Optimal fair value accounting standards

After studying the two opposite externalities of Level 3 reporting, I shall now conclude on the fair value accounting standards that a social planner maximizing investors’ surplus would set.30 _{To that end, I first state the following corollary that highlights} the optimal fair value standards.

Corollary 4. There is a tradeoff between comparability of financial statements and systemic risk:

• if the congestion effect is not too severe (δ >δ¯), equilibrium (B) is efficient and banning Level 3 reporting is optimal;

• otherwise, if the congestion effect is severe (δ < δ¯), equilibrium (A) is efficient and the optimal accounting standards use both Level 2 and Level 3 reporting. Further, ¯δ decreases in the probability qH to get a high quality asset.

Proof. Direct consequence of Propositions 3 and 4.

Increasing the use of Level 2 reporting may be socially optimal because it would increase the quality of the public information. However, when banks use only Level 2 reporting, the banks with the same type of assets report the same book value, which may lead to simultaneous inefficient liquidations. In other words, systemic risk is the price to pay to increase the consistency and the comparability of financial statements. The objective of standard setters is to increase transparency, whereas the objective of

prudential regulators is to increase financial stability (Barth and Landsman, 2010a). My model underscores that those two objectives are in conflict when the congestion effect is severe (δ <δ¯). In that latter case, a certain degree of opacity is optimal to in- crease financial stability. An interesting analogy can be drawn with Dang et al. (2017), who argue that banks are optimally opaque institutions to avoid runs. Goldstein and Sapra (2014) also accent this tradeoff between transparency and financial stability in the context of banks’ stress tests disclosure.

In normal times, there are more potential buyers for banks’ assets, which implies that banks face less monopsony power (higher δ). On the contrary, during distressed times, potential buyers are financially constrained and banks face more monopsony power (lower δ). As a result, it could well be the case that the efficient equilibrium is (B) under normal economic conditions and it switches to equilibrium (A) when a financial crisis hits. Having flexible accounting standards and letting banks move assets into the Level 3 category during financial collapses may therefore be the optimal policy to increase financial stability. Level 3 reporting then acts as a circuit-breaker and reduces systemic risk caused by mark-to-market. Laux and Leuz (2010) underline that banks were able to use this accounting discretion during the 2007-08 financial crisis. This policy could be more efficient to reduce systemic risk rather than isolating the effect of fair value measures on regulatory capital, which would strengthen the moral hazard problem between banks’ insiders and investors.31 _{I discuss in further} details the policy implications of the model in section 1.5.3.

One way for standard setters to influence the use of Level 3 reporting is to change the Level 3 reporting cost. For instance, accounting standard setters may impose standardized valuation models for Level 3 assets. This would increase the reliability of Level 3 valuations and reduce auditing costs associated to Level 3 reporting. On the other hand, increasing the reporting requirements would increase the cost.32 _Auditing standard setters can also influence the audit cost by requiring more due diligence and disclosure requirements from auditors for Level 3 assets, which would in turn increase audit fees. Increasing auditors’ litigation risks for mispricing would also increase audit fees and reduce the use of Level 3 reporting. For example, Botosan et al. (2011) provide evidence that auditors’ litigation risks limited the use of Level 3 reporting during the 2007-08 financial crisis. In the following corollary, I shall solve for the optimal distribution of the Level 3 reporting that maximizes the investors’ surplus.

31_{For instance by applying a filter on unrealized fair value gains and losses or by allowing financial}

institutions to reclassify assets into the HTM category and use amortized cost (Huizinga and Laeven, 2012).

32_{For example, the FASB is currently proposing to eliminate the disclosure of transfers between}

Level 1 and Level 2 category but they have increased the disclosure requirements for Level 3 (FASB_,

This distribution of the cost is fixed by a social planner at t = 0−_{, anticipating that} capital requirements are fixed at t= 0, based on the banks’ financial statements.

Corollary 5. In order to maximize the investors’ surplus:

• if δ > δ¯, the social planner sets the Level 3 reporting cost to infinity, i.e. bans Level 3 reporting;

• otherwise, if δ <δ¯, the social planner sets the minimum Level 3 reporting cost. Proof. Direct consequence of Corollary 4.

In the first case, when the market’s demand for the assets is sufficiently elastic (δ > δ¯), accounting standard setters and prudential regulators’ objectives coincide and it is then optimal to ban Level 3 reporting. Otherwise, when the congestion effect is severe (δ <δ¯), prudential regulators and standard setters should compromise on the objective. As discussed above, there is a tradeoff between transparency and financial stability. It is optimal to decrease the Level 3 reporting cost as much as possible. An interesting extension of this model, which lies beyond the scope of this paper, would be to consider explicitly two types of regulators, accounting standard setters and prudential regulators, to study their optimal behaviors and the implementation of the optimal fair value standards.33