3. Volatility is not risk: fluctuation versus permanent loss

Risk 1: Measuring risk

Warren Buffett has spent decades repeating that he would rather have a 15% return with bumps along the way than a placid 12%, and that defining risk as volatility strikes him as a category error. Howard Marks, in The Most Important Thing, puts it more usefully: risk is above all the probability of a permanent loss of capital. This module separates the two ideas and identifies the mechanisms by which the first turns into the second.

Two different things going by the same name

We give the name risk to two phenomena that have nothing to do with each other:

• Fluctuation. The market price of something you own goes up and down. It is a quoted figure. If you do not sell, no transaction has taken place and your wealth in euros has changed on the screen and nowhere else.
• Permanent loss. The capital does not come back. It may be because the underlying business deteriorated irreversibly, because you paid a price that was never justified, or because something forced you to sell at the bottom and you were no longer there for the recovery.

The confusion has a simple cause: fluctuation is measured exactly and permanent loss is not. Marks insists on the point: risk is not observable, not even after the fact. That an investor made money does not prove his position was low-risk, only that the bad scenario did not turn up that time.

Three mechanisms that turn fluctuation into definitive loss

The distinction is not philosophical. Fluctuation becomes a real loss when something stops you from waiting. There are three mechanisms, and it is worth keeping them apart because each one is fought differently.

First, leverage. You invest 100 euros of your own and 100 borrowed in an asset. The asset falls 40%: the position is worth 120, the debt is still 100 and your equity has gone from 100 to 20, a fall of 80%. And there is a second-order effect: your margin is now 20 out of 120, that is 16.7%. If your broker requires you to keep 25%, you get a margin call and are forced to sell at exactly the worst point. The asset may recover in full afterwards; you are no longer in it. The fluctuation has been made permanent by contract.

Second, the forced horizon. If the money has an exit date (a deposit on a house, retirement next month, redemptions by fund holders), fluctuation stops being an anecdote on a screen and becomes the actual price at which you are going to liquidate. Whoever does not choose when to sell cannot ignore the quoted price.

Third, giving up. It is the most frequent one and the least often declared. An index falls from 100 to 65 and, in time, comes back to 100. Clara holds on and ends up flat. David sells at 65 and decides to return once the market has 'confirmed' the recovery, say 20% above the low, that is, at 78. His 65 of capital, bought at 78 and valued at 100, ends at 83.3: he has lost 16.7% in a market that finished where it started. Nobody forced him; he forced himself. Morningstar's studies of the gap between a fund's return and the return its average holder actually obtains put that gap at around one percentage point a year, and this is its main cause.

When volatility is not real risk

The other side of this is taught far less. If you own an asset with solid economic value, you have no debt against it, you do not need the money by any given date and you are capable of not looking at the account, a 30% fall is a psychological nuisance, not a destruction of wealth. Better still: if the business has not changed, that lower price improves your expected return. Marks sums it up by saying that risk enters a portfolio through the price paid, not through the chart that comes afterwards.

This has an uncomfortable consequence for quantitative analysis: the same asset carries a different risk depending on who owns it, because the three mechanisms above describe the investor, not the instrument. No fund factsheet can capture that.

The opposite error: low volatility does not mean low risk

If high volatility does not imply risk, low volatility does not imply safety either, and that is the more expensive of the two mistakes.

• Assets that are not priced daily (property, private equity, private credit) report ridiculously low volatilities, largely because their valuation is a periodic, smoothed estimate rather than a market price. The absence of a price is not the absence of risk.
• The negatively skewed strategies of the previous module show little volatility right up to the day they show all of it at once.
• Fraud produces the calmest series of all. Bernard Madoff's scheme reported returns of around one percent a month for years with almost no negative months, and that impossible serenity was the alarm signal that almost nobody read.

Case study: Silicon Valley Bank, 2023

Silicon Valley Bank had invested a huge share of its deposits in Treasury bonds and agency mortgage-backed securities, assets with no appreciable default risk and classified for the most part as held to maturity. When rates rose in 2022, the price of those bonds fell and the bank built up unrealized losses of the order of 15 billion dollars, comparable to the whole of its own equity. On paper, that was pure fluctuation: if it waited until maturity, it would collect the face value in full.

It never got to wait. Deposit outflows forced it to sell around 21 billion of its available-for-sale portfolio, crystallizing a loss of roughly 1.8 billion. The announcement triggered a mass withdrawal and the bank was taken over in March 2023, in what was then the second largest bank failure in United States history.

The bonds did not fail: they paid everything promised to whoever was able to wait. What failed was the assumption that waiting was possible. That is exactly the border between fluctuating and losing.