ABSTRACT
The US government has tried to help high-tech start-ups with subsidy programs, the largest being Small Business Innovation Research (SBIR). Despite spending about $6 billion, there is scant evidence that it had any economic effect. That is, the US high-tech economy would have advanced as it has without any subsidy. Program evaluation by the government has avoided the hard questions and the programs continue undiminished. Only two bright spots have appeared, both in the Department of Defense, wherein new technology in young companies has apparently been substantially helped by SBIR. Only three start-up firms in the 80s can be held up as examples of SBIR investment leading to public shareholding in the 90s. SPIE companies can expect continued subsidy from which they can gather development funds at the risk of becoming addicted to investment that demands no market discipline.

Keywords: start-up company, government subsidy, SBIR, ATP, new technology, evaluation

2. YOUR GOVERNMENT IS HERE TO HELP YOU - THE VISIBLE HAND
The world marvels at the US high-tech start-up phenomenon that has made more people rich faster than any economic movement in history, even the Spanish plunder of the New World's gold and silver. The paper wealth in Silicon Valley in the form of stock options to ordinary employees beggars belief in Malyasia and France. The world's richest industrialist dropped out of one of the world's top few universities to start a software company whose market capitalization twenty years later is larger than most national economies. And yet the US Congress thinks these start-ups need help in the form of direct subsidy for research and development. Ain't politics a hoot?

Startups abound in America, and no type of startup gets more attention from financiers than high-tech startups. High-techs, better than the town pizza emporium, each have the chance to make dozens of millionaires. They are "the stuff that dreams are made on" combining explosive wealth appeal with the dream of a better life through gadgets.

With all that potential for start-ups' attracting more help than they can stand, need government do anything to help, other than what it does for any business entity? As midwife or as financier? Not really, but government has intervened anyway. Politicians need something to take credit for, as they did so gleefully in their 1997 balancing of the federal budget by lowering taxes and raising spending.

The US government hands out $85 billion a year1 in 125 business subsidy programs granted by an elected legislature that loves economic nationalism and helping its friends. The technology subsidies rely on a market-failure view that government should spend (invest) in technology to fill the gap left by inadequate, short term private investment. Without such subsidies, say these market-failurists, the nation would 1) lose the societal returns from innovation which Mansfield2 found 2-3 times the return to the innovator, 2) lose the technologies from within America (an increasingly outmoded view say Nelson & Wright3), and 3) lose jobs to foreign competitors. The hard question isn't so much whether start-ups have more innovation, but whether government subsidy can have any effect.

Helping startups more directly than just by capital gains tax cuts is politically appealing. But that does not mean that government does so efficiently nor that the help will show any more economic benefit than doing nothing (a trumpeted Conservative principle). In fact, the programs have little to boast about when examined through an economic lens.

Says one realist, The political game in Washington is a spoils system, The petitioners try to spread a verbal fig leaf over their desires. They say, "Gotta have this [capital gains tax] cut to encourage entrepreneurship and investment in stocks." Who are they kidding? Entrepreneurship is booming and so is the stock market.

3. THE BIG LITTLE GOVERNMENT PROGRAM - SBIR
The premier government program for high-tech start-ups is Small Business Innovation Research - with even a title that politicians can love - $1+B a year for R&D ($6B cumulative to date) with commercial potential.i Note that is not new R&D money, just a forced redirection of an agency's R&D spending. In principle, a typical SBIR gives a company $800K for an R&D project after which the company is expected to find its own product development financing. In practice, things don't work that simply as would be expected where free money can be had for learning to use the rules. GAO counted that in 1990-1997, 6500 companies had awards from the five biggest awarding agencies.

SBIR started with a market-failure excuse for small business's grabbing a bigger piece of the federal R&D expenditure pie partly by appealing to the paranoia of the 80s. Japan was "winning" according to a world view that looked only at the trade in hard goods while ignoring the global interdependence and an ever growing difficulty in defining a national firm.

The federal agencies, who both want and hate innovation, have made SBIR into an ambivalent program as the mission agencies (like DOD and NASA who need technology to perform their missions) spend most of their SBIR on incremental advances with little market appeal. SBIR therein has been captured by the bureaucracy who see no role in the private economy and don't yet believe in integrating commercial and government technology, especially in Defense where half the SBIR money is spent. Most agencies rationalize their incremental approach by accepting almost any commercialization story that SBIR companies spin and treat SBIR as a mere extension of ordinary R&D programs and use the spending guides in the law as an excuse not to fund the kind of start-up whose capital appetite grows as it succeeds. Thus they condemn SBIR to fail as a competitive venture program.

3.1 Fast Track
One exception to the insensitivity to market forces is a DOD variation within its SBIR - Fast Track - wherein DOD matches third party cash at a ratio that depends on the company's SBIR maturity. SBIR virgins get 4:1 and SBIR veterans get 1:1. The motivating idea wanted to speed those new concepts that had provable markets and market timing windows. One estimate by a Hewlett-Packard cognescentus says that six months sooner to market means 30% more total profit and six months later 30% less total profit and that one month sooner would cover the product's entire engineering and development costs. Third party cash provides a commercial validation that government cannot provide and separates story tellers from entrepreneurs.

DOD's first year of Fast Track had 53 winners of only 57 qualified applicants (about 10% of all proposals), with 70% of the companies getting the 4:1 match. In all, DOD put up $45M (of its $600M total SBIR) for the third parties' $18M 4. That's $45M where almost none would have been invested without Fast Track into such market-driven projects. As long as Fast Track proposers comprise only a small fraction of the total, the DOD agencies can afford to do so without abandoning their incremental approach to R&D.

With the exception of the Ballistic Missile Defense Organization (BMDO) which already had a Fast Track equivalent, the military thus dipped its toe into the market waters in a new way. Until Fast Track, DOD was satisfied with a linear model wherein government R&D was followed by strictly private commercialization. Such linearity served the autarkic Cold War approach to military R&D that the Clinton administration realized it could no longer afford. The military services, still wedded to the linear model, still resist Fast Track's model of integrating military and commercial R&D.

Even though there is no compelling evidence that SBIR makes any difference for start-ups, Fast Track does more than any other agency for the few market-driven entrepreneurs who come to government in the first-place. It enforces economic reality instead of self-deceptive fantasies by companies to win government contracts. And it lures well-heeled partners.

One other Fast Track has started at National Institutes of Health which focuses only on time (and not private capital) since many beneficiaries complain to Congress about slow government action hurting their companies (especially if they depend on SBIR for cash flow). They complain that the free money isn't handed out fast enough. Without third party capital investment as a criterion, NIH thus has only the chance to do the wrong thing faster. Results are not yet available for NIH's experiment.

3.2 SBIR Evaluation - Hah!
Since SBIR added no new money to national R&D, it has the burden is to show that a forced distribution produces better results in the technology industries than an unforced one. No study has yet shown such an effect. What are SBIR's results other than the targeted constituency getting a pile of money? Four different groups have had a chance to evaluate SBIR - 1) the GAO, 2) the SBA, 3) the federal R&D agencies, and 4) outsiders. Only the outsiders did an economically acceptable evaluation while the others merely served their institutional interests.

3.2.1General Accounting Office
In 1992 GAO found that two-thirds of projects got follow-on activity and that in total the firms got $1B on top of the $1B of SBIR funding with half of that coming from the private sector and half from more government sales. Further, the beneficiaries expected (unauditable, of course) another $2B in sales in the following 18 months plus $1B more in development funding.

While such numbers sound impressive to a Congress inclined to shelter small business, they do not suffice to evaluate an investment program. They have rather the usual government lack of an objective standard, no Return On Investment calculation, and no control group for comparison with doing nothing at all. Indeed, Roessner5 forecast such a control-free evaluation that guaranteed the program's future and avoided a hard market discipline. As a result GAO couldn't say much more that some good was done with the $1B. In politics, that's good enough for a favored group and quite insufficient for an disfavored group (such as welfare recipients).

3.2.2 Small Business Administration
SBA, charged with the Executive Branch oversight did what political entities typically do - measured percent participation. Its September 1997 testimony 6 said that 24% of projects finishing Phase 2 had commercialization (40% if products of more than one award are considered). Percent participation is hardly the right metric for investment. Investors measure total return, not percentage of units getting something. No VC boasts that 24% of the companies he invested in had sales. One conclusion from the evidence: a defendant getting that kind of representation in a criminal case could win an appeal on inadequate defense. And since SBA has an interest in small business programs, it is careful not to dash any cold water on the idea that SBIR serves America. SBA has not published any economic study of SBIR that uses a control group for comparison. One cannot believe that SBA has not considered the question and its silence must be judged as evidence that the results would not look good for SBIR. SBA also finds it a highly competitive program because it gets a lot of proposals, not an uncommon outcome when a lot of money is on offer with vague guidelines on what will win the money.

3.2.3 Federal R&D Agencies
Agencies don't try to measure against external standards, focusing instead on procedure. Mainly they do so because they see no need to sell their product to the customer class - the small business high-tech community. Nor indeed, do they accept the implication of the existence of the SBIR law that the high-tech small business community is the customer. As rule-driven bureaus they ask (sshh, quietly) only if they followed the rules correctly. Most don't even keep (or at least publish) any useful statistics such as: ROI, amounts co-invested, employment growth, fraction going to first-time winners, fraction going to multiple winners. They do publish what they can count easily and internally - proposals received, proposals funded, amounts spent, distribution by state or technology, etc.

One sub-agency, BMDO with about $40M a year, 5% of the total SBIR money, did look at external measures. Nelson7 reported capital raised by IPO, matching investment during the SBIR contract, even a hypothetical ROI for a hypothetical equity investment in the companies. By 1995, a dollar of BMDO SBIR money was being matched on average by a dollar of private sector investment in the projects. No other agency seems to have reported any similar analysis. But such BMDO's results came from a strategy that integrated government and commercial technology development for its small R&D companies.

3.2.4 Outsiders
All the government "evaluations" omitted a critical element in evaluation - a control group. The government evaluations looked only at the beneficiaries, and the federal agencies haven't even answered the question (publicly) whether SBIR was better for the agency than a control group of its non-SBIR contractors. Two outsiders did use a control group.

Josh Lerner, a former GAO auditor, started with the GAO results and asked the next crucial question - was the growth of those companies different from the growth of similar companies without SBIR. It was the first time in the life of SBIR that someone used the standard statistical approach of comparing results with a control group. Results, not input. Lerner8 found 56% more jobs and 98% more sales in companies that got SBIR 1983-1985 (the first years) which might convince anyone but the most hardened free-marketer of SBIR's value. But, the effect was almost wholly confined to zip codes where venture capital was already actively investing. Which fits Mokry's dictum that "The easiest solution for a policy maker who wants to promote enterpreneurship is to get elected in a place where natural entrepreneurial activity is already well established."9

One qualifier for Lerner's findings is that he did not subtract employment directly paid by any subsequent SBIR contracts - a large number for several companies and not a measure of job growth but only of job subsidy. After all, a government funded job in R&D is economically no different than a government funded job sweeping the sidewalk. He also found that "the presence of an SBIR award alone had little relationship with employment and sales growth."

Scott Wallsten, a Stanford economics student, asked a different question - did SBIR merely replace other R&D spending?. He found answers that the beneficiaries don't want to hear. He found that indeed it did, that SBIR had no beneficial effect on R&D spending nor jobs nor sales, a mere crowd-out effect.10 The publicly owned firms, for which such data are available, showed that they reduced their own funding of R&D dollar-for-dollar as SBIR was received.

3.3 What About Evaluation, Anyway?
It is, of course, wholly inadequate to argue that R&D spending did some good because it ignores the opportunity cost of not making some alternative investment with the funds. Each agency has long argued, in its both active and passive resistance to SBIR, that in effect the presumed benefit would outweigh the opportunity cost. The true test lies at the margin: the worst SBIR should have as much value to the agency's objectives and the worst non-SBIR. An unrestrained agency would inherently meet that test as all proposals are judged by the same criteria. If any agency has somehow grossly misjudged the value of small business innovation to its mission, the answer is not to blindly force more small business funding but to enlighten the agency about serving its own interests.

One interest that agencies do regularly overlook is the benefit from commercial adoption of new technology. The Cold War autarky dies hard because it has a hardened array of beneficiaries who are already sad at the demise of the Soviet Union let alone a re-think by the government about the sources of its next generation of technology. Perhaps the biggest disappointment to that system was the personal computer which invaded the military without any help from the established suppliers of defense systems. The era of special purpose computers from one or two vendors for each application was wiped out by a multi-purpose platform that needed only software tweaking to perform most of the functions that the specialized computer did. And quick turnaround and imagination in software came from the outside.

Evaluations of politically driven programs almost always and conveniently overlook the opportunity cost. While such an oversight fits the concentrated benefit-diffused cost syndrome, it mocks evaluation by any economically defensible standard. Indeed, even GAO succumbed to the oversight, no doubt to accommodate Congress. Wallsten, however, forcefully made the point in1997 testimony11 before Congress that cracked the wall of boosters that regularly testify.

One measure might be to compare SBIR's results with the venture capitalists' results. "Seven savvy prognosticators - VCs, technology execs, developers and industry watchers" picked hot prospects in seven categories for Upside [magazine] June 1997. The picks in semiconductor manufacturing and microprocessors and mobile communications - all of interest to government - were all California companies: Form Factor, Rambus, Chromatic Research, Synaptics, InterWave, Metricom, and Vadem. Now ask how much did SBIR invest in those companies in the three years 1993-1995? Synaptics got $0.4M of something like $500M SBIR poured into California. Many lists (Business Week, Inc, Forbes) of fastest growing high-tech companies show few SBIR companies regardless of the period chosen. here then did the SBIR go in California 1993-1995 SBA data say that Physical Optics lapped the field - $28M. Two other large recipients: Deacon Research - $6M, Mission Research - $6M. POC has been collecting huge sums since 1987; Mission Research is a government-style research house in four states.

Why so? Fast growth companies can get capital whereas slow growth and no growth companies must appeal to government. It's an old game: if you can't win in the market, win in the legislature. Appellants have plenty of models to follow: tobacco, ethanol, solar power, ....

Evaluation as an investment portfolio would have noticed that the by-far largest investment had fizzled long ago. If you had invested $60M over a decade in your largest investment what return would you expect from the company? The average VC got 42% last year. A stream of $5M a year for 12 years compounded at just 25% would be about $500M net worth for just the (presumed) minority share which implies a market cap in the billion dollar range. Has the largest SBIR recipient, still a private firm and still getting SBIR, returned such profit to the public coffers? Since it is a private firm, it can conceal its finances from public view and thus frustrate any attempt to evaluate its investment performance. Similarly the other largest "investments" are almost all private firms and likely to remain so since they do not make good IPO material. They thus have no chance to create the dozens of millionaires each with stock options. No matter to them; they like doing contract R&D.

Nor is Congress likely to mandate any hard evaluation criteria such as market capitalization since such a measure would not serve the present beneficiaries. Only a handful of SBIR companies have gone public and thus exposed themselves to public scrutiny. Those which had SBIR as a major role in their getting started make inspirational reading and the beneficiaries can use them as political cover for continued subsidy. Advanced Technology Materials, Cree Research, and SatCon Technology, each had fewer than ten employees when they first got SBIR and their combined market cap today approaches $1 billion.

4. OTHER PROGRAMS

4.1 Advanced Technology Program
Back in the Japan-scare days of the 80s, not unlike the late eighteenth century scare talk in Britain as Yardeni12 reminds us, the Commerce Department got into the competitiveness business by sponsoring purely commercial development in an innocuously named Advanced Technology Program. The Bush administration would not go to the wall for its free-market principles and when the Clinton administration wanted to "focus like a laser on the economy" ATP was a perfect vehicle. The politics of corporate welfare soon dominated the debate after 1994 and ATP claimed to then tack toward small business.

ATP gives R&D grants for technologies chosen by Commerce as meriting investment in the national interest - a tricky judgment at best that favors existing technologies at the expense of new technologies. Companies must match the government "investment", a total of 280 cost-sharing awards to individual companies or industry-led joint ventures to develop high-risk, enabling technologies that would stimulate the U.S. economy by making possible important new products, services, or industrial processes for the world's markets. Or so goes the prose. Nearly half (46 percent) of these have gone to individual small businesses or to joint ventures led by a small business.

4.1.1 ATP's Evaluation
A survey conducted by Silber & Associates13 found that of 125 companies participating in ATP projects during the first three years of the program, 70 percent said chances were slim to non-existent that they would have pursued the technology development at all without the ATP. Of the 30 percent that would have gone ahead anyway without the ATP's assistance, nine out of 10 said their goals and level of effort would have been scaled back significantly and progress would have been significantly slower. A parallel GAO study in 1995 for the House Science Committee found similar results. Imagine that! The beneficiaries said they benefited from free money for what had been economically uncompetitive projects. Which isn't necessarily bad but does impose a burden of proof to show that public funds produced a beneficial result that passes an opportunity cost test. So far, no such proof has appeared.

Sixty-one percent of those planning to commercialize the ATP-sponsored technology said they had uncovered an average of three new, unforeseen applications since the project began. Over 86 percent of the participants believe the ATP award will enable them to make a better product, in terms of quality and performance. (Companies must pay all costs for product development; ATP funds may not be used.) Fifty-five companies reported that they had adopted permanent process improvements in their own operations based on the ATP project. Eighty percent of those planning to commercialize the ATP technology reported that their companies have taken some steps toward marketing the products, processes, or services that ultimately are expected from the ATP project; one- third of these companies expected to earn revenue from ATP-based technology before the end of 1995; and 19 of the companies reported they were currently earning revenue at the time of the survey. None of which justifies government spending since a useful outcome is insufficient proof.

Fifty-six percent of the respondents to the Silber study reported that the project resulted in the creation of an average of six new jobs. Twenty-eight percent reported that the award enabled the company to retain an average of three jobs that otherwise would have been eliminated. Proof that government money pays for jobs. What happened to those jobs when the subsidy ended?

4.2.3 ATP's Future
Did the ATP companies merely benefit from the growing US economy and high-tech competitiveness or did ATP help cause the growth? Depends on your political view. Unlike SBIR, ATP has to fight an annual appropriations battle where all the enemies annually propose elimination. At least one serious proposition has been made to devolve the program to the states14. Despite the sub-optimization that would result, the politics could be ripe for doing such a thing. After the Silber "evaluation" The Secretary of Commerce issued the usual sounding statement that changes would be made for the better15 One implied change is more awards to small business, although not necessarily start-ups.

Note that the list includes neither Intel nor Microsoft nor Apple and surely not the Cypress Semiconductor of TJ Rodgers who repeatedly flails such corporate subsidy16. Yes: Motorola and Hewlett Packard. "It's just a political grab bag. Whoever can get a subsidy takes it", says WC Gale - a Brookings tax economist. And says Howard Gleckman in Business Week, July 14, "Never mind that those loopholes rarely encourage economic growth, as their backers claim. At best, they are a waste of taxpayers money. At worst, they actually discourage innovation by subsidizing old technologies"

4.3 Energy Related Inventions Program
The Commerce Department also has a tiny porthole for energy-related inventions, anything that would save energy. NIST funded 3% of the 31,000 applications (such percentages are meaningless in program evaluation and are usually a measure of how sloppily the government has described the actual competition). By the end of 1992 after $31 million of grant money disbursed, a fourth of the funded inventions had "entered the market" with total sales of $763 million and $531 million of saved energy expenditures form one millions of carbon not burned. The expenditures were small and the results open to substantial controversy in interpretation.

5. WHERE IS GOVERNMENT GOING AND HOW SHOULD PHOTONICS REACT?
Anyone now getting or planning to get an SBIR award would like SBIR to continue. Government has still not shown that SBIR nor any other such subsidy program has any demonstrable economic effect. And the political argument for more and more rests on a still unproved theory of market failure. Still, they will go on. The concentrated benefits to the few outweigh the small loss to the many.

Whether such an award is in the company's best interest remains debatable. Addiction often combines with passing up a source of management wisdom in the name of retaining control. Photonicians can use the matching SBIR requirements as a lever to attract strategic partners. They can also use the money to develop new technologies to the point of market competition. Unfortunately, SBIR's experience is that many companies adopt an outlook of getting more SBIR and prolonging the market introduction of a still immature technology. With that much cash flow on prospect it is easy to convince oneself that technological immaturity is the only barrier.

A few aggressive, market-driven photonics start-up beneficiaries (like CoreTek, Picolight, and Templex Technology) would weep at SBIR's loss. But not for long. With the chutzpah and tenacity to succeed with SBIR, they would succeed without SBIR because they would have gotten the government funding anyway from the many photonics research programs for strictly government purposes, from agencies like DARPA. If an agency is funding photonics in SBIR, it is also funding photonics in its mainline programs where the start-ups have just as competitive a story. The losers, who would be little loss to the high-tech world, would be the research-driven companies whose technology has marginal market value, usually because it costs too much. The funds that would have otherwise been diverted to them would revert to the open competition within the agency for its R&D mission.

Investment coffers are overflowing with money for products that have a profit potential. The gigantic money made in the stock market of the nineties has induced a flood of new VC funds and a choir of angels wanting to re-invest their new wealth. "The unprecedented flood of venture capital pouring into high technology start-ups is feeding a nationwide outbreak of entrepreneurial fever. There's a river of money out there", says Ann Winblad17, co-founder of software VC firm Hummer Winblad. Six new age business angels appeared in Silicon Valley recently dishing out money in tidbits under $1M. Aspen Ventures ($45M), Novus Ventures ($13M), Tech Farm ($35M), The Band of Angels (77 members have $21M), Interactive Minds ($manyM), New Vista Capital ($10M+$40M).18

Whether such programs as SBIR are good or bad depends on your politics and as the old bureaucratic saw goes "Where you stand depends on where you sit." And if, as Peters says, "If you're IBM's Akers, you continue to wonder why 50,000 competitors, most of them microscopic, are eating your lunch.19, why did the government need to also feed the lunchers? The programs won't go away even though they won't do any particular good nor harm because market-failure will persist as the excuse for political payoffs to interest groups. If the interest groups then waste the opportunity to live for the subsidy, they have only themselves to blame for the dependence and the lack of progress in any direction.

6. REFERENCES
i The money comes in two competitive stages. First a small amount, typically $70K to show feasibility, and then typically $700K for development. Nearly half the $70K winners go on to the $700K. Any economic evaluation of the program should consider only the $700K projects.

1. Moore S, Stansel D, "Ending Corporate Welfare As We Know It", CATO Institute Policy Analysis, CATO Institute, 1000 Massachusetts Ave NW, Washington DC 20001; and Anon, 1995, "Business Welfare", The American Enterprise, p82-83, Jul/Aug 1995

2. Mansfield E, "Social Returns from R&D: Findings, Methods and Limitations", Research and Technology Management, 34(6), p24-27, Nov/Dec 1991

3. Nelson RR, Wright G, "The Rise and Fall of American Technological Leadership: The Postwar Era in Historical Perspective", Journal of Economic Literature, XXX, p1931-1964, Dec 1992

4. http://www.acq.osd.mil/sadbu/sbir

5. Roessner JD, "Evaluating government innovation programs: lessons from the US experience", Research Policy, 18(1989)343-359, 1989

6. D O Hill, Testimony before House Subcommittee on Technology, September 4, 1997

7. C Nelson, Wavelength Division Mini-Money: Small Subsidy for Small Business", SPIE Photonics West, SPIE Volume 2690, January 1996, also at http://www.idsonline.com/business/cwnelson

8. J Lerner, "The Government as Venture Capitalist: The Long-Run Impact of the SBIR Program" Harvard Business School Working Paper #96-038 and NBER Working Paper No. 5753, 1997

9. BW Mokry, "Entrepreneurship and public policy: can government stimulate business startups", Greenwood Press, 1988

10. Wallsten, S, "Rethinking the Small Business Innovation Research Program", in Investing in Innovation: Creating a Research and Innovation PolicyThat Works, L Branscom & J Keller, eds, MIT Press, Cambridge, MA, 1997, forthcoming

11. SJ Wallsten, testimony before the House Subcommittee on Technology, September 4, 1997

12. E Yardeni, "The High-Tech Revolution in the US of @", CJL(awrence)/DBS Topical Study#25, 1995

13. Silber & Associates, Survey of Advanced Technology Program 1990-1992 Awardees: Company Opinion About the ATP and Its Early Effects, January 1996.

14. L Branscom, Issues in Science and Technology, Spring 1997

15. http://www.nist.gov/public_affairs/atp60day.htm

16. TJ Rodgers, "America's Corporations Should Swear Off Pork", Wall Street Journal, June 23, 1997

17. P Harris, Technology Business, May/Jun97

18 . San Jose Mercury, Jun 9, 1997

19. T Peters, "Rethinking Scale", Cal Mgt Rev, Fall 92, 35(1), 7-29