Herding phenomenon can also be explained as too many positive feedback loops to cause an oscillation. In investment area, we always preach on the virtue of diversification. Portfolio diversification means when you bet for upside in some products, you should also place protection elsewhere for downside. Investor diversity and heterogeneity is predicated on the notion that market needs both optimists and pessimists. Thus the system should not encourage all investors and their portfolios to behave in the same way. However, too many financial institutions outsourced the development of their internal control systems and the technical models used by their bankers and traders in internal assessment and risk control, including the program trading models that had been widely adopted at an earlier time. Outsourcing of system technologies at such a prevalent scale contributed to high degree of homogeneity in the financial system, which strongly added to pro-cyclicality. For complex financial products, most institutions use models built by a handful of quantitative analysts that get widely adopted throughout the industry. Such models tend to produce similar directional results at the same time when certain conditions prevail. In other words, outcomes from such models are highly correlated. When they are used by the whole financial industry world-wide, asset price boom is made much stronger and bust much more damaging. And due to high synchronizations of market participants' behaviors as a result of using the similar models, systemic risk arises. Regulators should require systemically important financial institutions to complement external pricing models with internally developed capabilities to exercise judgment. In addition, to give issuers of structured products more incentives to better assess their risks, regulators should ask them to retain a meaningful share of the underlying assets on their balance sheets in order to alleviate the myriad of problems associated with the "originate-to-distribute" business model, including moral hazards and fraudulent loan underwritings.

On the users' side of ratings, there is the long-standing moral hazard issue. Various rules have required investment management decisions and risk management practices to be benchmarked on financial instruments attaining certain ratings from the so-called Nationally Recognized Statistical Rating Organizations (NRSRO). This practice has enabled industry practitioners to piggyback on the external ratings and not to worry about the inherent risks once the instruments have achieved the threshold ratings. Over time, the financial industry has become accustomed to the practice and become complacent of the ratings they rely on so heavily. Some market players seem to have forgotten that the ratings are no more than indicators of default probabilities based on past experiences but were never meant to be guarantees for the future. Along with complacency, there is inertia and sloppiness on the part of investment managers to ask tough questions about the inherent risks of instruments sitting in their portfolios. Once problems take place, as we have seen during the current crisis, fingers are pointed to the rating agencies. The institutional users (e.g., the money managers and financial institutions) of credit ratings should be ultimately accountable to their customers and shareholders and should exercise their own judgment of risk, not just outsource risk assessment duties to the rating agencies. To the extent they have to use external expertise, internal and independent judgment has to be deployed as a complement. As a matter of fact, the problem has become so serious that regulators need to encourage financial institutions to enhance internal rating capability to rely less on external ratings, and that central banks and regulators should impose requirement whereby use of external ratings should not exceed 50 percent of business activities, at least for systemically important financial institutions. Internal capabilities should be developed to exercise independent judgment on credit risks at such organizations.

2) Fair value accounting, mark-to-market and mark-to-model

Both IFRS and GAAP require mixed value measurements of different type of assets and liabilities according to their features and the management's intentions of holding them, i.e. assets on the trading book and available-for-sale assets should be measured on the fair value basis, while hold-to-maturity assets, loans and liabilities without an objective fair value should be recorded at historical costs.