Historical Cost and Fair Value in the Gold Industry
Heather A. Wier*
University of Alberta
JEL Classification: M41
Keywords: revenue recognition; fair value accounting; historical cost
* Contact information: Heather A. Wier, Faculty of Business, University of Alberta, Edmonton, AB T6G 2R6.
Phone: 780.492.5752. Fax: 780.492.3325. E-mail: firstname.lastname@example.org.
I appreciate the comments and suggestions of Amy Choy, Peter Clarkson, Mark Huson, Tom Scott, and Christine
Wiedman. Additionally, I am indebted to Hemdat Sawh of Grant Thornton LLP for explaining how SAB 101
affected the revenue recognition practices of Canadian gold firms.
Historical Cost and Fair Value in the Gold Industry
Abstract: This research examines the information content of historical cost and fair value
reporting for gold firms. When revenue is recognized at the completion of production, finished
goods inventory of precious metals is marked to market on the balance sheet. Revenue
recognition at the point of completion of production was recently disallowed in Canada. This
represents a significant departure from the trend of current accounting standards, which have
been moving towards requiring mark-to-market accounting. I show that balance sheet data has
better explanatory power for firm value when firms recognize revenue at the completion of
production. Firms’ voluntary disclosure of the “cash cost of gold” provides value-relevant
information to investors; however, due to estimation errors, investors cannot use this disclosure
to convert inventory carried at cost under the sales method to a value-relevant estimate of its
market value. There is no evidence that firms who recognized revenue at the completion of
production did so for opportunistic reasons. I find evidence that income statement numbers are
more relevant for firms that did not recognize revenue as production was complete, but rather
used the traditional point of sale method as the critical event for revenue recognition. My results
highlight the trade-offs standard setters must make when considering the impact of accounting
standards on the value-relevance of the balance sheet versus the income statement.
This research examines the information content of historical cost and fair value reporting for
gold firms. The gold industry has had a long-standing history of relying on industry practice to
justify the recognition of revenue at the completion of production (the production method),
before the identification of a specific purchaser. The rationale for this practice has been that,
because the price of the commodity is relatively stable, selling costs minimal, and a ready market
available, the purchaser can be reasonably certain of the amount of profit to be realized from
production such that the ultimate sale constitutes a mere formality in the revenue recognition
process. The mechanics of this method of revenue recognition effectively mark inventory on the
balance sheet to market. As a result of recent accounting pronouncements, described later in this
section, this practice has changed such that gold producers are now required to carry inventory
on the balance sheet at cost, because they are disallowed from recognizing revenue on refined
gold before the point of sale (the sales method).
I use a sample of Canadian gold producers to study the trade-offs between fair value accounting
and conservatism in reporting. While accounting standard setters are increasingly moving
towards fair value recognition in financial statements, regulators are also eager to stem the tide of
what they perceive to be aggressive revenue recognition. Yet the two objectives are mutually
exclusive, at least in some instances. Clearly, in the case of the gold industry, fair value
reporting of gold inventories causes revenue to be recognized earlier than if inventory were
carried at cost.
Consistent with the stream of literature that examines the value-relevance of fair-value
accounting for balance sheet categories (see for example Barth, Beaver and Landsman’s 1996
study of banks’ fair value disclosures, Petroni and Wahlen’s 1995 examination of fair value
disclosures of property-liability insurers, and Dietrich, Harris and Mueller’s 2000 work on
investment property fair value estimates), I compare the explanatory power of firms’ inventory
of refined gold at cost and fair value for stock price. My tests provide a setting in which I can
test fair value accounting with minimal measurement error because gold trades in regulated
markets at known prices. My research therefore complements the above studies, which examine
categories that are more difficult to measure.
In conjunction with a comparison of the relevance of historical cost and fair value measure, I
also examine the relevance of the historical “cash cost of gold” disclosure. This disclosure
(described more precisely in Section 2) is calculated in accordance with industry standards, and
is reported in the management discussion and analysis (MD&A) section of firm annual reports.
I next turn to the income statement, and compare the value-relevance of income calculated under
the production and sales methods of revenue recognition for stock returns. While one might
expect fair values on the balance sheet to be more informative to the price-setting process than
historical costs, the explanatory power for stock returns of fair value adjustments on the income
statement is not so intuitive. Recording refined gold at market value on the firm’s year-end
balance sheet necessitates an adjustment to the following year’s income statement if gold prices
move, and will therefore trigger a series of transitory adjustments flowing through income unless
the price of gold remains stable from year to year.
As a final issue, I examine whether firms in my sample managed inventory levels in order to
maximize income. Inventory management should differ by accounting method if firms behave
strategically: Firms using the production method can accelerate income recognition merely by
building up finished goods inventory because their point of revenue recognition is the
completion of production, whereas firms using the sales method need to sell off all refined gold
at year-end in order to book maximum revenue.
My findings are as follows: The balance sheet amount of finished goods inventory of gold is
more value relevant for firms marking inventory to market than for firms carrying inventory at
cost. Firms’ voluntary disclosure of the “cash cost of gold” provides value-relevant information
to investors; however, due to estimation errors, investors cannot use this disclosure to convert
inventory carried at cost under the sales method to a value-relevant estimate of its market value.
I find evidence that the income of firms using the sales method of revenue recognition is more
value-relevant than the income of their counterparts who recognize revenue at the completion of
production. Finally, there is no evidence that firms in my sample managed inventory levels
The paper is organized as follows: Section 2 presents the hypotheses development, Section 3
describes the sample selection procedures and presents descriptive statistics, Section 4 presents
the empirical tests, and Section 5 concludes the paper.
2. Background and Hypotheses Development
In December of 1999, the Securities and Exchange Commission (SEC) issued Staff Accounting
Bulletin (SAB) No. 101, “Revenue Recognition in Financial Statements”. This bulletin clarified
and tightened existing rules to permit recognition of revenue on the income statement only when
all of several conditions were met. These conditions include persuasive evidence of a sales
arrangement and the delivery of the product. As a result of the document’s wording, with its
focus on both a sales arrangement and product delivery, the ability of American gold producers
to recognize revenue at the completion of production was eliminated beginning in fiscal 2000.
This movement away from mark-to-market accounting for the gold industry is somewhat of an
anomaly within current GAAP changes, with recent standards emphasizing use of fair values and
valuation on the balance sheet. The SEC’s motivation in SAB 101 was the desire to stem what
the Commission perceived to be an overly aggressive trend in revenue recognition. While there
is no indication in the Bulletin that the SEC intended to target the gold industry, the document’s
wording, in particular the requirement for the seller to demonstrate “persuasive evidence of
delivery” in order to recognize the sale of goods, resulted in a move back to historical cost
accounting in the gold industry.
Canadian gold firms remained free to recognize revenue at the completion of production until
2004. A number of Canadian firms, especially those trading on American exchanges, however,
elected to change from the production to the sales method of revenue recognition as a result of
the SEC bulletin. The option of recognizing revenue at the completion of production has not
been available to Canadian firms for fiscal years beginning after October 2003, when the new
Section 1100 of the Canadian Association of Chartered Accountants (CICA) Handbook, took
effect. In the absence of Canadian accounting pronouncements in a particular area, this
Handbook section gives the accounting standards issued by non-Canadian jurisdictions
preference over industry practice as a source of GAAP for Canadian firms. By 2004, all firms in
my sample used the sales method of revenue recognition, with 1/3 of firms switching from the
production to the sales method.
An unavoidable problem with studies that examine the competing value relevance of different
accounting policy choices is that firms self-select into their accounting policy category. Results
may therefore be driven by characteristics that determine the selection choice, and
inappropriately attributed to the accounting policy.1 The self-selection bias is mitigated in this
study because of the sample firms who switched revenue recognition methods over the study’s
time frame. This allows for comparison of the value relevance of the sales versus the production
method on some of the same firms. As a result, use of this data affords a unique opportunity to
explore various differences between fair value and historical cost accounting with a minimum of
sample selection bias.
Balance Sheet Tests
There are a variety of ways in which researchers can specify the relation between accounting
numbers and firm value. Use of the Ohlson (1995) model is prevalent in the recent literature.
1 A notable exception to the self-selection problem is Bryant (2003) in her examination of the relative value
relevance of successful effort (SE) and full cost (FC) methods in the oil and gas industry. Bryant restates SE firms
to obtain as-if FC figures, and vice versa, and uses a within-firm design to test the relative explanatory power of the
two methods for both stock price and returns.
This model describe returns as a function of the book value of the firm and the present value of
expected future abnormal earnings. The focus of my work is to document and contrast the value-
relevance of fair value versus historical cost reporting for price, using balance sheet disclosures,
and for the timeliness of information incorporated into returns, using income statement
disclosures. Therefore, instead of basing my empirical tests on Ohlson, I conduct my balance
sheet tests using a model that uses exclusively asset and liability data (both on-and off-balance
sheet) similar to the approach used by Barth, Beaver and Landsman (1993), Barth and
McNichols (1994), and bath Beaver and Landsman (1996). My research question is similar to
those of the preceding three papers, all of which focus on the valuation of fair value estimates of
assets or liabilities.2
The value of a gold firm is a function of its net on-balance sheet assets plus the present value of
the profit to be made on the gold reserves it has to extract. Gold firms do not disclose a dollar
amount of mineral reserves, as do oil and gas firms, but the quantity of reserves in ounces is
usually available in the MD&A section of firm annual reports. I thus begin with the following
baseline model to describe the market value of gold producers:
it = ?it + ?1TAit+ ?2TLit + ?3RESERVESitt + ?it
where PRICEit is the market price of the firm’s stock, measured three months after the firm’s
fiscal year-end, TAit,, and TLit , are, respectively, total assets at book value and total liabilities at
2 The three preceding papers examine, respectively, the effect of off-balance sheet pension amounts on share price,
the effect of the estimated fair value of environmental liabilities on price, and the effect of banks’ fair value
disclosures on price
book value, both scaled by the number of commons shares outstanding, and RESERVESit is
calculated as the quantity of year-end proven and probable reserves in ounces multiplied by the
year-end price of gold per ounce less the cash cost of gold per ounce (defined in the following
section), scaled by common shares outstanding.3 If we assume that net price (price less costs to
extract and refine) increases through time so as to exactly offset the discounting (Hotelling
1931), then RESERVESit captures the value of the firm’s reserves at time t.4
For my subsequent main analysis, I focus on empirical models that include the price and cash
cost of gold as separate explanatory variables, and drop the value of the reserves. As described
more fully in Section 3, this variable is highly collinear with other explanatory variables in the
model, and its inclusion also necessitates a reduction in sample size. I include a description in
the results section of how inclusion of the reserves affects the results of my regressions in order
to demonstrate that results are not driven by the omission of this variable.
I begin my primary analysis of the balance sheets of gold producers by documenting the
importance of the market price of gold and the firm’s cost to extract and refine gold (the “cash
cost of gold”) for firm value. I then introduce accounting variables, and compare the value
relevance of my sample firms’ inventory of finished goods carried at cost (the sales method) and
at market (the production method). Finally, I ask whether investors can use the cash cost of gold
3 Proven (probable) reserves are defined as follows by the Canadian Institute of Mining, Metallurgy, and Petroleum:
“A proven (probable) reserve is the economically mineable part of a Measured Mineral Reserve (an Indicated, and
in some cases. Measured Mineral Reserve) demonstrated by at least a preliminary feasibility study.
4 Magliolo (1986) studies reserve recognition accounting for oil and gas firms. Note disclosures of oil and gas
reserves present the present value of these reserves. Magliolo makes the point that such discounting, based on the
current prices and production costs, undervalues the reserves because it does not allow for an increase in net price,
and consequently he performs supplementary analysis using the undiscounted value of the reserves.
in conjunction with the per ounce market price of gold to convert refined gold carried at market
on the balance sheets of sales method firms to a value-relevant estimate of market value.
I structure my analysis in the stages described above because, firstly, if firms’ reported inventory
values possess little explanatory power for balance sheets after the inclusion of more
fundamental value drivers, the choice of inventory valuation method is less crucial. Secondly,
the importance of accounting policy diminishes to the extent that is it possible to undo the firm’s
choice using other disclosures.
Balance Sheet Tests - The Cash Cost of Gold
Two important drivers of value for gold producers are the market price of gold and the firm’s
cost to extract and refine gold. Since gold is a commodity, the former is publicly available and
common to all gold producers. The latter is a firm-specific disclosure that is often available in
the management discussion and analysis (MD&A) section of firm annual reports. I begin with a
discussion of the nature and importance of the cash cost of gold, and also test its explanatory
power for stock prices.
When disclosed, the firm’s per ounce cash cost of gold production is calculated in accordance
with the Production Cost Standard provided by the North American Gold Institute. The cash
cost of gold comprises direct mining costs, stripping and mine development adjustments, third-
party smelting, refining, and transportation costs, by-product credits, royalties, and production