FF Capital Partners

Managing Partner: Geraldo Filgueiras, PhD

"This Time is Different" vs. "How It Can Happen Next Time"

Economic data shows how the drivers of crises are of only a few types. Many economists will indeed argue that the "this time is different" perspective never is. Nevertheless from a risk management standpoint, significant losses are almost never repetitive.

We believe the proper perspective is that markets rarely blow up the same way more than once. Thus, part of the work of a risk manager is to understand how it could happen next time.

Furthermore, while difficult to determine, the use of early risk indicators is as old as commerce (e.g., besides the massive liquidity injection by central banks and some factors, a few institutions did much better than others in the 2008-09 period) If we believe we live in a black swam world (see the some of the flaws here), there is nothing to be done (no need for financial professionals except perhaps bank tellers). On the other hand, how should one deal with the political and regulatory risks of a financially repressed, economically recessionary world? (Which assets are more prone to reflect these risks? Which positions to take?)

There exist several early risk indicators some of a more subjective nature (e.g., economic, environmental, geopolitical and societal and technological) others more measurable (e.g., implied volatilities, volume of trading and price gaps). The former tend to play at much longer time horizons and the knowledge about it can be inferred from crowds (see for instance, the World Economic Forum – Global Risks report for 2013). The latter however is more transactional. The key questions for these risks (to be answered by risk managers) are:

How much of the risks are already embedded in prices (i.e., separating noise from information and statistical distributions from risk-aversion)?
Which assets are likely to be impacted? And
By how much should they be affected?
What to do (to optimize positioning)?

Value Added Risk Management

The objective of risk management should be to alert decision makers to the levels and sources of risk and to help them optimize the process of risk taking/reducing. The challenge of financial risk management is to produce and use forward looking analysis that are theoretically sound and market savvy.

There are many ways one can do this but they require market experience, risk science and art. As important as risk models (science and art), the risk management function cannot be performed if processes do not take into account the institutional constraints and the corporate culture surrounding them. That is, risk management must be comprehensive to be actionable.

Strategy and Risk Management

Businesses face at least three levels of risk problems: strategic (what will the competition be doing?), economic (e.g., where are those prices going?) and financial (how much liquidity will I need?). Financial firms face all of them at the same time. Effective financial risk management can frame and solve the problems associated with each of them. (e.g., monetizing relative value and/or properly positioning.)

Depending on one’s portfolio and risk-return targets, one’s risk management may require more or less complex tools. At first, a simple distinction, e.g., linear vs. non linear products, may seem enough of a divisor however -based on our experience- this is often not the case. For instance, large, less liquid positions of linear products may greatly benefit from non linear analyses (e.g., corporate credit risk vs. equity). Understanding the type of problem at hand and knowing how it can be quantified are key to determine the best governance (e.g., limits in VaR, Scenarios & stress test, ES, etc), positioning or hedges to be executed (e.g., forward vs. options vs. structured products).

One other important use of strategic risk management is to understand value across asset classes. We all know that different asset classes have very different liquidity and traded volume transparency but one should use market available information by consistently valuing the different factor components (e.g., bonds vs. cds vs. equities, their funding and embedded leverage elements).

Governance and the Old “New Normal”

With the acute crisis of 08-09 many players (on the buy and sell side) realized the importance of risk management as a function. Risk management is critical to the governance of a business. Depending on the economies of scale and the regulatory environment a firm faces, having an "independent but engaged group" for vetting, overseeing limits and challenging its separate parts adds significant value to the entire business. That is, a well functioning risk management group can add significantly more value than just keeping regulators happy (i.e., it is more than just making sure pillars 1, 2 and 3 are as prescribed).

For instance, those in the structure credit business know that using either the new standard economic capital formula or an approved internal model - accepted by regulators - is likely to crowd out that business. (Note that the standard formula would kill the business faster.) Thus the work of a risk manager in cases like this must be to: (a) prepare and get approval of internal models, (b) understand how the new models and rules will affect this as well as other asset classes and markets and (c) educate and negotiate improvement in the norms.

That is, many are rethinking the business model of those of their businesses most affected by the crisis (of ‘07-‘09). The crisis has brought governance and regulatory requirements as well as market opportunities to most risk-takers in the market place. One sound advice is to "Integrate strategy work with risk management. The planning process should begin with a top-down, risk-appetite-led view of the target balance sheet for the group. Balance sheet usage should then be allocated to divisions on the basis of both nominal and risk-weighted assets". (McKinsey & Co., "The Next Normal" March 2010.)

Risk Measurement and Models

Effective risk management must be guided by a deep understanding of potential P&L (size as well as sources of potential losses) and potentially available actions that can lead to effective action (e.g., pre-locking funding, hedging, positioning). Measurement is vital for this function to be performed. The widely held belief that such measurements and analyses are only useful for the determination of regulatory risk capital and governance is misguided. In fact, this has been proven to be destructive. Some of the very important uses are business performance, risk signals, risk budgeting, etc.

The central inputs to risk models are probability distributions. They are at the realm of any meaningful risk metric but they must be put in context of (risk-return) preferences and the short comings of pricing models

While some simplistic approaches e.g., "notional amounts" can be very useful to convey some issues and only rely on position data, this is often not enough. Factor sensitivity measures - driven by pricing models as well as historical data - are the bread and butter of risk analysis but they too are not enough. Three other critical elements of analysis are future plausible distributions, their relation structure (e.g., the simplistic linear correlation) and position aggregation. The latter may be classified more like a tool but all these elements (pricing model, historical data, probability distributions and aggregation mechanisms) form the core of P&L drivers and of risk measurement. However risk models, as any models, carry simplifications and caveats. Those models need to be carefully chosen, maintained and modified. In fact, this choice of models is what characterizes financial risk measurement very differently from the ones in physical sciences.

While most market practitioners have relied on VaR (and now SVaR as well), this metric has several shortcomings. Indeed, in the case of VaR not only does it carry significant mathematical problems (e.g., not a coherent metric in most cases) it assumes away the effect of liquidity or positional changes. Furthermore, since most VaR implementations use historical data it is supposed to be a tool for normal markets (i.e., no changes in regimes are contemplated). This is not totally useless but far from good enough.

Obviously VaR (and stress tests) are requirements for regulated entities and also, in part because of this, inevitable metrics. However other internal metrics, e.g., scenario based ones, may be much more useful (a clear success story used for margin calculations by clearing houses). The bottom line points are two: (a) while models are necessary and indeed central to risk measurement, the assumptions behind them are more important than their results and (b) since markets are in constant flux, models have to be adaptive to be continuously useful and not outrightly dangerous.