The impact of SFAS 157 on fair value accounting and future bank performance

• Published date: 16 July 2020
• Pages: 739-757
• Date: 16 July 2020
• DOI: https://doi.org/10.1108/IJAIM-11-2019-0135
• Author: Dimu Ehalaiye,Mark Tippett,Tony van Zijl
• Subject Matter: Accounting & Finance,Accounting/accountancy,Accounting methods/systems

The impact of SFAS 157 on fair

value accounting and future

bank performance

Dimu Ehalaiye

School of Accountancy, Massey University, Palmerston North, New Zealand

Mark Tippett

Business School, The University of Sydney, Sydney, Australia, and

Tony van Zijl

School of Accounting and Commercial Law, Victoria University of Wellington,

Wellington, New Zealand

Abstract

Purpose –The purpose of this paper is to investigate whether levels-classified fair values of US banks

based on SFAS 157: Fair Value Measurements, as recognised in the quarterly financial statements of the

banks over the period from 2008 until 2015, have predictive valuein relation to the banks’future financial

performance measuredby operating cash flows and earnings over a three-quarter horizonperiod. In addition,

we consider whether the global financial crisis (GFC) impacted the relationship between SFAS 157–based

levels-classifiedfair values and bank future financial performance.

Design/methodology/approach –We develop hypothesesconnecting the net levels-classified bank fair

values based on SFAS 157 with banks’future financial performance. We test the hypothesesby estimating

three-periodquarters’ahead forecasting models. We also use these modelsto test for the impact of the GFC on

the relationshipbetween the fair values and future financial performance.

Findings –Our findings suggest that the levels-classified net fair values based on SFAS 157 have predictive

value in relation to future cash flows for banks. There is significant variation, across the levels, in the predictive

value of levels-classified net fair values for future performance. Our findings indicate that the GFC has limited

impact on the predictive value for cash flows, but the GFC had a significant adverse impact on earnings, and, with

allowance for the effect of the GFC, the Level 2 net fair values have predictive value for the futuree arnings.

Originality/value –The study provides the first direct empirical evidence on the relationship between the

SFAS 157 levels-classified quarterly bank fair values recognised in publicly available financial statements and

banks’future performance. Our results are consistent with the findings from earlier research (Ehalaiye et al., 2017)

using annual data disclosed in the supplementary notes to the financial statements of US banks based on SFAS 107.

The study, makes a significant contribution to the question of frequency of reporting and to the disclosure vs

recognition debate. The studyhas implications for policy makers, regulators and accounting standards setters such

as the Securities and Exchange Commission and the Financial Accounting Standards Board in eva luating the use of

fair value measurement in financial reporting.

Keywords Fair value accounting, Predictive value, Operating earnings, Operating cash flows, SFAS 157

Paper type Research paper

JEL classification –G21, G28, M41

The authors would like to thank the two reviewers, the journal editor, the conference participants at the 2016

Annual Quantitative Accounting Research Symposium and Consortium in Wellington, New Zealand; the

conference participants at the 2016 Asian-Pacific Conference on International Accounting Issues in Maui,

Hawaii, USA and the conference participants at the 2017 Financial Markets and Corporate Governance

Conference in Wellington, New Zealand for their helpful comments. We are also grateful to Massey

University, New Zealand for the research funding provided.

Fair value

accounting

739

Received25 November 2019

Revised22 May 2020

Accepted6 June 2020

InternationalJournal of

Accounting& Information

Management

Vol.28 No. 4, 2020

pp. 739-757

© Emerald Publishing Limited

1834-7649

DOI 10.1108/IJAIM-11-2019-0135

The current issue and full text archive of this journal is available on Emerald Insight at:

https://www.emerald.com/insight/1834-7649.htm

1. Introduction

We investigate whether levels-classified net fair values of assets, based on SFAS 157: Fair

Value Measurements (hereafter SFAS 157), as recognised in the quarterly financial

statements of US banks over the period from 2008 until 2015, have predictive value for

banks’future financial performance. This is a significant issue because the Financial

Accounting Standards Board(FASB) Conceptual Framework identifies the predictive value

of financial information as a relevant metric for the potential inclusion of financial

information in firms’publicly available financial statements (FASB, 2010). We also use the

global financial crisis (GFC) yearsof 2008 to 2009, which coincided with the implementation

of the SFAS 157, to determine the extent to which levels-classified fair values have

predictive value ina periodof extreme market illiquidity and uncertainty.

Before 2008, US banks were required only to disclose the fair value of financial

instruments in the supplementarynotes to the financial statements. However, from 2008 the

requirement has been for recognitionof these instruments in the financialstatements at fair

value and disaggregated according to the three levels of estimation of fair value set out in

SFAS 157. The three levels vary in objectivity, with the most objective approach being for

level 1: exit prices in active markets. The less objective approaches are for Level 2: exit

prices of similar assets and liabilities in active markets,and Level 3: model estimation with

as much use as possible of market informationas the inputs to the estimation. The reporting

of this information quickly led to research to assess its impact.Initially, the studies focussed

on value relevance for shareprices using a model based on Ohlson (1995). For example, Song

et al. (2010), which used quarterly data on US banksfor 2008, found that the value relevance

of Levels 1 and 2 were similar but greater than for level 3. Similarresults were obtained in

Goh et al. (2015).

Subsequent studies have focussed on the predictive value of the disaggregated data on

fair value for future financial performance. Altamuro and Zhang (2013) found that the fair

value of mortgage servicing rights (MSRs) based on managerial inputs (Level 3) better

reflects the cash flow and risk characteristicsof the underlying assets than the fair value of

MSRs based on market inputs (Level 2) suggestingthat managers can use their information

advantage to generate higher quality fair value estimates than by using market inputs,

particularly where the market for the underlyingasset is inactive. Bratten et al. (2016) using

a sample of US bank holding companies,found that fair value adjustments included in other

comprehensive income (OCI) can predict earnings both 1 and2 years ahead.Yao et al. (2018),

using a sample of annual data on non-US banks, found that the nondiscretionary fair value

level 1 assets are positively associatedwith earnings persistence, whereas the Level 2 assets

and Level 3 assets are not associated with earningspersistence.

The use of fair value [1] as a basis of accounting measurement came under heavy

scrutiny and criticism duringthe GFC, as it was claimed that it worsened the financial crisis

due to its negative effect on the net assets of financial institutions, in particular, banks

(American Bankers Association, 2008;Wallison, 2008). Ryan (2008, p. 2) remarked that “[i]t

almost seems that the credit crunch was sent to serve as SFAS 157’s trial by fire”. A key

issue about the implementation of SFAS 157 [2] that has dominated academic debate is

whether the net levels-classified bank fair values based on SFAS 157 were value-relevant,

especially during the GFC period. This concern was strengthened by the argument that

during the GFC there was extreme market illiquidity, so much so that it was difficult to

apply the most objective approach to estimation of fair value (Level 1). This meant that

during the GFC period there was heavy use of the less objective approaches to the

estimation of fair value (Levels 2 and 3). Hence, followingthe period of the GFC, estimation

of the relative merits of the three levels of fair value measurement became a focus for

IJAIM

28,4

740