Please update your bookmarks. Taking Risk is now located at http://www.factset.com/blogs/takingrisk (or www.factset.com/riskblog). All the same great content, now in a new location.
If you subscribe to the blog by e-mail, your subscription will automatically be updated. If you subscribe by RSS, you will need to update the feed address to www.factset.com/blogs/takingrisk/RSS.
Thanks for reading!
Tuesday, December 21, 2010
Monday, November 22, 2010
Global risk models that aren’t skewed by asynchronous trading
by guest blogger Sebastian Ceria, Ph.D., President and CEO, Axioma, Inc.
In a recent abstract, Professor Bernd Scherer warned against the use of unadjusted daily stock market data to update global risk models, a practice he said results in “spuriously low correlations between stock markets.” As a provider of global, regional and multi-country risk models — models that are all updated daily — we couldn’t agree more.
Prof. Scherer correctly illustrates how the “use of daily accounting data would have underestimated the Value at Risk of an equal-weighted portfolio of G7 equity stock markets almost all the time. The use of unadjusted daily data becomes most troubling in periods of market crisis where underestimated correlations suggest a diversification benefit that is not real.
Prof. Scherer goes on to point out that this problem can only be overcome with “data synchronization models.”
Axioma has been acutely aware of this issue since it first began producing risk models, as all of its risk models are fully updated on a daily basis using daily closing return data. To address this issue, Axioma released in May 2010 a proprietary “returns-timing” adjustment methodology that specifically accounts for asynchronous trading between markets.
Most data synchronization models estimate a vector auto-regressive moving average (VARMA) model of returns. Axioma’s synchronization model uses a simple, first order vector auto-regressive (VAR) model. This model relies on just one day's lagged data (the first day is by far the most significant) and provides a robust estimate of the synchronized returns. To date, Axioma has shied away from using moving averages and higher order models as these are more difficult to estimate, require more modeling assumptions, and, in our experience, provide less robust results.
Axioma’s “returns-timing” model allows a number of important variables to be more accurately estimated in our risk models. First and foremost, it corrects the correlation underestimation between assets that trade at different times. Second, the model corrects the specific returns of ADRs and similar instruments whose underlyings trade at different hours than the ADRs. Third, the model allows returns to be decomposed into local market returns and global market returns. The global market returns can also be easily decomposed into industry and sector returns. This allows users to quantitatively assess whether a section of a market, such as banks, moving on one day in one part of the world moved in that same section of the market in a different part of the world later on the same day or on the next day. Finally, the model ensures that factor returns are synchronized, which improves ex-post attribution analyses and helps portfolio managers to understand the factors underlying performance.
For more, download Axioma’s research paper Returns-Timing: A Solution to Asynchronicity.
Receive new blogs by e-mail and follow @FactSet on Twitter.
In a recent abstract, Professor Bernd Scherer warned against the use of unadjusted daily stock market data to update global risk models, a practice he said results in “spuriously low correlations between stock markets.” As a provider of global, regional and multi-country risk models — models that are all updated daily — we couldn’t agree more.
Prof. Scherer correctly illustrates how the “use of daily accounting data would have underestimated the Value at Risk of an equal-weighted portfolio of G7 equity stock markets almost all the time. The use of unadjusted daily data becomes most troubling in periods of market crisis where underestimated correlations suggest a diversification benefit that is not real.
Prof. Scherer goes on to point out that this problem can only be overcome with “data synchronization models.”
Axioma has been acutely aware of this issue since it first began producing risk models, as all of its risk models are fully updated on a daily basis using daily closing return data. To address this issue, Axioma released in May 2010 a proprietary “returns-timing” adjustment methodology that specifically accounts for asynchronous trading between markets.
Most data synchronization models estimate a vector auto-regressive moving average (VARMA) model of returns. Axioma’s synchronization model uses a simple, first order vector auto-regressive (VAR) model. This model relies on just one day's lagged data (the first day is by far the most significant) and provides a robust estimate of the synchronized returns. To date, Axioma has shied away from using moving averages and higher order models as these are more difficult to estimate, require more modeling assumptions, and, in our experience, provide less robust results.
Axioma’s “returns-timing” model allows a number of important variables to be more accurately estimated in our risk models. First and foremost, it corrects the correlation underestimation between assets that trade at different times. Second, the model corrects the specific returns of ADRs and similar instruments whose underlyings trade at different hours than the ADRs. Third, the model allows returns to be decomposed into local market returns and global market returns. The global market returns can also be easily decomposed into industry and sector returns. This allows users to quantitatively assess whether a section of a market, such as banks, moving on one day in one part of the world moved in that same section of the market in a different part of the world later on the same day or on the next day. Finally, the model ensures that factor returns are synchronized, which improves ex-post attribution analyses and helps portfolio managers to understand the factors underlying performance.
For more, download Axioma’s research paper Returns-Timing: A Solution to Asynchronicity.
Receive new blogs by e-mail and follow @FactSet on Twitter.
Tuesday, November 16, 2010
Subscribe to:
Posts (Atom)
Posts
