PAUL M. ROMER Universityof California, Berkeley NationalBureau of Economic Research CanadianInstitute for Advanced Research Implementinga National Technology Strategy with Self-Organizing Industry InvestmentBoards THEMOST IMPORTANT LESSON from the study of research and development, economicgrowth, and the history of technology is that there are more waysto arrangethe objectsof the physicalworld than humans can possibly imagine.Ultimately, all increasesin standardsof living can be tracedto discoveriesof more valuable arrangementsfor the things in the earth's crust and atmosphere.The personalcomputer that I used to write this paperis made from almost exactly the same physicalmaterials as the PC thatI boughtten years ago-about thirtypounds of steel, copper,alumi- num,plastic, and silicon, with bits of gold, ironoxide, andmiscellaneous otherelements mixed in. In my new PC these materialsare arrangedin a slightlydifferent way that makes them aboutfifty times moreuseful than they were in the originalconfiguration. No amountof savings and investment,no policy of macroeconomic fine-tuning,no set of tax and spendingincentives can generatesustained economicgrowth unless it is accompaniedby the countlesslarge and small discoveriesthat are requiredto create more value from a fixed set of naturalresources. These discoveriesare the productof a complicatedset of marketand nonmarketinstitutions that constitute what has been called a nationalinnovation system. This paper considers both the economic The authorgratefully acknowledges comments from the discussantsand the editors. This materialis based on work supportedby a National Science Foundationgrant. 345 346 Brookings Papers: Microeconomics2, 1993 opportunitiesand the politicalrisks inherentin attemptsto strengthenthis system. It proposesan institutionalarrangement that could providemore financialsupport for innovativeactivity and direct it towardareas with largeeconomic payoffs. Even more impressivethan the availabilityof a much fasterPC to do wordprocessing is the fact that my family drinksmilk from a cow. If the milk cow did not exist, no one would ever believe that carbon,oxygen, hydrogen,and a few other types of atoms could possibly be assembled into a chemicalrefinery that automaticallyconverts grass, water, and air into a nearlyperfect liquid proteindietary supplement. This refineryop- erateswith almost no humansupervision, is mobile so it can searchout its own inputs, can heal most mechanicalfailures, and can detect and neutralizeany microscopicpathogens that enter the system. It even makes morethan milk. It can build severalreplacement factories out of the same rawmaterials: grass, water, and air. Comparedwith the cow, my PC has all the sophisticationof a Tink- ertoy. And if the particulararrangement of atoms that make a cow can cometogether through a blindprocess of mutationand selection-nature's versionof trial and error-imagine how many other ways atoms can be arrangedthat are as astonishingand valuableas a microprocessorand a cow. The fundamentalchallenge in economicgrowth is to findthese new arrangements. Thisoptimistic potential for exploitingwhat Vannevar Bush called "the endlessfrontier" of scientificand technological opportunity is limitedonly by the difficultyof organizingcollective action.I It takescollective action to encouragediscovery and sustainthe free flow of ideas, andthe political mechanismsused to undertakecollective actionsuffer from serious inher- ent defects. But people createnew institutions,just as they discovernew technol- ogies. Policy innovatorsdiscover better ways to undertakecollective ac- tion, just as scientists, engineers, and productdesigners discover better ways to arrangephysical objects. VannevarBush did more than write a reportabout the endless frontierof science. He midwifedthe birthof the NationalScience Foundationand peer-reviewedresearch grants for basic researchat universities.He identifiedan importantopportunity for collec- 1. Bush (1945). Paul Romer 347 tive action and then constructedan institutionalarrangement to exploit this opportunityat minimalpolitical cost. Any contemporarydiscussion of a nationaltechnology strategy should be basedon a balancedassessment of the potentialbenefits from collective actionto spurgrowth and the risksinherent in undertakingcollective action throughthe politicalprocess. Even die-hardfree marketeersshould con- cede that there is an opportunityfor collective action to supportsome formsof basic research.On the otherside, even the most ardentadvocate of an activistgovernment must admit that the nation's fifty-yearexperi- mentwith an explicit industrialpolicy for the savings and loan industry shows how badly awry well-intentionedcollective action can go when it is implementedthrough the existing politicalinstitutions. Once the tension between the large opportunitiesfor benefitingfrom collective action and limited political and institutionalcapabilities for undertakingeffective collective actionis recognized,it becomesclear that innovationin institutionalmechanisms could be very valuable.Surely, the opportunityto experimentwith new institutionsis as great as the oppor- tunity to experimentwith new arrangementsof physical objects. The standarddichotomy in economicpolicy debates-market exchangeversus governmentintervention-does not capturethe complexityof the kinds of social institutionsalready used to achieve common goals. It also fails to suggestthe broadrange of new institutionalarrangements that could be tried.In its purestform the marketis a mechanismfor allowingindepen- dentaction by all individuals,with no explicit coordination.The govern- mentis a mechanismfor explicitlycoordinating the actionsof all people. Mosteconomic activity is supportedby institutionalarrangements that are intermediatebetween these extremes. After all, the modern capitalist economy is organizedneither as a marketnor as a government.It lets large numbersof people exploit the benefitsof collective actionthrough explicit, hierarchicalcoordination as membersof a corporation,and in this sense large parts of a capitalist economyfunction like a government.But an economy with many firms subjectseach to the disciplineof competition,so in this sense the economy is organizedmore like a market.The diverse populationof corporations is constantlychanging, constantly finding new anddifferent ways for large groupsof people to work towardsome collective end. Some corporations findnew ways to structurethemselves. Others die off and are replacedby 348 Brookings Papers: Microeconomics2, 1993 new andbetter arrangements. What a nationalinnovation system needs is somethinglike this kindof dynamic,something that lets us takeadvantage of mutuallybeneficial coordinated action and that uses the pressuresof competitionand a markettest to shut downineffective institutional ar- rangementsand to rewardpromising ones. To show how such a process might work, I describea specific policy proposalconceived with this end in mind. Whateverthe finaljudgment is on this particularproposal, the general analyticalpoint will, I hope, be clear.To encouragetechnical change, it is not enoughto call on business leadersto be more innovative.Policymakers must themselvesbe willing to experimentwith new institutionalarrangements. Self-Organizing Industry Investment Boards This proposal specifies a process, not a specific policy. To see how it would work, consider a hypotheticalexample. Suppose you run the Acme Widget companyand I run ConsolidatedWidgets. You and I and many other widget makers come together and decide that there are industry-wideopportunities and problems that independentaction by individualfirms cannot address. You think, for example, thatuniversity professorscould be doing useful researchon questions about the basic principlesof widget design and manufacturingif they just had the funds and the incentive to do so. I believe that the upstreamindustry that manufactureswidget-making equipment could be designingmore useful specializedequipment. Underthe proposaloutlined in this paper, widget makerscould take collective action on these mattersby following the steps specified in a generalpiece of enabling legislation. They would start by petitioning the secretary of commerce, giving the argumentfor provision of an industry-specificpublic good. The secretarywould then hold a hearing to certify that the proposal for collective action addresses a genuine public need. If the proposal passes this test, an election would be held in which all widget manufacturersvote whetherto levy, say, a 1 percent tax on widget sales. If a large enough fractionof the industry(measured as a fractionof total sales, as a fractionof the total numberof firms, or some combinationof the two) votes in favor of this initiative, a tax Paul Romer 349 backedby the full force of law would be imposed on the entire industry. The proceeds from the tax would not, however, go to the government. Partof the presentationto the secretaryof commerceand then to the industrywould outline plans for creating industryinvestment boards. You will take the initiative to organize an investment board that will fund university-basedwidget research. I will organize a board that supportsthe developmentof widget-makingmachinery in the upstream industry.Both of these boardswill function as pass-throughs,accepting tax obligationsfrom contributingfirms and using these funds to support researchin universities in the first case or development in upstream firmsin the second. After the tax is passed, the two boardswould solicit the