Have accounting standards caused a decline in “matching” over time?
Many accounting academics think that matching expenses to their associated revenues produces an earnings pattern that is useful to decision makers. As a result, these academics think that any decline in matching will produce less useful earnings information for financial statement users. Along these lines, a recent study by Dichev and Tang (2008) in The Accounting Review examines the correlation between contemporaneous revenues and expenses and finds that this correlation has declined in the last 40 years. The authors suggest that this result indicates that matching has declined over the last 40 years and they provide indirect evidence that changes in account standards are primarily responsible for this decline.
However, a more recent study by Donelson, Jennings, and McInnis (2011) in The Accounting Review sheds important light on these earlier findings. Donelson et al. examine individual expense lines to discover which line items are most responsible for the decline in the revenue-expense correlation over time. They disaggregate total expenses into cost of goods sold, SG&A expense, depreciation, taxes, other income/expenses, and special items. They find that the decline in the revenue-expense correlation over the past 40 years is primarily attributable to one line item—special items, which consists of asset impairments, restructuring charges, and gains/losses from asset sales. When the authors factor out the effect of special items, they find that the changes in the revenue-expense correlation are substantially reduced (as much as 90% in some cases). The authors go on to provide additional evidence suggesting that the cause of most special items is from changes in the incidence of underling economic events, and not from changes in accounting standards.
So why should this matter to financial reporting standard setters? For those who don’t agree with the idea that matching is an important concept in accounting, the results of these two studies are probably not that useful. But for those standard setters who may still hold on to traditional ideas on matching, and who may deplore any decline in matching over the years, these two studies suggest that new accounting standards have contributed very little to any decline in matching. Instead, it is the changes in the underlying economics of businesses that have caused any decline in matching. Standard setters probably wouldn’t want to undo financial reporting that actually captures changes in underlying economics, would they?
I have always been confused by this issue of matching. So is it in the conceptual framework or not in it?
Lisa, based on my understanding of the current conceptual framework, expense recognition should not be deferred or accelerated solely to match the timing of expense recognition with that of related revenue recognition.
That said, expense recognition and revenue recognition often occur quite naturally in the same reporting periods because of the underlying economics of the transaction (as with a simple point of sale transaction), but this sort of “matching” is simply an indicator of good accounting.
On the flipside, matching for its own sake (i.e., simply to smooth reported earnings) is devoid of economic meaning and so obscures more than it reveals.
In order for that type of matching to be meaningful as a principle for standard setting, standard setters would need a definition of economic income from which they could derive definitions of assets and liabilities. But that type of income statement perspective was rejected back in the 1980s.
Yes, matching is explicitly mentioned in the current conceptual framework in a number of places (e.g., CON 6, para. 144-152; CON 5, para. 86).
While matching is mentioned in the framework to describe how to allocate costs over time, it is not mentioned as a principle that should dictate standard setting. Rather, if we can correctly identify and measure assets and liabilities, then net income (the classicial Hicksian view) will naturally fall out – and so will the proper matching of expenses with revenues.
At least, that’s my understanding – not speaking for the FASB! This issue is related to a question that will be asked to Bob Herz at this year’s AAA meetings.
My study extends DT and DJM by examining a wider sample of the firm population that includes new firms unexamined by those studies. I posit that population-wide decline in revenue-expense association, which I call “concurrence,” reflects dramatic increases in new listings and fundamentally different characteristics of new firms (Fama and French 2001, 2004). I find that new firms undertake riskier investments, specifically R&D outlays; therefore, a higher portion of their outlays is expensed immediately rather than capitalized. I also find that new firms bundle multiple products and services and deliver them over multiple reporting periods; therefore, a higher portion of their revenues is reported in future periods. These changes in firms’ investment and marketing policies reduce the likelihood that non-inventoriable outlays and corresponding revenues are reported in the same periods. Accordingly, I conclude that the overall decline in concurrence reflects changes in nature of underlying transactions and composition of the firm population.
http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1815562
Great comments Jeffrey and Lynn. I agree with your assessments of where matching fits into the framework (it’s just mentioned to describe a practice, but it is not a principle meant to guide standard setting) and that proper matching will naturally occur if (and that’s a big if) we can correctly identify and measure changes in assets and liabilities during the period.
Anup, thanks for bringing your paper to our attention! Definitely relevant to our discussion. I wonder how your results will change when (if?) the proposed standard on revenue recognition (which allows more performance obligations to be identified) goes into effect. Would you expect to see some of the deferred revenue getting to be recognized sooner as a result, and thus concurrence increasing after the proposed standard goes into effect?
Hi Jeff,
In summary, my study shows that the primary reason for decline in matching is changes in nature of firms’ underlying operating costs rather than changes in nature of firms’ revenue transactions. Note that operating costs constitute more than 95% of firms’ total costs.
Specifically, I find that inventoriable portion of total outlays have dramatically declined. The weight of COGS that used to exceed 80% for almost all firms in the early 1970s, has declined to less than 50% in recent times. The decline for new firms and high-tech firms is even more dramatic.
The most likely reason for decline in firms’ inventoriable outlays is the dramatic increase in firms’ R&D outlays. This finding holds for an average firm and not just for the high-tech firms. Figures 1 and 2 in my paper show startling trends in firms R&D outlays and weight of COGS. My study provides economic reasoning for this increasing R&D outlays, which trend is likely to strengthen further due to increasing globalization and “perpetual race” and winner-takes-it all nature of R&D payoffs.
Some may argue that concurrence has declined due to (potentially flawed) accounting practice of expensing rather than capitalization of rapidly increasing R&D outlays. I find that uncertainty of payoffs from R&D has increased across time, which indicates that capitalizing R&D outlays might (potentially) improve matching but would adversely affect earnings reliability (Kothari, Laguerre, and Leone 2002; Shi 2003).
Moreover, my study shows that it is simplistic to expect that expedited revenue recognition would solve (lack of) matching problem. Firms typically spend on R&D outlays before they initiate manufacturing. Thus, expedited revenue recognition might improve concurrence with manufacturing outlays, which are anyway inventoriable, but is unlikely to improve concurrence with R&D outlays. I introduce a “time-to-revenue-recognition” concept in my paper, which is naturally, longer for R&D outlays.
Thanks
Anup
Thanks Jeff for bringing some attention to our paper and summarizing it quite nicely.
For both academics and standards setters, the main takeaway from our paper is that the “matching” of revenues with operating expenses has not changed much over the last 40 years among economically significant firms. The biggest change in expense recognition has been an increasing frequency of asset impairments, restructuring charges, and disposal gains and losses. Much of this appears to be attributable to firms engaging in more asset sales and restructuring activities or facing an economic climate where impairments are more likely compared to 40 years ago. However, it could also be the case that firms now have more reporting incentives to recognize impairments or restructurings more rapidly in a single accounting period and deem them “one time” or “special.”
Putting our research paper aside, I’ve never taken a hardline view that either fully embraces or rejects the matching principle as a method to guide expense recognition. It’s just one way of thinking about accounting. While I personally think that viewing expenses as decreases (increases) in assets (liabilities) is a more tractable approach to standard setting, there are instances where embracing a balance sheet view of the world has limitations, and it’s nice to have the matching principle in your back pocket.
Here’s an example from the course I teach at UT. Suppose a retailer sends a mailer out at the end of period t offering 10% off all merchandise for a one day sale in period t+1. At the end of period t, should the retailer record a liability and offsetting expense for the expected costs of this promotion? Some might argue that this mailer is simply a marketing offer, and since customers have not paid any consideration or bought merchandise in exchange for this 10% off, the retailer has no obligation to the customer at the end of period t. However, looking to CON 6, I think a reasonable argument can be made that the retailer has a constructive obligation: a valid expectation has been created in the minds of customers that discounts will be offered and the retailed has very little “wiggle room” to back out of the promotion.
Here’s where I find the matching principle handy. Since the discount promotion is designed to stimulate sales in period t+1, shouldn’t the cost of the promotion be recorded in period t+1? From this perspective, it makes sense to record the expense in period t+1.
Thoughts?
John,
Personally, I’m not sure it makes sense to “gross up” the transaction (i.e., to view the original price as the “true” revenue amount and the 10% discount as an expense of doing business).
Rather, I think what is happening is that the retailer is simply agreeing to do business at a lower price (in hopes of doing more business in total).
When that additional business transacts in period t+1, the net amount of revenue is recorded (i.e., revenue that is 90% of the undiscounted price). This “matches” the “cost” of the coupon with the benefit, but not because of a matching principle.
Question on standard setters’ current on matching. Lynn Rees’ mail suggests that matching was considered an important tenet of accrual accounting (SFAC Nos. 5 and 6). I found a few documents in which FASB has denounced the revenue-expense system. I also noticed that the word matching is missing from the latest conceptual statement. But, have the board members denounced matching per se? I am looking for any minutes or any documentation on board members’ negative views on matching.
Thanks
Anup