Is the Minneapolis model the same as the original New York Fed CLASS model that was published in 2016?
The framework we use is a modified version of the 2016 CLASS model. Our changes were made to make it easier for others to replicate our results, because a change in law/reporting/policy required it, or simply to account for the time that has elapsed between the current period and the original CLASS model analysis. A detailed list of the exact changes that were made to the original CLASS model can be found in this article.
Are the scenarios contained in the stress test tool actual forecasts for how the economy is expected to evolve going forward?
No. The scenarios associated with the stress test tool are hypothetical scenarios. They were constructed to help illustrate how adverse economic conditions resulting severe economic recessions could impact the performance of large banking organizations.
Are the losses associated with operational risk or counterparty default calculated dynamically within the Minneapolis model?
The losses resulting from operational risk, counterparty default, and the global market shock to large bank trading portfolios are identical to the ones that were used in the official DFAST 2020 exercise. Alternative projections can be generated by the model that do not include these losses.
How is the level associated with the “capital threshold” determined?
Banks are required to maintain a minimum amount of capital equal to 4.5 percent of total risk-weighted assets. In addition, large organizations that are considered systemically important are required to hold additional capital (known as global systemically important bank, or GSIB, surcharges). The “capital threshold” shown in the chart is equal to one of the following:
- 4.5% if an individual bank is chosen that is not one of the eight systemically important firms;
- 4.5% plus the applicable GSIB surcharge if the firm is one of the 8 systemically important firms (the surcharge is 2.5% for JP Morgan Chase, 2.0% for Citigroup, 1.5% for Bank of America, Goldman Sachs, and Wells Fargo, and 1.0% for Bank of New York Mellon, Morgan Stanley, and State Street);
- 5.8% if the “21 firm total” is selected, which is aggregate amount of capital required by the set of firms to meet the 4.5% statutory minimum of risk-weighed assets plus the applicable GSIB surcharges for the 8 systemically important firms (this amount was roughly $530 billion as of 2020q4).
Does the Minneapolis model forecast that the assets and liabilities of the individual firms will grow or shrink over the projection horizon?
The projections shown on the website assume that each firm will hold a constant balance sheet over the projection period. This means that all assets, liabilities, and risk-weighted assets will not change over the projection period but remain constant at their fourth-quarter 2020 levels. This assumption is consistent with the stress capital buffer rule and mirrors the approach taken by the Fed in the official DFAST exercise.
Alternative projections in which individual firm balance sheets change over time can be produced using the program and related data files contained in this zip file.
The stress test tool only shows projections of total capital. Aren’t the individual components of capital, like net income, loan losses, and so on, also important?
Yes. We will consider ways of displaying this additional information in the future. For now, users of the website can find this information when they use the model’s code found in this zip file.
Are the capital projections lower or higher than those produced by the official DFAST exercises conducted by the Federal Reserve?
Capital levels projected by the Minneapolis model using the official DFAST scenarios are different from those that were generated by the Board of Governors for the official exercises. We would expect to see differences, even large ones, because each exercise uses a different model and approach. Each modeling approach comes with its own pros and cons. It is not obvious, given the nature of these exercises, if the lower losses produced by the Federal Reserve are “correct” or if the higher losses that result from the Minneapolis model are more “accurate.”
The forecasts created by models such as these are uncertain. One approach to address the uncertainty in modeling a relatively rare event like the effects of an economic downturn on large bank conditions is to use multiple modeling approaches. Looking to a variety of models can be particularly useful when uncertainty is especially large, as is the case today.
What is the significance of the “capital threshold”?
A crucial feature of the earlier stress tests conducted by the Federal Reserve was the use of simple “pass/fail” thresholds. These thresholds had the advantage of transparently indicating whether or not a firm would have enough capital to continue operations under negative economic conditions while continuing to pay dividends. We follow that approach in our model. The threshold is equal to the regulatory minimum amount of capital (4.5 percent of total risk-weighted assets) for all non-systemically-important firms. We elected to use the regulatory minimum plus the relevant GSIB-surcharge for the eight systemically important firms because we feel that level is a more appropriate benchmark for this exercise.
The Bank Policy Institute (an advocacy group that represents the nation’s leading banks) raised concerns about the magnitude of losses produced by the Minneapolis model, the plausibility of the macroeconomic scenarios, and the use of the “capital threshold.” Does the Minneapolis Fed need to adjust its approach given these concerns?
No. FAQ #7, #8, and #9 explain why our approach should not change.