What it means
Every business expects a certain level of loss in normal trading, and that expected loss should already be priced into products and provided for in the accounts. Economic capital covers what sits beyond that: the losses that arrive in a genuinely bad year and that no amount of ordinary pricing has paid for.
The calculation starts by modelling the distribution of possible losses over a set horizon, typically one year. Management then picks a confidence level, often tied to the credit rating it wants to hold, and reads off the loss at that point in the distribution.
Economic capital is the difference between that severe loss and the expected loss. This differs from regulatory capital, which is the minimum a supervisor requires using standardised rules.
Regulatory capital is a floor set by an outsider, while economic capital is an internal view that may be higher or lower and that reflects the actual mix of business being run. The practical value is in allocation.
Once you can express credit risk, market risk and operational risk in the same unit, you can compare a mortgage book with a trading desk and ask which earns more return per dollar of capital consumed. That comparison drives pricing, bonus pools and decisions about which businesses to grow or exit.
The obvious weakness is model dependence. A model calibrated on a calm decade will understate the tail, and correlations between risks tend to rise in a crisis exactly when diversification benefits were assumed to help.
Sensible institutions therefore run stress tests alongside the model and treat the output as one input to judgement rather than an answer.
In practice
Real-world examples
Example 1
A commercial bank compares two divisions that each earn $12,000,000 of profit. Corporate lending consumes $80,000,000 of economic capital while wealth management consumes $25,000,000, so the wealth business earns a far higher return on the capital it ties up and receives the growth budget.
Example 2
A general insurer models a one in two hundred year catastrophe scenario across its property book and finds it needs $340,000,000 of capital to survive it. Rather than raise new equity, it buys reinsurance that transfers part of the tail risk and reduces the requirement.
Example 3
A trading desk is charged an internal capital cost based on the economic capital it consumes. When a strategy's returns fall below that charge, the desk closes the position even though it is still nominally profitable.
Think of it
Economic capital is like the extra fuel and emergency supplies a pilot carries on a flight. You do not expect to use it on a normal journey, but you need it on board to survive unexpected extreme turbulence.
Formula
Calculation
Economic Capital = Loss at the chosen confidence level - Expected Loss
Consider a lender with a loan portfolio of $500,000,000. Its models suggest an expected loss of 1.2% of the portfolio each year, and a worst-case loss of 5.0% at the 99.9% confidence level, meaning a loss that severe should occur roughly once in a thousand years.
Expected loss = 1.2% x $500,000,000 = $6,000,000.
Loss at the 99.9% confidence level = 5.0% x $500,000,000 = $25,000,000.
Economic capital = $25,000,000 - $6,000,000 = $19,000,000.
That is 3.8% of the portfolio, since $19,000,000 / $500,000,000 = 0.038. If the portfolio earns $32,300,000 of net income after expected losses, the return on economic capital is $32,300,000 / $19,000,000 = 1.7, or 170%, which management would compare against the equivalent figure for its other business lines.Case study
Seen in the real world
Merrow Union Bank is an entirely fictional institution used here as an illustrative example. It held regulatory capital comfortably above the minimum and its board assumed this meant the bank was well protected.
An internal economic capital exercise told a different story. Because the loan book was concentrated in a single regional property market, the model showed that a severe downturn could produce losses of $210,000,000 against expected losses of $34,000,000, implying economic capital of $176,000,000, well above the regulatory figure the board had been watching. The standardised rules simply did not capture how correlated the exposures were.
Merrow Union responded over two years by capping new lending in that region, selling a portfolio of the most concentrated exposures, and setting internal pricing so that loans adding to the concentration were charged more. The illustrative moral is that meeting the regulatory minimum and being adequately capitalised for your own risk profile are not the same thing.
Watch out
Common mistakes
- Confusing economic capital with regulatory capital. One is an internal, risk-sensitive estimate and the other is an externally imposed minimum, and a business can satisfy one while being short on the other.
- Double counting expected losses. Expected loss belongs in pricing and provisions, so including it in the capital figure overstates what is genuinely needed for the unexpected.
- Trusting the diversification benefit in the model without question. Correlations between risk types tend to rise sharply in a crisis, which is precisely when the assumed offset is needed.
Questions
People also ask
Which confidence level should be used?
It usually reflects the credit rating the organisation targets, with 99.9% over one year being common for banks aiming at a strong investment grade rating.
Does economic capital apply outside financial services?
The full modelling machinery rarely does, but the underlying idea of sizing a cash buffer against a severe rather than an average year is useful for any business with volatile earnings.
How does it connect to pricing?
If a product consumes a lot of economic capital, it must earn a wider margin to deliver the same return to shareholders, so the figure feeds directly into how deals are priced.