“Stocks for the long run?
Evidence from a broad sample of developed markets”
Aizhan Anarkulovaa , Scott Cederburga , Michael S. O’Doherty 2022
,BACKGROUND: ARE STOCKS REALLY SAFE IN THE LONG RUN?
For decades, conventional belief has suggested that stocks are safe for long-term investors. The
historical equity premium (Mehra & Prescott, 1985) in the United States is large, and studies
such as Fama & French (2018) report a very low probability of loss over 20–30 year horizons.
Popular investment advice, therefore, encourages young investors to invest heavily in equities,
assuming that long horizons eliminate risk.
However, long-term losses are particularly harmful. An investor only has one working lifetime to
accumulate retirement savings. If stock markets perform poorly over a 20–30 year period, there
may be little opportunity to recover. Even though U.S. data appear reassuring, relying solely on
U.S. history may provide an overly optimistic view of long-term equity risk.
This paper questions whether the U.S. experience provides a complete and reliable picture of
long-horizon stock market risk.
, WHY U.S. EVIDENCE MAY BE MISLEADING
There are several reasons to be cautious when using only U.S. historical data:
First, the U.S. return history is relatively short. The commonly used CRSP sample covers less
than 100 years, which provides limited statistical information about 30-year investment
horizons.
Second, U.S. data may suffer from survivor bias (Brown et al., 1995) — the U.S. was one of the
most successful economies of the 20th century. Other developed markets experienced wars,
collapses, exchange closures, and even permanent stock market failures.
Third, there is concern about easy data bias (Dimson et al., 2002) — successful markets are
more likely to have continuous return records available. Example: Japan → From 1990 to
2019, Japanese investors experienced −21% real returns over 30 years.
These concerns motivate the need for broader international evidence on long-horizon stock
market risk.
Evidence from a broad sample of developed markets”
Aizhan Anarkulovaa , Scott Cederburga , Michael S. O’Doherty 2022
,BACKGROUND: ARE STOCKS REALLY SAFE IN THE LONG RUN?
For decades, conventional belief has suggested that stocks are safe for long-term investors. The
historical equity premium (Mehra & Prescott, 1985) in the United States is large, and studies
such as Fama & French (2018) report a very low probability of loss over 20–30 year horizons.
Popular investment advice, therefore, encourages young investors to invest heavily in equities,
assuming that long horizons eliminate risk.
However, long-term losses are particularly harmful. An investor only has one working lifetime to
accumulate retirement savings. If stock markets perform poorly over a 20–30 year period, there
may be little opportunity to recover. Even though U.S. data appear reassuring, relying solely on
U.S. history may provide an overly optimistic view of long-term equity risk.
This paper questions whether the U.S. experience provides a complete and reliable picture of
long-horizon stock market risk.
, WHY U.S. EVIDENCE MAY BE MISLEADING
There are several reasons to be cautious when using only U.S. historical data:
First, the U.S. return history is relatively short. The commonly used CRSP sample covers less
than 100 years, which provides limited statistical information about 30-year investment
horizons.
Second, U.S. data may suffer from survivor bias (Brown et al., 1995) — the U.S. was one of the
most successful economies of the 20th century. Other developed markets experienced wars,
collapses, exchange closures, and even permanent stock market failures.
Third, there is concern about easy data bias (Dimson et al., 2002) — successful markets are
more likely to have continuous return records available. Example: Japan → From 1990 to
2019, Japanese investors experienced −21% real returns over 30 years.
These concerns motivate the need for broader international evidence on long-horizon stock
market risk.
