The value relevance of inflation-adjusted and historical-cost earnings during hyperinflation

The value relevance of inflation-adjusted and historical-cost earnings during hyperinflation

The Value Relevance of Inflation-Adjusted and Historical-Cost Earnings During Hyperinflation Ran Barniv In Israel, publicly traded companies have bee...

79KB Sizes 70 Downloads 43 Views

The Value Relevance of Inflation-Adjusted and Historical-Cost Earnings During Hyperinflation Ran Barniv

In Israel, publicly traded companies have been required to present financial statements based on the real purchasing power of the Israeli currency since 1985. Supplementary historicalcost data are provided in detailed notes to the financial statements including the balance sheet and the income statement. In the U.S., the FASB (1986) made the requirement to disclose the effect of price changes on earnings optional (SFAS 89). Consequently, publicly traded companies began suppressing inflation-adjusted earnings (IAE) disclosures. Most studies conclude that current cost earnings and constant dollar earnings provided marginal or even no information content in the U.S. This study examines the value relevance of unexpected IAEs and historical-cost earnings (HCE) in the Israeli hyperinflationary environment. A sample of 106 publicly traded manufacturing firms is used. The sample comprises over 97 percent of the publicly traded manufacturers in Israel during the mid-1980s. Cross-sectional annual and pooled regression models are estimated across the triple-digit annual inflation rates from 1984 through 1985 and the double-digit rates from 1986 through 1988. The results show that unexpected IAEs are value-relevant beyond unexpected HCEs, which are not value-relevant to investors in the hyperinflationary Israeli economy. The findings tend to be statistically significant when unanticipated inflation rates are notably high. © 1999 Elsevier Science Inc. All rights reserved. Key Words: Inflation-Adjusted Earnings; historical-cost Earnings; Hyperinflation; Return (Price)-Earnings Relation; International Accounting; Alternative Accounting Procedures

INTRODUCTION During the early 1980s, the U.S. annual inflation rate decreased to less than five percent, which induced the Financial Accounting Standards Board (FASB) to

Ran Barniv ●Department of Accounting, Graduate School of Management, Kent State University, Kent, OH 44242; Phone: (330) 672–2545, ext. 379; E-mail: [email protected]. Journal of International Accounting, Auditing & Taxation, 8(2):269 –287 ISSN: 1061-9518 Copyright © 1999 by Elsevier Science Inc. All rights of reproduction in any form reserved.

270

INTERNATIONAL ACCOUNTING, AUDITING & TAXATION, 8(2)

1999

make price-level disclosures optional (SFAS 89; FASB, 1986). Consequently, publicly traded corporations suppressed inflation-adjusted earnings disclosures. Many other countries, however, experience triple-digit and double-digit inflation rates. Several of these countries require that publicly traded firms disclose accounting information adjusted to changing prices. For instance, most publicly traded companies in Mexico have been required to disclose some inflationary adjustments since 1980.1 Since 1976, Brazil has employed an inflation accounting practice known as monetary correction (Doupnik, Martins, & Barbieri, 1995).2 In Israel, all publicly traded companies have been required to present financial statements based on changes in real purchasing power since 1985. Supplementary historical-cost data are provided in detailed notes which include summaries of the balance sheet, income statement, and details on the adjustments necessary for deriving net income3. Prior to the current requirement, supplemental inflation-adjusted notes were reported during the period 1979 –1984. Inflationadjusted financial statements have been required since triple-digit annual inflation rates prevailed from 1979 through 1985 and double-digit annual inflation rates of less than 20 percent continued from 1986 to 1994. For example, during the period 1984 – 88, annual increases in the consumer price index (CPI) ranged from 445 percent to 16.1 percent. Since the crisis of Summer 1998, volatile economies and business environments in many countries have generated short-term inflationary or deflationary pressures across the globe. The expected shortage in energy sources by 2010, or at the latest, 2020, may yield an increase in long-term inflation rates. Therefore, inflation-adjusted accounting may provide useful information to investors in many countries. Increasing investments in foreign stock exchanges coupled with potentially high inflation rates may require worldwide inflation-adjusted disclosures. Several related international standards have been issued by the International Accounting Standards Committee (IASC). For example, IAS 29 (1989) was issued for financial reporting in hyperinflationary economies. IAS 29 requires the restatement of general price-purchasing accounting (similar to constant dollar accounting) in hyperinflationary countries, regardless of whether current cost accounting is already being used (Goldschmidt, 1992). The hyperinflationary environment in Israel during the 1980s provides a distinctive setting for investigating the value relevance of inflationary disclosures.4 The purpose of this study is to examine the incremental information content of inflation-adjusted disclosures for investors in Israel. The study focuses on the cross-sectional relationship between returns (prices) and unexpected historical-cost or inflation-adjusted earnings. The major objective is to examine whether inflation-adjusted earnings (IAE) are value-relevant beyond historicalcost earnings (HCE). The incremental information content of inflation-adjusted disclosures has been controversial, producing, for example, conflicting conclusions in U.S. studies (e.g., Beaver, Griffin, & Landsman, 1982; Bublitz, Frecka, & McKeown,

Value Relevance of Cost Earnings During Hyperinflation

271

1985; Espahbodi & Tehranian, 1989). Although prior research has not examined this issue in Israel, there is a common belief among professionals and users that inflation-adjusted financial statements include information essential to investors in the Israeli stock market (ICPAI, 1985). Annual regression models, correlation coefficients, and pooled regression models are estimated using data for all publicly traded manufacturing firms in Israel. It appears that the cross-sectional correlations between inflation-adjusted and historical-cost earnings (and unexpected earnings) increase when inflation rates decrease. While the cross-sectional correlations between returns and historical-cost unexpected earnings are not statistically significant, the correlations between returns and inflation-adjusted unexpected earnings tend to be statistically significant. One-stage and two-stage regression analyses suggest that inflation-adjusted unexpected earnings are value relevant, whereas historical-cost unexpected earnings are not. In addition, unanticipated annual inflation seems to enhance the value relevance of inflation-adjusted earnings compared with historical-cost earnings. The major conclusion is that in an hyperinflationary environment, historical-cost unexpected earnings have essentially no value relevance for investors, but inflation-adjusted unexpected earnings do have value relevance. The next section provides an essential background on inflation-adjusted accounting in the U.S. and Israel and is followed by a review of relevant literature and the incremental contribution of this article. The methodology, data, sample, and some aggregate descriptive statistics are presented in the fourth section and are followed by the empirical results for testing the value relevance of unexpected IAE versus HCE. The final section summarizes the findings and discusses policy implications.

PROFESSIONAL

AND

REGULATORY CONCERNS

In the U.S., the Security and Exchange Commission (SEC), the Accounting Principles Board (APB), and the Financial Accounting Standards Board (FASB) have been concerned with inflation-adjusted earnings.5 The SEC (1976) released Accounting Series Release No. 190 (ASR 190), which compels large publicly traded corporations to display replacement cost data for earnings, as well as cost of goods sold, inventory, property, plant, and equipment. Following the release of ASR 190, the FASB issued a discussion memorandum and later an exposure draft, and prepared to issue a new standard for inflation-adjusted earnings. The Big Eight CPA firms published numerous pronouncements concerning the standard (e.g., Scott, 1978; Peat, Marwick, Mitchell & Co., 1979). The FASB (1979) issued SFAS 33, which requires firms to disclose primarily current-cost and constant dollar earnings; certain other income statement items; and current cost of inventory, property, plant, and equipment, in notes to the financial statements.6 Although annual inflation rates were double-digit in the

272

INTERNATIONAL ACCOUNTING, AUDITING & TAXATION, 8(2)

1999

period 1979 –1980, they decreased to less than five percent by the mid-1980s. SFAS 33 was resisted for various reasons, including the complexity of its applications and its very limited usefulness to users.7 For example, Berliner (1983) and Norby (1983) indicate limited use of SFAS 33 by financial analysts, who had marginal interest in the inflationary-adjusted earnings. A FASB research report by Beaver and Landsman (1983) shows that IAE had no additional information content over HCE for explaining security returns. The FASB (1984) issued SFAS No. 82 amending SFAS 33 by essentially eliminating the requirement for disclosure of constant dollar information for those companies that presented such information. This amendment was issued primarily as a result of extensive lobbying by companies and external users and as a consequence of the Beaver and Landsman study. The FASB (1986) issued SFAS No. 89 and made voluntary the disclosure of current-cost information. As a result, publicly traded companies suppressed inflation-adjusted disclosures. In Israel, Israeli Opinions (ISOPs) 23 and 36 were issued by the Institute of Certified Public Accountants in Israel (ICPAI) during 1979 and 1985, respectively. ISOPs are equivalent to Statements of Financial Accounting Standards in the U.S. ISOP 23 (ICPAI, 1979) recommends the inclusion of inflation-adjusted earnings in notes to the financial statements. The annual increases in the CPI were 445 percent and 185 percent in 1984 and 1985, respectively. In 1985, the ICPAI issued ISOP 36. ISOP 36 requires corporations whose securities are traded on the Tel Aviv Stock Exchange (TASE) to fully adjust their financial statements to the real purchasing power of the Israeli currency. Consequently, publicly traded companies, including smaller firms, adjusted their financial statements for changes in real purchasing power for 1985, and restated inflation-adjusted financial statements for 1984. This inflationadjusted reporting has continued since 1985. The adjustments are based on a current monetary unit approach, which is a variant of general purchasing power accounting or constant dollar accounting. The ICPAI considered current value accounting unjustified during periods of very high inflation. One argument was that, in an importing economy, gathering and sorting information on current prices for both machinery and raw materials is very complex. Following the 1985 macroeconomic reform and stabilization program, the annual inflation rates declined and ranged between 16 and 20 percent from 1986 to 1990. Although the annual inflation rates declined, the ICPAI (1989) reinforced the current monetary approach by issuing ISOP No. 50, which mandated inflationary disclosures for the entire income statement, including extraordinary items.

LITERATURE REVIEW

AND

INCREMENTAL CONTRIBUTION

In the U.S., a few studies examine the association between companies’ historical earnings and inflation-adjusted earnings. Ketz (1983) finds high cross-

Value Relevance of Cost Earnings During Hyperinflation

273

sectional correlation between historical earnings and constant-dollar earnings, and concludes that when inflation rates were relatively stable, though moderately high, constant dollar earnings had essentially no information content. Divergence was likely to occur only when there was a sharp increase in the rate of inflation. Beaver and Landsman (1983) use a sample of 731 publicly traded firms collected by the FASB during the deliberations on SFAS 33 and following its issuing, and demonstrate high cross-sectional correlations between historical earnings and constant dollar or current-cost earnings for the years 1979 – 81. They conclude that the incremental information content provided by inflation-adjusted earnings is negligible. Most studies examine the incremental information content of inflationadjusted items (especially earnings) for investors in the U.S. capital markets. These studies investigate cross-sectional differences in security returns as functions of inflation-adjusted items versus historical-cost items. Beaver et al. (1982) and Beaver and Landsman (1983) suggest that the inflation-adjusted pre-holdinggains based on ASR 190 and inflation-adjusted earnings based on SFAS 33, respectively, have no incremental information content in explaining security returns. As indicated above, the results of their studies encouraged the issuance of SFAS 89. Watts and Zimmerman (1980) conclude that the benefits from inflationadjusted items based on ASR 190 and SFAS 33 would outweigh the costs associated with them. Further studies present inconclusive results on incremental information content for explaining security returns. Bublitz et al. (1985) find significant incremental explanatory power in inflation-adjusted variables beyond that provided by historical-cost variables. Freeman (1983) shows that the trend in historical-cost earnings in a given industry is reflected in security prices before the trend in current-cost earnings. Bernard and Ruland (1987) demonstrate that current-cost earnings convey incremental information only in industries where the correlations between historical-cost income and current-cost income are low. Nunthirapakorn and Millar (1987) conclude that historical-cost income explains systematic risk as well as or better than inflation-adjusted data. Espahbodi and Tehranian (1989) and Sami, Curatola, and Trapnell (1989) find mixed and inconclusive results for the association between security returns and inflationadjusted and historical-cost earnings, but predictive ability tests indicate that inflation-adjusted measures prevail moderately over historical-cost measures. Bildersee and Ronen (1987) conclude that current cost-data, on the margin, reflect productive activity information that might not be already contained in historicalcost data. Cormier (1989) summarizes most of the studies and concludes that only a few studies find that current-cost information has statistically significant information content. He delineates several possible reasons for this finding. Several studies explicate aspects of the information content of inflation-adjusted disclosures other than the direct impact on security returns (e.g., Ketz, 1978; Baran,

274

INTERNATIONAL ACCOUNTING, AUDITING & TAXATION, 8(2)

1999

Lakonishok, & Ofer, 1980; Bar–Yosef & Lev, 1983). Hopwood and Schaefer (1989) provide some empirical support of an association between the between unexpected current-cost income and risk-adjusted returns for firms maintaining operating income during periods of cost increases. In sum, in the U.S., most studies find: (1) high cross-sectional correlations exist between inflation-adjusted earnings and historical-cost earnings; (2) the correlations decrease as annual inflation rates increase (but even then they are relatively high); and (3) the incremental information content of inflation-adjusted data in explaining security returns tends to be low or even marginal. Recently, Friday (1997) finds that replacement-cost data are value-relevant, while Gordon (1997) determines that they are value-relevant beyond historicalcost estimates for samples of Mexican firms. Previous research has not examined the incremental information content of inflation-adjusted earnings based on ISOPs 23 and 36 for investors in Israel. The incremental contribution of this study stems from the scrutiny of this relationship and the explanatory power of historical-cost earnings versus inflation-adjusted earnings in explaining security returns and prices during the period of triple-digit annual inflation in Israel, and the doubledigit inflation rate of less than 20 percent in 1986 and beyond. Since the results of previous studies in the U.S. during periods of moderate to low inflation are mixed, it is important to examine this relationship in a different country during hyperinflationary periods. The study does not examine other aspects of inflationadjusted data compared with historical-cost data.

METHODOLOGY, DATA,

AND

SAMPLE

Research Methodology The value relevance of unexpected inflation-adjusted earnings versus historical-cost earnings is examined using cross-sectional regressions of security returns on unexpected earnings. First, one-stage cross-sectional regression models are examined using annual and pooled data: RETj ⫽ ␭ 1 ⫹ ␮ 1DHCEj ⫹ ␪ Hj,

(1a)

RETj ⫽ ␭ 2 ⫹ ␮ 2DIAEj ⫹ ␪ Ij,

(1b)

RETj ⫽ ␭ ⫹ ␮ 1DHCEj ⫹ ␮ 2DIAEj ⫹ ␪ j,

(1c)

where: RETj ⫽ (Pricej,t ⫺ Pricej,t⫺1 ⫹ Dividendsj,t) / Pricej,t⫺1, is the annual return for security j in year t on the TASE;

Value Relevance of Cost Earnings During Hyperinflation

275

DHCEj ⫽ (HCEj,t ⫺ HCEj,t⫺1) / HCEj,t⫺1, is the unexpected (percentage change) historical-cost earnings, and HCEj,t is the historical-cost annual earnings for firm j in year t; DIAEj ⫽ (IAEt ⫺ IAEt⫺1) / IAEt⫺1, is the unexpected inflationadjusted earnings, and IAEj is the inflation-adjusted annual earnings for firm j in year t. Other measures of unexpected earnings, such as forecast errors, are not available, as financial analysts did not operate in Israel during the research period. The explanatory power of unexpected historical-cost earnings (DHCE) relative to unexpected inflation-adjusted earnings (DIAE) is also compared by 2 2 computing the ratio RH /RI2, where RH and RI2 are obtained from regressions (1a) and (1b), respectively. Comparing the R2s using the ratio, however, does not provide statistical tests to examine whether regressions using DIAE are superior to regressions using DHCE. Therefore, the Vuong (1989) z-statistic is used. The statistic assumes that neither model (1a) nor model (1b) are necessarily true (see Dechow, 1994). Second, a two-stage regression approach is used (Beaver et al., 1982). The first set of two-stage regressions examines whether unexpected inflation-adjusted earnings (DIAE) provide incremental value relevance conditional on knowledge of unexpected historical-cost earnings (DHCE). Cross-sectional regressions by year and pooled regressions are presented. In the first stage, DIAE is regressed on DHCE as follows: DIAEj ⫽ ␣ ⫹ ␤ DHCEj ⫹ Z j,

(2)

and in the second-stage, annual security returns (RET) are regressed on DHCE and the residuals (Zs) obtained from the first regression: RETj ⫽ ␥ ⫹ ␦ 1DHCEj ⫹ ␦ 2Z j ⫹ ⑀ j,

(3)

The regressions then are transposed to determine whether DHCE provides incremental value relevance conditional upon knowledge of DIAE. The first-stage regression is: DHCEj ⫽ ␣ ⫹ ␤ DIAEj ⫹ Z j,

(4)

and the second-stage regression is: RETj ⫽ ␥ ⫹ ␦ 1DIAEj ⫹ ␦ 2Z j ⫹ ⑀ j. Finally, a one-stage price model is used. The regression model is:

(5)

276

INTERNATIONAL ACCOUNTING, AUDITING & TAXATION, 8(2)

1999

TABLE 1

Annual Inflation Rates and Stock Returns 1980 –90 Annual Inflation Rate

Year 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990

132.9% 101.5 131.5 190.6 444.9 185.2 19.7 16.1 16.4 20.4 17.6

TASE Annual Returns (Excluding Banks)a Nominal Return

Real Returnb

356.9% 149.9 454.6 7.9 408.9 199.2 66.2 23.3 ⫺19.1 96.0 16.6

96.2% 24.0 135.6 ⫺62.9 ⫺6.6 4.9 38.9 6.2 ⫺30.5 62.2 ⫺0.8

a

Value-weighted average annual returns for firms traded on the Tel Aviv Stock Exchange (TASE), excluding banks. b The annual real return is defined as: (1 ⫹ R r ) ⫽ (1 ⫹ R n )/(1 ⫹ I), where R r is the real return, R n is the nominal return, and I is the inflation rate. c Unanticipated inflation rates are estimated as 1/4, 1/3, 1/5, 1/3, and 1/20 of the annual inflation rates for 1984, 1985, 1986, 1987, and 1988, respectively (The 1989 Annual Report of the Central Bank of Israel).

DPRj ⫽ ␳ ⫹ ␾ 1HCEj ⫹ ␾ 2CHCEj ⫹ ␾ 3IAEj ⫹ ␾ 4CIAEj ⫹ ␻ j

(6)

where: DPRj ⫽ the annual change in price of security j; CHCEj ⫽ HCEj,t ⫺ HCEj,t⫺1; and CIAEj ⫽ IAEj,t ⫺ IAEj,t⫺1. Data and Sample Most publicly traded companies in Israel adjust financial statements to changes in the general purchasing power of the Israeli Shekel based on changes in the CPI in Israel. Relatively few companies adjust financial statements to changes in the foreign exchange rate8. Table 1 presents the annual inflation rates in Israel and the value-weighted average returns on the TASE from 1980 through 1990. The triple-digit annual inflation rates were extremely high during the period 1983–1985. The 1985 macroeconomic reform and stabilization program reduced annual inflation rates to 20 percent in 1986 and subsequent years. In addition, the Israeli economy has been characterized by indexation of wages, savings, and long-term debt, which have been linked to changes in the CPI (Fisher, 1985, 1993).9 The annual ratio of unanticipated inflation to actual inflation is presented in Note c under Table 1. Both nominal and real annual returns on the TASE are also presented in Table 1. Following the market crash on the TASE in Fall, 1983, investors in Israel tended

Value Relevance of Cost Earnings During Hyperinflation

277

TABLE 2

The Manufacturing Firms Traded on the Tel Aviv Stock Exchange (TASE) Firms Traded on the TASE 1984–1988 Less: Financial & Insurance Real Estate & Agriculture Retail & Service Investment Oil & Gas Ext.

262

Firms Excluded

153

Manufacturing Firms Missing Data

109 3

Useful Firms

106

41 42 31 33 6

to gain positive real annual returns between 1984 and 1990, but not in 1984 and 1988. Table 2 shows the composition of the 262 companies traded on the TASE during the research period 1984 – 88. The sample includes all manufacturing companies, except three firms with missing data, yielding a final sample of 106 manufacturing firms. Other industries are excluded due to missing values and also to avoid potential noise generated by industry effect. Price information from the end of the months when annual financial statements were released and annual earnings were obtained from Halevey Financial-Service File and TASE publications.10

RESULTS Table 3 presents the means of, and cross-sectional correlations between, unexpected historical-cost earnings, unexpected inflation-adjusted earnings, returns, and prices. Annual and pooled correlations between the unexpected earnings tend to be low, but most are statistically significant. Cross-sectional correlations between returns and DHCE range from 0.007 to 0.109 and are not statistically significant. In contrast, DIAE has significantly higher correlations with returns for most years, ranging from 0.139 to 0.440. In particular, higher correlations between RET and DIAE are obtained during the period 1985– 87. A rationale for this result is provided below. The cross-sectional correlations between DPR and CHCE range from 0.005 to 0.153 and are not statistically significant for the specific years but are significant for the pooled results. In comparison, CIAE, except for 1987, is significantly correlated with price changes, ranging from 0.186 to 0.448. The initial regression analysis compares the association between stock returns and unexpected historical-cost earnings to the association between stock

278

INTERNATIONAL ACCOUNTING, AUDITING & TAXATION, 8(2)

1999

TABLE 3

Cross-Sectional Univariate Analyses and Pearson’s Correlations Between Inflation-Adjusted, Historical-Cost Earnings and Returns: Manufacturing Firmsa Pooled 1984 Sample Size

b

31

1985

1986

1987

1988

1984 –1988

1985–1988

106

106

105

105

453

422

42.3 18.3 23.7 115.6 70.2 0.016

17.8 ⫺47.5 ⫺241 ⫺65.2 58.2 1.015

Panel A: Means for Univariate Variables DHCE (%) DIAE (%) CHCE (IS 000) CIAE (IS 000) RET (%) DPR (IS 000)

27.7 33.4 179.8 29.6 33.5 ⫺199.1 776.3 561.7 262.4 157.7 2.393 2.645

46.6 77.5 509.4 153.6 58.6 1.261

34.7 ⫺7.2 64.4 ⫺100 19.1 0.753

⫺43.5 ⫺124 ⫺1342 ⫺1351 ⫺2.49 ⫺0.601

Panel B: Correlation Coefficients (Rs) DHCE*DIAE RET*DHCE RET*DIAE

⫺.405d .109 .242

⫺.019 ⫺.065 .440c

.198d .025 .277c

.079 .007 .403c

.232d .075 .139

.148c .023 .148c

.117d .009 .194c

CHCE*CIAE DPR*CHCE DPR*CIAE Annual Inflation Rate (%)

⫺.147 .039 .448d

.591c .051 .253c

.219d .005 .378c

.163 .064 .186

.829c .091 .207d

.616c .129c .223c

.563c .153c .329c

16.1

16.4

444.9

185.2

19.7

a

Annual returns and earnings intervals are from the end of the month of the financial statements’ releases through the end of the month of the financial statements’ releases in the following year. Few earnings announcements precede financial statements releases by one trading day and have no impact on the reported results. RET ⫽ (Pricet ⫺ Pricet⫺1 ⫹ Dividendst )/Pricet⫺1 . DHCE ⫽ (HCEt ⫺ HCEt⫺1 )/HCEt⫺1 , where HCE is the historical-cost earnings. DIAE ⫽ (IAEt ⫺ IAEt⫺1 )/IAEt⫺1 , where IAE is the inflation-adjusted earnings. DPR ⫽ Pricet ⫺ Pricet⫺1 . CHCE ⫽ HCEt ⫺ HCEt⫺1 , where HCE is the historical-cost earnings. CIAE ⫽ IAEt ⫺ IAEt⫺1 , where IAE is the inflation-adjusted earnings. b Inflation-adjusted earnings are available for only 31 firms in 1983; this data is necessary for computing DIAE and CIAE for 1984. c Significant at p ⬍ 0.01. d Significant at p ⬍ 0.05.

returns and unexpected inflation-adjusted earnings. Table 4 presents the results for the annual regressions from 1984 through 1988 and pooled regressions for both the 1984 – 88 and 1985– 88 periods. The pooled regression for 1985– 88 is reported separately because results for 1984 are based on fewer observations and, therefore, considered less reliable. In 1983, only 31 firms reported the IAE needed to compute variable DIAE for 1984. The t-statistics on the estimated coefficients of DHCE are not statistically significant for all regressions, whereas the t-statistics

Value Relevance of Cost Earnings During Hyperinflation

279

TABLE 4

One-Stage Regressions: Dependent Variable–RETj on DHCE or DIAEa Pooled Variable Sample Size

1984

1985

1986

1987

1988

1984 – 88

1985– 88

31

106

106

105

105

453

422

RETj ⫽ ␭ ⫹ n 1 DHCEj ⫹ ␪ Hj

␮1 t-statisticsb

.050 (1.03)

⫺.021 (⫺.67)

.019 (0.25)

.001 (0.08)

.009 (0.77)

.009 (0.57)

.003 (0.19)

R2 Adj-R 2

.012 .0007

.004 ⫺.005

.0006 ⫺.009

.0001 ⫺.009

.006 ⫺.004

.0005 ⫺.001

.0001 ⫺.002

.321 (1.34)

.195 (4.99)

.066 (2.94)

.126 (4.49)

.004 (1.44)

.021 (3.30)

.024 (4.05)

R2 Adj-R 2

.059 .026

.194 .186

.077 .068

.162 .155

.019 .010

.022 .020

.038 .035

2 /R I2c RH

.204

.021

.008

.0006

.316

.023

.0026

.21

5.04

2.11

3.63

.70

2.23

3.84

RETj ⫽ ␭ ⫹ n 2 DIAEj ⫹ ␪ Ij

␮2 t-statisticsb

Vuong-Z-statisticsd

RET ⫽ (Pricet ⫺ Pricet⫺1 ⫹ Dividendst )/Pricet⫺1 . DHCE ⫽ (HCEt ⫺ HCEt⫺1 )/HCEt⫺1 , where HCE is the historical-cost earnings. DIAE ⫽ (IAEt ⫺ IAEt⫺1 )/IAEt⫺1 , where IAE is the inflation-adjusted earnings. b All t-statistics greater than 1.98 and 2.62 are significant at p ⬍ 0.05 and p ⬍ 0.01, respectively. c 2 R H /R I2 —This ratio examines the explanatory power of DHCE relative to DIAE. While examining the ratio of Adj-R 2 s is preferable, most Adj-R 2 s are negative for DHCE as an independent variable; therefore, the adjusted-R 2 s ratio is not applicable. d The Vuong-Z-statistics (1989) match unexpected IAE with unexpected HCE as competing nonnested models. Vuong-Z-statistics greater than 1.96 and 2.58 are significant at p ⬍ 0.05 and p ⬍ 0.01, respectively. a

on estimated coefficients of DIAE are positive and statistically significant for the 1985, 1986, 1987, and the pooled regressions. For each regression presented in Table 4, the R2 is relatively greater when DIAE, rather than DHCE, is included in the regression. For example, in 1987 the R2 (adjusted–R2) is 0.162 (0.155) for the regression using DIAE as an independent variable, compared to 0.0001 (⫺0.009) for the regression using DHCE. The 2 ratio RH /RI2 is only 0.021 and 0.0006 for the 1985 and 1987 regressions, respectively. In 1988, however, the ratio increases to 0.316. The results for the Vuong z-statistics are reported at the bottom of Table 4. The results demonstrate that the explanatory power of DIAE is significantly greater than that of DHCE. Specifically, DHCE is rejected in favor of DIAE for regressions using data from 1985, 1986, 1987, and the pooled regressions, but not for the 1984 and 1988 regressions. Results for the one-stage regressions, reported in Table 5, demonstrate that none of the estimated coefficients for DHCE are statistically significant, whereas

280

INTERNATIONAL ACCOUNTING, AUDITING & TAXATION, 8(2)

1999

TABLE 5

One-Stage Regressions: Dependent Variable–RETj on DHCE or DIAEa Pooled Variable

1984

1985

1986

1987

1988

1984 – 88

1985– 88

Sample Size

31

106

106

105

105

453

422

⫺.002 (⫺0.26)

.005 (0.45)

⫺.006 (⫺0.36)

⫺.004 (⫺0.29)

.126 (4.46) 9.95 (.0001) .163 .147

.004 (1.28) 1.11 (.332) .021 .0002

.028 (4.13) 8.52 (.0002) .036 .032

.024 (4.05) 8.22 (.0003) .038 .033

RETj ⫽ ␭ ⫹ n 1 DHCEj ⫹ n 2 DIAEj ⫹ ␪ j ␮1 .167 ⫺.018 ⫺.025 (0.18) (⫺.64) (⫺0.33) t-statisticsb

␮2 t-statisticsb F-statistics (p-value) R2 Adj-R 2

.341 (1.28) 0.88 (.423) .059 ⫺.008

1.94 (4.97) 12.6 (.0001) .197 .181

.067 (2.93) 4.33 (.016) .078 .060

RET ⫽ (Pricet ⫺ Pricet⫺1 ⫹ Dividendst )/Pricet⫺1 . DHCE ⫽ (HCEt ⫺ HCEt⫺1 )HCEt⫺1 , where HCE is the historical-cost earnings. DIAE ⫽ (IAEt ⫺ IAEt⫺1 )/IAEt⫺1 , where IAE is the inflation-adjusted earnings. b All t-statistics greater than 1.98 and 2.62 are significant at p ⬍ 0.05 and p ⬍ 0.01, respectively. a

the estimated coefficients for DIAE are statistically significant except for 1984 and 1988. The t-statistics for DIAE in the pooled analyses are highly significant and positive in both regressions. In addition, the regressions are highly significant and contain substantial explanatory power (R2) except for 1984 and 1988. While these results are expected from the correlations and regressions reported in Table 3 and Table 4, respectively, the results reported in Table 5 further suggest that unexpected inflation-adjusted earnings are value-relevant, while unexpected historical-cost earnings are not. Table 6 reports the two-stage annual and pooled regressions. In the first stage, presented in Panel A, DIAE is regressed on DHCE to obtain the residual Z. The t-statistics are positive and significant (at a 5 percent level) for 1986, 1988, and for the pooled regressions, but are negative and significant for 1984. These results are anticipated, given the correlations reported in Table 3. The results in the second stage demonstrate that none of the estimated coefficients for DHCE are statistically significant. The estimated coefficients of the residuals, Zs, from the first-stage, however, are of substantial statistical significance for 1985, 1986 and 1987, and the pooled regressions. Thus, DIAE provides incremental value relevance, conditional upon knowledge of DHCE. Panel B of Table 6 shows the two-stage regressions while reversing panel A and regressing DHCE on DIAE in the first stage. The results are consistent with the first stage presented in Panel A. As expected for the second stage, the estimated coefficients for RET on DIAE are statistically significant for most years and for the pooled regressions. None of the estimated coefficients for the Zs in the

Value Relevance of Cost Earnings During Hyperinflation

281

TABLE 6

Two-Stage Regressionsa Pooled Variable

1984

1985

1986

1987

1988

1984 – 88

1985– 88

Panel A: Inflation-Adjusted Earnings as Dependent Variable in the First Stage DIAEj ⫽ ␣ ⫹ ␤ DHCEj ⫹ Z j ␤ ⫺1.38 ⫺.136 (⫺2.39) (⫺0.19) t-statisticsb .165 .0004 R2 .136 ⫺.009 Adj-R 2

.673 (2.07) .039 .030

.232 (0.81) .006 ⫺.003

.878 (2.42) .054 .045

.298 (2.48) .013 .011

.300 (2.42) .014 .011

Second Stage RETj ⫽ ␥ ⫹ ␦ 1 DHCEj ⫹ ␦ 2 Z j ⫹ ⑀ j ␦1 ⫺.305 ⫺.021 (⫺0.37) (⫺0.74) t-statisticsb ␦2 .341 .194 (1.28) (4.97) t-statisticsb .059 .197 R2 ⫺.008 .181 Adj-R 2

.019 (0.26) .067 (2.93) .078 .060

.0007 (0.09) .126 (4.46) .163 .147

.009 (0.77) .004 (1.28) .021 .002

.002 (0.12) .028 (4.13) .037 .032

.003 (0.19) .024 (4.05) .038 .033

Panel B: Historical-Cost Earnings as Dependent Variable in the First Stagec DHCEj ⫽ ␣ ⫹ ␤ DIAEj ⫹ Z j ␤ ⫺.119 ⫺.027 (⫺2.39) (⫺0.19) t-statisticsb

.059 (2.07)

.027 (0.81)

.061 (2.42)

.045 (2.48)

.046 (2.42)

Second Stagec RETj ⫽ ␥ ⫹ ␦ 1 DIAEj ⫹ ␦ 2 Z j ⫹ ⑀ j ␦1 .321 .195 (1.32) (4.97) t-statisticsb ␦2 .167 ⫺.018 t-value (0.18) (⫺0.64)

0.065 (2.93) ⫺0.025 (⫺0.33)

.012 (4.45) ⫺.002 (⫺0.26)

.004 (1.42) .005 (0.45)

.028 (4.11) ⫺.006 (⫺0.36)

.024 (4.05) ⫺.004 (⫺0.29)

RET ⫽ (Pricet ⫺ Pricet⫺1 ⫹ Dividendst )/Pricet⫺1 . DHCE ⫽ (HCEt ⫺ HCEt⫺1 )/HCEt⫺1 , where HCE is the historical-cost earnings. DIAE ⫽ (IAEt ⫺ IAEt⫺1 )/IAEt⫺1 , where IAE is the inflation-adjusted earnings. b All t-statistics greater than 1.98 and 2.62 are significant at p ⬍ 0.05 and p ⬍ 0.01, respectively. c 2 R s and adjusted-R 2 s are identical in both panels. a

second stage, however, are significant. Thus, DHCE provides no incremental value relevance, conditional upon knowledge of DIAE. Both correlation and regression results, reported in Tables 3 through 6 are significant for 1985– 87 but are not for 1984 and 1988. These findings may be attributed to the small sample used for 1984 and the lower unanticipated inflation rate (anticipated inflation was almost equal to actual inflation) during 1988. The cross-sectional correlations between RET and DIAE and the estimated coefficients for DIAE tend to be higher when the annual inflation rates decrease in comparison with the previous year, but the unanticipated inflation rates increase. It is estimated by the 1989 Annual Report of the Central Bank of Israel that about one-third of the annual inflation rates during 1985 and 1987 and about one-fifth

282

INTERNATIONAL ACCOUNTING, AUDITING & TAXATION, 8(2)

1999

TABLE 7

One-Stage Regressions: Dependent Variable–DPRj on HCE, CHCE, IAE, and CIAE Pooled Variable Sample Size

1984 31

1985 106

1986 106

1987 105

1988 105

1984 – 88

1985– 88

453

422

DPRj ⫽ ␳ ⫹ ␾ 1 HCEj ⫹ ␤ 2 CHCEj ⫹ ␾ 3 IAEj ⫹ ␾ 4 CIAEj ⫹ ␻ j

␾1 t-statisticsb

1.383 (2.72)

.167 (1.40)

.173 (1.97)

⫺.039 (⫺0.80)

.068 (1.04)

.038 (0.79)

.068 (1.82)

␾2 t-statisticsb

.613 (1.53)

⫺.379 (⫺3.44)

⫺.283 (⫺1.96)

.153 (2.76)

⫺.082 (⫺1.50)

⫺.143 (⫺3.06)

⫺.115 (⫺2.95)

␾3 t-statisticsb

⫺1.984 (⫺2.71)

.299 (2.07)

.048 (0.62)

.110 (2.59)

⫺.081 (⫺1.39)

.094 (2.11)

.141 (1.77)

␾4 t-statisticsb

1.832 (3.24)

.231 (2.76)

.125 (1.56)

.150 (3.37)

.155 (2.80)

.138 (3.32)

.141 (4.36)

F-statistics (p-value)

4.57 (.006)

29.5 (.0001)

7.90 (.0001)

10.3 (.0001)

2.25 (.069)

12.8 (.0001)

29.4 (.0001)

.404 .316

.539 .521

.283 .208

.289 .261

.082 .045

.096 .089

.163 .155

R2 Adj-R 2

a DPR ⫽ Pricet ⫺ Pricet⫺1 . CHCE ⫽ HCEt ⫺ HCEt⫺1 where HCE is the historical-cost earnings. CIAE ⫽ IAEt ⫺ IAEt⫺1 , where IAE is the inflation-adjusted earnings. For presentation only, all estimated coefficients are multiplied by 1000. b All t-statistics greater than 1.98 and 2.62 are significant at p ⬍ 0.05 and p ⬍ 0.01, respectively.

of the annual inflation rate during 1986 comprise unanticipated inflation. The estimated unanticipated inflation for 1988, however, is negligible. Therefore, the results are highly significant for 1985, 1986, and 1987, but not for 1984 and 1988. Results obtained using the price model tend to support those obtained from the return model. The regressions presented in Table 7 are statistically significant for the annual and pooled regressions, except the regression for 1988. The estimated coefficients for CIAE are positive and statistically significant. Contrary to expectations, however, the estimated coefficients for CHCE are negative and statistically significant for the 1985 and the pooled regressions. The estimated coefficient for CHCE is positive and statistically significant only for the 1987 regression. These results further suggest that inflation adjusted earnings are value-relevant, while historical-cost earnings are not. In sum, the cross-sectional incremental value relevance of inflation-adjusted earnings is significant, while the value relevance of historical-cost earnings is negligible for investors in Israel. During the hyperinflation of 1984 and 1985 and

Value Relevance of Cost Earnings During Hyperinflation

283

the double-digit inflation from 1986 to 1987, inflation-adjusted unexpected earnings provide incremental value relevance, given unexpected historical-cost earnings are known. DHCE, however, provides no additional value relevance conditional upon knowledge of unexpected inflation-adjusted earnings (DIAE). Unanticipated inflation seems to reinforce this conclusion since the results tend to be highly significant when it exists and insignificant when it is lower.

SUMMARY

AND

CONCLUDING REMARKS

This study is the first to examine the value relevance of inflation-adjusted earnings versus historical-cost earnings for investors in Israel. The study focuses on a period of triple-digit annual inflation rates and the convergence to doubledigit annual inflation of less than 20 percent. Since 1985, Israeli publicly traded companies, including smaller firms, have adjusted their entire financial statements for changes in real purchasing power. The adjustments are based on the current monetary unit approach, a variant of constant dollar accounting. Historical-cost balance sheets and income statements continue to be reported only in notes. The combination of hyperinflation and a financial reporting system that provides both historical-cost and inflation-adjusted financial statements presents a unique opportunity to examine the value relevance of inflation-adjusted earnings versus historical-cost earnings. Because the results of previous studies in the U.S. during periods of moderate to low inflation are mixed, it is important to examine this relationship in a different country during hyperinflationary periods. Multivariate analyses demonstrate that the explanatory powers of the inflation-adjusted regressions are statistically more significant than the poor explanatory powers of the historical-cost regressions. In addition, the results show that the estimated coefficients for unexpected inflation-adjusted earnings (DIAE) are statistically significant in explaining security returns (RET), whereas the estimated coefficients for unexpected historical-cost earnings (DHCE) are not. Furthermore, two-stage regressions demonstrate that the estimated coefficients of DIAE are statistically significant in explaining RET once DHCE is known. The estimated coefficients of DHCE, however, are not significant even when inflationadjusted earnings are known. In sum, during the hyperinflation of 1984 and 1985 and the double-digit inflation of 1986 – 88, the empirical results show that the value relevance of inflation-adjusted earnings was significant, while that of historical-cost earnings was negligible. Two potential explanations for the results reported in this study are as follows. First, during hyperinflation, historical-cost data distort measurement accuracy, and reported earnings are not reliable. Second, relevant and reliable measurement of inflation-adjusted accounting is enhanced with higher unanticipated inflation. The results presented in this study suggest that unexpected inflation-adjusted earnings are significantly more value-relevant than unexpected

284

INTERNATIONAL ACCOUNTING, AUDITING & TAXATION, 8(2)

1999

historical-cost earnings during years with high unanticipated inflation. The differences in value relevance diminish and are less significant when unanticipated inflation rates are relatively low. Thus, unexpected inflation-adjusted earnings are significantly value-relevant beyond unexpected historical-cost earnings, particularly during years with high unanticipated inflation. Unexpected historical-cost earnings, however, are not value-relevant during the entire research period. The results suggest that inflation-adjusted earnings may be more valuerelevant than historical-cost earnings for countries experiencing hyperinflation or even moderate double-digit inflation. It appears that inflation-adjusted measurements, such as those recommended by IAS 29, are essential for investors during periods of hyperinflation or even moderate double-digit inflation. During such periods, investors may ignore historical-cost measurements and prefer valuerelevant inflation-adjusted reporting systems. The results presented in this study will guide standard setters and regulators in establishing the relevance of inflation-adjusted reporting systems. Three limitations of the study should be noted. First, due information missing from the database, only prices, returns, and earnings are reported and analyzed. Inclusion of other financial statement data may improve the analyses. Second, data are available for the period 1983–1988. Results for more recent years, however, may be less informative due to declining inflation rates during the 1990s. Finally, only manufacturing firms are included in the empirical analyses. Further examinations of other industries may be useful, though they will probably introduce noise in the empirical analyses. Future research should address these limitations.

Acknowledgment: I acknowledge the helpful comments of two anonymous reviewers, Kreag Danvers, Olusegan Wallace (the discussant) and other participants in the American Accounting Association, International Section Midyear Meeting, Orlando Florida, 1999, and participants in the American Accounting Association Midwest Regional Meeting, 1999. Data collection efforts by B. Shmila and S. Asulin are also appreciated.

NOTES 1.

2.

In Mexico, more comprehensive disclosures have been required since 1985. Publicly traded companies are required to report the impact of inflation on nonmonetary assets, certain income statement items, and shareholders’ equity. Most items are adjusted to general purchasing power or current-cost disclosures. Inventory and cost of goods sold are restated using replacement cost. In Brazil, a second system known as integral correction was introduced in 1987. Both monetary correction and integral correction are required for publicly traded firms (Doupnik et al., 1995).

Value Relevance of Cost Earnings During Hyperinflation 3.

4.

5.

6.

7.

8.

9.

10.

285

In Israel, all publicly traded companies have issued inflation-adjusted financial statements routinely since 1985. The initial costs of implementing such adjustments were not substantial, since inexpensive off-the-shelf software and programs were available. Although hyperinflation is not precisely defined, the FASB defines a highly inflationary economy as one that has a cumulative inflation rate of 100 percent or more over a three-year period, and studies have defined double-digit inflation rate as being in the range of 10 –25 percent (Boatler, 1992). Some economists argue that annual inflation rates over 50 percent may be considered hyperinflation. The APB recommended (APB Opinion 3, 1968) the inclusion of inflation-adjusted items as supplements to the financial statements. The recommendation did not have a meaningful impact on the financial statements, since publicly traded firms did not follow it. SFAS 33 (FASB, 1979) applied to medium-size and large publicly-traded companies that had either total assets net depreciation of $1 billion, or inventories, property, plant, and equipment of $125 million, before depreciation and amortization. SFAS 33 was issued to be effective for five years. Larger companies opposed the requirement to comply with SFAS 33, since they estimated substantial adjustment costs, and, primarily, because their inflation-adjusted earnings were much smaller compared with historical earnings. Paragraph 29 of ISOP 36 permits the use of foreign currency adjustment if one of the following conditions applies: (1) most revenues, income, and fixed assets are received in foreign currency, and (2) securities of the public company are traded on a foreign stock exchange. The Israeli economy has been characterized by complete indexation since the early 1970s. Saving accounts, financial debt (including bonds), and labor markets have all been indexed to inflation. Trading on the Tel Aviv Stock Exchange (TASE) was influenced by banks and mutual funds during the 1980s. Banks were also the major brokers. However, many sophisticated investors have been active in trading, and information has been flown in on a daily basis. The data include annual earnings, prices, and market information. Other items which are beyond the scope of this study are not available. Only annual earnings, prices, and some dividends are used. ISOP 43 (1986) recommends issuing semi-annual or quarterly interim financial statements. Many publicly traded companies, however, did not report interim financial statements during the 1980s.

REFERENCES Baran, A., Lakonishok, J., & Ofer, R. A. 1980. The information content of general price level adjusted earnings: Some empirical evidence. Accounting Review, 55: 22–35. Bar-Yosef, S., & Lev, B. 1983. Historical-cost earnings versus inflation-adjusted earnings in the dividend decision. Financial Analysts Journal, 39: 41–50. Beaver, W. H., & Landsman, W. R. 1983. Incremental information content of statement 33 disclosure. Stamford, CT: FASB. Beaver, W. H., Griffin, P. A., & Landsman, W. R. 1982. The incremental information content of replacement cost earnings. Journal of Accounting and Economics, 4: 15–39. Berliner, R. W. 1983. Do analysts use inflation-adjusted information? Results of a survey. Financial Analysts Journal, 39: 65–72. Bernard, V. L., & Ruland, R. J. 1987. The incremental information content of historical and current cost income numbers: Time series analysis for 1962–1980. Accounting Review, 62: 707–722. Bildersee, J. S., & Ronen, J. 1987. Stock return and real activity in an inflationary environment: The information impact of FAS No. 33. Contemporary Accounting Research, 4: 89 –110.

286

INTERNATIONAL ACCOUNTING, AUDITING & TAXATION, 8(2)

1999

Boatler, R. W. 1992. When inflation is not high enough: Disappearance of real assets under FAS 52. International Journal of Accounting, 27: 302–317. Bublitz, B., Frecka, T. J., & McKeown, J. C. 1985. Market association test and FASB Statement No. 33 disclosures: A reexamination. Journal of Accounting Research, 23 (Suppl.): 1–23. Cormier, D. 1989. Descriptive models of inflation accounting adopted by standard setting bodies: Empirical results. Journal of Accounting Literature, 8: 293–309. Dechow, P. M. 1994. Accounting earnings and cash flows as measures of firm performance: The role of accounting accruals. Journal of Accounting and Economics, 18: 3– 42. Doupnik, T. S., Martins, E., & Barbieri, G. 1995. Innovations in Brazilian inflation accounting. International Journal of Accounting, 30: 302–317. Espahbodi, R., & Tehranian, H. 1989. Stock market reaction to issuance of FAS 33 and its preceding exposure drafts. Contemporary Accounting Research, 5: 575–591. FASB. 1979. SFAS No. 33. Financial reporting and changing prices. Stamford, CT: Financial Accounting Standards Board. . 1984. SFAS No. 82. Financial reporting and changing prices: Elimination of certain disclosures. Stamford, CT: FASB. . 1986. SFAS No. 89. Financial reporting and changing prices. Stamford, CT: FASB. Fisher, S. 1985. Israeli inflation and indexation. In J. Williamson (Ed.), Inflation and indexation. Cambridge, MA: MIT Press. . 1993. Israeli growth, stabilization and reform. Tel Aviv, Israel: The Association of Banks in Israel & Tel-Aviv University. Freeman, R. N. 1983. Alternative measures of profit margin: an empirical information content of current cost accounting. Journal of Accounting Research, 21: 42– 64. Friday, D. F. 1997. Accounting valuation and the current cost disclosures: The case of Mexico. Working paper. Presented at the annual meeting of the AAA, Dallas, TX. Goldschmidt, Y. 1992. International Accounting Standard 29: formulation and clarification on income measurement in hyperinflationary economies. International Journal of Accounting, 27: 137–150. Gordon, E. A. 1997. An examination of the value relevance of historical-cost, price level and replacement cost accounting using firms listed on the Mexican stock exchange. Working paper. Presented at the annual meeting of the AAA, Dallas, TX. Hopwood, W., & Schaefer, T. 1989. Firm-specific responsiveness to input price changes and the incremental information in current cost income. Accounting Review, 64: 313–328. ICPAI. 1979. ISOP No. 23. Disclosure in financial statements of the effects of the changes in the general Purchasing power of the Israeli currency on the operating result of companies. Tel Aviv, Israel: Institute of Certified Public Accountants in Israel. . 1985. ISOP No. 36. Principles for preparation of financial statements adjusted for changes in the purchasing power of the shekel. Tel Aviv, Israel: ICPAI. . 1989. ISOP No. 50. Preparation of income statements fully adjusted for changes in the purchasing power of the Shekel (An extension of ISOP 36). Tel Aviv, Israel: ICPAI. Ketz, J. E. 1978. The effect of general price-level adjustments on the predictive ability of financial ratios. Journal of Accounting Research, 16: 273–284. . 1983. Are constant dollar disclosures informative? Financial Analysts Journal, 39: 52–55. Norby, W. C. 1983. Application of inflation-adjusted accounting data. Financial Analysts Journal, 39: 33–39. Nunthirapakorn, T., & Millar, J. A. 1987. Changing prices, accounting earnings and systematic risk. Journal of Business Finance and Accounting, 14: 1–25. Peat, Marwick, Mitchell & Co. 1979. FASB takes its stand on issue of changing prices. Executive Newsletter. Sami, H., Curatola, A. P., & Trapnell, J. E. 1989. Evidence on the predictive ability of inflationadjusted earnings measures. Contemporary Accounting Research, 5: 556 –574.

Value Relevance of Cost Earnings During Hyperinflation

287

Scott, G. M. 1978. Research study on current-value accounting measurement and utility. Touche Ross Foundation. Vuong, Q. H. 1989. Likelihood ratio tests for model selection and non-nested hypotheses. Econometrica, 57: 307–333. Watts, R. S., & Zimmerman, J. L. 1980. On the relevance of replacement cost disclosures for security prices. Journal of Accounting and Economics, 2: 95–106.