Papers Commissioned for a Workshop on the Federal Role in Research and Development (1985)

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.

FEDERAI" Fu1IDI`IG OF RESEARCH ANTI) DE1JELOPME^NT IN TRANSPORTATION: THE CASE OF AVIATION David C. Mowery* Carnegie-flellon University The U. S . commercial aircraft industry has long beer a maj or beneficiary of federal research and development (R&D) programs managed by the National Aeronautics arid Space Ad~ninis Oration (NASA) and the ensued services. I-his paper discusses the impact of federal research investment on the technological and economic performance of the commercial aircraft industry, focusing on the role of such investment within a policy structure that has affected both the supply of innovations and the demand for the embodiment o f those innovations in new aircraft des igns . Government policy in the aircraft industry not only has supported precommercial research in civilian and military aircraft technologies, but also has played a major role in supporting the diffusion of the results of that research. The focus of this symposium is the nature and measurement of the economic payoff to the federal government's R&D investment. Clearly, this payoff is influenced heavily by the structure of the institutions and policies mediating the expenditure of public funds on research. Does the policy structure associated with federal support of innovation in the commercial aircraft industry serve as useful model for similar federal programs in other industries? Does NASA represent a prototype of an industrial generic technology research program? What factors contributed to NASA's success in supporting technological development within ache aircraft industry during the past 70 years? The apparent success of the aeronautics research programs of NASA and its predecessor, the National Advisory Committee on Aeronautics (NACA), contrasts sharply witch ache results * Data and other important assistance were provided by Gene Kingsbury, Department of Commerce; Virginia Lopez, Aerospace Industries Association Research Center; Joan Winston, Congressional Research Ser~rica; and Margaret Grucza, National Science Foundation. This paper draws on research funded by the American Enterprise Institu~ce and the Cen~cer for Economic Policy Research arc Stanford Unifiers ity . Mark Kamlet provided useful comments and suggestions, and Pamela Reyner provided secretarial assistance. None of these individuals or organizations are respons ible for the errors, interpretations, or conclus ions of this paper . - 301 —

of the federal go~rernment's attempts to intervene in ache innovation process in ache U. S. au~comobile industry. Evaluation of the rote that federal support has played requires some assessment of the structure of the innovation process wi~c}~in the commercial aircraft industry. If federal programs are Deco be successful in supporting incus trial innovation, they must be sensitive to ache requirements and characteristics of ache innovation process in specific industries Therefore, this paper incorporates some discussion of the relationship between the characteristics of ache innovation process within the aircraft industry and the structure of federal R&D programs in the incus try . Technological progress, economic perfo ~ Lance, and inds~s~crial structure in the commercial aviation industry during the postwar period are discussed first. What section is followed by an examination of the sources of technological change in Ache industry, including a discussion of R&O investment. then, an examination of "he structure of pa licy toward the commercial aircraft and air transpor~ca~cion industries considers the role of the Civil Aeronautics Board, as well as the various public and private sources of R&D suppor~c, in contributing to technical progress. A brief comparison of publicly supported research in the automotive and commercial aircraft industries is followed by a conclusion. ILN~OVATION AND ECONOMIC PE:RFORMAN7CE IN lisle U. S . COtMERCIAL AIRCRAFT INDUSTRY The U.S. commercial aircraft industry has been ir~no~rati~re End internationally competitive throughout the postwar period. In 1983, total aerospace incus cry sales ~ including missiles and spacecraft) amounted to nearly $76 billion ($28 billion in 1972 dotiars)! more than 2 percent of the gross national product. Within this total, sales of military and civilian aircraft, engines, and parsecs were valued at $41. 2 billion ($15. 2 billion in 1972 dollars ~ . The contribution of aircraft to U. S . foreign trade in 1983 was important as well; exports of aircraft, engines, and parts equaled $15. ~ billion ($7 billion in 1972 dollars), the largest single category of U. S . manufactured exports. The aircraft industry is a maj or investor in research and development. Expenditures for R&D (nearly 74 percent of which were financed by federal funds in 1983) amounted to 14 percent of the value of 1983 shipments, a level exceeded only by the electronics industry. The aircraft industry also has important links (through its demand {or components and parts) with other high-technology industries. Indeed, a central reason for the rapid technological progress characteristic of this industry is its ability to draw on and benefit from technological developments in numerous other industries .4 — 302 -

The ability of the commercial aircraft industry to benefit born technological developments in a wide range of ocher industries reflects the fact that ~ given aircraft or engine design integrates number of exceedingly complex subsystems, involving electronics, hydraulics, and materials technologies. The interaction of these individually complex systems or components is crucial to the performance of a design, yet often is extremely difficul~c to predict . Cons iderable technological uncertainty thus pervades the development of a new airframe or engine design, rendering the systems integration and des ign phases critical to the introduction of a successful new product. Therefore, R&D investment within the industry is dominated by the integration of components and prototype design and testing, rather than by basic or fundamental research (see below for further discuss ion) . The continually changing character of the market and of commercial aircraft technology contributes deco the length of the des in phase in aircraft inno~ra~cion. In an effort to accou=oda~ce the broadest possible group of purchasers, ma: or firms produce dozens of "paper airplanes ~ prior to the decision ~co launch the development of a specific design. S teiner cites the excruciating pain of trying deco achieve a common denominator among varying airline requirements. All commercial programs go through a s imilar process and the engineers mus ~ work wi eh a great many aired nes, notSjust the few who are most likely to become launch customers. In the design of the Boeing 727, this process took two and one-half years and produced al least nine separate complete designs for the aircraft. The design definition phase for the Boeing 767 lasted nearly six years. Once a producer decides to introduce a specific design, however, speed is essential. 6 Another reason for the tmpor~canca of product des ign in this industry is ache fact that an ~ircraf~c design is produced for a remarkably tong time. The Boeing 727 was produced for 20 years, and the manufacture of the DC-S extended from 1957 through 1972. While these aircraft were produced over a lengthy period, their designs were modified in major ways, through "stretching" the fuselage to accommodate additional passengers, or retrofitting an airframe with new engines. Producers now devote considerable at~cen~cion to des igning aircraft that are amenable to such s trenching . The des ign of aircraft with a high potential. for screeching requires, among other things, ache development of wing das igns that will allow significant increases in aircraft payload without major mod: Cations. In some cases, this feat may intone the substi~cu~ion of new materials, such as composites or alumin~-lithium alloys, for older ones in an existing wing design. Production facilities also must be designed to accommodate variations in fuselage length.

The economic significance of stretching an aircraft design is difficult to overstate. It allows the high fixed costs of design and development to be defrayed over additional sales in a new market segment; the incremental costs of stretching an aircraft design rarely exceed 25 percent of the original development costs. And, it allows the bulk of the cost reductions from the movement down the learning curve in the production of the original design to be applied to an essentially new aircraft. Thus, design decisions influence both the success of the initial model and its potential for the stretching that may enable a single aircraft design to serve additional markets. Other incremental modifications are made throughout the life of given aircraft or engine. Such changes rely heavily on information gained from close monitoring of operating experience after the introduction of an aircraft. The importance of this monitoring function and of product support ( spare parts supply and field service) makes the establishment or existence of a global marketing and product suppor ~ organization critical to marke ~ acceptance of a new aircraft des ign . lye need for a global product support and marketing network alsO7acts as a barrier co entry of new firms into the aircraft incI~;stry. Another source of entry barriers is the high and rap idly growing cost of new product development. Development costs have risen dramatically, increasing ( is constant dollars ~ at an average annual rate of nearly 20 percent during ~ 930~1970, considerably greater than ache average annual rate of growth in aircraft weight of ~ .5 percent. Development of the Dou'1as DC- 3 in the 1930 ' s cost roughly $3 million ~ sea Miller and Sawers ~ . The DC - 8, introduced in 19 58, cost nearly $ii2 million, while development of the Boeing 747, production of which began in the early 1970's, cost $l billion. More recently, development of the Boeing 767 is estimated to have cost $~.5 billion, while estimates of the development costs for a new-technology, 150-seat transport range as high as $2 billion. the V2500, a new h~gh-bypass-ratio engine, is expected to require $~.5 billion for development. The rapid growth of such costs means that an increasing proportion of the costs of introducing a new aircraft is incurred during the phase of greatest uncertainty concerning market prospects and technical feasibility. In addition to their sheer magnitude, which is appreciated best by comparison with total stockholders' equity of a firm such as Boeing (in 1984, roughly $2. 7 btIlion), these high fixed casts result in a falling short- run average cost curve . The behavior of costs in the industry is governed by Two other factors. The first is the relatively small production runs of most aircraft. Since the introduction of the commercial jet transport in the early 1950's, only 4 aircraft designs (the DC-9/MD-80 and the Boeing 707, 727, and 737) out of 23 have sold more than 600 units. The aircraft business is not one characterized by high throughput. While, typically, average annual production rates are low, they are subject to wide - 304 -

£1uc~u~ ~ions, pith the peak production rate being as much as eight .irs~es as large as the trough output rate. A final dimens ion of cost behavior in the industry is the importance of reductions in variable posts as a function of cumulative output- - the well -known learning c:yre, first documented in the production of airframes in 'world War II. Cost reduction offer the course of an aircraf~c's production history is dr~atic--most estimates suggest typic a doubling of output reduces uni~c.costs by as much as 20 percent. trHE RECORD OF TECHNICAL PROGRES S The U. S . air transportation industry is the primary beneficiary of technological progress in commercial aircraft. Accordingly, one index of ache rate of technical change in commercial aircraft is the growth of productivi ty in air ~cranspor~cation . To cal factor productivi~cy in that industry has grown more rap idly than in virtually2any other U. S . industry during the postwar period. Kendrick' concluded that the average annual rate of growth in tctal factor productivity in air transportation was 8 percent during 1948-1966~ higher than for any ocher industry. Fraumeni and Jorgenson 3 found ~cha~c total factor productivity growth in air transportation was exceeded only by that o f telecommunications . Clearly, such product~vi~cy growth reflects more than innovation in commercial aircraft Air traffic control improvements, insertions in ground-based navigational equipment, airfie Id expansion and moderniza~cion, and other enhancements o f the overall air transportation network, many of which were financed by the Federal Anacin Administration (FAA), have played major roles also. Therefore, analysis of technological progress in commercial aircraft requires the use of indices of aircraft perfo seance, rather than coccal factor productivity measures for the tndus fries employing aircraft. Two such indices of commercial aircraft perfo~-=anc~ are available: seats multiplied by cruising Speed (AS*VC), and direct operating costs per available seat mile. While the two measures do not translate into an index of total factor productivity, they have the advantage of being available separately for different aircraft designs. Table ~ displays the evolution of the two aircraft performance measures for a sample of piston and jet engine aircraft during 1938-1983. With the introduction of successive generations of aircraft, AS*VC has risen, and costs per available seat mile ~ ~ n 1972 dollars) have fallen. the sharp drop in operating cages represented by the OC-3 can be seen clearly in the table. Another maj or drop in seat-mile costs came with the introduction of the wide -body transpor~cs ~ such as the Boeing 747', the Lockheed L- 101l, and the McDonnell Douglas DC- 10), incorporating large, - 305 -

TABLE 1 Measures of Aircraft Performance. 1938-1983 fat A, ~~ 1 C ~—_47 A. O. ~ . 97. ~11 ars 9_~3, 10. 8 OC-+ ~ t94`j' Foci; ~ t9~9' 7. 47 —,`:? ~19~-) 94,j,j ~ i9=,j, L7C ( 19~4) _J_tj') · ( 1959) _ Tic ~—7~7 ( thy) 7_~:)`jC, ( 19~9) _. a' ~ —727—I (,Y~) - 13~. ~_, -, - 2,;,,;, ~ IS_ ) ~—,(1t}~;, ~ . &~ ~—7_~7 - t'tj ( 199., 4'`;)00 5-747 ( t963. - .) ~1 St~tjO 1 c At:, B-757 ( 1993) 77-S:,':) ~ . '4. , E-767 ~ 198~) 91.~O t. 65 OC-9 ( 1 966 ) I. 68. OC-9-80 ~ 198~) 6~100 I. 59 ~ continued ~ . . 306 -

·.~L~ ~ ~ cornea l:C3-b' ( 1'78~) 86604 OC-3 ( 1 95~') 1 o SO 20 23* D121~-10 ( 198-) 131290 10 - 7 L-~1 1-Si`:~O ( t~S ) 11~2t:~O 1.74 -~` ,<~-94 ( 1 9153- ) 84 1 2C, 2. 73 F: rst Year of aperat: an. Data for pre- 1983 period taken from Rosenberg et al . Technological Change and Produc~ivi By Growth in the Air Transport Indusery. NASA Technical Memoranda 78505. Washington, DC: National Aeronautics and Space Administration, 19 7 8 ~ Data for 19 8 3 taken from Aircraft Operaring Cose and Performance Re?ore, Volume 7 . Washington , 1: C: Civil Aeronautics Board , ~ 9 84 . 2 Data for pre-1983 period taken from R. Miller and 3. Sawers. The Technical Developmen~c of Modern Aria con. London: RoutI~dge & Kegan Paul., 1968, Appendix T. Date for 1983 Fallen from Citrti Board, op. civic., above. Aeronauts _ 307 -

high-bypass-ratio j et engines . Awhile the latest generation of aircraft, including the Boeing 757 and 767 and the Airbus A300, exhibit higher direct operating costs per avail able seat mile than did the first wide-body aircraft, this is an obvious result of the fact that fine number of available seats in these aircraft is much smaller than is true of the early wide-body transports. Cruis ing speed and capacity (ASPIC ) declined sharp ly with the introduction of four- engine transports t~ediat~ly after World War II, and they dropped further still with ache introduction of j et-powered transports . Typically, speed and capacity gains have been obtained at the expense of operating costs; only the first wide-body transports combined major increases- in available seat Precocity with significant declines in direct operating costs per sea. mile. This record of performance improvements in commercial aircraft can be summarized ~n the estimate by Rosenberg et al. 6 that between the appearance of the monocoque airframe in 19 3 3 and the introduction of the 747, costs per seat mite dectinet7tenfold, while passenger capaci~cy and speed rose by ~ factor of 20. voucher measure of technical progress estimates the resource savings associated with improved technical performance. A calculation of the " social savings " resulting from technological progress in commercial aircraft compares the costs of air transportation using ache 1983 U. S . fleet in scheduled domestic service with the costs; associated with the exclusive use of OC-3' s in that service. lo lathe choice of the DC-3 for comparison is a fairly conser~rati-ve one, inasmuch as that aircraft was characterized by low operating cos Is relative deco other. contemporary designs; moreover, by 1939, ache base year for the comparison, operating costs had declined from the levels experienced immediately after the introduction of the aircraft in 1933. In 1983, on the other hand, both ache Boeing 767 and 157 were skill relatively new aircraft ~1983 was the first full year of operation for the 757 ~ and, therefore, exhibited operating costs above then r long-run level. In additions of course, the substitution of the 21- seat 1)C- 3 for the current fleet of larger aircraft almost certainly would produce gridlock at ache nation's airports due to the huge increase in flights, landings, and takeoffs chat would be necessary. Moreover, the value of more rapid travel is ignored in this calculation. lathe calculation may somewhat overstate ache acted savings, however, inasmuch as the higher real costs of transporting current traffic toads in a fleet of DC-3's would be reflected in much higher prices for air transportation and, consequen~cly, lower levels of demand. Nonetheless, the calculation suggests that the 1983 dolce of passenger traffic would cost nearly $24 billion (in 1972 dollars), rather than the current costs of more than 65. ~ billion (also in 1972 dollars) . According to this measure of technical progress, then, innovation in commercial aircraf~c has saved more than 75 percent of the cost of achieving todays Clever s of domestic air passenger traffic volume with 1939 equipment.

An additional feature of technical progress in aircraft Is revealed in fable 1, which presents operating costs ~ in 1972 dollars for both ache year of introduction and 198 3 for such aircraft as the Boeing 727, the DC- 8, and the DC- 9 . An ~ mportant element of technical change and performance improvement in co~erc ial aircraft occurs during the operating life of a given airfare design, in the "best phase' of the i nnovation process ~ see Enos and Rosenberg en al O for further discuss ton) . For the Boeing 247, the first monocoque passenger airframe des ign, seat-mile operating costs decl ined by more than 25 percent during 19 3 3 -1940 . The Lockheed Electra L- 188, a four-engine turboprop, exhibi~ced an annual rate of cost decline of roughly 7 percent during tics period of service, while operatin,2costs for the Boeing 707 declined at an annual rate of 8.7 percent . Successive stretches of the fuselage of the DC- 8, as well as modifications in the wing design, increased AS*,IC from 6 2 . 50O2 for the original DC - 8/DC - 10 des ign to nearly 90, 000 in i983. 3 Similarly, direct operating costs for the DC-9 declined by nearly 50 percent during 1966-1983, due to a succession of stretches of the fuselage and the employment of new, fuel-efficient engines. These operating cost reductions reflect modifications in aircraft des ign and improvements in aircraf ~ utilization and maintenance, bo th of which incorporate important elements of leaving in use (see Rosenberg for further discussion). Operating cost reductions depend heavily on the gradual accumulation of knowledge during flight operations about the-2erformance characteristics of an airplane and its components. Fo.r example, it is only through extensive use that detailed knowledge is developed about engines' maintenance needs, minims servic ing and overhaul requirements, and the like . Such learning by using is significant for several additional reasons. It is not unique to commercial aircraft, but characterizes a number of complex capital goods and even software technologies. In addition, it raeans chat ache product support networks of established producers of airframes and engines are important sources of innovations. Careful monitoring of aircraft performance is necessary so that subsequent models of the same design can incorporate the lessons learned from the operating experiences of early models. Finally, much of ache learning by us tng in ache aircraft incus try is an excellent example of the ways in which an industry organizes to take advan~cage of "user-ac~civen patterns of innovation, as discussed by von Hippel. 5 Airlines, along with technologically sophisticated purchasers in other industries, are ma; or sources of suggestions for changes in aircraft des ign, maintenance, and even production practices. The integration of after-sales maintenance and marketing data with design engineering that ache films in this industry have developed to take advan~cage of their users' experience is an excellent example of ache ways in which Ache characteristics of this industry' s product technology have influenced the org~niza~cion of R&D . - 309 -

R&D I~1EST!£ENT This section focuses primarily on the support of commercial aircraft R&D by botch public and private sources. However, eve commercial aircraft indus~cry' s innovative performance has been aided by additional factors, including innovations in other industries, such as materials and en ectronics g and the ability of aircraft firms to exploit operating experience (nlearning by yoking," noted above) as a source of incremental technological change. To assess the significance and composition of the overall federal research investment in commercial aircraft, this section reviews ache sources of financial support for aircraft in;'ova~cion, including ~ ndus~ry- f inanced R&D, military procurement and research funding, and federal funding for research in civil aeronautics and propulsion techno] og' es. DATA SOURCES AND CONSTRUCTION Analysis of it&l) investment in ache commercial aircraft industry ~ s hampered by ache absence of consistent time series da~ca. The modest size of Ale pre-945 R&D investment, the substantial position shift in the technology of commercial aircraft as piston engines were replaced by turbojets, and the absence of reliable data for that period have resulted in the omission of the pre - i945 period from this discussion. Construction of data on R&1) investment during 194:-1983 relied on two inconsistent sources. R&D expenditures for i945-1969 are well documented in the study by Booz, Allen and Hamilton Applied Research, Inc., for the joint NASA/Depar~ment of Transportation analysis ,; technological change in the commercial aircraft industry. This source is particularly useful because of its detailed breakdown of aircraft industry R&D expenditures by source, by research activity (for example, basic versus applied), and by functional area of research (for example, p~gpuision). The data were discussed at length in Mower~gand Rosenberg arid were employed in modified form in Teriecky; 's analysis of productivity growth in air ~cranspor~catton. The Booz-Allen data also are useful because of the ir separation of incus try ~ financed research expend) Scores wi Chin the aircraft industry from expenditures covered by federal reimbursements for military procurement and development contracts. Prior to t959, such reimbursements were restric~ced to allowable overhead charges on military research, de~re3-opment, and procurement contracts. However, after 1959, reimbursement was allowed for "independent R&D" (IRKS) as a part of the overhead costs of Military contracts (see Reddy, Levy and Teriecicy;, and Terieckyj 3 ) . While it represents an important source of federal support for industrial research in fines producing commodities for the armed forces, JR&I) typically is reported by those firms to the National Science Founda~cior~ (tISF) as industry- financed research investment . - 310 -

Me pus I- 1069 data are less detailed and lack the functional and o ther disaggregated breakdowns provided in the Booz -Allen study . Data on the Department of Defense (DOD), NASA, and other federal research support for aeronautics were taken from Aerospace r aces and Figures . Industry- financed R&D expenditure data after 1970 were taken from the product field data compiled by NSF on the "aircraft and parts" industry (SIC 372~. 4 The NSF data have several problems- - they do not include basic research activity, and the JR&D reimbursements received by firms in the industry presumably are reported as industry- financed research investment ~ To deal with these problems, the reported basic research expenditures for ache aerospace industry (a broader category than aircraft and parts ~ were added Taco the figures, after which the sun was deflated by the average share of reimbursed R&D expenditures for industry- financed and industry-reimbursed R&D investment for the 1945-1969 period, as reported in the Booz-Allen study. The excluded portion of " industry- financed" R&D investment was added to reported militias R&3 expend) Cures . The result of these various procedures is a time series that is imperfect but consistent within the 1945-1969 and 1970-1982 periods. Table 2 and Figure 1, respectively, contain tabl,lar and graphic representations of these R&D series, which are reported in terms o f 1972 dollars . Estimates of ache c~uiati~re R&D investment from pub kc and private sources also are reported in Table 2.3 Total R&D expenditures from all sources rose by more than 224 percent in real terms, from $963 million in 1945 to roughly S3 ~ billion in 1982. Bill tary R&D investment grew rapidly in the aftermath of World Car II and the Korean War mobilization, reaching ~ plateau in the 1950's and early 1960's. Following 1962-1963, however, military research funding declined through the late 1960' s. Hili~ca~ research funding increased from $820 million in 1945 to $2. ~ billion in 19B2--if anything, this lancer figure may overstate total military research funding, as a result of ache effor~c to adj use ache pose- 1969 figures for T R&D . Throughout the pos twar period, the military portion of total R&D expenditures never fell below 65 percent. Throughout the postwar period, NASA research funding grew at ~ very modest rate, and it has remained essentially constant since the late 1960' s . The diminishing importance of NACA research funding in aircraft also is apparent in Table 2. Whereas, in 1945, NACA research support exceeded industry- financed industry R&1) investment, by the mid-1950's, NACA accounted for less than 20 percent of infuse financed R&D expenditures. By ache late 1970's, however, as industry-financed R&D Declined in real hems, the relative importance of NASA funding increased substantially, even as the rate of growth of that funding declined. Expenditures by the Atomic Energy Commiss ion supported research on nuclear propuls ion of aircraft and space vehicles, while the Federal Aviation Administration supported work on avionics and ache supersonic transport during the 1960' s. - 311 -

I\BLE 2 Annual and Cumuiat~ve R&D Investment, 1945-1982 (1972 dollars in ;:illions ) Federal C: v: l Total Ind~s-r`~— Aer~naut: cs Pt: l i tary Feceral F: nar-ec: Year NASA R.'. 3 ~'~D R&D ~'-: ~ incl . I~SA' 1945 79.16 631.79 3-~. :iS 9(':. _7 it':'O S~ 194it 84 S a~o.'t6 9~2.16 1':l~8.7~ ~t~.78 1947 6~).6tl `',:~ 6~ 7':~ O', 767~ ^t8 746 7- 1948 79.25 8,. C)- *,a~. ~- 766. u4 9,:l. '7 1949 l(:)v.9:i l~t4.70 71538.'7 89~.'- :~.~' 1950 97.01 111.94 3'-.76 5_4.7() 16t9.7i3 t 9'1 l C'S. '3 1 -7.8- l l e5. ~4 1 1.,49 297 -~ 19.- 195.1`o -19.~4 1984.-~8 -l(j'.~t~ 475.41 195;~ 1-9.~5 17Ct.07 -574. a. 2744.9'\ '7~.~3 19:i4 9 .44 1,4.4:i _79_. ~9 ~9_7.7' .7c.47 19S~ 77._CI 1_~.~;6 _.87. l7 -711:~.:iZ '2~.3Z 19:i~ Sl.~1 16(:'.a~ -'6'.1'j 7~:.9- 5~t~. ;(, 19:i7 77.~4 -~)lj.,1 -ofi4.3S _8:iS.1~ o':~4.~)1 1958 oat l8 _,jl e_ -78t)._C, -°,81.5 - 53?.~5J 19:i9 71.C)l '_4.85 ~9.'_ _794.~S 't:)1.48 190':, 46.58 10.89 2190.S1 -41~._9 4~,a.a9 19~1 56.-8 ~_~).79 2_9'.3' 2'16.:i9 4~.'= 1 9 0 _ ~ _ . _ 1 5 _ . 9 7 . _ 3 6 . 1 ~ _ 4 T 9 , ~ ~ 9 ~ ~ ~ , ~ _3 196_ °~. cfi ~tJv) c 84 277~.85 '977. ~,a ~-s. -0 1964 l l S. -S 192. - 1 o^ ~.4d, g-~.77 437. '3 1965 1_~7.1C1 ,~:~5.6'S ''t:)5.38 _711.~_ 474.a6 1966t 14, . ~ ~ ~29.4, -6- 1. C'9 ~9-'j . ' '79.43 1967 1~9.41 45 .adt ~441.-1 -89 u7 714 9 196t8 C'7.~7 _-6.v~ 24~9.:19 ^7SS.1S 91'.76 1964 .48.3:i _93. ~t_ .111.7S _~;1U. -7 7~:) t . i't 197CJ - 17.7-- ,~. ~IS -41 t).95 -6t74. o ~ ~t79.7'' 1971 ~ l S.7'' ~94.79 -_8~ .44 -S77. 2_ -~6t .1 ~, 197- 2_it . OO _~ ~ .0~) 4 9. ~O ~76tO . dttO 513.4~ 197 , 296t.1_ '~St7.08 _c~S'.6~- -449.71 419.73 1974 -41. ~- ~I;,''.3~ 18~0. Sl _1(:~6,. ~ 378.16 1977; 249.6~t 3~:,9.4- 1 ~71.56. 1379.99 3':l6.81 1976 245.6tti =`39.9fJ 1779.350 2Ct89.4v :;44.46 1977 -70.00 :5 ~d,.43 19:i3.17 __89.6'C, 37~t.83 1978 -91. ~3 -:S4.0t) 2-~8.99 26t9~.99 .l- 6,a 1979 317. ~3 -7_. I' 19-6t.5'7 -_~(;1. :9 o;!4.23 1980 313.90 ~67.15 19-·.::i 2~30.50 6138.ti':, 1';131 268.9- _:_.11 20~1.8S -~44.9~ 733.al 1992 ~48.79 '87.ElS ~102.72 ~390.357 71;;:.14 Cumu- lative b~j9S.8S 't')l-.21 77~,S.93 8~49.14 1749:.47 F<~D S ourc es: See ~cext . — 312 —

FIGURE 1 Annual R&D Ir.vestment, 1945 -1982 ( 1972 dollars, in millions ~ ~ soot Al\ ',-1 2,500- 2,000- 1, 500 - 1, 000- 500 ~ 300 200 ll ~J \~\ To ta 1 Rag Industry- : ~ ,_ financed RSD; ., if .'' """is' :"' A "a ~ / '~1' I'm i_ .. Federa l Civi 1 Aeronautics RID ( incl . KASA) ~ . . . ~ . 1945 1950 195; Sources: See ~cex~c. 1960 1965 1970 1975 19a' · 313 -

Industry- financed research expenditures display an oscillating pa~ctern of growth and decline during this period, in some contrast to the pattern of privately financed R&D in U. S . manufacturing overall. The successive waves of investment in the development of three generations of a~rfr~es - and engines during the postwar period are revealed clearly: R&~) expenditures grew rapidly during the early 1950's, the period of developmen~c of the first commercial jet aircraft; during the late 1960' s, as the first wide-body transports and high-bypass engines were developed; and during the late 1970' s, with the development of the most recent generation of smaller aircraft equipped with downs iced high-bypass engines . The to tal R&D inve s Mend in aircraft from all s ourc ~ s during 1945-1982 amounts to nearly $104 billion in 1972 dollars. Of this total, almost 75 percent, $77 billion, was provided by militar-y sources. Industry-f~nanced R&D during the period amounted to $11.4 billion, roughly 15 percent of the total. Federal nonmilitary research funding was a small portion of the total investment, totaling some $9 billion. Clearly, this enormous public investment was not directed solely toward the support of technological innovation in commercial aircraft; national security considerations dominated the vast majority of the expenditures. Nonetheless, the federal investment in military aircraft technology has had a significant impact on the course af innovation in commercial aircraft. ME ROLE Or NACA AND NASA The commercial aircraft industry is virtually unique among manufacturing industries in that a federal research organization, the National Advisory Committee on Aeronautics (and, subsequently, the National Aeronautics and Space Administration), has for many years conducted and funded research on airframe and propulsion technologies. Both NACA and NASA have been cited by scholars as models for publicly supported, precommerc~ai6r,~earch cooperation between industry and government (see Nelson ~ ), and NASA's aeronautics research program has received the endorsement of the Reagan A ~inistration's Office of Science and Technology Policy (oSTP).3 This section discusses the record and characteristics of NACA and NASA research funding and activities. World War ~ resulted in the establishment of a number of organizations intended to bring together leading academic, business, and government figures in an effort to analyze Important problems of national security in the areas of industrial mobilization, research, and technology. The National Research Council (NRC) was one such body; NACA was another, more firmly under federal control than the NRC. Established in i915, NACA was intended to "investigate the scientific problems involved in flight and to give advice to the military air services and other aviation services of the

government . " 3 9 During its early years, COCA did no ~ confine i sacs research solely to military aircraft, but worked on problems of aerodynamics and aeronautics con on to both military and commercial sectors . using expert mental facilities at Langley Field, Virginia, and, after 1940, at .Moffe~ct Field, California, and Cleveland, Ohio, NACA was an important source of performance and other test data in aeronautics . The committee p ioneered in the construction and use of large wind tunnels, completing one in 1927 that could accommodate full - scale airframes . This and other facilities provided a s teady sacred of test results that ted deco maj or improvements in airframe design. The famous "NACA cowl" for radial air-cooled engines reduced wind resistance and cut airframe drag by nearly 75 percent. Also, NACA research demonstrated ache superior performance of airframes with retractable landing gear and led Deco import he modifications in the positioning of engines in aircraft wings .4 Total appropriations for NACA from 1915 to 1940 amounted to Al mlilion in 1972 dollars, less than one third of NASA's annul appropriation for aeronautics research in the late 1970' s . Before World War II, NACA operated primarily as a Incest center, providing exceilen~c facilities deco both civilian and military users. S. ign~fican~cly, and in co4ir.ast to the pos twar period, the prewar SACK owned few test aircraft. Reflecting; its limited budget and staff, however, NACA carried out very little research during this period that could be described as "basic. ~ Indeed, one account of the development of the j et engine characterized the United States prior to 1940 as ~ backwater of research in theoretical aerodynamics, attributing the failure of American engineers to apprecia~ce ache possibilities of j e~c-powered aircraf~c to :heir ignorance of aeronautical des ign theory ~ see Constant4 ~ . As World War II approached, NACA focused increasingly on military aircraft design, Deco the partial exclusion of civilian aircraft research. After World tier IT, during which NACA work was exclusively military in character, the structure of ache aeronautics research system in the United States changed considerably. The maj or aircraft producers had Required substantial in-house facilities of their own during the war ; NACA' s research infrastructure was now less critical. Military support of industry R~ also became Crassly more important after World War II. Despite its considerably larger annual budget, which by 1944 exceeded the cumula~ci~re tomcat of appropriations during 1915-1940, NACA declined in importance. The agency remained an important sponsor of fundamental academic research, however, and continued to conduct empirical research on ~ scale that was now dwarfed by military- supported research. During the early postwar period, NACA expanded its research activities in rocketry. Then, in 1953, in ache wake of the Soviet launch of Sputnik, NACA was absorbed by ache National Aeronautics arid Space Administration. As NASA undertook a massive expansion of space exploration activities, its aeronautics research declined in - 315 —

importance . in i966, ~ Senate study noted, n Space budget demands have probate ly hampered what night have been expected to be ~ normal growth of ache level of effort in aeronautics within the agency. . . n44 In the aftermath of ache Apollo program, as NASA's operating budget increasingly was hostage to the fortunes of the space shuttle, budgetary pressures on aeronautics research programs mounted. Appropriations for aeronautics research continued to grow during the 1970's, but at ~ much lower race than had been true of the 1960' s . Moreover, the modest growth in real spending for NASA aeronautics programs during the 197O' s and 1980' s masks an apparent decline in the RED component of NASA's aeronautics research program (see National Research Council 5) . . Growth in NASA's overall aeronautics RED expenditures thus includes increases in the costs of staff, management, and construction, at the expense of research programs . Despite its reduced importance in the aftermath of World War II, NACA, and later, NASA, played an important strategic role in supporting research and managing a major component of the research infrastructure of the commercial aircraft industry. In addition, NASA proj ects frequently have involved two or more ers awhile competitor firms, encouraging, to ~ modest extent, the pooling of research efforts and results within the industry. Further, NACA sponsored a liberal system of cross- licensing of pa~cen~cs, disbanded in 157S due deco the obj ect~ons of the Antitrust Division of the Justice Department. That system aided In ache development of a- wide! shared ee:tnology base among firms in ache U.S. commercial aircraft incus try . Both NASA asked tJACA did more than simply support research yielding results that were diffused widely within the industry; they underwrote a portion of ache costs of the research infrastructure associated with innovation In airframes and engines. The ne~c social loss associated with private firms' pursuit of duplicative, parallel R&D programs has been noted by Hirshieifer 7 and Nelson,4 among others. Industry-wide research facilities of the sort maintained by HASA and NACA have considerable potential deco reduce such duplication. Of course, NASA test facilities are by no means the only ones available in ache U. S . commercial aircraft industry. Individual firms maintain extensive testing and design facilities, to preserve proprietary data and results. Nonetheless, NASA facili~cies complemen~c the privately funded R&D infrastructure and reduce ache total costs of R&D to the industry. Estimates prepared by a subcommittee of NASA's Advisory Committee on Aeronautics suggested that if the NASA aeronautics research program were terminated and private commercial aircraft and engine firms individually supported only one half of the NASA research programs of relevance to their segment of the aerospace industry during the 1982-1991 decade, while collabora~cing on 18 percent of these programs, the ne~c addi~cional costs of maintains ng paratIel arid duplicati~?:gresearch programs would amount to nearly $l billion in 1972 dollars. - 3 t6 —

MI LI MARY - S PONS ORE) RES ~~RCEI A final source of ex Vernal support for commercial aircraft innovation is the research and procurement of the U. S . armed forces. As ache data ~ n Table 2 indicate, military sources have provided the vast maj ority of the considerable research in~restmen~c in the aircraft industry during the postwar period. With ache possible exception of JR&D, this research investment was not intended ~co support innovation in any but mili~cary airframe and propulsion technologies. It has, nonetheless, yielded indirect, but very important, technological spillovers to ache commercial aircraft industry, notably in aircraft engines . From the Prat ~ and Wh.i~cney Wasp of 19 25 to the high-b~rpas s turbofans of the 1980' s, commercial aircraft engine development has beneficed from, and frequently has followed, the demands of military procurement and mili~cary-supported research. The development of ache first U.S. jet engine was financed entirely by the military. More recently, military- supported research on turbofan engines for the C- 5A transport fed to the developmen~c of the high-bypass -rancid engines that power the latest generation of commercial ~cransports, including the Air:Dus Industrie A300 and A310, as well as the Boeing 737 - 300, 747, 757, and 767 . Mil~tary-ci~r'lian technological spillovers of this type have been most important in aircraft propulsion technologies. However, the development of commercial aircraft has beneficed also from military- sponsored research and procurement in airframes . The importance o f technological spillo~rers in airframe des ign and development has fluctuated over time, since these advances are most pronounced in military transports and tong- range bombers . Thus, periods during which military procurement or development programs were active in these areas have wit nessed significant mili~cary-civilian spillovers . In the aftermath of World War II, ache development of jet-powered s bate gic bombers and tankers allowed airframe makers to apply knowledge gained in military proj ects to commercial aircraft design, tooling, and production. The Boeing 707, for example, was based closely on the design of a tanker, ache KC-135, developed by Boeing deco provide in- flight refueling for the strategic bombers ~ ache B-47 and B-52) developed previously by ache firm. A major share of ache development costs for ache 707 was borne by the KC-135, as a comparison of these costs with those for the 1)C- 8 reveals: Douglas lost SIO9 million in the two years 1959 and 1960, having written off S298 million for development comics and production losses up to ache end of 1960. Boeing did not suffer so badly. They wrote off $165 million on the 707 by then; some of the development cost may have been carried by the tanker program, which give provided a few of the tools on which the airliner was built . - 317 —

Increasing divergences be Preen civilian and military aircraft technologies, as well as the absence of maj or defense procurement and development programs in large transports since the late 1960' s, have reduced the amount and s ignificance of military- clvil~an technological spillover. Obviously, these two factors are related--military and civilian materials and engine requirements have diverged, for example, in part because fighter aircraft do not require long- term durability or operating efficiency to ache same extent as do both civilian and military transports. The magnitude of the shift in the relationship between military and civilian technologies is striking. In many cases,- technologies now flow from civilian to military appl ications. Whereas the Boeing 707 was a derivative of a military tanker design, the primary military tanker now being purchased ~ the KC- 10) is a derivative of the DC- i0, an aircraft designed originally as-a commercial transport. Similarly, as ~ recent National Academy of Engineering (NAE) study of the U. S. commercial aircraft industry noted, "Commercial engines gain service experience 10 to 15 times faster than miti~cary engines, even military transport engines .... For example, some of the improvements in the CF6 turbofan engine (derived from the TE39 used in DOI)'s large CSA cargo airplane ), developed during commercial servile, are being incorporated in later versions of the IF39. n Spillovers from military deco civilian applications resnain significant in Ache areas of propulsion, avionics, and flight control systems, but their importance has declined. In addition, according to the NAE panes , " for ache last IS years D(3D has tended deco define its interests more narrowly, to fund less generic research, and to insist on a specific demonstrable relevance.to present or proposed weapons systems for all DOD-sponsored R&D. For all but advanced supersonic aircraft or highly specialized mission requirement, OOD is largely prepared to buy off-the-shelf engine technology. n Therefore, both aircraft and engine producers must rely more heavily on industry-financed R&D. Assumption of a greater share of this financial burden by the private firms in the industry has increased the financial risks of developing new aircraft considerably. INDUSTRY- FINANCEI) R&l:) The commercial aircraft industries Red) contribution has been a strikingly small share of total R&D in the industry throughout the postwar period, despite the rapid growth of industry-funded research. Industry- financed R&D investment throughout 1945-1982 has never accounted for more than 23 percent of total R&D spending, and it stood below 20 percent of the coed for most of the period. Also, this estimated share also does not appear to be sensitive to ache shift from the BOON -Allen time series deco the adj ushered NSF data for industry- financed R&D after 1969; the low points in the indusesy- funded R&D time series during the 1960's are no lower than those of the 197C's. 318 -

Industry expenditures grew subs antially as a share of nonmi? itary research expenditures during the early postwar period, reflecting the growth of large in-house research establishments and soaring development costs for commercial aircraft. From 42 percent of nonde fens e R&D spending in 1946, the industry share rose to nearly 64 percent by 1969. During ache 197Q's, however, the NASA research budget grew substantially relative to that of industry. THE COMPOSITION OF PIES The Booz -Allen study contains da-a on the compos Scion of RED expend) tures in the aircraft industry ~ including both military and commercial aircraf~c) from both public and priorate sources for the t945-1969 pert ad (see Table 3) . While these data are Limited ire their coverage of ache years 1945 through 1982, the relative shares are fairly stable throughout the 1945-1969 period and, therefore, are ~ ikely deco describe industry R&D investment patterns accurately during the 1970' s and 1980 ~ s as well. One of ache most striking findings is ache small portion of total industry R&D (both privately and publicly funded) that goes to basic research. That category accounted for less than 10 percent of total R&D expenditures from 1945 through 1969 . Moreover, the share of teas ic research accounted for by industry funds also is less than 10 percent- - in other words, incus try- financed basic research accounted for less than ~ percent of total aircraft R&D during Chat period. Public sources, primarily the Air Force, Navy, and NASA (in the 1960's) supported most of fine basic research that was performed. Applied research accounts for a greater share of total research investment; ache portion of this category accounted for by industry funds in i969 was 34 percent, exceeding substan~cially the industry share of basic research. By far ache largest share of Local R&D investment is development expenditures throughout 1945 - i969; they never fall below 60 percent of the total. Mili~cary funds account for the largest share of development expenses. lathe Air Force atone supported more than 50 percent of ail development expenses during 1953 - 1966, and development expenditures accounted for more then 70 percent of total Air Force research support during the entire l945- 1969 period. Industry funds never were more than 15 percent of total de~relopmen~c expenses. The shares of total R&D investment accounted for by various components of the aircraft are given also in ache Booz-Allen study ~ see Table 4) . Although airframes account for the the largest s ingle share of total R&D investment from all sources, 40-45 percent, avi onics are responsible for a larger share of the total in~resment than are engines throughout the postwar period. Surpris ingly, in view of the rapid growth in ache sophistication and cost of avionics during this period, the shares of total R&D accounted for by avionics were quince stable. As the discussion above indicates, the maj or - 319 —

~ as, ~ Don `] Q. ~ To TV .? In o 5 - - C, ~5 I" C' - ~: - He X o o - - 0- C o Cal Hi 45: ~0 6 , ~ .: _ ~ ~ _ ~ ~ ~ . - _ ~ `1 ,1 ~ , ^ t -, ^~ ~ ~ 3 ~ -1 — ~ 3 — ~ _ -4 ~ ;^, _ _ a: _ ~ `~' ,., ~ ,: _ ~ ~ ~ r] ~ ~, M) M, ~ ~ ~ t~ ~ ~ ~ ~ s] ~ 41 ''' !"' ~ .` ~ ~ ~ ~ ~ ~ _ _ _ _ _ _ _ d _ _ _ — _ _ _ _ _ · ~ . ~n . ~ ~ ~ - - ~ ~ ~ o~ : O ce] ;4 :4 c] rl ~ ~ ~ ~ U] ~—~ ~ ·- ~ ~ ~ ~ ~ ~ ~' L`_ _ — _ ~ r, ~ -` ~ ~,, .7 _ ,4 ~'' - , ~~a~. ~t't r . ,, ~o~. 3 .- - '~ C. ^1 ~ n -. _ ~ ~ ~ ~ -. ~ _ ~ ,~ -. ,,' ~ ^' r, -' ~ ~ ~ ~ . ~ '', ~n ;> — :o ~ `1 ~ ~1 ~ ~ ~ ` ~ ~> ~ ', `~ :4 ' ~ ~ ~ ~ ~ ~ '_' ~ ._' ~ — — ~ ~ ~ !~ '' ' ~4 ~' i" ~ ^4 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ,~ ~ ~ ~ ~ ~ ~ ~ t~ ~ ~ ~ ~ ~ ~ ~ ~ ~' ~ ~ ~ ^' ~ ~ ~ ~ ~ . _ r~ c~ r, `1 r~ ~, ~' ~ _ ~ _ ~ _ ~ rJ ~ ~ ~ .` J1 ~ 41 - 3 _ c. r, r' r~ ?~ ,` ~ ,~ r! :. .~ c~ .~ :. ~ ~S ~n t. ~° .,, ~ ~` - ~ ~O :d ~ (] 3 ~ ~ ~ —01 ~ ~ ~ ~ — ~ ~ — 3 - — — ~ 3 ~ ~ ~ ~ ~ J7 — ~ ~ ~ _ ~ ~ ~ ~ ~ ~ ~ '~, J1 — J1 ~ -~ .` _ _ r~ ~r `1 ~n `: `: .~ ~ ~1 ~ ~ ~ at J: ~ ~ ,` ~ ~ o?~' 4 o~,,~.~ r~ ~ 3 ~ ~ ~ _ ~ ~ ~ O, ~ ~ ~ ~ ~' ~ ~ n ~ ~ -` _ _ _ ~ _ _ _ _ — '. Cd ~) ~. ~ ~' ~ · ~ ~ ~o ~ - _ - - - ,~ ~ ~ ~ 3 ~ ~t ?- 3 ~ ~ ~ ~ _ In 33 ~ ~ — 3 — .` ~ ~ '_' ~ ~ ~ 2 ! ~ -: _ ~ ~ — ~ — _ _ _ _ _ _ ~ J _ — _ _ (4 :4 ~d (~ ~ ~ ~ J — _ ^ ~ ~ ~1 et ~ ~ ~ ~ ~ t~ ~ ~ ~ ~ ~ ~~ ~ ~ — ~ ~ ~? -~ ~ ,,,~ ~,~ ;~ c~ t~ ~ ~ 51 ~ 33 ~~? ;~1 ~ ~ ;1 ;: ~1 -~ — ~ ~ ~ ~ ~ a'. —c' ,~, ~ ~ r' t~, -', r) t~ r, ~ r~ -, r ~,1 ;~~~<'0 : e - ~. . ~.~.e',. <,, ~. 4 ;e,,~ t.1~ ~ ~ ~ ~ ~ 3 ~n ~ r' ss `1 5~ ~ ,` o~ c' ~ ~o ~ - ~ -, c~ t' r~ Cl ~ r, .1 3 _ _ _ _ _ - _ ~ ~ ~ r~ r~ c~ er _ _ _ _ _ _ _ = _ _ ~ ~ ~ ~ ~ ~ _ ~ ~ ~ a: ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ _ _ _ _ _ _ _ ed ;~ U~ Qb .~: ~~ <~ C~ g] ~ ~ u~ _ ~ ~ ~r ~ —~ ~ k? o1 ~ —~ ~ 3 _ _ _ _ _ _ rd ~r ~ ~ ~: ~ 43 r~ r~ ~ -~] ~] 33 ~ ~~ ~ P~ ~` f ed ~ ~ ~ ~ ~ ~ ~ i] ~ ~ ~ ~ r] - .: . - ~ ~ ~ - - ~ ~ _ ci ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ M' ~ ~ 41 ~ 61 - ~e ~ ~ ~ ~ ~ ~ - r' ~ ~ M] ~ ~ ~ ~ .' - ci ~r ~ ~ ~1 ~? ~ ~1 J, V, J1 `~1 ~ ~1 ~ ~ ~ ~ ~ ~ ~ 3 ~ ~O Q.> ~ ~ ~ ~.t~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 0 - 320 - - a: ~ _ _ '" — _ c: ~ _ _ C~ "J '_ = C _ ~ ~ O- '~ ;O _ C: ~: ~ _ _ ~ : q; _ C} . - ~ ~ ~ - c~ ~— ~— ~ t - - - _ ~ ~ s u, c) ~ ~ ? _ cn C ~ c~ ~ · - O == Ll ~ _ _ _ 0 ~ ~5 ~5 ~ _ ._' ~ ~ 3 ¢ _ ~ ~ ~ ~S O ~ ~ ~: · ~ c; 5- ,=

TABLE 4 Aeronautical it&i) Funds Used by Industry, Classified by Aircraf~ Component ~ annual expend) cures in millions of dollars F1 seal Year ~: rt'-ame Enc: ne Revlon: c 5 Tc" a_ ~945; 113 on 79 _~ Id- as lop aid cr47 1 Ma 7~ 9 1 A;' 1 94iB 1 4a ~ ~ He `. 1 949 1 634 1. C~~ 1 _~~ 41.1~ 19~:, AS_ 1 17 141 4,t, 19~1 _~ '84 ~1 7;z 1 9 _ ~ 'Jo ~;~5 —SO ~ ~ _ ci5 ~ 7' 6 _ 97 477 ~ -he., 1 9= ~ 7S9 4__ "~- 1 Ado 195S 715 -9' 476 1'S8 ~ 95~ 749 4 1 ~ 49~ 1 b~- 19J7 81t 4C_ -4 ~ tS1 1 BAJA ~ 4 4~ See 1~_ 1 9ti; 79= 44 1 "; C) 1 ~ =a 194:~) 7 . ~ -9= 47_ 1', 19~: 7_)sj A; 48~ ' 5~_ 902 7~9 4`j5 49" 10 '; cra 6:s. 47 ~ Sea 1 3~ ~ t9~4 at 4ba 5a lo $~; 845 469 -6, ~ ~ / 7 9~a 98- 54~ 552: ~ ~ 8 9~7 1 c)~;6 -;57 7~) ~ I, t9~8 1i~9e Al tj - 7 ~ A. -4 1 9~9 ~ ;?~5 57ti Cam Age Source: Booz, Allen and Hamilton Applied Research, Inc. A Historical Study of the Benefits Derived from Application of Technical Advar.ces to Commercial Aviation. Washington, DC: U. S . Government Printing Office, 19?i Table C- 21 . - 321 -

share Of the aircraft industry' s R&D in~restmen~c goes toward the integration of the various components of an aircraft, represented , n the testing and debugging of prototypes, rather than their separate development. To some extent, this characteristic of aircraft R&D inves tment also reflects the fact that the costs of developing many of the maj or components and systems in an aircraft are borne by other indus~cries; while those costs are reflected in the cost of components, they are excluded from aircraft industry R&D investment data . R&D INl7ESI~ENT: CONCLUS ION During the postwar period, the commercial aircraft industry has benefited frown substantial direct (NASA and NACA) and indirect (military research and JR&D) federal financial support for research. The size of "he federal R&D investment, as wel~i as the existence of a dedicated civilian technology development program, renders the aircraft industry unique among U. ~ . manufacturing industries . Both NACA and NASA played important roles as centers for generic research, reducing total industry R&D costs through operation and construction of costly testing and research installations. Moreover, both civilian and military research programs encouraged the wide diffusion of techno~ og' cat knowledge within the aircraft industry, supporting the development of a -readily access ible industry knowledge base . In -his way, the federal programs operated in a fashion that resembles c] osely the cooperative R&~) pro grams in the Japanese economy. S4 None~cheless, federal R&D priorities in the aircraft industry resemb le ache broader pattern of federal research funding throughout the U.S. economy, insofar as they were motivated throughout: the 1945-i982 period primarily by national security requirements. The civilian share of the Coccal federal R&D investment ~ largely NASA funded) is, after all, only slightly more than 10 percent. Technological development in this Undue try, as in others, has relied heavily on technological spillovers from military to civilian app lications . However, such spillovers have grown less s ignifican~c in recent years, and their very direction seems deco have been reversed in a number of aircraft: technologies. THE DEMAND FOR INNOVATION: THE ROLE OF GO1J-~RNME~ REGUIATION While the federal government played an important role in supporting air.c-af~c indus~cry research, thereby expanding the supply of relevant ~cechnology, federal policies also affected the speed with which that technology was embodied in commercial aircraft innovations. Civil Aeronautics Board (CAB) regulation during 1938-1978 supported rapid adoption of such innovations. Indeed, U. S. go~rerrment policy toward — 322 -

the Ados cry throughout Ale pus twar period is unusual in i as impact on both the supply of technical knowledge and the demand for application of that knowledge in innovation In the face of recent suggestions that federal science and technology policies devote greater attention to the diffus ion and utilization of research ~ see David:: and ^Mowery56), the operation and costs of this "diffusion support" policy in the commercial aircraft industry are worth examlulng . Congressional dissatisfaction with passenger safety and regulatory policy led to Ache establishment of ache CAB in 1938. 57 Through its issuance of operating cer~cifica~ces and oversight of airline fares, the board effectively controlled airline pricing policies, as well as entry into or exit from the air transportation industry, from 1938 to 1978. Its powers were used throughout that period to prevent entry into scheduled trunks ine air transportation and to prevent price competition. That regulatory environment motivated a high level of competition in service quality-. One result of that competition was rapid adoption of new aircraft des igns by ache ma; or carriers, in Ache be l ie f that introduction of s ~a~ce-of- the-art aircraft was an effective marke~cin~strategy where price competition was prohibited. Jordan's study compared California's intrastate air carriers (not regulated by the CAB, and subj ect to price compet~&cion and easier entry) with the interstate carriers in ~ he speed of adopt) on of cabin pressur~za--on and j et aircraft: The trunk carriers were consistently the first to introduce each innovation. In fact, they introduced all but two of tne over 40 aircraft tykes operated by all three carrier groups between 1946 and 1955 . In addition, they adopted these innovations rapids y and extens ively . The local carriers, on the other hand, were slow to introduce the two innovations and their rates of adoption were tow. lie drive Deco be first witch a new design was one of the central motives for the willingness of maj or airlines deco malce early purchase commitments deco aircraft manufacturers as ~ means of obtaining the earliest posse ble delivery. lathe importance of an early position in the delivery queue also conferred considerable ~ average upon producers in extracting advance orders from ache airlines, effectively defraying a portion of the costs of developing a new aircraft with advance customer payments. Thus, CAB regulation encouraged a rapid pace of innovation and adoption in the commercial aircraft and air cransportation incus tries . That rapid race of innovation, however, and the assoc! ated impressive productivity growth in U.S. domestic air transpor~cation came at cons iderable cost . Despite their protection from entry during the period of regulation, the airlines earned, on average, only a normal rate of return on capital; as Keeler: noted, disc - 323 -

could appear that with fares set at high, cartel levels, the airlines have competed away profits through excess capacity. " Service-quali~y cors~pet~tic; caused costs to rise to the level of fares, as Douglas and Miller O noted, and largely pre~rer~ted airlines from exiling their protested position for priorate profit. Consumer welfare also was impaired by the lack of variety in service quality and price. Government regulation restricted the range within which consumers were free to trade off Mace against juali~cy. The U. S. Genergi Accounting Office (GAO), employing a model based on Keeler, concluded that an efficien~c, deregulated air ~cransportation system would have cost consumers $1.4 to $1. 8 billion (in nominal dollars) less during each year between 1969 and 1974. Moreover, in view <'f the high probability that tower fares would lead to more traffic, the GAO calculation almost certainly unders~cated the total annual costs. In other words, -a cos ~ equal to nearly trio thirds of the tomcat annual R&D investment flow in this industry ~ in 1970, slightly more than $3 billion in nominal dollars ~ was borne by the Craning public as a cost for the rapid diffusion of commercial aircraft innovations. The direction of innovation was also affected by this regimen of regulation and service-quality competition. The growth of the U. S . market for commuter aircraft was stunted severely by -CAS policies. The modest demand for aircraft with 60 or fewer seats refiec~ced the fact that ache route structures developed by the maj or carriers emphasized tong-haul, "point- to-point" service, without a mat or role for smaller feeder carriers. Indeed, for much of the. regulated period, trunk carriers subs ionized their short-h:~t routes from profits ear fled it' long-haul service ~ see Keeler ), res tricking further the possibilities for entry into short-haul service. As a resut A, throughout the postwar period, the development of short -haul aircraft with more than 19 seats was confined largely deco Europe and Canada, where Fokicer, Aerospatiale, Shorts Brothers, British Aerospace, DeHavilland, and other firms developed commuter aircraft. Rapid growth in ache U. S . commuter market since t978 has benefited these and other foreign producers priority. [Deregulation of domestic air transport has affected Ache market for aircraft and engines in several ways. Service-quali~cy competition has declined greatly in importance. As a result, major U. S . airlines are less eager now to adopt new aircraft without significant improvements in seat-mile operating costs. The early years of the deregulated era also were characterized by upheaval in the U. S . airline industry as low-cost entrants ~ eopardized the position of established carriers. The initial Shakeouts phase appears to be ending, and numerous carriers now are enj oying healthy profits . Nonetheless, ache transition was characterized by considerable uncertainty and canceled orders for aircraft. An enduring characteristic of the deregulated era is likely to be a much greater reluctance of maj or airlines to commit funds (in the form of advance orders, etc. ~ for major new aircraft development. Aircraft producers now are less able to share — 324 _

the risks of new product development with maj or cus Comers . W-hile an explicit link is not apparent, the exit of Lockheed from commercial aircraft manuEac Cure, as well as the cautious policies of McDonnell Douglas in new product development, are related closely to the new realities of a deregulated U. S . air transportation industry. AN EVALUATION OF THE TECHNOLOGY POLI CY STRUCTURE IN THE COMMERCIAL AIRCRAFT INCUS TRY This section assesses the economic costs and benefits Deco commercial aircraft of the federal policy structure, using a frame'~oric that draws on Fogel64 and Mansfield et al. 63 in an illust~~ ore calculation. Following the quantitative assessment of the impact the "technology push's and "demand pull" combination of policies toward aircraft innovation is ~ qualitative discuss ion of the possible implications of this policy structure for industry-wide research programs in other sectors. THE COSTS AND BENEFITS OF CAB ANI) NASA POLICIES, 1966-1993 The structure of policy toward research and innovation in ache 1~'. S. commercial aircraft industry has affected both the supply of technical developments and the incentives for their embodiment in new cornn~ercial aircraft. In evaluating this policy structure, therefore, it is important to incorporate the costs of both the " supply-push" and "demand-pulin components. Measuring the economic benefits of technical change in commerc ial aircraft is exceedingly cliff icily and is hampered by ache absence of reliable time series on operating costs and performance characteristics. ° Thus, the estimates presented below should be viewed as provisional. One calculation ~ discussed earlier) of ache economic benefits from technical change in commercial aircraf~c was the estimated social savings due to declines in operating costs between 1945 and 1983, employing the DC- 3 as the representative 1945 aircraft . That analys is suggested that subs tantial social benefi ts have flowed from innova~cion in corms ercial aircraft. The c~ulative investment of S26 billion from industry and federal civilian R&~) programs, according Deco this measure, has yielded an annual flow of social savings, as of 1983, of $13 billion in 1972 dollars. Lens et al. 67 treat this one-year flow of savings as a net return on the public and private it&l) investment in the commercial aircraft industry, concluding that the return is quite high. While interes ting, the ir analys is overlooks several dimens ions of R&D investment in the industry. Presumably, the annual flow of savings represented in the calculation is a benefit that will persist indefinitely. As such, the annual savings flow should be capitalized - 325 -

for comparison with the original investment. In addition, the computation by Lend et al. of the magnitude of the cumulative public and private Rat) investment as a simple son of annual inves=men~c f: ows overlooks the time value of money. Lenz et al. also ignore the casts of CAB regulation in computing the social returns to the R&D investment in commercial aircraft O If CAB regulation played a central role in the rapid adoption of new commercial aircraft, the costs of such regulation surely should be included in any evaluation. The analysis of returns to the public and private investment in commercial aircraft R&D is presented here to illustrate relative orders of magnitude. The basic analysis is similar to that used by Mansfield et al.6 tO compute social and private rates of return to innovation. Rather than computing the social savings resulting from improvements in commercial aircraft performance at two widely separated times, the annual CAB surveys of operating costs and fleet mix were used to compute the annual savings during 1966-1983 resulting from the employment of curren~c-year equipment, rather than ache fleet that existed in 1965. Those benefits were compared with the annual stream of public R&D investment from NASA and FAA, as well as that financed by industry, contained in Table 2. The inclusion of ache costs of CAB regulation ~ n this analysis is based on the above arguments concerning ache importance of such regulation for the diffusion of aircraft innovations. The costs of regulation were computed from the GAO study, which estimated that 1972 costs to consumers resulting from regulaticn amounted to I. 8 billion. Those costs were assumed to be proportional deco annual domestic passenger traffic ~ ignoring changes ire the mix of long- and short-haul traffic ), and fatal annual domestic revenue passenger miles fit own during 1966-1983 were used deco translate the 1972 benchmark estimate of the cos Is of CAB regulation into an annual estimate for each year during the period of regulation ~1966 -1978 ~ . These various components of the annual stream of. casts and benefits during 1966-1983 were used deco compute an internal rat. of re~curn. As elsewhere in this paper, all costs are reported in 1972 dollars. Following Mansfield et al.,70 the stream of annual cost savings was extended through i993, reflecting the fact that the social savings are assumed to be realizable permanently. In addition, the current aircraft fleet mix is likely to be used through the early L990' s, approximately. The treatment of military R&D expenditures also poses some lifficult issues . Clearly, these huge expenditures have yielded substantial benefits to the commercial aircraft industry. Equity [early, fiche existence of spiders has rarely, if ever, governed :he policy of the armed forces toward their R&D programs. In tried of he contrasting motives for federal support of military and civil eronautics R&D, it seems inappropria~ca to incorporate the entirety f the mili scary R&D inves tment in this analys is . Obvious ty, Even the size of the annual military R&D investment, this decision - 326 —

-~as mat or im~Tica~ions for the ultimate computation of the internal race of return to che federal R&D i nvest:ment. An additional issue of methodology concerns the rate at which the results of R&D carried out in public and priorate research facilities are embodied in the actual operating costs of the U. S. commercial aircraft fleet; that is, how rapidly are ache productivity- enhancing and cost - reducing results of federal R&D realized in air transportation? While the "embodiment ~ age may have been lower in this industry than in others during much of ache postwar period, thanks in part to CAB regulation, it can affect the measureme72 of benefits from R&D investment greatly. Fortunately, Terlecicyj has computed estimates of that lag, which he fixes at six to eight years . In the computations ~epor ~~d be low, the embodiment lag was assumed to be s even years . Table 5 presents estimates under various assumptions of the internal rate of return to public and private R&I) investment in commercial aircraft from 1966 to 1993. Not surprisingly, in view of the substantial public support of both Red and diffusion, the returns to industry- financed R&D investment, which do not incorporate the costs of CAB regulation or NASA RED, are substantially higher than the social returns. Under ache assumption that R&D is embodied ins tantaneous ly in the commercial aircraft fleet, ache internal race of rescue to private Red:) investment amounts to more than 3 3 percen If a seven-year lag in the realization of the economic benefits of privately financed R3~ is assumed, the returns increase dramatically, to roughly 60 percent. =" returns deco public and priorate research investment in ache industry are lower, but still substantial. Under the assumption Chat R&D is embodied immediately in the aircraft f? eel, ache internal rate of return to NASA, FAA, and privately financed R&D stands at 11. 5 percent, substantially above ache usual returns to other public investments such as water proj ects . Moreover, if the assumption of a seven-year embodiment lag is employed, this rat. of return increases substantially, to more than 24 percent. Thus, ache retune its to the public investment in the supply of civilian technologies in con~nerciai aircraft are quite high. If one assumes that 2S percent of the military R&D investment benefited commercial aircraft directs y and solely, however, the internal rate of return drops sharply, to sligh~cly more than l.7 percent. }Iowever, as has been stressed repeatedly ~ the perfo tmance characteristics of the current aircraft fleet hairs been influenced substan~cially by the lengthy period of CAB regulation. Cohen the costs of CAB regulation are included with those of ptl~rately and publicly financed R&D investments, the social rate of return becomes negative, regardless of the assumptions concerning embodiment 1= gs . If the estimate of CAB- induced costs is reduced to 50 percent of the GAO study' s findings, the estima~cedi integral rate of return is less than ~ percent. If the costs of CAB regulation are reduced deco 25 percent of the GAO estimate, the internal rate of return on ache funds — 3 2 / —

I.\BL~ 5 Races of Returrt on R&i3 Investmen~c, 1966-1953 Industry—~ ~ nanc:ed 2~/. . research i I n d~ str y—i ~ n an c ad r esear ch 6Ij% . Ind~stry—~ i nanced and c: wi 1 s an emt~cci men t 1 ag: 1 ~ ~ 5/~. D) Industry—~inanced and c:~: l san emt:¢d~ment 1ag: 4.4/ nvestment, ~nves~ment, 7-year f ede~al f;&O eder al f<~) Industry—~ ~ nanced and c: vi ~ s an ;ec~ral c c:st s, n o emt~ac: ~ men t 1 aq: -4 . '% . Ind~stry—~f i nanced . and c ~ v ~ 1 i an ~ ader a1 c ost s, 7-y ear en,~ c~d s men t 1 ~: -_` . a z.. . no e~ncod:.~ent 1 ag: emb cd ~ men t 1 aq: ~ nvestment, nc s nvestment, 7—year f<&D ~ nvestment anc -~E ~'.<D i nvestment and C~E G) 2ndustry-+i~a~ced R&1D, ctv: lien '`~de~al ~ 'D investment ~ and ~./. cf m: 1 ~ tary F<~.~D, no emeoc3tment 1 aq: 1. '%. H) Industry—~f ~ na~c~d ~'-~D and c: v: 1 ~ an ~f - -e-al -~% cf cAa cos~s, nc: emacd~n~ent 1ag: `:~. ;~. I ) Industry—~ ~ nanc~d Rt,:) and c: v: 1 i an f ede~al =% af CA~ costs, no embad~,nent 1 aq: b. ~f . Source: See text. R&O ~ ~estment and R.~O ~ nvestment and - 328 -

.~. Used in Rig) and d Effusion support is more than 6 percent Clears I, this estimated rate of return is quite sensitive to one' s assumptions ~oncer.~ing the moti~ra~cion and impact of CAB regulation of domestic air transportation. Thus, ache costs of the "diffusion support" policy embodied in CAB regulation are sufficiently large to offset entirely the pos irate rate of return to R&D investment in the industry. If nothing else, these findings suggest that the CAB' s positive impact on diffusion does not nearly offset its negative impact on consumer welfare. These illustrative computations employ a l' mited definition of benefits. Benefits resulting from technological improvements are restricted entirely deco operating cost reductions, rather than improvements ~ n aircraft safe tv or speed. The omission of improvements in aircraft speed would indeed understate benefits in a comparison of an all-piston engine fleet with one powered largely by j et engines . In fact, however, the average cruis ing speed of the U. S . domestic f' eel has increased only modestly during the ~ 966-1984 period, reflecting ache widespread adoption of j et aircraft by 1966 . Incon~ora~cing estimates of ache value of savings in craver time does not affect these results materially. More significant, perhaps, is ~ he failure of the estimates to cons ider ache larger benefits to the national economy of the federal policy structure' s support for bye U. S . commercial aircraft industry. Througnout the pos-~-~ar period, the United S cates has been by far ache largest single market for cogency al aircraft indeed, only recently has the U. S ~ market come to represent less than an absolute maj orgy of scored aircraft demand. While a low tariff was imposed on U.S. imports of foreign aircraft and parsecs for much of the postwar period, formal protection of the U. S . commercial aircraft market was modest. Nonetheless, the importance of close producer-purchaser contacts in the finance, design, and performance monitoring phases of new aircraft development conferred substantial advantages on U. S . producers in this marke~c. By supporting U. S . domestic demand for long-haul commercial aircraft, CAB regulation enabled U.S. producers to gain experience in the production and des ign of aircraft with a large foreign as well as domestic market. In view of the great importance of learning effects in the des ign and production of commercial aircraft, CAB support for domestic market demand may have operated analogously to an export subsidy, in the fashion outlined by Krugman. 74 Federal support for research, as we ll as for diffus ion and adoption of innovations, in the domestic commercial aircraft market thus may have enhanced the international Competitiveness of the industry and expanded its export arkets. In view of the fact that aircraf ~ make up the largest single category of manufacturing exports from the United States, this impact on competitiveness was extremely important. To assess ache significance of such benefits, however, one must compare the income and other benefits of; ob creation in the commercial aircraft Endue try with those resulting from Ache employment o f Ache highly skilled commercial aircraft industry labor pool in other sectors of the economy. It is unlikely - 329 -

that the human resources devoted to the production of commercial aircraft for . export and domestic use would have remained unemployed in the absence or federal policies supporting research and adoption in commercial aircraft. Moreover, while CAB regulat: on aided U. S producers of large commerce al transports, it simul~car~eously discouraged domestic demand and innovation in the commuter segment of the aircraft industry, contributing to the current failure of U. S . producers to enter that rap idly expanding world market . IMPLICATIONS FOR GENERIC RESEARCH POLICY. The impact of federal R&D investment on the postwar commercial aircraf ~ industry has been influenced heavily by the con: unction of that research support with a regulatory policy that facilitated the rapid diffusion of innovations. Perhaps the most significant implication of such ~ policy structure for other indus~cries, then, is that it a£fects both inno~ra~cion and diffus ion. Nonetheless, the apparent ef£icacy of CAB regulation in encouraging rapid adoption cannot hide the fact that the regulatory policies were an extremely inefficient mechanism for such diffusion support, for they imposed high costs on the traveling public. Eisewnere, ache author has sugges ted that the failure of o the' federal research programs to invest more heavily in the support of utilization and difrus ion of research resul es ~ an exception being the Department of Agriculture ' s research programs ~ reflects the limited relevance for policy of the theoretical framework employed by economists to analyze R&D investment ~ see Mowery 6) . Largely assigning away the difficulties of know] edge transfer and utilization, the conventional economic snalys is of R&D in~resment focuses on the imperfect nature of property rights in knowledge, especially in basic ~ esearch, and recommends a maj or government role in supporting such research acti~ri By. The undersupply, rather than the utilization, of technical knowledge is ache critical policy problem, in this view. How do NASA' s actual programs match such a concep~cnalization of the appropriate role for government research? Throughout the history of go~rerstment- supported research in civic aeronautics and propulsion, the ~client" industry has supported a large in-house R&D capability. lye internal expertise of firms in the industry facilica~ced their absorption and utilization of the. results of NASA-sponsored research. The absence of such in-house technical expertise in client fists in other industries has contributed to the fat lure of some publicly supported extramural resear~b~ Programs in the United S tates and Great Britain (see flowery ~ ). The direct participation of firms in numerous NASA- sponsored research proj acts enhanced the ease witch which the results of such research could be absorbed and utilized by a firm. Moreover, the liberal dissemination policies of NASA with respect to - 330 -

technical d=.a, the .Manufac'urers' Avi~c~on Association patent licensing scheme, and the liberal licensing requirements for patents obtained wi'h federal military research funding all enhanced the intraindustry diffusion of technical data and results. Indeed, one way in which federal policy has affected the innovation process in the commercial aircraft industry is the fact chat a s igrtificant portion of industry- financed R&D ~nvesment is intended deco support absorption and adaptation of research and technological developments funded by either NASA or the military services. The commercial aircraft industry is an excellent example of an industry in which high spillovers from one firm's knowledge base to other fines are associated with high levels of research investment, as a means of absorbing such spillovers (Cohen and Lsvinthal provide a formal analysis of this phenomenon). .Mareo~rer, the experience of the aircraft industry suggests that this aspec ~ of R&D investment behavior may be influenced by policy as much as by exogenous aspects of industry structure and technology or demand condi tenons . The NASA and NACA research programs departed from the policy prescriptions of the neoclassical model of research and. development in another fundamental way. Basic research, while important in both FACE and NASA aeronautics R&D, by no means constituted the sole or even the primary focus of the research programs. No~ced above was NACA's relative inattention to basic research acti~rit'es (by compare son with the aeronautics research estate' ishments of such nations as Germany and Great Britain) prior to World War II. During the postwar period as well, the research programs of MESA and NACA have extended well beyond basic research into research concerned with the demonstration of the feasibility of a specific combination of systems or materials . In ocher words, NASA/NACA research resembled the R&D carried out by private firms in the industry in its focus on systems integration and overall design questions. By virtue of its ability to pursue both basic and preco~ercial technological "proof of concepts research, the NASA R&D program was able to exploit ache links between basic research and ether stages of :he research process that contribute to successful innovation. )esp i te the frequent portrayal of this process as one that is essentially linear, in which technological developments draw on prior Basic research breakthroughs, the relationship among the stages in act is complex and interactive--in many cases, basic research questions and agenda are influenced powerfully by Surlier :=chno loge failures and findings ~ see Rosenberg and Kline and .osenberg for further discussion). In pursuing research ctivities hat extended beyond basic research, NACA and NASA were ble to accommodate the requirements of the commercial aircraft nnovat~on process and exploit ache interactive rela~cionships among he early stages of the innovation process. - 331 -

R&3 SI:PPORT WIl~.nOLT O7FFIJSION SUPPORT: AUTOMOBILE E~£ISSIONS CONTROL The structure of ache U. S . automobile incus ~~ resembles that of the commercial aircraft industry in ~ nuder of ways--both industries are highly concentrated oligopolies, with a high reliance on components and parts supp liars as sources of innovation . Indeed, many of the large suppliers of automo~ci~re components, such as Rockwell, TRW, and the Bendix Corporation, are also maj or manufacturers of components for commercial aircraft. Since the 1960's, moreover, ache federal government has exerted a maj or influence on innovation in the automotive industry. Federal and state regulatory requirements for emis s ions, fue 1 economy, and safety kave mushroomed in number and stringency. S imultaneously, federal spending for research and development within the industry has increased, accounting for perhaps $250 million within a total expenditure on R&D by the three maj or mar£ufactur::s and parts suppliers of more than $4 billion in 1919 (see White ). Federal policy in the automotive industry has combined support for technological development with regulatory incentives for adoption. Has this policy structure had a positive e effect on innovation that is comparable to that observed in the commercial aircraft industry? The apparently s isnilar policies have in fact yielded ve~ different results . Research cooperation between ache U. S . automotive industry and ache federal ga~rernment has been far less smooth and successful than for the commercial aircraft industry. Federal initiatives Deco increase financial support for automo~ci~re R&D, such as the Cooperative Automotive Research Program (CARP) of the Carter Administration, have met with cool industry reception; the lack of incus try support contributed to the demise of CARP under the Reagan Administration. The performance of the U. S . automotive industry in meecing ''technology- forcingn statutory targets fo~Spollu~cant reduced ons in particular has been poor. As White noted, the timetable for target achievement has been revised repeatedly, and progress in meeting the goals of the Clean Air Act has been shower than anticipated. A central problem of federal research policy in automotive emissions technology appears to be poorly designed diffusion policy. While ache federal government has provided financ tat support for research in emissions control technologies, the resistance of U. S. producers to the adop~cion of new and existing technologies has contributed deco repeated showdowns with federal and state policymakers oared ache ability of the inductor to meet emissions standards. The central flaw in the diffusion policy is its reliance on industry-wide standards. An al~ce~:ative system of "axes based on emissions levels ~ such as that advocated in White, ~ would affect the relative prices of low- and high-emissions automobiles, thereby influencing ache incentives of manufacturers to develop and incorporate new emiss ions control technologies in their products . - 332 -

Industr-; so underdo remove any incentive for interfirm technological -compet':ion in the adoption of new pollution control methods. This situation contrasts sharply with the commercial aircraft industry during the period of regulation, where producers facet clear incentives for the embodiment of innovations in new aircraft des ign. Since individual automotive firms have no incentive to adapt emissions con~cro: technologies, the payoff to collective industry resistance to emissions standards within this oligopot is tic industry is substantial. The absence of policies encouraging the adoption of emissions control technologies thus has delayed and complicated the implementation of the Clean Air Act, while contributing to a poisonous relationship between federal policymak~rs and industry officials that has undercut the success of other research collaboratior~. Clearly, the design of emissions regulations is not the sole factor behind the very different character of industry- government research cooperation in the automotive and commercial aircraft industries . Nor has the inno~ra~ci~re performance of ache U. S. automotive industry, ester tally during ache pas t decade, been uniformly dismal. Nonetheless, the tangled and conflicted history of federal automotive emissions con~cro~ regulation underlines forcefully the importance of linking federal technology support with appropriately designed policies deco support ache adoption and diffusion of the technologies developed witch public funds. COttCLUS ION The history of federal research ~n~restnent in the commercial aircraft industry suggests that public Rag) programs can exert a powerful and pos incite influence on the innovative performance of an incus try . The unique circumstances of the commercial aircraft industry must be Rep in mind clearly, however, before drawing conclus ions about the ease with which the policy framework associated with that industry can be transferred to other sectors. lrhe contribution of the aircraft industry to the military strength of the United S tares has meant that a substantial and ultimately unquantifiable influence on its innovative performance has operated through military procurement and R&D support. With the possible exception of microelectronics during the 19SO' s and 1960' s, few other industries in the United S~cates have benefited so greatly from military-civilian technological spillovers. Lee foundation of NACA and the continued support of aeronautics research in NASA reflected the perception that a strong civil aircraft technology base also yielded benefits for military aircraft. In acdi~cion, the regulatory regime of the Civil Aeronautics Board played a s ignificant role in the demand for innovations in commercial aircraft that in Burro supported rapid growth and an impress ive record of innovation. - 333 —

None' heless, the elements of the policy framework that characterized commercial aircraft innovation, particularly ache combinati on of support for R&D and adaption, have considerable potential for more general application. Indeed, a great deal of Japanese industrial policy can be explained in precisely these tee as . However, the particular policy instrument that operated support such diffusion, CAB regulation, was hardly an efficient cos tress on" . Nor can the policy frame be described as one was planned consc Pious ly taco support innovation, or as one whose distributional consequences (especially Chose of CAB regulation) even well understood prior to the 1960' a and 1970' s . Indeed, had distributional consequences been more Iris ible or more clearly understood, it is lively that deregulation might have occurred cons iderab ly scone- . to' or that were the Analysis of the actual roles played by ache CAB, NACA, and CASE in supporting innovation within the commercial aircraft industry also suggests the limitations of the neoclassical economic theory of R&D for policy analysis. This theoretical framework focuses primarily on the puta~cive undersupply of such research and bases its recommendations for pa licy on this market failure e However, for pa licy purposes, the distribution and utilization of the results of research and development are crucial . An exclus ive focus on the R&D support policies of the federal government, without some cognizance of the substantial diffusion support component of the policy structure, yields conclusions that differ substantially from those of an analysis that attempts ~ 0 incorporate both the technology supply and technology adoption incentives operating within ache overall policy framework. MorecYer, this neoclassical trier leads to a somewhat distorted view of the appropriate and actual role of public agency es in supporting research. Rather than focusing exclusively on basic research, NACA and NASA clearly went well into preco~ercial "proof of concepts work, the so-called "grey area" between basic research and applied development. The effectiveness of both the applied and the basic research agendas pursued by those agencies was enhanced considerably by the fact that the links between these phases of the innovati on process were not cut. The creation of an industry knowledge base and the tailoring of R&D support policies to the specific requirements of the industry's innovation process are considerably more impor~can~c than abstract debates over where basic research, for which federal support is Justified, ends, and where applied research, which theoretically is to be handled by the private sector, begins. The federal rol e in the development of the structure and ce~hnology of the commercial aircraft industry and of Ache air transportation industry means that recent federal policy changes will exert a maj or influence on the future developmen~c of both. The corn~nercial aircraft industry has become truly global in the panic decade, with extensive cooperation between U.S. and foreign firms in a number of product development and manufacturing ventures. The implications of such global alliances for the research support

policies of VASA and, indeed, for the strategies of individual fir as are not well understood. Should NASA, for example, further restrict international dis tribu~cion of technical resul ts and data? Is the result of cooperation a purely one-way transfer of technology, ultimately eroding U. S . competitiveness? That are the implications of in~cernationalization for ache U. S . firms supplying components and assemblies ? Should NASA contemplate a more ac rive role in supporting research activities within the supplier segment of ache industry? Deregulation raises similarly important issues, at least from the view of continued innovation within the industry. The transition Deco a deregulated air transportation system appears to have been substance ally completed. Nonetheless, the implications of this new environment for the innovative ~ and ultimately, therefore, the competitive) performance of the U. S. commercial aircraft industry have not been examined Anile regulation' s demise was an unambiguous victory for both consumer welfare and economic efficiency, the implications of deregulation for the adoption of innovations by the domestic air transportation industry and, therefore, for the inno~rati~re performance of the U. S . co~ercia~ aircraft industry are understood poorly. In the absence of either ~ maj or military research and procurement initiative in the area of transports or some. al~cernative support for the adoption of innovations in commercial aircraft, it seems likely ocher the pace of innovation in the industry may decline from its postwar leered. - 335 -

NOTES AND REFERENCES This section draws on discussions in D. C. Mowery and N. Rosenberg. "=e Commercial Aircraft Industry. ~ In Governmen~c and Tec.—ice ~ Progress: A Cross-Indus~cry Comparison. Edited by R. R. Nelson. New York: Pergamon Press, 1982; and 1) . C . Mowery and N. Rosenberg. Commercial Aircraft: Cooperation and Competition Between the U. S and Japan, n Cal if ornia Management Review, (~985~. ~ .2 . Al 1 f igures are from Aerospace Industries Association. Aerospace Faces and Figures 1984/1983. New York: McGraw- Hill, 1984. 3. A 1981 study of aeronautics R&D by the Office of Science and Technology Policy concluded that, ~ . . . the aeronautics industry is characterized by high research intensity and ~ wide technology base. That is, aeronautics depends on R&T (Research and Technology) performed within the aeronautics industry and on R&T performed by virtually every other high-~cechnology industry. See Aeronautical Research and Technology Folksy, Volumes I and II. Washington, DC: Office of Science and Technology Pot icy, 1982, p. V-28. 4. A stray by the Ir~ernational Trade Administration. found that the aircraft industry ranked third among U.S. manufacturing industries in 1980 in its level of ~ embodied research intensity, that is, R&D expenditures incorporated in purchased inputs. The aircraft industry was exceeded only by missiles, spacecraft, and electronics in this measure. See U. S. International Trade Administration, Department of Commerce . An Assessment of U. S . Compeci riveness In High Te~hno logy Industries . Washington, DC: U. S . Government Printing Office, 1983, p . 42. J . E. Steiner. "How Decisions Are `Hade: Ma; or Considerations for Aircraft Programs. ~ American Institute of Aeronautics and Astronautics, 1982 Wright Brothers Lectureship in Aeronautics, Seattle, Washington, August 24, 1982, p. 14. "Uben tile market is ready, the successful. manufacturer may have to go. The eventual prize sometimes goes to the company which is fast on its feet.... Ibid., p. 3l, emphasis in original. Moreover, the high fixed costs of supporting such ~ network create ~ strong incentive for producers to market the widest possible range of a~rcraf ~ and engines . Potential purchasers are influenced also by ache lower maintenance cos ts and spare parts inventories associated wit h standardization around the products of a s ingle manufacturer of airframes or engines . - 336 -

i. R. d. .~i~ let and D. Sawers. ~,'~e Tec.~ica: Development of Code Aviaric.~. London: Route edge & Kegan Paul, 1968, p. 267 . 9. Partly in response deco these dramatic increases in development costs, the subcontracting of production of new aircraft and engines has grown substantially in recent years, and increas ingly involves foreign files as partners of ma: or U. S . aircraft and engine producers. The role of Japanese firms as partners in recent multinational j oint ventures in airframe and engine development is discussed in Mowery and Rosenberg, 1985, op. cit. ~ see Reference ~ above ~ . 10. See A. A. Alchi an. "Reliability of Progress Curves in Airframe Production, " S:conomerrica, (1963~; and A. Hirsch. "Fir Progress Ratios, " Economerrica, (1956) . ll. As McCulloch has noted, the high fixed costs that characterize the aircraft industry' s cost structure, as well as the strong learning effects, can glare rise to pricing below average costs. See R. McCulloch. " International Compe~ci~cion in High-Technology Industries: The Consequences of Alternative Trade Regimes for Aircraft. ~ Presented at the Na~ciona~ Science Foundation Workshop an the Economic Implications of Restrictions deco Trade in High- Technology Goods, "Washington, DC, October 3, 1984. In addition to such "predatory pricing, n of course, ache presence of s Prong learning effects means that support or protection of a domestic market can move domestic farms rapidly down the Or leas fling curves, effectively operating as export subs idles ~ in the fashion outlined by Krugman . See P . U . Krugman. n Import Protection as Export Promotion: Interna~cional Compe~cition in the Presence of Oligopoly and Economies of Scale." In Mor~opo~ise~c Competition and InCernational Trade. Edited by H. Kierz~cowsici. Oxford: Oxford University Press, 1984. 12. J. W. Kendrick. Postwar Productivity Trends in the tanned Stares 1948-1969. New York: Columbia University Press, 1913. 13 . B. W. Cravens and 1). Jorgenson. "The Role of Capital in U. S . Economic Growth, 1948-76. n In Capital, Efficiency, and Growth. Edited by G.~. von Furstenberg. Cambridge, MA: Bat linger, 1980. 14. These measures are employed by Miller and Sawers, op. cit., and by N. Rosenberg, A. Thompson, and S. Belsley. Technological Change and ProducritJicy Growth in the Air Transport Industry. NASA Technical Hemor~ndum 7 8505 . Washington, DC: National Aeronautics and Space Administration, 1978. 15 . As Phillips noted, the seat mile costs of the DC- 3 aircraft were "so much lower than those of aite~ate aircraft that even with a - 337 -

.el~t,-vely low load factor its passenger mile costs were often lower than those for other planes. 't See A. W. Phil' ips. Technology and Mariner Struceure. Lexington, MA: Heath, 1971. 16 . Rosenberg, Thompson, and Bels fey, op . cit . ~ see Reference 14 above ~ . 7. Improvements in propulsion technology have been responsible for much of the progress in aircraft performance. Fuel consumption per hour per pound of thrust has dropped by at least 30 percent during the postwar period ~ see Boeing Commercial Airplane Company, Document B- 7210- 2 -418, 1976 ~ p . 41) . Moreover, as temperatures have inc. eased and new materials have been employed the thrust- to-weight ratio of modern turbofan engines has increased dramatically- -by nearly 50 percent during l960- I98O . See National Academy of Engineering. The Coruperi rive Scacus of the U. S. Civil Aircraft Manufacturing Industry. Washington, DC: National Academy Press, 19 BS ~ p . 123 . 18 . The genera ~ concept of n social savings n is discussed extens ive ty in R0 W. Fogel. Railroads and American Economic Growth: Essays in £conometr~c Riscory. Baltimore: Johns Hopkins University Press, 1964; and critically evaluated in P. A. David. "Professor Fogy On and Off ache Rails. n In Innovation, Technica] Choice,- adze Growth. New York: Cartridge University Press, 197S. Direct operating costs per seat mile for the OC- 3 in 1939 were taken from Phillips , op . cit ~ see Reference 15 above); direct operating costs per seat mile and available seat miles for the 1983 fleet were taken from Citric Aeronautics Board. Aircraft Operating Cose and Performance Report. Washington, 3)C: U. S . Go~rers~ment Printing Office, 1966-'984, volumes I-XVII. The calculations were based on available seat miles, rather than actual revenue passenger miles flown. Therefore, they tgnora possible differences in load factors between the 1939 and 1983 aircraft fleets. 20. J. Enos. Petroleum Progress and Profits. Cambridge' MA: MIT Press, 1962. 21. Rosenberg, Thompson, and Belsley, op . cit . ~ see Reference 14 ate ova ~ . 22 . For additional discussion, see Rosenberg, Thompson, and Belsley, op . cit . ~ see Reference 14 above) or Mowery and Rosenberg, 1982, op. cit. (see Reference ~ ablate). 23 . AS*t7C figures for the DC- 8/DC- 10 are taken from Rosenberg, Thompson, and Belsley, op. cit. (see Reference 14 above); that for 1983 is from ache Civil Aeronautics Board, op. cit. (see - 338 -

Reference 19 above) . The 3C- 8 airframe has proven to be so rugged that the original engines recently have been replaced we Ah fuel-efficient C~S6 engines, reducing the aircraft' s operating costs and extending its operating life greatly. Indeed, the ease with which tne DC- 8 can be stretched and re -engined has led some observers to the conclusion that the decision of McDonnell Douglas to close the production line and destroy the tooling for the DC- 8 was exceedingly unfortunate ~ for example, see W. H . Demisch, C. C. Demisch, and T. L. Concert. The Jetliner Business. New York: First Boston Corporation, 1984~. 24. Nathan Rosenberg. "Learning by Using.' In Inside the Black Box: Technology in Economics. Edited by Nathan Rosenberg. New York: Cambridge University Press, 1982. 25. E. van Hi?pel. The Dominant Role of Users in the Scientific c Instrument Innovation Process. n Research Policy, (1976) . 26. These sources of technical change are discussed at greater length in Mowery and Rosenberg, ~ 982, op. cit. (see Reference ~ above) . 21. Boon, Allen and Hamilton Applied Research, Inc. A Historical Study of the Benefits Derived from Application of Technical Advances to Cor~ercia] Aviation. Prepared for the j oint Department of Transportation/Na~cional Aeronautics and Space Administration Ci-~'il Aviation R&D Policy Study. Washington, DC: U. S. Government Printing Office, 1971. 28. Mowery and Rosenberg, i982, op. cit. (see Reference ~ above). 29. N. Terieckyj. "The Time Pattern of the Effects of Industrial R&D on Productivity Growth. ~ Presented at the Conference on Interindustry Differences in Productivity Growth, American Enterprise Institute, Washington, DC, October It- 12, 1984 . 30. J. V. Reddy. The IT Program of the Department of Defense. Peace Studies Program Occasional Paper No . 6 . I thaca: Cornell Uniters ity, 19 7 ~ . 31. D. W. Levy and N. Teriecky]. "Effects of Government Red) on Private RED In~restment and Production: A Macroeconomic Analysis, ~ Be21 Journal of Economics, (1983) . 32. 'reriecky], op. cit. 33 ~ Aerospace Industries Association, op . cit. ~ see Reference 2 above ~ . 34. Chile these product field data also report federally funded research, the category is restricted to publicly supported R&D carried out within incus try and, therefore, understates - 339 -

35 . cons iderably the to Cal public R&D inves-=ment . lathe NSF data are collected only for alternate years after 1977. =e 1978, 198O, and la82 entries for industry- financed R&D in Table 2 therefore were based on linear interpolations. These "R&D capital stock" estimates were not depreciated for several reasons . Pres~mab By, the knowledge resul~cing from the public research investment does not depreciate, nor is its value subj ect to erosion through spillovers to competitors, as is assumed frequently in analyses of priorate Ret) tnvesment (see, for example, H. G. Grabowsict and O. C. Mueller. "Industrial Research and Development, Intangible Capi~cai Stocks, and Firm Profit Rates, n Be11 Journal of Economics, (1978) ~ . rhe physical research plant employed by NASA and the al wed forces does depreciate prest'mably--however, obtaining a reliable Dime series for the physical capital portion of this public investment proved impose ible . As for the private R&D inves tment s Cock, were this analysis considering the "~&I) capital stock" from the point of view of an ~ ndi~ridual fine, deprecia~c~on of this stock would be appropriate . However, inasmuch as this s Cock is be ing seed across all of ache firms in the industry, ache n sp i iloVer" j Ossification for depreciation of this component of the R&D capital stock also seems weak. 36 . R. R. Nelson. "Government S timulus of Technological Progress: Lessons from Amend can History, " In Government and Technical Progress: A Cross-Industry Comparison. Edited by R. R. Nelson. New York: Pergamon Press, 1982 - 37. R. R. Nelson "Assessing Private Enterprise: An Exegesis of Tangled Doctrtr~e, ~ Be]] Journal of Economics, (198-L) . 3 8 . See ache report o f OS ~P on f ederal ly supported aeronautics R&D, op . cit . ~ see Reference 3 above) . 3 9 . S tatement of Or . Joseph Ames, NACA Chairman, before the President's Aircraft Board, 1925. 40. "By a comprehensive survey of the net efficiencies of various engine nacelle locations, the optimum position in the wing was found. This N.A.C A. engine location principle, together with other refinements' had a revolutionary effect on military and commercial aviate on ache worn' over. It changed military aviation tactics, made long- range bombers pass ible, and forced the development of higher speed pursuit planes. In the commercial field it permitted the speeding up of cruising schedules on the air lines from i20 miles per hour of ache Fords to the lB0 miles per hour of the new Douglas planes. The overnight transcontinental run became possible and the air lines vastly increased the, r appesi to ache public.. Even in the midst of the depression, air line traffic boomed. n See J. C. Hunsaker. - 349

Research in Aeronautics. n In ScLtiona~ Resources Placr.ing Board. .Researcn--A National Resource. Washington, DC: U. S. Government Prin~cing Office, 1941, p. 139. 41. "The old NACA was s'rictly a research, test, and advisory organization . I t built, modified, and owned few aircraft and in the main only tested aircraft submi Ted to it for evaluation, or 'debugging. ' " See Legislative Reference Service. Policy Planning for Aeronaurical Research and Development. Prepared for the Senate Committee on Aeronautical and Space Sciences, U. S. Congress. Washington, DC: U. S. Government Printing Office, 1966, p. 107. 42 . E. W. Cortstan~c. The Cretins of the Tu~boje~c Revolur ~ on . Baltimore: Johns Hopkins University Press, 1980. 43 . Me National Research Council' s surveys of industrial research laboratories before and after World War lI convey some idea of the scale of the expansion of the aircraft industry' s in-house research capabilities. The in-house research staff at Douglas Aircraft grew from 22 in 1940 to ti: in 1946; the research staff at the Glenn .Mar~cin Company grew from 42 to 76; Loclcheed's grew from 10 to 314; Consolidated Vultee's grew from 12 to 195; United Aircraft, which included Pratt and Whitney, Hamilton S tandard, and Sikorsky, increased its research staff from 80 to 732; and. Curtiss-Wright expanded its research employment from 14 in 2 940 to lS9 in 1946. These figures are all the more im?ress:,re ~ n view of the fact that these firms' military aircraft sales, especial y to Britain and France, had already expanded considerably by 1940. See National Research Council. Indus Aria ~ Researc.h Laboratories of cbe lJni ted Stares . Washington, DC: National Research Council, 1940, 1946. 44. Legislative Reference Service, op . cit ., p . 20 ~ see Reference 41 above ) . . Committee on NASA Scientific and Technological Program Reviews. Aeronautics Research and Technology: A Review of Proposed Reduc Lions in the FY 1983 NASA Program. Washington, DC: National Research Council, 1982. lye other components of ache NASA aeronautics budget are "Construction of Facilities " and "Research and Program Management . n 46 . Miller and Sawers, op . cit ., pp . 255 - 256, described the Manufacturers' Aviation Association (MAA) licansir~g system as a system Sunder which all aircraft manufacturers agreed to bet all their competitors use their patents. No member can have a patent monopoly on any inventions which his staff can make, or even for any invention t}~.at he may license from an outs ide inventory . If he takes an exc lus ive license, the patent has to be available to the other members of ache MAA; but the original licensee can claim 341 -

Chat he should be granted compensation by the other licensees. Manufacturers apparent' y bell eve that it is a good bargain to give up their right to a patent monopoly in return for the protection from litigation wash other companies in the incus try that the right to use their patents brings " (pp . 2SS - 256 ~ . Roland discusses the - origins of the cross- 1 icensing agreement. See A. Rot and. Model Research: The National Advisory Commi tree for Aeronautics, 1915-58 . Washington, DC: U S . Government Printing Office, t985, pp . 37-43 . . J. HirshisifPr. "lathe Private and Social Value of Information and the Reward to Inno~ra~cion, n American Economic Review, (1971) . . Nelson, op. cit. 49. See the "Draft Interim Report of the Ad Hoc Tnfo~maL Subcommittee on NASA Aeronautical Projects," NASA Aeronautics Advisory Committee, 1983. These estimates should be viewed as i llustra~c1 ve of general orders of magnitude ~ rather than precise figures. In addition, of course, any assessment of the net benefits resulting from the in~restmen~c in NASA research infrastructure mus ~ consider the returns from alternative inve s tments o f pub ~ i c funds . 50 Indeed, the 707 airframe design followed that of the KC-135 so closely that =he -first 707 to be "rolled out" of the Seattle factory did not have windows in the fuselage. :l . Miller and Sawers, op . cat., pp . i93- 194. 52. National Academy of Engineering, op. cit., p. 101 (see Reference ' ? above) . . Ibid. p. 102. 54. For additional discussion of Japanese cooperative R&I) programs, see D. T. Okimoto. "Pioneer and Pursuer: The Rote of the State in the Evolution of the Japanese and American Semiconductor Industries. n Occasional Paper, Northeast Asia-United States Forum on International Policy, Stanford University, 1983; and IS. J. Peck and A. Gato. "Technology and Economic Growth: lathe Case of Japan," Research Policy, (19811. The analogy between U.S. aeronautics R&D and these Japanese programs is developed in greater detail in flowery and Rosenberg, 1985, op. cit. (see Reference ~ above). 55. P. A. Oavid. "Technology Diffusion, Public Policy and Industrial Competitiveness. ~ Presented at the National Academy of Engineering Conference on Economics and Technology, S tanford University, March IB- 20, 1985 . - 342 -

6 . ~ . C . .Mot~ery . " Economic Theory and Government Techno logy Policy, n Policy Science, ~ ~ 983) . . 57. The federal government has played a critical rot e in the evolution of the al rline industry's structure throughout the industry's history. Divestiture of airlines by aircraft and engine manufacturers was mandated by law in 19 34, following a series of congress tonal in~restiga~cions of political influence in mail contract awards . Those awards during ache 1920' s had been employed in part to build up large transcontinental carriers, which in Curie would provide a market for advanced aircraft. For further details, see Mowery and Rosenberg, 1982, op . cite . ~ see Reference 1 above ~ . 58 . W. Jordan. Airline Regulation in America . Baltimore: Johns Hopkins Un_~rersity Press, 1970, ?. S3. 59. T. Keeler; "Airline Regulation and Economic Performance, " Bel Journal of Economics, (1972), p. 421 60. G. W. Douglas and J. C. Miller. Economic Regulation of Domes tic Air Transport . Washington , DC: Brookings Institution, 19 74 . 51 . General Ace ounting Off ic e . Lowe ~ Airy ~ He Cos As Per Passe.~2ge - Are Possible and Would 2-stllc in Lower r arcs. r-ashirlgton, DC: IJ. S . Government Printing Office, 1977. 62. Keel er, op. cit. 63. Ibid. 64. Fogel, op. cit. (se" Reference ~ 8 abjure) . 6~. E. Mansfield et al. The Production and Application of New Industrial Technology. New York: W. A. Norton, 1977. 66. We invaluable Civil Aeronautics Board series begins only in 196S . 6 7 . R. C . Len=, J . A . Machnic, and A . W . Elkins . l.he influence of Aeronautical R&D Expenditures lJpon Ohm Productivity of Air Transportation. Dayton: University of Dayton Research Institute, 198~ 68. Mansfield et at., op. cit. 69. General Accounting Office, op. cit. 70. Mansfield et al., op. cit. — 343 —

7i - 4.e adapts tion and application of military research inves Moment to commercial aircraft also requires a substantial investment in incus Cry - f inane ed res earch . 72. Terleckyj, op. cat. Terlec~rj 's analysis fails deco account for publicly financed research not performed in industry. In addict on, in view of the fact that his estimates of the embodiment lag were computed for the 1959~1976 period, one during which CAB regulation presumably minimized such lags, his estimates icky be Coo low for the post- 1978 period. 3 . The study by Phillips, op. cit., of the demand by trunk airlines for transports found that controlling for other performance characteris tics, regardless of the national identity of the producer, foreign commercial ~cranspor~cs were less attractive than those manufactured in ache United States to the U. S . domestic airlines during the postwar period (p . 102 ~ . 74 . Krugman, op . cite . (see Reference t: above ~ . Once again, the similarities wi Oh Japanese industrial policy, which in many industries has worked initially to develop competitive firms serving domestic markets that are formally- or informally protected, followed by the movement of such firms co international export markets, are very strong, an argument developed further in Mowery arid Rosenberg, 1985, op. cit. (see Reference ~ above ~ . 76. flowery, t983, op. cit. 77. Ibid. 78. D. C. Mowery. "Firm Structure, Government Policy, and eke Organization of Industrial Research: Great Britain and the United States, 1900 1950, n Business Sorer Review, (] 984) . 79. The unusual combination of high concentration and occasional dramatic changes in the market shares of firms in ache commercial aircraft industry noted by Phillips, op. cit., also may reflect ache existence of an indus~cry-wide knowledge base that was accessible by a number of different fines. Rather than a single firm establishing on unchallengeable technological lead, ache industry has witnessed recurrent technological "leapfrogging" of one firm by another. While Phillips attributes this tendency to the existence of a substantial exogenous research effort funded by NASA and the military, the ease with which incumbent finds could tap that external knowledge pool also has been important. 80. Teriecicyj, op. cit., suggests that this absorption function of self - financed research investment within the industry may explain ache apparent lack of sta~cist~cal significance for publicly funded · 346 _

research ~ r.vestment in explaining productivity growth in the air transportation industry: "It is also possible that ache private R&D expenditure es made to adapt the results of government R&D Deco private products already incorporate the effects of governmen~c R&D. The military R&D which is represented in the government and to a large degree in the total R&D expenditure for aircraft and parts by itself does not produce products sufficiently developed for private use . " (p . 32 ~ . 8 ~ . W . M . Cohen and D . Le~rinthal . " The Endo gene ~ ty ~ f Appropriability and Red) Investment. n forking paper, Carnegie -Mellon L'ni~rers i By, 19 8 S . 32. Rosenberg, op. cite. 8 3 . S . J . Kline and N . Rosenberg . "An O~rerv~ew of Innovation . n Presented at the National Academy of Engineering Conference on Economics and Technology, Stanford University, March 18-20, 1985. 84. L. J. Bite. "The Motor Vehicle Industry. n In Government and Technical Progress: A Cross-Induscry Anai~sis. Edi ted by R. Nelson. New York: Perg~mon Press, 1982. . L. J. Chide. ~.Au~omobile Emissions Control Policy: Success Story or wrongheaded Regulations? " In Government, Technology, and the Feature of ire Automobile . Evinced by ~ . H. Ginsburg and W. J . Abernathy. New York: McGraw-Hill, 1980. . Ibid. a 8 7 . n ~ T ~ he standards approach encourages coil us ion among the companies. Though the incentives facing the individual company might lead it to make an all-out effort, it is clear that the incentives for the industry jointly are to delay. To the extent that the industry presents a united front and says that the technology is not available, brinicsmanship becomes yet easier, delays in standards enforcement become yet more certain, and public relations advantages become yet greater. The policymakers and ache public are dependent to a great extent on the companies Co inform them as to the state of technology and its feasibility.... Delaying the development of emissions-control technology and delaying the reporting of information about that techno~ ogy are clearly in the industry' s j oint interests . n Ib id., p . 406, er~phas is in original . 88. lathe recent report of the Office of Science and Technology Policy, op. cit. (see Reference 3 above), adopts this view of NASA's . role. 34-

89. `ToT1 and Owen suggest Chat economic analyses of the dis tributional and efficiency impacts of regulation p rayed a maj or role in a number of deregulatory policy episodes of the 1970's. See Roger G. Noll and Bruce M. Owen. "The Political Economy of Deregulation: An Over''iew. " In The Political Economy of lDeregula~ion. Edited by Roger G. Noll and Bruce M. Owen. Washington, DC: American Enterprise Institute i 1983 . 346 -