American University’s Jeffrey Harris and two co-authors say their novel, more timely statistical approach can be applied to sectors beyond banking
More than 10 years since the global financial market meltdown, regulators and central bankers are confident that a stronger and well capitalized banking system is better able to withstand another major systemic shock. Yet proven measures of systemic risk, and particularly predictive tools that cut through market noise and volatility, remain hard to come by.
Bank Holdings and Systemic Risk, a paper published last year on the Federal Reserve Board website, puts forward what its co-authors claim is a unique and effective statistical approach that can monitor for systemic risk in banking. It can also be utilized for tracking change and risk in other complex sectors such as mutual funds, real estate investment trusts (REITs) and broker-dealer holdings.
The approach is based on what the paper calls a novel “statistical model and estimation framework” that regulators can use to “better assess, in a timely manner, concentrated risk within a bank without having to directly examine bank balance sheets. Moreover, the similarity of bank portfolios indicates interconnectedness, an important measure for the propagation of shocks.”
[...] Plans are under way to improve and expand upon the initial research and validation testing, with an updated paper to be released this summer.
“Many of the regulations we currently have in place in the financial sector somehow miss many of the risks out there in the real world,” says Prof. Harris, who served from September 2017 through May 2018 as the Securities and Exchange Commission's chief economist and director of its Division of Economic and Risk Analysis (DERA).
“If we can shed new light on those financial institutions and the obligations or counterparty obligations that they hold,“ Harris adds, “the better it will be. This paper is a way to add to the multiple dimensions of risk management out there.”
Past Attempts
Academics and official and government bodies such as the Financial Stability Board, the SEC's DERA and U.S. Treasury Office of Financial Research (OFR) have been hard at work on systemic risk indicators and tools. Pre-existing literature on the subject was compiled in a January 2012 paper, the first in the OFR's working paper series, by Massachusetts Institute of Technology professor Andrew Lo and three co-authors. A Financial Stress Index and Financial System Vulnerabilities Monitor are among the monitoring tools subsequently developed and maintained by the OFR.
[...]
Balance-Sheet Estimates
Brunetti, Harris and Mankad propose a new statistical method estimating the portfolio concentration or stock returns on balance sheet within each bank, along with an estimate of the common asset holdings across all banks. The former provides a measure of each bank's asset diversification; the latter, an indication the overall banking system's susceptibility to shocks.
It relies on an analysis of daily inter-bank trades and stock returns for individual banks and across all banks, culled from the e-MID European interbank deposit market, and publicly available stock return data, culled from annual reports and other, more current sources.
What's new about this approach, Harris explains, is that it focuses on the asset side of the balance sheet and identifies the concentration risk within each bank – the degree of concentration in one or a few assets. Other approaches tend to focus on the liability side or on capital adequacy, which is what the MES (marginal expected shortfall) and SRISK systemic risk monitoring approaches tend to do.
Faster Readings
Harris and his co-authors, in their paper, describe the asset-based approach as “more timely” and “a robust forecasting tool.”
They say that their testing indicates that the standard deviation and skewness of their measures generally lead, or are more predictive than, data published by the European Central Bank – the Composite Systemic Risk Index, the Simultaneous Default Probability and Sovereign Composite Systemic Risk Index, as well as EU macroeconomic indicators such as the Consumer Confidence Index (CCI) and Purchasing Managers' Index (PMI).
Harris says that risk insights can be produced with greater frequency than with quarterly or annual bank earnings statements.
“Instead of waiting for a quarterly report, you can see the buildup of risks within a bank much earlier,” Harris says. “This allows an auditor or central bank to investigate bank holdings and financial stability before the next quarter comes around,” and before significant risk starts to build up.
One factor contributing to the new formulation is the fact that “we now have access to incrementally better data about the interbank market, and so, any central bank would be able to replicate our approach.” However, Harris cautions that the e-MID does not provide 100% of market data related to the European bank sector, but rather 30% of interbank data in the European sphere. This limitation is overcome with machine learning tools.
Bayesian Framework
As described in the paper, the approach involves eight categories of bank data – cash, commercial loans, intangible assets, interbank assets, residential loans, investments, other holdings and remainder holdings.
A “novel Bayesian estimation framework” utilizes the two sets of data: stock returns and interbank lending data. This allows for the creation of a concentration index, which captures the degree of diversification of each bank's portfolio, and a similarity index, which captures how similar portfolio holdings are across banks.
“We view these two as indicators of system risk,” Harris says.
This data enables a “daily assessment of whether banks are strong or weak and what asset classes may be pushing them into a position of strength or weakness,” Harris says. Regulators may prefer to assess monthly tallies.
Finally, the authors have tested and validated their approach from both a statistical perspective through various simulation exercises, and from an accounting perspective.
Technological Assist
Charles Kane, a senior lecturer at MIT Sloan, says that new approaches for monitoring systemic risk are welcome.
“What the developers of this new approach and others are trying to do now is apply the newest technology to be able to measure risk as quickly as possible,” Kane says. “I applaud any new technology or technology-based approach that can measure the liquidity of banking institutions.” He believes that aside from regulators, credit rating agencies could benefit.
Sam Malone, director of research at Moody's Analytics (not the credit rating business of Moody's Investors Service), says, “This new approach is a good idea because it allows for greater frequency in systemic risk measurement.” He also applauds the use of bank stock returns as a data input, something that Moody's employs in its own Systemic Risk Monitor tool, and the triangulation of this information with interbank trading data.
“When the credit cycle begins to turn, which it soon will in the U.S. and is already happening in China, we will need tools such as these to help us get a handle on what is going on,” Malone says.
European Research
Another recently proposed systemic risk monitor, with a standard set of financial stability indicators, is in “A new financial stability risk index to predict the near-term risk of recession,” a European Central Bank paper by senior financial stability expert Peter Welz and two others.
A 2016 paper from Paolo Giudici of the University of Pavia, Italy, and two others, “The multivariate nature of systemic risk: direct and common exposure,” looks at network structures as a way of identifying systemic risk in the integrated design of financial systems.
Sean Campbell, executive vice president and director of policy research at the Financial Services Forum, which represents the biggest, diversified U.S. financial institutions, believes that a clear definition of systemic risk and its causes – a prerequisite for better monitoring tools – is still lacking.
“We should be more demanding of the regulators,” Campbell insists, “asking them to explain more precisely what they mean by systemic risk. When we talk about bank capital, we know exactly what we mean, and when we talk about volatility, we know the meaning, but in the context of systemic risk, we are still in the early days of understanding.”
Campbell says he does appreciate efforts to produce “a more evidence-based way of assessing for systemic risk.”
Policy Limitations
In a forthcoming American Economic Review article, “Macroprudential Policy: What We've Learned, Don't Know and Need to Do,” Kristin J. Forbes, Jerome and Dorothy Lemelson professor of management at the MIT Sloan School, considers whether policymakers have done enough to prevent the next crisis.
“There are key issues around macroprudential policy about which we do not have sufficient understanding, such as on the new risks generated from the leakages and spillovers, on how to calibrate the different regulations (especially given political incentives), and on the potential risks to financial stability outside the mandates for most macroprudential authorities,” Forbes writes.
MIT's Kane says that no single approach is adequate to this complex task, and we still need to measure the sustainability, credit ratios and liability side of the banks' balance sheets and regulatory policies to realize the overall risk.
The Brunetti-Harris-Mankad framework is “not a silver bullet,” Kane says, noting that it is more focused on commercial banking as opposed to investment banking, where sophisticated, hard-to-measure derivative instruments can be a source of extreme risk.
Leverage and Stress Testing
Malone of Moody's Analytics says one limitation of the asset-based approach – in its initial iteration and testing – is that it does not look closely at leveraged loan activity in the U.S. and the way in which “as an asset class, we have seen a suboptimal amount of crowding,” a trend of concern to systemic risk watchers.
“It would be wonderful if they could extend their assessment and look at the rising risk in the leveraged loan asset class in the U.S., how U.S. banks' exposure is evolving in this area, and how this activity may be interconnected,” Malone says.
He believes that any thorough effort to monitor for systemic risk would require the inclusion of systemic risk metrics with U.S. CCAR regulatory stress tests. “It is curious that we haven't seen that,” Malone says.
Complementary Strengths
Kane says he is concerned about rising risks in the insurance sector, shadow banking, fintech and in cryptocurrency markets, and whether approaches are or are not being developed to measure such risks that may rise to the level of systemic concern.
Harris and colleagues acknowledge their approach's limitations, which they intend to address in future testing and iterations. The initial testing was limited to an analysis of 40 to 60 European banks. Future tests will include larger quantities of data specific to the U.S. banking system.
Harris is optimistic about the new tool's ability to advance systemic risk monitoring within banks, in particular when it is used in combination with approaches that monitor for network effects and interconnectedness.
“Our methods complement other approaches for assessing systemic risk that build on network science techniques,” Harris says, which is important at a time when both banks and regulators consume and analyze massive quantities of data.
“Integrating data from myriad products across various regulated and unregulated markets remains a significant challenge,” Harris says, adding that this new method provides a practical means for assessing complex financial institutions that trade hundreds of financial products in markets around the world.
[...]
Posts
No comments:
Post a Comment