Equity Valuation and Current Cost Disclosures: the Case of Mexico

Transcription

1 Journal of International Financial Management and Accounting 12: Equy Valuation and Current Cost Disclosures: the Case of Mexico Paqua Y. Davis-Friday Universy of Notre Dame, 386 Mendoza College of Business, Notre Dame, IN 46556, USA Abstract This study uses an accounting-based valuation model to investigate the relation between the market value of publicly traded Mexican firms and their disclosures of price-level adjusted accounting information. The model is estimated on a sample of Mexican companies during , when annual inflation rates in Mexico decreased from 130 per cent to 20 per cent. The results indicate that general price level-adjusted and current cost disclosures explain a significant portion of the cross-sectional variation in the market-to-book ratios of the sample firms. Further, the explanatory power of holding gains is robust to decreases in the general level of inflation, which suggests that current cost and constant peso disclosures are relevant for determining firm value over a wide range of inflation rates. These results are particularly important now since the Mexican Instute of CPAs has proposed eliminating the measurement of holding gains in order to make Mexican financial statements more comparable to US and Canadian GAAP. 1. Introduction In 1994, the Canadian Instute of Chartered Accountants (CICA), the Mexican Instute of CPAs (IMCP), and the Financial Accounting Standards Board (FASB) completed a joint study to analyze similaries and differences in accounting standards across the three countries (FASB, 1994). The authors of the study concluded that the implementation of the North American Free Trade Agreement (NAFTA) should lead to profound changes in accounting rules, especially in Mexico. In response to these conclusions, the Mexican Accounting Principles Commtee agreed that beginning January 1, 1999, comparative Mexican financial statements This paper is based on my dissertation completed at the Universy of Michigan. I wish to thank my dissertation commtee members, Gene Imhoff (Co-Chair), Russ Lundholm, Gunter Dufey, and Robb Muirhead, for their helpful comments and suggestions and especially the late Vic Bernard (Co-Chair) whose contributions to this project and to my development as a researcher have been immeasurable. The paper has also benefed from the comments of Ed Swanson, Tom Stober, Doug Skinner, Juan Rivera, Ram Ramanan, Grace Pownall, Connie Kertz, Rajib Doogar, Michael Clement, workshop participants at Emory Universy, and participants at the 1997 Indiana/Notre Dame/Purdue Research Conference, the 1997 American Accounting Association Annual Meeting, and the 1999 AAA International Accounting Section Mid-year Meeting. The financial support of the AICPA, the Center for International Business Education and Research at the Universy of Michigan, the KPMG Foundation, and the Paton Accounting Foundation is gratefully acknowledged. Blackwell Publishers Ltd. 2001, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA.

2 Equy Valuation and Current Cost Disclosures 261 would no longer be updated to pesos of current purchasing power. In addion, any adjustments to revalue fixed assets would be made using the National Consumer Price Index instead of specific price indices or appraised values. These changes would specifically eliminate the calculation of the gain or loss from holding non-monetary assets in Mexico s inflationary economy. However, the evidence presented here demonstrates that holding gains/losses are relevant for determining firm value in Mexico over a wide range of inflation rates and therefore eliminating these adjustments would result in a loss of important information. This research is of particular importance given the current debate surrounding the potential acceptance of International Accounting Standards (IAS) by the US Securies and Exchange Commission (SEC). Pownall and Schipper (1999) suggest that studies documenting significant associations between stock market variables and country-specific characteristics of non-us GAAP systems, such as inflation, may speak to the issue of comprehensiveness one of the three main creria to be used by the SEC in assessing the acceptabily of IAS. The implication from this research is that any attempt to standardize accounting practices across the US, Canada, and Mexico by eliminating inflation adjustments would lead to less comprehensive standards from the perspective of Mexican investors. One objective of the International Accounting Standards Commtee (IASC) is to promote worldwide acceptance of IAS while at the same time harmonizing regulations, accounting standards, and procedures relating to the presentation of financial statements (IASC, 1998). If the outcome of this process is the elimination of useful (i.e., value-relevant) information, then the standards will fail to reflect real economic differences across countries. Mexico provides a unique setting for investigating the value relevance of general price level-adjusted (GPLA) and current cost accounting disclosures because of s hyperinflationary economy. Inflation rates in Mexico fell from 132 per cent and 114 per cent in 1987 and 1988, respectively, to around 20 per cent in 1989 and The variation in inflation rates during this period allows tests of how the magnude of inflation affects the value-relevance of GPLA accounting disclosures. I estimate an inflation-adjusted accounting valuation model that explains a significant portion of the cross-sectional variation in the market-to-book ratios of Mexican firms. The results indicate that holding gains in particular are informative for valuing Mexican equies even when inflation rates are relatively low. These results are of particular interest now because of the proposed accounting changes in Mexico as a result of NAFTA.

3 262 P. Y. Davis-Friday The remainder of the paper is organized as follows. Section 2 reviews related research and Mexican financial reporting requirements. Section 3 develops the accounting valuation model in the context of an inflationary economy, derives hypotheses regarding the model s coefficients, and outlines the sample selection procedures. Section 4 provides descriptive statistics of the data and empirical results. Section 5 summarizes the paper. 2. Mexican Accounting and Financial Reporting Mexican GAAP has already been heavily influenced by US accounting practices and financial statements of the two countries are similar in many respects. The primary difference between the two countries accounting principles is the restatement for inflation. 1 Mexican inflation accounting is the primary focus of this study because of s implications for the valuation of Mexican firms and because the Mexican Instute of CPAs has proposed changes that would eliminate several current cost disclosures. Since 1984, Mexican accounting principles have required all primary financial statements to be restated to reflect the effects of inflation, but revaluing fixed assets was a common practice prior to the first standard on inflation accounting. Mexican companies use a comprehensive current cost/constant peso method of presenting comparative financial statements that requires the balance sheet and income statement to be restated in terms of constant pesos as of the balance sheet date using the National Consumer Price Index (NCPI). Addionally, publicly held companies make current cost adjustments (revaluations) for inventories, depreciable assets, and related expenses (cost of goods sold and depreciation). They are required to use the lower of replacement cost or net realizable value for inventories and their fixed assets can be revalued by independent appraisal or by the use of specific price indices. 2 Rivera (1987) examines the relation between market returns and various combinations of price level/replacement cost earnings measures as substutes for historical earnings during a period when both sets of data were available in Mexico ( ). He finds that inflation-adjusted earnings are better predictors of returns than historical cost earnings, but his results regarding the incremental information content of inflationadjusted earnings beyond historical cost earnings are mixed. Gordon (2000) examines the value-relevance of adjustments between historical cost financial statement data and price level and replacement cost-adjusted data using a sample of Mexican firms during the period Since

4 Equy Valuation and Current Cost Disclosures 263 historical cost data have not been available in Mexican financial statements since 1983, she estimates as-if-disclosed historical cost accounting data and finds that price level/replacement cost adjustments to book value and earnings are incrementally value relevant beyond the estimates of historical cost book value and earnings. This study differs from Gordon (2000) in that investigates the existence of a significant relation between the inflation accounting data recognized in the body of Mexican financial statements, in particular holding gains, and the value of the firms publicly traded equy in periods of high and low inflation. While this research does not attempt to assess the value relevance of current cost disclosures relative to historical cost disclosures, does investigate the value relevance of the individual components of current cost earnings an issue not addressed in Gordon (2000). The results indicate that holding gains are value-relevant throughout the sample period in spe of substantial decreases in Mexican inflation rates. 3. Research Design 3.1 Inflation Variables The primary research question in this paper is whether the current cost/constant peso disclosures currently provided in Mexican financial statements are value-relevant regardless of the level of inflation. If they are, then the results provide direct evidence against the efficacy of the proposed policy changes in Mexico that would eliminate those disclosures. Therefore, the analysis focuses on the valuation role of current cost/constant peso earnings and book value found in the body of Mexican financial statements. Current cost/constant peso earnings are a combination of current cost operating income (CCOI), purchasing power gains (PPG), net interest expense (NINT), net exchange losses (NEXL), and the annual increment to holding gains ( HG). Current cost operating income (CCOI) consists of sales or revenues less the cost of goods sold and current cost operating expenses. For simplicy, Mexican firms generally assume that all transactions take place at month end, and monthly figures are updated for general inflation from the date of the sale to the balance sheet date. Plant, property, and equipment, as well as construction in progress, are stated at their current cost using the specific cost method, and the related depreciation expense is based on the current cost of the related assets.

5 264 P. Y. Davis-Friday Purchasing power or monetary gains/losses (PPG) result from constant peso accounting. Under Mexican GAAP, the effect of the general inflation rate on a company s net monetary posion (cash and receivables less current liabilies) during the accounting period is included in income as a component of the company s net financing costs. For example, as inflation slows, the monetary gain/loss will shrink. 3 Having monetary or purchasing power gains indicates that the company is in a net debtor posion. The amounts that borrowed in a previous period are being repaid later wh pesos which cost less than they did at the origination of the debt. This will make the company s borrowing cheaper to service. Therefore, overall profabily will be increased in the long run provided rising interest expense does not offset the shrinking cost of the debt. Purchasing power gains (gains on net monetary posion) are correlated wh nominal interest expense. Both are predictable based on their common component: the expected rate of inflation. The monetary gain or loss offsets nominal financing cost (interest income and expense and exchange gain/loss) to produce real financing cost. 4 Nominal net interest expense (net of interest revenue) (NINT) is a function of the firm s net debt posion and s expected average nominal interest rate. The interest expense represents the actual cash cost of financing. If inflation is anticipated, lenders raise interest rates to compensate for the diminishing value of their principal. The monetary gain from borrowing during a period of inflation offsets the inflation premium that lenders charge. Net interest expense is included in the income statement as a component of the company s net financing cost to recognize that interest earned on monetary assets and interest paid on monetary liabilies include compensation to the lender for the loss of purchasing power. Net exchange gains or losses (NEXL) are also included in the cost of financing and they are calculated by translating monetary assets and liabilies, denominated in foreign currencies, into pesos at the exchange rate in effect at the end of the month. Transaction gains and losses on long-term debt and other long-term monetary ems are included in income of the period in which the exchange rate changes. During the sample period, a portion of the recorded exchange differences is determined based on the economic theory of exchange rates known as technical or purchasing power pary. 5 The combined net interest expense, foreign currency, and monetary (purchasing power) gain or loss approximate the company s real (inflation-adjusted) financing cost.

6 Equy Valuation and Current Cost Disclosures 265 One important component of current cost accounting is not included in current cost earnings ( does not flow through the income statement). The excess value of a non-monetary asset determined by the specific price method over the asset value based on general price movements represents an unrealized holding gain that is not included in current cost income. It is instead included in an account tled surplus or defic on the restatement of capal and is reported directly in stockholders equy. This variable is of particular importance because the Mexican Instute of CPAs has proposed eliminating non-monetary asset revaluations based on specific price indices and appraisals in response to recent decreases in the Mexican general inflation rate. Under the current accounting principles, a cred balance (a surplus or gain) in this account represents the amount by which the specific values of fixed assets and inventories exceed their historical cost adjusted by the NCPI. A deb balance (a defic or loss) simply indicates that the replacement cost of these assets increased less than the general rate of inflation. Because of this accounting convention, holding gains (losses) might exist no matter what the level of general inflation. Therefore, is possible that this variable might be valuerelevant even in the absence of general inflation. Given s potential importance in valuation, s elimination may make Mexican financial statements as uninformative wh respect to holding gains as those in the US and Canada are. 3.2 The Accounting Valuation Model I address the issue of firm valuation in an inflationary economy by using a framework introduced in the recent accounting-based valuation lerature. Edwards and Bell (1961) and Ohlson (1995) argue that given the assumption of clean surplus accounting, the value of the firm equals s current book value plus the present value of s expected future excess earnings. In empirical work, Easton et al. (1993) use the following model that expresses the market value of equy as a function of book value and earnings in order to investigate the value-relevance of revaluations of tangible long-term assets: * * P = α + BV + α E + ε 0 1 (1) where P is the market value per share of common equy of firm i at time t, BV is the book value per share of common equy, E is the earnings per

7 266 P. Y. Davis-Friday share information in the financial statements, and ε is the unexplained portion of price per share. Their empirical analysis is based on the work of Ohlson (1995), however neher the Ohlson study, nor subsequent empirical studies which use Ohlson s accounting-based valuation technology, address the issue of specific price level adjustments as suggested by Edwards and Bell (1961). This issue is investigated here by incorporating current cost and general price-level adjusted data provided in Mexican financial statements into the valuation model. To maintain the clean surplus assumption under inflation accounting, the annual increment to holding gains ( HG) must also be included as a component of net income. 6 Wh these modifications, the mechanics of clean surplus accounting perm value to be wrten in terms of accounting information whout making any assumptions about the amount of or lack of conservatism in the accounting numbers. 7 Accounting book value in Mexican financial statements is already adjusted for increases in the general price level. To investigate the valuerelevance of current cost disclosures, I decompose earnings into current cost components (current cost operating income, purchasing power gains/losses, net interest expense, net exchange gains/losses, and holding gains). Decomposing the earnings figure also eases the restriction that all earnings components maintain the same degree of persistence. Lipe (1986) demonstrates that market participants implicly assign different multiples to the components of earnings when assessing the market value of common equy. The multiplier on each component of earnings reflects differences in the information that the component contains about the firm s future earnings potential. Starting wh equation (1), a number of empirical models are developed that examine the valuation implications of constant dollar and current cost disclosures whin an accounting valuation framework. In s most basic form, the current market value can be expressed as a function of inflationadjusted book values and disaggregated inflation-adjusted earnings as follows: BV CCOI PPG NINT NEXL HG = α 0 + α1 + α 2 + α 3 + α 4 + α 5 + ε (2) BV BV BV BV BV where is the year-end market value of common equy of firm i at the end of year t, BV is the current cost/constant dollar book value

8 Equy Valuation and Current Cost Disclosures 267 of equy, CCOI is the current cost operating income, PPG is purchasing power gains, NINT is net interest expense, NEXL is net exchange loss, and HG denotes the change in unrealized holding gains. 8 Equation (2) scales all variables by the book value of equy at time t to control for differences in firm sizes. The dependent variable is therefore the market-to-book ratio. If the inflation adjustments reflect value-relevant changes in equy and income, then the ratio of price to inflation-adjusted book value should be closer to one than whout constant dollar/current cost accounting and variation in the inflation accounting variables will explain variation in the price-to-inflation adjusted book value ratio. The effect of unrecorded net assets is captured in the intercept (the coefficient on the scaled book value, α 0 ) and is often used as a measure of the degree of accounting conservatism (Feltham and Ohlson, 1995). Since the intercept does not vary across firms, this specification assumes that conservatism is eher constant across firms or completely reflected in the form of higher abnormal earnings whin the forecast horizon. The model also assumes that discount rates, which are embedded in the coefficient estimates, are constant across firms. The use of annual cross-sectional regressions restricts coefficients to be constant across firms in a given year but allows them to vary across years. As each component of current cost/constant dollar operating income has s own coefficient, this estimation method allows the coefficient of each earnings component to vary cross-sectionally as a function of the individual effect of inflation on firms net assets. 3.3 Regression Analyses and Hypothesis Development Current cost operating income (CCOI) is the primary component of inflation-adjusted earnings. The estimated coefficient is expected to be posive and to vary wh the persistence of earnings. Since the coefficient on PPG captures both the persistence of inflation and expected gains in future profabily, the expected sign of s estimated coefficient is ambiguous. The estimated coefficient on NINT should capture the effect of nominal interest rates on the value of equies; therefore, is expected to be negative. If recognized exchange losses (the variable is defined in terms of net exchange losses) actually reflect free exchange rate movements, then the coefficient on NEXL should capture the effect of market exchange rate fluctuations on asset values and is expected to be

9 268 P. Y. Davis-Friday negative. However, if the Mexican accounting convention of adjusting for the difference between market rates and the technical pary exchange rate introduces noise into the measurement of net exchange gains/losses, the expected sign of the coefficient is ambiguous. Finally, if the replacement cost of assets has increased by more than general inflation, may be because the assets can be used to earn more money and thus someone is willing to pay more for them. If this is true, a holding gain may indicate a future increase in a company s cash flows and earnings. Alternatively, a holding gain can be viewed as unfavorable since represents a price increase (when ems on hand are used, replacing them will cost more). Likewise, a holding loss does not necessarily indicate a decline in profs and cash flows. On average we would not expect specific asset prices or price changes to differ from general price changes. However, while the coefficient on the increment to holding gains in the valuation model may be different from zero, the expected sign of the coefficient is ambiguous. The balance in this account is cumulative; therefore the change in the balance from one period to the next is the relevant variable for investigation in terms of current cost earnings. Three research questions are investigated in this study. The first question is whether the components of current cost/constant peso earnings and the book value of equy are value-relevant. The components of current cost earnings are related so that their individual effects on price are suppressed by including them separately in the regression models. The primary valuation model includes each of the components separately, however, Edwards and Bell (1961) suggest that one of the values of current cost accounting lies in distinguishing between income from everyday business operations (operating income) and that derived from holding assets (holding gains). Therefore, the second valuation model aggregates all of the components except holding gains into current cost/constant dollar earnings (defined as the combination of current cost operating income (CCOI), net interest expense (NINT), net exchange losses (NEXL), and purchasing power gains (PPG)). I test the hypothesis that if current cost operating earnings have different implications for future profabily than holding gains then the market will weight current cost earnings differently than holding gains. The first hypothesis, stated in the null form, follows: H1: The market will price current cost operating earnings and holding gains equally (β 1 = β 2 in equation (3) below).

10 Equy Valuation and Current Cost Disclosures 269 A reduced valuation model, estimated in terms of aggregated current cost/constant dollar income and holding gains, offers a direct test of the first hypothesis: BV CCOI PPG NINT NEXL HG = β0 + β β2 + ε BV BV BV BV BV (3) where all variables are defined as in equation (2). 9 Finally, the second hypothesis is about the relevance of the components of comprehensive financing cost. Mexican accounting principles suggest that the net exchange loss together wh net interest expense comprise nominal financing cost. Addionally, to the extent that inflation is expected, nominal financing cost and purchasing power gains/losses should offset to produce the real cost of financing. This is true because the difference between nominal financing cost and purchasing power gains/losses should be the unanticipated portion of inflation. That is, if the expected rate of inflation is properly measured, the unexpected gain/loss would be zero and an ex-post gain/loss in a given year would not be expected to recur. The common components should completely offset, and the residual should receive a coefficient of (negative) one because the expost gain/loss would affect market value of equy dollar for dollar. Equation (4) offers a direct test of this hypothesis: CCOI PPG NINT NEXL HG = γ 0 + γ 1 + γ γ 3 + ε BV BV BV BV BV BV (4) where all variables are defined as in equation (2). H2: If comprehensive financing cost measures real financing cost, then γ 2 0. These tests constute the primary empirical analyses regarding the relation between the market value of equy and current cost accounting disclosures of the sample of Mexican firms. 3.4 Sample Selection The data used in this study come primarily from the Mexican Stock Exchange (Bolsa). The Bolsa is a private instution that is owned equally by s member brokerage houses. At the close of 1990, the aggregate market capalization of stocks traded on the Bolsa totaled billion

11 270 P. Y. Davis-Friday pesos (40.94 billion dollars), representing an increase of per cent over the previous year. The Bolsa is much smaller and less liquid than the large markets in the US. At the end of 1990, there were only 217 companies (excluding mutual funds) listed on the exchange. 10 Stock price data were acquired from the Anuario Bursátil (Annual Market Statistics). Financial statement data were acquired from the Anuario Financiero (Annual Financial Data). Where possible, these data have been verified directly wh firms financial statements and annual reports. The sample consists of all firms listed on the Mexican Stock Exchange for which financial statement and stock price data are available during each of the years Therefore, only firms wh data available during the entire sample period are used. The 44 sample companies investigated in this study are drawn from both the industrial and service sectors of the Mexican economy. Banking instutions, insurance companies and brokerage firms are eliminated since they have different regulatory and reporting requirements than the rest of the firms. Although the sample is small in absolute terms, represents over 35 per cent of the entire population of industrial and service firms listed on the Mexican exchange, and almost 25 per cent of the total population of firms listed on the exchange. In 1987, the total market capalization of the sample firms is 10 billion Mexican pesos, which accounts for more than 50 per cent of the total Mexican market capalization. In 1988, 1989, and 1990, the total market capalization of the sample is 13.5, 24.2, and 38.6 billion pesos, respectively. The sample firms represent 38.6 per cent of the total Mexican market capalization in 1988, 33.5 per cent in 1989, and 31.4 per cent in Empirical Results 4.1 Descriptive Statistics Panel A of Table 1 details the sample selection procedures and panel B lists the industry membership of the sample. The most frequent cause for eliminating a firm from the sample is the lack of a recent stock price or missing stock prices for firms wh multiple series of shares. If the firm s shares have not traded for several weeks before the end of the year, the Bolsa does not report the share price. 12 Table 2 provides annual descriptive statistics for the sample data. These data are expressed in terms of Mexican pesos and the notes to Table 2 contain the average and year-end exchange rates during the sample period. The mean (median) market

12 Equy Valuation and Current Cost Disclosures 271 Table 1. The Sample Panel A: Sample Selection Procedures Number of enties listed on the Bolsa a Less: Financial instutions b Mutual funds Brokerage firms Commercial banks Total industrial and service firms Missing financial statement data c Missing stock price data d Missing prior year(s) data N/A Final sample a This total includes only those enties wh equy securies (excludes debentures). b Financial instutions include insurance, bonding, leasing, and financial holding companies. c The Anuario Financiero (Annual Financial Data) does not contain the required financial statements. d The Anuario Bursátil (Annual Market Data) does not contain the required price data for all share series. Panel B: Industry Distribution of the Sample SIC Industry Name Total Firms 10 Mining and Extractive Industries 2 15 Building Construction General 3 20 Food and Kindred Products 4 22 Textile Mills Products 4 26 Paper and Allied Products 5 27 Printing and Publishing 1 28 Chemical and Allied Products 3 32 Stone, Clay, and Glass Products 1 33 Metals 4 35 Machinery excluding Electric 3 36 Electrical and Electronic Equipment 3 37 Motor Vehicle Parts and Accessories 2 38 Instruments and Related Products 1 40 Transportation 1 45 Transportation by Air 1 49 Electric, Gas and Water Utilies 1 53 Department Stores 2 70 Hotels and Motels 1 99 Various/Unclassifiable 2 Total 19 44

13 Table 2. Sample Descriptive Statistics Variable Mean Std. Dev. Median Mean Std. Dev. Median Mean Std. Dev. Median Mean Std. Dev. Median (228) (255) (131) (307) (387) (146) (550) (682) (296) (878) (1,701) (287) BV (399) (520) (202) (568) (719) (289) (709) (876) (393) (856) (1,043) (498) CCOI (46) (77) (20) (80) (130) (25) (77) (128) (32) (110) (208) (31) PPG ( 51) (144) ( 5) ( 26) (108) ( 3) ( 10) (32) ( 0.37) ( 29) (82) ( 4) NINT (37) (110) (7) (20) (76) (5) (9) (62) (2) (22) (142) (3) NEXL (16) (81) (0) (1) (10) (0) (5) (12) (225) (8) (14) (0.52) HG (0.45) (196) ( 0.49) ( 95) (260) ( 21) ( 126) (325) ( 11) ( 50) (393) ( 44) HG ( 0.62) (152) ( 0.11) ( 96) (180) ( 28) ( 34) (115) ( 9) (76) (268) (24) PB

14 Table 2. Continued Variable Definions, Market value of equy (fiscal year-end stock price times the number of shares outstanding). BV, Year-end constant peso book value of equy (excluding holding gains). CCOI, Current cost operating income. PPG, Purchasing power gains. NINT, Net interest expense. NEXL, Net exchange loss. HG, Cumulative holding gain. HG, Annual changes in the cumulative holding gain. PB, Price-to-book ratio (/BV). 1 There are 44 firms in the sample. The top numbers are deflated by book value. Amounts in parentheses are totals in terms of millions of Mexican pesos. US Dollar to Mexican Peso Exchange Rates ( ) December Average Mexican Inflation Rates ( ) % 114% 20% 30% Source: Statistical Abstract of Latin America (Vol. 30) and the International Monetary Fund s International Financial Statistics Yearbook (1996).

15 274 P. Y. Davis-Friday value of equy of the sample firms increases from 228 (131) million pesos in 1987 to 878 (287) million in This translates into approximately US$103.2 (59.5) million in 1987 and US$297.8 (97.3) in Purchasing power gains and holding gains were on average negative (or losses) throughout the sample period. Purchasing power losses decreased from 1987 to 1989 as did the rate of inflation, but holding gains became losses from 1987 to This implies that specific asset prices failed to keep pace even wh the lower rates of inflation in Univariate tests show that the mean price-to-book value ratio is significantly less than one in 1987 (0.70) and 1988 (0.58), but not significantly different from one in 1989 (0.74) and 1990 (0.91). There are at least two alternative explanations for why the price-to-book ratio is less than one during the beginning of the sample period. It is possible that the market anticipated that future profabily and/or the growth in book value of Mexican firms would not keep pace wh the high levels of inflation. In this case, the ratio would be expected to approach one as the level of inflation moderated. Alternatively, the market might have considered firms accounting methods to be aggressive. In this case, the effect of the aggressive accounting ultimately reverses, leading to lower earnings and book values and higher price-to-book ratios. 13 It is not clear from the tests performed here which explanation best describes the Mexican market during the period. Table 3 reports pooled Pearson and Spearman correlation coefficients among the earnings components, the book value of equy, and the market value of equy. The market value of equy and book value of equy are significantly posively related. Current cost operating income is significantly posively associated wh the market value of equy. There are also significant associations among the components of current cost earnings, especially current cost operating income and net interest expense. Because of these associations, the estimation of equation (2) may be affected by collineary. 4.2 Valuation Models The results from the cross-sectional regressions are summarized in Tables 4 through 6. 14,15 Under comprehensive revaluation, and assuming no measurement error, current cost accounting should display a straightforward relationship to the market value of the firms common stock. The ratio of the market value of common equy to the book value of common equy would be one. Empirically, this means that one would expect the

16 Equy Valuation and Current Cost Disclosures 275 Table 3. Pooled Correlation Analyses ( ) BV CCOI PPG NINT NEXL HG HG (0.00) (0.00) (0.22) (0.50) (0.95) (0.19) (0.18) BV (0.00) (0.00) (0.00) (0.00) (0.15) (0.00) (0.42) CCOI (0.00) (0.00) (0.00) (0.00) (0.67) (0.02) (0.21) PPG (0.66) (0.02) (0.24) (0.00) (0.00) (0.65) (0.39) NINT (0.48) (0.00) (0.00) (0.00) (0.00) (0.41) (0.33) NEXL (0.44) (0.06) (0.90) (0.00) (0.08) (0.44) (0.27) HG (0.68) (0.81) (0.35) (0.88) (0.54) (0.57) (0.70) DHG (0.90) (0.61) (0.21) (0.19) (0.03) (0.52) (0.00) Upper (lower) diagonal: Pearson (Spearman) correlation coefficients. Numbers in parentheses represent p-values, two tailed tests (N = 176). Refer to Table 2 for variable definions. coefficients on current book value to be close to one or greater than one if conservative accounting practices are followed. If there is measurement error, or if there are unreported net assets, the coefficient of book value under eher current cost or constant dollar accounting might differ from one. The primary valuation model, equation (2), is estimated in Table 4. The estimated coefficient on the current book value of equy (measured as the intercept) is posive and significant in all years, but is significantly less than one. Across all years, the coefficient on CCOI, when significant, is greater than zero as expected. The coefficient on PPG is significantly less than zero except in 1990 when is not significantly different from zero. This result is consistent wh the expectations regarding the persistence of inflation and the magnude of interest expense overwhelming expectations regarding future profabily generated by cheaper borrowing costs. This explanation is also supported by the consistently negative coefficient on net interest expense (NINT). Net exchange losses (NEXL) are measured and reported using the theory of Purchasing Power Pary (PPP) in an attempt to align exchange rates between the US and Mexico based on the inflation rate differential between the two countries. In 1987, when inflation rates in Mexico were

17 276 P. Y. Davis-Friday Table 4. Association between Market-to-Book Ratios and the Components of Inflation-Adjusted Earnings BV CCOI PPG NINT NEXL HG = α 0 + α1 + α 2 + α 3 + α 4 + α 5 + ε BV BV BV BV BV Adj. α 0 α 1 α 2 α 3 α 4 α 5 R 2 Obs Coefficient estimate (t-statistic) (6.10) ( 0.13) ( 1.69) ( 2.29) ( 1.98) (4.90) p-value Coefficient estimate (t-statistic) (6.56) (2.95) ( 2.41) ( 3.02) ( 1.16) (2.31) p-value Coefficient estimate (t-statistic) (2.97) (3.69) ( 1.95) ( 2.73) ( 0.61) (0.64) p-value Coefficient estimate (t-statistic) (2.33) (4.45) ( 0.48) ( 2.01) (1.19) (3.80) p-value All variables are defined in Table 2. The data are scaled by book value. the highest during the sample period, the coefficient on NEXL is significantly less than zero as hypothesized. During the remainder of the sample period, the coefficient is not significantly different from zero. This result suggests that the controlled exchange rate system in Mexico caused the largest discrepancy between exchange rates in the US and Mexico when inflation rates were high, therefore warranting the PPP adjustment. As inflation rates decreased, the PPP adjustment to net exchange losses simply introduced noise into the measurement of exchange gains and losses so that NEXL no longer provides significant explanatory power for the market value of Mexican firms. Finally, the coefficient on the annual change in holding gains ( HG) is significantly greater than zero, except in Taken together wh the results for CCOI, appears that there may be a valuation trade-off between the current change in holding gains and CCOI in 1987 and In determining firm value in 1987, the market places more weight on holding non-monetary assets than income from operations (the coefficient

18 Equy Valuation and Current Cost Disclosures 277 on CCOI is not significantly different from zero while the coefficient on HG is significantly greater than zero and not statistically different from one). In 1989, when the inflation rate is the lowest in the sample period, there is a significant relation between market value and CCOI, while the relation between market value and the change in holding gains is not significant. These results are also consistent wh the contention that the recent change in holding gains provides information about future changes and therefore reflects inflation expectations. If the decrease in inflation from 1988 to 1989 was unexpected, then holding gains might have temporarily lost their value-relevance. In 1990, inflation rates are similar to those in 1989 and the change in holding gains is once again significantly related to market values. 16 To investigate the relative value-relevance of earnings and holding gains and to migate the effects of collineary among the earnings components, equation (3) aggregates all current cost earnings components, except holding gains, into a single explanatory variable. The results from the estimation of equation (3) are presented in Table 5. The coefficient estimates on the scaled book value of equy and holding gains are posive and significant throughout the sample period. In 1989 the estimated coefficient on current cost earnings (consisting of CCOI, PPG, NINT, and NEXL), is not significantly different from zero. Again, the significant drop in the inflation rate from 1988 to 1989 appears to affect the results. This year is particularly interesting since the coefficient estimate on holding gains, which is consistently between zero and one in 1987, 1988 and 1990, is greater than one in Unlike equation (2), there is a significant relation between market values and holding gains but not between market value and operating earnings. This difference most likely results from the considerable multicollineary present in equation (2) and because the holding gains are only indirectly related to the general level of inflation. Hypothesis 1, which asserts that current cost earnings are valued the same as holding gains, is supported in the annual estimations. However, in 1989 the coefficient on current cost earnings is not significantly different from zero, but the coefficient on the change in holding gains is. Equation (4) measures the relation between comprehensive financing cost and the market value of equy while controlling separately for current cost operating income and holding gains. The results, presented in Table 6, are consistent wh those from the estimation of equations (2) and (3). However, equation (4) helps disentangle the effect of the significant drop in the inflation rate from 1988 to 1989 on the regression coefficients.

19 278 P. Y. Davis-Friday Table 5. Incremental Abily of the Change in Holding Gains to Explain Market-to-Book Ratios Beyond Aggregate Inflation-Adjusted Earnings BV CCOI PPG NINT NEXL HG = β0 + β β2 + ε BV BV BV BV BV β 0 β 1 β 2 Adj. R 2 Obs Coefficient estimate (t-statistic) (4.99) (2.25) (1.78 p-value H 0 : β 1 = β 2 F-value: 0.26 Prob > F: Coefficient estimate (t-statistic) (6.11) (2.74) (2.01) p-value H 0 : β 1 = β 2 F-value: 0.13 Prob > F: Coefficient estimate (t-statistic) (6.28) (1.30) (2.40) p-value H 0 : β 1 = β 2 F-value: 0.36 Prob > F: Coefficient estimate (t-statistic) (4.08) (2.00) (4.16) p-value H 0 : β 1 = β 2 F-value: 2.05 Prob > F: 0.16 All variables are defined in Table 2. The data are scaled by book value. The coefficient on CCOI in equation (4) is similar in magnude to the coefficient from the estimation of equation (2). Addionally, the coefficient on the change in holding gains is significantly different from zero, as is in equation (3). Only in 1989 is comprehensive financing cost (consisting of PPG, NINT, and NEXL) not significantly related to the market value of equy. In all other years, hypothesis 2 is supported and the coefficient on comprehensive financing cost is negatively related to market value as expected. Comprehensive financing cost failed to measure the real cost of financing when the inflation rate dropped and therefore lost s association wh the market value of equy. Across the annual specifications, the adjusted R 2 s decrease only marginally (if at all) from

20 Equy Valuation and Current Cost Disclosures 279 Table 6. Association Between Market-to-Book Ratios and Current Cost Operating Income, Comprehensive Financing Cost, and the Change in Holding Gains BV CCOI PPG NINT NEXL HG = γ 0 + γ 1 + γ γ 3 + ε BV BV BV BV BV γ 0 γ 1 γ 2 γ 3 Adj. R 2 Obs Coefficient estimate (t-statistic) (6.23) (0.13) ( 2.71) (4.93) p-value Coefficient estimate (t-statistic) (7.83) (3.37) ( 3.35) (3.21) p-value Coefficient estimate (t-statistic) (7.17) (5.44) ( 1.23) (2.18) p-value Coefficient estimate (t-statistic) (3.57) (4.20) ( 2.30) (3.71) p-value All variables are defined in Table 2. The data are scaled by book value to This is consistent wh the explanatory power of the model being robust to changes in the rate of inflation. Given the small sample size and the relations among the independent variables, may not be reasonable to aggregate several of the independent variables, thereby restricting their coefficients to be the same. Therefore, the results from these analyses are viewed primarily as descriptive given the limations of the data. 4.3 Sensivy Analyses Analysis of price-to-book ratios focuses on the extent to which inflation adjustments align book and market values but does not consider whether Mexican GAAP recognizes changes in value. I estimate a returns model on the pooled sample of data to investigate whether the results are consistent across levels and changes specifications. Since stock prices are

21 280 P. Y. Davis-Friday serially correlated, the data cannot be pooled across years in the levels. Therefore the data are first differenced and deflated by beginning-ofperiod prices: R BV CCOI PPG NINT = δ 0 + δ1 + δ 2 + δ NEXL HG δ 4 + δ 5 + ε 1 1 (5a) where is the difference operator (e.g., R = ( 1 )/ 1 ). Pooling the data substantially increases the sample size and testing the abily of the model to explain annual returns should help to confirm the results from the levels estimations. Equation (5a) provides a direct comparison wh equation (2). Equation (5b) is an alternative estimation of equation (5a) that attempts to reduce the effects of the collineary among the earnings variables (similar to (3)). The model is estimated in the following form: R BV [ CCOI + PPG + NINT + NEXL ] = φ0 + φ1 + φ 2 HG ε (5b) Finally, equation (5c) examines the importance of the comprehensive financing cost specification as in equation (4): R BV CCOI [ PPG + NINT + NEXL ] = λ0 + λ1 + λ2 + HG λ ε (5c) Equations (5a), (5b), and (5c) attempt to explain annual changes in the market value of equy based on the annual change in the book value of equy and the components of current cost earnings; the results are presented in Table 7. In the estimation of equation (5a) only the change in current cost operating income is not significantly related to the change in market value. If this model is affected by multicollineary as equation (2) is, then the estimation of the model in terms of aggregate earnings (the combination of current cost operating income, purchasing power gains,

22 Equy Valuation and Current Cost Disclosures 281 net exchange losses, and net interest expense) and the annual holding gain should reduce the effects of the collineary. The estimation of equation (5b) examines the relation between changes in the market value of equy and changes in aggregate earnings and holding gains. The results, presented in panel B of Table 7, are consistent wh those from equation (3). Finally, the results from the estimation of equation (5c) are presented in Table 7, panel C. These results are also similar to those from the annual estimations of equation (3). The sensivy analyses confirm that there is an association between inflation-adjusted accounting data and the market value of the firm. Changes in holding gains are significant explanators of the annual change in the market value of equy. Throughout this period of inial hyperinflation and later moderate inflation, the market appears to value holding non-monetary assets as well as earnings from operations. The importance of holding gains is pervasive across all of the regression analyses. This finding suggests that is important to separate income from operations from holding gains because the market appears to value the information in the annual change in holding gains as much as, and in some cases more than, current cost operating income. This result is especially important in light of the recent iniatives to eliminate the accounting for holding gains in Mexico in order to simplify Mexican accounting principles for US investors (FASB, 1994). The evidence presented here suggests that decreases in the rate of inflation do not significantly affect the relation between holding gains and firm value. 5. Summary and Implications This study investigates the relation between the market value of Mexican firms and their general price level-adjusted and current cost accounting disclosures. The results should help policymakers, specifically the Canadian Instute of Chartered Accountants (CICA), the Mexican Instute of CPAs (IMCP), and the Financial Accounting Standards Board (FASB), who have suggested altering Mexican Generally Accepted Accounting Principles (GAAP) to make them more comparable to US GAAP. This study documents that inflation adjustments, impounded in the book value of equy and captured by current cost operating income, the integral cost of financing, and holding gains, are value-relevant, even when inflation rates are relatively low. These results call into question the efficacy of eliminating a consistently value-relevant disclosure in Mexican financial statements: the gain or loss from holding non-monetary assets.

23 282 P. Y. Davis-Friday Table 7. Sensivy Analyses Pooled Returns Model ( ) Panel A: Returns Model wh the Components of Inflation-Adjusted Earnings R δ 0 δ 1 δ 2 δ 3 δ 4 δ 5 Adj R 2 Obs. Coefficient estimate (t-statistic) (6.76) (0.60) (2.38) (1.10) (3.34) (5.59) p-value Panel B: Returns Model wh Aggregate Inflation-Adjusted Earnings and Holding Gains R φ 0 φ 1 φ 2 Adj. R 2 Obs. Coefficient estimate (t-statistic) (11.86) (8.81) (7.86) p-value Panel C: Returns Model wh Current Cost Operating Income, Comprehensive Financing Cost, and the Change in Holding Gains R BV CCOI PPG NINT NEXL = δ 0 + δ1 + δ 2 + δ 3 + δ HG δ 5 + ε 1 BV [ CCOI + PPG + NINT + NEXL ] HG = φ0 + φ1 + φ2 + ε BV CCOI [ PPG + NINT + NEXL ] HG = λ0 + λ1 + λ2 + λ3 + ε λ 0 λ 1 λ 2 λ 3 Adj. R 2 Obs. Coefficient estimate (t-statistic) (7.50) (0.83) ( 2.56) (6.05) p-value Variable Definions R, Annual change in the market value of equy (fiscal year-end stock price) scaled by the beginning market value. BV, Annual change in ending constant peso book value of equy (excluding holding gains) scaled by the beginning market value. CCOI, Annual change in the current cost operating income scaled by the beginning market value. PPG, Annual change in the purchasing power gains scaled by the beginning market value. NINT, Annual change in the net interest expense scaled by the beginning market value. NEXL, Annual change in the net exchange loss scaled by the beginning market value. HG, Annual change in the cumulative holding gain scaled by the beginning market value.