The rate of return on everything – Comparing investments in housing and equity

What is the real rate of return on investment in the economy? A recent study, The Rate of Return on Everything, analyses data on total real returns on equity, housing, bonds and bills between 1870 and 2015 across 16 advanced economies. This post reviews a few of its key findings, including its analysis on investment returns between safe and risky assets and factors that may bias our comparison of investment returns between different asset classes.

Return on risky and safe assets

The Rate of Return on Everything compares the long run returns and risk premiums of risky assets relative to safe assets. Their findings appear consistent with practitioners such as Joel Greenblatt in that ‘risky’ assets (i.e. housing and equity) appear to yield higher returns at more tolerable volatility relative to ‘safe’ assets in the long run. We encourage you to read the full paper for a proper account of their findings but discuss a few interesting data points below.

Risky assets

The study found the total real returns, in investment income (yield) and capital gains (price changes), for real estate and equities to be similar at an average of about 7% a year between 1870 and 2015. While housing outperformed equities prior to World War II, equities have since outperformed housing on average. However, this is with greater volatility and co-movement with the business cycle. The lower covariance in returns from housing and equity after World War II may point to diversification benefits from holding a mix of asset classes.

Safe assets

Finance theories argue that preferences to consume now than later, risk aversion and covariation with consumption risk mean that investor demand a premium return for investing in riskier assets (i.e. housing and equity) over safer assets (i.e. bonds and bills). Surprisingly, while safe assets (bonds and bills) have yielded a real return between 1% to 3% during peaceful times, the study finds safe asset returns to be very volatile over the long run.

Risk premiums

The findings suggest that risk premiums (the difference in return between risky and safe assets) have varied between 4% to 5% during peacetimes. While risk premiums exhibit high volatility over the long run, the study observes some evidence of mean reversion. More interestingly, temporary increases in risk premiums during wartime and its interwar years were attributed to collapses in safe returns, as opposed to strong increases in risky returns.

Biases in comparing investment returns

When comparing the relative long-run performance of various asset classes, it is important to remember that the indexes and data sets that track such investments are imperfect. Comparing different indexes at face value can lead to incorrect conclusions and poorer investment decisions. The paper in particular highlights biases in weighting, sample selection, survivorship and geography. These factors can have disproportionate effects on reported aggregate returns.

Weighting and selection bias

When comparing return on investment between asset classes, it is important to consider weighting and sample selection. Weighting bias can occur when the weights of an equity index does not correspond to the representative investor within the economy of interest. Selection bias can occur when the choice of stocks in a portfolio or index do not correspond to the portfolio of a representative investor.

Survivorship and missing data bias

Survivorship bias and missing data bias may also affect comparability of returns. Samples that count only surviving or successful firms are likely to overstate investment returns. Likewise, returns that exclude rare disasters (e.g. when markets are closed during wartime) may overstate investment results as well. In housing, properties that decline in investment value are more likely trade less frequently due to loss in liquidity, and hence hold lower weights in indexes. Similarly, such indexes may ignore the destruction of capital during rare events such as wartime or natural disasters (e.g. flood, fires, etc.).

Geography and replication

Geographic coverage and replicability may also affect the comparability of returns. For example, urban areas may bias reporting returns on housing due to the availability of data. Furthermore, what appears less risky or volatile at the national or aggregate level, may not necessarily be replicable at the individual investor level. While the long run rate on return on housing may appear less volatile in aggregate, it is also relatively trickier for the investor to own a diversified portfolio in housing.

Factors that affect return on housing and equity

It is actually quite difficult for the average investor to properly compare the realised net real returns on investment of different asset classes. The paper discusses how transaction costs, leverage and taxation can complicate the relative investment returns between assets very quickly. They are very important investment considerations nonetheless.

Maintenance and transaction costs

We shoould also consider maintenance and transaction costs when comparing returns. For housing, this may include maintenance, repairs, interests, utilities, application fees, management fees and insurance fees. The OECD estimates the transaction costs for housing, across 13 countries of the study’s sample, at 7.7% of the property’s value. Note that this estimate ranges from 1% to 15% if we inspect the cross-section. Note too that the average holding period for housing varies from 10 to 20 years. This lowers the impact of transaction costs on total annual returns.

The study reported that the average transaction cost for stocks at 80 bps (a half bid-ask spread of 30 bps and a commission rate of around 50 bps), based on NYSE data between 1900 and 2001. It also estimated average portfolio turnover at about 60% a year, leading to an annual average equity transaction cost of around 40 bps. While both asset classes face some level of transaction costs, housing returns for individuals may understate a ‘shadow utility cost’, in which investors engage in less optimal capital allocation due to the high upfront transaction costs.

Leverage

Undertaking debt can increase both the risk and return on investments. Today, individual investors can take greater leverage on houses than on equity. However, investors should be aware of embedded leverage positions in their stock holdings since the balance sheets of underlying business are likely to include both debt and equity. We may have to unlever or lever both assets to equivalent levels to compare returns more fairly.

Taxation

While asset returns are typically reported in pre-tax terms, taxes can have very big impact on the net returns we realise. Owner-occupied homes for example enjoy many tax exemptions and advantages. Likewise, the time frame for realisation of capital gains at point-of-sale for stocks and investment properties can have material effect on realised net-return on investment.

Reference

Jordà, Òscar, Katharina Knoll, Dmitry Kuvshinov, Moritz Schularick, and Alan Taylor. “The Rate of Return on Everything, 1870-2015.” NBER Working Paper Series (2017): 24112. Web. Available at < https://economics.harvard.edu/files/economics/files/ms28533.pdf >