NL Long-run investment returns
DUTCH LONG-RUN TOTAL RETURN INDEXES
Lange-termijn beleggingsrendement Nederlandse aandelen, obligaties, geldmarkt.
Index values monthly 1946-2018.12
(last update 14 June 2019)
Prof. Jeremy Siegel’s seminal book Stocks for the Long Run (1994, 2nd ed. 1998) provided long-run total return or wealth indexes to demonstrate the long-run outperformance of stock investments over other investments such as bonds, bills, gold or cash/currency. His work has created a lasting impression and popularized a graph showing long-run total return indexes for various investment classes. It continues to have a strong influence on the investment management industry. The graph on the RHS shows US data until 2002.
I have collected data to present Prof. Siegel’s charts with data for the Netherlands. The long-run U.S. data start in 1800, but for the Netherlands my data start after WW2 (i.e. 1946:12). Some time series exist for the pre-war period, but usually only annual data and/or stock prices without dividends. I am interested in extending series further backwards in time, but the lack of Dutch stock dividend data to create a total return series severely restricts progress on this work. Some students' thesis work on this issue remained somewhat unsatisfactory.
The Dutch historical results confirm that stocks are the best investment for the long run, although, as an investor, you sometimes need a very long investment horizon to ride out the disappointing years -- even decades. We also observe that postwar real returns on the bond market and money market (and, arguably, stock market) were on average close to zero for long periods. Real bond and real money market returns have turned to positive values only after 1980, possibly due to financial market market deregulations at that time.
Gold is not a serious long-run investment vehicle. Long-run average return is close to zero, although in certain periods values are influenced by short-lived price increase (i.e. first oil crisis, second oil crisis, and recent financial crisis).
TABLE: Value 1946:12=10. Trend growth rate is measured by the linear trend fitted to log values (=continuous compounding) and annualized (*1200), using monthly data. A linear trend avoids dependency on specific high or low start and end points in the data. Real value of currency equals inverse of CPI price index.
DATA
The graph shows long-run data for two types of Dutch government interest rate investments. First, the total return (accrued interest plus capital gains) for a hypothetical 10-year constant maturity government bond investment, with rebalancing on a monthly basis. Second, the total return for a 3-month government bill/deposit investment, with monthly rebalancing. Monthly accrued interest equals 1/12 of previous month yield or interest rate. Capital gains are calculated from interest rate changes using a modified-duration approach.
The 10-year Dutch government bond rate is a composite series based on available time series for bonds with (average) remaining maturities close to 10 year (selected from historical CBS, DNB series).
NLGVT10Y: 1947.01-1948.05 NLPERP, 1948.06-1956.12 NL1948-II, 1957.01-1958.12 NL1937, 1959.01-1962.12 NL3NEW, 1963.01-1966.12 NLAVG3, 1967.12-1973.12 NLAVG2, 1974.01-1979.12 NL8-15Y, 1980:01-1984:12 NL3NEW, 1985:01-1985:07 NL5-8Y, 1985:08-1986:04 NL5LNG, from 1986:05 NL10Y (latest 10yr)
This series is close to the DNB long-run composite 10-year government bond series, but the NL3NEW series used by the DNB from 1960 has a maturity that went up to 17.5 years (and down) over time.
The 3-month rate is a composite series based on 3-month Treasury bills, 3-month loans to local authorities and 3-month Euribor. Dutch T-bills were not frequently traded and published prices were mostly based on broker advisory rates. T-bills even disappeared completely for a period during the 1990s. Interest rates on loans to local authorities are a little higher than central government T-bill rates (up to 50bp), but lower than interbank rates (up to 10bp, lower in recent years).
NLGVT3M: 1947.01 - 1954.11 TREASURY PAPER 3-MONTHS, 1954.12-2011.12 LOANS TO LOCAL AUTH 3-MONTHS, from 2012.01 EURIBOR 3-MONTHS.
Stock market returns are calculated from several sources and various indices available for the historical period. Total return series are based partly on daily actual dividends and a proxy based on dividend yield series.
STOCK PRICE INDEX PI: Before 1952.12 monthly average ANP-CBS (spliced), 1952.12-2011.12 end of month ASE All Share price index (spliced CBS, ASE), from 2011.12 AEX headline price index (Note: AEX appears to resemble All Share Index closely, more closely than MSCI AC).
STOCK PRICE TOTAL RETURN INDEX RI including gross dividend reinvestment. 1947-1969 Total Return Index constructed from price index grossed up with dividend yield data (AMRO data and partly estimated). RI = RI(-1)*[PI/PI(-1)+(1/1200)*DY*PI/PI(-1)]. From 1969.12 ASE All Share price return and implied monthly dividends from MSCI Total Return index (Note: MSCI historically based on smoothed dividend yield gross up method). RI = RI(-1)*[PI/PI(-1)+MSCI DIV/ MSCI PI(-1)]. From 1980.01 ASE All Share with daily dividend reinvested. From 2012.01 AEX Total Return Index.
GOLD: IMF, DNB, CBS, Kito.com data on London gold price converted into euro (Historical HFL data converted into euro using fixed conversion rate, 2.20371).
CURRENCY: CPI, spliced historical series for various base years.