Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
ECONOMI CS, POLITI CS, AND GOVERNMENT RESEARCH ANI) DEVELOPMENT Roger G. Noll* Stanford University and Lint R. Cohen* University of Washington Public off-imials and policy analysts long ago developed logically compelling ra~cionales for active government promotion of research. The basic argument has two components. Firs~c,- research and development (Rho) is desirable because it promotes economic growth, s treng~chens national defense, and contributes to national pres tige, not deco mention that it creates new knowledge that may be a valued end in its own right . Second, R&D tends to be undertaken insuffic iently by the priorate sector if left to its own devices. The second component, although having numerous elements, boils down for ache most part deco an argument about the appropriabili~cy of new knowledge People who ir~ves~c their resources in R&~) are unable to reap all the benefits to socie~cy that the ir discoveries create and, hence, will have insufficient incentives to inves~c as much in Rho as is socially des irable . Moreover, even if a result of Rail:) is appropriable, this can mean that those who possess it hairs a monopoly in the new technology or product resulting from the R&D. This is inefficient to the extent Chat monopoly prices are charged for the output of ache R&D . The argument described above in used to justify the echoes ma] or types of government-supported research and development. First, government suppor~cs R&D to imp rove the performance or to reduce the costs of the goods and services that it uses. For goods produced virtually en~cirely by go~rernmen~c, such as defense weapons systems, this makes sense, because government ~chereby can reduce itch own costs and direct R&D in ways ~chat serve its own needs. Second9 government can contribute ~co the general technological base of the nation by * This paper is a synopsis of the authors ' j oint research pro] ect and forthcoming book, The Technology Pork Barrel. The authors are grateful to research support from Ache Caltech Energy Policy Studies Program, . _ ~ ~ ~ the Brookings Institution, the Guggenheim Foundation, ache Center for Advanced S tudy in ache Behavioral Sciences, the National Science Foundation, and the Hoover Institution. - ;9 -
-supporting basic research in science and technology. Advances in basic knowledge can be applied widely in numerous industrial settings and so have great potential economic value. At the same time, new teas ic knowledge usually becomes pub kc quite quickly and so is difficult to convert to profit if done entirely by private enterprise. Consequently, there is reason to expect that coo li~ctle teas ic research will be undertaken if government does not support i Tic actively. Moreover, by supporting basic research governmen~c can assure that the results are available freely and disseminated widely to produce maximal economic benefits. third. governments may elect to undertake te~relopmental. work to produce new commercial ~cechnology for a specific sector of the economy if that sector is especially weak at innovating on its own. For examples the films in the industry may be so small and specialized that they cannot undertake commercial development of new technology efficiently, as has been argued to be the case in agriculture since the establishment of the land- grant colleges and ache Agricultural Extension Service in ache 19th century. For these rationales to be valid reasons for public action, two additional issues otust be resolved satisfactorily. First, a plausible case must be made Chat the government is able to select ache R&~) acti~ri~cies that most clearly fit the rationales . Second, government agencies must be capable of adept management of the R&D programs that are undertaken. Both of these issues refer to more than the cognitive capaci ties of government officials- -henceforth, it will be assumed here that elected and bureaucratic public officials are capable of effective program selection and management. She primary concern in this paper is with the institutional structure in which government officials operate, the incentives that affees decisions in the public sector, and the information flows on which decisions must be based. The principal message is that scheme are serious political barriers to the efficient implemen~cstion of some types of R&D policies. These political. phenomena, wines, taken into account, can alter substantially a prospective analysis of the expected benefits and costs of a go~rers~ment R&D program. POI-ITICAI. INCENTIVES AND R&D POLICIES lathe point of departure for this analysis is the recognition that certain broad incentives regarding policymaking are created by the political institutions of ache American system of governmen~c. The focal point of the American political system is elections. Whereas public officials have complex motives and may even seek to secure such public-spiri~ced objectives as improving ache nation's defense, prestige, and economic performance, all face the staric reality of reelection. A public official cannot achieve either altruis~cic or selfishly personal goals in public service unless he or she takes account of the necessity for elected politicians to be successful in seeking reelection. This fact not only influences the strategies of 60
elect ed o£fic~1s and their political appointees, but also Chose of career civil servants who must establish enduring working relationships with political officials in both Congress and the executive branch. Recent research on how the need for reelection affects policy choices has focused on three especially important aspects of the electoral process. lathe first is the inability of Forcers or politicians to make creditable or enforceable commitments for long periods of time (more than one berm of office). Because one session of Congress cannot bind subsequen~c sessions except by eons titutiona1 amendment, public officials cannot enter into longs ten contract with the public to complete a program as established and promised initially. Me second important feature of ache . electoral system is the relatively low informa~cion content of voting . to Pacers have only a s imp le. dichotomous s ignal to send to incumbent officials - - simple approval or disapproval ~ Moreover, as far as national policymaking is concerned, voters often castnot assign credit or blame accurately for a change in policy. Power in the American go~rer~ental system is so diffused that rarely can it be said that any single public official, even the president, is clearly responsible for a policy decision. Hence, in making decisions about voting' it makes sense for voters deco focus on ~ relatively small number of highly impotent issue'--and issues for which individual roles in policy can be assessed. The third element of the electoral system is that the net social consequences of a pa licy are only one element of its overall political importance. Individuals also care whether their communities and businesses receive a fair share of the go~rernment's expenditures for goods and services. In a few industries, economic weil-being is highly dependent on who gets government contracts. Hence, voters will evaluate political leaders in part on the basis of the distribution of program expenditures as well as the net Social benefits of the program that Me expenditures make possible. Obese basic features of the electoral process produce several incentives that are importan~c from the panic of view of government Red:) policies. Impartence. The political process is biased against programs that take a tong the to produce a payoff. First, elected officials who undertake such progrA--a face several intermediate reelection campaigns before ~ very long-te~ project will. produce benefits for which credit can be claimed. All other things being eq''=l, political actors will prefer to allocate the budget to policies which a more immediate payoff to help them get through ache next election. Plovers, ~coo, wil.l be wary of programs promising long-term payoffs, for they will have to take into account the prospect that, an election or two down the road, politics will change, ache program will be canceled, and they will be left having floated some of the costs without receiving any of ache benefits. Contracting firms and their employees - 61 -
also will lee skeptical of the desirability of undertaking long- term commitments to participate in a program for the same se e of reasons . This implies that firms will require extra financial inducements to commit their resources to a pro gram that may be canceled mid-stream. Risk Aversion. Public officials will be reluctant to undertake a program if they cannot identify its potential beneficiaries, both ache users of the program's ultimate outputs and ache recipients of its contracts. Remote and uncertain benefits are unlikely deco motivate voters. Uncertain~cies abou~c the ultima~ce contrac~cors undermine an official ~ s ability to take political credit for the program with the people whose incomes will benefit, and, in any case, the contrac~cing benefits may well go to constituencies that do not serve the reelec~cion interests of the poll ticians . Consequently, public officials would prefer to appropriate funds to proj ecus whose beneficiaries are Mown at ache beginning, so political credit can be claimed effectively for the next election. Risk aversion is induced further by eke fact that elected officials usually face ~ very high probability of reelection. Inc~zbas~ts are far more likely to win than lose. Hence, undertaking risLy programs--progr~s that have a high expected political value but also a substantial probability of failure--presents a relatively unattractive gamble If the pro gram fails, it can undermine a high reelection probability. If the program succeeds, it cannot add very much to a reelection probabili~cy that, in the absence of the risk . . . program, would be very high. Do No Positive }tarry. All else being envy, political leaders favor programs that are rela~ci~rely uncontroversial, where ache results are perceived widely to be benefits (and not really costs), and ache act of contracting does not inflict significant Teenage on competitors of the contractor. The latter is accomplished by having most competitors share in the program, either through extensive subcontracting (as with defense) or by dividing the program into a large number of separate, small components (as with basic research grants to universities). lThe way these three factors relate to R&D is as follows. A defining characteristic of R&D is that its results are uncertain and normally take cons iderable time to bear fruit . Research is a search for unknown new pieces of knowledge. Until the research is done, one does not know what new knowledge will emerge, or what use of it is likely to be made. Moreover, the act of pursuing an R&D program deco its ultimate end can take a very long acid, especially for ache radical changes in ~cechnology that are most likely to produce a fun~?amen~cal change in industrial practice. Finally, fundamentsI changes can do positive harm. For example, the firm. undertaking government- sponsored research may gain a substantial advantage over its competitors, even if the results are made publ~ico Indeed, whole industries may be made obsolete when the new technology is adopted. 62 At'
T".PLI~ATIONS FOR THE SET F.CTION OF R&D PROGRAMS The principal implication deco be drawn from the preceding summary of the political aspects of R&D decks ions is that very long- term, fundamental work to transform technology is unlikely to appear attractive politically. The most attractive R&D programs will have the following characteristics: (1) They can be connected readily to one of the very few salient political issues or which elections normally turn. One example is the stance of national defense; another was energy policy after the rapid increase in the price of oil in the 1970' s . (2) They can be spread easily to all the important components of the contracting industry. This is easiest to accomplish when the industry is concentrated (so that a few con~cracts and subcontracts do the job) or, in an unconcentrated industry, when the program can be subdivided into numerous small pt:oj ects without a substantial loss of efficiency . (3) They promise relatively shor~c-term payoffs in politically visible social benefits and expendi~cures to politically impor~cant constituencies . (4) They are unlikely to produce an embarrass ing failure that will lead to investigations and scandal. This is most easily assured when the government itself can~crols the decision to adopt the new technology and, thereby, to declare it a success. The type of R&D that is most attractive politically is short term in nature; is directed at the production of government goods ~ defense or space exploration, for example) so ~chat it can lead to utilization regardless of the shortfall in performance or overrun in CQStS; iS addressed to a widely accepted, generally us~contro~rersial national objective; and is undertaken without altering substantially the distribution of market atvaneages in the private economy. Least attrac~cive are program that address unsensational, long- term goals (such as long-term economic grouch); that require very large contracts for a relatively small fraction of an industry; Chat not only have uncertain results but can end in obvious failure; and that are unlikely to produce tangible evidence of success for a long period. For each of the three categories of research mentioned at the outset, ache implications are quite different. For research aimed at the production of government goods and services, prospects are best, but 8 few problems do emerge. Because all research is necessarily uncertain, some project are bound to fail, yet poli~cical incentives militate against allowing failure. Hence, the go~rers~ment will be prone to continue projects Chat have gone sour, even though Weir continuation can create difficulties in the efficient provision of - 63 -
governmen~c goods. This tendency is enhanced further by the political unattractiveness of canceling economically signs ficant contracts enj oyed by ma; or groups of constituents . This type of R&D also faces the problem cuff malting ache best trade-off between short- term proj ecus promising relatively quick but small technical advances versus riskier, longer acerb proj eats that may produce revolutionary changes in technology. However, because the government is the consumer of the product and because political officials will be held accoun~cable for the state of technology in important government areas (such as defense) over a sustained period, impatience asked risk aversion are likely ~co be least important in this particular category of proj ects . The principal difficulty with the second category of R&D activities, ache promotion of basic research in science and technology, is that rarely is baste research ~ response to a salient political issue. A notable exception is the connection between biomedical research and medical care policy--especially the war on cancer. But, usually, basic scientific research is difficult to connect to a symbolic national issue and, so, conveys less promise as a device for clan ming political credit O And, of course, basic scientific research usually is long tern in character. Nevertheless, the go~rernmene has succeeded in making it appear more immediate by relying on short- tern grants to researchers . At lasst, this creates frequent opportunities for announcing major grants ant for taking credit for each step along the path Coward the creation of important new scientific knowledge. Most problematic is ache third R&D category, the development of new technology for a specific industry or sector O Here ~ the best indicator of success ts whether the industry adopts the new technology freely, which happens only if it is a technical and economic success. The government, therefore, has relatively Is~ctle leverage on whether the image of the program is successful. As a resul~c, such activity carries greater political risks. In addition, if the industry is not concentrated or the R&D cannot be fragmented efficiency, some firms will not be able to participate in the program. Such firms are likely deco oppose the program, and SO9 t00, will their elected political leaders. Hence, the prospects for the program to surprise multiple appropriations through several future elections are dim. As.with programs for the production of government goods, commercially oriented R&D projects will tend to favor relatively short-term, low-risk ac~civities. But, there is no countervailing tendency working against the pursuit of incremental advances because of the absence of a salient national political issue and because of the inability of the go~rernmen~c to use procurement expenditures to cover at least some of the failures. The problem is especially acute for commercial R&D because, typically, the kinds of ~cechnology development proj ecus Chat are easiest to rationalize as requiring government ass issuance are ones that are risky, long term, and large compared to the capabilities of firms in the industry.
h~?LICATIONS FOR PROJECT MANAGEMENT In addition deco the forces influencing the selection of government R&D proj acts, two political factors affect ache choice of management s strategies . In the first and third categories of government research programs, the obj ective is to develop a useful new technalogy, either for ache government to use or for the private sector to adopt. Usually, such projects involve a sequence of activities, beginning with parallel research to test al~carnati~re design concepts before a commitment is made to develop prototypes and working models of the chosen method. The purpose of the early research is deco undertake relatively low- cost activities Chat increase the probability that a right choice will be made when the Acme comes to commit mad or resources to ache program. However, since the latter step is so highly valued by political decisionmakers as the fistic "risible sign of progress, political incentives push program management to foreshc~r~cen the research phase of ~ de~relopmen~cal research program and commit too early deco a design. The consequence is that programs, on ache average, cost more and perform less well than they would if they were managed witch maximal efficiency (chat iS9 t0 maximize the present value of benefits net of costs. Thus, the factors Chat make such programs generally una~ctractive- - the need for protracted research and the uncertainties surrounding the pursuit of the unknown- - also work to undermine optimal proj ect management by encouraging management to minimize Chose very disad~rantagesO Foreshortening research cuts the duration of the prod ect and resolves ache uncertainties regarding identification of some of the beneficiaries ~ including the contractors ~ . A second management choice is the extent to which the work in the program is done in-house by goverr~men~c agencies. Specifically, if the program is designed to advance technology in an industry in which there are more firms than can be supported efficiently, the political incentives are to build the research capability in-house to avoid, if possible, advantaging one competitor over another Generally, it is more efficient to hare the users of the technology develop the capacity for research. Undertaking research in one organization to advance technology that will be embodied in the product of another crevices ~mpor~cant informational problems. The users are likely to be more know3~edgeabl~e about their own needs than they are able or willing to communicate across organizational lines, especially for public consumption. Of course, if the users have subs~can~cial research capabilities already, and the industry is highly concentrated, the government research efforts can be in the form of contracts with industry research labs. lichen, organizations need not share information to assist in selecting research topics; however, this will work to lessen the extent to which new technical discoveries from government contracts are diffused through industry. 65
The third management issue is the periodic evaluation of the progress of a program and the decisions that music be made about con~cinuing, altering, or canceling it. The key paint to recognize is ache asymmetry of political incentives between the prospec~ci~re decision deco undertake a ma] or expendi~cure and the later decision deco cut it back. Because constituents and9 hence, political leaders discount the future excessively and evaluate programs heavily on current activity levels, the political benefits of promising an expenditure is the future are less than the political costs of canceling a current program. This leads to the creation of a threshold in the life of ~ mad or developmental research proj epic . During the relatively tow-cost exploratory research and concept de~relopmen~c phase, programs can be evaluated relatively cleanly because the political costs of killing them are low. Only a decision to extend the research phase faces difficulty' because of the impatience and risk-a~rersion incentives acting upon political decisionmalcers. Butch once the large expenditures associated with constructing prototypes and demonstra~cion projects are undertaken, ache distributive aspects of the program become much more significant. Important political benefits are ~co be reaped from granting the contracts - - and impor~can~ political costs to be suffered if they are canceled. Thus, technology development programs are likely to be coo sensi~cive deco bad news in the research phase, but too prone deco ignore bad news in the de~relopmental/demonstration phase. ILLUSTRATIONS FROM RECENT PROGRAMS Far too many examples of government support in developing new ~cechnology are available to make a definitive, exhausti~re treatmen~c possible in a single paper. Yet, empirical research on these types of programs has bean quince vigorous since the mid-1970' s. Al~chough government suppc~r~c for R&D has been common for 200 years, the intensity of federal support for other Ian defense systems has increased dramatically since ache early 1950' s ~ when the federal government launched the program to commercialize nuclear power for electricity generation. Me energy problems of ache 1970's caused a dramatic, if temporary, increase in go~rernmen~c support of nonde fens e R&l), directed primarily at the development of a wide variety of alternatives to oil and gas. The following examples of government R&D activities illustrate some of the points made in the preceding sections. ME CLINCH RIVER BREEDER REACTOR Am) THE SPACE SHUS:I~5 Although facing different fiscal fates ultimately ~ the Clinch River Breeder Reactor and the Space Shuttle programs had quite similar lives. Both were conceited as part of important, but not the most 66
crucial, national issues . I.he Space Shuttle was ache follow- on deco the highly successful and widely popular Apollo program that enabled Americans to explore the moon. Clinch River was a follow-on to the development of light-water nuclear reactors for generating electric power, a program t.ha~c in the early 1970's also was regarded widely as a success. Both programs were justified deco Congress as economically warranted: Clinch River as a means to guarantee relatively cheap fuel for nuclear reac~cors in anticipation of enormous growth in their use for genera~cing electricity combined with pessimistic forecasts about the availability of uranium from reliable sources; the Space Shuttle as a means of lowering the cost of launching very large payloads and of retrieving payloads for repair and relaunch . Bo th programs eventually ran into significant problems - - after the maj or parts of the expenditure programs were under way. Clinch River faced large delays and cost overruns, and, eventually, it became clear that it would not even be at the forefront of breeder technology. Moreover, ache demise of the nuclear power industry in the late 1970' s, owing to a slowdown in ache growth of demand for elec~cricity and unexpectedly high costs for nuclear pi~tS9 undercut the economic rationale for immediate commercialization of breeders. Nevertheless, the program continued for more than fire years after it was clear that it should be either canceled outright or redirected more toward research and away from early commercialization. lbe Space Shuttle also faced cost overruns but, more impor~can~cl~, a critical performance underrun: the demise of the Aspect Tug. " In the initial benefi~c-cos~c studies to justify ache program, approximately half of the benefits of the Space Shu~ctle were to be cap~cured from its ability to retrieve and repair satellites. The Space Tug was to be carried aloft by the Shuttle and used to transport astronautics to geosynchronous orbit, where most of the investment in satellites resides. By the late 1970' s, The Space Tug idea had been scrubbed as technically infeasible. This eliminated most of the Shuttle' s repair and retrieval benefits, which, in tub, mate the net expected benefits of continuing the program to negative. In addition, the costs and launch capability of the Space Shuttle turned out to be much worse than had been predic~ced an the outset. The crisis period was from about 1978 until the first launch in 1981. During that period, the optimal management decision probably was to stretch ou~c the program, produce fewer orbiters, undertake more research, and terminate the plan for excI''-Cive reliance on the Shuttle as America's launch vehicle for spacecraft. But, witch production expenditures in high gear, the program sur~ri;ved the crisis. These two programs illustrate the following. Both were tied to relatively important national issues that enter the normal poli~cical debauch. Both were directed at. relatively concentrated industries: aerospace and nuclear reactors. Both appeared economically attractive to political decisionmakers at the time they were proposed 67
and for most of the early research phase. Both then suffered sufficiently serious setbacks so that they should have been rethought and reprogrammed. But, the se~chaclcs took place after ache high-cost expenditure phase was under way, and, so, like supertankers, they continued to lunge ahead preachy much according to the old plan-°but with reduced capabilities and higher costs. Eventually, one was killed, but long after many useless dollars had been spent and only after the performance become almost laughably short of initial expectations. The other continued, primarily because it did produce a significant visible product, the television pictures of astronauts in space. Because of the relatively low Isvel of detailed knowledge that people have generally about program performance, the limi~ced but highly visible success of the Space Shuttle was ample Deco keep it going despite its poor performance as a commercial eechnology CO~UNI CATIONS SATELLITES 6 In the wake of Sputnik, ache U . S . go~rers~ment launched a maj or effort to make use of earth-orbiting satelli~ces. Ouring the 1960's, several generatior~s of new technologies for satellites were developed with go~rer~ent support. Whereas ache actual launches by ache government were of mixed success, the program was regarded widely as successful. Many technical innovations developed under the auspices of ache program were adopted by commercial sate111te users and manufacturers. Then, in the early 19709 s, the program was promptly killed witch Congress sally removing from the budget the developmental feds for the next generation of expert mends . To find ~ technical. or economic ~ ustification for the demise of the program is difficult, for the canceled projects do not appear to have any less promise than the ores that went before. Indeed, a major emphasis in ache canceled programs was the development of new methods for using satelli~ces for broadcasting--an innovation that ache private sector and other nations began deco explore on their own in the mid-1930's. Thus, it appears hat Ache government research program had detected a logical next seep for an important commercial development . What really happened deco the satelli~ce program was that ache industry turned against it. When the program began, the satellite business was highly concentrated--indeed, only Hughes Aircraft Company was capable initial.ly of punting satelli~ces in geosynchronous orbit. But, by the Come the basic project was being designed, ache industry had become quite competi~csve. Several firms were capable of building ache next experimental sa~cell~2-es. Moreover, the National Aeronautics and Space Administration (NASA), which administered the program, had attempted explicitly ~ to its credit) to encourage competition, to the point Chat Hughes was no longer even attempting Deco acquire the experimental contracts. Men NASA did award the contracts for its last proposed experiment, the ~ osers obj acted 68
s wrongly . The NASA inves~ciga~cors found the dec is ion to be a close call but certainly not tainted, and proposed to continue. But, ache issue had become ~ cause celebre with members of Congress who represented districts in which the losing companies were located. Congressional in~restigati-ons were launched' and the contract award decis ion was repressed. Meanwhile, telecommunications utilities grew deco fear that NASA was emerging as a competi~cor. Once the satelli~ces were launched, they could be used for years - ~ long past the period of Ad. Hence, the telecommunications ind',-ctry conditioned its support for the program oft the presence of severe limits deco the uses that NASA could make of its satellites. Amid this growing opposition from both manufacturers and users of satellites, the program was killed. The record of this controversy indicates that no technical or economic rationale for cancellation was developed. Some pointed with alarm to failures in past launches' but research projects are inherently riser, and these did not seem excessively us~produc~ci~re-- especially considering that the technologies developed by NASA were used subsequently in commercial satellites. The key argument appears to be that the new satellites would put the government in the business of competing with public utilities and determining which firms in a competitive manufacturing industry would have ~ lead in the next generation of satellites. Because the losers outnumbered the winners, and because the winners had no really substantial technical advantages over the losers, political actors decided that it was inequitable to. have such a program, regardless of the technical benefits it might produce. Because the nature of the project precluded letting everyone in on the program, NASA had no effective response Deco this opposition. lithe scale of the project was simply too small to allow the use of subcontracting and project segmenta~cion to overcome the disadvantage of harming key firms in the indus~cry. The unfortunate consequence, however, was Chat American R&D fell. precipitously after the program was canceled, allowing other nations to overcome their ~cechnical deficits by pursuing their own R&13 programs. SYNTHETIC lIJELS lye energy shocks of the 1970's gave rise to a major effort by the federal goverrusen~c to produce development of technologies for creating liquid hydrocarbon fuels from coal, oil shale, and other exotic sources. Obviously, at inception these programs were part of the national response to a highly importan~c political issue, the energy security of the nation. In the late 1970' s, funds were appropriated to begin major research program-=, a few of which were for construction of prototypes and demonstrations, but most of which were to begin the more basic and conceptual world that would be a prelude to the expensive demonstration phase. Soon thereafter, ache Synthetic Fuels Corporation was created deco finance the me jor proj ects . 69
Technically and economically, most synthetic fuels proj acts were not very exciting, and the early work offered no happy surprises. Moreover, not long after the 1979 oil shock occasioned by the Iran- Iraq war, energy supplies loosened substantially, real energy prices began to fall, and energy fell in priority as a national issue. All this took place before major commitments had been made to large construction proj eats (with a few notable exceptions ~ . Consequently, the budgetary plans for subsidizing synthetic fuels plants and production could be scaled back relatively easily without doing harm to a large number of contractors and employees. Unlike the Clinch River Breeder Reactor and the Space Shuttle, changed circumstances and sober reflection =bout the program' s merits could produce a reprogramming and scaling down because the most visible stage of the proj ecu had no~c yet been entered. SEMICONDUCTORS ANI) COMPUTERS / A maj or factor in the devel~opmen~c of the modern computer and microelectronics industry was support from the Departmen~c of Defense (DO1)) in the 1940' s and 1950' s ~ the Department of Defense sought high performance and was not terribly concerned with costs, because the new systems would be embedded in far more valuable weapons and communications systems. Hence, DOI) was a predictable, relatively sure market for any inventor or firm that could meet the performance requirements of the military. There are two key features of these programs. First' as a consumer of ache products to be produced, DOD had a clear idea of wham it wanted and could work with industrial suppliers to push technology toward its well-articulated objectives. Second, DOD would decide on its awn whether a new technology was wor~while--i~c did nor blare deco face an independent j udgmen~ from firms in the private sector, as was the case with the projects tisc~sed above. Hence, firms could be sure hat if the technical objectives announced by Ache government were met, they would enjoy commercial. success--~che agency specifying the obj ecti~res was also Ache potential customer O That these activities produced a commercially successful industry that was far larger than Ache market for defense electronics was something of a fortuitous stroke. Indeed, in the early 1950' s, conventional wisdom held that defense would dominate the trivet indefinitely. An interesting aspect of the DOD pro; ects is that, however imp organ they were in developing commercial production capabili~cies, Don was not particularly successful in identifying the most promis ing basic research. Fortunately, the agency was flexible enough to begin supporting projects once it was apparent from their early successes that they were promising. But, in computers and, especially, semiconductors, many of the major breakthroughs first were supported privately, by firms or universities, with DOD jumping in only after some progress had been made This provides some evidence of the 70
difficulties of communicating the relative merits of alternative lines of basic research and prospects for revolutionary advances across organizational boundaries. CONCL1JS IONS lye primary lessors to be learned from the marriage of the economic and technical analysis of R&D with the political factors affecting public decisions are as follows. First, whereas government has a relatively low tolerance for risk and failure, research and development programs ir~herently are uncertain ventures. Indeed, a program that does not have some failures is probably not being sufficiently innovative. Public officials are likely to be leery of programs that might fat1- - and especially of programs in which the failure plausibly can be laid on their doorsteps. The lesson: Governmen~c is not likely to be a very effective source of support for developing radical nest technological assurances for ache private sector. Probably, the most palatable fold of such policies is to distance govers~men~e from the results of any s ingle proj ect, such as by providing support to general industrial R&D facilities. lathe Synthetic Fuels Corporation is an example of this kind of organization, although it was created to pursue Technical ideas that were not very promising. Second, whereas government is impatient, a well-managed R&D program requires some long- term research and the flexibility to speed up or slow down as the results from the research begin to emerge. The lesson: When government does become commit~ced to develop a radical. new ~cechnology, it is likely to foreshorten the research period and thereby produce a less than satisfactory result-- including both performance undermans and cost overruns. Me most likely successes will be in areas where the duration of each generation of technologies is relatively short ~ satellites, for example) or where there will be some valuable' visible success indicators along the way (such as pictures of astronauts in space ~ . Third, there is no automatic match between the conditions that glare rise to inadequacy priorate R&D and the political conditions that malce a program possible and effective. If competi~cive industries with small firms are least likely to be inno~rati~re, ~chey also may be least at~eracti~s as a target of a federal research effort because of ache problem associated with creating winners and losers in the techno logy race . If radical new technologies present risks and gestation periods too great for pri~rs~ce actors deco accept, they equally may be una~ctrac~ive to impatient, risk-averse public officials . Fourth, ache distributional aspects of a program affect its fate . Even tong-ter~ programs can survive if they provide short-t~rm 71 ~
political benefits, through interim successes and contract awards that are visible publicly and that allow public officials to claim credit ~ But this two - edged sword is dangerous ~ Firs c, it can keep a bad program going, especially if ache primary political benefit is large, ongo ing government contracts rancher than technical performance. Second, it can kill a good program that has the unfortunate side effect of harming part of the industry. Me lesson: Do not overlook the distributional consequences of a program when it is being des igned ~ and give special attention to the extent to which the work is done iras ide the government, diffused among a large rtumber of small enter prises, or concentrated in a few large contrac~cs . Finally, by no means is continued support for teas ic research assured. It is most bountiful when it is tied to salient national political issues, as high energy physics is tied Deco defense and as biomedical research is tied to health. This suggests that continued support for basic research depends on the efforts of its supporters deco publicize ache results and their potential utile ity. It also is more viable politically if it is distrib-utionally conscious; that is, if tt is carried out by constituen~cs of a large number of poll tical ~ eaders. Heavy institutional geographical, or even topical concentration of effort invites attack by Chose left out and can lead to reduced support for all. Unfortuna~cely, this strategy is not likely to be ache most efficient one for advancing the base of technical knowledge. - _ ~ / c
NOTES AND REFERENCES 1. Morris P. Fiorina. Retrospective Voting in American Narional Elections. New Haven: Yale University Press, 1981. 2. Morris P. Fiorina and Roger G. Noll. "Ma; ority Rule Models and Legislative Elections, n Journal of Poli~cics, Vol. 42~1979), pp . 108 1 - 1104 . Barry R. Weingast9 Kenneth A. Shepsle, and Christopher Johnson. clue Political Economy of Benefits and Costs," Journal of Political Economy, Vol 0 89 (1981), pp . 642-664. Linda R. Cohen and Roger G o Noll. The El eccoral Connection co Intercempora1 Policy Evaluation by a legislator. Stanford, CA: Stanford University, Center for Economic Policy Research Publication No. 36, September 1984. See Jeffrey Banks. -Political Influence in Government R&D Programs. The Case of the Space Shuttle, ~ and Linda Cohen. "lathe Clinch River Breeder Reactor," revised versions of which will appear in Linda R. Cohen and Roger G. Noll. The Technology Pork Barrel, forthcoming, and current versions of which are available from the Caltech Environmental Quality Laboratory. 60 Linda R. Cohen and Roger G. Noll. 'the Political Economy of Government Research. The Communications Satellite Program. n Presented at the Annual Meeting of the American. Political Science Association, 1984. 7. See Barbara Goody Katz and Almarir, Phillips. "lThe Computer Industry, ~ and Richard Levin. lithe Semiconductor Industry. Me Both in Gove~menc. and Technical Progress. Edited by Richard Re Nelson. New York: Person Press, 1982. 73 - .