From YalePressWiki

[edit] Appendix

Economists like to do two things: theorize and measure. The first is a job increasingly left to the most mathematically inclined or gifted and tends to be the most highly rewarded in the profession. Rare is the Nobel Prize winner who has demonstrated or explained what may seem like an elementary proposition without using some elegant mathematical tools. Although sometime derided by economists who don’t have these skills and who believe that it is more important to focus on institutions, culture, and history of economies, the application of mathematics to economics is also useful because it can reveal some surprising and important insights.

Most economists who do not theorize do “empirical work”—testing theories through the use of (increasingly sophisticated) statistical techniques. We have already discussed the limits of such analysis when it comes to assessing the causal factors driving economic growth. Nonetheless, we concede that propositions about economic growth will not be widely accepted within the profession until there is at least some degree of statistical validation (although those outside the profession generally do not apply such exacting standards). For example, one would like to be able test various theories related to our four categories of capitalism. We would like to answer questions such as: Is it really true that oligarchic economies grow less rapidly than other economies? And at what point do state-guided and bureaucratic systems, which can generate growth for substantial periods of time, run out of gas? Is it inevitable that these economies reach that point and, if not, why not? And, as we asked earlier, what mix of entrepreneurial firms and large firms—perhaps measured by the relative output of each type—is likely to maximize growth? Does the answer depend on a country’s stage of economic development (measured by its per capita income)?

These are among the questions one would like to examine if only one could measure and/or classify the extent to which capitalist economies fall into one or more of the four categories we have outlined. Alas, that job is difficult, conceivably impossible, but in any event, it is one that we do not undertake here. We trust that if our arguments resonate with at least some in the profession, others more expert than we will expend the effort and resources to collect and analyze the relevant data. But if we are correct that most economies exhibit some combination of several or even all of the categories of capitalism, then it may be a fool’s errand to put entire economies into one box or the other. Instead, it would be more revealing to be able to track the mix of the different types of capitalism present in any economy, perhaps by value added generated by sector or by firms classified into one of the four categories.

Even this may be difficult, if not impossible, to do within any reasonable length of time. As we mentioned in chapter 5, the World Bank has just in the past two years assembled a set of indicators to measure the strength of legal systems and regulatory burdens, more than a decade after Peruvian economist Hernando De Soto popularized the notion that such factors are central to encouraging or handicapping growth in developing countries. But since the World Bank is unable collect such data for earlier periods, it will be many years before it has a series of data spanning a period of sufficient length to be useful for statistical analysis. The same would be true of any effort to collect data on the shares of different economies characterized by the various forms of capitalism we have outlined.

Until then, policy makers and economists who agree with us that our typology is useful will have to be satisfied with impressionistic views of different economies, supplemented by various nuggets of hard statistical data. Such data may be hard to come by for state-guided economies since, as we have noted, state guidance comes in many forms and is often too subtle for outside observers to detect, let alone measure. Data measuring or indicating oligarchic capitalism are a bit easier to assemble. Again, as we noted earlier, such economies tend to be characterized by high income inequality, the presence of a substantial underground (that is, informal) sector, and an abundance of corruption. Income inequality can readily be measured; corruption and the size of the informal sector can be approximated.

[edit] Testing Big-Firm Theories

It is tempting to conclude that measurement of big-firm capitalism is easier. For example, why not look to the share of value added or employment by the largest firms in the economy, say the top 100 or 500? The drawback to this deceptively simple approach is that country size matters greatly. In small economies, relatively few firms may account for a significant share of value added or employment, but the firms may be small by international standards. In other words, the firms in such economies may be competing in oligopolistic settings but still may be small in comparison to the large enterprises in bigger and richer countries. On the other hand, in such countries as the United States, India, and China, the largest firms might account for a relatively small share of total value added or employment but still would be characterized, on the basis of their size alone, as bureaucratic. Depending on the industry in which they do business, they may or may not be competing in oligopolistic markets. For these reasons, cross-country comparisons of the shares of value added by a fixed number of companies are likely not to be very revealing and, worse, could even be misleading. Instead, what may be more instructive is the turnover or churn among the top companies in an economy. In a dynamic big-firm economy, one would expect to see a reasonable amount of shifting of rankings among the largest companies over time, as some firms grow rapidly while others recede. In less dynamic settings, there would be little churn.

Table 19 illustrates the kind of data we have in mind, covering just the top twenty U.S. companies, by revenue, since 1955, with snapshots every ten years. Interestingly, although General Motors and Exxon Mobil topped the list for most of that period, by 2005 both had been eclipsed by Wal-Mart, a company that ranked fourth in 1995 and didn’t even make the top twenty in 1975. The table shows an especially large movement of rankings among companies below the top five throughout the period. Expanding the focus to a larger group of companies reinforces the message in the table. According to one calculation, only about half of the one hundred largest manufacturing and industrial firms in the United States in the 1970s survived until the year 2000. The rest disappeared via takeovers or bankruptcies (Micklethwait and Woolridge, 2003, 130–31).

We do not know what a similar chart or data on the status of the largest companies over several decades would look like for Europe or Japan, or for other economies. Our educated guess, however, is that churning of the rankings of large companies in rapidly growing economies generally would be more evident than in more slowly growing ones. In any event, our impressionistic view of the U.S. data indicates that within the big-firm sector, the U.S. economy has been reasonably, and perhaps even remarkably, dynamic over the past five decades (if not more). The U.S. economy looks even more dynamic once account is taken of its entrepreneurial sector.

[edit] Entrepreneurship Data and Theory

And so we come to the polar opposite form of capitalism, calling for measures or indicators of entrepreneurial activity. Again, for reasons explained throughout the book, it would be a mistake to measure the vitality or the degree of entrepreneurial capitalism simply by counting the number of entrepreneurs, or small businesses, or those who identify themselves as self-employed. Such a measure is much too broad for our purposes since it would probably include a far greater number of replicative entrepreneurs than innovative entrepreneurs, the number of primary interest here. In principle, surveys can be designed to collect data on the number of innovative entrepreneurs and the growth in that number, and we describe one such effort at the end of this appendix. Given the extensive data collection on firms available in Scandinavian countries and in France, we suspect that such information already exists or could be generated for those countries without overwhelming difficulty. We doubt, however, that such data exist for the rest of the world.

[edit] Entrepreneurship Data

The entrepreneurship data that do exist are spotty and, unfortunately, of limited value for purposes of testing the hypotheses advanced in this book.^[1] Nonetheless, there is a wealth of information available related to entrepreneurship that serves other useful purposes. We are able, for example, to distinguish between the numbers of employed versus self-employed individuals, which may be relevant for understanding social behavior and learning about responses to regulatory or other incentives or disincentives. Similarly, we can use data on new firm formation to understand the effects of the regulatory and institutional climate in a given country for a particular type of entrepreneurial activity.

Self-employment data. In the United States, the Current Population Survey (CPS) and Census of Population provide estimates of the number of self-employed business owners annually and every decade, respectively. Data from these household surveys allow us to estimate the number of self-employed individuals at a particular point in time and to track changes in the number of self-employed individuals over time. The new Kauffman Index of Entrepreneurial Activity, compiled by Professor Robert Fairlie (of the University of California at Santa Cruz) takes these analyses one step further, using the matched basic monthly files from the CPS to learn about trends in the rate of business creation at a national level.

It is inherently very difficult, however, to find comparable self-employment data sources for cross-country comparisons. In addition to cultural differences that influence survey responses, definitions of self-employment may vary from country to country. There is variation among nations in reporting unpaid family workers as self-employed (likely a function of the particular tax regime and welfare system), and not all countries consider owners of larger businesses to be self-employed.

The Organization for Economic Cooperation and Development (OECD), an organization of largely rich countries (and thirty in all) has made considerable efforts to create comparable cross-country self-employment data. The OECD Labour Force Statistics, generally based on household labor force surveys, provide self-employment data for all OECD member countries. As the majority of these countries (with the exception of Iceland, Mexico, and Turkey) use the International Labor Organization Guidelines definition of self-employment for measuring employment, most of these statistics are comparable across countries. The population statistics that serve as the denominator when calculating these self-employment rates are from a mixture of labor force surveys, administrative records, and population censuses.

Unfortunately, neither the U.S. datasets nor the OECD data can be used to test the theories presented in this book. It is impossible to distinguish the portion of the selfemployed population that started businesses for lack of better options for work from those who are taking advantage of an entrepreneurial opportunity. We cannot differentiate between individuals who seek business growth versus those who only have an interest in maintaining their market share. And there is certainly no way to identify those self-employed individuals who are creating truly innovative new entities.

New firm formation. The United States also maintains data on new firm formation and on the number of small businesses in the country. Using business tax returns and administrative records, the U.S. Census Bureau maintains various programs that both extract relevant data from these files and use them as sampling frames for surveys of businesses.

These data are, however, of limited use for the study of entrepreneurship and certainly do not provide additional information for investigations of innovation and growth. Entrepreneurial activity may be overstated in those datasets that include all firms with receipts of $1,000 or more, which may include side or casual businesses. And efforts to use small businesses as a proxy for entrepreneurs may also overstate the number of new businesses. The Small Business Administration’s definition of small businesses as those with fewer than five hundred employees means that this classification includes firms that may be much larger than what we think of as entrepreneurial companies and that may have been in existence for decades.

International data, global entrepreneurship monitor. The Global Entrepreneurship Monitor (GEM) is, in fact, designed to provide international data that would answer some of the important questions raised in the text. This survey of approximately forty countries is intended to provide comparative entrepreneurship data that include measures of innovation and distinguish between opportunity entrepreneurs and necessity entrepreneurs. GEM presents a Total Entrepreneurial Activity (TEA) index that measures both nascent and early-stage entrepreneurship, capturing individuals between the ages of eighteen and sixty-four who are involved in either the start-up phase or manage a business that is less than forty-two months old. In addition, GEM seeks to measure innovation by asking respondents if their product or service is completely new, and the dataset includes both a TEA-Opportunity and a TEA-Necessity measure, based on questions regarding the reasons for entrepreneurs’ decision to start businesses. With this information, the principal investigators have concluded that there is a positive relation between entrepreneurial activity and economic growth.

Although GEM begins to identify the questions that must be part of future data collection efforts, methodological problems with the GEM data and inconsistent results over time suggest that this dataset is not appropriate for investigating the questions raised in the text in a rigorous and meaningful way. It is not clear that GEM’s definition of entrepreneurial activity is sufficiently nuanced for scientific inquiry, and it is possible that interpretations of this definition may vary significantly across countries. Furthermore, while the response rate for these surveys is within the operational range for commercial marketing surveys, it is not necessarily high enough for academic analysis.

Inconsistencies in GEM survey results also cast doubt on the credibility of these data for academic research. The significant changes in the entrepreneurship rate for a single country from year to year, and conflicts between the GEM findings for the United States and U.S. Census Bureau data for the same time periods suggest that the measure is problematic for this type of research.

[edit] Improving Data Collection on Entrepreneurial Activity

The Kauffman Foundation is making significant efforts to improve the state of data collection related to entrepreneurial activity in the United States and in the world. Since we see the distinction between replicative and innovative entrepreneurship as fundamental to relating entrepreneurship and growth, several of the foundation’s data collection initiatives are explicitly intended to bring greater clarity to this contrast.

First, the foundation is supporting a National Academies study of U.S. federal business statistics. The Committee on National Statistics (CNSTAT) has established a panel of experts to review existing data sources in light of researchers’ need for better measurement of younger and smaller businesses, their evolution over time, their economic performance, and their role in the larger economy. The panel’s final report, to be completed in 2007, will present recommendations for improving the sources and accessibility of data on high-growth firms. In addition, it will suggest new data collection efforts that will give researchers better information for measuring and analyzing the early life cycle dynamics of businesses and for evaluating theories of business formation, selection, and growth.

Similar efforts will need to take place at the international level in order for meaningful comparison to take place. The foundation is sponsoring a study through OECD that will begin to identify needs and make recommendations for international data collection. This assessment of existing data sources and identification of those entrepreneurship- related questions for which there are no international data available will call attention to the significant gaps in data sources for comparative analysis of high-growth companies and will provide a roadmap for future work.

As both of these studies are now in their early stages, the foundation has also taken steps to start collecting the type of data that will allow for greater insight into the difference between replicative and innovative entrepreneurship in the United States. The foundation is funding the Kauffman Firm Survey, a multiyear longitudinal study of new businesses started in the United States in 2004. An oversample of high-technology businesses is expected to yield a greater number of innovative firms than most business surveys, and a focus on the financial development of the firm will offer researchers new insight into new business financing and growth.

While the foundation can begin to identify the gaps in existing data sources and fund data collection efforts that further efforts to answer the important questions raised in the text, the comprehensive data collection efforts that are truly needed will require the support of multinational organizations like the United Nations or the World Bank. These organizations alone have the broad resources and the network of relationships that are required for this vitally important and incredibly difficult task.

[edit] Financial Data on Entrepreneurship

Some analysts find it useful to measure entrepreneurship by the financing— specifically third-party equity financing, venture, or angel capital—that supports it. Information about venture capital is available for the United States and Europe and, to a more limited degree, for other countries. Little or nothing is known about the magnitudes of angel investing among countries, though some effort is being expended on collection of such data in the United States.^[2]

There are two caveats that must be mentioned regarding the venture figures, however. One is that since the bursting of the dot-com bubble in 2000, venture funds in the United States seem to be going predominantly to existing companies rather than to support firms that are in their “early stages.” So even venture money raised or invested may no longer be a good indicator of the risk-taking associated with start-up companies. For the United States, angel money invested now is likely to be a better indicator of entrepreneurial activity (but so far comprehensive data for this measure are not yet available). A second caveat is that although venture money can be important—Paul Gompers and Josh Lerner of Harvard Business School have argued that it is critical for many “innovative” companies—many innovative firms get their start without it (Gompers and Lerner, 1999).

Figure 7. U.S. Venture Capital Investment by Year. Source: PricewaterhouseCoopers/Thomson Venture Economics/ National Venture Capital Association MoneyTreeTM Survey.

One financial indicator of entrepreneurial activity that does not have these difficulties is the number of companies raising money through initial public offerings (IPOs), whether on local or foreign stock exchanges. But even IPO data have their limits because they do not include many privately held, but rapidly growing, companies in the United States and elsewhere around the world.

Furthermore, money going into venture (angel) funds and activity in the IPO market reflects only the willingness of particular types of investors to take on higher risks. There certainly is much more variation in these measures than in underlying entrepreneurial activity. Figure 7 displays the gross flows into venture funds in the United States since 1980. The graph shows the peak in 2000 at roughly $100 billion, with flows dropping dramatically thereafter and recovering modestly in 2003–4. It is difficult to believe that innovative entrepreneurial activity—or the kind that is most attractive to venture (and angel) capital—has varied, and fallen, as much during this period.

Meanwhile, large companies in the United States do not appear to be abandoning the financing of early-stage companies via venture funds and IPOs.^[3] It is true that this sort of financing may be less supportive of entrepreneurial activity in the long run since the less structured environment that leads to successful start-ups may often clash with the

more bureaucratic structures and, more important, the incentive systems of the large companies that acquire them. Nevertheless, the existence of an active market in earlystage companies tempers the view that just because venture or IPO funding may have declined, entrepreneurial activity by implication must have dropped along with it.

All of this is not to dismiss the relevance of venture and IPO funding to entrepreneurial economies. As we discussed in chapter 5, as economies mature, the depth and complexity of an economy’s formal financial system appears to play an essential role in enhancing entrepreneurship. Thus, the fact that the early-stage capital market— whether in the form of angel groups, venture funds, initial public offerings, or the outright purchase of early-stage firms by larger enterprises—is more developed in the United States than in other countries suggests strongly that it is more entrepreneurial than the others.

[edit] Indicators to Avoid

Finally, it is important to know what measures not to use as indicators of the degree of entrepreneurial activity in an economy. Here we have in mind various technological indicators, such as number of patents issued by the government or numbers of scientists and engineers.^[4] Both of these measures may shed light on how innovative an economy may be, but not necessarily on how entrepreneurial it is. Patents can be obtained—and in the United States this seems increasingly easy to do, as we discussed in chapter 8—but a patent may sit on the shelf for years until some entrepreneur, in a current company or forming a new one, actually licenses or purchases it and puts it to use. Similarly, economies such those as India and China are turning out ever greater numbers of engineers, but many (perhaps most) of these highly trained individuals are going to work in existing bureaucratic enterprises, to tinker with incremental refinements, rather than to develop and take to the marketplace the truly radical innovations that characterize entrepreneurial economies. Japan is the world leader in the number of patents granted, but even the Japanese would concede that their economy is not a model of entrepreneurial capitalism. Rather, Japanese entrepreneurship essentially is synonymous with small retail stores, which are ways for former managers in large Japanese companies to finish their working careers and to supplement their retirement income.

[edit] Summary

In sum, existing data do not permit the kind of econometric testing of the various hypotheses we have advanced in this book. But we believe that the broad outline of our argument—and, specifically, our distinctions among the different types of capitalism—will strike a chord among many readers. It is sometimes true that informal evidence available to the trained eye is just as revealing as a mountain of numbers.

[edit] Notes

We are grateful for the extraordinarily able assistance of Alyse Freilich in preparing this appendix.

1. ↑ For similar complaints about the adequacy or usefulness of existing data on entrepreneurship, see Audretsch, Keilbach, and Lehman, 2006, 7–9.
2. ↑ That effort has begun with the formation of the Angel Capital Association, an organization that, as of June 2006, had approximately 130 angel investing groups as members. The Ewing Marion Kauffman Foundation provided the initial funds for the organization.
3. ↑ For an extensive study of large company acquisitions of newer, smaller enterprises, see Christiansen and Raynor, 2003.
4. ↑ Gross patent counts have several shortcomings even as measures of innovation. Patents do not include innovations that are not formally patented but nonetheless entail important breakthroughs in know-how, especially in production or specific products. These innovations may be protected by “trade secret” law, but government agencies (or even their private equivalents) do not, and cannot, count trade secrets. Another limitation is that mere counts of patents do not reflect their importance. One patent may generate billions of dollars in revenue, while another may lie in a drawer and never be used. There is no good way to distinguish between the two in official government statistics that count the number of patents issued.

This site made possible through the generous support of the Ewing Marion Kauffman Foundation