Anup Srivastava (Northwestern) presented a research paper at today’s FASRI Office Hours. In this post, I briefly recap the takeaways from the paper and some of the questions that were asked by participants in today’s office hours.

Anup’s paper examines how the constraints on revenue recognition imposed by SOP 97-2 Software Revenue Recognition affect the informativeness of earnings and the prevalence of earnings management. Earlier studies examine constraints introduced by SAB 101 Revenue Recognition, which focuses more on determining when a good or service is delivered. Anup’s paper extends the prior literature by examining the constraints placed on revenue recognition when a company lacks observable selling prices for all components of a multiple element arrangement.

Consistent with prior research examining delivery constraints on revenue recognition, this paper documents a decline in the earnings-returns association after SOP 97-2 imposed valuation constraints on revenue recognition. This result is observed even after controlling for the increase in deferred revenues created by SOP 97-2. Additional tests suggest that the market prices the SOP 97-2 created deferred revenues as if they were still revenues. Also, a pro-forma earnings that adds back the SOP 97-2 created deferred revenues more strongly corresponds with market returns than GAAP earnings. Finally, the paper does not observe any change in the measures that are generally interpreted as indicating earnings management, which is consistent with SOP 97-2 having no effect on earnings management.

The paper concludes that the IASB/FASB’s recently proposed revenue recognition model would be an improvement for the firms in this study (and other similar firms) because the proposed model would allow estimates of selling price when determining how to allocate consideration across multiple elements. This would in turn make earnings more informative in that earnings would depict the transfer of goods and services to customers at the times those goods and services are transferred to customers.

Participants in today’s discussion raised a number of questions, including the following:

  • Is there a way to identify a priori which firms have greater incentive to be opportunistic? Similarly, can we identify which firms have greater constraints already (perhaps through better corporate governance) on their revenue recognition practices. I think the aim of these questions was to see whether we might improve the power of the tests that examine whether earnings management is affected by SOP 97-2 implementation.
  • Rather than describing SOP 97-2 as removing flexibility to be opportunistic, could you instead describe it as removing the ability to credibly signal a firm’s strength? For example, a strong firm might have already been delaying all revenue recognition until the final element was delivered, even though it had valuation information that would have permitted revenue recognition on delivered items. By deferring all revenue until delivery of the last item, these firms may have been signaling that they were in a relative position of strength. By forcing all firms to postpone revenue recognition in the absence of VSOE (vendor specific objective evidence) of selling price, such a firm would no longer be able to signal its strength as easily.
  • In adopting SOP 97-2, companies may have decided to cookie-jar as much revenue as possible because the market was already expecting a one-time hit due to implementation of SOP 97-2. How might this affect the conclusions reached in this paper? Would it bias toward finding less informative earnings after adoption of SOP 97-2?
  • To what extent is the observed decrease in the earnings-returns association explained by the fact that investors are accustomed to the old revenue recognition approach and had not yet adjusted to the new valuation-related constraints on revenue recognition? If you were to rerun the analysis today for firms that follow SOP 97-2, would you see the same result? This may speak to the implications for standard setters.
  • To what extent are start up firms priced based on cash flow? If many of the firms in your sample were start up firms and investors valued them based on cash flow rather than earnings, then implementation of SOP 97-2 may not be that important to investors. They would continue to value the firm based on cash flow, which would make the earnings-returns association decrease and would also explain why deferred revenues appear to be priced as if they were still being recognized as revenue. Do the implications for standard setters change if this is the story about what is going on?
  • To what extent does an earnings-return association tell us anything about the usefulness of recognized revenue? Is there a way to focus just on the association between revenues and returns? I suppose that adding back the deferred revenues to your pro-forma earnings analysis deals with this to some extent.
  • This paper analyzes software companies. If the proposed revenue recognition model were applied to construction companies, some of them would no longer be able to recognize revenue during the construction process (because some of these entities do not transfer an asset to the customer until the end of the construction process). Would you expect the informativeness of earnings (or revenue) to decreases for these companies? Would you interpret that as evidence that the proposed revenue recognition model is faulty, or that investors have a different idea of when they think value is created for construction companies—regardless of when an asset transfer to the customer?

Anup provided insightful response to many of these questions, but I did not capture most of them in my notes. So, I invite Anup to respond again to any of these questions. In fact, I invite any of today’s participants or any reader to respond to these questions or suggest any additional questions on the topic. I look forward to your thoughts.