What Is Systemic Risk?
Systemic risk is the danger that a failure spreads through the financial system rather than staying contained in one institution or asset. It cannot be diversified away, because diversification relies on holdings moving independently and in a systemic event correlations converge toward one.
Most risks can be spread out until they stop mattering. This one cannot, and understanding why is the difference between a portfolio that is diversified and one that merely looks it.
How it works
A systemic event is one that propagates. One institution fails, its counterparties take losses, they fail or retrench, and the failure travels through the connections rather than stopping at the first holder.
Diversification is the standard defence against risk and it does not apply here. Spreading money across many holdings works because those holdings move somewhat independently. Systemic events remove the independence.
Correlations converge toward one. In a genuine crisis, assets that normally have little to do with each other fall together, because the reason they are falling is common to all of them — everyone needs cash at the same moment.
So the protection evaporates exactly when it is called on. That is not diversification failing to work; it is diversification working as specified, against a risk it was never designed for.
Systematic and unsystematic
Unsystematic risk is specific to one company. A factory fire, a failed product, a fraud. Hold thirty companies and any one of those becomes a small dent.
Systematic risk belongs to the market itself. Rates, recessions, liquidity. Holding more companies does not reduce it, because every one of them is exposed to the same thing.
And that distinction has a consequence for returns. The market pays you for bearing risk that cannot be diversified away, and does not pay you for bearing risk you could have removed for free — which is the argument for diversifying and the reason concentration is not rewarded.
A worked example
Take this site’s shared series. The deepest drawdown ran 3.76%, 95% of bars sat below a prior peak, and the longest recovery took 73 bars.
Now imagine that drawdown occurring across every holding at once. A portfolio of twenty positions does not experience twenty independent 3.76% drawdowns averaging out — it experiences one, simultaneously, and the diversification contributed nothing.
That is what a systemic event does to the arithmetic. The expected benefit of spreading risk assumes the events are separate draws. When they are the same draw, twenty positions behave like one.
And the correlations measured in calm conditions are the ones people plan with. They are accurate for the period they were measured over and misleading about the period that matters.
What actually protects you
Position size. If a simultaneous decline across everything you hold is survivable, you survive it. That is the only defence that does not depend on holdings behaving differently from one another.
Holding assets with genuinely different mechanics, not merely different labels. Two equity funds in different countries are two equity positions.
And having cash. Not as a return-seeking position, but because the defining feature of a systemic event is that everyone needs cash at once, and whoever already has it is not forced to sell into it.
The original data
On this site’s shared series: deepest drawdown 3.76%, 95% of bars below a prior peak, longest recovery 73 bars, stretch finished +3.61%. A round trip costs 0.0098, about 2% of the median bar range of 0.493.
The 95% figure is the honest backdrop. Being below a prior peak was the ordinary condition, so a portfolio built on the assumption that declines are occasional and offsetting was mispriced from the start — before any systemic event was involved.
Why the connections are the mechanism
Systemic risk is a property of the network, not of any institution in it. A large bank failing in isolation is a large failure; the same bank failing while owing money to twenty others is a different event entirely, and the difference is the connections rather than the size.
That is why the phrase “too big to fail” is slightly wrong. The real criterion is too connected to fail — an institution sitting at the centre of many obligations can be moderate in size and still bring down everything attached to it.
Leverage multiplies the propagation. A borrowed position forced to liquidate sells into a falling market, which pushes prices further down, which forces the next borrower to liquidate. That chain is mechanical rather than informational, and it runs at the speed of margin calls.
And it is why liquidity vanishes in exactly these conditions. The intermediaries who normally stand between buyers and sellers are themselves exposed, so they withdraw — which removes the shock absorber at the moment the shock arrives.
When it fails
The characteristic failure is measuring correlation in calm conditions and sizing against it. The numbers are computed over the available history, most of which is ordinary, so they describe ordinary behaviour. A portfolio sized to those correlations is implicitly assuming the relationship holds in the one scenario it was assembled to survive — and in that scenario the correlations are the first thing to change. The diversification is real right up to the moment it is needed.
A second failure is treating central clearing as elimination. It concentrates counterparty exposure in one institution, which makes that institution systemically important rather than making the risk go away.
A third is holding many funds that hold the same things, which is diversification on the statement and concentration in the portfolio.
A fourth is using leverage against a diversified book, since simultaneous declines and a margin requirement combine badly.
And a fifth is assuming regulation removed it. Rules change where the risk sits; they do not delete the connections that make propagation possible.
Related
Counterparty risk covers the bilateral version that propagation is built from. Risk management covers position sizing, the defence that survives correlation. And diversification covers the protection that stops working here.
Every diversification argument contains a hidden assumption: that the things you hold will behave differently from each other. That assumption is reliable in ordinary conditions and fails precisely in the conditions it was bought for, which is not a detail — it is the entire risk.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.