Why R and D Is Not A Discretionary Expenditure In the Digital Era

In the digital economy, where every company is now, at it's core, a tech company, the calculations of costs and benefits is changing.

Investment - and the assessment of returns to those resource allocations - must be focused on intangibles like research and development as well as the people who inspire, conduct and manage those initiatives because without them, such enterprises would quickly be surpassed by rivals.

A challenge is that hidebound accounting strictures continue to treat R&D as discretionary, meaning that they often negatively impact short term financial results and the compensation systems based on them. Successful companies increasingly recognize that to prevail they have to innovate continuously and that to optimize the results of those efforts, they must invest. JL


Vijay Govindarajan and colleagues report in Harvard Business Review:

Research suggests a significant component of digital companies’ R&D costs are necessary operating expenses whose curtailment might stop the companies’ operations. In 2017, Facebook and Alphabet spent 19% and 15% of sales on R&D, compared to 2% for General Motors, zero for Walmart. Digital companies must invest to keep pace with the technological progress, competitive offerings, and ever-increasing customer expectations, or become obsolete. Economists must revisit the notion of declining marginal utility of costs, as R&D can have increasing returns to scale.
Paul Romer won the 2018 Nobel Economics Prize for integrating technological innovation into macroeconomic analysis and showing that research and development (R&D) investments are becoming essential to a nation’s progress. However, accounting textbooks, academic literature, national budget plans, and management books continue to consider R&D as a discretionary expense, defined as cost that is not tied to operations and can be curtailed or even eliminated in the short run without impacting current revenues. Even the IRS tax code’s tax credits for R&D expenditures are based on the idea that R&D is a non-essential expenditure and managers would avoid those investments without fiscal incentives.
We question the idea that R&D is a discretionary expense, at least for digital firms. Our research suggests that a significant component of digital companies’ R&D costs are necessary operating expenses whose curtailment might stop the companies’ operations. We therefore recommend that the notion of R&D as a discretionary expense must be revisited as ideas, strategy, software, algorithms, and innovation become the foundations of economic activity.
Revenues for digital companies depend on the network of members and their engagement with the company’s platform. We surveyed financial disclosures of Facebook, Alphabet, Twitter, LinkedIn, Spotify, Netflix, and Yelp to investigate how R&D affects their network and user engagement for the fiscal years of 2013 to 2017. We distilled five insights from our investigation.
First, R&D is an economically significant expense for digital companies, much larger than for physical product companies. For example, in 2017, established digital companies such as Facebook and Alphabet spent 19% and 15% of their sales, respectively, on R&D. Smaller digital companies might even spend more than 50%. For example, in 2013, Twitter spent 76% of its revenues on R&D. Compare these numbers to 2% for General Motors, zero for Walmart, and 12% for Tesla in 2017.
Second, a large component of R&D costs for digital companies consists of employee costs for engineering, product management, and information technology personnel. Given the global quest for scientific talent and manpower, it is unlikely that digital companies would fire those employees in the short term with the hope of rehiring them in future. In a previous HBR article, we found that digital companies plan their activities around scientific manpower, which they consider their biggest constraint for growth, instead of around financial capital, which they can find in abundance. Stated differently, scientific talent is as essential to a digital company’s operations as machine operators are for a manufacturing company.
Third, digital companies consider product development as a necessary activity to survive. For example, Facebook continually innovates to address the changing needs of its customers and marketers. Its R&D costs include those for expanding computing needs and protecting the security and integrity of its platforms and systems. LinkedIn relentlessly improves its proprietary systems for feature extraction, information retrieval, and matching to provide the most relevant ads to its marketers while improving the power of its datasets for personalized member search.  In addition, it invests to improve its mobile strategy. Yelp invests in advertising and e-commerce products, explores new platforms and distribution channels, and develops partner arrangements, to encourage members to increase their usage and encourage marketers to apportion greater advertising budgets to its platform. For Netflix, technology and development expenses includes those incurred on testing, enhancing, and modifying user interface and recommendations, streaming delivery technology and infrastructure, as well as computer hardware and software costs. Salesforce.com improves security and enhances its enterprise cloud computing services. In all cases, digital companies must invest in product development to keep pace with the technological progress, competitive offerings, and ever-increasing customer expectations, or fear becoming obsolete in no time.
Our fourth finding is that digital companies routinely rely on third-party software, algorithms, development tools, cloud services, security and data integrity systems, client monitoring, customer databases, and cross-selling platforms for their day-to-day operations. This reliance on external partners is no different than leasing assets or purchasing raw materials for a physical product company. Therefore, digital firms’ R&D expenses include the costs of obtaining outside services, consulting, and code modification and integration, all of which are necessary for seamless operations. A related fifth finding is that digital companies continuously scout for acquisition targets to obtain readymade R&D and talent teams. The costs of those acquisitions are also amortized as R&D expenses.
For the five reasons described above, digital companies must spend considerable amounts on R&D to run their day-to-day operations, while relentlessly improving the depth and width of their capabilities in order to compete and survive. Therefore, unlike physical product companies for which R&D expenses can be curtailed or even eliminated in the short term without any immediate impact on revenues, the curtailment of R&D investments for a digital company could halt its operations or cause quick exodus of its members and marketers. Hence, R&D for digital companies is an essential operating activity, not any different than producing or purchasing inventory for a physical product company. The only difference is that these operating activities for digital companies rely on the principle of virtual scalability, where the variable costs of servicing an additional customer approaches zero, while the variable cost for physical product companies runs from 60 to 90 cents per revenue dollar.
We therefore conclude that R&D costs should not be considered as discretionary, at least for digital companies, as is often the case in accounting, finance, economics, and taxation literature, textbooks, practice, and regulation. Our conclusion could have important policy implications. For example, R&D often requires graduates trained in science, technology, and mathematics (STEM). More than half of Silicon Valley STEM graduates are immigrants. This explains digital companies’ lobbying for visa reform or planning to open offshore centers for product development.
Furthermore, our analysis suggests that economists must revisit the notion of declining marginal utility of costs, as R&D, the main costs for digital companies, can have increasing returns to scale. Financial analysts and management cost accountants must revise their notions of variable costs, gross margins, and fixed cost allocations. CEO compensation plans and calculation of non-GAAP earnings, that often exclude stock-option expense, must reconsider those calculation methods when scientific manpower is paid mainly by stock options and is essential for the survival of the company.
To sum it up, the notion that R&D is not a necessary expense must be revisited as digital companies play an increasingly important role in the economy and as digital innovation and technological advancement become the key drivers of a nation’s progress.