Time and again, I've seen endowments analyse correlations and Sharpe ratios using monthly returns data to design their portfolios. This method relies on the Capital Asset Pricing Model, which assumes that volatility captures risk and returns are normally distributed — both flawed assumptions. Putting aside these obvious flaws, there is another issue that does not get enough attention. Endowments care about long-term returns, not monthly ones. For example, while Microsoft and Tesla may move together in the short-term, due to macro news and the like, their long-term drivers are fundamentally very different. Using monthly data captures short-term correlations financial market that aren't relevant to endowments and overlooks long-term trends.
Notes: In statistical terms, relying on monthly returns is only valid if returns are independently and identically distributed, which they aren't.
This flaw leads to various misguided portfolio decisions, such as overvaluing commodities for their short-term low correlation with public equities despite their poor long-term return potential. Some commodities even have negative real returns over the long term. Portfolio rebalancing doesn't solve this issue, as it requires consistently predicting market conditions to be overweight and underweight at the right times — something few can do reliably. This issue also affects real estate portfolios, where publicly listed real estate is often avoided for its short-term correlation with equities, despite having similar long-term correlations to private real estate. A final example is the idea that there is no alpha in large cap equities. The argument for this is that the dispersion of returns for large-cap managers is lower than for small cap managers or even private equity[1]. However, such arguments usually show annual dispersion. Dispersion in the shorter-term will naturally be related to financial market correlations. However, two large-cap stocks in different industries will have very large dispersion over a decade and this can be a meaningful source of outperformance (or underperformance).
So, what's the solution? We can't replace monthly returns with 10-year returns due to insufficient data (100 years would mean only 10 data points). Instead, we need a qualitative, first-principles approach to understand the true long-term return drivers. This approach reveals that even the concept of an "asset class" is flawed and should be one of many risk-management lenses, not the primary driver of portfolio construction.
[1] Note: All else equal, private equity managers also have greater dispersion because their returns are reported as IRRs not time-weighted returns (”TWRs”). Mathematically, the IRR method drives dispersion relative to TWRs. The two should never be compared.