The Trillion-Dollar R&D Fix(HBR）

Artwork ricky Allman, Fluid Redux 2010, acrylic on panel, 36" x 48"Spotlight Spotlight on InnovatIon For the 21st Century The Trillion- Dollar R&D Fix Most big companies should spend more on R&D. But how much more? by Anne Marie KnottHow does a company know what kind of re-turn it’s getting from R&D? Is it better at R&D than the competition? How much should it be spending, and what can it do to improve the effectiveness of its investments? Existing measures of R&D effectiveness—for in- stance, amount of spending or number of patents— don’t answer those questions or reliably predict mar- ket value. Year after year Booz & Company publishes “The Global Innovation 1000.” And year after year the consulting giant points out that R&D spending does not correlate with market value or growth. The 2010 report argues, “Spending more on R&D won’t drive results. The most crucial factors are strategic Ph ot og ra Ph y: C ou rt es y of t h e By ro n C . C oh en g al le ry hBr.org May 2012 harvard Business review 77 The Theory Behind RQ Definition The RQ method for measuring productivity of R&D investment uses the well-known economic formula for measuring labor and capital pro- ductivity. This equation defines the relationship between a firm’s inputs (what it spends) and output (its revenues). The formula typically looks like this: The RQ method expands the calculation to include another input, R&D (R): CalCulation To calculate your RQ, you need several years’ data on revenues and annual expenditures on PP&E (property, plant, and equipment), labor, and R&D. Those data are converted into logs, a standard transforma- tion needed to run the regression analysis that produces RQ. The spreadsheet looks something like this: When you run the regression analysis, you get productivity levels for PP&E, employees, and R&D. An analysis of the full set of U.S. publicly traded compa- nies yielded an average R&D exponent of 0.109—which means that increasing R&D spending by 1% would increase revenue by 0.11%. Comparison Finally, to compare research productivity across firms and to help firms track changes in their R&D productivity, you rescale the exponent number relative to the mean of all U.S. traded firms to create an index num- ber. (An RQ of 100, therefore, denotes the average R&D exponent across all firms.) Most companies (67%) have RQs between 85 and 115. Revenue PP&e emPloyee R&D ln(Rev) ln(PPe) ln(emP) ln(R&D) 2007 $4,847 $976 $8 $108 8.49 6.88 2.05 4.68 2008 $5,273 $960 $8 $111 8.57 6.87 2.12 4.71 2009 $5,450 $955 $8 $114 8.60 6.86 2.12 4.74 2010 $5,534 $979 $8 $119 8.62 6.89 2.12 4.78 alignment and a culture that supports innovation.” The trouble is that it’s also hard to measure strategic alignment and culture, let alone link them to profit- ability or market value. R&D is thus an easy target when firms face quar- terly earnings pressure. Since it is expensed rather than capitalized, cuts yield immediate increases in profit, while the detrimental impact of those cuts aren’t felt for a few years. In marginal trade-offs between investments in, say, physical capital or advertising, whose returns are more quantifiable, R&D loses out. In response to the recent prolonged recession, for instance, firms with revenues greater than $100 million reduced their R&D intensity (R&D spend divided by revenue) by 5.6%, on average, whereas capital intensity at those firms fell only 4.8% and advertising intensity actually increased 3.4%. A new metric for R&D productivity—which I call RQ, short for research quotient—can change all that. RQ allows you to estimate the effectiveness of your R&D investment relative to the competition and to see how changes in your R&D expenditure affect the bottom line and, most important, your company’s market value. My research—which includes a com- prehensive analysis of all publicly traded companies in the U.S.—suggests that if the top 20 firms traded on U.S. exchanges had optimized their 2010 R&D spending using the RQ method, the collective in- crease in market cap would have been an astonishing $1 trillion. The longer-term benefits are even greater, as RQ also allows companies to link changes in R&D strategy, practices, and processes more closely to profitability and value. This is a story we’ve seen played out before: Thirty years ago, W. Edwards Deming’s quality met- rics inspired the TQM movement, which revolution- ized the way companies manufacture. I believe that RQ can do the same for R&D—and that the payoff will be even greater. The Measure Calculating RQ doesn’t involve fancy new math. Economists have been calculating capital and labor productivity for years—that is, determining the mar- ginal value of increasing either one. R&D productiv- ity can be determined using the same method, al- though few, if any, analysts or academics have done so at the level of individual companies. Essentially, the equation defines the relation- ship between a firm’s inputs (what it spends) and its output (its revenues). The formula typically con- HBR.ORG To see the RQs of all U.S. publicly traded companies that do R&D (more than 1,500 firms), go to the online lookup tool at www.hbr.org/magazine/rq. Y=K𝛂l𝛃 Y=K l r ThE ExPonEnTS inDicATE hoW PRoDUcTivE EAch inPUT iS in gEnERATing oUTPUT. SPEciFicAlly, ThEy ShoW ThE PERcEnTAgE incREASE in A FiRM’S REvEnUES RESUlTing FRoM A 1% incREASE in cAPiTAl (AlPhA) oR lABoR (BETA). ThE nEW ExPonEnT, gAMMA, TEllS yoU hoW MUch oF A PERcEnTAgE incREASE in oUTPUT yoU WoUlD gET FRoM A 1% incREASE in R&D SPEnDing. $ in millions OuTpuT (Revenues) capiTal laBOR OuTpuT capiTal laBOR R&D 78  harvard Business Review May 2012 spOTliGHT on innovATion FoR ThE 21ST cEnTURy siders two costs, capital and labor. Of course, those aren’t the only determinants of revenue, and most economists would accept that the equation could be expanded to include another central input: R&D. Us- ing standard regression analysis, the calculation tells us in a very precise way how productive each of the inputs is in generating output. It tells us, for instance, how much a 1% increase in R&D spending would in- crease a firm’s revenue. A precise estimation of RQ examines thousands of firms simultaneously using fairly sophisticated software, but a coarse estimate of a single firm’s RQ can be run on an ordinary spreadsheet using histori- cal data easily obtainable at most large companies— revenue figures and spending on PP&E (property, plant, and equipment), employment, and research. (See the sidebar “The Theory Behind RQ” for more on the method.) Once you know your RQ—how effective your company is at R&D—you can determine the amount of R&D spending that would produce the maximum profits. That calculation involves a standard piece of math, called a partial derivative, that can be easily embedded in a spreadsheet. In essence, it’s an exer- cise in marginal returns—determining at what point an additional dollar spent on R&D begins to reduce revenues and profitability. Why It Works Good measures have three properties: universality, uniformity, and reliability. Uniformity means the measure is interpreted the same way in all contexts; universality means it applies to all relevant entities (in this case, firms); and reliability means that its predictions confirm what theory says should hap- pen. The easiest way to explain why these properties are important is to show why another measure often used to gauge R&D effectiveness—patent counts— fails because it lacks them. First, patent counts aren’t universal in that not all firms doing R&D patent their innovations. In fact, fewer than 50% of firms engaged in R&D file patents in any given year. Moreover, even among patenting firms, few of them patent all their innovations. It’s often more effective to protect intellectual property by keeping it a trade secret. Patents aren’t uniform, either. Compare, for example, the economic value of the patent for copying DNA with that of the 97% of patents that are never commercialized. On aver- age, 10% of patents account for up to 85% of the value of all patents. Finally, higher patent counts don’t reliably predict higher profits and market value—the outcome companies expect from R&D investments. In contrast, RQ exhibits all three properties. RQ is estimated entirely from standard financial data, so it can be calculated for any firm doing R&D. And because RQ is a ratio, its interpretation is uniform across firms regardless of currency. Most important, RQ is reliable. It confirms what you would expect it to: (1) that firms with higher RQ—those that are bet- ter at R&D—spend more on R&D than firms with low RQ; (2) that R&D spending beyond the optimal limit identified by RQ reduces firm market value; and (3) that firms with higher RQ have higher profits and market value for a given set of inputs. My colleagues Carl Vieregger and James Yen and I have demonstrated all three effects rigorously across all publicly traded U.S. firms from 1981 through 2006. Our analysis of the data shows that a 10% increase in RQ—that is, in R&D productivity—results in an in- crease in market value of 1.1%. The Payoff Using the RQ measure has immediate benefits. Firms—not to mention the financial analysts who track them—can now identify the marginal returns to R&D and the level of R&D investment that gen- erates the greatest market cap. As the exhibit “The Trillion-Dollar Opportunity” illustrates, the gains from bringing R&D closer in line with optimal levels prescribed by RQ are enormous. For most companies, RQ will call for significant increases to R&D budgets. A few firms, of course, will find the opposite. To reach its optimal level, Pfizer would have to cut its R&D spending by $3 bil- lion a year, money that would be freed up for invest- ment in other, more productive things. Once com- panies adopt RQ as a standard metric and align their spending accordingly, market caps should rise very quickly since increases to market value typically ma- terialize as soon as beliefs about future performance change. Traditional measures of R&D productivity are only loosely linked to profits or market value. That makes it difficult for executives to know whether they’re spend- ing as much (or as little) as they should, let alone make improvements in the way they spend. A new measure, RQ or research quotient, derived from classic regression analysis, al- lows managers to make these judgment calls. It also enables managers to estimate the optimal amount they should spend on R&D. RQ is a very accurate predictor of profits and share price perfor- mance. If companies were to adjust their R&D budgets in line with the levels appropriate to their RQ, the positive impact on corpo- rate value would be enormous. For most companies, RQ will call for significant increases to R&D budgets. A few firms will find the opposite. Idea in Brief hbR.oRg May 2012 harvard business Review 79 The TrIllIon-Dollar r&D FIx Over the long term, RQ will improve the quality and effectiveness of R&D initiatives. Managers will be able to determine, for instance, whether a given change in R&D strategy translates over time into a higher or lower RQ. And as managers and analysts get better at measuring the success of initiatives, they will be able to make better judgments about the quality of firms’ management decisions and start to understand which R&D practices create the most value in various contexts. Since the measure is new and not widely used, I can’t make conclusive statements about which prac- tices and processes improve RQ. However, a National Science Foundation study I conducted with Bruno Cassiman, of IESE Business School, suggests three preliminary insights: RQ rises with the breadth of a firm’s activ- ity. RQ is positively correlated with the number of markets a company sells to outside its home region (export breadth), the number of locations for R&D The Trillion-Dollar Opportunity increase r&D spenD 10% or cut excess reVenue rQ 2010 r&D spenD optimal r&D spenD unDer spenD expecteD profit increase price-to- earnings ratio (Jan ’12) expecteD market Value increase EXXON MOBiL $341,578 108 $1,012 $136,486 $135,474 $18,190 10.3 $188,089 CHEVRON $189,607 106 $526 $56,690 $56,163 $7,151 8.1 $58,208 CONOCOPHiLLiPS $175,752 112 $230 $100,350 $100,119 $10,112 8.7 $87,471 GENERAL ELECTRiC $149,060 102 $3,939 $19,947 $16,008 $4,515 15.3 $68,857 GENERAL MOTORS $135,592 105 $6,962 $15,570 $8,608 $6,264 4.4 $27,749 FORD $128,954 105 $5,000 $14,405 $9,405 $5,904 6.5 $38,084 HEWLETT-PACKARD $126,033 114 $2,959 $43,907 $40,948 $13,891 8.8 $121,548 MCKESSON $112,084 117 $407 $111,598 $111,190 $10,231 7.8 $79,393 iBM $99,871 100 $5,720 $10,359 $4,639 $2,124 14.5 $30,782 PROCTER & GAMBLE $78,938 101 $1,950 $7,816 $5,866 $1,946 16.9 $32,959 PFiZER $67,791 104 $9,538 $6,304 $-3,234 $3,235 15.0 $48,614 APPLE $65,225 105 $1,782 $9,468 $7,686 $2,573 14.6 $37,637 BOEiNG $64,306 104 $4,121 $8,142 $4,021 $2,441 13.9 $33,878 MiCROSOFT $62,484 107 $8,714 $9,210 $496 $3,198 9.4 $30,192 ARCHER-DANiELS-MiDLAND $61,682 111 $56 $29,947 $29,891 $2,386 14.0 $33,336 JOHNSON & JOHNSON $61,587 101 $6,844 $5,371 $-1,472 $1,473 16.0 $23,562 DELL $61,494 108 $661 $16,218 $15,557 $2,887 7.5 $21,740 UNiTED TECHNOLOGiES CORP. $54,326 103 $1,746 $6,192 $4,446 $1,654 15.2 $25,113 DOW CHEMiCAL $53,674 107 $1,660 $9,356 $7,695 $2,459 16.4 $40,425 KRAFT FOODS $49,207 103 $583 $7,254 $6,671 $1,310 20.4 $26,695 I then looked at the amount each firm actually invested in R&D... First, I calculated RQ for the 20 largest U.S. firms (by 2010 revenue). My research suggests that if the top 20 firms traded on U.S. exchanges had optimized their R&D spending in 2010 using the RQ method, the collective increase in market cap would have been an astonishing $1 trillion. ...and calculated the optimal R&D spend for the firms based on their RQs. I then estimated the change in profits result- ing from either increasing spending by 10% (for un- derspenders) or reducing it (for overspenders). The result: a whopping $1 trillion Finally, I applied the firms’ current price-to-earnings ratio to the notional extra profits to calculate the corresponding increase in market value. $ iN MiLLiONS $1,054,335,000,000 80  Harvard Business Review May 2012 HBR.ORGSpoTlIghT ON iNNOVATiON FOR THE 21ST CENTURy activity (technical breadth), and the number of prod- uct lines (product breadth). In-house research trumps outsourced R&D. RQ is negatively correlated with cooperative R&D, and the correlation between RQ and R&D is higher for internal R&D than external R&D. RQ varies for different types of innovation. RQ is positively correlated with product (versus pro- cess) innovation. Also, RQ is higher for companies that do incremental innovation (those that are new to the firm) rather than radical innovation (those that are new to the world). And RQ is positively cor- related with organizational innovations that comple- ment product innovations. These insights, while useful, don’t reveal why or how they act to improve R&D productivity. Nonethe- less, managers—and analysts—who use RQ will be able to tell quite a lot about what’s likely to happen to any company’s share price in response to changes in strategy and management practice. The case of Trimble Navigation provides an interesting example. The company was founded in 1978 to develop po- sitioning and navigation products utilizing LORAN (and subsequently GPS) technology. While Trim- ble’s initial markets were in military applications for which the technology was originally developed, it quickly applied the technology to commercial mar- kets, such as surveying and mariner navigation. Trimble’s net income grew fairly rapidly from 2000 to 2007 but then suffered a steep decline in 2009, largely owing to the recession. Trimble’s profit collapse of 54% was on par with the 62% average decline for U.S. publicly traded firms. But Trimble failed to bounce back. The average net income for U.S. firms is now 8.5% higher than the pre-recession peak, whereas Trimble’s net income remains 27% below its peak. Trimble’s RQ history reveals what may account for this prolonged slump. Trimble’s RQ steadily increased though 2004—the peak in RQ preceded the peak in market cap by about three years and in- creases in net income by about four years. In 2004, however, Trimble’s RQ fell 40%. Three years later, market cap plunged; the following year, profitability took a hit. The drop in RQ can be linked to changes in Trim- ble’s strategy. Throughout the 1990s Trimble devel- oped and patented many technologies, reaching a peak of 94 patents in 1997. In addition, Trimble was rapidly expanding the product markets in which this technology was deployed, according to the “Company History” on the firm’s website. In 2000, however, the company appears to have changed its strategy from one of in-house development to ac- quisition. This switch is documented in the “Com- pany History” (up until 2000, each year’s summary described a technological development; after 2000 there is no mention of developments, only of acqui- sitions) and by the decline in patents obtained per year, which dropped to almost zero. A shrewd ana- lyst might pick up on either or both of these patterns, but without RQ it would be difficult to tell whether the change in strategy was value-enhancing or value-destroying. Using RQ, a manager or analyst could easily see that the shift was value-destroying. The Promise At first, RQ might be viewed with suspicion, even hostility, at many corporations, appearing to be a device the R&D community could use to line its cof- fers. But concern over providing yet another metric for company managers to abuse should not out- weigh the substantial long-term benefits that RQ can deliver. Improving manufacturing operations creates value, but R&D is a basic engine of economic and so- cial growth. If enough firms adopt RQ and align their R&D spending and strategies accordingly, we should see a systematic improvement in overall corporate R&D effectiveness. The benefits for us all would be remarkable. HBR Reprint R1205D Firms can identify the marginal returns to R&D and the level of investment that generates the greatest market cap. Anne Marie Knott is a professor of strategy at Wash- ington University’s Olin Business School in St Louis and a director of the Berkeley Research Group in Los Angeles. 82  Harvard Business Review May 2012 HBR.ORGSPoTligHT On InnOvAtIOn FOR tHe 21St CentURy Harvard Business Review Notice of Use Restrictions, May 2009 Harvard Business Review and Harvard Business Publishing Newsletter content on EBSCOhost is licensed for the private individual use of authorized EBSCOhost users. It is not intended for use as assigned course material in academic institutions nor as corporate learning or training materials in businesses. Academic licensees may not use this content in electronic reserves, electronic course packs, persistent linking from syllabi or by any other means of incorporating the content into course resources. Business licensees may not host this content on learning management systems or use persistent linking or other means to incorporate the content into learning management systems. Harvard Business Publishing will be pleased to grant permission to make this content available through such means. For rates and permission, contact permissions@harvardbusiness.org.