctuation in asset prices than to fluctuation in prices for goods and services. A sharp fall in asset prices following substantial increases has often had severe economic implications. The price of an asset depends on the expectations of the future course of the economy. Minor changes in the prevailing economic situation can entail considerable shifts in expectations of future trends. Asset prices therefore naturally fluctuate more than other economic variables. In certain periods expectations can be too optimistic, or too pessimistic, causing asset-price fluctuations which deviate considerably from movements consistent with the economic fundamentals (speculative bubbles). A speculative bubble in an asset is typically followed and nourished by an increase in borrowing, e.g. borrowing against the free mortgageable value of a property, or gearing of the rising market value of a stock portfolio. The financial sector is therefore vulnerable when the bubble bursts. During the last two decades inflation has been reduced successfully in Europe and the USA. Nevertheless, this period has seen more financial crises than previously in the last century [1]. Significant increases in prices for stocks and housing in Denmark during the economic expansion in the mid-1980s were followed by strongly falling housing prices and a prolonged recession. The other Nordic countries saw even stronger increases in asset prices at the end of the 1980s, and the subsequent price drops had a severe impact on the financial sector, leading to a recession at the beginning of the 1990s. Beyond the Nordic region, a similar scenario prevailed in the UK at the beginning of the 1990s. In Japan, the financial sector is still suffering the consequences of a substantial portfolio of non-performing loans after a speculative bubble in property and stock prices burst at the end of the 1980s. Japan's inflation was low in 1986-87, and the authorities were therefore able to lower interest rates to curb the strengthening of the yen [2]. The lower interest rates led to further increases in prices for stocks and housing. In this case focusing solely on the primary objective proved to be inadequate [3]. The disruptive effects are most severe if asset prices deviate far from their fundamental level before the bubble bursts. The problems can be reduced if the bubble can be deflated before the deviation becomes large. However, in practice bubbles are difficult to identify before they have burst, and their effect on the overall economy is subject to great uncertainty. If many indicators point to the same, however, this could signal that the development is not sustainable. Under such circumstances an economic-policy response may be needed to ensure fiscal and economic stability in the longer term. Stock prices in a number of countries, including Denmark, are historically high, even after the fall in prices in recent months. This may reflect that the current estimates of the potential real growth rates of the economies are too low, for example because the effects of the development of networks and information technology are not taken sufficiently into account. It may also reflect that investors in stocks are willing to accept a reduction of the risk premium, e.g. due to the liberalisation and integration of financial markets and the improved macroeconomic environment in the 1980s and the 1990s. It is still unclear, however, whether these effects are of such magnitude that they justify the current level of stock prices. Property prices have risen considerably in Denmark since 1993. These increases followed a period of price lags in the housing area compared to the general price development. As a ratio of construction costs, the cash price index is close to the relatively high level in the mid-1980s. This article considers three issues: firstly, what determines the prices of stocks and housing? Secondly, how do substantial fluctuations in asset prices affect financial activity? And thirdly, what role should monetary policy play in countering threats to economic and financial stability? Stock pricesLike other assets, stocks are a means to postpone consumption from today to a later date. This allows the investor to determine the most suitable consumption pattern over time. The yield on the asset is the compensation for postponing consumption. To obtain the optimal consumption pattern, a person must, at the expected yield by postponing consumption, be indifferent as to whether consumption takes place now or is postponed. If the investor is not indifferent, the consumption pattern can be changed, and overall affluence improved. The price which the investor is willing to pay for the asset thus depends on the expected yield on the asset, as well as the investor's time preference regarding the consumption pattern (discount factor). Over a company's lifetime a stock's yield is the sum of the dividend in the individual periods. The stock price therefore depends on the sum of the expected future dividend payments, discounted to present value by the discount factor. A mathematical expression of this price is [4]: Pt is the asset price at time t, E(Dt+i) is the expected future dividend payments in current prices, and b is the discount factor of the investor. Since neither the future dividend payments nor the discount factor can be observed, pricing entails certain problems. It is often assumed as an approximation that the actual dividend payments are subject to a constant growth rate and that the discount factor is closely associated with the interest rate. Application of these assumptions gives the following expression of the stock price [5]: g is the nominal growth rate for dividend payments, r is the nominal interest rate and r is the risk premium on holding the stock rather than government bonds. The discount factor is approximated in terms of 1/r. In the financial literature r is also known as the risk-free interest rate, and often the government-bond yield is used. The bond yield can also be interpreted as an opportunity cost of the stock investment. An alternative to buying the stock is to invest, almost without risk, in a government bond. If this yield rises, buying the stock becomes less attractive compared to buying the government bond. The equation, also referred to as the Gordon-Shapiro equation, is very simple, but also widely used. The central element of pricing stocks is that the price depends on the expectations of the company's earnings and the interest rate. The expectations can be excessively pessimistic or optimistic, and can lead to considerable fluctuations in stock prices, with substantial deviation from the fundamental value in certain periods. One of the explanations for non-fundamental fluctuations is that some investors do not act rationally. Examples are herd behaviour, myopic behaviour (e.g. the investor knows that the increases in the market are not sustainable, but he believes that he can buy and sell again before the market turns around) or the application of an incorrect model (e.g. that a high return in the preceding period will be achieved again in this period). However, it can be very difficult to determine the fundamental value. One approach to assessing stock prices is to compare them with the real rate of growth in the economy, cf. Chart 1. In the long run, real stock prices will normally follow the real growth rate in the economy, and thereby also the rate of growth in real income [6]. Since the mid-1990s the rate of increase in real stock prices in Denmark, Germany and the USA has been considerably higher than the rate of growth in real output. Chart 1 Real stock prices and GDP in Denmark, Germany and the USA
Development in the USA Chart 2 shows the development in stock prices as a ratio of business enterprises' earnings per share (price earnings or P/E) and stock prices as a ratio of dividend (P/D). The price earnings ratio is the price an investor has to pay for a share of one unit of the company's profit [7]. Chart 2 Price/earnings ratio (P/E) and price/dividend ratio (P/D) in the USA
The average P/E ratio in the USA for the broad S&P 500 index in the period 1957-2000 is around 16. This entails an E/P (earnings yield) of approximately 6 per cent, which is relatively close to the average real yield for S&P 500 of 6.5 per cent p.a. throughout this period. In practice, the development in the earnings yield for broad stock indices has been relatively close to the average real yield on stocks, and this real yield has been stable for long periods of time. Estimates from the USA show that the real yield on stocks has been around 6.5-7.5 per cent p.a. in various sub-periods for the last 200 years [8]. The price earnings ratio has risen considerably from the level at the end of the 1980s. This should be viewed against the background of an unusually long period of high growth in private consumption and investments, falling unemployment and a relatively low inflation rate in the USA. For the last 2 years the P/E ratio for S&P 500 has fluctuated at around 30 (27 in mid-November 2000), which corresponds to an earnings yield of 3.5 per cent, i.e. only half the historical average real stock yield. The P/E ratio for S&P 500 covers a wide range of values for the individual companies, where the P/E ratio of certain of them is higher than 50. The stock price of certain companies in the information technology sector (computer technology and communications) is very high. The NASDAQ index, which primarily comprises technology enterprises, has shown a P/E ratio of more than 100 in 2000, cf. Chart 3. The ratio has fluctuated considerably in the last two years. This average value also conceals large differences between the individual enterprises. Certain companies have a P/E ratio of more than 300. Unless there has been a significant change in the discount factor, a P/E ratio of 300 implies that the company's earnings must increase very strongly in order to fulfil expectations. An example of how expectations are met is that the company's earnings increase by approximately 40 per cent p.a. for 10 years, and then level out to approximately 6 per cent p.a. in all subsequent years, i.e. close to the nominal growth rate for the rest of the economy. Chart 3 Price/earnings ratio (P/E) for Nasdaq
Viewed in a historical perspective, these values appear to be extreme. However, it is difficult to apply traditional valuation methods to companies of this type, which often have a low capital stock. A large proportion of their value is associated with their organisational structure and human capital, which are difficult to value. On the basis of the Gordon-Shapiro equation an implicit risk premium can be derived, cf. Table 1. The Table shows historical averages and current levels for key economic indicators in the USA and Germany. The yields on 10-year government bonds (deflated by the consumer-price index) are used as the risk-free interest rate of the Gordon-Shapiro equation. An investment horizon of 10 years for investors in stocks is often applied in the financial literature. The real growth in the economy is a fair approximation of the growth in real corporate earnings and dividend payments, assuming that profits and dividends are stable percentages of GDP. The potential rate of growth in the economy is an approximation of investors' expectations of the future rate of growth in corporate earnings. The historical average is limited to the period 1987-99. This relatively short period is chosen primarily on the grounds of data availability. Furthermore, the period chosen was characterised by relatively stable macroeconomic conditions, in contrast to the 1970s and the beginning of the 1980s. Calculating the average over a longer period results in higher risk premiums [9]. Table 1 Implicit risk premium in the USA and Germany
The current stock market levels are historically high. There are at least four interpretations of the results. Firstly, if the current values in the stock markets are sustainable, this could imply that investors accept a considerable permanent reduction in the risk premium. The current implicit risk premium in the USA and Germany is respectively 1.2 and 0.5 percentage points lower than the average for 1987-99, according to the above calculation. The current risk premium is also considerably below the previous averages, viewed over a longer period. It is difficult to say whether a decrease in the risk premium is sustainable in the longer term. Liberalisation and integration of the financial markets, as well as the financial innovation during the last two decades, may have contributed to a lower risk premium. Investors now have greater opportunity to diversify and hedge risk, as well as easier and cheaper access to the financial markets via e.g. mutual funds and the Internet. The risk premium can also be influenced by demographic trends. Higher pension savings in the 1990s by large year-groups can increase demand for stocks, and lead to a reduction of the risk premium. This mechanism will be reinforced if pension funds in the long term increase the proportion of stocks in their portfolios. An improvement in the macroeconomic conditions may also affect the risk premium. The return to low inflation, balance in government budgets and a reduction of public debt may also have contributed to a lower risk premium. However, these factors can only lead to a permanently lower risk premium to the extent that investors expect permanent improvements. It is not yet clear, however, whether these effects are sufficient to justify the current level of stock prices. The risk premium has tended to fluctuate with the business cycle, so that increasing risk premiums can be expected in the event of a downturn in the economy. Another interpretation is to assume that the risk premium has not been reduced, but that the current values reflect expectations of future higher earnings growth rates and dividend payments. If this is the case, the rate of growth in dividend payments must be respectively 1.2 and 0.5 percentage points higher than the potential real growth in the USA and Germany, according to the calculations in the table. The ratio of corporate earnings and dividend payments to the overall economy has been seen to rise in the short term, but has been remarkably stable in the long term. The stability of this ratio is normally regarded as a well-established element of economic growth theories [10]. If this applies, the potential growth rate in the economy must rise to the level corresponding to the expectations of growth in earnings, or the growth in earnings will be below current expectations. This indicates a third interpretation. The current estimates of the potential real growth rate in the economies may be too low. Large-scale investments in networks and information technology in recent years can improve production processes, stock management and distribution, and enhance competition and productivity. If new technologies increase productivity, the potential real growth rates are currently undervalued. Productivity rose by 2.6 per cent p.a. on average in 1995-99 in the USA after a prolonged period of relatively low productivity growth at 1.7 per cent p.a. in 1972-95. However, it appears from an estimate of the contributions to the increase in productivity growth that half of the increase can be attributed to adjustments to the statistical methodology and normal cyclical effects, while the other half relates solely to the computer sector, which accounts for only 1.2 per cent of the US economy [11]. However, the introduction of new technology may require changes in production processes and supplementary training, so that the effect is apparent from the data after a certain time lag. The increase in the productivity growth rate in the USA has occurred at a relatively late stage of the expansion, which is unusual. Previous periods have shown that there is no natural connection between technological progress and strong increases in stock prices. The advance of electricity-based industries in the 1920s resulted in considerable productivity increases, but did not entail the strong rises in stock prices seen for IT stocks in recent years [12]. A fourth interpretation of the results is that the current values in the stock market are excessively high and that the stock markets will see a negative correction. A correction of the broad stock indices does not rule out the possibility that some business enterprises will be able to fulfil the current high expectations. The development in Denmark Chart 4 Stock prices in Denmark, Germany and the USA, 1995-2000
As in the USA, the P/E ratio for the broad stock index in Denmark is relatively high at present. The average P/E ratio was approximately 18 in the period 1987-99, while it was around 30 in mid-November. This corresponds to an earnings yield of approximately 3.5 per cent and implies an annual real yield on stocks of the same magnitude. This would be a considerable reduction from the historical averages. Since 1980 the average annual real yield, including dividend, from the total index has been around 13 per cent. Including the 1970s, the average is approximately 9 per cent p.a. Table 2 illustrates that the pricing can vary considerably among the individual sectors. Furthermore, the overall figures reflect a relatively wide spread between individual shares, of which few have a P/E ratio that exceeds 200. Table 2 P/E and D/p for the Danish Stock Market, October 2000
Calculation of the risk premium for Denmark by means of the Gordon-Shapiro equation, as in Table 1 for the USA and Germany, gives a negative risk premium for the period 1987-99. This result appears implausible. One reason may be that the equation is a simplified approximation to "true" values. For example, the equation does not take the taxation factor into account. The pension fund tax which was adopted in Denmark in 1983 did not include yields on stocks, and thus favoured investments in stocks rather than bonds. This may have reduced the pre-tax risk premium for stock investments. The average dividend yield in this period was approximately 1.5 per cent p.a. in Denmark, which is somewhat lower than in the USA and Germany. This can be attributable to a greater tendency for Danish companies to retain dividends during this period. Variations in corporate dividend policies among various countries can be affected by such factors as how disbursed dividends are taxed compared to retained dividends. No value is given to retained dividends in the Gordon-Shapiro equation. Surveys over a longer period have estimated a positive risk premium in Denmark [14]. to be continued on next page Footnotes[1] Michael Bordo and Barry Eichengreen, Is the Crisis Problem Growing More Severe?, presented at the conference Asset Markets and Monetary Policy, Sveriges Riksbank, Stockholm, 16-17 June 2000. [2] Inter alia to comply with the international foreign-exchange agreements (the Plaza and Louvre Accords). [3] The major fluctuations in asset prices in a number of countries, including Denmark, during this period are described in C. Borio, N. Kennedy and S. Prows, Exploring Aggregate Asset Price Fluctuations Across Countries – measurement, determinants and monetary policy implications, BIS Economic Papers, 1994. [4] When the time horizon is infinite, capital gains are zero (the transversality condition). If a company has a policy of retaining dividend, e.g. if distributed dividend is subject to higher taxation than retained dividend, a price based solely on dividend will be undervalued. In that case, the profit for the individual periods will be a more correct variable than dividend. A dividend policy leaning towards retained dividend is often pursued by newly-established companies. [5] From Myron J. Gordon, The Investment, Financing and Valuation of the Corporation, Irwin, 1962. [6] If the earnings and dividend payments of the companies included in the stock index are a stable proportion of GDP. [7] The P/E ratio can be shown theoretically by a simple adjustment of the Gordon-Shapiro equation where it is assumed that dividend disbursements constitute a stable proportion, , of earnings E (D=E):
[8] Jeremy Siegel, Stocks for the Long-Run, McGraw-Hill, 1998. [9] See e.g. Olivier Blanchard, Movements in the Equity Premium, Brookings Papers on Economic Activity: 2, 1993 and Sushil B. Wadhwani, The US Stock Market and The Global Economic Crisis, Special Paper No. 110, Financial Markets Group, London School of Economics, 1999. [10] See e.g. Olivier Blanchard, The Medium Run, Brookings Papers on Economic Activity: 2, 1997 and Robert Barro and Xavier Sala-i-Martin, Economic Growth, McGraw-Hill, 1995. [11] Robert Gordon, Has the 'New Economy' Rendered the Productivity Slowdown Obsolete?, Northwestern University Working Paper, 1999. [12] Nicholas Crafts, Globalization and Growth in the Twentieth Century, IMF Working Paper 00/44, 2000. [13] In empirical research corporate earnings and bond yields are robust fundamental variables to explain the development in stock prices, see e.g. Jan Overgaard Olesen, A Simple Explanation of Stock Price Behaviour in the Long Run: Evidence for Denmark, EPRU Working Paper Series, 09, University of Copenhagen, 2000. [14] Tom Engsted and Carsten Tanggaard, The Risk Premium on Danish Stocks, (in Danish) Nationaløkonomisk Tidsskrift, 1999, estimates the risk premium at 3.7 per cent on average in the period 1922-96. |
|
|
|
Version 1.0 December 2000 Nationalbanken. |
