An Accountant's Measurement Primer!

A primary function of accountants is to develop, implement, and monitor measures of economic resources, business processes and utilization of resources by such processes. Wide varieties of stakeholders use these measures in making a range of economic decisions about business organizations. Pension fund managers, as do stockholders and financial analysts, use accounting information reported by companies to assess the economic performance of those companies and form expectations of future performance. Production managers use accounting information to estimate the cost of products they produce and thus, the relatively profitability of producing more of some products and less of others. Executives use accounting information to budget economic resources among operating units. Stakeholder decisions generally have in common the objective of allocating scarce economic resources to their most productive uses, whether it is managers allocating organizational resources among subunits (or alliances of organizations) or capital owners allocating their capital among organizations (or portfolios of organizations). Consequently, it is vital for accountants to understand measurement concepts (theories) in order to achieve accounting's measurement objectives. We start with a working definition of measurement. Measurement is the assignment of numerals and other symbols to represent the magnitude of an attribute of a phenomenon. 2 A phenomenon is an object of interest to those who make decisions based, at least in part, on or about the phenomenon. In business, a relevant phenomenon can be a thing such as a particular asset, a person such as the manager of a business' production plant, and a unit of organization such as a corporation. A phenomenon also can be an event or transaction such as a business' purchase of an asset and a series of interrelated transactions/events such as a business' earning of income. The important point is that the starting point for measurement is the determination of the phenomenon of interest. Many phenomena of interest in business are social constructions - organizations and the processes by which organizations operate. Most social phenomena, however, are exceedingly complex. These phenomena have multiple characteristics or qualities that we will label attributes. Take, for example, the organization known as Starbucks. Starbucks, as an organization, is a phenomenon, but what sense could we make of a request to "measure Starbucks?" Indeed, we require more specificity as to the object of interest - what attribute (or attributes) of Starbucks are of particular interest? This begins a "peeling of the onion," in which attributes of Starbucks as an organization become phenomena of interest in their own right. For example, we might be interested in measuring the quality of Starbucks corporate governance. Thus, corporate governance, an attribute of Starbucks, becomes a phenomenon in its own right. But wait, corporate governance is a complex thing and as such has numerous attributes itself. What are the corporate governance attributes in which we have interest? One such attribute is Starbucks' board of directors. The onion peels again - the board of directors, an attribute of corporate But again, the Starbucks board of governance, now becomes a phenomenon of interest.

I Prepared by Prof. Clifton Brown, Department of Accountancy, University of lIIinois at Urbana-Champaign. All rights reserved, 2007. 2 Modified from Miller and Mosso, p. 29.

27

An Accountant's Measurement Primer

2

directors is a complex thing and has numerous attributes itself. In which attributes of the board of directors do we have interest? Perhaps, one such attribute is the extent to which the board is composed of "independent directors.,,3 So you see, a phenomenon's attributes peel away, becoming phenomena themselves until we reach the level in which our interest lies. That level is only knowable from a clear definition of our objective in measuring in the first place. Returning to our working definition of measurement, magnitude is the extent to which the phenomenon possesses the measured attribute. For example, as of 2007 eight of eleven (72.7%) Starbucks directors meet the definition of independence. "Eight," "eleven," and "72.7% each are numerals. Numerals tend to have greater precision than non-numeral measurement However, non-numeral measurement is possible. Thus, Starbucks disclosure on its corporate governance website states, "Currently, our board has 11 directors, a substantial majority of whom meet all of the independence requirements." "A substantial majority" is a non-numerical measurement that is less precise than is "72.7%" Another aspect of measurement is whether the phenomenon and the attribute(s) of interest are directly observable or must be estimated through substitution. For example, current accounting standards require inventory be valued at (measured at) the lower of its cost or market. Further, FASB Statement of Financial Accounting Concepts No.5, Recognition and Measurement in Financial Statements of Business Enterprises, discusses different measurement attributes of assets and liabilities: historical cost, current cost, current market value, net realizable value, and present value of future cash flows. Let us consider the phenomenon of interest to be a particular item of inventory and the attribute of interest to be the current market value of this item of invent­ tory. Exhibit 1 illus­ trates direct observation and estimation of the relevant phenomenon (the item of inventory) and the relevant attri­ bute (the current market value). Direct observa­ tion of the relevant phenomenon requires observation of the par­ ticular item of in­ ventory on hand (or an identical item). A sub­ stitute for this phenomeExhibit I - Measurement of the Current Market Value of an Item of Inventory Relevant Phenomenon (Item of Inventory) Direct Observation Relevant Attribute (Current Market Value) Measure amount of cash that would be received from selling the item currently in the market Estimation Substitute Phenomenon Estimation Measure amount of cash that would be received from selling a similar item currently in the market and adjust for item differences Estimation Measure amount of cash required to purchase currently a similar item and adjust for item differences

Substitute Attribute

Measure amount of cash paid for the item when it was purchased in the past

3Independent

for purposes of corporate governance generally means that a director does not have a material relationship with the organization other than as a director and a stockholder.

28

An Accountant's Measurement Primer

3

non might entail obser-vation of a similar, but not identical, item of inventory. For example, our inventory item may be a used 200 I Chevrolet Lumina sedan with certain optional equipment (e.g., 4-wheel, anti-lock brake system) and with certain mileage (e.g., 75,247 miles), A substitute phenomenon might be a similar Lumina sedan with different optional equipment and different mileage. Direct observation of the relevant attribute requires observation of the item's current market value (i.e., the amount of cash that the company would receive if it were to sell the item currently). A substitute attribute might entail observation of the phenomenon's historical cost (i.e., the amount of cash the company paid to purchase the item of inventory) or the cost for the company to purchase the item currently (known as current, or replacement, cost). In many instances, direct observation may not be possible, requiring estimation through substitution. For example, our specific Chevrolet Lumina (with its specific optional equipment and its exact mileage) may not have an active market and thus, we may not be able to observe its current market value. In such a case, we would need to measure a substitute phenomenon, a substitute attribute, or possibly both. A substitute phenomenon might be a similar Chevrolet Lumina that does have an observable sales market, which requires us to estimate and adjust for the differences between the observed Chevrolet Lumina and the one we have in our inventory. A substitute attribute might be the item's current purchase cost rather than its current market value. Notice that the essential difference in these attributes is whether the observed prices come from an input, or purchase, market (i.e., current replacement cost and historical cost) or from an output, or sales, market (i.e., current market value and net realizable value).

Qualitative Characteristics of Measures and Measurement
Qualitative characteristics deal with the intrinsic, or essential, nature of a thing. From this per­ spective, economic information only has value if it is useful for making economic decisions. Qualitative characteristics of decision usefulness are attributes with which accountants have a reasonable tradition. FASB Statement of Financial Accounting Concepts No.2, Qualitative Characteristics of Accounting Information, is an example of this tradition. This concept state­ ment enumerates the attributes of decision usefulness: Relevance Relevant information is information capable of making a difference in a decision by helping users to form predictions about the outcomes of past, present, and future events or to confirm or correct prior expectations. An ancillary of relevance is timeliness, having information available to decision makers before it loses its capacity to influence decisions. Reliability Reliability is the extent to which the information represents, or corresponds with, what it pur­ ports to represent. That is, the extent of the information's representational faithfulness. Scientists refer to this characteristic as a measure's content validity. Content validity is the correspondence of the measure with the substance of what we seek to measure. Ancillaries of reliability are verifiability and neutrality. The extent of consensus among indepen­ dent measurers using the same measurement methods is an indicator of the information's verify­ ability. Neutrality means that the accountant should base his or her choice among measures or measurement methods on the concepts of relevance and reliability, and not on the preferences of a particular interested party over the preferences of all interested parties.

29

An Accountant's Measurement Primer Comparability

4

Most economic information has limited value in an absolute sense. Its value is relative; that is, in comparison with a benchmark (for example, similar information about other entities and simi­ lar information about the same entity at other points in time). Consistency in the application of measurement methods over time increases its comparability. I direct readers to FASB Concepts No.2 for more detailed discussion of the qualitative charac­ teristics of accounting information, as well as threshold concepts such as materiality and the costlbenefit nature of information as an economic commodity.
Quantitative Characteristics of Measures and Measurement

Quantitative characteristics deal with what are known as levels of measurement, or forms of quantification. Generally, measurement theorists refer to four forms nominal, ordinal, interval and ratio. Nominal Level of Measurement Some measures refer only to discrete states that have no obvious relation with each other. Counting the instances that a phenomenon of interest falls into such discrete states is known as nominal measurement. States of a phenomenon measured at a nominal level can only be labeled and nothing can be said about the differences between the states other than they are different. For example, gender is a phenomenon with two states - male and female. Thus, in our accounting class we may count that there are 10 female and 13 male students, but such count implies nothing about differences between females and males. Listing on the NYSE is a phenomenon with a variety of possible nominal states - listed and not listed, listed US company and listed non-US company, etc. Likewise, accounting standards require companies to report segmental information, which some companies such as Dell Computers, define as geographic regions. Thus, the categories of "Americas," "Europe," "Asia," etc. represent nominal measurement (although information within categories may be at a higher level of measurement). Ordinal Level of Measurement When the phenomenon of interest has states that can be rank-ordered, such ordering conveys information. However, the state labels do not convey information about the difference between states. An ordinal measure familiar to most students is letter grades. A letter grade of A is better than a B, which in tum is better than a C, and so forth. Letter grades are ordinal measures, but the difference between an A and a B does not necessarily equal the difference between, say, a Band a C. The same is true of bond ratings. For example, Moodys three highest bond ratings are Aaa, Aa and A, but the difference between bonds rated Aaa and Aa is not equal to the difference between those rated Aa and A. Likewise, accounting standards require companies to report any contingent liabilities 4 they may have. The nature of a contingent liability's disclosure is dependent upon the likelihood of the future event occurring upon which the liability is
Contingent liabilities are present obligations that are dependent upon the occurrence, or not, of one or more future events to confirm an attribute of the liability (e.g., the amount, the existence. the payee, etc.). An example of a con­ tingent liability is a product liability lawsuit by customers against a manufacturing company the existence and amount of the liability is contingent on the outcome of the future trial.

4

30

An Accountant's Measurement Primer

5

contingent. Accounting standards indicate that companies should categorize this probability as remote, reasonably possible, or probable. These probability categories are examples of ordinal measurement - probable> reasonably possible> remote, but differences between the categories have no particular meaning. Interval Level of Measurement When states of the phenomenon of interest not only can be rank-ordered, but also the difference between the states has a particular meaning, such states represent an interval level of measure­ ment. The meaning of the state differences is that the interval between adjacent states is con­ stant. Another attribute of interval measures is no natural zero point exists (i.e., whatever zero point the measure uses is arbitrary - it has no unique meaning). An interval measure familiar to most people is temperature. Seventy degrees Fahrenheit is less than 80°F, which in tum is less than 90°F. Additionally, the difference between these adjacent states also has meaning - it is 10°F. Although interval measures abound in the physical sciences, they are less common in the social sciences (which inc1ude economics and accounting). Sometimes social scientists dissemble with their measures, calling them interval measurement when they may not be. For example, say we measure customer satisfaction asking customers to rate some aspect of our product or service on a five-point scale such as:
2 Not At All Satisfied 3

4

5
Completely Satisfied

While it is unlikely that the difference between people responding 5 and 4 (a value of one) is the same as the difference between people responding 1 and 2 (also a value of one), many mar­ keting researchers will pretend the differences are the same in their analysis and presentation of the data. Ratio Level of Measurement When the states of a phenomenon have all the properties of an interval scale, plus have a natural zero point, such states represent a ratio level of measurement. Examples of ratio level measure­ ment abound. Weight, for example, is a phenomenon that we can measure using a ratio scale the concept of zero weight has meaning. Thus, a man weighing 200 pounds weights twice as much as a woman weighing 100 points. Many business phenomena have the properties of ratio level measurement. For example, a com­ pany with $3 bil1ion in cash and cash equivalents has twice as much of this phenomenon as a company with only $1.5 billion in cash and cash equivalents. Likewise, a company with $150 million annual net income has three times more net income than does a company with only $50 million annual net income. We must take care, however, with the manner in which we measure these phenomena different measurement attributes will result in non-comparable measurements that are not on a common ratio scale. For example, measuring marketable securities in terms of current market value while measuring property, plant and equipment in terms of un-depreciated historical cost would suggest the addition ofthese two asset measures (in a balance sheet) may not have meaning.

31

An Accountant's Measurement Primer 7 variable either reduces or changes the association between the other variables, or the intervening variable increases or exacerbates the association between the other variables. For example, a supervisor's insults to a subordinate may invoke hostility in the subordinate, but the subordinate's self-interest (in career management) may moderate his or her otherwise hostile response to the supervisor. Self-interest is a moderating intermediary variable in this instance. In another example, assume a company pays an employee a fixed wage of $15 per hour, expecting the employee to complete correctly a minimum number of tasks. Modify this situation by introducing a bonus for each correctly performed task above some number. The subordinate's self-interest now increases his or her task performance. Again, self-interest is an intermediary variable, but in this instance, it amplifies the relation between level of performance and compensation. Causal Relationships Causal relationships allow us to divide variables into what social scientists refer to as inde­ pendent and dependent variables. An independent variable is the cause and a dependent variable is the affected variable. The independent variable produces an effect in the dependent variable. For example, an aspirin (the independent variable) taken orally reduces the pain in my head (the dependent variable). Aspirin and pain in the head are in a causal relation in which taking the aspirin causes the headache to diminish (in a leading temporal relation). Likewise, calibrating a pin-setting machine in the manufacture of integrated circuits causes the number of pins the machine will set on an integrated circuit (i.e., 36 pins versus 72 pins). Scientists formally model causal relationships using a variety of mathematical techniques, such as regression. When modeling causal relationships, scientists generally divide the measured variation of the dependent variable (or variables) into three categories: • Variation directly related to each independent variable by itself, which measures the strength of the relationship between each independent variable and the dependent variable; • Variation related to the interaction of the independent variables (i.e., the modifying or amplifying effects of independent variables in combination), which measures the strength of the relationship between the independent variable combinations and the dependent variable; • Variation caused by measurement error, omission of relevant independent variables, and random fluctuation in the dependent variable. The omission of independent variables raises the concepts of necessity and sufficiency. In complex causal relationships, some independent variables may be necessary for a particular effect to occur, but not sufficient. That is, other independent variables must also be present for the effect to occur. In other situations, an independent variable may be sufficient, but not necessary. That is, the independent variable may be adequate to produce the effect, but other independent variable also will produce the effect without the first independent variable. In still other situations, an independent variable may be both necessary and sufficient. Probabilistic Relationships So far, all the relationships are of the form, "Under conditions ZI, Z2, ... , Zn, Y will occur" (either because Y is associated with the Z or the Z cause V). However, in many, if not most,

33

An Accountant's Measurement Primer

8

situations, such a statement is too strong. The best we can say is that "Under conditions Zt, Z2, .. . , Zn, Y will occur with probability p." The earlier statement is known as a deterministic statement the dependent variable (Y) is determined completely by the independent variables (Zn). In other words, the probability of the dependent variable given the independent variable is one. The later statement is known as a probabilistic statement - the probability of the dependent variable given the independent variable is some chance p, where 0 < p < 1. For example, if! draw one card from a fair, 52-card deck, the probability I will draw an ace is 0.0769. Drawing a card and the value of that card are variables with an associational relationship; unless I am a card cbeat, in which case it may be causal (but then no longer a probabilistic relation, rather a deterministic one). What is the market value of Pfizer's Lipitor patent, which it introduced in 1997? Lipitor is the leading prescription drug on the market with sales of $6.5 billion for 2003. Assuming production and distribution costs of about 55%, Lipitor earns a margin of $2.9 billion per year. Even if one discounts this net cash flow over the remaining life of the patent (2017), the result is probabilistic (although likely causal, not associational). Estimates of future sales, sales prices, production and distribution costs, are probabilistic - many omitted variables may result in different prices and costs in the future years. But wait, Ranbaxy Laboratories Ltd. (India's largest pharmaceutical company) currently is challenging the patent in court! If Ranbaxy were successful in its challenge, a generic Lipitor could be on the market as soon as 2006, effectively cutting its patent life in half. In social science, most variable relationships are probabilistic. Typically, this is because most concepts of interest are very complex and exceedingly difficult to measure. In many instances, a single measure is not possible because the phenomena and associated attributes have so many relevant facets or dimensions, each with varying relationships. In social science, we face not only identifYing the necessary and sufficient variables, but also assessing the strength of association the probability - of any relationship.
Estimation and Properties of Estimators

The concept of estimation has many uses in everyday life. Some examples include: • An opinion or judgment of the qualities of a phenomenon of interest, possibly formed without using precise data (e.g., in my estimation, the judge is a fair-minded person); • An approximate calculation of the qualities of a phenomenon of interest, generally using data (e.g., in my estimation that house, given recent selling prices, would sell for $200,000 in today's housing market); • A statistical procedure that assigns to an unknown quantity a plausible value based on relevant and reliable data collected in a random sample (e.g., according to a recent Harris Poll measuring U.S. adults' trust in different professions to give advice that is best for them, the professionals trusted completely by the greatest number of adults are doctors); For our purposes, we define estimation as the approximation of an attribute of a phenomenon that is useful for one's purposes, even if the input to the estimation process may be incomplete, uncertain, or noisy. Assume we own a 2001 Chevrolet Lumina sedan. Such an object (Le., phenomenon) has many attributes including color, mileage, original purchase price, and current market value. If the attribute of interest is current market value, then an estimation process may entail consulting Edmunds. com where we learn that the national base retail price for such a car is

34

An Accountant's Measurement Primer 9 $5,699 as of July 2006. This value is an estimate for a clean vehicle of this age and type with standard equipment and normal mileage. However, this estimate does not reflect the specific mileage, condition and optional equipment of our sedan. We can improve the estimate of our sedan's market value by specifying the missing attributes that affect value. In doing this, Edmunds.com might estimate the following:
Dealer Retail National base price Optional equipment: Keyless entry system 4-wheel ABS Color adjustment (light blue) Mileage adjustment (75,000 miles) Condition adjustment (clean) Total 74 224 -12 -68 $5,699

o
$5,917

Again, the $5,917 price may be a more accurate estimate, but it is still an estimate. Before we sell the sedan, the actual market price is unobservable and is only knowable if we were to sell the sedan. Estimation theory is a branch of statistics that deals with estimating the parameter values (e.g., mean, variance, skew, etc.) for a particular attribute of a population based on measured data. For example, assume a marketing manager desires to know the reaction of customers to a 5% increase in a product price. Specifically, assume that the reaction of interest is the proportion of customers who would decrease their consumption of the product by, say, 10% or more. In this case, the proportion is the parameter of interest. Prior to actually raising the price, that proportion is unobservable. The marketing manager might base the estimate on the responses of a small (possibly random) sample of customers to a survey asking them for their most likely reaction. Assume the survey indicates that 40% of the sampled customers would reduce their consumption by 10% or more if the company raised the product price by 5%. The 40% number, known as an estimator, is an estimate of the parameter of interest. Estimators have properties that govern their value, or usefulness, to users such as our marketing manager. These properties include bias, efficiency, and consistency. Bias In every day usage, bias generally means prejudice in terms of point of view about something. While this sounds pejorative, estimation theory uses bias in a different way. A biased estimator may be difficult to analyze or may lead to an inaccurate or wrong inference by the decision maker. A biased estimator is one that on average misestimates the attribute being estimated. Bias is the extent to which the estimator fails to produce correct, or accurate, estimates (i.e., the magnitude of the error in the estimate). Consider a pistol competition in which the target is set at 50 yards and competitors shoot 5-shot strings. Exhibit 2 presents the target results for two shooters shooting one string. Given the objective of hitting the target's bul1's-eye (i.e., the center of the target), Shooter 1's shots have greater bias than do Shooter 2's shots. Shooter 1 is less accurate than is Shooter 2; indeed,

35

An Accountant's Measurement Primer

10

Shooter 1 seems biased low and to the right, whereas Shooter 2 seems unbiased. Ifboth shooters were rookie police officers and you were the Chief of Police, which would you rather hire? Exhibit 2 Target Bias Shooter I Shooter 2

To illustrate bias in an accounting context, consider the package delivery firm FedEx. In its 2006 annual report to stockholders, FedEx discloses an intangible asset labeled "customer relationships." FedEx further discloses that this asset has a net book value of $48 million ($77 million gross carrying value less $29 million accumulated amortization). Is this asset valuation biased? Accounting only recognizes intangibles as assets when externally purchased, not when internally developed. Thus, this measure may be biased to the extent that it represents only those customer relationships that arise from FedEx's acquisition strategy and not those that arise from its internal operations strategy. Efficiency Efficiency is the extent of the variation of the estimator's estimates. unbiased and attains the smallest error variance is the most efficient. The estimator that is

Consider, again, our pistol competition. Exhibit 3 presents the target results for two additional shooters shooting one string. Both Shooter 3 and Shooter 4 are unbiased in their pistol shooting. On average, they hit the center of the target. Shoot 3, however, is less efficient that is Shooter 4. Efficiency in this situation has to do with the distances between the shot holes. Measuring this attribute produces the parameter called variance. The variance of Shooter 4's shots is smaller than the variance of Shooter 3's shots. Further, we can compare all four shooters. Shooter 1 is an efficient, but biased marksman. Shooter 3 is an unbiased, but inefficient marksman. Shooters 2 and 4 are both unbiased and efficient. Again, if you were the Chief of Police, you would prefer Shooters 2 and 4. However, if your choice were constrained to Shooters 1 and 3, which shooter would you prefer?

36

An Accountant's Measurement Primer Exhibit 3 Target Efficiency Shooter 3 Shooter 4

11

To illustrate efficiency in an accounting context, again consider the package delivery firm FedEx and the customer relationships intangible asset disclosed in its 2006 annual report. Is this asset valuation efficient? The accounting valuation likely is inefficient in that it recognizes only a portion of its customer relationships. Further, FedEx values that portion at historical cost. Depending on a stockholder's purpose for this information, historical cost measurement may further reduce the efficiency of this measure. For example, if the stockholder desires to assess the future cash flows likely to result from the asset, fair value measurement might be more efficient than historical cost measure. On the other hand, if the stockholder desires to assess FedEx's return on invested capital, the historical cost attribute may be more efficient than a fair value measure. Consistency Consistency is the extent to which the estimator's estimates converge to the true (population) parameter as the number of observations increase. Estimators with greater estimation convergence are more consistent than are those with lesser convergence. Consider our pistol competition. Exhibit 3 presents the target results of two shooters, each shooting two strings (recall, each string is five shots). The black target holes are the first string and the yellow target holes are the second string. Both Shooter 5 and Shooter 6 have similar bias and efficiency, but Shooter 6 appears to have greater convergence (i.e., Shooter 6's second string appears to have less spread than does hislher first string) than does Shooter 5 (i.e., Shooter 5's second string appears to have the same spread for both his/her first and second string). Continuing the accounting context illustration of customer relationships reported in FedEx's annual financial statements, is this asset valuation consistent? The accounting valuation is inconsistent to the extent that it does not converge, through repeated measurement, to the population value of interest - all of FedEx's customer relationships. However, its consistency increases to the extent that we redefine the phenomenon of interest as those customer relationships that FedEx acquired externally, ignoring those developed internally.

37

An Accountant's Measurement Primer Exhibit 3 - Target Consistency Shooter 5 Shooter 6

12

In summary, estimation in accounting involves measurement of an attribute of an event (e.g., a business transaction) or object (e.g., an economic resource) that is useful for the information user's purposes, even if the input to the estimation process is incomplete, uncertain, or noisy. The primary users of accounting information are investors/creditors and managers. Accounting labels usefulness as relevance and the properties of estimators - bias, efficiency, consistency ­ largely are elements of accounting's reliability concept. Subjectivity Most human problems - such as "what is the fair value of my used car?" are ambiguous. This ambiguity arises, at least in part, from the fact that our problems represent both an objective state of affairs and a subjective state of mind. Applying this to estimation, an estimator has both objective and subjective components. The objective component arises from the phenomenon! attribute of interest, whereas the subjective component arises from the person making the estimate. The risk is that the measurer's subjectivity may introduce a bias into the estimate, a bias that the user generally is unable to separate from the objective component. Such bias is even more dysfunctional when the user is unaware of, or ignores, the fact that bias is present in the estimate. The potential for subjectivity exists when any aspect of the measurement, or estimation, process involves human judgment. Take, for example, a game of chance that involves rolling a pair of dice. What is the probability of a "1" occurring on the next roll of the dice? Logic instructs us that there are 36 possible outcomes for a pair of six-sided dice. Since a seven can occur in six of these outcomes, the probability (assuming the dice are fair) of rolling a seven is 6/36 = 116 = 16.67%. This estimate largely is objective in that it arises from properties of the object (the pair of dice) and its attributes, and not the person making the estimate. On the other hand, take the issue of my car starting on the first frozen morning this fall (my car is outdoors overnight). 1 might assert that the probability of my car starting is 90%. Intertwined in this estimate are both objective and subjective components. It is objective in that I base my estimate, in part, on the knowledge that 1 have just installed a new, heavy-duty car battery. It is subjective in that it

38

An Accountant's Measurement Primer

l3

represents my general optimism, for I have failed to incorporate into my estimate issues such as the car's six-year age and my tardiness in having its engine tuned. Subjectivity is a significant issue in accounting estimation; most accounting estimations involve subjective judgments by the accountant. For example, consider the fair value of a drug patent owned by a pharmaceutical company. Assuming the company intends to hold the patent and to manufacture and sell the drug, the fair value is in its use, not in its exchange. 5 Thus, to measure the fair value of an asset in-use, the accountant might estimate the future net cash flows from this drug and then discount such cash flows to the present using a risk-adjusted interest rate. Contrast this situation, however, with a short-term investment in, say, the common stock of another company. The accountant might estimate the fair value of such an investment as the closing price per share of that common stock on the NYSE (assuming the stock trades on this exchange). While the fair value estimate of the investment still has a subjective component, its objective component (observable market prices) is much larger than is the objective component of the drug patent's fair value estimate (present value of estimated cash flows).

5

The patent's value in-exchange would be how much I could obtain from selling the patent in a market transaction.

39

An Accountant's Measurement Primer References

14

Financial Accounting Standards Board. Statement of Financial Accounting Concept No.2: Qualitative Characteristics of Accounting lriformation. Norwalk, CT: Financial Accounting Standards Board, 1980. Downloaded from in January 2007. Financial Accounting Standards Board. Statement of Financial Accounting Concepts No.5: Recognition and Measurement in Financial Statements of Business Enterprises. Norwalk, CT: Financial Accounting Standards Board, 1984. Downloaded from in January 2007. Miller, Paul B. W. and David Mosso. Financial Accounting Measurement: Why Things May Not Be What They Seem To Be. Financial Executive, December 1983. Suggested Further Reading Kerlinger, Fred N. and Howard B. Lee. Foundations of Behavioral Research, Orlando, FL: Harcourt College Publishers, 2000.
4th

edition.

Reynolds, Paul Davidson. A Primer in Theory Construction. New York, NY: Macmillian Publishing Company, 1971. Savage, Leonard J. The Foundations of Statistics, 2nd revised edition. New York: Dover Publications, Inc., 1972.

40