The rate of return on everything

The return, and the risk beside it

Housing returned 6.48% a year in real terms against 4.43% for equities, at less than half the volatility (10.11 against 22.36 points). Bonds were more volatile than housing and returned almost nothing. That comparison is only honest with both columns present, which is why they sit together here.

Source: Jorda-Schularick-Taylor Macrohistory Database, Release 6 (Jorda, Knoll, Kuvshinov, Schularick & Taylor) Compound annual real return, pooled over every country-year. Volatility is the standard deviation of the annual real return, in percentage points. "Excluding the wars" drops 1914-18 and 1939-45. Pooling weights each country-year equally, so a country with 150 years of data counts for more than one with 121: this is the return of the average country-year, not of a portfolio. Methodology

Housing is not as safe as this table makes it look

The volatility column is measured from national house-price indices, and those are smoothed by construction: houses trade rarely, valuations lean on appraisals and on comparable sales that are already months old, so the index moves less than the underlying asset does. Housing risk is therefore understated here, and by an amount this data cannot measure. The index is also a diversified national portfolio that no actual owner holds: a real household owns one house, in one street, financed with leverage, and pays transaction costs and maintenance that these series do not net out. Read the housing column as the return on the national housing stock, not as the return you would have earned.

What each asset did in a banking crisis

Real returns in the years JST flags as banking crises, against every other year. Equities lose 16.65% in a crisis year; housing still returns 4.14%. This is descriptive. It says what these assets did in years already known to be crises. It is not a claim that any of them can tell you a crisis is coming, and nothing on this page should be read as a forecast.

Source: Jorda-Schularick-Taylor Macrohistory Database, Release 6 (Jorda, Knoll, Kuvshinov, Schularick & Taylor) Crisis years are JST's own crisisJST flag, not the FinObservatory crisis atlas: a window statistic computed off a different chronology than the table it is joined to would not reconcile. The crisis-year count differs by asset because not every crisis year has a return reported for every asset. Methodology

Where the return actually came from

Total return splits exactly into an income leg and a price leg (JST's own identity: total = capital gain + yield, which holds to the decimal in the released data). Housing earns most of its return as rent. Equities earn most of theirs as price appreciation. These legs are nominal, because the point is the share of the return each leg contributed, and deflating both by the same index would not change that share.

Source: Jorda-Schularick-Taylor Macrohistory Database, Release 6 (Jorda, Knoll, Kuvshinov, Schularick & Taylor) Compound annual NOMINAL return. Top eight countries by total return; every country is on its own page. Methodology

Every country

Source: Jorda-Schularick-Taylor Macrohistory Database, Release 6 (Jorda, Knoll, Kuvshinov, Schularick & Taylor) Methodology