Menu
Welcome, Tim : view your dashboard at My WisdomTree.

 

Stress Test Methodology

 

Factor-based stress testing methodology

 

Factor-based stress testing and scenario analysis is a technique for calculating and visualizing hypothetical risk and return in a portfolio. Stress tests do not represent actual fund performance and are hypothetical in nature. The stress tests are based on observed historical sensitivities, and there is no guarantee that the assets will perform in a similar manner to the various factors in the future; additionally, while the historical scenarios presented represent observed historical factor shifts, there is no guarantee that the factor shifts in the future will be the same as the historical scenarios. Past performance is not indicative of future results.

 

Please see below for methodologies regarding stress testing.  Additional information regarding the stress testing methodology can be given upon request.

 

Workflow to calculate stress tests

 

  1. Select a set of factors that drives the performance of an asset. The DPD tool utilizes 16 factors; see the Global multi-asset factor set below to explain each asset’s performance variations.
  2. Determine a source for the daily history of those factors.
  3. When necessary, transform the factor from its raw form into one more suitable for regressions.
  4. Measure the sensitivity of each asset to each of the factors by performing a multi-variate linear regression that evaluates the historical total return of the asset against the historical factor set.“Assets” in this situation would be the individual funds in the portfolio being evaluated. The sensitivity could also be defined as the historical beta of an asset to a given factor.
  5. Determine the shift of each of the factors during a stress test scenario. The shift can be based on a historically observed situation, or a hypothetical scenario in which the user themselves (or a third party, such as FinMason) determines the factor shifts.
  6. Multiply the factor shifts for the scenario by the sensitivities of each asset to each factor and sum across factors to calculate the hypothetical return of that asset in that scenario.
  7. Aggregate the performance of each asset up to the portfolio level using its weight in the portfolio.

 

Source Data

 

DPD’s stress testing output relies upon two primary databases of financial data:

 

Thomson Reuters Lipper for Investment Management: This data subscription is used to retrieve historical data necessary to calculating monthly ETF and Mutual Fund-level performance (e.g., Price and NAV, distributions), aggregate constituent composition (e.g., Sector exposures, Country exposures), as well as static reference information such as launch date and market identifiers for more than 25,000 funds domiciled in the US. All of the data is entirely historical and based on the most recent past date available for reporting.

 

FinMason’s “FinRiver” API: WisdomTree has an agreement with analytics firm FinMason to utilize its investment analytics platform to calculate and retrieve several portfolio-level data points for real-time presentation in DPD.  FinMason’s architecture considers millions of securities globally against nearly 2,400 factors across 40,000 economic inputs. FinRiver provides the 16 Factor Model used by the Stress Tests to calculate portfolio returns for historically observed scenarios.

                       

* Note that the FinMason also uses Thomson Reuters Lipper as a data source for its analytics 

 

The Global multi-asset factor set

 

The 16 Factor Model Components

 

The 16 Factor Model is the default factor set FinMason employs in its monte carlo simulations, scenario analysis and associated stress testing. The factors break down into 4 distinct categories:

 

  1. Global Equity factors
  2. Fama French US equity market factors
  3. US Treasury Yield Curve factors
  4. Other factors

Global Equity factors

 

FinMason currently employs five geographically-based broad equity market indices representing the largest national/regional equity markets globally: The five equity factors in the 16 Factor Model are:

 

1. US equity market (SP500): S&P 500 Total Return Index.

2. European equity market (SPEURO350): S&P Euro 350 EUR Index (USD). 

3. Japanese equity market (SPJP150): S&P/Topix 150 Market Index (USD).

4. Chinese equity market (FTSECHINA): FTSE China 50 Index.

5. Emerging Markets equities (MSCIEM). MSCI International EM Net Index (USD).

 

The factor shifts are calculated as the change in the factor observed during each particular scenario, measured as the cumulative percentage change over the course of that scenario. The exact period over which a drawdown or change is measured may differ somewhat across factors. For example, in the crash of 2008 episode, US equity markets did not peak until October of 2007, whereas a number of other markets – most notably US mortgage securities and US money markets – began adjusting much earlier. 

 

Fama French US equity market factors

 

FinMason’s factor set includes benchmark returns for two US equity market factors defined by Eugene F. Fama and Kenneth R. French in their seminal research paper, “Common risk factors in the returns on stocks and bonds,” Journal of Financial Economics (1993):

 

6. Small Minus Big (SMB) - The performance of small stocks relative to big stocks

7. High Minus Low (HML) - The performance of value stocks relative to growth stocks

 

 

These factors are calculated from US corporate financial data obtained from the Thomson Reuters WorldScope Fundamentals database.

 

FinMason calculates benchmark returns to Fama-French size (SMB) and value (HML) factors using the same methodology described in Fama-French (1993) and on the website of Kenneth French. Our dataset includes all firms incorporated in the US with ordinary common equity listed on the NYSE, Amex, or NASDAQ. This methodology can be provided on request.

 

US Treasury Yield Curve factors

 

A core component of FinMason fixed income factors are US Treasury Curve Yields. Constant Maturity Treasury Yields are provided by the Board of Governors of the Federal Reserve System,. FinMason uses several points on the curve as factors:

 

8. 3 month (UST3M)

9. 1 year (UST1)

10. 5 year (UST5)

11. 10 year (UST10)

12. 30 year (UST30)

 

Yields on Treasury nominal securities at “constant maturity” are interpolated by the U.S. Treasury from the daily yield curve for non-inflation-indexed Treasury securities. This curve, which relates the yield on a security to its time to maturity, is based on the closing market bid yields on actively traded Treasury securities in the over-the-counter market.

 

The FED publishes Constant Maturity Treasury Yields as % Yield to Maturity. FinMason converts these interest rates into % return for the purposes of using in the regression models. Essentially, FinMason uses the day-to-day change in Treasury yields, as published by the FED, to estimate the return to a holder of a particular Treasury, had that person held the bond from the close of one day to the close of the next. This methodology can be provided on request.

 

Similar to the methodology for international equity market factors, FinMason uses month-end data to calculate month-on-month yield changes, in basis points, which are then transformed into monthly total returns. Factor shifts for scenarios are calculated in the same manner, using daily data if possible, starting with maximum changes in yields over the course of the scenario in question.

 

Other factors

 

13. US corporate bond yield spread (SPREAD). The US corporate bond spread is represented by the BofA Merrill Lynch US Corporate BBB Index Effective Yield minus the 10-Year Treasury Yield. This index represents a subset of the BofA Merrill Lynch US Corporate Master Index tracking the performance of US dollar denominated investment grate rated corporate debt publicly issued in the US domestic market. The subset includes all securities with a given investment grade rating BBB. FinMason uses the spread, in basis points, between the published effective yield on the BBB Index and the US Treasury 10-Year Note Constant Maturity yield. Both series are sourced from FRED (Federal Reserve Economic Data, published by the Federal Reserve Bank of St. Louis). After calculating the basis point change in the spread over the course of a scenario (or a month), this yield change is transformed into a total return using the same methodology as used for Treasury securities, which represents the actual factor shift used. 

14. Oil Prices (OIL). Wholesale spot crude oil prices for the West Texas Intermediate benchmark traded in Cushing, Oklahoma are sourced from the US Energy Information Administration, published daily.

15. US Dollar Index (USD). The US Dollar Major Currency Index, calculated and published weekly in the H.10 Release by the Federal Reserve Board, is a trade-weighted weighted average of the foreign exchange value of the US dollar against currencies of the Euro Area, Canada, Japan, the UK, Switzerland, Australia, and Sweden.

16. Gold Price (GOLD). Gold prices are represented by the ICE Benchmark Administration Limited (IBA), Gold Fixing Price 3:00 P.M. (London time) in London Bullion Market, based in U.S. Dollars. FinMason obtains these prices from FRED (see above).

 

Creating Stress Test Scenarios

 

The Stress Tests in the DPD tool are based on factor shifts observed during historically observed scenarios. Historical scenarios use the actual observed factor shifts during the specific time period for that scenario . For example, the Financial Crisis stress test uses the actual factor shifts for the 16 factors during the 2008 Global Financial Crisis. These shifts were measured from the market peak to trough time period of October 9, 2007 (peak) to March 9, 2009 (trough) in the case of US equities, and applicable peak/trough dates for the other factors during the 2007-09 time period.

 

The ‘Custom Stress Test’ is a feature allows financial professionals to create their own hypothetical scenarios by selecting their own ‘factor shifts’ described above for the 16-Factor Model, and seeing the resulting calculated return for their portfolio. This capability increases the financial professional’s visibility of how the Stress Tests work, allowing them to test hypotheses about potential scenarios.  Historical stress tests are loadable into the Custom Stress Test capability, making the factor shifts fully transparent to the user.

 

Example Stress Test calculation

 

The following example illustrates how a return is calculated for a portfolio consisting of 40% WisdomTree Barclays Yield Enhanced U.S. Aggregate Bond Fund (AGGY), and 60% WisdomTree Quality Dividend Growth Fund (DGRW), based on the Financial Crisis hypothetical stress test scenario in DPD. Data as of 1/31/2017.

 

Overview

 

Portfolio Return =

(Weight of AGGY in portfolio * Return for AGGY in Financial Crisis scenario [see below]) +

(Weight of DGRW in portfolio * Return for DGRW in Financial Crisis scenario [see below])

 

Return for AGGY in Financial Crisis scenario

 

Return for AGGY =

Constant1  +

Beta_SP500 factor sensitivity * SP500 factor shift +

Beta_USD factor sensitivity * USD factor shift +

Beta_SPREAD factor sensitivity * SPREAD factor shift +

Beta_SMB factor sensitivity * SMB factor shift +

Beta_HML factor sensitivity * HML factor shift +

Beta_UST5 factor sensitivity * UST5 factor shift

 

Factor Sensitivities of AGGY

 

(Note: all other factors sensitivities [betas], were 0 for AGGY)2

 

Factor shifts, Financial Crisis Scenario

 

 

AGGY Return in Financial Crisis Scenario

 

AGGY Return =

 0.00209602

+ 0.135273734 * (-0.552503680)

+ -0.073779153 * (0.245737941) 

+ 0.253071934 * (-0.259426467)

+ 0.025705872 * (-0.002893345)

+ -0.047769047 * (-0.113878308)

+ 0.59528765 * (0.174206510)

          -0.04736 or -4.7%

 

Return for DGRW in Financial Crisis scenario

 

 

Return for DGRW =

Constant1 +

Beta_SP500 factor sensitivity * SP500 factor shift +

Beta_SMB factor sensitivity * SMB factor shift +

Beta_HML factor sensitivity * HML factor shift

 

Factor Sensitivities of DGRW

 

 

(Note: all other factors sensitivities [betas], were 0 for DGRW)2

 

Factor shifts, Financial Crisis Scenario

 

DGRW Return in Financial Crisis Scenario

 

 

DGRW Return =

0.000873002

+ 0.135273734 * (-0.552503680)

+ -0.073779153 * (0.170788678)

+ 0.253071934 * (-0.242499022)

-0.5597 or -56.0%

 

Return to Portfolio (40% AGGY and 60% DGRW), Financial Crisis Scenario

 

The portfolio return is the weighted average of the calculated returns for each fund, based on the DPD user’s weighting for each fund in the portfolio.

 

Portfolio return = 40% * -4.7% + 60% * -56.0% = -35.5%

 

***

 

Additional information regarding the stress testing methodology can be made available upon request.

 


 

 

1 It’s possible that for a given asset some factors do not impact its returns, and therefore have a factor sensitivity for that asset of zero. For example, it is quite possible that a US fund is not influenced by changes in the Chinese stock market (to a statistically significant level), and therefore the FTSECHINA factor sensitivity for that fund would be zero. Each asset’s factor sensitivities are calculated independently of all other assets so will have different betas.

2 The constant, also called the intercept term, is the estimated return on the asset when all of the other factor shifts in the equation are equal to zero.