OTC financial 20171030

Debt Financing of Entrepreneurial Firms: Evidence from the OTC Market Claire Y.C. Liang Southern Illinois University Ca...

0 downloads 90 Views 672KB Size
Debt Financing of Entrepreneurial Firms: Evidence from the OTC Market

Claire Y.C. Liang Southern Illinois University Carbondale [email protected]

Rengong (Alex) Zhang City University of Hong Kong [email protected]

October 2017

Preliminary draft Please do not circulate or distribute without permission.

Abstract We examine uses of debt in a large sample of entrepreneurial firms trading on the over-the-counter (OTC) market. Overall debt is a substantial source of capital, accounting for nearly 20% of total financial capital provided in our sample. Further, debt usage increases as firms develop. During the process, debt composition changes as well. We observe that positive sales and positive cash flows mark two important milestones in entrepreneurial firms’ debt financing: having positive sales broadens firms’ access to various segments of the debt market as evidenced by reduced debt specialization; and having positive cash flows deepens firms’ relations with conventional lenders, such as banks. Our study indicates that debt capital, with its diversified sources and flexible features, supplements equity in funding entrepreneurial firms’ development.

1

1. Introduction How small emerging firms survive and thrive in the capital market is crucial to our understanding of entrepreneurial finance (Robb and Robinson 2014). While venture capital (VC) is widely regarded to play an important role in financing startup companies, research and anecdotal evidence suggest that VC backs less than 2% of entrepreneurial firms and accounts for only a small portion of total entrepreneurial financing (Berger and Udell 1998; Moskowitz and VissingJorgensen 2002; Puri and Zarutskie 2010; Rao 2013; Robb and Robinson 2014). In contrast, recent studies show that debt also plays an important role in funding startups besides private equity (Berger and Udell 1998; Cosh, Cumming and Hughes 2009; Robb and Robinson 2014). Debt financing has gained increased attention in the entrepreneurial finance literature. Conventional wisdom deems startup companies unsuitable for debt, as they often lack the collateral or cash flows needed to repay the loans. However, survey evidence shows that debt is prevalently used by startups and can be obtained through alternative means such as the guarantees of founders or large shareholders (Robb and Robinson 2014). Further, debt has the benefits of avoiding over-dilution of ownership (Ibrahim 2010). For example, a biotechnology firm awaiting the results of clinical trials and anticipating a significantly higher equity valuation upon success of the trials may use debt as bridge finance before the next round of equity fund raising. Debt also offers greater flexibility in its features, such as convertibility, seniority, and schedule of payments. Recent literature development also highlights the differences in the fine structure of debt, such as different debt types or different debt features (Colla et al. 2013; Custodio et al. 2013). However, evidence on the nuanced debt structures and features of entrepreneurial firms remains scarce due to data limitations.

2

In this study, we make a first attempt in documenting the debt features and security types employed by a sample of entrepreneurial firms that trade on the over-the-counter (OTC) market. Similar to the vast majority of small emerging firms, most OTC firms are not heavily backed by venture capital and do not qualify for listing on major stock exchanges such as NYSE, Amex, or Nasdaq.1 With more than 8,000 domestic firms trading on the venue, the OTC market has a long history of hosting startup companies from Walmart in its infancy decades ago to recently the first crowdfunded IPO firm Elio Motors (Feldman 2016). Despite its large universe of constituents, the OTC market only provides limited access to equity due to poor liquidity and lack of participation by institutional investors (Brüggemann, Kaul, Leuz, and Werner 2016; Liang and Zhang 2017). As a result, OTC firms rely heavily on debt financing, as is the case for most startup companies (Berger and Udell 1998; Robb and Robinson 2014). We obtain debt and financial data for 3,037 OTC firms from fiscal year 2002 to 2013 based on the the audited financial statements filed with the SEC and the financial data collected and compiled by the Capital IQ database. 2 Our final sample contains a total number of 12,831 firm-year observations. Our study sample is unique compared to previous studies. Prior studies of entrepreneurial finance either focus on firms backed by venture capital or on firms from small business surveys. We complement previous literature by introducing a novel sample set that straddles both camps. Compared to VC-back firms, entrepreneurial firms on OTC better resemble the broad universe of

Capital IQ provides the ownership structure as of firms’ most recent status. Among the 4,405 U.S. firms traded on the OTC market and covered by Capital IQ as of February 2017, only 234 firms (5.3%) has VC/PE ownership. The average VC/PE ownership conditional on receiving VC/PE financing is only 11.5%, in contrast to around 50% in the surveyed firms in Kaplan and Stromberg (2002). 2 A firm is required to file audited financial statements with the SEC if it has total assets greater than $10 million and more than 500 shareholders or if it is quoted on the OTC Bulletin Board (OTCBB) according to the Securities Exchange Act of 1934 and the “the Eligibility Rule” enacted by the SEC in 2000. The criteria of 500 shareholders was further relaxed to 2,000 shareholders under the JOBS Act in 2012. The SEC filings provide detailed information on listing venue and capital structure. 1

3

startup firms in the sense that both lack strong support from outside equity. Relative to the small business survey samples, firms in our OTC sample are more similar in nature to public firms, which VC-backed firms often aim to become, in terms of industry composition and willingness to access outside equity. Insights drawn from previous studies on VC-backed startups are often distinct from those obtained from the small business survey samples likely due to differences in various aspects, such as company format,3 willingness to access outside equity, growth ambition, and industry composition. The unique features of our sample allow us to draw insights from the two distinct strands of prior literature in entrepreneurial finance. We compare and contrast debt usage in three developmental stages: (i) pre-sales, (ii) postsales with negative cash flows, and (iii) positive cash flows. Our classification is inspired by the life cycle framework of small business finance proposed by Berger and Udell (1998) where “financial needs and options change as the business grows” (p.622). About a third of our sample do not have sales. We expect these firms to have a more difficult time raising debt capital. When firms reach the second stage with positive revenues, they demonstrate good potential of their ventures and can borrow against receivables and inventories. This would likely enhance their access to the debt market although access may still be limited due to lack of cash flows. In the third stage when firms have positive cash flows, they show to lenders that they have working business models and good debt repayment capability. This would further increase their access to the debt market, especially to the segments that are more demanding on borrowers’ credit quality. We first examine the usage of debt as a source of capital. Our results show that debt is a substaintial component of capital in our sample. It accounts for nearly 20% of total financial capital provided, and is the most important source of non-equity capital. We also find that the importance

3

Many firms in small business surveys are sole proprietorships, rather than corporations (e.g., Robb and Robinson 2014). 4

of debt increases as firms develop. Debt constitutes 13.6%, 16.1%, and 27.5% of total capital for stage 1, 2, and 3, respectively. The increasing reliance on debt persists after controlling for firm characteristics and in different specifications. We next examine the usage of various debt types. We divide debt into seven mutually exclusive types: commercial paper, revolving credit, term loans, senior bonds and notes, subordinate bonds and notes, capital leases, and other, following the classification in Colla et al. (2013) and the Capital IQ database. These securities differ in their sources, formality, and structures. Revolving credit and term loans are two subcategories of bank debt. Capital leases are provided by leasing companies. Bonds and notes in our sample, including both senior and subordinated bonds and notes, are all likely privately placed because none of our sample firms has public bond ratings. Privately placed bonds and notes are typically less formal than bank debt and more flexible in their features such as convertibility. 4 With this classification, we are able to construct debt specialization measures, following Colla et al. (2013), in order to examine how debt concentration evolves in our sample. We find that overall debt specialization decreases as firms develop, with the most significant decline taking place in stage 2. The result suggests that having sales is an important milestone for entrepreneurial debt financing as it has the biggest impact on debt composition. Our descriptive evidence reveals that in stage 1 (no sales), debt is concentrated in senior bonds and notes, term loans, and other unclassified debt, together accounting for 95% of total debt. This ratio decreases to 85% in stage 2 (positive sales; negative cash flows) as firms increasingly use other types of debt, such as subordinated bonds and notes, capital leases and revolving credit. The results indicate that having positive sales seems to broaden firms’ access to various segments of the debt

4

We observe no usage of commercial paper in our sample, which confirms with Berger and Udell (1998) financial growth cycle that commercial paper is mainly employed by large companies in mature stage of development. 5

market. This pattern is also evidenced by the percentage of firms that predominantly employ a single debt type. In stage 1, 77% of the firms have more than 90% of their debt concentrated in a single debt type, and this ratio drops to 59% in stage 2. This pattern is consistent with the explanations of debt specialization provided by Colla et al. (2013). First, as firms grow and have proven records of business operations, they gain access to debt instruments that were previously unavailable or too expensive to obtain in their earlier stages of development (Faulkender and Petersen 2006; Rauh and Sufi 2010). Second, reduced information asymmetry in later stages decreases the need for concentrated monitoring. Third, firms in later stages have lower risk of bankruptcy, and thus have less demand for concentrated debt to reduce bankruptcy negotiation (Ivashina, Iverson, and Smith 2011). Although reaching positive cash flows in stage 3 does not appear to greatly reduce debt specialization, it significantly increases a firm’s total debt amount as well as its access to bank debt. Bank debt, including both revolving credit and term loans, becomes the dominant debt type in stage 3. It contributes to nearly half (49%) of total debt in stage 3, in contrast to 33% in stage 1. The increased importance of bank debt in stage 3 also persists after controlling for firm characters and in different specifications. Taken together, the results discussed above indicate that positive sales and positive cash flows are two important milestones in entrepreneurial firms’ development in regards to debt financing: Having positive sales broadens firms’ access to various segments of the debt market; and having positive cash flows deepens firms’ relations with conventional lenders, such as banks. To complement our understanding of the aforementioned seven mutually exclusive debt types, we also observe two binary features of debt: convertible (vs. non-convertible), and secured (vs. unsecured). Previous studies and anecdotal evidence suggest that convertible debt is an

6

important source of financing for startup firms (Berger and Udell 1998; Deeb 2014). Prior literature also shows that lenders often prefer collateral when lending to low quality borrowers (e.g., Leeth and Scott 1989). Note that different debt types can share a common feature. For example, privately placed notes and bank debt can both be structured as secured debt as long as borrowers pledge assets as collateral. Hence, our analysis of debt features complements that of debt types. Together, they provide a comprehensive picture of debt structure in our sample of entrepreneurial firms. We find that convertible debt usage peaks in stage 2 and sharply declines in stage 3. The pattern of convertible debt resembles that of bonds and notes. As other debt types, such as bank debt and capital leases, typically do not have convertible features, the similarity between convertible debt and bonds and notes suggests that bonds and notes employed by early-stage firms often have convertible features. Perhaps more interestingly and more surprisingly, we find increasing usage of secured debt for firms at a later stage of development, and this pattern mirrors that of bank debt. The evidence indicates a transition from unsecured non-bank debt to secured bank debt as a firm progresses to a later stage of development. Our study is closely related to Robb and Robinson (2014), which examines the financing patterns of entrepreneurial firms using the Kauffman Survey. First, we differ in the sample space examined. Overall, firms in the Kauffman survey are very small and have limited operations. For example, 36% of the firms in the Kauffman Firm Survey are sole proprietorships and 50% operate at home. In contrast, firms in our sample are corporations and larger on average. Compared to the Kauffman survey, our sample firms are also more skewed towards information technology and

7

biotechnology industries, while such industries have little presence in the Kauffman Survey (Robb, Ballou, DesRoches, Potter, Zhao and Reedy 2009).5 Our study makes several contributions to the literature. First, we contribute to the scarce literature on debt financing of entrepreneurial firms. Due to data limitations, prior literature relies on surveys and focuses on the sources (i.e., insider versus outsider debt) of financing (Berger and Udell 1998; Robinson 2012; Rob and Robinson 2014). We go beyond the insider/outsider distinction and examine the debt security types employed. Different security types from the same source may have different implications for the firm. For example, bonds and notes tend to have a balloon principal repayment at maturity while term loans have a pre-specified repayment schedule throughout the term. As our data come from audited financial statements compiled in a standard format, we are able to compare and contrast usages of various debt types and features as well as debt specialization at different stages of development. Second, we add to the financial growth cycle of small firm financing literature. Berger and Udell (1998) hypothesize that as firms grow, they employ more traditional means of finance. We provide empirical evidence to their framework by documenting that firms rely more on bank debt at later stages of development. We also find empirical support that firms use mezzanine finance, such as convertible debt, the most during the middle stage of development. Lastly, we contribute to the scarce literature on OTC firms. Prior literature has examined stock returns and liquidity (e.g., Ang et al. 2013; Eraker and Ready 2015), and disclosure regimes (e.g., Jiang, Petroni and Wang 2016; Brüggemann, Kaul, Leuz, and Werner 2016; Bushee and

5

2.6% of firms in the Kauffman Survey are in IT industry, and we found no presence of bio-tech/pharmaceutical firms. The largest four industries in the Kauffman Surveys are professional, management, and educational services (15.5%); retail trade (15.6%); Administrative and support, and waste management and remediation services (11.4%) and construction (9.8%). The top four industries in our sample are business services (15.5%), petroleum and natural gas (8.18%), pharmaceuticals (7.70%) and computer software (7.50%). 8

Leuz 2005). Ours is the first study to provide detailed descriptions of firm characteristics and their financing patterns with a focus on debt. More importantly, we demonstrates that the OTC space contains a rich set of startup firms at various stages of development and can be a fruitful ground for studying non-VC backed startups. The remainder of the paper is organized as follows: Section 2 discusses the background and related literature. Section 3 describes sample and data, and provides summary statistics on the usage of various debt security types and features. Section 4 presents the results of multivariate regression analysis. Section 5 discusses and summarizes our findings. 2. Background 2.1 Entrepreneurial Finance and Related Literature Entrepreneurial firms are important for economic growth, innovation, and employment (OECD 1998, Sargeant and Moutray 2011). They account for 66% of net new job creation since the 1970s in the U.S.6 A growing literature also suggests that they are more likely to embrace innovation than large established firms (Akcigit and Kerr 2016; Nanda, Younge and Fleming 2015). However, due to limited track record and high information asymmetry, enterprenerial firms are long believed to be financially constrained. Their financing pattern is of great interest to policy makers. There is great heterogeity among enterprenerial firms. Some specialize in biotechnology or information technology that requires intense investment in intellectual capital and has a potential big market. Others are small scale businesses serving the lifestyle needs of local customers (e.g. restaurants, lawn services). As a result, enterprenerual finance literature is fragmented (Cuming

6

https://www.sba.gov/managing-business/running-business/energy-efficiency/sustainable-business-practices/smallbusiness-trends 9

and Vismara 2017), with different studies examining different enterprenerial samples based on the sources of data. In general, enterprenerial finance literature follows two streams. One stream of the literature examines firms funded by ventural capital. These firms are the most promising group of firms in the startup universe given the strict vetting processes of VC financing.7 This stream of literature either relies on survey data with small sample sizes but rich contract details, or commercial databases such as Thomson One (formerly VentureExpert) with larger sample sizes but less refined data. These studies highlight the staging of capital investments and prevalent use of convertile preferred equity in VC financing (e.g., Kaplan and Stromberg 2003, 2004; see Hall and Lerner 2010; Dabin, Hellmann and Puri 2013; Cumming and Johan 2009 for comprehensive reviews). Although venture capital plays an important role in enterprenerual finance, they are very selective in the industry and type of firms they sponsor (Zider 1998). Research and annecotal evidence suggest that only a very small portion of enterperurial firms (ca. 2%) are VC-backed (Moskowitz and Vissing-Jorgensen 2002; Puri and Zarutskie 2010; Rao 2013; Robb and Robinson 2014). Thus the insights drawn from the VC literature may not be applicable to the broader universe of entrepreneurial firms. The second stream of the literature examines startup firms based on the surveys conducted on small businesses. Earlier studies relied on National Survey of Small Business Finances (NSSBF) conducted every five years from 1987-2003. Berger and Udell (1998) summarize and review these earlier studies. Together, these studies highlight owners’ equity, bank debt, and trade credit as the most important sources of financing for startups. Owners often use their personal assets, such as houses, as collateral to obtain bank financing. They also rely on informal debt from family and

7

Less than 1% of startup firms received VC financing according to the Kaufmann Survey (Robb and Robinson 2014).The lack of access to VC financing was also confirmed by practitioners (Rao 2013). 10

friends although to a lesser degree. Berger and Udell (1998) provide a conceptual framework for the financial growth cycle of entrepreneurial firms, with firms in later stages of development employ more conventional forms of debt or debt with less intense monitoring. Robb and Robinson (2014) examine the financing sources of nascent firms using Kauffman Firm Survey. They document heavy reliance on bank debt even for firms at the very early stage. These surveys focus on the providers of debt financing such as owners, commercial banks, trade suppliers, and family and friends. Less is known about the features or security types of debt instruments employed (e.g., convertible debt, term loans, senior or subordinate notes). In addition to these two aforementioned subsamples of entrepreneurial firms, the over-thecounter (OTC) market houses a large number of entrepreneurial firms. Many of them are in technology or biotech industries, but are not fully funded by venture capital and do not qualify for trading on the main stock exchanges, such as NYSE, Nasdaq, and Amex. The OTC market provides a platform for these firms to access outside equity although their access to equity is more restricted relative to main-exchange firms due to low liquidity and limited participation of insititutional investors (Brüggemann et al. 2016; Liang and Zhang 2017). Appendix A privides a brief description of the institutional details of the OTC market. 3. Sample and Data 3.1 Sample Construction We start with U.S. firms that file annual reports with the SEC from 1994 to 2014. 8 We identify OTC firms with the help of Intellegize, an EDGAR search tool, by searching for keywords of “OTC”, “Over-the-Counter”, “Pink Sheets”, “Over-The-Counter Bulletin Board”, “OTCBB”,

8

These include Form 10-K, 10-KSB and 10-KSB40 filings. Form 10-KSB filings are annual reports filed by small business issuers before March 2009. Since then, 10-KSB has been eliminated and small business issuers file 10-K and tick their status on the front page. Form 10-KSB40 is an optional form for annual and transition reports of small business issuers under Section 13 or 15 (d) of the Securities Exchange Act 1934. 11

“OTCQB”, or “OTCQX” in Item 5 of the annual reports, which contains information on the filer’s common equity including the trading venue.9 Among these firms, we further identify a subset of firms whose common equity was once traded on main exchanges through keyword searches on “NYSE”, “AMEX”, “NASDAQ” or the non-abbreviated variations. We then manually check the annual reports to construct the trading venue history for each firm in the subset. In particular, we identify years when OTC firms switch to or from the main exchanges (i.e., “rising stars” and “fallen angels”). To focus on entrepreneurial firms trading on the OTC market, we remove firms that first appear in our data as main-exchange listed but later trade down to the OTC market because these are likely established firms undergoing financial distress. We also remove an OTC firm from our sample once it obtains main-exchange listing (but we keep the OTC years) or when its total assets exceed $100 mm. Finally, we exclude firms in the financial or utilities industries (SIC codes 6000–6999; 4900–4949) from our sample. We obtain financial and detailed capital structure data from Capital IQ corporate intelligence platform, which provides comprehensive coverage on such data starting in 2002 (Colla, et al. 2013).10 We match our OTC firm list obtained through SEC filings search described above with Capital IQ through the SEC filer number (Central Index Key or CIK), company name, headquarter location, and manual checks. As annual reports filed in calender year 2014 do not have complete coverage for fiscal year 2014, we end our sample period at fiscal year 2013. Hence, our final sample covers fiscal years 2002-2013 with 12,831 firm-year observations. 3.2 Sample Overview

9

In addition to Pink Sheets, the OTC Markets Group (OTCMG) established another two tiers, OTCQB and OTCQX, to compete with OTCBB in recent years. These two tiers are subject to more financial standards or reporting requirements imposed by the OTCMG than Pink Sheets. See Brüggemann et al. (2016) for detailed descriptions. 10 In addition to more detailed data on capital structure, Capital IQ also has better coverage of OTC firms overall than Compustat. We are able to match 64% of our OTC firm list with Capital IQ, but only 50% with Compustat. 12

Panel A1 of Table 1 shows the sample distribution by year. The sample starts with 746 firms in 2002, and rises to 1,004 firms in 2005. Afterwards, the sample size remains relatively stable, around 1,100-1,200 every year. This contrasts the declining number of firms trading on main exchanges during the same period (Doidge, Karolyi and Stulz 2016). Overall, there are 3,037 unique firms in our sample. Among these OTC firms, 7.9% or 239 firms succeeded to obtain listings on one of the three major stock exchanges (“rising stars”) during our sample period (Panel A2). Panel B of Table 1 presents the industry distribution of our sample. The top four industries are business services, petroleum and natural gas, pharmaceuticals, and computer software, accounting for approximately 40% of the sample. These industries feature high R&D or fixed assets investments and have large financing needs. The industry distribution resembles that of firms trading on the main exchanges. This contrasts the samples from the Small Business Finance Survey or Kauffman Firm Survey, where the establishments primarily provide lifestyle services (Robb, Ballou, DeRoches, Potter, Zhao and Reedy 2009). We provide the summary statistics of firm characters in Panel C of Table 1. We divide the sample into three stages of development: (1) pre-sales; (2) post-sales with negative cash flows; and (3) positive cash flows. We use operating income plus depreciation as the proxy for cash flow. Firms in stage 1 have not started generating revenues (sales = 0); in stage 2, sales are positive but cash flow is still negative; in stage 3 both sales and cash flow are positive. Overall, firms in later stages are larger and more profitable. They also have higher asset tangibility and lower risk of bankruptcy as reflected in the Altman Z-score. The mean (median) total assets is $2.295 (0.195) mm for firms in stage 1, $5.171 (1.529) mm for stage 2, and $15.415 (7.857) mm for stage 3. Tangible assets (net property, plant, and equipment; NPPEA) account for 17.9% of total assets on

13

average for stage 1, 21.7% for stage 2, and 22.8% for stage 3. It is also worth noting that R&D intensity, measured with R&D expense/total assets (RDA), declines from 20.8% in stage 1, to 14.5% in stage 2, and 2.7% in stage 3. In contrast, firms in stage 2 have the highest advertising intensity, measured with advertising expense/total assets (ADA). 3.3 Capital Composition Table 2 reports the frequency and magnitude of usage of non-equity instruments. Following Robb and Robinson (2014), we scale each source of finance by the total capital provided, referred to as total capital injection hereafter. Total capital injection is calculated as the sum of total debt, preferred equity, warrants, and equity injection. Equity injection is the total value of common equity less retained earnings (losses). 11 Prior literature (Berger and Udell 1998; Robb and Robinson 2014) shows that trade credit is an important source of financing for startups, so we include trade credit (accounts payable) in Table 2 alongside other non-equity instruments for comparison. Overall, debt and trade credit are the most frequently used means of non-equity financing: 81.8% of observations in our sample use debt and 94.7% have a non-zero balance of payables; when scaled by total capital injection, debt accounts for 17.0% of total capital injection on average, greater than 10.6% for accounts payable. Preferred equity and warrants are used in more than 20% of the observations (22.6% and 28.4%, respectively), but they account for much smaller portions of total capital (1.4% and 4.8%). The magnitude of debt and trade credit usage also increases as a firm develops. The mean values of accounts payable scaled by total capital injection are 7.6%, 9.1%, 22.7% for stages 1, 2,

11

As start-up firms have large negative retained earnings relative to assets, values scaling by total asset will largely capture the profitability of the firm and often generate extreme values. By construction, profitability increases from stage1 to stage 3. Our contrast of three stage is less meaningful if our ratios largely reflect a profitability effect. 14

and 3 respectively. Debt constitutes 13.6%, 16.1%, and 27.5% of total capital from stage 1 through 3. This suggests that when a firm grows in sales and cash flow, it has increasing access to debt and trade credit. In contrast, the amounts of preferred and warrants increase from stage 1 to 2 when a firm begins generating revenues, and then decline from stage 2 to 3 when a firm becomes cash flow positive. The results in Table 2 suggest that debt is the most important source of non-equity capital for entrepreneurial firms trading on the OTC market. In the next section, we further examine the composition and features of debt capital. 4. Debt Composition 4.1 Descriptive Statistics Colla et al. (2013) and Capital IQ group debt into seven mutually exclusive categories: commercial paper, revolving credit, term loans, senior bonds and notes, subordinate bonds and notes, capital leases, and other debt12, based on security and lender types. We find no existence of commercial paper in our sample, consistent with the financial growth cycle framework proposed by Berger and Udell (1998) that commercial paper is available only to most established firms. Hence, we focus our discussions on the remaining six types. Table 3 summarizes the frequency (Panel A) and magnitude (Panel B) of debt usage in our sample. In addition, we also report the statistics of three binary debt features: short-term debt (vs. long-term debt), convertible debt (vs. straight debt), and secured debt (vs. unsecured debt). 4.1.1 Frequency of debt usage The results in Panel A of Table 3 show that overall senior bonds and notes are the most frequently used debt type: over half of the observations in our sample (57.8%) employ senior bonds

12

Other debt is debt that cannot be classified into any of the other six debt types. 15

and notes in their capital structure, followed by 37.5% of term loans and 19.4% of revolving credit. Together, term loans and revolving credit consist of bank debt reported by Capital IQ. Bonds and notes in our sample, including both senior and subordinate, are likely all privately placed, based on lack of public bond ratings. Our readings of 100 randomly selected annual reports indicate a broad spectrum of investors, including investment funds, small investor groups, significant shareholders, executives, vendors, and suppliers. Many of these bonds and notes also have convertible feature. Overall, the results indicate that bank debt and privately placed bonds and notes are the most important sources of debt capital for entrepreneurial firms in our sample. The usage of bank debt increases steadily throughout the three developmental stages. Drawn revolving credit are used by only 6.3% of observations in stage 1; the ratio rises to 22.9% in stage 2, and then reaches 35.4% in stage 3. Similarly, term loans are employed by 32.2%, 39.3%, and 43.1% of observations in stages 1-3 respectively. The evidence is consistent with prior studies that bank debt is an important source of financing even at the very early stage of development (Berger and Udell 1998; Robb and Robinson 2014). It also suggests greater presence of bank debt in later stages as firms solidify operations by starting to generate revenues and cash flows. Similar patterns hold for capital leases (stages 1-3: 2.6%, 17.3%, 27.1%) and secured debt (stages 1-3: 21.0%, 53.6%, 64.6%). Overall, the evidence suggests that uses of conventional debt sources, such as bank debt, increase as firms have greater cash inflows. In the process, they also shift towards collateralized or secured debt, which are less risky from lenders’ perspectives, such as capital leases or debt secured by receivables or other firm assets.13 In contrast, the usages of senior bonds and notes, other debt, short-term debt, and convertible debt do not increase monotonically; their frequencies of usage peak in stage 2 (sales >

13

Capital leases are collateralized by capital equipment, so they are secured in nature and less risky than unsecured debt. 16

0 but cash flow < 0), indicating that having at least some revenues allows firms to access a greater variety of debt capital. However, the frequencies of usage of these four debt types decline in stage 3 when firms turn cash flow positive. This seems to suggest that firms view these four debt types as unfavorable means of bridge financing, so they transition out of these types when credit quality improves. The results are consistent with the findings in the prior literature that firms of the lowest credit quality borrow from non-bank private lenders (Denis and Mihov 2003), and the importance of convertible debt for startup firms (Berger and Udell 1998). Other debt seems to fall into this category of less conventional and less favorable debt types. Subordinate bonds and notes show a slightly different pattern from senior bonds and notes in that its frequency of usage increases throughout the three stages. However, subordinate bonds and notes are much less prevalent than the other five debt types (4.6% of observations). Overall, the usage of bonds and notes, including both senior and subordinate, also peaks in stage 2. 4.1.2 Magnitude of debt usage After discussing the frequency of usage of various debt types and features, we next examine the magnitude of usage as a percentage of total debt amount. Table 3 Panel B presents the results. The patterns are broadly consistent with those in Panel A. Overall, senior bonds and notes are the largest component of debt (52.7%), followed by term loans (25.0%) and revolving credit (8.9%). Together bonds and notes, including both senior and subordinate, constitute 54.7% of total debt amount; bank debt, including revolving credit and term loans, account for 33.9%. By development stage, revolving credit occupies only 3.0% of total debt for firms in stage 1, but the number increases to 8.9% in stage 2 and 20.4% in stage 3. Term loans constitute 30.0%, 21.5%, and 28.3% for stages 1-3 respectively. Together the high percentage of bank debt in stage 1 suggests the importance of bank debt in launching a new business. Bonds and notes remain the

17

largest source of financing for stage 1 (54.0%) and stage 2 (57.2%), but in stage 3, its share (33.5%) is surpassed by bank debt (48.7%). The proportion of other debt declines steadily throughout the three stages (11.2%, 6.3%, and 4.8%), similar to short-term debt (68.2%, 48.2%, 31.9%), suggesting that these debt types become less popular as firms become more profitable. 4.1.3 Debt specialization The results so far indicate that as firms develop, they employ a greater amount of debt in their capital structure. In earlier stages, privately placed bonds and notes, especially senior bonds and notes, constitute more than 50% of total debt, but their dominance is overtaken by bank debt in the third stage. These results indicate changing debt dynamics in a firm’s early life cycle. We next examine how debt specialization varies by stage. We construct two debt specialization measures following Colla et al. (2013): HHI and Excl90 as shown in eq. (1) and (2) below. 𝑆𝑆𝑖,𝑡 in eq. (1) is the sum of the squares of the weight in total debt of each of the seven debt types discussed above; hence, HHI is based on similar concept of the Herfindahl-Hirschman Index. Excl90 is a dummy variable that equals 1 if the amount of one debt type accounts for at least 90% of total debt. 𝐻𝐻𝐼𝑖,𝑡 =

𝑆𝑆𝑖,𝑡 − 1/7 1 − 1/7

(1)

𝐸𝑥𝑐𝑙90𝑖,𝑡 = 1 𝑖𝑓 𝑜𝑛𝑒 𝑑𝑒𝑏𝑡 𝑡𝑦𝑝𝑒 ≥ 90%; 0 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒

(2)

Table 3 Panel C reports debt specialization statistics by stage. Overall, debt specialization decreases steadily, and the largest decline occurs at the transition from stage 1 to stage 2. HH1 is 0.881 in stage 1, lowers to 0.780 in stage 2, and then 0.733 in stage 3. Excl90 exhibits a similar but more prominent trend: in stage 1, 76.7% of the observations rely almost exclusively on single debt type; the ratio drops to 59.0% in stage 2, and reaches 52.1% in stage 3.

18

The next two columns show the two most important sources of debt as a percentage of total capital injection. Bank debt accounts for 4.5% of total capital injection in stage 1, and the ratio more than tripled to reach 14.9% in stage 3. Bonds and notes constitute a greater percentage of capital relative to bank debt in stage 1 (7.5%), but by stage 3, its weights shrinks to 9.4%, 5.5% lower than bank debt. The last two columns report the ratios of convertible and secured debt to total capital injection. Secured debt exhibits a pattern similar to that of bank debt. On the other hand, convertible debt shares similar fate as bonds and notes, but its waned importance in stage 3 is even more pronounced. 4.2 Multivariate Evidence Having shown the descriptive statistics, in this section we test whether the patterns still hold if we control for known determinants of debt. All regressions include industry and year fixed effects with standard errors clustered by firm. 4.2.1 Total debt Table 4 reports the regression estimates with the ratio of total debt to total capital injection as the dependent variable. In addition to known determinants of debt, we also include two dummy variables in the regressions: (i) Sale_Pos = 1 if a firm has positive revenues in the year, and zero otherwise; (ii) CF_Pos =1 if a firm has positive cash flows in the year, and zero otherwise. Hence, firms in stage 1 (sales = 0) have a value 0 for both variables; in stage 2 (sales > 0; cash flow 0) both variables = 1. The factor loading on Sale_Pos measures the difference in debt ratio between firms in stages 1 and 2, and CF_Pos estimates the difference between stages 2 and 3. We include the following control variables in our regressions following the prior literature (Graham and Leary 2011): size (logarithm of total assets; LAT), growth (growth of total assets;

19

ATG), asset tangibility (net plant, property, and equipment/total assets; NPPEA), profitability (operating income/assets; EBITA), volatility (standard deviation of operating income/assets from the past three years; EBITA_VOL), bankruptcy risk (Altman’s z-score; ZSCORE), R&D expense (R&D expense/assets; RDA), advertising expense (advertising expense/assets; ADA), and capital expenditure (capital expenditure/lagged assets; CAPEXA). We use assets growth to proxy for growth, not sales growth or market-to-book ratio, because many of our observations do not have sales or have missing market values in the Capital IQ database. Table 4 column 1 shows that there are significant differences between pre-revenue and post-revenue firms, and firms that are post-revenue with negative cash flows and those with positive cash flows in the cross-section. The result is similar if we include firm fixed effects in the regressions (column 4). Although short-term debt constitutes a significant portion of total debt, the results continue to hold when we exclude short-term debt from the debt-to-capital calculation (columns 2 and 5). The results are also similar if we exclude convertible debt from the analysis (columns 3 and 6). Some of the control variables have same signs as those documented in the previous studies conducted on main-exchange firms: firms with high asset tangibility use more debt. However, some of the controls have different signs from those found on large public firms. For example, growth has positive signs in most regressions, indicating that debt usage is associated with higher growth. Several reasons may contribute to this: (i) startup firms may use debt to signal their quality and growth potential; (ii) high-growth startup firms anticipating reaching certain value-enhancing milestones (e.g. success in drug clinical trials) in the near term may use debt as bridge capital before the next round of equity financing to reduce ownership dilution (Ibrahim 2010). Some of the controls change signs when firm fixed effects are included. For example, profitability has a

20

negative sign in columns 1 – 3, but the sign turns positive in columns 4 - 6. The result suggests that although more profitable firms tend to use less debt in the cross-section, for a given firm, debt usage tends to increase with profitability. Overall, the results in Table 4 are consistent with the patterns of descriptive statistics discussed in Section 3. Debt plays an important role in funding entrepreneurial firms in our sample, and its usage increases with development, especially from no revenue to positive revenues, and from negative to positive cash flows. 4.2.2 Debt types and features In Section 3, we show that entrepreneurial firms in our sample source their debt capital from two broad categories: (i) privately placed bonds and notes, including both senior and junior bonds and notes; and (ii) bank debt, including revolving credit and term loans. Together these two categories account for nearly 90% (88.6%) of debt capital in our sample. However, their usages exhibit distinct dynamics throughout the three developmental stages: while bank debt increases steadily with development, bonds and notes peak in stage 2. Besides, debt with secured feature shows patterns similar to those of bank debt, and convertible debt resembles bonds and notes. We next examine these patterns under a multivariate regression setting. Table 5 presents the regression results. We scale the amount of various debt types by total capital injection to facilitate comparison with total debt. Another reason we choose to scale each debt component by total capital injection rather than total debt is that the latter would not capture the dynamics of moving from zero debt to some debt because zero-debt firms would be excluded from the analysis if total debt is the scalar. Panel A columns 2 and 3 show that cross-sectionally, firms with positive revenues use significantly more both bonds and notes and bank debt than firms without revenues, and firms with positive cash flows use more bank debt compared to firms with

21

only positive sales. We include in column 1 the results of total debt from Table 4 to help visualize the changes. Together, columns 1 – 3 indicate that the increase in debt usage in stage 2 (Sale_Pos) comes from increases in both privately placed bonds and notes and bank debt while in stage 3 (CF_Pos), the increases most source from bank debt. In Panel B, we include firm fixed effects in the regressions. The results are qualitatively similar. These results in Table 5 indicate that positive sales and positive cash flows are two important milestones of debt financing for entrepreneurial firms in our sample: Each is associated with a significant increase of debt usage. Both cross-sectionally and within-firm, positive sales is associated with greater uses of privately placed bonds and notes and bank debt, and positive cash flows is associated with greater uses of bank debt. The debt increase in stage 2 can also be attributed to greater usage of debt with convertible feature (column 4). However, the importance of convertible debt decreases in stage 3. The pattern is more pronounced in the cross-section and less so within firm. In contrast, secured debt increases steadily from stage 1 to 3. Its patterns mirror those of bank debt. Overall the results in Table 5 are consistent with the patterns discussed previously in Section 3. Debt is an important component of capital for entrepreneurial firms in our sample and the types and features of debt employed vary throughout the early life cycle. In earlier stages when cash flows are still negative, firms rely more on debt from private non-bank lenders or on debt with convertible feature, consistent with Berger and Udell (1998) and Denis and Mihov (2003). However, within these firms, there are still noticeable differences between firms with sales and without sales. The former group uses more debt compared to the latter, and the increases are across the board. In the last stage when firms reach cash flow positive, debt increases primarily come

22

from more conventional sources, such as bank debt, and from debt that is less risky from lenders’ perspectives, such as secured debt. 4.2.3 Debt specialization The results in Table 5 indicate that firms use more debt, with debt coming from a greater variety of sources, starting in stage 2. Previously the descriptive statistics also show that debt specialization decreases as firms develop. In Table 6, we examine the pattern controlling for known determinants of debt specialization. Our dependent variables are the two debt specialization measures defined in Colla et al. (2013): HHI and Excl90. We use the same set of control variables as in Colla et al. (2013) except for unrated and dividend payer dummies because no observations in our sample are rated and very few pay dividends (0.7%). We use OLS regressions instead of Tobit or Probit to facilitate interpretation of the results. The results in Table 6 suggest that debt specialization exhibits a significant drop from stage 1 to stage 2 and stabilizes from stage 2 to stage 3. In columns 1-2 without firm fixed effects, the factor loadings on Sale_Pos are -0.052 and -0.093 for HHI and Excl90 respectively, both statistically significant at the 1% level, while the loadings on CF_Pos are insignificant. Adding firm fixed effects yield qualitatively similar results (Sale_Pos coefficients: -0.029 for HHI; -0.062 for Excl90, both significant at 1% level). These results are consistent with the patterns observed previously that debt becomes more diversified when firms start to generate revenues. Most of the control variables have signs same as those documented in the literature. Larger firms are less specialized. High growth firms have higher debt concentration. Firms with high asset tangibility have less debt concentration. 5. Summary and Discussions

23

We study the uses of debt in funding entrepreneurial firms trading on the OTC market in this paper. We document that debt plays a substantial role in funding entrepreneurial firms. Further, the importance of debt increases as firms develop. During the process, the types and features of debt employed change as well. We divide our sample firms into three stages of development: (i) pre-sales, (ii) post-sales with negative cash flows, and (iii) positive cash flows. We find that positive sales and positive cash flows are two important milestones in entrepreneurial firms’ debt financing: Having positive revenues broadens firms’ access to various segments of the debt market; and having positive cash flows deepens firms’ relations with conventional lenders, such as banks. As a result, debt specialization declines steadily throughout the three stages, with the most significant drop observed in stage 2. We ration that when firms start generating revenues, they demonstrate that their products have demand in the market place, which sends a positive signal to lenders. Further, having sales enables firms to borrow against receivables and inventories. This grants them greater access to the debt market. However, since these firms are still cash flow negative, they rely more heavily on private non-bank debt, such as privately placed bonds and notes, which tend to have lesser requirements on credit quality than bank debt. They also use more hybrid securities, such as convertible debt, which allows lenders to convert to equity and capture the upside if firms perform well in the future. When firms reach the second milestone of being able to generate positive cash flows, their ability to repay lenders is further enhanced because now they can borrow against future cash flows. Thus firms gain increased access to the debt market, especially bank debt, which tends to be more demanding on borrowers’ credit quality. We also document that as firms develop, they use more

24

secured debt. Together, the evidence suggests that in earlier stages, firms have limited cash inflows and less valuable assets that can serve as collateral, so they have to rely on less conventional sources of debt financing, such as privately placed bonds and notes with convertible features. In later stages, firms have more cash flows and more valuable assets, which can serve as collateral and facilitate bank borrowing. Prior research on entrepreneurial firm finance has primarily focused on the importance of equity capital. A growing number of studies show that debt also plays an important role in funding startup firms. In this study, we extend the research on debt capital in entrepreneurial firms and show that many parts of the debt market participate in funding entrepreneurial firms. Most previous studies view debt as a monolith but an increasing number of studies explore the fine structures of debt. Our findings add more evidence to the diversity and versatility of debt capital.

25

References Akcigit, U., and W. R. Kerr, 2016, Growth through heterogeneous innovations, Working Paper. Ang, A., Shtauber A.A., and P. C. Tetlock, 2013, Asset pricing in the dark: The cross-section of OTC stocks, Review of Financial Studies 26, 2985–3028. Kerr, W. R., and R. Nanda, 2015, Financing innovation. Annual Review of Finance and Economics, 7,445-462. Berger, A. N., and G. F. Udell, 1998, The economics of small Business finance: the roles of private equity and debt markets in the financial grwoth cycle, Journal of Banking and Finance 22: 613-673. Bruggemann, U., A. Kaul, C. Leuz, and I. M. Werner. 2016. The Twilight Zone : OTC Regulatory Regimes and Market Quality OTC Regulatory Regimes and Market Quality. Working Paper Bushee, B. J., and C. Leuz, 2005, Economic consequences of SEC disclosure regulation: Evidence from the OTC bulletin board, Journal of Accounting and Economics 39, 233–264. Cumming, D. J., and S. Vismara. 2017. De-segmenting research in entrepreneurial finance. Venture Capital 19 (1–2): 17–27. Custodio, C., M. A. Ferreira, and L.Laureano, 2013, Why are U.S. firms using more short-term debt? Journal of Financial Economics 108, 182-212. Colla, P., Ippolito F., and K. Li, 2013, Debt specialization, Journal of Finance 68, 2117–2141. Cumming, D., and S. Johan 2009, Venture Capital and Private Equity Contracting: An International Perspective, Elsevier Science Academic Press. Da Rin. M., Hellmann, T., and M. Puri, 2013, A survey of venture capital research, Handbook of the Economics of Finance 2(A), eds. Constantinides, George M., Milton Harris, and Rene M. Stulz, Amsterdam: North Holland. Denis, D. J., and V. T. Mihov. 2003. The choice among bank debt, non-bank private debt, and public debt: Evidence from new corporate borrowings. Journal of Financial Economics 70 (1): 3–28. Doidge, C., A. Karolyi, and René M. Stulz, 2017, The U.S. listing gap, Journal of financial Economics 123, 1–65. Eraker, B., and M. Ready, 2015, Do investors overpay for stocks with lottery-like payoffs? An examination of the returns of OTC stocks, Journal of Financial Economics 115, 486–504. Feldman, A. (Mar 1, 2016). Elio Motors, First Equity-Crowdfunded IPO, Soars Past $1B Valuation Days After Listing Shares. Retrievd from 26

https://www.forbes.com/sites/amyfeldman/2016/03/01/elio-motors-first-equity-crowdfundedcompany-soars-past-1b-valuation-days-after-listing-shares/#13b988c97844 Faulkender, M., and M. A. Petersen. 2006. Does the source of capital affect capital structure? Review of Financial Studies 19 (1 SPEC. ISS.): 45–79. Graham, J. R., and M. T. Leary, 2011, A review of empirical capital structure research and directions for the future, Annual Review of Financial Economics 3, 309-345. Ibrahim, D.M. 2010. Debt as venture capital. Univ. Illinois Law Rev. 2010(4): 1169-1210. Ivashina, V., B. Iverson, and D. C. Smith, 2011, The ownership and trading of debt claims in Chapter 11 restructurings, Working paper, Harvard University. Jiang, J. X., K. R. Petroni, and I. Y. Wang, 2016, Private Intermediary Innovation and Market Liquidity: Evidence from the Pink Sheets Market, Contemporary Accounting Research 33, 920– 948. Kaplan, S. N., and Per Stromberg, 2003, Financial contracting theory meets the real world: evidence from venture capital contracts, Review of Economic Studies 70, 281–315. Kaplan, S. N., and Per Stromberg, 2004, Characteristics, contracts, and actions: evidence from venture capitalist analyses., Journal of Finance 59, 2177–2210. Kerr, W. R., and R. Nanda. 2015. Financing Innovation. Annual Review of Financial Economics 7 (1): 445–462. Lerner, J., and H. B. Hall, 2010, Handbook of The economics of innovation, Vol. 1, Handbook of the Economics of Innovation 1, 609–639. John D. L. and J. A. Scott 1989, The Journal of Financial and Quantitative Analysis, Vol. 24, No. 3: 379-394 Rao, D., (July 22, 2013) Why 99.95% of Entrepreneurs should stop wasting time seeking venture capital, Forbes Magazine. Retrieved from https://www.forbes.com/sites/dileeprao/2013/07/22/why-99-95-of-entrepreneurs-should-stopwasting-time-seeking-venture-capital/#7cddab8546eb Robb, A., B., J., DesRoches, D., Potter, F., Zhao, A., & Reedy, E. J. 2009. An overview of the Kauffman firm survey: Results from the 2004-2007 data. Rauh, J. D., and A. Sufi. 2010. Capital structure and debt structure. Review of Financial Studies 23 (12): 4242–4280. Robb, A. M., and D. T. Robinson, 2014, The capital structure decisions of new firms, Review of Financial Studies 27, 153–179. Robinson, D. 2012. New Perspectives on Entrepreneurial Capital Structure. Cumming (ed.), 27

Oxford Handbook of Entrepreneurial Finance, pp. 153-166. Reeb,G. (Mar 19, 2014). Comparing Equity, Debt and Convertibles for Startup Financings. Forbes. Retrieved from https://www.forbes.com/sites/georgedeeb/2014/03/19/comparing-equityvs-debt-vs-convertibles-for-startup-financings/#5fe2a39d69ff Sargeant, W., and C. Moutray. 2011. The small business economy: a report to the president 2010. Washington, D.C.: United States Government Printing Office. Zider 1998. How Venture Capital Works. Harvard Business Review. November-December, 131139.

28

Table 1 Sample Descriptions Panel A1: Number of observations by year Year

Number of Firms

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

746 833 919 1004 1113 1200 1153 1174 1179 1187 1187 1136

Total

12831

Panel A2: Number of unique firms “Rising star” is a firm that trades on the OTC market but later obtains listing in one of the three main stock exchanges (NYSE, Nasdaq, and Amex) during the sample period.

Total Rising star

No. of Unique Firms

% Total

3037 239

7.9%

29

Table 1. Sample Descriptions Panel B: Industry distribution: Top 20 industries Industry grouping follows Fama-French 49 industry classification. Rank

Industry

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18.5 18.5 20

Business Services Petroleum & Natural Gas Pharmaceuticals Computer Software Mining Retail Wholesale Electronic Equipment Communications Medical Equipment Entertainment Chemicals Machinery Personal Services Precious Metals Consumer Goods Healthcare services Automobiles & Trucks Food Products Measuring and Control Equipment

%

Cumulative %

15.7% 8.3% 7.8% 7.6% 5.3% 4.4% 4.1% 4.0% 3.7% 3.2% 3.2% 3.0% 2.3% 2.1% 1.9% 1.8% 1.6% 1.5% 1.5% 1.4%

15.7% 24.0% 31.8% 39.4% 44.7% 49.1% 53.1% 57.1% 60.8% 64.1% 67.2% 70.2% 72.5% 74.6% 76.6% 78.3% 80.0% 81.5% 83.0% 84.5%

30

Table 1. Sample Descriptions Panel C: Descriptive statistics of the sample Firms in stage 1 has no sales; in stage 2, sales are positive but cash flow is still negative; in stage 3, cash flow is positive. Cash flow is measured with operating income plus depreciation (EBITDA). SALE: total revenues. CF: operating income plus depreciation (EBITDA). AT: total assets. ATG: total assets growth = log(AT/lagged AT). SALEG: sales growth = log(SALE/lagged SALE). NPPEA: net plant, property, and equipment/total assets. EBITA: operating income/total assets. EBITA_VOL: the standard deviation of EBITA from past 3 years. ZSCORE: Altman z-score = (3.3*earnings before taxes + total revenues + 1.4*retained earnings + 1.2*(total current assets – total current liabilities))/total assets. RDA: research and development expense/total assets. ADA: advertising expense/total assets. CAPEXA: capital expenditure/lagged total assets. Stage

SALE

EBITDA

AT

ATG

SALEG

NPPEA

EBITA

EBITA_VOL

ZSCORE

RDA

ADA

CAPEXA

N Mean p25 p50 p75

1

4205 0.000 0.000 0.000 0.000

4205 -2.072 -2.049 -0.612 -0.105

4205 2.295 0.019 0.195 1.248

3614 0.181 -0.890 -0.019 1.072

0 . . . .

4205 0.179 0.000 0.004 0.230

4205 -13.463 -14.523 -2.923 -0.718

2947 19.895 0.692 3.628 22.313

4205 -312.847 -248.369 -39.345 -7.766

4202 0.208 0.000 0.000 0.000

4205 0.008 0.000 0.000 0.000

3614 0.142 0.000 0.000 0.023

N Mean p25 p50 p75

2

6695 4.166 0.117 0.679 3.068

6695 -2.902 -3.523 -1.500 -0.521

6695 5.171 0.383 1.529 5.081

6241 0.214 -0.388 0.008 0.633

5451 0.218 -0.321 0.108 0.677

6695 0.217 0.019 0.093 0.323

6695 -4.507 -3.214 -1.079 -0.373

5421 6.150 0.254 0.790 2.741

6695 -93.615 -47.749 -14.193 -3.982

6690 0.145 0.000 0.000 0.064

6694 0.025 0.000 0.000 0.007

6240 0.139 0.000 0.014 0.083

N Mean p25 p50 p75

3

1931 22.610 3.610 9.917 24.158

1931 2.083 0.263 0.871 2.349

1931 15.415 2.838 7.857 19.895

1784 0.246 -0.039 0.103 0.393

1707 0.274 -0.004 0.140 0.400

1931 0.228 0.030 0.112 0.356

1931 0.066 0.014 0.073 0.164

1619 0.922 0.041 0.097 0.240

1931 -7.634 -1.014 0.941 2.638

1930 0.027 0.000 0.000 0.000

1931 0.014 0.000 0.000 0.006

1784 0.096 0.006 0.024 0.076

N Mean p25 p50 p75

Total

12831 5.576 0.000 0.219 2.980

12831 -1.880 -2.521 -0.777 -0.069

12831 5.770 0.182 1.208 5.340

11639 0.208 -0.406 0.033 0.662

7158 0.231 -0.201 0.122 0.582

12831 0.206 0.004 0.066 0.309

12831 -6.754 -4.182 -0.994 -0.186

9987 9.358 0.190 0.800 4.243

12831 -152.522 -63.213 -13.126 -2.284

12822 0.148 0.000 0.000 0.024

12830 0.018 0.000 0.000 0.000

11638 0.133 0.000 0.009 0.067

31

Table 2. Capital Composition: Frequency and Magnitude of Usage Firms in stage 1 has no sales; in stage 2, sales are positive but cash flow is still negative; in stage 3, cash flow is positive. Cash flow is measured with operating income plus depreciation (EBITDA). In Panel A, a value of 1 is assigned if a non-equity instrument is present in the firm’s capital structure; 0 otherwise. In Panel B, an instrument’s amount is scaled by total capital injection. Total capital injection = total debt + preferred equity + warrants + common equity – retained earnings (or losses). The number of observations in Panel B is less than Panel A because we exclude firms with negative total capital injection from the analysis. Trade credit, measured with the amount of accounts payable, is not a formal component of total capital, but we include it alongside other non-equity instruments for comparison because prior literature shows that trade credit is an important source of financing for entrepreneurial firms (Berger and Udell 1998). Panel A: Frequency (Yes: 1; No: 0)

Panel B: Magnitude (scaled by Total Capital Injection)

Stage

Debt

Preferred

Warrant

Trade Credit

Debt

Preferred

Warrant

Trade Credit

N Mean p25 p50 p75

1

4205 0.727 0.000 1.000 1.000

4205 0.149 0.000 0.000 0.000

4205 0.236 0.000 0.000 0.000

4205 0.891 1.000 1.000 1.000

4194 0.136 0.000 0.033 0.164

4194 0.008 0.000 0.000 0.000

4194 0.042 0.000 0.000 0.000

4194 0.076 0.004 0.018 0.057

N Mean p25 p50 p75

2

6695 0.875 1.000 1.000 1.000

6695 0.279 0.000 0.000 1.000

6695 0.339 0.000 0.000 1.000

6695 0.973 1.000 1.000 1.000

6658 0.161 0.013 0.070 0.207

6658 0.018 0.000 0.000 0.000

6658 0.057 0.000 0.000 0.021

6658 0.091 0.012 0.030 0.076

N Mean p25 p50 p75

3

1931 0.819 1.000 1.000 1.000

1931 0.209 0.000 0.000 0.000

1931 0.199 0.000 0.000 0.000

1931 0.980 1.000 1.000 1.000

1900 0.275 0.018 0.146 0.454

1900 0.013 0.000 0.000 0.000

1900 0.026 0.000 0.000 0.000

1900 0.227 0.023 0.068 0.195

N Mean p25 p50 p75

Total

12831 0.818 1.000 1.000 1.000

12831 0.226 0.000 0.000 0.000

12831 0.284 0.000 0.000 1.000

12831 0.947 1.000 1.000 1.000

12752 0.170 0.007 0.063 0.225

12752 0.014 0.000 0.000 0.000

12752 0.048 0.000 0.000 0.003

12752 0.106 0.010 0.029 0.081

32

Table 3. Debt Composition Panel A. Frequency of Usage Firms in stage 1 has no sales; in stage 2, sales are positive but cash flow is still negative; in stage 3, cash flow is positive. Cash flow is measured with operating income plus depreciation (EBITDA). This panel reports the frequencies of usage of various debt types and features. A value of 1 is assigned if a specific debt type/feature is present; 0 otherwise. In the left panel, we show the usages of seven mutually exclusive debt types according to Colla et al. (2013) and Capital IQ. Because there are no commercial paper observed in our sample, we report statistics on the remaining six debt types. In the right panel, we report the frequencies of usage of three binary debt features. Sr Bonds & Notes is senior bonds and notes. Sub Bonds & Notes is subordinate bonds and notes. Cap. Leases is capital leases. Short-term is short-term debt. Convertible (Secured) is debt with convertible (secured) feature.

Stage

Revolving Credit

Mutually Exclusive Debt Components Term Sr Bonds & Sub Bonds & Loans Notes Notes

Cap. Leases

Other Debt

Binary Debt Features ShortTerm Convertible Secured

N Mean p25 p50 p75

1

4205 0.063 0.000 0.000 0.000

4205 0.322 0.000 0.000 1.000

4205 0.477 0.000 0.000 1.000

4205 0.016 0.000 0.000 0.000

4205 0.026 0.000 0.000 0.000

4205 0.145 0.000 0.000 0.000

4205 0.609 0.000 1.000 1.000

4205 0.347 0.000 0.000 1.000

4205 0.210 0.000 0.000 0.000

N Mean p25 p50 p75

2

6695 0.229 0.000 0.000 0.000

6695 0.393 0.000 0.000 1.000

6695 0.670 0.000 1.000 1.000

6695 0.051 0.000 0.000 0.000

6695 0.173 0.000 0.000 0.000

6695 0.160 0.000 0.000 0.000

6695 0.668 0.000 1.000 1.000

6695 0.482 0.000 0.000 1.000

6695 0.536 0.000 1.000 1.000

N Mean p25 p50 p75

3

1931 0.354 0.000 0.000 1.000

1931 0.431 0.000 0.000 1.000

1931 0.480 0.000 0.000 1.000

1931 0.091 0.000 0.000 0.000

1931 0.271 0.000 0.000 1.000

1931 0.122 0.000 0.000 0.000

1931 0.469 0.000 0.000 1.000

1931 0.210 0.000 0.000 0.000

1931 0.646 0.000 1.000 1.000

N Mean p25 p50 p75

Total

12831 0.194 0.000 0.000 0.000

12831 0.375 0.000 0.000 1.000

12831 0.578 0.000 1.000 1.000

12831 0.046 0.000 0.000 0.000

12831 0.140 0.000 0.000 0.000

12831 0.149 0.000 0.000 0.000

12831 0.619 0.000 1.000 1.000

12831 0.397 0.000 0.000 1.000

12831 0.446 0.000 0.000 1.000

33

Table 3. Debt Composition Panel B. Magnitude of Usage This panel reports the magnitude of usage of various debt types and features. The amount of a specific debt type/feature is scaled by total debt.

Stage

Revolving Credit

Mutually Exclusive Debt Components Term Sr Bonds & Sub Bonds & Loans Notes Notes

Cap. Leases

Other Debt

Binary Debt Features ShortTerm Convertible Secured

N Mean p25 p50 p75

1

3043 0.030 0.000 0.000 0.000

3043 0.300 0.000 0.000 0.796

3043 0.540 0.000 0.714 1.000

3043 0.008 0.000 0.000 0.000

3043 0.010 0.000 0.000 0.000

3043 0.112 0.000 0.000 0.000

3043 0.682 0.191 1.000 1.000

3043 0.349 0.000 0.000 0.870

3043 0.182 0.000 0.000 0.145

N Mean p25 p50 p75

2

5767 0.089 0.000 0.000 0.004

5767 0.215 0.000 0.000 0.325

5767 0.572 0.053 0.718 1.000

5767 0.021 0.000 0.000 0.000

5767 0.040 0.000 0.000 0.000

5767 0.063 0.000 0.000 0.000

5767 0.482 0.005 0.414 1.000

5767 0.363 0.000 0.132 0.802

5767 0.382 0.000 0.208 0.823

N Mean p25 p50 p75

3

1562 0.204 0.000 0.000 0.322

1562 0.283 0.000 0.025 0.554

1562 0.335 0.000 0.071 0.757

1562 0.040 0.000 0.000 0.000

1562 0.090 0.000 0.000 0.017

1562 0.048 0.000 0.000 0.000

1562 0.319 0.000 0.057 0.693

1562 0.128 0.000 0.000 0.010

1562 0.614 0.125 0.814 1.000

N Mean p25 p50 p75

Total

10372 0.089 0.000 0.000 0.000

10372 0.250 0.000 0.000 0.452

10372 0.527 0.000 0.612 1.000

10372 0.020 0.000 0.000 0.000

10372 0.039 0.000 0.000 0.000

10372 0.075 0.000 0.000 0.000

10372 0.516 0.000 0.517 1.000

10372 0.323 0.000 0.000 0.747

10372 0.358 0.000 0.084 0.820

34

Table 3. Debt Composition Panel C. Debt Specialization and Debt Components as Percentage of Total Capital Injection HHI and Excl90 are debt specialization measures as defined in Colla et al. (2013). HHI is calculated in a way similar to the Herfindahl-Hirschman Index. Excl90 takes the value of 1 if one debt type accounts for at least 90% of total debt, and 0 otherwise. Bank debt = revolving credit + term loans. Bonds & Notes = senior bonds and notes + subordinate bonds and notes. Convertible (Secured) is debt with convertible (secured) feature. Total capital injection = total debt + preferred equity + warrants + common equity – retained earnings (or losses). Debt Specialization

Debt Component/Total Capital Injection

Stage

HHI

Excl90

Bank Debt

Bonds & Notes

Convertible

Secured

N Mean p25 p50 p75

1

3043 0.881 0.836 1.000 1.000

3043 0.767 1.000 1.000 1.000

4194 0.045 0.000 0.000 0.011

4194 0.075 0.000 0.000 0.057

4194 0.041 0.000 0.000 0.018

4194 0.025 0.000 0.000 0.000

N Mean p25 p50 p75

2

5767 0.780 0.542 0.915 1.000

5767 0.590 0.000 1.000 1.000

6658 0.056 0.000 0.000 0.034

6658 0.094 0.000 0.025 0.114

6658 0.052 0.000 0.000 0.055

6658 0.065 0.000 0.002 0.062

N Mean p25 p50 p75

3

1562 0.733 0.480 0.815 1.000

1562 0.521 0.000 1.000 1.000

1900 0.149 0.000 0.018 0.214

1900 0.094 0.000 0.003 0.097

1900 0.029 0.000 0.000 0.000

1900 0.172 0.000 0.041 0.269

N Mean p25 p50 p75

Total

10372 0.803 0.583 0.975 1.000

10372 0.631 0.000 1.000 1.000

12752 0.066 0.000 0.000 0.037

12752 0.088 0.000 0.010 0.097

12752 0.045 0.000 0.000 0.034

12752 0.068 0.000 0.000 0.049

35

Table 4. Total Debt This table reports how the amount of total debt, scaled by total capital injection, changes with a firm’s sales and cash flow positions. Columns 4-6 differ from 1-3 with firm fixed effects also included in the regressions. In columns 2 and 5, short-term debt is excluded. In columns 3 and 6, convertible debt is excluded. Total capital injection = total debt + preferred equity + warrants + common equity – retained earnings (or losses). Sale_Pos: 1 if total revenues>0; 0 otherwise. CF_Pos: 1 if cash flow>0; 0 otherwise. Cash flow is measured with operating income plus depreciation (EBITDA). For firms in stage 1, Sale_Pos=0, CF_Pos=0; stage 2: Sale_Pos=1, CF_Pos=0; stage 3: Sale_Pos=1, CF_Pos=1. Therefore, Sale_Pos (CF_Pos) measures the effect when firms transition from stage 1 to 2 (2 to 3). Other control variables are as defined in Table 1 Panel C. Industry and year fixed effects are included in all regressions. Standard errors are clustered on firm. T-statistics are shown in parentheses. ***, **, and * represents 1%, 5%, and 10% statistical significance respectively. Columns 4-6 have less observations than columns 1-3 because firms with only single period appearance are dropped from the analysis when firm fixed effects are included in the regressions. (1) Dep. Variable

Total Debt

(2) Total Debt (ex. ST)

Sale_Pos

0.020*** (2.86) 0.103*** (8.61) -0.006*** (-2.79) 0.013*** (5.84) 0.101*** (7.14) -0.001* (-1.73) -0.000 (-0.49) 0.000*** (3.40) -0.023*** (-3.10) -0.015 (-0.33) -0.002 (-0.20)

0.016*** (3.09) 0.076*** (7.87) 0.010*** (6.26) 0.001 (0.89) 0.076*** (6.35) -0.000 (-0.06) -0.000 (-0.00) -0.000*** (-2.78) -0.009* (-1.72) -0.011 (-0.36) -0.002 (-0.22)

0.012* (1.94) 0.119*** (10.41) -0.006*** (-2.85) 0.008*** (3.98) 0.097*** (7.23) -0.000 (-1.51) 0.000 (0.13) 0.000*** (3.44) -0.030*** (-4.99) -0.057 (-1.37) -0.006 (-0.70)

0.023*** (3.70) 0.029*** (3.80) -0.000 (-0.03) 0.007*** (3.32) 0.048*** (3.94) 0.000* (1.87) -0.000 (-0.01) -0.000 (-1.47) 0.009* (1.72) 0.010 (0.25) 0.011 (1.21)

0.019*** (3.63) 0.027*** (4.14) 0.005** (2.38) 0.002 (1.18) 0.030*** (3.06) 0.000* (1.88) 0.000 (0.02) -0.000** (-2.29) 0.004 (1.16) -0.018 (-0.71) 0.008 (1.18)

0.017*** (2.97) 0.028*** (3.69) -0.005* (-1.89) 0.006*** (3.13) 0.048*** (4.28) 0.000 (0.67) 0.000 (0.14) 0.000 (0.21) -0.003 (-0.80) 0.023 (0.58) 0.009 (1.12)

9,936 0.102 No

9,936 0.142 No

9,936 0.122 No

9,354 0.723 Yes

9,354 0.717 Yes

9,354 0.716 Yes

CF_Pos Log(AT) ATG NPPEA EBITA EBITA_VOL ZSCORE RDA ADA CAPEXA

Observations R-squared Firm FE

(3) Total Debt (ex. CONV)

(4)

36

Total Debt

(5) Total Debt (ex. ST)

(6) Total Debt (ex. CONV)

Table 5. Debt Types and Features This table reports how different types and features of debt, scaled by total capital injection, change with a firm’s sales and cash flow positions. Column 1 in Panel A (B) is same as column 1(4) in Table 4 to facilitate comparison. We use total capital injection, not total debt, as the scalar to capture the dynamics when firms move from no debt to some debt use. Bonds & notes = senior bonds and notes + subordinate bonds and notes. Bank debt = revolving credit + term loans. Convertible (Secured) is debt with convertible (secured) feature. Sale_Pos: 1 if total revenues>0; 0 otherwise. CF_Pos: 1 if cash flow>0; 0 otherwise. Cash flow is measured with operating income plus depreciation (EBITDA). For firms in stage 1, Sale_Pos=0, CF_Pos=0; stage 2: Sale_Pos=1, CF_Pos=0; stage 3: Sale_Pos=1, CF_Pos=1. Therefore, Sale_Pos (CF_Pos) measures the effect when firms transition from stage 1 to 2 (2 to 3). Other control variables are as defined in Table 1 Panel C. Industry and year fixed effects are included in all regressions. Standard errors are clustered on firm. T-statistics are shown in parentheses. ***, **, and * represents 1%, 5%, and 10% statistical significance respectively.

Panel A. No Firm Fixed Effects Dep. Variable Sale_Pos CF_Pos Log(AT) ATG NPPEA EBITA EBITA_VOL ZSCORE RDA ADA CAPEXA

Observations R-squared

(1) Total Debt

(2) Bonds & Notes

(3) Bank Debt

(4) Convertible

(5) Secured

0.020*** (2.86) 0.103*** (8.61) -0.006*** (-2.79) 0.013*** (5.84) 0.101*** (7.14) -0.001* (-1.73) -0.000 (-0.49) 0.000*** (3.40) -0.023*** (-3.10) -0.015 (-0.33) -0.002 (-0.20)

0.016*** (3.00) 0.004 (0.58) -0.001 (-1.05) 0.008*** (4.74) 0.038*** (3.80) 0.000 (0.27) -0.000 (-1.01) 0.000 (0.11) 0.001 (0.14) 0.007 (0.20) 0.001 (0.11)

0.007* (1.72) 0.081*** (8.40) -0.002 (-1.12) 0.005*** (3.31) 0.040*** (3.96) -0.001*** (-3.61) 0.000 (0.80) 0.000*** (5.36) -0.019*** (-6.07) -0.016 (-0.52) 0.002 (0.35)

0.009*** (2.63) -0.017*** (-5.21) 0.000 (0.41) 0.004*** (3.51) 0.003 (0.69) -0.000 (-1.29) -0.000 (-1.26) 0.000 (0.81) 0.007* (1.66) 0.040* (1.68) 0.003 (0.79)

0.011*** (2.89) 0.077*** (8.27) 0.013*** (11.73) -0.000 (-0.26) 0.060*** (5.86) -0.000*** (-3.20) -0.000 (-1.12) -0.000*** (-4.49) -0.005 (-1.01) 0.036 (1.16) 0.001 (0.20)

9,936 0.102

9,936 0.042

9,936 0.097

9,936 0.035

9,936 0.183

37

Table 5. Debt Types and Features Panel B. Firm Fixed Effects Panel B differs from Panel A with firm fixed effects also included in the regressions. Panel B has less observations than Panel A because firms with only single period appearance are dropped from the analysis when firm fixed effects are included.

Dep. Variable Sale_Pos CF_Pos Log(AT) ATG NPPEA EBITA EBITA_VOL ZSCORE RDA ADA CAPEXA

Observations R-squared

(1) Total Debt

(2) Bonds & Notes

(3) Bank Debt

(4) Convertible

(5) Secured

0.023*** (3.70) 0.029*** (3.80) -0.000 (-0.03) 0.007*** (3.32) 0.048*** (3.94) 0.000* (1.87) -0.000 (-0.01) -0.000 (-1.47) 0.009* (1.72) 0.010 (0.25) 0.011 (1.21)

0.011** (2.07) 0.010 (1.54) 0.002 (0.87) 0.004*** (2.70) 0.020** (2.20) 0.000 (1.48) -0.000 (-0.32) -0.000* (-1.78) 0.006 (1.36) 0.002 (0.06) 0.002 (0.32)

0.012*** (3.39) 0.013** (2.09) 0.001 (0.45) 0.003* (1.92) 0.023*** (3.10) 0.000 (0.73) 0.000 (1.42) -0.000 (-0.42) 0.002 (0.78) 0.024 (0.96) 0.009 (1.52)

0.006** (2.00) -0.000 (-0.12) 0.005*** (3.73) 0.000 (0.14) 0.002 (0.46) 0.000* (1.83) -0.000 (-0.14) -0.000*** (-2.66) 0.010*** (2.79) -0.019 (-0.76) 0.002 (0.41)

0.012*** (2.85) 0.017*** (2.90) 0.009*** (5.98) 0.001 (0.58) 0.024*** (2.98) -0.000 (-1.01) 0.000 (0.31) -0.000*** (-3.96) 0.004 (1.61) 0.037 (1.53) 0.011** (2.23)

9,354 0.723

9,354 0.650

9,354 0.709

9,354 0.611

9,354 0.719

38

Table 6. Debt Specialization This table reports how debt specialization changes with a firm’s sales and cash flow positions. HHI and Excl90 are debt specialization measures as defined in Colla et al. (2013). HHI is calculated in a way similar to the HerfindahlHirschman Index. Excl90 takes the value of 1 if one debt type accounts for at least 90% of total debt, and 0 otherwise. Columns 3-4 differ from columns 1-2 with firm fixed effects also included in the regressions. Sale_Pos: 1 if total revenues>0; 0 otherwise. CF_Pos: 1 if cash flow>0; 0 otherwise. Cash flow is measured with operating income plus depreciation (EBITDA). For firms in stage 1, Sale_Pos=0, CF_Pos=0; stage 2: Sale_Pos=1, CF_Pos=0; stage 3: Sale_Pos=1, CF_Pos=1. Therefore, Sale_Pos (CF_Pos) measures the effect when firms transition from stage 1 to 2 (2 to 3). Debt/Cap = total debt/total capital injection. Other control variables are as defined in Table 1 Panel C. Industry and year fixed effects are included in all regressions. Standard errors are clustered on firm. T-statistics are shown in parentheses. ***, **, and * represents 1%, 5%, and 10% statistical significance respectively. Columns 3-4 have less observations than columns 1-2 because firms with only single period appearance are dropped from the analysis when firm fixed effects are included in the regressions.

VARIABLES Sale_Pos CF_Pos Debt/Cap Log(AT) ATG NPPEA EBITA EBITA_VOL RDA

Observations R-squared Firm FE

(1) HHI

(2) Excl90

(3) HHI

(4) Excl90

-0.052*** (-5.76) 0.003 (0.22) -0.129*** (-6.51) -0.017*** (-8.15) 0.008*** (3.14) -0.036** (-2.51) 0.001*** (4.03) -0.000 (-0.72) 0.026*** (2.81)

-0.093*** (-5.33) 0.019 (0.87) -0.202*** (-5.46) -0.029*** (-7.31) 0.012** (2.41) -0.056** (-2.00) 0.002*** (3.49) -0.000 (-0.35) 0.042** (2.24)

-0.029*** (-2.63) 0.006 (0.50) -0.108*** (-3.91) -0.013*** (-3.68) 0.002 (0.57) -0.031* (-1.87) 0.001** (2.55) 0.000 (0.54) -0.014 (-1.27)

-0.062*** (-2.79) 0.018 (0.80) -0.193*** (-3.60) -0.023*** (-3.27) 0.001 (0.25) -0.048 (-1.39) 0.001* (1.93) 0.000 (0.43) -0.044* (-1.83)

8,244 0.091 No

8,244 0.066 No

7,653 0.559 Yes

7,653 0.502 Yes

39

Appendix A. The Over-the-Counter (OTC) Market The OTC market comprises two main venues: the OTC Bulletin Board (OTCBB) and the Pink Sheets.14 Most firms have a market value below $20 million and are “penny stocks” with share prices below $5.00. They concentrate in technology, biotech, and service industries such as financial, pharmaceutical, computing software and media industries (Brüggemann et al. 2016). The Penny Stock Reform Act of 1990 required the SEC to establish a trading venue for small cap stocks, which gave birth to the OTCBB. It is an electronic interdealer quotation system that is currently owned and regulated by Financial Industry Regulatory Authority (FINRA). Unlike the main exchanges (NYSE, Nasdaq, Amex), the OTCBB does not have minimum size or governance requirements. Since year 2000, all non-financial firms quoted on the OTCBB are required to file audited annual reports, i.e. Form 10-K, with the SEC (“the Eligibility Rule”; Bushee and Leuz 2005).15 The Pink Sheets, established as early as the 1910s, compete with the OTCBB in quoting stocks that do not qualify to be listed on main exchanges. It obtained its name because stock quotations were originally displayed in pink paper sheets. Currently, Pink Sheets are owned and operated by the OTC Markets Group (“OTCMG”). Under Section 12(g) of the Securities Exchange Act of 1934, firms with total assets greater than $10 million and more than 500 shareholders are required to file audited annual reports with the SEC. Such criteria were further relaxed to 2,000 shareholders under the JOBS Act in 2012. Hence we were able to obtain audited financial statements for pink sheet firms under Section 12(g) and for all OTCBB firms since 2000.

14

Brüggemann et al. 2016 provides institutional background. The study also documents a grey market that features stocks that are not traded on any venue. We do not consider this category as they do not file with the SEC. 15 Small firms can elect to file Form 10-KSB or 10-KSB40. 40