Why is a measured correlation not the same thing as a hedge? This note separates the statistic from the position, using the periods when U.S. stocks and long-term Treasuries have moved together as the public example.
A statistic describes a window
Correlation says how two series moved together over a window someone chose. A hedge is a position intended to offset a loss in another position in the scenario the holder cares about. The statistic can inform the intention. It cannot be the intention. A correlation estimated on ordinary days may be silent about the day the holder actually needed the offset. Calling the statistic a hedge promotes a description of the past into a promise about the scenario that was left out of the window.
The sign can flip. Two series that moved in opposite directions for years can move in the same direction when the thing that hits both is the same thing, such as a rise in the discount rate applied to distant cash flows. The historical opposite sign was real. It was also conditional. A hedge that depends on the sign remaining opposite has imported that condition without writing it down. When the condition leaves, the position that was supposed to offset can add to the loss. The arithmetic is ordinary. The surprise is only in the name.
A correlation is a summary of how two series moved together in a chosen window. A hedge is a position meant to offset another position in a scenario that has to be named.
Stocks and Treasuries on the same side
U.S. equities and long-term Treasury bonds have, in several public periods including parts of 2022, declined in overlapping windows. The example does not need a private dataset and does not need a claimed precision. It needs only the observation that a relationship many people had summarized as opposite was not opposite in that weather. A portfolio note that said bonds were the hedge for stocks, because a long window of correlation said so, was using a statistic outside the condition that made the statistic true.
None of this says bonds failed as bonds, or stocks failed as stocks. Each did what its cash flows and its buyers produced. The failure is in the sentence that treated one as the automatic offset of the other. A hedge has to name the scenario. If the scenario is a growth scare that sends money toward government bonds, the historical opposite move may be the relevant sample. If the scenario is inflation that pushes yields up and compresses equity valuations together, that sample is the wrong comfort. The 2022-style window is a public reminder to name the scenario before inheriting the correlation.
There is a measurement trap beside the economic one. Correlation depends on the frequency. Daily returns, monthly returns, and overlapping windows do not answer the same question. A hedge that has to work on a particular afternoon is not tested by a monthly correlation. A hedge that has to work over a year is not tested by one violent week. The stock-and-Treasury example can be computed at several frequencies, and the frequencies will not recite one number. Publishing one number is a choice of question. The choice should be visible.
What a hedge statement has to include
A sentence that claims an offset should state the position, the scenario, the window used as evidence, and the frequency. If any of those are missing, the sentence is a correlation wearing a more confident name. It can be rewritten without shame. The rewritten sentence will be longer and less comforting, which is a sign that it has become a description.
Correlation is not a hedge because one is a summary of co-movement and the other is a purpose under a stated scenario. Stocks and Treasuries are the public pair that makes the distinction memorable. The pair will co-move differently in some future window. That possibility is the reason the distinction is worth keeping.
This piece does not claim that bonds cannot offset stocks, that any year will repeat, or that a portfolio should hold a stated mix. It refuses to let a historical co-movement inherit the job of a hedge without a scenario.
