Stocks and Bonds for the Long Run
Data and text in this post taken from the chapter House Prices, Stock Returns, National Accounts, and the Riksbank Balance Sheet, 1620–2012 in the book Historical Monetary and Financial Statistics for Sweden, Volume II.
- The long-run evolution of Swedish financial market returns over the past one hundred and fifty years
- The short-term risk-free rate of return is presented from 1856 and a representative long-term government bond yield from 1874
- A preliminary version of a new annual stock price and returns index for the period 1870–2012, using previously unexplored evidence of historical stock prices, dividends and equity capital from the last decades of the 19th century
Part 2: Swedish Markets – Stock Returns
“The Swedish bond market emerged in the middle of the 19th century. At first, the most important borrowers on the domestic market were mortgage associations and industrial corporations (Gårdlund, 1942). The Swedish government did issue bonds, but these were floated almost exclusively in foreign markets up until the 1920s. From the interwar period onwards, the Swedish government has been the main borrower and almost all of its loans have been issued to the domestic market.
Secondary bond trading in Sweden has traditionally been conducted outside the organized stock exchange. Investors have traded bonds over the counter at banks or in bilateral block transactions. The Stockholm Stock Exchange has regularly listed bond prices since the late 19th century. Within the Exchange, however, bond trading takes place in several markets, depending on the type of loan. Convertible public sector loans are quoted alongside stocks on the A (main) and O (subsidiary) lists. Premium lottery bonds are traded separately. Finally, there is a retail market – the SOX market – in which bonds are transacted freely. The volume of secondary trading on the Exchange has generally been smaller for bonds than for stocks.
Although bonds were not always traded actively on the Stock Exchange, quoted bond market prices have been published in several listings by the Stock Exchange as well as by stockbrokers and banks. In the postwar period, the Swedish secondary bond market has been dominated by trading in government and mortgage loans. Corporate bonds have been relatively few and are normally held by investors until maturity.
Stock returns and bond prices and yields are reported in both nominal and real terms, using inflation data from various sources. Annual consumer price index data are collected from the Riksbank project on Historical Monetary and Financial Statistics for Sweden, described in Edvinsson and Söderberg (2010). A monthly price index does not exist for the entire period. Frennberg and Hansson (1992a) use the Swedish National Board of Health and Welfare’s cost-of-living index to construct a monthly price index back to 1918.”
“Figure 6.3 presents stock returns on the Swedish stock market since 1901. The series are indexed so that they all equal one in October 1901, the starting date for the stock price index at the reorganized Stockholm Stock Exchange. Several interesting patterns emerge. First, the long-run picture differs quite markedly between prices and returns. Real prices were virtually unchanged throughout the 20th century up to 1980, when they started to increase. In contrast, real returns steadily increased, which emphasizes the historical importance of dividends to Swedish investors.”
“The level of stock returns has not been constant over time. Business cycles and periods of financial and economic crisis have led to fluctuations in corporate profits as well as in investors’ income and wealth returns. This warrants an investigation into the extent to which stock returns in Sweden have varied over time. Table 6.1 provides a set of summary statistics of Swedish stock market returns for different sub-periods. Using the arithmetic average to calculate the mean return since 1901 gives an annual real return of 7.6 per cent. There were, however, some decades when the real market return was much higher than this; in the 1950s it was almost 13 per cent and in the 1980s and 1990s it was 20–25 per cent. In other decades it was lower and in the 1910s and the 1970s the average return was even negative.
Table 6.1 decomposes the stock returns into the price change, or capital gain, component and the dividend income component. The average nominal annual continuously compounded stock return over the entire study period, 1901–2012, is 7.8 per cent. Adjusting for consumer prices almost halves the return to 4.2 per cent. Decomposing the total nominal return of 7.8 per cent, about half comes from nominal capital gains (3.8 per cent) and half from the dividend yield (3.9 per cent). What the table also shows, however, is a considerable variation in stock returns across eras. Investing in the stock market portfolio in either 1930 or 2000, and then selling a decade later, resulted in losses. In contrast, an investment in the portfolio in 1980 yielded an average return of almost 30 per cent each year! It is also noteworthy that almost all of the variation across decades comes from differences in stock prices, whereas dividends have been fairly stable over time (see further on dividends below).
Another way of assessing the variation in stock market returns is to look at the individual months and years when returns were extremely high or low. Table 6.2 provides such evidence by listing the top and bottom ten months (Panel a) and years (Panel b) based on a ranking of all returns during the entire period studied. The list of dates in the table also provides a snapshot of the historically important events in the history of the Swedish stock market.
Looking at the highest returns, it is striking that most of them stem from the period after 1980. Nine out of ten top years and eight out of ten top months occurred during the 1980s, 1990s or the 2000s. This reflects the long boom on the Swedish stock market that occurred after 1980. However, some of the top months represent bounce-backs during recessions. The highest monthly return, on November 1992, occurred in the middle of the 1990s financial crisis but reflects the sudden drop in the Swedish exchange rate after the Riksbank decided to leave the ERM’s fixed exchange rate system. The high returns in single months during the banking crisis of 1921 and 1922 likewise reflect bounce-backs.
Looking at extremely bad months and years, the share of years from earlier historical eras is higher. The worst month on the Swedish stock market over the past 112 years was March 1932, when the industrialist Ivar Keuger committed suicide and his conglomerate collapsed. The other bottom return months also reflect important economic or political events, including collapses of financial firms that mark the beginning of financial crises (November 1907, September 1990, October 2008), global stock market crashes (October and November 1987) or political events such as Germany’s invasion of Denmark and Norway during the Second World War (April 1940). The picture of the lowest yearly returns is similar; they are typically associated with the major economic crises and political turbulence during the historical period under study.”
“The volatility of stock prices is one of the most distinct features of this form of investment, distinguishing it from most other forms of financial investment such as bonds or bank deposits. Figure 6.4 shows the evolution of stock market volatility on the Stockholm Stock Exchange since 1901. Volatility is calculated as the 12-month standard deviation of the nominal capital gain, presented as a rolling window over the entire period. There are three peaks in volatility: the early 1920s (deflation and a banking crisis), the early 1930s (international financial crisis and the Kreuger crash) and the early 1980s (devaluations and stock market boom).
As can be seen from Figure 6.4, stock market volatility varies over time. Determining whether this variability is significant or merely a matter of nuances requires a more systematic take. I therefore applied the estimation methodology for detecting and measuring structural breaks in time series proposed by Bai and Perron (1998, 2003). The results of this estimation are indicated by the horizontal solid line. Stock price volatility on the Swedish stock market did indeed increase significantly from the early 1980s onwards. In this estimation the recent era been the most volatile in the modern history of Swedish stock markets, which may contradict some preconceptions about the role of technological development in stock market volatility. The second most volatile period was the interwar era, which is hardly surprising considering the extreme economic and political turbulence in this period. The quietest periods were the early 1900s and the postwar era up to 1980.”
“As apparent from the analysis above, dividends have been a significant part of total stock returns in Swedish stock markets. Figure 6.5 displays the evolution since 1901 of dividend yields on the Stockholm Stock Exchange, calculated as the annual average dividend level divided by monthly stock prices. Although the monthly variation in the series is considerable, dividend yields follow a fairly clear secular decreasing trend. An application of the Bai and Perron (1998, 2003) time series method for estimating structural breaks revealed two significant breaks in the mean dividend yield. Dividends hovered around five per cent up to the early 1940s; they were somewhat lower, around four per cent, in the postwar period up to around 1980 and then decreased again to a level of around three per cent. It is difficult to tell just what explains the level of dividends. One hypothesis, put forward by Baskin (1988), is that in early financial markets firms used dividends to overcome information asymmetries vis-à-vis stock investors. By making stocks more similar to fixed-interest securities, e.g., bonds, a smaller premium was sufficient when firms acquired external finance.”