Stress testing is becoming an evermore critical aspect of portfolio risk management. As outlined in our recent Cyclical Outlook, “Synching Lower,” we believe investors should prepare for a synchronized global slowdown amid tighter financial conditions and increased political and economic uncertainties. At the same time, we foresee increasing economic divergence and differentiation between and within asset classes. Stress testing offers a way to help identify and assess myriad risks.
Stress testing is a basic building block of prudent portfolio and risk management, but PIMCO has evolved a distinct approach. We seek to understand how a variety of market shocks and scenarios may affect portfolio returns and sensitivities. We explore the ramifications of isolated market factors and the impact of nonlinear exposures such as options, which may be difficult to quantify using traditional risk factors. Stress testing also plays an important role in our liquidity management process, as Courtney Garcia and Sudi Mariappa detailed last October in “PIMCO’s Rigorous Approach to Managing Portfolio Liquidity Risk.”
History plays a role in defining scenarios. It provides perspective on the magnitude of potential shocks and market correlations. However, historical scenarios are not sufficient. Today’s initial conditions, market dynamics and economic and geopolitical relationships are unique. Therefore, we place a much greater emphasis on forward-looking scenarios.
Scenario construction begins with discussions within the risk management team and PIMCO’s Investment Committee. We outline potential but plausible macroeconomic events over the coming six to 12 months. They typically fall into two groups:
- Event-driven, such as Brexit, a breakup of the eurozone, elections and trade conflicts
- Economics-driven, typically mapped to one of four quadrants in a 2x2 matrix defined by above- or below-market expectations for growth and inflation
Next, we determine how these scenarios would affect various market factors. We gather input from the risk management team, senior portfolio managers and sector specialists to identify specific risk-factor shocks. These are entered into PROTEUS, a proprietary global risk model designed by PIMCO’s portfolio analytics team. The stress testing engine leverages the correlations and covariances embedded in PROTEUS to generate additional shocks across factors a portfolio may have exposure to in order to produce a comprehensive simulation of market movements across regions and asset classes (see sidebar).
Given the breadth of coverage in the PROTEUS risk model with over 5,000 market factors, the creation of a new scenario is sometimes an iterative process. On occasion, the shocks estimated from the model may not be intuitive or in line with user expectations. When these instances arise, additional factors may be added to the scenario to improve the model estimates.
For sector-specific strategies, which include a number of bottom-up relative value or basis trades, risk managers and portfolio managers develop more granular stress tests that better assess potential risks to performance. For example, in credit portfolios PIMCO runs stress tests that isolate cash versus credit default swap basis; in emerging market portfolios, our stress tests simulate a breakdown in regional correlations.
We also use stress tests to assess the degree of convexity or nonlinearity in a portfolio. This could entail modeling portfolios being shocked by, say, moves in interest rates of 25-basis-point increments or swings in equity markets of 5%, 10%, 15%, etc.
With each shock, our models reprice instruments such as options, agency mortgage-backed securities and callable bonds. This process can identify risks that may not be obvious from factor exposures alone. It also helps determine “breaking points” at which a seemingly well-behaved portfolio may experience significant drawdowns.
Although stress tests provide meaningful insights, the bigger benefit comes from applying this knowledge in the investment process.
For instance, one approach compares the results for each scenario, especially those deemed more likely by the Investment Committee, to the intended risk and return profile for each portfolio. If a given scenario represents a multi-sigma drawdown for a particular strategy, then it is flagged for further escalation and review, beginning with risk specialists and continuing to portfolio managers and PIMCO’s chief investment officers or the Investment Committee. They assess what exposures and positions may cause the potential drawdowns and discuss the best way to mitigate these risks. Although PIMCO does not apply explicit drawdown limits on the basis of stress test results, we often make adjustments in positioning and trade sizes.
We also use stress tests to monitor consistency across accounts within a given strategy. Due to a variety of factors such as the timing of cash flows, client or fund guidelines and portfolio manager tilts, multiple portfolios within a strategy may have moderate differences in exposures. To help ensure consistency and control dispersion within a strategy, we consider a series of model portfolio targets and ranges based on guidance from the Investment Committee.
PROTEUS is a mythological Greek sea-god who continuously changed form while wrestling the hero Menelaus. We named our risk model architecture after Proteus as a reminder of the ever-changing nature of the risks in financial markets (and out of sympathy for those who would find Menelaus hard to pronounce).
PROTEUS provides integrated risk measurement across rates, credit spread, agency and non-agency spreads, emerging market spreads, equities, commodities and currencies. It produces the firm’s ex ante portfolio risk, tracking error and value-at- risk measures. PROTEUS addresses the dual goals of providing robust, intuitive risk reporting and flexible, detailed factor structures that comprehensively identify investment risks to which PIMCO accounts are exposed. The system produces risk reporting at the top level (i.e., account, index and account-index-relative) as well as sector, asset class and security levels. It helps PIMCO understand risk at broad scales and in terms of individual positions or trade ideas.
If the risk factors and concentration limits of a particular strategy’s portfolios are aligned, then one would expect similar stress test results. This is the case in the vast majority of instances. However, on occasion our forecasts signal an outlier. In these cases, PIMCO’s portfolio risk management team investigates to identify the cause of the potential dispersion, and uses the findings to further enhance the risk management process.
Finally, stress tests may inform the investment process following sharp moves in the market. Given the breadth of input PIMCO specialists can provide, the scenarios can serve as ex ante market forecasts; they also help identify sectors that may have over- or under-reacted.
Over the past several years, forward-looking stress tests have been key to PIMCO’s navigation of multiple periods of market-induced volatility – including China’s surprise currency devaluation in August 2015, the collapse of oil prices in late 2015 and early 2016, the Brexit vote in 2016, and multiple bouts of eurozone market volatility stemming from the French elections in 2017 and the recent elections and budgetary issues in Italy.
Stress tests have been, and remain, a critical input into PIMCO’s investment process. Consistent with our team approach, we combine insights from the risk, portfolio management and analytics teams to help deliver better risk-adjusted returns for clients.