4 Comparison and discussion.

As noted in the previous section, no general measure of entrepreneurship contained in a single number can be constructed. As such, researchers need to consider specialized measures which focus on specify dimensions of entrepreneurship. According to our theoretical discussion, these aspects include uncertainty-bearing, management, decision-making, opportunity-seeking, and innovative activity. Obviously, entrepreneurship is difficult to measure and operationalize for empirical work.

In Chapter 3, we have reviewed some of the most commonly used indicators of entrepreneurship. These can roughly be divided into three categories: (1) stock measures (self-employment rates), (2) flow measures (firm and self-employment entry/exit rates); and (3) indirect indicators of entrepreneurship such as, e.g., patents or growth measures.

We have argued that because the stock measures focus on business owners, they are most closely related to Knight's entrepreneur who acts as a residual claimant. Despite the fact that modifications of the self employment rate may provide indications of innovative entrepreneurship, the flow measures seem more appropriate for capturing the process of change embedded in the theories of Schumpeter and Kirzner. Ideally, an indicator of Schumpeterian entrepreneurship should measure all five of his entrepreneurial tasks, but we are forced to settle for less because of lack of data. Self-employment and firm entry and exit rates are used frequently in the literature and with the current data available this is probably as far as we can go.

To illustrate that the distinction between the different empirical measures matters empirically, we have tried to compare entrepreneurial activity across the OECD countries using the different measures. Table 4.1 sums up this information by comparing the cross-country rankings according to the different measures. Although, the rankings are based on estimates and hide important information, Table 4.1 still leads to a number of interesting observations. The overall impression is that the variations in rankings of countries are substantial and cannot be ascribed exclusively to sampling variation.

While Greece is consistently one of the most entrepreneurial countries according to the self-employment rates, it has a low degree of entrepreneurship according to the self-employed transition measures in columns 8 and 9. The reverse applies to, e.g., Germany and the United Kingdom. However, an important observation is that Greece and the United Kingdom have comparable levels of entrepreneurial activity when the self-employed transition measures are computed using the labor force as the denominator.

The stock-based measures in columns 1-5 also show much within variation with respect to the ranking. Examples include Belgium, Portugal, Germany, and Sweden. Most of this within variation can be attributed to differences between the high-skilled self-employment rates (columns 4 and 5) and the overall self-employment rates in columns 1-3. Similarly, the flow-based measures of entrepreneurship also exhibit some within variation.

The TEA indices in columns 12-14 suggest yet another ranking of the OECD countries. For instance, while the United Kingdom is more entrepreneurial than the United States according to the self employment rates, this ranking is reversed when using the GEM measures.

To illustrate the empirical relationships between the different measures, Table 4.2 presents a correlation matrix of the different measures. (1) The fist thing to note is that the correlations within the stock-based (columns 1-5) and flow-based measures (columns 6-11) are all positive and generally strong. However, the correlations between the stock- and flow-based measures taken from the literature are in many cases negative.

This negative correlation may be due to the fact some of the flow measures consider new self-employed relative to the stock of self-employed. That is, the stock of self-employed enters the denominator when calculating the flow measures. This will, as already noted, tend to downgrade countries with a large self-employment rate and thereby induce a negative correlation between these measures. To illustrate this point, we see in Table 4.2 that the correlations between the alternative flow measures, which use the labor force in the denominator (columns 10 and 11), and the stock measures are positive and somewhat high.

This again illustrate why precision is important. The entry and exit rates in columns 8 and 9 indicate the degree of innovative activity within the self-employment sector. The alternative flow measures using the labor force as denominator reflect the degree of activity relative to the potential of the economy and are, in this respect, perhaps a better indicator of the level of, e.g., Schumpeterian entrepreneurship in the economy.

While we have argued that the TEA index combines both the Knightian stock measures and the flow measures, Table 4.2 shows that the empirical correlation between the TEA index and the firm entry/exit rates is the strongest. Not surprisingly, both the NEA and YFEA indices are also strongly correlated with the firm entry/exit rate. No clear picture emerge with respect to the measures of latent and innovative entrepreneurship.

In sum, the correlation matrix in Table 4.2 underscores the importance of distinguishing between the different empirical measures. This in turn requires researchers to be precise about what dimension of entrepreneurship is analyzed in a given context as the choice of indicator will implicitly specify their view of entrepreneurship.

(1) To check the robustness of our results we recalculated the correlation matrix using the rankings of countries instead of the measures. These calculations showed a similar picture.

Jens Iversen (1), Rasmus Jorgensen (2) and Nikolaj Malchow-Moller (3)

(1) Centre for Economic and Business Research (CEBR) and University of Aarhus, Denmark, jiv@cebr.dk

(2) Centre for Economic and Business Research (CEBR) and University of Copenhagen, Denmark

(3) Centre for Economic and Business Research (CEBR) and University of Southern Denmark, Denmark Table 4.1 Ranking of countries according to different measures.

Non- Non-

agricultural agricultural

Self- self- business

employment employment ownership

rate (2002) rate (2002) rate (2002) Australia 14 13 5 Austria 20 24 16 Belgium 13 10 9 Canada 23 21 8 Czech Republic 12 9 -- Denmark 27 25 21 Finland 16 19 15 France 25 26 14 Germany 21 18 18 Greece 2 2 1 Hungary 15 15 -- Iceland 11 11 7 Ireland 10 12 10 Italy 6 5 2 Japan 18 20 19 Korea 4 3 -- Luxembourg 30 28 23 Mexico 3 1 -- Netherlands 19 17 11 New Zealand 8 7 3 Norway 29 29 22 Poland 7 13 -- Portugal 5 6 4 Slovak Republic 26 22 -- Spain 9 8 6 Sweden 24 23 12 Switzerland 22 -- 20 Turkey 1 4 -- UK 17 16 13 USA 28 27 17

Self-

High- employment

skilled rates in

self- high- Firm Firm

employment skilled entry exit