Sensitivity and Scenario Risk Measures and VaR

If you've ever sat down to a nice dinner (and you probably have at some point) you've had at least a fork, spoon, and knife around your plate. Each serves a different purpose but can work together to accomplish a specific purpose. In the same way, sensitivity, scenario, and VaR risk measures can work together to provide specific risk information about your portfolio.

To start, think about how VaR differs from sensitivity analysis. Which one would provide a probability of a large loss?

You got it!

That's not it.

VaR's the way to go if you're looking for the probability of a large loss, while sensitivity measures tell you how the value changes in response to another value's change. VaR may give you the full picture of risk, but sensitivity measures can drill into the data to reveal the "why" of the estimated loss. So putting them together can yield some great results!

But it wouldn't be so great to combine VaR and scenario risk measures. Why might that be the case?

Exactly! Both measure the estimated loss, so the risk measures overlap, but they take different approaches to get to the same spot. So you can use one approach to offset the weaknesses in the other. For example, VaR has trouble calculating the correlations during significant downturns, and the market volatility might not represent the actual situation. So scenario analysis could be taken from an even greater event to project a worst-case scenario. That way you'd eliminate some of the issues with VaR.

No. That specifically refers to the scenario method.

That's not it. That would be the VaR method.

Similarly, scenario analysis can be based on time periods that might not be accurate for the future. In that case, VaR maintains a disciplined approach with emphasis on all factors. It's a perfect pairing with the weakness of the scenario method.

But sensitivity and scenario methods also have independent strengths and weaknesses. For example, sensitivity measures do a great job of addressing issues in weighting the actual portfolio holdings. Why is that?

That's it! It's any of these reasons. A simple position size doesn't capture any of those risks, while sensitivity analysis can address them all. It also can be used to test an unknown market experience, so it doesn't rely on history. But one thing sensitivity measures can't do is measure risk by volatility because the measures don't use standard deviation or other confidence loss measures. So that's a definite drawback.

Additionally, scenario analysis doesn't need to rely on history to estimate the loss and can be tailored to the most exposed position, so you can understand how that position will drive your risk. Another great feature is that it can take on multiple different distributions, so options are good to go on the scenario analysis. But there are also some negatives, including history. What might cause that to be a drawback?

That's it! The chosen time period might not actually reflect the future, so you need to be careful with a historical scenario analysis. Other disadvantages for hypothetical scenarios include that you could incorrectly assume the correlations of various assets, along with the fact that hypothetical scenarios could become a never-ending guessing game. So it's difficult to cut the analysis at a certain point or choose a specific situation to analyze.

No. Creation isn't needed when it comes to historical periods.

Not quite. The historical time period should reflect the correlation you want to test, so there are no assumptions.

So as you can see, no one risk measurement is the be all and end all to understanding a portfolio's risk. Each one has advantages and disadvantages, but together a picture of risk can come together.

To summarize: [[summary]]

VaR

Sensitivity

Both measure the estimated potential loss

Both measure the estimated loss with volatility and return parameters

Both measure the estimated loss through a use of a hypothetical situation

Position size can't capture interest rate risk

Position size can't capture a multi-asset portfolio

Position size can't assess hedging from options or short positions

The chosen time period might not capture the future

Historical scenarios are difficult to create and maintain

The historical time period might assume an incorrect asset correlation

Continue

Continue

Continue

Continue

Continue

Continue