Over the past couple of years, we have struggled to find a common language and measurement scheme to manage risk. After countless meetings, interviews with major banks and Fortune 500 corporations, we have arrived at a framework. We would like to spell it out in hopes of testing our structure and soliciting your opinion.
In our enterprise risk committee, we are proposing a "Risk Appetite Statement" to our board that is an expression of desired return and a given level of risk. It is management's job to present a matrix of
potentially obtainable returns each year and the corresponding level of target risk. Upon selection by the Board, management is then charged with maximizing return, while minimizing risk under the constraints. In addition, management also is operating under a maximum level of risk that can be taken for a given amount of capital at risk.
Now, coming up with the return portion of the equation is easy. For simplicity, we are expressing return in a standardized nominal return on equity. While risk-adjusted and excess capital return numbers were considered, we opted for an accounting return measure due to its transparency and experience of the board in working with return estimates.
Coming up with a language to frame and discuss risk was infinitely more difficult. What really mattered to us was the negative risk to capital. That is, while volatility measures (both positive and negative) are a good proxy for risk, we needed something that could easily be tracked and so settled on a strict capital measure that only impacted a downside case.
Once potential negative capital swings was chosen as the measure of choice, we picked a time horizon and a measuring frequency. Here, we chose a 1Y time horizon with quarterly measuring intervals. This decision is still subject to testing, but we wanted to pick a suitably long horizon to measure a material amount of a risk cycle, while not too long as to sacrifice gross accuracy. While a year is much longer than most measuring horizons, we limited potential compounded movement by only measuring projected risk quarterly. Over time, we hope to become more efficient at the process and move to a monthly measuring cycle.
Finally, a methodology had to be chosen. Here we selected not one, but a set of 3 different approaches in order to take the best of all. First, we chose to look at risk utilizing a value at risk statistical model utilizing two different calculation methodologies (variance-covariance and Monte Carlo simulation). This approach gives us an objective statistical valuation of all revenue streams that basically serves to mark every revenue stream to market to the best of our ability. We set a negative side "confidence interval" that basically says that we want to be 95% certain to remain in business one year from now.
In addition to the value at risk model, we also have a scenario model where we are busy developing a set of credit, interest rate, liquidity and market shocks that will be applied to a starting balance sheet and income statement. Incorporating our Loan Pricing Model contained in our BIG Profit application, as well as our Credit Stress and ALM Model to provide inputs and
parameters for our scenario shocks is central to this step. We hope to standardize these set of shocks and use it as a template to measure given risk types.
Finally, we have the panel of the Risk Committee that needs to certify that they are comfortable with the level of risk, the modeling assumptions and model's power. This gives us a subjective "gut" or experience check to highlight risk that cannot be captured in a model.
While we understand our structure has its shortcomings, we know by defining our methodology and monitoring our accuracy, we can improve over time. If you have thoughts or comments on our approach, we would love to hear it, as we look to get some external feedback. In addition, we look to keep you updated to see how the plan develop.