While risk management may make bankers feel better, it does not necessarily make the banking system safer.

Sophisticated risk management systems failed banks during the financial crisis just as they failed JPMorgan Chase last year during its London Whale trading losses. This is surprising given the large investments banks have made in risk management over the past years. It is easy to blame malicious clueless bankers for the problems, but behavioral economics suggests the problem may have more to do with the limitations of risk management itself.

The current physics-based risk models, like value at risk, neglect the human element inherent in their use. They tend to downplay uncertainty with their mechanical attachment to history. Regulatory efforts reinforce the problem. They have converted risk management into a check-the-box compliance function.

Risk is dynamic and does not exist in a vacuum. We adjust our behavior to perceived risk changes. This leads to the dark side of risk management; namely, risk management success encourages unintended, offsetting risky behavior. The perception of safety creates risk whether based on calm markets (the Volatility Paradox) or a belief we can control our environment through risk management (Peltzman Effect).

Faith in risk management motivates bankers to take more risk than they would otherwise assume because they feel they are in control. For example, we tend to drive faster when wearing seat belts. It is difficult to see what could happen when you believe it cannot happen. This is especially true when this bias is reinforced by incentives. Risk management morphs into risk justification when this occurs.

This is arguably what occurred in the JPM London Whale situation. The bank had understandable confidence in its risk management models based on its performance during the financial crisis.Success breeds the disregard of failure. The belief in complex VaR-based, risk-weighted asset measures created an illusion of safety. This supported the creation of a large illiquid structured instruments portfolio. Unfortunately, the black boxes became black holes. The JPM bankers were neither evil nor dumb. Rather, they were just humans falling prey to the behavioral bias of believing they could control the uncontrollable. The bias was amplified by a generous eight-figure incentive compensation package that made declining such investments difficult to resist.

Containing this problem requires fighting complacency as it is easy to assume the future will look like the past after an extended period of success. Reverse stress tests can help in this fight. Traditional stress tests start with a selected scenario and explore the outcome on the bank. These models can be gamed to make the institution seem safe by choosing scenarios that are insufficiently negative to be effective.

Reverse stress tests start with an outcome in which the institution suffers a catastrophic loss, however unlikely, to determine the circumstances under which this could occur. These tests are more of a qualitative exercise that is independent of probabilities and allow banks to prepare for a true worst-case scenario by removing behavioral blinders to see what could actually occur.

The biggest model risk in banking is incentive compensation. It is hard to see something if you are paid not to see it. Incentives reflect an organization's risk culture. A high-performance culture reflected in lavish incentives encourages bankers to take more risks. This is especially true when they believe their risk management models can protect them from the consequences of their actions. Thus, incentives for both line and risk managers must be changed to deliver the intended level of risk by encouraging good behavior.

Finally, effective board oversight is needed to curb the tendency to accept excessive risk in the pursuit of business objectives. The focus should be on risk appetite changes from large and growing asset exposures. The board can nudge management in the desired direction through choice architecture involving monitoring, incentives and sanctions.

Risk is managed by people exercising judgment not models. Judgment requires being aware of the paradox of risk management. This requires identifying behavioral biases and offsetting them through appropriate rules and policy. Such a process places risk management within its proper context as part of the institution and not an independent specialist function, and reduces the unintended consequences of risk management.

J.V. Rizzi is a banking industry consultant and investor. He is also an instructor at Loyola University Chicago.