Measurement in accounting is the process of determining the monetary amounts at which assets and liabilities should be recognised and carried in the balance sheet. Traditionally such measurement has been carried out on the basis of historical costthe consideration given or received at the time the asset or liability first entered into the firms balance sheet. 
 
These historical values, adjusted for depreciation, impairment or partial fulfilment of obligations, forms the basis on which the financial health of the firm and its future prospects are judged. 
 
At first sight, this convention of measurement at historical costs appears contrary to a basic principle of decision-making--that of sunk cost. 
 
This principle says that only facts which can be influenced by current decisions are relevant. Payments received or costs incurred in the past cannot be of any use to the decision-maker. 
 
The principle of sunk cost says that all decision-relevant information must be forward-looking. If we go by this, historical costs cannot have any place in financial statements that aim to help outsiders evaluate the financial situation of the firm. But then why have accountants used historical costs? 
 
There are two fundamental reasons. The first is reliability. While the management must make forward-looking projections for its own decisions, such projections are by their very nature subjective. 
 
They reflect expectations and risk preferences of the management and as a result are not open to verification. Allowing such unverified projections in financial statements would be an open invitation to managements to manipulate outsiders by not revealing truthfully the current state of the firm. 
 
The second reason for use of historical costs is that traditionally, accounting has also been required to fulfil the stewardship function of monitoring owners' control over managers and this control function requires historical values in order to compare actual performance with predictions. 
 
For these reasons, accountants had chosen to base their measurement system on verifiable historical costs while at the same time developing rules of thumb to adjust historical costs to give them greater decision relevance. These rules of thumb worked well in the relative static environment of the formative years of modern accounting. 
 
However, there is a growing realisation within the accounting profession that such rules of thumb no longer work as well in today's dynamic business environment. The complexity of transactions that modern corporations enter into, fast changing prices, the rapidity of the technological cycle, all conspire to defeat the old rules. 
 
Thus, decision-usefulness demands that accounting move towards a more explicitly forward-looking basis. The challenge is to do so without sacrificing reliability. 
 
Also, such a move might open up a contradiction between decision-usefulness and the stewardship function. Both these issues are being intensely debated by accountants. 
 
The greatest chance of finding values useful for decision-making is where the asset or liability in question or similar assets or liabilities are priced by an active market. In fact, accountants define fair valueas the best estimate of market value-- the amount for which an asset or liability could be exchanged between willing, knowledgeable parties in an arms length transaction. 
 
A stock exchange is an example of an active market, a place where prices cannot be influenced by individual entities. At the same time, the prices reflect the expectations of the market participants about future cash flows and hence satisfy our requirement of being forward-looking. 
 
Indeed, in the strongly efficient markets of financial economists, prices reflect not just all public but also all private information that exists about the asset. Other than observed prices, replacement costs, reproduction costs and realisable value are also used to measure market value. 
 
More difficult is the case of non-contractual assets including tangible assets like property, plant and equipment, and intangible assets like product brands. The cash flows that these assets are expected to generate in the future are not contractually defined. 
 
They must be estimated on the basis of forecasts about the firms business environment, the characteristics of the assets, their condition on the measurement date, the nature of the ownership rights the firm has over them etc. Active markets with arms length transactions do no exist for such assets. 
 
Moreover, these assets do not generate cash flows independently but do so only by working with other assets and inputs. Therefore the sum of the values of individual assets taken separately may be less than the value of all the assets taken as a unit. 
 
The different degrees of reliability with which fair value can be measured has been recognised by the Statement of Financial Accounting Standards (SFAS)-157 on Fair Value Accounting, issued by the FASB in September 2006. 
 
SFAS-157 establishes a fair value hierarchy. It gives the highest priority (Level 1) to quoted prices (unadjusted) of identical or similar assets in active markets and the lowest priority (Level 3) to unobservable inputs. 
 
At Level 2 are quoted prices of a date other than the measurement date, adjusted for intervening changes in the economic environment. 
 
At Level 3, entities are required to estimate fair value using multiple valuation techniques consistent with the market approach, income approach and cost approach whenever the information necessary to apply those techniques is available without undue cost and effort. 
 
In most situations, it is difficult to collect market information without undue cost and efforts. Therefore, entities determine market value based on their expectations and their own risk perception. 
 
Fair value models have been developed for some significant contractual assets and liabilities, but no reliable and well accepted model is available for valuation of non-contractual assets. 
 
Therefore, value for a non-contractual asset determined using an economic model fails to pass the test of verifiability, and hence that of reliability. 
 
Even for contractual assets and liabilities, value determined using an economic model cannot be called fair value because, the value so determined does not have the quality of fair value determined in an active market. 
 
Moreover, the inputs and methods for estimating their fair value are more subjective and, therefore, the valuations less reliable. For example, fair value estimates for bank loans can vary greatly depending on the valuation inputs and methodology used. 
 
Therefore, value for a contractual assets or liability so determined fails the criterion of faithful representation and neutrality, and hence the test of reliability. 
 
As a result we may have to continue using historical cost or hybrid models for certain classes of assets and liabilities. But it is time to acknowledge such measurements for the compromises they are and set fair value measurement as the ideal towards which we must move. 
 
The transition will not be easy. Yet this transition must be carried out. Till now, accounting has overemphasised reliability at the cost of relevance. To continue on that path can only make accounting irrelevant for business and investment decisions. 

Asish K Bhattacharyya
The writer is professor of finance control at IIM-C