funding gaps: a symposium - CiteSeerX

4 downloads 1596 Views 70KB Size Report
venture capital trusts, soft loan schemes, government grants, government equity ..... Berger and Udell's contribution to this symposium, “Small Business, Credit.
FUNDING GAPS: A SYMPOSIUM 1 By

Robert Cressy City University Business School London, England Email: [email protected]

ABSTRACT The theoretical and empirical foundations of government policies designed to plug alleged business funding gaps are highly controversial but rarely subject to wide-ranging and in-depth debate. This symposium from major scholars in the field provides a counterbalance to this tendency. Topics addressed cover the theory of lending under asymmetric information and its implications for overlending; relationship lending as a market solution to credit market information asymmetries; government emulation of private sector decision-making to eliminate underprovision of high tech equity; and theoretically-based empirical work testing for funding deficiencies in the high tech sector. Despite the very different and potentially contradictory contributions to the debate the result is, surprisingly, a set of mutually agreed policy conclusions.

I. Introduction Do firms, especially small, young firms, get enough money to fund viable projects? If, not, what, if anything, should be done about it? These questions are beguilingly simple to ask, and, needless to say, much less easy for economists to answer. However, to gauge the importance of doing so, we should appreciate that huge sums of money (many billions of dollars) are spent annually in attempts by governments across the globe to plug alleged debt and equity gaps in national economies (Cressy, 2000). Thus we find a growing proliferation of loan and equity guarantee schemes, venture capital trusts, soft loan schemes, government grants, government equity investment, and so on, with a focus on small, young, often high tech firms perceived to be the main victims of such funding shortages. In geographical terms these schemes 1

I am indebted to the Editor, Steven Machin, two referees of The Economic Journal, contributors to the Warwick conference on International Funding Gaps Controversies and the other Symposium contributors for helpful comments. David de Meza in particular has provided me with a number of theoretical insights into the debate that follows whilst Robert Carpenter and Bruce Petersen both

1

have traditionally been found in Western Europe, the United States and Japan, but their use has now percolated through to the developing countries of Eastern Europe, South America, Asia and Africa who often see small firms as a solution to the development or unemployment problem.

Is there general agreement amongst economists on the need for these interventions and if so, a consensus on the type of interventions required? The answer on both accounts is, unfortunately, No. The concept of a funding gap is by no means straightforward, as exchanges with symposium contributors amply demonstrate. Two approaches emerge from the debate which we term the normative and the positive. Funding Gap (positive or ‘P’ definition). An equilibrium, in which the volume of lending is below that which would emerge in a competitive capital market with costless and complete contracting, no private information and rational expectations. Funding Gap (normative or ‘N’ definition). A market failure, the appropriate policy response to which is an increase in the volume of lending. As the symposium makes clear, for N to hold it is necessary but not sufficient that P applies 2 . Furthermore, as we shall see, P is consistent with, but does not imply, a related outcome, namely, credit rationing, a situation in which there is excess demand for loans and the market for fails to clear.3 As regards policy making, an important

offered detailed comments on earlier drafts. In the last resort, however, none of these individuals may have saved me from myself, as both they and the general reader may now judge. 2 For example, suppose that it is costly for banks to monitor borrower behaviour so that problems of moral hazard are inevitable. As a result banks do not lend to a certain class of loan applicant. Whilst subsidising lending may bring its volume closer to the level prevailing were information problems absent, since the policy cannot eliminate moral hazard it may nevertheless lower efficiency. de Meza’s paper in this symposium highlights such possibilities. 3 This situation, arising from adverse selection, was first identified as a theoretical possibility, by Stiglitz-Weiss, 1981; henceforth SW. Here, the supply curve of funds is backward-bending and lies everywhere to the left of the demand curve. Under competitive market conditions with symmetric information the supply curve would be horizontal at the opportunity cost of bank funds and excess demand would be eliminated in the usual way. In SW theory the excess demand for loans is not

2

observable feature of a funding gap is that the deficit (or, as we shall see, surplus) is a persistent rather than a temporary phenomenon4 . Whilst these definitions are clear, so controversial is the funding gaps literature that papers sometimes arrive at exactly opposite conclusions, as attested to by the de Meza-Webb (1987) response to the theoretical models of Stiglitz- Weiss(1981) (a surplus versus a deficit of funds), recent interchanges in the Quarterly Journal of Economics5 (empirical evidence for US funding deficits) and the conference on Funding Gaps Controversies held at Warwick Business School’s Centre for Small and Medium Enterprises in 1999 (both theoretical and empirical disagreements). There is moreover, dispute as to what (if anything) should be done about deficits or surpluses should they exist. The papers in the present Symposium arose from the Warwick conference, organised and Chaired by myself. Four different criteria for the identification of funding deficits were found to be used by economists in practice: (i) the labour market state switching criterion where, roughly, a positive relation between individual assets and self employment decision is taken to indicate the existence of a debt gap (see the seminal paper by Evans and Jovanovic,1989) 6 ; (ii) the net worth -investment criterion where (controlling for demand influences) a positive relationship between companies’ cashflow and their investment decisions is taken to indicate the existence of a ‘wedge’ between the cost of internal and external funds, thereby suggesting the factoring in of agency costs by fund providers (see the seminal paper of Fazzari, Hubbard and

eliminated as adverse selection provides banks with a disincentive to raise interest rates. In such a situation, credit is allocated by quota. 4 A temporary ‘gap’ would, for example, be associated with a disequilibrium under competitive conditions as markets adjust to some unpredictable shock to the system. This would not be a market imperfection and be no cause for intervention. 5 See Kaplan and Zingales(1997), Fazzari, Hubbard and Petersen(2000), Kaplan and Zingales(2000). 6 Papers in this area were presented at the conference but the issues are not discussed in this Symposium. The reader interested in other papers from the conference may contact the author via the email address on the title page of this Introduction.

3

Petersen, 1988 hence, forth FHP); (iii) the Euler equation approach, based on structural estimation of a firm’s capital stock equations, to test for deviations from the classical competitive model of capital markets 7 ; and, simplest of all, (iv) the selfassessment criterion (e.g. “Do limitations on availability of funds influence your decision to startup/invest/etc.”), used in several studies (see for example Aston Business School, 1990; Egeln, Licht and Steil, 1997; Kaplan and Zingales(1997, 2000).) In debating the validity of these criteria, the conference identified some market and government-generated solutions to rationing issues and evaluated their effectiveness. In the case of perceived funding deficits, these included the evolution of long-term relationships between firms and creditors (“relationship banking”) and the role of government as venture capitalist. The current Symposium represents, then, (a) a summary and evaluation of some important issues in the theory of funding under asymmetric information; (b) a summary of the theory and empirical work on relationship lending together with some new theoretical analysis; (c) an empirical examination of the existence and nature of funding gaps in the US high tech sector; (d) an evaluation of the role of government as provider of equity support to high tech firms in America; and (e) an attempt to arrive at some overall policy conclusions based on the analysis, evidence and debate. Before we move on to a detailed discussion of the controversies arising from the Symposium it is worth reflecting on the nature of the debate that follows. Firstly, of course, it is fundamentally a debate about the existence or non-existence of funding gaps and their conditional extent and location8 . Since our interpretation of the evidence is the only way of deciding whether there is a funding gap, the debate is both 7

This is an alternative method to estimating an investment demand function and uses the Euler equation derived from the firm’s dynamic optimisation problem rather than the investment demand function used in FHP. See the excellent survey of Hubbard(1998) for details and references.

4

about interpretation of the evidence and about existence and (conditional) extent of gaps. Secondly, the debate is sometimes about the appropriate empirical methodology dataset or criteria for testing a particular theoretical model. Thirdly it is sometimes about the a priori plausibility of one theoretical model against another. Finally, there is the debate about appropriate policy measures (if any) that should be adopted in the light of the evidence and its interpretation. Enjoy!

II. The theoretical debate The opening salvo by David de Meza, provocatively entitled “Overlending?”, sets the theoretical scene for the debate. Summarising history, prior to de Meza-Webb(1987) the funding deficit theory of Stiglitz-Weiss (1981) had remained unchallenged. SW had argued that if firms’ projects have the same mean return but differ in borrower quality (probability of project success), observable only to the firm, banks could end up supplying less credit than firms desire and that there would be no tendency for the market to resolve this problem over time. The price of credit (the interest rate) would not rise to equate demand and supply as higher interest would be paid only by the high risk (low quality) borrowers whose projects had the higher return. This would in turn reduce the bank’s profits and turn breakeven into a loss. This analysis seemed to identify a classic example of adverse selection in credit markets and the result was a funding deficit. 9 However, De Meza-Webb (1987) (henceforth DMW) upset the SW applecart by showing that a small change to their assumptions (namely, allowing the mean return more realistically to vary across firms) could completely reverse the results, with the market then characterised by credit surplus rather than deficiency.

8

By location I mean, for example, whether they affect a certain class of firm, a section of the economy, a particular country or a certain stage of economic development. 9 In this case, demand and supply do not intersect and rationing occurs.

5

De Meza, in this symposium contribution, argues there is now ample empirical evidence that individuals with low net worth or from disadvantaged groups are excluded from capital markets. Whilst these might be explained by ‘standard theories of lending under asymmetric information’ (i.e. the SW logic discussed above) he argues that ‘more plausible’ models of the credit market suggest that ‘curbing lending [my italics] would raise enterprise quality sufficiently to yield social benefits.’ This implied surplus of credit could happen if the bank could not distinguish good projects from bad and so priced all at average quality. Good projects would then subsidise bad. The marginal borrower 10 being below average quality would then yield a loss to the bank. De Meza’s argument for surplus is strengthened by adding the ingredient of borrower optimism about own-project quality. Optimism is the much-documented human tendency to overestimate one’s chances of success - apparently an affliction of the younger car driver11 and the younger entrepreneur. In de Meza, it is a theoretical construct responsible for a further lowering of the average quality of the borrowing pool, as the marginal borrower now makes her entry decision based on the same perceived quality but one which is now an overestimate of the true quality12 . The model is robust to a number of variations in its underpinnings. Allowing for moral hazard and the correlation of ability with project risk, for example, leaves the overlending conclusion unchanged. In particular, allowing variations in ability to (positively) influence both return and risk the marginal borrower will now be defined in terms of a cutoff ability level with entrepreneurs above the threshold borrowing from the banks. An increase in wealth will require that the marginal borrower be 10

The marginal borrower is defined as an entrepreneur who is just breaking even on his project when borrowing to fund it. 11 Absenting cross-subsidisation, car insurance premiums are higher for younger drivers. This is partly a result of their relative optimism about prospects of having an accident.

6

charged a lower interest rate to keep his profits at zero 13 . This in turn will lower the cutoff ability level (positively correlated with return net of interest) and generate entry. However, lower ability means higher risk and lower profits to the bank, and so average ability in the borrower pool will decline. Since bank pricing of loans is at average ability the marginal borrower will be loss-making to the bank, once again implying too many entrepreneurs and too much lending. Thus de Meza controversially (but in fact in line with the empirically drawn conclusions of Cressy, 1996b 14 ) argues that the exclusion of borrowers at the lower end of the wealth spectrum is not a bad thing: they are signalling their low quality15 . He therefore proposes either the subsidisation of non-entrepreneurs or taxation of entrepreneurs that will curb excessive lending to these low quality, overoptimistic borrowers. What are policy-makers to make of this analysis? How realistic are these models for policy decisions? How does the analysis sit with the other papers in the symposium? Firstly, the realism of the models. Theoretically, the de Meza-Southey (DMS) model of the optimistic entrepreneur is based on the assumption that there are no costs of 12

Another way of stating this is that optimistic would-be borrowers believe they have a lower cost capital than realistic borrowers. Hence they continue to apply for funds when a realist with the same chances of success would have regarded the application as uneconomic. 13 This change is required because the return on wealth not invested in the business constitutes an opportunity cost of entrepreneurship. 14 My conclusions were based also on the role of human capital in both the lending decision (as a determinant of asset levels used in collateral) and in business survival, and therefore follow a rather different kind of reasoning. The logic is that human capital predicts asset levels, bank funding and survival of the typical (not fast growth) UK startup. Firms get the funding they ‘need’, determined by the bank’s correct perception of their worth via the observable human capital in the business (this influences the likely length of the relationship). Funding is not excessive or deficient unless there is, e.g., intervention of government, thereby distorting the market process. Restrictions on entry into selfemployment would be worthwhile if government-funded training schemes raised human capital and therefore survival sufficiently to offset the cost to the exchequer without crowding out other entrepreneurs. In this model firms run by (on average) younger entrepreneurs will get less money than those run by older ones because the loan officers are realists (in de Meza’s sense) and firms’ contracts are not influenced by the optimism of the younger borrowers. 15 Although de Meza’s paper does not assume a distribution of wealth amongst the population of potential borrowers, or a correlation of wealth and ability, it is easy to show that the model’s results hold up when these assumptions are relaxed, e.g., when wealth and ability are (as in Cressy, 1996) perfectly positively correlated.

7

bankruptcy associated with debt. Once we introduce this real-world phenomenon into the picture it creates a deadweight loss that both firm and bank will wish to avoid. One way this can be done is by the use of equity, since equity has no bankruptcy costs associated with it. If, then, there is an outside equity provider the firm will prefer (cet. par.) to take funds from her than borrow from the bank. Of course this all-equity funding is not observed in practice - except for fast growth businesses with significant capital gain associated with the investment – see the discussion of Carpenter and Petersen below. The dearth of external equity may also result partly from the existence of adverse selection, as poor quality businesses are more likely to offer equity than high quality ones 16 . However, this argument shows, I believe, that the optimism models are less than robust to relatively small (and realistic) variations in their assumptions, just as are the Stiglitz-Weiss models were found to be. Looking at the empirical evidence for DMW versus SW, de Meza himself cites two pieces of empirical evidence against the SW theory. (i) If the market is biased, risky rather than safe projects go unfunded. Black, de Meza and Jeffreys (1996) are cited as evidence. They found that UK firms starting in periods of high interest rates have better, rather than worse, survival rates, thus suggesting that ability is the most important form of heterogeneity17 . (ii) SW implies that the equilibrium mode of financing is equity rather than debt. The facts, however, show debt to be by far the largest source of external finance for SMEs 18 . He argues that in most cases such debt will also have been supplied after scrutiny designed to eliminate asymmetric information. Therefore, this counts as evidence against SW and for DMW. 16

Still other reasons include the fact that most SME owners are averse to losing control of their businesses. More on this later. 17 This finding also has support from Cressy(1996b) who established, using a large sample of UK startups, that ability (human capital in the form of experience and team size) is the primary determinant of startup survival. Cressy(1993) also found that the quality of startups in a recession was higher than that in a boom.

8

However, de Meza also recognises that the empirical argument is not conclusive as there are other theories of debt that can be ‘grafted on’ to the hidden types-model and would yield debt financing even if heterogeneity were to reflect risk. 19 Another approach to assessing the realism of the models is to examine the implications of SW and DMW and to assess their descriptive accuracy. One assumption common to both classes of model is that they claim the existence of a credit market pooling equilibrium, or in layman’s terms, that there is no difference in the interest rate and other charges for credit offered to borrowers of different project quality20 . This common conclusion seems inconsistent with the UK empirical evidence since the spread of small business interest rates is larger than one would expect if banks really could not distinguish borrower types 21 . On the other hand, such a spread of rates does not prove that rate formation is competitive i.e. actuarially fair and based on borrower types, as would be the case under symmetric information22 . An empirical point in favour of the DMW models, however, is that they do not assume that projects differ in a mean-preserving spread of returns. There is no empirical 18

See the discussion of Carpenter and Petersen below. Examples of such alternative explanations can be found in Townsend (1979) and Gale and Hellwig (1985), where debt functions to economise on costly state verification. 20 This is a simplification and is not true of all models discussed by de Meza, for example, of the random interest rate model which he argues is the true outcome of SW assumptions. De Meza’s later model assumes a distinction between borrower quality (ability) and project risk. 21 Cressy and Toivanen (1998) (henceforth CT) in a large sample study of UK start-up term loan conditions in the early 90s found that even after controlling for the industry of the entrepreneur there was ‘wide variation’ (in some well-defined sense) in loan contract terms. For example, 68.9 % of the industries had wide variation in the interest rates charged on loans. In fairness to de Meza, who is discussing primarily start-up businesses, it should be mentioned that some of these loans are to established smaller businesses. It is also true that the wide variation in contract terms we observe is consistent with variation in unobservable characteristics of borrowers and so does not of itself imply a first best regime. This is in fact tested for in CT by deriving a theoretical relationship that holds only under a symmetric information regime. Finally, if relationship banking operates as described in Berger and Udell’s contribution to the Symposium, we should expect a higher correlation of quality and pricing of loans for the CT sample than for one based on start-up businesses. These points perhaps underline the importance of dataset differences in explaining divergent empirical results. Unfortunately, much of the theoretical literature seems not to distinguish businesses along the business development (age) dimension. 22 There is some evidence that this is the case. CT found that better quality small business borrowers got lower interest rates and larger loans, a finding consistent with a symmetric information or first best 19

9

evidence whatsoever to suggest that projects follow this pattern, although it is difficult to assess empirically the gross revenue and success chances of projects rather than those of loans or firms 23 . The DMW models of surplus lending incorporating moral hazard as well as heterogeneous ability and optimism also have significant real-world resonance. From one of the few studies examining the empirical characteristics of UK small firm loans (Toivanen and Cressy, 1999) such loans seem in practice to be subject to moral hazard rather than adverse selection. 24

III. A market response to gaps: relationship lending Berger and Udell’s contribution to this symposium, “Small Business, Credit Availability and Relationship Lending”, examines the banking sector’s response to the informational opaqueness of the small, young firm and its consequent disadvantage in credit markets. Their paper builds on the seminal empirical work of Petersen and Rajan (1994) and their own empirical findings (e.g. Berger and Udell¸1995). Petersen and Rajan found that a major contribution of relationship lending lay in the increased availability of credit to customers with longer links to their bank, thus ameliorating any tendency to credit rationing of those customers. In this paper Berger and Udell (henceforth BU) examine in depth the role of long-term relationships between lender and borrower in ameliorating credit market deficiencies. Lending activity by the banks is divided into two principal kinds. The first, transactions-based lending, is where the evaluation of the loan application is made over a short space of time and is based on quantitative criteria (balance sheet and income statement information, SME credit scoring etc.). The second, relationship-

regime. More robust tests of exogeneity in Toivanen and Cressy(1999) still rejected the presence of adverse selection but indicated the influence of moral hazard on the lending contract. 23 The literature usually assumes that the firm and project are identical and, in the absence of moral hazard, that the quality of the entrepreneur is identical with the quality of the project. Whilst this is credible for small, especially start-up firms, it is much less so for medium-sized or large firms.

10

based lending, is one in which a loan decision is based on information that takes time to accumulate and is relies on qualitative facts that are both relationship-specific and difficult to transfer or communicate to others (including members of the bank). Banks use the transactions ‘lending technology’ in dealing with informationally transparent, larger, more established firms with whom they can conduct a more arms-length relationship and where monitoring plays a relatively limited role. Banks employ relationship lending technology for dealings with smaller, and younger firms who are more informationally opaque and where the association is closer and more personal. The repository of information on specific relationships is then the bank’s loan officer. She will have built up a network of local contacts via shared interests, membership of clubs, boards, social committees, etc, that will provide her with qualitative data on the entrepreneur’s performance in other areas of activity, but which also yields information about behaviour under certain conditions relevant to her bank’s lending decisions. However, the very intimacy of this relationship and the non-transferability of the data extracted means that she also has scope for being compromised by it. Since her remuneration is partially a function of the volume of lending, she may even overlend 25 to customers within her network. She may also be tempted to overlook warning signs on problematic loans, especially if by doing so this may increase her job opportunities with the firm under her umbrella. All this, BU argue, requires a bank 24

Coupled with this, however, Toivanen and Cressy(1999) found that superior bargaining power on the behalf of the bank seems also to play an important role in the determination of contract characteristics. 25 Such over-lending does not arise because of considerations discussed by de Meza in the present Symposium. His argument for over-lending is based on considerations of informational asymmetry between banks and firms - banks know less than firms about the latter’s prospects and so offer contracts that do not depend on their being able to observe characteristics, qualitative or otherwise. This combined with the tendency to moral hazard and optimism on the behalf of potential borrowers generates too much lending. Over-lending in BU’s model by contrast arises from a divergence of interest between the loan officer and her superiors in the bank . It does not arise from the standard information asymmetry discussed above. Indeed, she may be very well informed about both entrepreneur and firm after 5-10 years of dealing with the entrepreneur’s network. de Meza’s theories in this Symposium can be reconciled with BU’s position by regarding them as a description of lending

11

organisational structure that puts checks and balances in place against abuse by lending officers. Banks significantly involved with relationship lending will therefore be structurally disparate from those that are involved in primarily transactions-based lending. Among other things, this disparity will reflect the need for closer need for monitoring of loan officer activities. In this paper BU also identify for the first time a hierarchy of principal-agent relationships in the lending decision starting with the loan-officer-firm relationship and ending with the stockholder-bank regulator interaction. Thus in the chain of relationships: firm-loan officer-senior management-shareholders-regulators, each link has its associated agency costs, principally costs resulting from the monitoring of activities at the next lowest level. Their main conclusion is that the phenomenon of relationship banking is an important source of mitigation of informational opaqueness and thus should ameliorate potential funding deficiencies 26 . However, because of the agency dimension, relationship banking is also expensive to deliver, and is substantially affected by changes to the legal environment, financial policy, macroeconomic shocks and so on. So does the widespread practice of relationship banking dispose of the problem of credit deficits or rationing? 27 . Once again, things are not so simple. The concept of

behaviour within a rather different market niche: that of entirely new ventures . Under this interpretation, the learning process crucial to relationship banking has not yet been initiated. 26 Referring to supply curve of funds described in FHP and Carpenter and Petersen (this Symposium), the effect of relationship banking is to flatten the upward sloping part of the curve and thus reduce the wedge between the cost of internal and external funds. 27 In a simple model it can be shown that relationship banking will eliminate any initial information asymmetry with the result that firms will never switch banks, and all banking is therefore ‘relationship banking’. The argument goes as follows. Suppose there are two types of borrowers, High and Low, with qualities (success probabilities) p H > p L . Each project has a two-point distribution of returns, as usual, and requires $1 to get off the ground. Borrowers join a bank now with the prospect of borrowing in 1 year’s time to fund their project. Hs and Ls occur with equal probability in the borrower population. To fix ideas suppose that pH =3/4 and p L =1/4. There are n competitive banks each with opportunity cost of lending r. Wealth of borrowers, and therefore collateral, is zero. Banks know the distribution of borrower quality and can learn about their own borrowers’ quality after lending to them. Under complete information on borrower types the bank prices according to the rule α = r / p where r is

12

relationship lending, as we have seen, presupposes the existence of a learning process by the bank with respect to borrower characteristics, and one established over an extended period of time - an average of some 9.4 years, according to BU on their American data. For policy purposes we therefore need to know what proportion of relationships are still on the lower reaches of the learning curve. Clearly, the greater the proportion of new borrowers coming on to the market, and the greater the proportion of entrepreneurs switching banks, the greater the proportion of early-stage relationships. However, Carpenter and Petersen in this Symposium (henceforth CP) argue that relationship lending may not work well as a solution to funding shortage in the hightech sector of the economy. Because of the dynamic nature of the high-tech sector new firms may have to obtain financing at a very young age. CP point to evidence that firms are typically only a few years old when their receive their first equity financing, a time frame much shorter than the average lending relationship (9.4 years) reported in BU’s dataset. Thus there appears to be an argument for the existence of a funding deficiency based on the trio of: relationship shortness, lack of track record

the bank’s cost of capital and α the interest rate it charges on the risky loan. This yields α H = 4r / 3 and α L = 4r . In the present, borrowers are allocated randomly to banks. In one period’s time the firm’s own bank learns with certainty their quality whilst other banks know only the average quality of the borrower pool. It would seem that the Hs would stay with their own banks whilst the Ls would prefer to search, since the latter’s return to switching would be a lower interest rate α L/ = 2r < 4r . However, the fact of switching conveys information about borrower type: only the Ls switch. Hence the welcoming bank instead of charging 2r will now charge 4r to the new applicant - just as did the firm’s own bank. This means that both the Hs and Ls in fact stay put and all lending is done on a relationship basis. The length of relationship a borrowing firm has with its bank (namely the time taken to gain perfect knowledge about the borrower) is arbitrarily set here at 1 period. We have also assumed that the cost to the bank of learning the information about borrower quality is zero, but making it a fixed cost per $ lent simply alters the bank’s cost of funds from r to r+f and interest rates to (r+f)/p. All else remains the same. If the distribution of borrower quality is made continuous we get exactly the same result. Under SW assumptions the marginal borrower is profitable and so there is initially deficient credit in the market. Once relationships are established the marginal borrower is then offered funds that are priced correctly and no rationing occurs. Under DMW assumptions this same logic leads to elimination of surplus funds.

13

and the absence of collateral. Should this problem exist in practice, it would seem to be accentuated by the problems of technical complexity added by high tech firms. In conclusion, it seems that relationship lending functions to some extent as a way of learning about firm’s success probabilities and thus eliminating informational asymmetries affecting small, young and high tech firms, but it takes time to deliver. In that time, gaps (positive or negative) may exist. Deficiencies of funding, if they exist, are most likely to occur the higher the proportion of early-stage relationships and this in turn will be influenced by the rate of switching and the rate of entry of first-timers into the self-employment market.

IV. A test case for funding gaps: high- tech young firms One area where many writers feel asymmetric information is at its most damaging to fund-raising is therefore that of young, small high tech firms, commonly known as Technology-Based Small Firms or TBSFs. The potential of small, potentially fast growth, high tech firms to fall foul of a banking system would seem to be high. Additions to bank debt by such firms also raise the firm’s the debt-equity ratio thereby rendering it financially more fragile 28 . By contrast, an injection of outside equity lowers the firm’s riskiness and may encourage offers of more debt to boot, thus encouraging faster growth than would be possible from debt alone. Since the firms in question are not quoted companies, outside equity in the form of venture capital (‘private equity’) might seem therefore to be the solution to the problem of funding rapid growth. However, the small scale of operation and apparent ly low returns, combined with opaqueness of their technology and products suggests that venture

28

Higher fixed payments at regular intervals now need to be made, regardless of the state of the firm’s earnings before interest, if the firm is not to be brought to insolvency.

14

capitalists may also give them a wide berth. 29 Carpenter and Petersen’s present paper “Capital Market Imperfections, Hi-Tech Investment and New Equity Financing” (henceforth CP) uses a US panel of 2,600 such firms to examine whether these nostrums hold good in practice. Carpenter and Petersen begin by providing a comprehensive overview of the reasons why the marginal cost of debt finance in particular should increase rapidly after a certain point. In particular, the absence of inside collateral (many assets are intangible) and the costs of financial distress, play a major role in this process. Thus, high tech firms are likely to suffer from a ‘rapidly rising marginal cost of debt schedule, [with]…. a large wedge between the cost of internal and external funds’. This suggests that there is a gap between the competitively supplied quantity of finance (along a horizontal supply of funds curve) and that delivered by the market. Moreover, the position of this supply curve will depend on the firm’s internal funds so that firms with more funds of their own will experience less of a gap than those with fewer. Their analysis relies on a large panel of data on high-tech small firms most of which went to IPO over the period 1981-1998 They examine the firms’ use of debt and equity in the financing of their assets and their employment size, before and after floatation. Theory predicts that high-tech small firms use comparatively little debt for

29

In fact the circumstantial evidence for this is striking: 75% of venture capital in the UK and some 50% in the United States is invested in Management Buyouts or Buy-ins (i.e. in funding changes of ownership of large established firms ). By contrast, the figures for early stage (Startups and Other early stage investments) in the two countries are some 10% and 20% respectively. (See BVCA , 2000). What evidence there is suggests that annual rates of return on early stage investments in the UK are lower than those on later stage. This is less the case in the US where early stage investments seem to enjoy returns several-fold greater than that of the UK. (Bank of England, 2001). Whilst this evidence is consistent also with supply-side constraints (too few good early stage projects), many commentators would argue that it is a result of a combination of this with capital constraints (too little money). Perhaps in the UK it is also due to a smaller pool of available expertise to evaluate early stage hightech propositions, a lower degree of involvement of the financial institutions in venture capital investment and a smaller size of market for firms’ products (as a result of cultural barriers in the EU). See Bank of England, 2001.

15

the reasons mentioned above 30 . Theory also predicts that if they are subject to finance constraints they will experience a substantial increase in size on floatation as these constraints are relaxed. But do high-tech firms use much debt before IPO? And does floatation in practice permit a significant increase in firm size? And if so, are these the only hypotheses that may explain their results? CP find that their new empirical results are consistent with the predictions of theory. Firstly, the TBSFs in their sample obtained little debt financing for their activities: Over the whole sample, at the time of the IPO, the median debt-to-asset ratio is only 6%, and in the year of floatation firms actually retire debt, as would be expected if they were denied finance by the institutions. Comparing this with the data on SMEs in general from Berger and Udell (1998) it is clear that young high-tech firms are much smaller users of debt than the typical counterpart in the SME population31 . By contrast, TBSFs’ use of equity is substantial: new equity issues constitute 136% of assets in the early part of the period studied. Secondly, the high-tech firms in the sample (especially the smallest) pay virtually no dividends in the period considered and rely heavily on internal funds for the financing of expansion: at the median internal funds were at least 20% of assets for small firms, and typically exceeded the firm’s total expenditure on physical and R&D investment. Thirdly, new equity financing in the form of Initial Public Offering (stock market floatation) did indeed permit ‘a major increase in firm size’ (in fact a doubling or tripling of their asset values) consistent with a freeing up of financial constraints faced before floatation. 30

The test of whether high-tech firms are relatively disadvantaged of course presumes a control group. For some of the comparisons (e.g. proportions of debt and equity in the asset structure) such a group exists: it is the population of small firms studied in Berger and Udell (1998). However, for other comparisons CP need to be interpreted as testing whether high tech firms are absolutely disadvantaged or equity-constrained before and after IPO. 31 BU’s data show that 27% of total US SME finance in 1993 is provided by debt (excluding trade debt). A slightly lower proportion is holds for the UK. The difference between the high tech and conventional sectors is explained by (a) the existence of relationship banking in the conventional sector

16

Fourthly, after going public ‘only a small fraction of firms continue to make significant use of external finance’, despite the fact that these firms continued to grow significantly in size and therefore might be expected to make use of external sources of finance to do so. One of the implications of these results CP argue is that in countries where equity markets are under-developed, the ability of TSBFs to increase their size will be severely limited. These findings are rigorously derived, empirically novel and, in the context of the current debate, unquestionably controversial. For example, one perennial problem facing all writers in this area and a function of potentially fast growth firms in general, is the difficulty of disentangling demand- and supply-side constraints. Do firms fail to use equity because they cannot obtain it from outside investors (except perhaps at unacceptable prices - measured by the share of the firm relinquished), or do they simply wish to retain control of the firm, despite the possibility that this will reduce their level of long-term profits? In other words, are firms more concerned with the disutility of control-loss (implying utility- rather than profit-maximisation) than with supply-side constraints in the form of a wedge between internal and external funds? This is a grey area in economics: theory might suggest that if it is a demand-side problem, then either such ‘inefficient’ firms would be taken over and managed by less-control-averse managers who would maximise shareholder value; or that it is economically irrelevant. However, closely held firms are not subject to takeover in the way quoted firms are, thus cutting off this particular avenue of market discipline and raising the issue of government intervention once more 32 . Looking at the theoretical issues surrounding the CP paper, de Meza-Webb (1987) showed that Stiglitz-Weiss assumptions result in the optimality of equity- rather than

precluded from the young high tech firms in CP’s study and (b) the absence, in young, high-tech firms

17

debt-funding. If one believes the general logic of the SW model then a crucial point flowing from it for the present discussion is that there is no under-provision of equity33 . This could be important in interpreting the Carpenter-Petersen results in the Symposium since their position is clearly that there is a deficiency in the provision of funds to high-tech SMEs. 34 However, the assumptions of the SW model as we have seen are empirically questionable. Of course, all-equity funding would also be the result under the DMS optimism model, but again this means there is no funding shortage. Venture capitalists, the primary source of funding for small firms, are only interested in firms with growth potential and where there is an ‘exit route’ 35 (trade sale or IPO). CP’s policy conclusion is that an area for intervention is in lowering the barriers to listing of high tech firms on public security markets, particularly in economies where the financial system is underdeveloped 36 . Needless to say, the debate about the interpretation of the FHP original findings and the current CP results will probably continue. Recent exchanges in a well-known American journal attest to this (See Kaplan and Zingales, 1997; Fazzari, Hubbard and Petersen, 2000, and Kaplan and Zingales, 2000 for details).

V. Government intervention: desirable or undesirable? So, making the assumption that hi tech firms are financially constrained, what if anything should government do about it? Josh Lerner’s paper “When Bureaucrats meet Entrepreneurs…” is a fascinating review of US policy efforts to support the

of collateral and positive cash flow to secure and service debt. 32 See Cressy and Olofsson(1997) for an extended discussion of this issue. 33 This occurs because there is no marginal entrepreneur with equity funding. Her role was pivotal in generating the advantageous selection result, since it is at the margin that the bank is supplying too much credit: the marginal borrower is loss-making to the bank, rather than yielding zero profit, as required for efficiency. 34 This does not necessarily imply that FHP intervention is appropriate. See the later discussion of this issue. 35 ‘Exit route’ is the commonly accepted jargon for ‘a means of cashing in an investment’.

18

financing of small high tech firms over the last few decades and an attempt to identify future directions for policy initiatives. He begins by reviewing the long-term role of the US government in support for high tech firms, casting the economist’s usual sceptical eye over the notion of government intervention to supply allegedly marketdeficient equity to young, high-tech firms. How can officials expect to outperform the financial intermediaries (venture capitalists, specialist investment banks) in identifying and plugging perceived equity gaps? Reviewing the arguments for and against intervention, including the private sector’s response to perceived funding gaps, Lerner concludes that the dominant argument for intervention is the existence of R&D spillovers - positive externalities to innovation. The rents from such innovation will be only partly captured by the firms themselves, and small, young firms are the least effective at doing this. (Their ability to patent effectively, for example, is lower than that of their larger counterparts). 37 Lerner also raises the very important question of the selective allocation of venture capital funds. These tend, as we have seen above, to be generally concentrated away from early stage funding (seed capital, startup) and more towards transfers of ownership in management buyouts. However, Lerner shows that in the United States they are also concentrated geographically (California and Massachusetts getting the lion’s share) and by subject area (with IT and the Life Sciences dominating). These biases may well be the result of political manoeuvring. Finally, a scale bias in the allocation of government funds is apparent, with larger projects being preferred over 36

Countries where this has already occurred include the USA, UK and Germany all of whom have specialist second-tier markets for high tech firms with much lower listing requirements than the main markets. 37 If correctly identified, this deficiency of R&D would be an argument for industrial policy intervention rather than capital market intervention since it does not constitute an imperfection in the capital market. However, as has de Meza argues in correspondence, even in the industrial policy area there is scope for disagreement: the theoretical industrial organisation literature leans towards the view that there is excessive entry into the industry via the phenomenon of patent races, which result in a

19

smaller. Importantly, as with debt financing discussed earlier, there is no apparent tendency for these perceived deficits to be removed over time. So, is there a role for government or will intervention in equity markets simply pour oil on the fire? Lerner’s conclusion is that whilst there is an array of market responses to perceived equity gaps that may help to offset perceived equity deficiencies, and a similar array of issues associated with abuse of government subsidies that militate against intervention, a selective legitimate role for the government is indeed indicated in recent research. Government must intervene according to a very carefully defined set of criteria. It should (a) sometimes intervene in areas that are not currently ‘sexy’ to the private sector, supporting developments in areas that have promise but may otherwise be ignored; (b) be prepared to selectively offer follow-on financing (second-stage rather than simply first-stage funds) when the market is unwilling; (c) adopt a flexible approach to funding and a long-term commitment to a firm despite the inevitable vicissitudes of the business; (d) adopt subsidy selection criteria that have emerged from recent empirical research in the area 38 ; (e) avoid the tendency to invest in firms that are already recipients of other government awards just because they are recipients; (f) evaluate companies’ track records and management skills, with a view to identifying the effects of past awards on subsequent performance; (g) examine the role of exogenous factors 39 that may additionally influence a firm’s grant-related performance that may otherwise bias the allocation of subsidy; (h) examine carefully the commercialisability of the firm’s technology; (i) avoid pushing companies into purely pre-commercial research funded from the public purse which may result in their owners ignoring critical feedback from customers.

winner-takes-all outcome. Thus it is arguable from a theoretical standpoint that there is too much R&D relative to the social optimum. 38 For example, basing judgements more on management flexibility and experience rather than on a particular product or service offered by the company.

20

This is a substantial raft of policy proposals, but, distilled to its essence, Lerner argues that there is a role for government in supporting young, small high-tech companies if it can learn lessons from the operations of the venture capital firms themselves and from academic research in the area. This learning should include mimicking the best practice of the venture capital industry, especially where this approach is supported by relevant empirical research. But it should also adopt methods of evaluating public programs, in a more hands-on and ongoing way, thereby reducing the scope of the private sector abuse of public support and assisting in the targeting of public funds into genuine equity gaps. These findings are also controversial. For example, consider implementation issues. Whilst mimicking the behaviour of the private sector in evaluating investments is a fine idea, it may be difficult to implement in practice. 40 Similarly, governments come and go, so that a longer-term commitment of funding to specific firms may be politically difficult to sustain. There are also issues regarding the interpretation of Lerner’s evidence. De Meza, for example, seems to view the Lerner paper as providing inconclusive evidence for the need for intervention, particularly in capital markets. He sees an excess of R&D as likely as a deficiency. Adding to the funding opportunities of high-tech firms, which are (in his view) probably already gorged with money, will only exacerbate the problem41 . Lerner’s arguments for intervention are also rather different from those of CP. The latter’s case for intervention rests not so much on the likely positive spillovers of R&D uncaptured by the market system but 39

For example, legal disputes over patents. In a slightly different area, a private initiative of some of the big UK banks in the 1990s to break down cultural barriers to understanding by getting bank managers to role-play small business owners has met with a modicum of success. 41 Having recently performed a literature review on this subject for the UK’s Department of Trade and Industry, is that UK policies offering public funds for small high-tech projects probably do have economic and finance additionality and provide a certification role along the lines identified by Lerner in other US studies. Certification here consists in a positive signal that government funding can provide 40

21

more on the relative costs of obtaining external finance for young, small, high-tech firms.

VI. Conclusions As recent exchanges between economists have highlighted, the funding gaps debate is far from over. There will continue to be discussion over the criteria appropriate for identifying and measuring the existence, direction and extent of such gaps over the coming years. Whilst this symposium has highlighted important theoretical and empirical differences as well as identified areas where conflict is virtual rather than real, in terms of policy recommendations, there seems to be a reasonable consensus of opinion. If we were to ask the contributors to this symposium the following key questions the similarities of response would, I think, be surprising: 1) With regard to finance for SMEs in general, would it be best for government to a) subsidise them? b) tax them? c) do nothing? 2) With regard to selective intervention towards finance for young, small, hi-tech SMEs, would it be best for government to a) subsidise or b) tax or c) do nothing? My guess is that the Symposium consensus answer to question (1) would be (c): Do nothing. The answer to question (2) would be less obvious, but for the western European economies and the United States, the answer would probably also be (c): Do nothing. However, there also seems to be some sympathy from contributors for the idea of providing support of a different kind: mimicking the evaluative style of the private sector (Lerner), entry restrictions to self-employment (de Meza); breaking down of barriers to floatation (Carpenter and Petersen) and supplementation of Boards of Directors of young companies with experienced personnel to boost early stage ‘team human capital’ (Cressy; 3i-CBI, 1996).

to the private sector regarding investment in a particular firm. Substantial private follow-on funds have

22

If, finally, the broader consensus of economic opinion turns out to be against the support of small firms our prediction is that government intervention will nonetheless continue for purely political reasons: small firms are seen as big business in the war against unemployment and poverty, and in an increasingly non-interventionist world, as a means of pump-priming enterprise rather than providing long-term subsidy. However, one can only hope that the empirical results and policy advice embodied in the present Symposium will come finally to be built into the administration of these programs. One last academic word - even if this may sound like special pleading. It really is clear that whilst governments have begun to wake up to this fact, more independent and rigorous evaluations need to be made of the effectiveness of government schemes offered in this area.

Robert Cressy City University Business School London

been recorded in the UK and the US from this source.

23

References 3i-CBI, 1996, Tech Stars: Breaking the growth barriers for technology-based SMEs, Confederation of British Industry, London Aston Business School, 1990, Constraints on the Growth of Small Firms, Department of Trade and Industry, UK Bank of England, 2001, The financing of technology-based small firms, Domestic Finance Division, February Berger, Alan N. and Gregory Udell, 1995, Relationship lending and lines of credit in small firm finance, Journal of Business 68, 351-382 Berger, Alan N. and Gregory Udell, 2001,Small Business Credit Availability and Relationship Lending: the Importance of Bank Organisational Structure, paper presented to the International Conference on Funding Gaps Controversies, CSME, University of Warwick, April 11-12, 1999 and prepared for this Symposium Black, J., De. Meza, D., and Jeffreys, D., 1996, House Prices, the Supply of Collateral and the Enterprise Economy, The Economic Journal, vol 106 no.434: 60-75 BVCA, 2000, Report on Investment Activity 1999, British Venture Capital Association, May Carpenter, Robert and Bruce Petersen, 2001, ‘Capital Market Imperfections, Hi-Tech Investment and New Equity Financing’, paper presented to the International Conference on Funding Gaps Controversies, CSME, University of Warwick, April 11-12, 1999 and prepared for this Symposium Cressy, Robert C. ,1993, The Startup Tracking Exercise: Third Year Report, the National Westminster Bank of Great Britain, November

24

Cressy, Robert C., 1996a. “Commitment Lending under Asymmetric Information: Theory and Tests on U.K. Startup Data.” Small Business Economics 8: 1-12 Cressy, Robert C., 1996b, Are business startups debt-rationed?, The Economic Journal, 106, No. 438, September, 1253-1270 Cressy, Robert C., 2000, European Loan Guarantee Schemes: Who has them, who pays and who gains?, in Green, 2000 Cressy, Robert C., and Christer Olofsson (Eds) (1997), European SME Financing: Small Business Economics, Special Issue on European SME Financing (Cressy and Olofsson, eds.). Cressy, Robert C., and Otto Toivanen, 1998, Is There Adverse Selection in the Credit Market?, CSME Working Paper, Warwick Business School, University of Warwick, forthcoming in Venture Capital: An International Journal of Entrepreneurial Finance, 2001 Egeln, Jürgen, Georg Licht and Fabian Steil, 1997, Firm Foundations and the Role of Financing Constraints, in Cressy and Olofsson( Eds) (1997) Evans, David and Boyan Jovanovic. 1989. “An Estimated Model of Entrepreneurial Choice Under Liquidity Constraints.” Journal of Political Economy 97,4: 808827 Fazzari, Steven M. , R. Glen Hubbard and Bruce Peterson, 1988, "Financing Constraints and Corporate Investment", Brookings Papers on Economic Activity, 1, 141-195 Fazzari, Steven M. , R. Glen Hubbard and Bruce Peterson, 2000, Financing Constraints and Corporate Investment: Response to Kaplan and Zingales, Quarterley Journal of Economics, February, 169-215 Gale, David and Martin Hellwig, 1985, “Incentive compatible debt contracts: the one period problem” Review of Economics Studies, Vol. 52: 647-63. Green, Bo (Editor), 2000, Risk Behaviour and Risk Management in Business Life, Kluwer, Netherlands 25

Hall, Bronwyn, 1992, "Investment and Research and Development at the Firm Level: Does the Source of Financing Matter?", National Bureau of Economic Research Working Paper #4096, Cambridge, Massachusetts Hubbard, Glen, Capital-Market Imperfections and Investment, 1998, Journal of Economic Literature Vol XXXVI, March, 193-225 Kaplan, Steven N., and Luigi Zingales, 1997, Do investment –cashflow sensitivities provide useful measures of financing constraints?, Quarterly Journal of Economics, February, 168-215 Kaplan, Steven N., and Luigi Zingales, 2000, Investment –cash flow sensitivities are not valid measures of financing constraints, Quarterly Journal of Economics, May, 707-712 Lerner, Josh, 2000, When Bureaucrats Meet Entrepreneurs: The Design of Effective “Public Venture Capital” Programs, paper presented to the International Conference on Funding Gaps Controversies, CSME, University of Warwick, April 11-12, 1999 and prepared for this Symposium de Meza, David and David Webb, 1987, Too Much Investment: A Problem of Asymmetric Information, Quarterly Journal of Economics, 102, 281-292 de Meza, David, 1999, ‘Overlending’, paper presented to the International Conference on Funding Gaps Controversies, CSME, University of Warwick, April 11-12, 1999, and prepared for this Symposium Peterson, Mitchell A. and Raghuram G. Rajan, 1994, The benefits of lending relationships: Evidence from small business data, Journal of Finance XLIX (1), 3-37, March Ritter, Jay. 1991. “The Long-Run Performance of Initial Public Offerings,” Journal of Finance, 46, 3-27

26

Toivanen, Otto, and Robert Cressy, 1999, Lazy Entrepreneurs Or Dominant Banks? An Empirical Analysis of the Market for SME Loans in The UK, CSME Working Paper, Warwick Business School Townsend, R. (1979). “Optimal contracts and competitive markets with costly state verification” Journal of Economic Theory, vol.21 pp 265-93.

27