Start-up funding sources and biotechnology firm growth

View the most recent version.

by Brian Cozzarin, Associate Professor, University of Waterloo

Although private investors and government funding agencies have learned that the biotechnology sector requires a funding model different from that of traditional manufacturing, there is a paucity of empirical research investigating the links between characteristics of the funding model and firm performance.  The purpose of this article is to examine which funding sources have the greatest influence on firm growth.

About this article
Context
Methodology
Findings
Summary
References
About the author

About this article

For the purpose of this analysis, data from the Biotechnology Use and Development Surveys (BUDS) 1999 and 2001 were used to examine firms that raised capital and their subsequent performance. The 1999 survey was mailed to the 3,377 Canadian biotechnology firms from selected North American Industry Classification System (NAICS) codes.  After accounting for non-respondents, 358 firms remained¹.  Observations were retrieved based on two criteria.  First, firms must have raised capital in 1999, so that their subsequent performance in 2001 could be assessed.  It turned out that 178 firms attempted to raise capital, but only 138 were successful.  Second, of those 138, only those firms sampled in BUDS 2001 were retained.  After the above selection procedures, 52 firms remained for analysis.

More information about the Biotechnology Use and Development Survey is available here, choose ‘Other reference periods' for 1999 and 2001 questionnaires and metadata.

All financial variables are in millions of 2001 Canadian dollars adjusted by the Industrial Product Price Index for the pharmaceutical industry. All coefficients have been corrected for heteroskedasticity².

Top of Page

Context

Biotechnology is one of the world's fastest growing sectors, expanding almost four times faster than the G-7 average for all sectors.  Canadian biotechnology revenue grew from $1.9 billion in 1999 (25% more than in 1998) to more than $3.6 billion in 2001.  In 2001, there were 375 biotechnology companies operating in 10 Canadian provinces, up from 358 in 1999 (Statistics Canada 2001, McNiven 2001).   Furthermore, in 1999 biotechnology firms raised $2.147 billion; $644.1 million (or 30%) from venture capitalists, $579.7 million (or 27%) from angel investors, $493.8 million (or 23%) from collaborative alliances, $150.3 million (7%) from government sources, $150.3 million (7%) from conventional sources such as banks, $42.9 million (2%) from initial public offerings (IPO), and the remaining $85.9 million (4%) from ‘other' sources (Traore 2005).

Using data from the 1999 and 2001 Biotechnology Use and Development Survey, this article examines the effect of funding sources on firm growth in the Canadian biotechnology sector.

Top of Page

Methodology

R&D investment and capital structure

Biotechnology is a young sector where new firms face unknown markets for their products.  The success of each stage of development, whether R&D, pre-clinical trials, regulatory or production, is subject to great uncertainty and the firm itself has little or no prior track record on which to base forecasts. Explanations for financing preferences of the owner/manager corroborate the static tradeoff and pecking order frameworks of Myers and Majluf (1984).  In this context, ‘pecking order' means that firms exhibit an explicit preference ordering over the set of possible financing sources. 

According to the pecking order hypothesis, internal sources are preferred to debt, and debt is preferred to external equity (Myers and Majluf 1984).  This result is borne out by the above funding data for Canadian biotechnology firms—the bulk of their financing comes from venture capitalists and angel investors; collaborative alliances and conventional funding sources are tied for third place, ‘other' sources are fourth, while initial public offerings are last. Jeng and Wells (2000) also predict that such a sector is best suited to venture capital financing.

Empirical model

To estimate the effect of funding sources on firm growth the following two regression equations were modelled:

The dependent variables for the regressions were: revenue growth and the ratio of revenue growth to R&D capital (hereafter denoted as revenue growth/R&D capital).  The revenue growth variable was constructed by subtracting total revenue in 1999 from 2001.  Since R&D capital is not directly observable, it was constructed via three methods using the values for total R&D spending in each year reported in BUDS 1999 and 2001. The methods included summation of total R&D spending, adjustment by straight-line depreciation and adjustment by double-declining balance depreciation (see Three methods for calculating R&D capital).  Although the summation method was adopted in the estimation of R&D capital and works well, the other methods are conceptually more sound. 

Three methods for calculating R&D capital

In the summation method, the variable was calculated simply by summing the values of total R&D spending from 1998 to 2001 without considering depreciation. In the straight-line depreciation method, it was assumed that the salvage value of total R&D capital available in each year was 10% of the total R&D spending in each year, as there was no direct information available.  Usually the salvage value of R&D capital ranges from 10% to 20% of R&D capital in R&D intensive firms but the results would not differ significantly if it were set to 20% (Hall et al. 1998).

For double-declining balance depreciation, it was assumed that the salvage value of R&D capital available in each year was 10% of the total R&D spending in each year, and that the estimated useful life of R&D capital was only four years.  The value of total R&D capital available in 2001 was obtained by summing the values of total R&D spending of the firms in each year after depreciation adjustment in 2001.

The independent variables included six firm-level funding sources in 1999—angel investors or family or friends; government loans or grants; venture capital; conventional sources such as banks and/or trust companies; initial public offering (IPO); and collaborative alliance.  They also included the total amount of capital raised in 1999, and ownership type (public or private).  Biotechnology products and processes were identified by four different stages of development (ranging from least developed to fully commercialized): R&D; pre-clinical trials or confined field trials; regulatory phase/unconfined release assessment; and approved or on the market or in production³.  If a firm had an IPO in any year between its establishment year and 2001, then the variable ‘Public' was set to one, otherwise it was zero. 

Top of Page

Findings

For the first regression model (revenue growth), the results show that of the six funding sources only the coefficients on angel, venture, and conventional capital are significant (Table 1).  Conventional capital has the greatest impact on revenue growth between 1999 and 2001.  Angel capital has the second greatest impact, while venture capital is third.  The results also show that of the four stages of product or process development, only the coefficient on R&D is not significant.  The coefficients on the remaining stages (i.e., pre-clinical, regulatory, and on the market) are all significant at the 10% level and have the expected signs.  Likewise the results for total capital raised in FY99 and whether the firm is publicly traded are positive and statistically significant.

Table 1
Ordinary least squares (OLS) regression results (dependent variable is firm RevenueGrowth)

Table 2 presents the regression results for the second model (revenue growth/R&D capital).  Of the six funding sources only the coefficients on angel, venture, and conventional capital are significant and have positive signs.  If we exclude the summation method (because it is less conceptually sound), in order of importance it is conventional capital, angel capital and venture capital that contribute the most to firm growth. 

Table 2 Ordinary least squares (OLS) regression results (dependent variable is firm RevenueGrowth/R&DCapital)

Of the four stages of product or process development only the coefficient on R&D is not significant under all three methods of R&D capital. The coefficients on pre-clinical, regulatory, and on the market are all significant and have positive signs.  In order of magnitude, the on the market/in production stage, followed by the regulatory stage and finally the pre-clinical stage affects sales growth the most.  This finding seems to make more sense theoretically than the results in Table 2, since products and/or processes that are closer to the market/in production stage should contribute more to firm growth.  The coefficients on total capital raised in 1999 and public or private ownership are significant and have positive signs.

Top of Page

Summary

The results indicate that of the funding sources, only angel, venture, and conventional capital have contributed significantly to R&D capital formation and revenue growth.  Conversely, the contribution of funding from government, IPO, and alliance capital sources were found to be less important for the given sample of biotechnology firms.  There are counter intuitive results on the importance of conventional capital for firm growth rates.  It was expected that venture capital would be most important; however, it may be that older firms with more mature products or products ready for market are chosen by banks.  In that case, perhaps it makes sense that bank capital is correlated with higher growth rates. 

The results provide insights for policy makers and investors (both private and public)—angel capital, venture capital and conventional capital all have a significant role to play in biotechnology firm growth.  One caveat is that venture capitalists, angel investors and banks may use superior selection criteria to pick prospective start-ups.  It may also be that once chosen, these firms receive critical managerial input (not available from government, alliance capital or an IPO) to the new venture which accounts for their success.  These caveats can only be answered with further research.

Top of Page

References

Hall, B. H., J. Mairesse, L. Branstetter and B. Crepon (1998). Does cash flow cause investment and R&D: an exploration using panel data for French, Japanese, and United States scientific firms. University of California at Berkeley, Working Paper, January.

Jeng, L.A. and P.C. Wells (2000). The determinants of venture capital funding: evidence across countries. Journal of Corporate Finance, 6(3): 241-289.

Jensen, M.C.  (1993). Presidential address: the modern industrial revolution, exit and failure of internal control systems. Journal of Finance, 48(3): 831-880.

McNiven, C. (2001). Practices and activities of Canadian biotechnology firms: Results from the biotechnology use and development survey 1999. Statistics Canada, Catalogue no. 88F0006XIE, no. 011, August.

Modigliani, F. and Miller, M.H.  (1959). The cost of capital, corporation finance and the theory of investment. The American Economic Review, 49(4): 655-669.

Myers, S.C. and Majluf, N.S. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics, 13: 187-221.

Statistics Canada (2001). Biotechnology use and development survey 2001. Science, Innovation and Electronic Information Division.

Traore, N. (2005). Access to financing capital by Canadian innovative biotechnology firms. Statistics Canada, Catalogue no. 88F0006XIE, no. 010, April.

Top of Page

About the author

Brian Cozzarin is an Associate Professor with the University of Waterloo. For more information about this article, please contact sieidinfo@statcan.gc.ca.

Notes

1. Firms with fewer than five employees and less than $100,000 in R&D expenditures were excluded from the sample.
2. Heteroskedasticity is a common problem with survey data.  It means that when survey responses are taken across different firms the resulting data appear to be drawn from different distributions (as opposed to all responses being drawn from the same distribution, which is usually assumed to be normal).  The regression estimates in Tables 1 and 2 have been corrected for this problem.
3. The stages-of-development variables were the total number of products or processes at each stage of development (R&D, pre-clinical, regulatory, and on the market) for all firms in 1999.