Guest Column

William C. Wood

Show Me the Money

Despite extravagant claims, small business programs stimulate little economy-wide job formation or wealth creation. Mainly, they help lucky beneficiaries at the expense of their competitors.

Editor's note: The Warner administration caught a lot of grief this year when it slashed funding for the Department of Business Assistance and cut state support for small business development centers around the state. Small business programs pay for themselves many times over, program advocates claim, by helping launch businesses that create jobs and generate taxes.

In a 1994 article in the Journal of Private Enterprise, William C. Wood, an economics professor at James Madison University, questioned the putative returns on most government small-business assistance programs. Wood referred his article to the attention of Bacon's Rebellion, suggesting that its conclusions are as valid today as they were nine years ago.

Is entrepreneurship a delicate flower that wilts unless watered by the government? Or is it a kudzu vine, overgrowing everything in its way despite being hacked, sprayed with herbicides and grazed by livestock? These competing metaphors for entrepreneurship symbolize an important debate. The issue is whether, and to what extent, the government should use tax-funded programs to help small businesses.

This debate should be conducted in full light of the facts. However, the terms of the debate are today being altered by politically distorted cost-benefit analyses. These analyses claim that small business assistance programs achieve benefit-to-cost ratios as high as 23 to 1, leading to the obvious conclusion that small business assistance should be continued and expanded. As this paper will show, those ratios are greatly overstated. Entrepreneurship flourishes without government aid -- or, perhaps, even in spite of government aid.

Existing small business assistance programs do not distinguish "entrepreneurs" from "small business owners." The word "entrepreneur" implies something genuinely new and innovative, while a small business could have any degree of innovation -- or none at all. In keeping with their charters, small business assistance programs as currently structured do not distinguish "entrepreneurs" from "non-entrepreneurs."

This essay deals most directly with programs in which service providers assist small businesses. Within the states, Small Business Development Centers have just this mission (SBDC Act 1980). Programs claiming to assist small business are politically popular; however, a sensible evaluation of such programs calls for careful consideration of their specific benefits and costs.

Real Primary Benefits and Illusory Secondary Benefits

The primary measure of benefit for any good or service is the willingness of demanders to pay for it (Mishan 1976, p. 24). When an entrepreneur creates value, that value is reflected in the venture's ability to get people to voluntarily pay for its output. The market provides a stern test for private enterprises, in that they cannot make money by simply claiming, however eloquently, that people are willing to pay a lot for the product. But what if a private enterprise's customer goes on to engage in some new and separate value-creating venture: Is that value a benefit of the original private enterprise? The answer is negative except in a few cases, as outlined below. The benefit of the original venture is reflected in what people will pay for its output -- in whatever way they go on to use that output.

The picture is only slightly different for public enterprises. Their benefit is also measured by the amount demanders would pay for their services, though the measure is more theoretical because customers do not actually pay. Because there is no market test, it is more important to persuade government officials of high value than it is to actually provide high value.

In the case of small business assistance, the persuasion takes the form of arguing for large benefits by claiming some or all of the value creation engaged in by successful clients. Providers of small business assistance have not focused on measuring clients' willingness to pay for service, arguing instead for measuring clients' increased sales and employment after receiving assistance. Such sales dwarf the dollar amounts of willingness to pay for assistance, making the case for large benefit-cost ratios more plausible to legislators.

However, the increased sales and employment indicate a genuine secondary benefit only if the sales and employment are new to the economy. The question needs to be asked: Where do a client's new sales come from? Small businesses compete most directly with other small businesses, so it is quite likely that the new sales come from competitors. If that is so, there has been no increase in the economy's sales, only in the client's sales. As helpful as this may be to the client, it generates no net benefit attributable to the small business assistance program. Universities could as easily claim that the increments to graduates' salaries are net benefits to the economy and entirely attributable to the universities.

Existing evaluations of small business assistance programs (Robinson 1982; Chrisman, Nelson, Hoy and Robinson 1985; and Chrisman, Hoy and Robinson 1987, for example) treat clients' sales increases as entirely new to the economy. They also treat client hirings as job creation, neglecting the possibility that jobs have been lost at non-client firms. To generate sales and jobs that are genuinely new to the American economy, clients must fulfill conditions such as:

• Gaining all their new sales from export demand previously satisfied by a non-American firm, or

• Relieving genuine involuntary unemployment from workers who otherwise would have remain unemployed.

There is little indication that the clients of small business assistance programs ordinarily meet either condition.

Examining existing evaluations of small business assistance programs, we find a near-total disregard of the possibility that clients' new sales and employment are at the expense of non-clients. A study by Chrisman, Nelson, Hoy and Robinson (1985) made no argument for non-canceling secondary benefits, implicitly assuming that there were no revenue losses by non-clients. After the release of the 1985 study, there was a published exchange with Elstrott (Elstrott 1987; Chrisman, Hoy, Robinson and Nelson 1987). In that exchange, a variety of issues in methodology were discussed, but the possibility of canceling effects was not considered. One exchange about control samples -- to determine how much more growth client firms had achieved than a control group -- placed the problem in sharp relief (Elstrott 1987, p. 69; Chrisman, Hoy, Robinson and Nelson 1987, p. 73). If clients gain at the expense of non-clients, their growth rate will be higher. Yet that differential growth rate is exactly the problem: the assistance program may have only redistributed sales rather than creating sales.

Several adjustments to secondary benefits were made in a later and more sophisticated study by Chrisman, Hoy and Robinson (1987, p. 319). The adjustments were made to avoid attributing to an assistance program all the credit for new business, and to account for the counseled ventures that failed. However, there was again no consideration of revenue losses to non-clients.

Turning from sales gains to employment effects, we find that existing evaluations of small business assistance (such as Bent 1993, Chrisman, Nelson, Hoy and Robinson 1985, and Chrisman, Hoy and Robinson 1987) tend to characterize client hirings as job creation, without further qualification. For there to be productive job creation, there must be genuine new sales. If small business assistance programs succeeded in increasing overall hiring without any increase in sales, by definition there would be a decrease in productivity as measured by sales per employee. This serves to emphasize that "number of jobs" is not a reliable indicator of welfare; it is positively but imperfectly correlated with economic well-being.

Therefore, although there may be instances when public assistance programs create jobs (Feller 1988), substantial job creation from small business assistance depends on the creation of new sales, as by exports. Exports are a minor part of small business clients' sales (Burr and Solomon 1977, p. 5; Bent 1993); and the conditions for productive job creation from an assistance program (as opposed to job creation from economy-wide growth) are seldom met.

Overstating Benefits: A Political Tradition

Certainly the practice of overstating benefit-cost ratios was not invented by small business assistance programs. The practice is as old as cost-benefit analysis itself. Exaggerating secondary benefits goes back at least to the time shortly after World War II, when the Army Corps of Engineers and the Bureau of Reclamation were claiming large secondary benefits to justify major public water projects. By overstating secondary benefits, they generally "became adept at producing benefit-cost ratios well in excess of unity for any project that had the support of the senators and representatives who controlled agency budgets" (Frederick 1991, p. 40). As early as the 1960s, the literature in cost-benefit analysis was unanimous in condemning agency practice for inflating benefits or deflating costs, or both (Marshall 1966, quoted in Frederick 1991, p. 40). The Bureau of Reclamation in particular distorted benefits "by claiming secondary benefits, often exceeding the primary benefits, that most analysts would not accept as legitimate project benefits" (Frederick 1991, p. 40).

In view of this research background, ignorance of benefit estimation methods provides little excuse to those who would overstate the benefits of small business assistance programs. The faults now identified in 1980s and 1990s assessments of small business assistance programs have been known for years. Their recognition as errors is commonplace in the literature on cost-benefit analysis.

Why have such obvious errors in cost-benefit analysis of small business assistance gone uncorrected? The answer arises from a consideration of the public choice aspects of small business assistance. The providers of assistance have themselves become a special interest group, complete with their own professional associations and lobbyists (Caruthers 1992). It is in their interest to engage in rent-seeking behavior, to get taxpayers to finance their programs.

Their task is to persuade the public and legislators that they are promoting the public interest rather than their private interests. In one especially strong form, this amounts to the claim that small business assistance is self-financing, in that it generates more than enough tax revenue among successful clients to pay for the program (Chrisman, Hoy and Robinson 1987). As shown above, this claim is based on an assumption of no revenue losses by non-clients, and therefore is invalid. However, the claim of a self-financing program is politically appealing and the claim has not been challenged to date in the journals of entrepreneurship and small business.

The published cost-benefit analyses of small business assistance programs often are originated by those who are paid to deliver the assistance. Because they involve surveys using confidential client lists, they cannot easily be verified or replicated by outside researchers. The resulting studies often are peer-reviewed by others who have an interest in seeing small business assistance programs continue. The inherent conflicts of interest constitute a serious obstacle to their credibility. Even well-constructed studies might be suspect under such conditions. As this paper has shown, the studies are not well constructed, containing errors that bias upward the benefit estimates.

As small business assistance has become a special interest of its own, it has begun to advocate what is good for small business assistance providers but not necessarily for small businesses themselves. This has led to the paradox of a small business assistance program (Small Business Development Centers) being opposed by an advocacy group for small businesses, the National Federation of Independent Business (Caruthers 1992).

Suggestions for Improvement

Those who criticize have an implicit obligation to suggest improvements. In keeping with this spirit, the following recommendations for improved cost-benefit analysis of small business assistance programs are offered:

1. Future cost-benefit analyses should be constructed by independent outside researchers with no stake in increased funding for small business assistance.

2. Future cost-benefit analyses should explicitly separate out the primary and secondary benefit estimates, making clear the assumptions underlying the estimation. Only a portion of client hirings and new sales can be claimed as net benefits to the economy; future studies should make clear exactly what percentage is being claimed.

3. Future evaluations of small business assistance must make clear the rationale for government intervention into this area. Some studies claim a role for intervention because small businesses create jobs; but for this to be a reason for government assistance, it must be true that small business in the absence of government assistance fails to create jobs. No such case has been made in the literature of small business and entrepreneurship.

In this paper, small business assistance programs have been considered to have a positive but overstated effect on the economy. Before concluding, we must consider consider two additional possibilities: that the programs may have no effect at all, or even a negative effect.

First, consider the possibility that programs have no effect. In a market economy, firms expand and contract; jobs are created and destroyed every year (Davis and Haltiwanger 1992). In the process, firms which do not meet consumer preferences well or control costs diligently go out of business. As traumatic as this is to those involved, there are benefits in the workings of this process: When it operates properly, it generates outputs that satisfy consumer wants at reasonable costs. Further, the real probability of traumatic failure may be much less than has previously been thought (Kirchoff 1993).

If shrinking and failing firms seek small business assistance less often than growing firms, then the clients of assistance programs will (as a group) do more hiring than other firms. Therefore assistance programs' clients will increase their hiring more than the economy-wide average even if the assistance program has no effect at all. Thus it is possible that high figures for client hiring are simply a product of the client mix.

Second, consider the possibility that small business assistance may actually have a negative effect on the economy. The taxes used to fund the programs create a burden on the economy, just as the taxes for any program create a burden on the economy. This burden may not be totally offset by benefits, particularly if assistance programs only shuffle sales around among small businesses without creating new sales.

Over the long term, small business assistance rewards those who play the game of receiving assistance. The game is stacked in favor of those with political influence and in favor of those willing to comply with the small business bureaucracy. It is stacked against the independent entrepreneur who has little patience for the political process but a strong desire to serve customers. It is not surprising that clients of some programs are highly educated (Weinstein, Nicholls and Seaton 1992, p. 67), or that they are already relatively large "small businesses" (Elstrott 1987, p. 70). Small business assistance, delivered through a politically dominated system, may well reward the most influential clients rather than the clients who most "need" assistance.

Entrepreneurship is at its best when it is left free to satisfy consumer wants. Public policy must be quite careful when adding new criteria for entrepreneurial success that do not relate directly to customer satisfaction or efficient performance.

Small business assistance programs are today being promoted by an interest group: small business assistance providers. This interest group is using cost-benefit techniques that were discredited years ago when used by a different set of special interests in favor of government water projects.

To return to the opening metaphor, small business assistance providers are trying to persuade the public and policymakers today that entrepreneurship is a frail flower. The providers will water the flower for us, if we open the public treasury to them. But the more true picture is that instead of a flower in a pot they have a wild-growing kudzu vine which will flourish on its own. If we want entrepreneurship to thrive in our economy, the best policy may be simply to get out of the way.

REFERENCES

Bring Home the Bacon

William C. Wood is an economics professor at James Madison University. This article originally appeared as  

"Cost-Benefit Analysis of Small Business Assistance: Do Entrepreneurs Really Need ‘Assisting’?" Journal of Private Enterprise, Vol. 10, No. 1 (Summer 1994): 13-21

His e-mail is woodwc@jmu.edu