Crowding Out Private Equity: Canadian Evidence∗

Douglas J. Cumming School of Business University of Alberta Edmonton, Alberta Canada T6G 2R6 Tel: (780) 492-0678 Fax: (780) 492-3325 E-mail: [email protected] Http://www.bus.ualberta.ca/dcumming Jeffrey G. MacIntosh Toronto Stock Exchange Professor of Capital Markets Faculty of Law University of Toronto 78 Queen’s Park Toronto, Ontario Canada M5S 2C5 Tel: (416) 978-5785 Fax: (416) 978-6020 E-mail: [email protected]

First Draft: September 2000 This Draft: September 2003

∗ We are grateful for comments from Mark Huson, Aditya Kaul, Janet Payne, Corrine Sellars, Wolfgang Stummer, Ralph Winter and the seminar participants at the Canadian Law and Economics Association 13th Annual Conference (Toronto, September 2001), the Eastern Finance Association (Baltimore, April 2002), the Academy of Entrepreneurial Finance and Business Ventures Conference (New York, April 2002), the Tilburg University Conference on Regulatory Competition (The Netherlands, September 2001), the Northern Finance Association (Banff, September 2002), the Financial Management Association (San Antonio, October 2002), and the CESifo Conference on Venture Capital and Public Policy (Munich, November 2002). The Schulich School of Business National Research Program in Financial Services and Public Policy provided generous financial support. In preparing the grant application, preliminary versions of the paper with all the figures that are presented in this current draft (but based on data only up to 1999) were distributed to various Canadian academics for comments in the fall of 2000 and the spring of 2001. We also owe thanks for the (anonymous) comments received through the Schulich application process. This paper is scheduled for presentation at the American Law and Economics Association Annual Conference (Toronto, September 2003).

Crowding Out Private Equity: Canadian Evidence

Abstract In this paper, we examine a Canadian tax-driven venture capital vehicle known as the “Labour Sponsored Venture Capital Corporation” (LSVCC). As a theoretical matter, we suggest that the LSVCCs can be expected to have higher agency costs and lower profitability than private venture capital funds. We present data that is consistent with this view. The central question that we analyze, however, is whether the tax advantages conferred on LSVCCs have resulted in LSVCCs “crowding out”, or displacing other types of venture capital funds. Empirical analysis of our data (which covers the 1977-2001 period) is highly consistent with crowding out. The data suggest that crowding out has been sufficiently energetic as to lead to a reduction in the aggregate pool of venture capital in Canada, frustrating one of the key governmental goals underlying the LSVCC programs; namely, the expansion of the aggregate pool of capital. In the course of our analysis, we confirm the importance of macroeconomic factors (the performance of the stock market, real interest rates, and changes in real gross domestic product) in affecting the supply of and demand for venture capital.

We also generate evidence that is consistent with the proposition that

entrepreneurs in the market for venture capital prefer to incorporate their businesses federally, rather than provincially.

Key words: Venture capital cycle; Government sponsorship; Tax; Crowding out JEL classification: G24, G28, G32, G38, K22

1. Introduction The importance of venture capitalists in providing funds for small and medium sized enterprises (“SMEs”) has been well documented (see, e.g., Barry et al., 1990; Berglof, 1994; Black and Gilson, 1998; Gompers, 1997; Gompers and Lerner, 1999; Hellmann, 1998; Hellmann and Puri, 2000a; Kaplan and Stromberg, 2000; Kirilenko, 2001; Kortum and Lerner, 2000; Lin and Smith, 1997; MacDonald, 1992; Megginson and Weiss, 1991; among others). Venture capitalists play a significant role in enhancing the value of their entrepreneurial investments (e.g., Sahlman, 1990; Gompers and Lerner, 1999; Hellmann and Puri, 2000b). Venture capitalists affect the product market strategies and outcomes of entrepreneurial firms (Hellmann and Puri, 2000a) and the variability of equity financing can affect the performance of entrepreneurial ventures (Lerner and Tsai, 1999). In a collection of their groundbreaking work, Gompers and Lerner (1999) document the existence of a “venture capital cycle” in the United States, and the factors that affect venture capital fundraising, investing and exiting. In response to the perceived importance of venture capital to the funding of entrepreneurial firms, many governments have mounted programs that seek to foster venture capital financing. Such programs have been the subject of previous scholarly examination; see, e.g., Cressy (1996, 2002), DeMeza (2002), Keuschnigg and Nielsen (2002a). Lerner (1999) evaluates the success of the U.S. government’s SBIR program. Lerner (1999, 2002) and Gompers and Lerner (2001) discuss the appropriate role of governments in private equity markets, and suggest that government programs ought to complement – and not compete with – private venture capital investments. Recent advances in the literature have considered the role of informed versus uninformed venture capitalists and government policy towards venture capital (Kanniainen and Keuschnigg, 2000, 2001; Keuschnigg, 2002; Keuschnigg and Nielsen, 2001, 2002a,b). In this paper, we examine a particular form of government subsidization of venture capital, the Canadian “Labour Sponsored Venture Capital Corporation” (LSVCC). In contradistinction to private funds, LSVCCs are set up as corporations, and may be incorporated in a variety of Canadian provinces or federally. They must have a union sponsor, which must by statute control the fund, but which has no ownership interest1 in the fund, and will merely ‘rent’ its name to the fund in return for a fixed fee or a small percentage of net asset value (Osborne and Sandler, 1998). In many cases, the impetus to establish 1

Formally, the union will hold a class of shares in the fund that does not receive dividends and is not entitled to share in the assets upon winding up, but which will be entitled to elect a majority of the directors.

2 the fund comes not from the union sponsor, but from the company that will manage the fund (Osborne and Sandler, 1998). Only individual investors may invest in a LSVCC, and there is typically either no minimum (or a modest minimum) investment, although investors must typically remain invested for a period of eight years. Contributions are encouraged by extremely generous tax benefits, consisting of a combination of provincial and federal tax credits and deductions. These benefits yield investors an immediate 30% return on investment in the year of the contribution. The LSVCC concept was first introduced in the province of Québec in 1983. In 1985, the federal government matched the Québec tax credit and provided deductability of the amount invested for investments up to a certain ceiling. In the late 1980’s and early 1990’s, most of the other Canadian provinces followed suit and introduced their own LSVCC programs, which, like the Québec program, receive the benefit of both provincial and federal tax credits and federal deductability. The federal government also introduced legislation allowing for the creation of federal LSVCCs. These funds may operate in any province that specifically permits the operation of a federal LSVCC. Since the early 1990’s, LSVCCs have become the dominant form of venture capital organization in Canada, now controlling over half of the total pool of Canadian venture capital. The focus of this paper is on whether the LSVCCs have “crowded out”, or displaced other forms of venture capital organizations, and on whether the various government sponsors have achieved their goal of expanding the Canadian pool of venture capital. At a theoretical level, we suggest that LSVCCs are an inferior organizational form that exhibits high agency costs and low returns. The generous tax subsidies underlying the LSVCC programs lower the LSVCCs’ required rate of return, allowing LSVCCs to out-bid other types of funds (even those with tax-exempt investors), drive up deal prices and lower returns in the market. If institutional investors are risk averse and commit capital prior to knowing the increase in LSVCC fundraising in any given year, then institutional investors overestimate the extent LSVCC funding, and reduce their commitments to private venture capital funds. As a result, government sponsorship through the particular tax breaks exclusively to LSVCCs may paradoxically decrease the overall supply of venture funds. We find evidence that is highly consistent with this view. We build upon previous research on government sponsorship of venture capital and the demand and supply for venture capital in a number of interrelated ways. First, using a Canadian data set covering the period 1977-2001, we use simultaneous equations econometric methodology to estimate the Canadian supply and demand equations for venture capital (and hence the determinants of both the demand for, and

3 supply of venture capital). We examine whether these equations differ across different geographic regions within Canada; namely, British Columbia, Alberta, Saskatchewan and Manitoba (jointly), Ontario, Québec, and the Maritime provinces (jointly). We test the robustness of our estimates by separately estimating the equations for start-up, expansion, buyout, and turnaround stages of financing. We also use a bootstrap experiment to test the efficiency of the estimation procedure in a finite sample. Second, in estimating the supply equation, we focus on the determinants of venture capital investing (i.e., capital flows from venture capitalists to entrepreneurial firms), rather than venture capital fundraising (i.e., capital flows from institutional investors to venture capitalists, the specification used by Gompers and Lerner (1998) and Jeng and Wells (2000)). This difference is motivated by the unique characteristics of the LSVCCs, and in particular, the tendency of the LSVCCs to invest a non-trivial portion of their capital in relatively low risk instruments such as treasury bills and short-term corporate debt obligations, and in some cases to maintain a significant part of their invested portfolios in publicly traded securities, rather than private entrepreneurial companies. In estimating supply and demand equations, there is some overlap in independent variables. For example, we hypothesize that interest rates, real gross domestic product, stock market performance, and market trends over time are likely to affect both supply and demand.

Our empirical results offer

confirmation of this hypothesis. On the demand side, we test whether changes in the number of both provincially and federally incorporated firms affect the demand for venture capital. We find that changes in the number of federally incorporated firms are significantly and positively related to demand, while changes in the number of provincially incorporated firms are not. This suggests that entrepreneurs seeking venture capital tend to incorporate their firms federally, rather than provincially. On the supply side, in furtherance of the central thrust of this paper, we use dummy variables to determine whether the introduction of LSVCC legislation in various Canadian jurisdictions between 1983 and 1994 crowded out other types of funds and/or affected the size of the overall pool of Canadian venture capital. We find evidence not merely that LSVCCs have crowded out other Canadian funds, but that they have led to a reduction in the overall size of the venture capital pool. This evidence, which is strongly at odds with the widely held perception that LSVCCs have greatly added to the pool of venture capital in Canada, supports the theoretical predictions of Kanniainen and Keuschnigg (2000, 2001),

4 Keuschnigg (2002), and Keuschnigg and Nielsen (2001, 2002a,b) with respect to public policy towards venture capital. Our results are somewhat different from those in Leleux and Surlemont’s (2003) study of public venture capital programs in Europe over the 1990-1996 period. In contrast to our results, Leleux and Surlemont find that European public venture capital programs have led to an increase in venture capital funding in Europe. The difference between Leleux and Surlemont’s European results and our Canadian results may well be attributable to the different nature and method of delivery of the government assistance, the different institutional settings, the comparative maturities of the private funding sectors in the different economic settings, different balances between different types of funds, and/or other factors. An examination of these differences may be able to shed new light on when government assistance to venture capital is likely to succeed, and when it is not. We thus take care to describe in some detail both the nature of the assistance and the legal and institutional setting in which it occurs. While we focus on a unique institutional setting, we nonetheless believe that our results have general implications for the efficacy of focused government subsidization of venture capital. Tax breaks to particular types of venture capital funds may exacerbate, not mitigate, capital gaps. We note that venture capital organizations similar to Canadian LSVCCs have been introduced in other countries, such as the U.K.2 Further research is warranted. This paper is organized as follows. Section 2 provides a description of the Canadian venture capital market and Canadian venture capital data over the 1977-2001 period. Section 3 briefly discusses the structure and governance of LSVCCs, and outlines the legislation governing their operation. Section 4 describes LSVCC performance. Descriptive statistics on the Canadian venture capital industry are provided in section 5. We present data on the amount of venture capital financing in Canada, both before and after the introduction of LSVCCs. In section 6 we provide econometric estimates of aggregate demand and supply equations for venture capital in the Canadian provinces. A bootstrap experiment is presented in the Appendix to illustrate robustness and the suitability of the econometric specifications. Section 7 provides a summary and conclusion.

2

The ‘Venture Capital Trust’ was introduced in the U.K. in 1997. We believe it would be fruitful to conduct a study of Venture Capital Trusts along the same lines as in this paper.

5 2. The Canadian Venture Capital Industry: 1977-2001 To provide a perspective on venture capital in Canada, we begin by presenting aggregate Canadian venture capital industry statistics collected by Macdonald & Associates, Ltd. (Toronto) for the Canadian Venture Capital Association (“CVCA”) (1978-2002) (for the years 1977 – 2001). The data are presented in Figures 1 – 3 and described below. Most of the data encompass the full 1977-2001 period, but some of the statistics presented below cover a shorter time frame where the full data are not available. Macdonald (1992) previously discusses some of the data. [Figures 1 – 3 About Here] As illustrated in Figure 3, there are six types of venture capital funds in Canada (see also Macdonald, 1992; MacIntosh, 1994; Amit et al., 1997): private independent, corporate, government, institutional direct (formerly known as hybrid funds), foreign, and labour-sponsored. Private independent funds (or simply “private funds” below) are similar to U.S. private funds (described, inter alia, by Gompers and Lerner (1991), and Sahlman (1991)), and are organized as limited partnerships. They tend, however, toward a lower degree of specialization in investment activity than their U.S. counterparts (MacIntosh, 1994; Amit et al., 1997). Canadian corporate VCs are analogous to U.S. corporate VCs (as described by Gompers and Lerner, 1999), but tend to finance a somewhat more heterogeneous group of entrepreneurial firms (Cumming, 2000). Government funds (which comprised 5% of the overall pool of capital in 2001) are managed by independent professional venture capital managers and also finance a wide variety of different entrepreneurial firms. The “hybrid” category consists of public pension funds that invest money directly in entrepreneurial ventures, rather than investing through the medium of a private fund. As of 2001, all institutional actors that make direct investments are reported as “institutional direct” funds, although this category largely corresponds with funds previously reported as hybrid funds (i.e. most are public pension funds). Foreign funds are mostly U.S.-based private funds. The geographic distribution of venture capital investments in Canada is presented in Figure 1. There has been a significant increase in the number of venture capital investments in Canada since 1990, and most of this increase has been concentrated in Ontario and Québec, showing the growing importance of the Canadian venture capital industry for small firms in Canada.3 3

The trends in Canada for dollar values (available upon request from the authors) parallel the trends for

6

Figure 2 presents data from the CVCA indicating capital under management, capital available for investment and new venture funds for the 1988-2001 period. The capital available for investment reflects the extent to which contributions to venture capital funds have outstripped the funds’ ability to invest these contributions. On the basis of the CVCA data, it can be seen from Figure 2 that, historically, there has been a large “overhang” of uninvested capital in Canada. The CVCA data, however, seriously understate the size of the overhang of uninvested capital. The CVCA excludes between 20% and 40% of all LSVCC capital contributions (depending on the jurisdiction of incorporation4) from the overhang, on the basis that the governing legislation requires LSVCC funds to maintain these amounts as uninvested reserves placed in relatively risk-free short-term government and corporate debt obligations. However, the legislation only states that 60-80% of contributed capital (again, by jurisdiction) must be invested within certain deadlines, and that any balance that remains uninvested must be placed in specified low risk instruments. Thus, for example, in 2001, the recorded overhang of $1.3 billion for LSVCC funds should in fact be in excess of $3.8 billion, or 45% of all LSVCC capital under administration. 5 The magnitude of the overhang is likely a consequence of two factors. First, LSVCCs are structured as open-ended mutual funds subject to investor redemptions at any time. The likelihood of redemption is substantially lowered by the fact that premature withdrawal (i.e. prior to the elapse of the statutorily specified hold period) will result in forfeiture of previously claimed tax credits - in addition to contractually stipulated fund penalties that are typically in the vicinity of 6%. Nonetheless, the typical LSVCC lock-in (both for tax credits and contractual penalties) is eight years, which is somewhat shorter than that of private funds (typically 10 years, with the possibility of extensions with approval of the limited partners). This creates a correspondingly greater need than for private funds to maintain liquid investments against the event of redemptions.

numbers of investments. Similar trends in venture capital are documented elsewhere. Previous research on the distribution of venture capital investments within the U.S. includes Gompers and Lerner (1998) and Sorenson and Stuart (1999); research across countries appears in Black and Gilson (1998) and Jeng and Wells (2000). 4 In particular, 40% for funds registered federally or in Quebec, 30% for funds in Ontario, Saskatchewan and Altantic Canada, 25% for Manitoba funds, and 20% for B.C. funds. 5 The overhang for other recent years should be increased as follows (all amounts approximate); 2000 - $1.9 billion; 1999 - $1.5 billion; 1998 - $1.3 billion; 1997 - $1.1 billion; 1996 – $0.9 billion; 1995 - $538 million.

7 Second, and more importantly, we believe that the size of the overhang is accounted for by the comparative lack of skill of LSVCC managers (Cumming and MacIntosh, 2002)). While no systematic evidence exists, anecdotal evidence suggests that many of the LSVCC fund managers have little or no background in venture capital investing. This has led to an inability to identify promising investments. Figure 3 indicates the amount of funds under administration at each fund type from 1992 to 2001. In the past decade, it is clear that most of the growth in the industry can be attributed to the LSVCCs. The capital under administration at the private funds - the LSVCC’s main competitor – remained largely static between 1992 and 1999. While Figure 3 indicates substantial increases in the capital administered by private funds in 2000 and 2001, this increase is illusory. Each fund type in Figure 3 is a net of five factors; capital contributions, returns of capital and/or profits to investors, changes in capital under administration resulting from revaluations of fund portfolios, changes resulting from realizations on the sale of investments, and the composition of the respondents to the CVCA’s annual information-gathering survey. According to the data gatherer for the CVCA, much of the apparent increase in private funding in the year 2000 is attributable to the private fund’s realized capital gains, as well as the fundraising from a large private fund introduced to the database in that year.6 Of the total $3.4 billion increase in private capital reported in 2000, only $1.6 billion can be attributed to new commitments to private funds.7 In addition, it is noteworthy that LSVCC capital under administration in 2000 was significantly affected by substantial redemptions that came due at the end of the investors’ mandatory 8-year holding period (see item 4 in section 3 below), particularly at Working Ventures, one of the largest LSVCCs.8 Thus, while Figure 3 suggests that the relative balance between private and LSVCC funds changed substantially in 2000, the apparent change in balance is also attributable LSVCC redemptions.

6

E-mail correspondence to authors dated 19 Feb. 2003 from Kirk Falconer, Director of Research and Analysis, Macdonald & Associates, Ltd. The CVCA also indicates on its website that the increase in private funding in the year 2000 came from “a handful of large private funds”. See Canadian Venture Capital Association, “Venture Capital Industry Shatters All Records in 2000”, available at http://www.cvca.ca/downloads/2000overview.pdf. Macdonald & Associates, Ltd., notes that their industry coverage have been very comprehensive and stable over the years in which they have recorded data, but less stable over the bubble period in 1999-2000. 7 Ibid. On its website, the CVCA indicates only that the increase in “private” funding in 2002 came from “a handful of large private funds”. See Canadian Venture Capital Association, “Venture Capital Industry Shatters All Records in 2000”, available at http://www.cvca.ca/downloads/2000overview.pdf. The “increase” referred to would appear to include the widening of the CVCA database in 2000. 8 Kirk Falconer, Director of Research and Analysis, Macdonald & Associates, Ltd., private communication to authors, Feb. 20, 2003.

8 While Figure 3 discloses significant proportional increases in corporate, institutional direct and government funds, these increases nonetheless represent a relatively modest increase in aggregate dollar value. Most of the increase in institutional direct funds in the years 2000 and 2001 can be attributed to a single fund (CDP Capital) that is a subsidiary of the Caisse de depot et placement du Quebec, the country’s largest public pension fund.9 Because the Caisse de depot is statutorily mandated to pursue objectives other than profit maximization (and apparently does so quite vigorously10), CDP is thus perhaps more closely analogous to an LSVCC fund than a private fund. As discussed in section 4 below, the increase in LSVCC capital is surprising in light of the low returns realized by these funds. It seems reasonably clear that the capital flows to LSVCCs are best explained by strong tax incentives that are not shared by investors in other types of funds. Section 3 provides more detail. 3. The Legislative, Contractual and Governance Structures of Labour-Sponsored Venture Capital Corporations (LSVCCs) Contrasted with Private Funds Elsewhere, we explore the identity of LSVCC investors, in addition to LSVCC mandates, structure, governance, and taxation (Cumming and MacIntosh, 2002c). In this section, we briefly summarize this and related research on both LSVCC funds and private funds. 1.

A LSVCC may be incorporated in any province that has passed legislation specifically allowing for the creation of a LSVCC. It may also be incorporated under similar federal legislation, in which case it may operate in any province that has passed legislation specifically authorizing a federal LSVCC to operate in that jurisdiction. Of the 50 funds in our sample, 14 are federally incorporated and 29 are incorporated in Ontario.

2.

As noted above, only individuals may invest in LSVCCs. However, any individual, regardless of his or her net worth, may invest, and only a small minimum investment (typically less than or equal to $1000) is required.

9

See Canadian Venture Capital Association, The Venture Capital Industry in 2001: An Overview, available at http://www.cvca.ca/downloads/The%20Venture%20Capital%20Industry%20in%202001%20-%20Overview1.pdf 10 See, e.g., Arbour (1993).

9 3.

Most LSVCC investments are small (