Analytics beyond patterns
All successful Financial Service providers are exploring analytics to manage business. An important aspect of analytics is to use historic patterns to determine future trends. Thus the knowledge of fraud patterns can be used to strengthen fraud management.
Regulations changing risk analytics needs
Maintaining an organization or individual risk score card was once confined to larger organizations. Bureaus filled the gap by offering a unified approach to scoring based on behavioral attributes, which appealed to smaller firms. The evolution of business intelligence and analytical tools has helped smaller organizations adopt analytical insights and risk-grid-based pricing to reduce their risk exposure. What was once a competitive advantage is becoming a regulatory requirement thanks to the Basel I-II-III roadmap, which is forging the convergence of governance, regulation and technology?
Currently predictive analytics in risk models has gone beyond individual customer behavior and needs – and can now create certainty rather than probable scenarios when predicting macroeconomic conditions or capitalization requirements as an essential component of the ICAAP (Internal Capital Adequacy Assessment Process).
Assessing capital adequacy is a key challenge that needs predictive analytics. Even if risk and capital requirements from attributes such as LGD, VAR, PD and portfolio concentration could be appropriately computed by statistical measures, calibrating scenarios based on macroeconomic forecasts could still provide misleading capital targets. Inaccurate economic forecasts and off-track randomness can then create a wide gap between Tier I capital availability and the actual requirement. People are fooled into believing that random things have some meaningful causality. Randomly, economic forecasts defy causal reasoning – making them statistically less accurate. In an increasingly more converged global economy, it’s even more challenging to look beyond a short span of time and logically predict the impact of internal factors.
Quantifying risk for certainty
Some risks, by their very nature are quantifiable, while others are not. Risks such as market risk and credit/country risk are quantifiable. Risks like goodwill/reputation and business risks such as new ventures however, are difficult to quantify. These may require alternative models such as neural networks because these have better classification accuracy.
When in 2008 the economic conditions slid precipitously, how many stress tests effectively sent early warning signals so institutions could hold onto sufficient capital to remain solvent? Over exposure to risk and the following Wall Street death knell required economists to rethink future scenarios with sounder methods. Moreover, not many predictive models had factored in the importance of AAA bond rating in their stress tests. The unforeseen downgrading of the U.S. credit rating provided a severe jolt to the global markets demonstrating little certainty in the way ahead.
Small is beautiful with cloud and open source
Big banks had an unfair advantage against smaller banks due to their early adoption of the Basel requirements. However, developments in cloud computing and solutions for small businesses will address the needs of small banks to allow them to more easily adopt regulations.
By the time the Basel III timeline arrives, enriched risk frameworks will have analytics embedded in them with data warehousing features and computing decision cycles of one-day value at risk (VAR) — and credit spread VAR will become true, real -time VAR. There will also likely be a convergence of embedded analytics with next generation data warehousing and cloud computing to support branch level monitoring of capital adequacy.
Future with certainty
Risk management is becoming increasingly intelligent, and future bank success will depend on advanced predictive capabilities as a key competitive advantage.
Predictive analytics, by looking beyond the historic trends to simulate the near past into the future, is likely to provide the much needed foreknowledge to risk managers. The future will be guided by sophisticated models that optimize capital allocation across portfolios. Increasing usage of Meta-heuristic optimization algorithms will be a key trend in deriving optimal solutions. Further research will explore new algorithms to estimate the lower bounds, new parameters and strategies that can enhance the performance of stress testing, analytical framework and algorithm automation strategies.