New methodology from MSCI ends the debate as to whether Expected Shortfall can be back-tested
MSCI has developed a methodology to back-test Expected Shortfall, a risk measure that has been proposed by the Basel Committee on Banking Supervision as an alternative to Value at Risk (VAR). Expected Shortfall is defined as an average of losses in excess of a given VAR level.
VAR has been used by banks since 1996 to calculate regulatory capital requirements. In 2012, the Basel Committee proposed as part of its first consultative paper, Fundamental Review of Trading Book Capital Requirements, to change the measurement method for calculating losses to Expected Shortfall because they believe it will better capture the extreme losses that can occur in times of systemic turmoil.
There has always been a concern about using Expected Shortfall, however, because critics claimed it couldn’t be back-tested due to a theoretical debate around its mathematical property called elicitability. Because of this debate, in its second consultative paper in October 2013, the Basel Committee suggested adopting Expected Shortfall to measure risk, but to continue to back-test using VAR.
New groundbreaking research from MSCI solves this dilemma by demonstrating that it is not only possible to back-test Expected Shortfall, but the methodology MSCI has created is a more informative test of model performance than the current VAR back-testing methodology.
In their December 2014 paper, Backtesting Expected Shortfall, Carlo Acerbi, PhD and Executive Director at MSCI and Balazs Szekely, PhD and Senior Researcher at MSCI, propose three back-testing methodologies for Expected Shortfall and observe that elicitability is irrelevant when it comes to the choice of a regulatory standard.
MSCI is a leader in providing risk and performance solutions to the world’s most sophisticated investors. MSCI’s multi-asset class models and tools help investors manage investment risk and meet the demands of regulatory reporting.