Você está na página 1de 35

CROSS-INDUSTRY DIFFERENCES IN FINANCING BEHAVIOUR OF

SMEs FROM AUSTRALIA’S BUSINESS LONGITUDINAL SURVEY

Mr Pal C. Johnsen,
Special Investigator,
National Authority for Investigation and Prosecution
of Economic and Environmental Crime (OKOKRIM),
Pb 8193 Dep,
N-0034 Oslo,
Norway.
Telephone: +47 23 29 10 00
Facsimile: +47 23 29 10 01
Email: paalchristian@yahoo.no
and

Professor Richard G.P. McMahon,*


Head, School of Commerce,
The Flinders University of South Australia,
GPO Box 2100,
Adelaide South Australia 5001.
Telephone: +61 8 82012840
Facsimile: +61 8 82012644
Email: Richard.McMahon@flinders.edu.au

SCHOOL OF COMMERCE

RESEARCH PAPER SERIES: 03-7

ISSN: 1441-3906

Acknowledgments

The permission of the Australian Statistician to use confidentialised data from the federal government’s

Business Longitudinal Survey, and to publish findings based on analysis of that data, is gratefully

acknowledged. Responsibility for interpretation of the findings lies solely with the authors.

* Author for correspondence.


2

CROSS-INDUSTRY DIFFERENCES IN SME FINANCING BEHAVIOUR

Abstract
The principal objective in this paper is to ascertain the extent to which industry appears to influence
the financing behaviour of several thousand SMEs distributed across eleven industries, using data
taken from the Australian federal government’s Business Longitudinal Survey for four financial
years from 1994-95 to 1997-98. The research reported in the paper provide substantial empirical
evidence that cross-industry differences in financing behaviour do exist even after controlling for
other relevant influences on SME financing choices such as enterprise size, business age,
profitability, growth, asset structure and risk. The key finding is that industry does not simply proxy
for one or more of these other factors, but is an important influence in its own right.

Introduction

The scholarly literature concerned with the financing behaviour of business concerns, including that of

small and medium-sized enterprises (SMEs), has for some time shown interest in the influence industry

may have upon funding choices. In the simplest terms, it has been conjectured that firms within a particular

industry tend to adopt a similar financing pattern that essentially represents a consensus on what is

appropriate given prevailing circumstances in the industry. Holmes et al. (2003, p. 112) have recently

summarised extant knowledge on this matter, with emphasis on SMEs, as follows:

Industry as an explanatory variable is supported by the equilibrium theory of capital structure which
suggests ‘the economic sector a company belongs to can be an important factor . . . when explaining
. . . financial behaviour’ (Lopez-Garcia & Aybar-Arias 2000, p. 57). Sectors with strong tangible
asset holdings are expected to have higher average debt levels than is evident in sectors associated
with intangible or risky assets. However there has been controversy and debate concerning the
association between industry and debt structure (Jordan, Lowe & Taylor 1998, p. 3). This debate has
ranged from comments suggesting differences across industries but consistency within industries
(Harris and Raviv, 1991), to claims that industry is not as important as firm-specific aspects
(Balakrishnan & Fox 1993). Jordan, Lowe and Taylor (1998) support this latter view, while Cassar
and Holmes (2001), Hall, Hutchinson and Michaelas (2000), Lopez-Gracia (sic) and Aybar-Arias
(2000), Romano, Tanewski and Smyrnios (2001), Michaelas, Chittenden and Poutziouris (1999) and
Bennett and Donnelly (1993) find some support for an association.

Apart from highlighting that the issue of whether industry influences financing behaviour is, as yet,

unsettled, this quotation raises the possibility that industry is simply a proxy for one or more firm-specific

characteristics that are the real underlying determinants of funding choices (Van Auken and Neely, 1996;
3

Gibson, 2002). These characteristics could include enterprise size, business age, profitability, growth, asset

structure and risk.

The preceding discussion suggests the following research question that has not hitherto been

explicitly addressed in published scholarly works employing the same data:

In a multivariate context in which other known influences are included essentially as controls, does
industry still appear to have a consistent statistically significant influence upon the financing
behaviour of small and medium-sized enterprises from Australia’s Business Longitudinal Survey?

This research builds upon and extends prior studies using the same data source and addressing aspects of

the financing behaviour of SMEs undertaken by Gibson (2001, 2002), Holmes and Cassar (2001) and

Cassar and Holmes (2001, 2003). Gibson (2001, 2002) uses an entirely different research design and relies

on a bivariate analysis to test for industry influence upon financial choices. Nevertheless, he finds

statistically significant industry effects. Industry is incidental to the work of Holmes and Cassar (2001) and

Cassar and Holmes (2001, 2003) which also does not incorporate the full range of independent/control

variables included in the present study. They find only limited industry effects.

The paper proceeds as follows. After reviewing prior research on industry and financing behaviour

amongst SMEs, the research method is outlined. Thereafter, the findings of the research are presented,

followed by conclusions arising from this investigation.

Prior Research

Why Industry May Influence SME Financing Behaviour

Recent decades have seen the emergence of a number of often overlapping theories of business financing

that are believed to have relevance to SMEs. These include Static Trade-Off Theory, Agency Theory,

Pecking Order Theory, Life-Cycle Theory and Alternate Resource (or Bootstrapping) Explanations. Each

of these is briefly examined below for what it may suggest about a possible industry influence upon SME

financing behaviour. In the following section, recent empirical evidence on this matter amongst SMEs is

reviewed. Thereafter, consideration is given to empirical evidence on other firm-specific characteristics that

might influence financial choices.

According to Static Trade-Off Theory, financial leverage should be related to observable firm

characteristics such as business risk and asset structure (Peirson et al., 1995). Businesses with mostly
4

intangible assets should borrow less, other factors being equal, than concerns with mostly tangible assets,

because of the collateral provided by these assets (Jordan et al., 1998; Michaelas et al., 1999). With Static

Trade-Off Theory’s emphasis on the costs of financial distress, there should clearly be an inverse

relationship between business risk and leverage. One could therefore suggest that the Static Trade-Off

Theory implies variations of debt to equity ratios across different industries as business risk typically varies

between industries. Bradley et al. (1984) indicate that, if a firm’s cost of financial distress is significant, its

optimal capital structure will be inversely related to the variability of earnings. They further argue that the

theory of optimal capital structure suggests a strong relationship between industry classification and

average firm leverage ratio.

Agency Theory suggests that firms in growing industries will experience higher agency costs due to

the fact that equity-controlled firms have a tendency to invest suboptimally to expropriate wealth from debt

holders (Jordan et al., 1998). Hall et al. (2000) point out that agency costs may vary across industries and

lead to inter-industry differences in financial structure.

In proposing his Pecking Order Theory of business financing, Myers (1984, p. 590) indicates that

‘average debt ratio will vary from industry to industry, because asset risk, asset type, and requirements for

external funds also vary by industry’. He rejects the idea that a long-run industry average will be a

meaningful target for individual businesses within industries. Myers (1984) illustrates this by arguing that a

very profitable firm in an industry with slow growth will end up with an unusually low debt ratio compared

to the industry average. This profitable firm, he argues, will not ‘go out of its way to issue debt and retire

equity to achieve a more normal debt ratio’ (Myers, 1984, p.589). According to Pecking Order Theory, an

unprofitable firm in the same industry will choose a higher debt ratio than a profitable firm and, if this debt

ratio is high enough to cause financial distress, the firm may rebalance its financial structure by issuing

equity.

In expanding on Life-Cycle Theory, Weston and Brigham (1981) claim that life-cycles differ

between growth and non-growth industries, and between new and traditional industries. As an example,

they indicate that SMEs in growing industries, where forecasts of future sales are optimistic, should

increase the proportion of fixed assets in order to reduce variable costs in the future, and thereby give them

a competitive advantage. Weston and Brigham (1981) suggest that these investments are most likely
5

undertaken with short-term debt, and that this will lead to poor liquidity over some period. This increase in

fixed assets will, according to Weston and Brigham (1981), alter firms’ capital structures and influence

their debt ratios.

In their exploration of Alternate Resource (or Bootstrapping) Explanations of SME financing, Van

Auken and Neeley (1996) point out that small firms in capital-intensive industries, with traditionally high

proportions of fixed assets, are less likely to make use of bootstrap financing than firms in less capital-

intensive industries. This, they argue, is because fixed assets acts as collateral and thereby simplify access

to traditional sources of finance. Besides this, the bootstrap financing literature does not seem to consider,

in depth, whether industry has an effect on the use of these alternative sources of finance.

Empirical Evidence on Industry and SME Financing Behaviour

This section of the paper briefly reviews five international and three Australian studies that provide recent

empirical evidence on the influence of industry upon the financing behaviour of SMEs. Overall, the studies

referred to here provide mixed support for the view that industry may be a determining factor for SME

financial behaviour.

Michaelas et al. (1999) examined 3,500 United Kingdom small firms randomly selected from the

Lotus One-Source Database in 1995 and representing 10 industries. Profit and Loss Accounts and Balance

Sheets for a ten-year period (1986-1995) were examined in order to ascertain the influence of operating

risk, age, profitability, size and industry on financial structures. Michaelas et al.’s (1999, p. 123) findings

provide strong support for their hypothesis that industry has an influence on the capital structure of small

firms:

. . . almost all of the industry dummy coefficients are significantly different from zero at the 5%
level of significance, indicating that industry exhibits a significant effect on the capital structure of
small firms. . . . industry has an effect on the total level of debt in small firms as well as the maturity
structure of debt.

Michaelas et al. (1999) suggest that, in all industries examined, the industry effect is more pronounced on

short-term debt ratios compared to long-term debt ratios. The difference between the magnitude of the

industry effect on short-term and long-term debt, however, varies greatly across industries. For the

construction industry and the wholesale and retail trade industry, the industry effect on short-term debt
6

relative to long-term debt is estimated to be 9.6 and 11.2 respectively; whereas the comparable figure for

education, health and social work is 1.2.

Hall et al. (2000) employed the same database and the same number of businesses as Michaelas et

al. (1999), but used a slightly different approach. Hall et al.’s (2000) dependent variables were short-term

debt to total assets and long-term debt to total assets. Amongst their independent variables, they included a

total of nine constant and 45 slope dummies for industry. The conclusion reached by Hall et al. (2000, p.

311) is that:

Across all industries profitability does not affect long-term debt. The effects of the other variables
do vary. With short-term debt the effects of profitability, asset structure, size and age vary across
industries but that of growth, whilst strong, remains the same.

Hall et al. (2000) report similar results to Michaelas et al. (1999) suggesting that the wholesale and retail

trade-industry, on average, uses very little long-term debt; and that the education, health and social work

industry uses almost equal amounts of short-term and long-term debt. Overall, Hall et al. (2000) indicate,

like Michaelas et al. (1999), that leverage ratios vary across industries, but that the effect is much larger for

short-term debt when compared to long-term debt

Lopez-Garcia and Aybar-Arias (2000) examined 1,000 Spanish SMEs in the years 1994-1995. Their

results suggest a significant influence of industry on short-term debt, but indicate little support for a

significant effect on long-term debt. Their findings also suggest that enterprise size is an important

influence on financial behaviour, and that medium-sized concerns act in a significantly different way from

micro-businesses.

Van der Wijst and Thurik’s (1993) study examined financial statements of 27 former West German

industries in retailing over a 20-year period for time- and industry-specific effects. Using debt ratio as the

dependent variable and total assets, asset structure, inventory turnover, depreciation charges, and return on

investment as independent variables, they find that, for total debt, the industry-specific effects dominate the

time-specific effects for almost all variables tested. Their results suggest that SMEs in various industries

tend to match the maturity structures of debt with their asset structures, leading to industries with high

proportions of fixed assets using more long-term debt and vice versa. Van der Wijst and Thurik (1993) do

not, however, report the same predominant industry effect on short-term debt compared to long-term debt,

as suggested by Michaelas et al. (1999), Hall et al. (2000), and Lopez-Garcia and Aybar-Arias (2000).
7

In Jordan et al.’s (1998) study, strategy, financial policy and capital structure were linked using

questionnaire-based data from 275 SMEs in a number of industries in the United Kingdom. Three different

measures of capital structure were used as dependent variables. Amongst the independent variables were

nine dummies, each representing a one-digit level SIC industry code. Jordan et al.’s (1998) results suggest

that industry-specific effects are unimportant in the context of SME capital structures. Jordan et al. (1998)

explain their finding by pointing out that small firms often operate in niche markets and that this is likely to

reduce broad industry influences on capital structures.

In recent years, some Australian evidence regarding industry effects on SME financing behaviour

has been provided by Romano et al. (2000), Gibson (2001, 2002), Holmes and Cassar (2001) and Cassar

and Holmes (2001, 2003). As indicated in the introduction to the paper, Gibson (2001, 2002), Holmes and

Cassar (2001) and Cassar and Holmes (2001, 2003) use data from Australia’s Business Longitudinal

Survey.

Romano et al. (2000) suggest, after having analysed the financial structures of 1,490 Australian

family businesses, that industry is not a significant predictor of debt as a source of financing. They do,

however, indicate that, overall, industry is one of several factors that can influence some financial decisions

of family business owners.

In the research of Gibson (2001, 2002), exploratory cluster analysis is used to identify five distinct

clusters of SMEs based around key funding sources: trade credit debt, bank loan debt, related persons debt,

other debt and equity, and working owner equity. Gibson (2001) finds that there is no industry effect on the

bank loan cluster. In addition, he suggests no strong industry association for the ‘other debt and equity’

cluster and ‘working owner equity’ cluster, even if these are the dominant sources for the finance and

insurance and mining and manufacturing industries respectively. Gibson (2001) suggests that, for each of

the years tested, there appear to be significant differences in cluster membership according to industry

sector. The mining and manufacturing industry in particular, show persistently high debt ratios for all three

years; whereas the finance and insurance industry shows persistently low debt ratios. Mining and

manufacturing industry seems, however, to rely heavily on ‘working owner equity’; whereas the finance

and insurance industry tends to rely more on ‘other debt and equity’. Gibson (2001) further suggests that
8

the trade credit (short-term) debt cluster is associated with the wholesale and the retail trade sectors. These

sectors are characterised by low asset structures and high reliance on current assets.

As indicated earlier, industry is incidental to the research of Holmes and Cassar (2001) and Cassar

and Holmes (2001, 2003). Holmes and Cassar (2001, p. 14) point to the mixed results in previous research

on an industry effect upon SME financing and suggest that:

The problems with drawing inferences from these studies include the varying use of other control
variables between studies and the different proxy for industry effects employed. Consequently,
industry may be proxying the securability of assets or asset risk, depending upon the choice of
variables included in the empirical models.

Holmes and Cassar (2001, p.14) go on to indicate that ‘the control of industry grouping [in] the regressions

had limited effect on the inferences found, although industry effects were generally found to be significant’.

Other Influences on SME Financing Behaviour

This section provides some empirical evidence on other firm-specific influences that are likely to affect the

financing behaviour of SMEs. These influences are: enterprise size, business age, profitability, growth,

asset structure and risk. Omission of other possible factors bearing on SME financial choices suggested by

prior research are acknowledged as being limitations of the present study in the concluding section of the

paper.

Regarding enterprise size, Michaelas et al. (1999) find evidence indicating that larger firms use

higher gearing ratios than smaller firms, and they suggest this is a result of smaller firms facing higher

financial barriers. This view is supported by, amongst others, Chittenden et al. (1996), Hall et al. (2000

Holmes and Cassar (2001), and Cassar and Holmes (2001, 2003) who provide evidence suggesting that size

is positively related to long-term debt and negatively related to short-term debt. Romano et al. (2000) and

Gibson (2001, 2002) also find an important relationship between size and capital structure; and Lopez-

Garcia and Aybar-Arias (2000) suggest that size significantly influences the self-financing of smaller

companies. Amongst recent studies, only Jordan et al. (1998) find no relationship between financial

structure and enterprise size.

Concerning business age, Chittenden et al. (1996) and Hall et al. (2000) argue that younger firms

rely more on short-term finance than more mature firms, as a result of a positive relationship between age

and liquidity. Chittenden et al. (1996) further suggest that the use of both short-term and long-term debt
9

falls with age. Michaelas et al. (1999) argue that younger firms have higher average gearing ratios than

older firms as a result of the latter being more profitable and having accumulated internal sources. Romano

et al. (2000) find that business age is not a significant predictor of debt as a source of financing, although

their review of prior research caused them to hypothesise that the opposite would be true. Curiously,

despite this evidence and the fact that business age data were available to them, Gibson (2001, 2002),

Holmes and Cassar (2001), and Cassar and Holmes (2001, 2003) did not include this variable amongst the

independent variables in their studies.

With respect to profitability, Myers (1984) argues that profitable firms are less likely to borrow as a

result of their preference for and the rewards of retaining earnings. Chittenden et al. (1996), Michaelas et

al. (1999), Gibson (2001, 20020, Holmes and Cassar (2001), and Cassar and Holmes (2001, 2003) provide

evidence supporting Myers (1984), as they indicate profitability is negatively related to total gearing. They

suggest that SME owner-managers prefer to use retained profits, and that they only raise debt when

additional financing is essential. Unprofitable concerns will therefore be more likely to increase borrowings

as a result of their lack of internal funds.

A negative relationship between profitability and short-term debt is reported by Chittenden et al.

(1996), Hall et al. (2000) and Gibson (2001, 2002). Chittenden et al. (1996) also find that profitability

appears to be negatively related to long-term debt; whereas Hall et al. (2000) suggest that profitability is

not statistically significantly related to long-term debt. Jordan et al. (1998) provide no support for the

negative impact of profitability on debt, and they argue that profitability is not important in determining

SME financial structures.

Regarding growth, small firms with rapid growth are less likely to be able to finance their activities

by using internal funds and will therefore need to borrow funds (Ang, 1992). Empirical evidence for this

view is provided by, amongst others, Michaelas et al. (1999), Hall et al. (2000), Holmes and Cassar (2001),

and Cassar and Holmes (2001, 2003). Holmes and Cassar (2001) and Cassar and Holmes (2001, 2003)

further suggest growth as being an important factor for the composition of SME financial structures. Hall et

al. (2000) find a positive and highly significant relationship, especially between short-term debt and

growth. Chittenden et al. (1996), however, suggest that the influence of growth itself is insignificant for

SME capital structures, but that the combination of rapid growth and lack of access to capital market is an
10

important factor. Gibson (2002), on the other hand, does not find any consistent statistically significant

relationship between growth and debt levels. This view is to some extent supported by Jordan et al. (1998)

who argue there is no positive association between sales growth and debt levels.

Concerning asset structure, Michaelas et al. (1999) suggest that it is reasonable to expect firms

with predominant fixed asset structures and high collateral values to have easier access to external finance

than firms with lower proportions of fixed assets. This, they believe, will probably lead to differences in

capital structures, with the latter firms obtaining less debt. Jordan et al. (1998) argue that higher proportions

of fixed assets and better collateral values are particularly important in the case of SMEs because this will

assist them in obtaining higher levels of debt at lower cost.

Chittenden et al. (1996), Hall et al. (2000), Gibson (2000, 2001), Holmes and Cassar (2001), and

Cassar and Holmes (2001, 2003) suggest that asset structure is positively related to long-term debt and

negatively related to short-term debt. However, Holmes and Cassar (2001) and Cassar and Holmes (2001,

2003) argue that, overall, asset structure is negatively related to leverage. Michaelas et al. (1999) find

strong support for the hypothesis that asset structure is positively related to gearing. Their results indicate

that a high component of fixed assets and a high inventory level are associated with both higher short-term

debt and higher long-term debt. This, they argue, is a result of the significant effect information asymmetry

and agency problems have on small firms, inducing them to use fixed assets as collateral for obtaining debt.

This view is supported by, amongst others, Chittenden et al. (1996) and Jordan et al. (1998).

With respect to risk, Harris and Raviv (1991) review a number of studies and conclude that most

research tends to support the view that profit volatility is negatively related to debt. This is in line with the

suggestion that firms with high business risk tend to have less leveraged capital structures. Pettit and Singer

(1985) indicate that small firms will have a negative relationship between risk and gearing, due to higher

bankruptcy costs. The probability of bankruptcy will depend on both the firm’s business risk and its

financial risk; but, at any given level of business risk, the higher the firm’s financial risk, the higher will be

the probability of bankruptcy.

Holmes and Cassar (2001) and Cassar and Holmes (2001, 2003) find, at first, no support for the

proposition that risk influences the level of debt financing in SMEs. By re-running their regressions using

an alternative risk measure, they find, however, evidence suggesting a relatively strong negative effect.
11

That is, high-risk firms find debt less attractive. Overall, their study provides weak evidence for risk being

an influencing factor on SMEs’ financing and capital structure choices. In contrast, Jordan et al. (1998)

surprisingly find a positive relationship between risk and gearing in small firms. This result is supported by

Michaelas et al. (1999, p. 121) who suggest that this is because ‘agency costs are lower in more risky firms,

due to lower underinvestment problems, allowing such firms to rely on higher gearing ratios’.

Research Method

Research Data

The data employed in this research are drawn from the Business Longitudinal Survey (BLS) conducted by

the Australian Bureau of Statistics (ABS) on behalf of the federal government over the four financial years

1994-95 to 1997-98. Costing in excess of $4 million, the BLS was designed to provide information on the

growth and performance of Australian employing businesses, and to identify selected economic and

structural characteristics of these businesses.

The ABS Business Register was used as the population frame for the survey, with approximately

13,000 business units being selected for inclusion in the 1994-95 mailing of questionnaires. For the 1995-

96 survey, a sub-sample of the original selections for 1994-95 was chosen, and this was supplemented with

a sample of new business units added to the Business Register during 1995-96. The sample for the 1996-97

survey was again in two parts. The first formed the longitudinal or continuing part of the sample,

comprising all those remaining live businesses from the 1995-96 survey. The second part comprised a

sample of new business units added to the Business Register during 1996-97. A similar procedure was

followed for the 1997-98 survey. Approximately 6,400 business units were surveyed in each of 1995-96,

1996-97 and 1997-98. The BLS did not employ completely random samples. The original population (for

1994-95) was stratified by industry and business size, with equal probability sampling methods being

employed within strata. Further stratification by innovation status, exporting status and growth status took

place for the 1995-96 survey.

Data collection in the BLS was achieved through self-administered, structured questionnaires

containing essentially closed questions. Copies of the questionnaires used in each of the four annual

collections can be obtained from the ABS. The questionnaires were piloted prior to their first use, and were
12

then progressively refined after each collection in the light of experience. As well as on-going questions,

each questionnaire also included once-off questions dealing with certain matters of policy interest to the

federal government at the time of the collections. Various imputation techniques, including matching with

other data files available to the ABS, were employed to deal with any missing data. Because information

collected in the BLS was sought under the authority of the Census and Statistics Act 1905, and thus

provision of appropriate responses to the mailed questionnaires could be legally enforced by the Australian

Statistician, response rates were very high by conventional research standards – typically exceeding 90 per

cent.

The specific BLS data used in this study are included in a Confidentialised Unit Record File

(CURF) released by the ABS on CD-ROM in December, 1999. This CURF contains data on 9,731 business

units employing fewer than 200 persons – broadly representing SMEs in the Australian context. Restricted

industrial classification detail, no geographical indicators, presentation of enterprise age in ranges, and

omission of certain data items obtained in the BLS all help to maintain the confidentiality of unit records.

Furthermore, all financial variables have been subject to perturbation – a process in which values are

slightly varied to provide further confidentiality protection.

Data Analysis

Variables used in this research are either categorical in nature or, if metric, have irregular distributional

properties (that is, they are non-normally distributed and/or have noticeable extreme values or outliers).

Transformation of metric variables to produce normal distributions is not undertaken because of difficulties

of interpretation often created by such procedures. Somewhat arbitrary censoring of extreme values or

outliers is similarly avoided. Accordingly, non-parametric/distribution free techniques of statistical analysis

are employed exclusively. Amongst other attractions, these techniques emphasise the median as a measure

of centrality that is resistant to extreme values or outliers.

The principal modelling procedure used in this research is logistic regression (also referred to as

‘logit analysis’). The main reason for choosing this multivariate technique is the decision made to use

categorical (that is, non-metric) forms of the dependent variables. As Hair et al. (1995, p. 130) point out:

. . . discriminant analysis is also appropriate when the dependent variable is nonmetric. However,
logit analysis may be preferred for several reasons. First, discriminant analysis relies on strictly
13

meeting the assumptions of multivariate normality and equal variance-covariance matrices across
groups, features not found in all situations. Logit analysis does not face these strict assumptions,
thus making its application appropriate in many more situations. Second, even if the assumptions are
met, many researchers prefer logit analysis because it is similar to regression with its straightforward
statistical tests, ability to incorporate nonlinear effects, and wide range of diagnostics. For these and
more technical reasons, logit analysis is equivalent to discriminant analysis and may be more
appropriate in certain situations.

The assumptions underlying logistic regression are undemanding and its use with the irregularly distributed

(that is, non-normal) data available to the present study is entirely appropriate (Aldrich and Nelson, 1984).

Further information on logistic regression as a statistical modelling technique is presented in an Appendix

to the paper.

Research Hypotheses

After having reviewed the literature and its conflicting conclusions relating to industry and SME financing

behaviour, the following null-hypotheses are proposed for testing in this research study:

H0A: After controlling for other relevant influences on SME financing behaviour, there are no
consistent statistically significant differences in short-term debt vs total funding across industry
sectors.
H0B: After controlling for other relevant influences on SME financing behaviour, there are no
consistent statistically significant differences in long-term debt vs total funding across industry
sectors.
H0C: After controlling for other relevant influences on SME financing behaviour, there are no
consistent statistically significant differences in total debt vs total funding across industry
sectors.
H0D: After controlling for other relevant influences on SME financing behaviour, there are no
consistent statistically significant differences in times interest earned across industry
sectors.

Specifying Independent and Dependent Variables

The analytical model for this study, derived from the prior research reviewed earlier, is as illustrated in

Figure 1.

INSERT FIGURE 1 ABOUT HERE

This model represents industry (one-digit ANZSIC code) as the key independent variable having 11

categories1, with controls for enterprise size (employment, sales or assets), business age (two-year ranges),
14

profitability (return on total assets), growth (employment growth, sales growth and asset growth), asset

structure (fixed assets as a percentage of total assets) and risk (absolute value of coefficient of variation

over four years for return on total assets), as possibly influencing the financing behaviour of the SMEs

studied. The significance values for a series of Kolmogorov-Smirnov one-sample tests suggest that all the

metric independent/control variables are not normally distributed. A series of associative tests suggests the

possibility of multicollinearity between the enterprise size measures: number of employees, annual sales

and total assets. As will become evident, simultaneous use in modelling of multicollinear independent

variables has been precluded. For reasons of space, descriptive statistics for the various independent

variables are not included in this paper, but they can be provided by the authors on request.

Financing behaviour is captured separately using four dependent variables: short-term debt / total

funding, long-term debt / total funding, total debt / total funding, and times interest earned (net profit before

tax and interest / interest). The use of leverage variables is common in research in the area. While

consideration of times interest earned is less usual, it is nevertheless a valid reflection of financing choices

made by businesses. Following the lead of prior research (Van der Wijst and Thurik, 1993; Hutchinson et

al., 1998; Michaelas et al., 1999; Hall et al., 2000), employing three leverage variables allows the research

to examine influences on the maturity structure of debt as well as the total debt position of sample SMEs.

The significance values for a series of Kolmogorov-Smirnov one-sample tests suggest that the metric forms

of the dependent variables are far from being normally distributed. Selected statistics for the dependent

variables for each year of the study are presented in Table 1.

INSERT TABLE 1 ABOUT HERE

Examination of Table 1 reveals that in each of the four years of the longitudinal study, the sample SMEs

maintained quite high levels of short-term debt vs total funding, in the range 38 to 45 per cent. By contrast,

long-term debt vs total funding is in the range 5 to 9 per cent. Overall, total debt vs total funding in the

range 71 to 72 per cent reveals that the financial structure of the SMEs examined is clearly debt-oriented.

Times interest earned in the range 4 to 6 times is not exceptional for businesses in general. Table 1 also

reveals that for all dependent variables in all years there are highly statistically significant differences

across industries. It is for the research to discover whether such cross-industry differences remain evident

in an appropriate multivariate context.


15

Research Findings

Short-Term Debt vs Total Funding

The first stage of the multivariate logistic regression modelling undertaken employed a dichotomous

dependent variable indicating whether short-term debt vs total funding (per cent) is above or below the

median value for this ratio amongst the SMEs in the datafile. Separate modelling was undertaken for each

of the four years considered in the study. For any one year, avoiding the joint inclusion of multicollinear

variables measuring enterprise size meant producing three models.

Results from this modelling effort, expressed in terms of the sign and statistical significance of the

coefficients for the chosen independent/control variables, are presented in Table 2.

INSERT TABLE 2 ABOUT HERE

Comments below focus mainly upon the independent/control variables in bold italics in Table 2, for which

there appear to be consistent statistically significant relationships with the dependent variable in a

multivariate context. It would appear from the modelling results that short-term debt vs total funding for the

business concerns studied is particularly influenced by:

• Industry which, overall, is consistently statistically significant at the 0.01 level or better across the

four years examined. It seems that an SME being in the construction or wholesale trade industries

increases the likelihood that its short-term debt vs total funding ratio will be above the median value

for the businesses studied. Note also that, although not consistent over all four years, there is a

suggestion of a significant industry effect for cultural and recreational services. In this case, SMEs

in the industry are less likely to have a short-term debt vs total funding ratio above the median.

• Business age, for which the sign of the regression coefficient is negative. The implication is that the

younger an SME is, and therefore the less time it has had to become self-sufficient through

reinvestment of profits and/or to build a sufficient track record to attract longer-term development

funding, the more likely it will need to depend upon short-term debt financing for its assets and

activities.

• Asset structure, for which the sign of the regression coefficient is negative. The implication is that

the lower the proportion of fixed assets held by an SME, the more likely it will be that it depends
16

upon short-term debt funding for its assets. This is in accord with the dictates of the matching or

hedging principle of business financing.

Long-Term Debt vs Total Funding

The second stage of the multivariate logistic regression modelling undertaken employed a dichotomous

dependent variable indicating whether long-term debt vs total funding (per cent) is above or below the

median value for this ratio amongst the SMEs in the datafile. The extent and pattern of modelling

undertaken were similar to those already described for short-term debt vs total funding. Results from this

modelling effort, expressed in terms of the sign and statistical significance of the coefficients for the chosen

independent/control variables, are presented in Table 3.

INSERT TABLE 3 ABOUT HERE

Comments below focus mainly upon the independent/control variables in bold italics in Table 3, for which

there appear to be consistent statistically significant relationships with the dependent variable in a

multivariate context. It would appear from the modelling results that long-term debt vs total funding for the

business concerns studied is particularly influenced by:

• Industry which, overall, is consistently statistically significant at the 0.01 level or better across the

four years examined. It seems that an SME being in the manufacturing, retail trade and transport and

storage industries increases the likelihood that its long-term debt vs total funding ratio will be above

the median value for the businesses studied. Note also that, although not consistent over all four

years, there is a suggestion of a significant industry effect for the wholesale trade and finance and

insurance industries. In the case of wholesale trade, SMEs in the industry are more likely to have a

long-term debt vs total funding ratio above the median. In the case of finance and insurance, SMEs

in the industry are less likely to have a long-term debt vs total funding ratio above the median.

• Enterprise size as measured by number of employees, for which the sign of the regression

coefficient is positive. The implication is that the larger an SME is in terms of employment, the

more likely it will depend upon long-term debt financing. This would be the case if access to longer-

term debt financing is dictated, to some degree, by the size of the business.

• Asset structure, for which the sign of the regression coefficient is positive. The implication is that

the higher the proportion of fixed assets held by an SME, the more likely it will be that it depends
17

upon long-term debt financing for its assets. This is in accord with the dictates of the matching or

hedging principle.

Note also that, although not consistent over all four years, there is a suggestion of a significant positive

influence of business age upon long-term debt vs total funding. The implication is that the older the

concern, the more likely it will be that it is above the median for the proportion of long-term debt in total

funding for the SMEs investigated. This would be the case if older businesses have a stronger equity base

and/or a longer track record of successful operation that facilitate access to long-term debt financing.

Total Debt vs Total Funding

The third stage of the multivariate logistic regression modelling undertaken employed a dichotomous

dependent variable indicating whether total debt vs total funding (per cent) is above or below the median

value for this ratio amongst the SMEs in the datafile. The extent and pattern of modelling undertaken were

similar to those already described for the previous stages. Results from this modelling effort, expressed in

terms of the sign and statistical significance of the coefficients for the chosen independent/control

variables, are presented in Table 4.

INSERT TABLE 4 ABOUT HERE

Comments below focus mainly upon the independent/control variables in bold italics in Table 4, for which

there appear to be consistent statistically significant relationships with the dependent variable in a

multivariate context. It would appear from the modelling results that total debt vs total funding for the

business concerns studied is particularly influenced by:

• Industry which, overall, is consistently statistically significant at the 0.01 level or better across the

four years examined. It seems that an SME being in the construction, retail trade and transport and

storage industries increases the likelihood that its total debt vs total funding ratio will be above the

median value for the businesses studied. Note also that, although not consistent over all four years,

there is a suggestion of a significant industry effect for cultural and recreational services. In this

case, SMEs in the industry are less likely to have a total debt vs total funding ratio above the

median.

• Business age, for which the sign of the regression coefficient is negative. The implication is that the

younger an SME is, and therefore the less time it has had to become self-sufficient through
18

reinvestment of profits and/or to build a sufficient track record to attract longer-term development

funding, the more likely it will need to depend upon debt financing for its assets and activities.

• Asset structure, for which the sign of the regression coefficient is negative. The implication is that

the lower the proportion of fixed assets held by an SME, the more likely it will be that it depends

upon debt financing for its assets. This finding is counterintuitive if, as seems probable, debt

providers rely upon fixed assets as security or collateral for loans. However, the explanation may

well lie with the predominance of short-term debt in the balance sheets of the businesses studied

(see findings presented earlier).

Note also that, although not consistent over all four years, there is a suggestion of a significant positive

influence of employment growth upon the total debt vs total funding ratio. The implication is that the

higher the growth rate in employment, the more likely it will be that the SMEs investigated depend upon

debt financing for their assets and activities. This would be the case when business growth exceeds the rate

of accumulation of undistributed profits and/or the ability and willingness to raise new equity capital.

Time Interest Earned

The final stage of the multivariate logistic regression modelling undertaken employed a dichotomous

dependent variable indicating whether times interest earned is above or below the median value for this

ratio amongst the SMEs in the datafile. The extent and pattern of modelling undertaken were similar to

those already described for the previous stages. Results from this modelling effort, expressed in terms of

the sign and statistical significance of the coefficients for the chosen independent/control variables, are

presented in Table 5.

INSERT TABLE 5 ABOUT HERE

Comments below focus mainly upon the independent/control variables in bold italics in Table 5, for which

there appear to be consistent statistically significant relationships with the dependent variable in a

multivariate context. It would appear from the modelling results that times interest earned for the business

concerns studied is particularly influenced by:

• Industry which, overall, is consistently statistically significant at the 0.01 level or better across the

four years examined. It seems that an SME being in the mining and retail trade industries increases
19

the likelihood that its times interest earned ratio will be above the median value for the businesses

studied.

• Return on total assets, for which the sign of the regression coefficient is positive. The implication is

that the more profitable an SME is, the more likely it will be that its interest coverage is above the

median for all businesses studied. This clearly follows from the fact that operating profit is the

numerator in calculating the times interest earned ratio.

• Asset structure, for which the sign of the regression coefficient is negative. The implication is that

the lower the proportion of fixed assets held by an SME, the more likely it will be that its interest

coverage is above the median for all businesses studied. An explanation for this finding is that the

lower the proportion of fixed assets, the greater use is likely to be made of short-term debt that

normally commands a lower rate of interest than long-term debt. The lower interest expense is the

denominator in calculating the times interest earned ratio, causing the ratio to be higher.

Note also that, although not consistent over all four years, there is a suggestion of a significant positive

influence of business age upon the times interest earned ratio. The implication is that the older the concern,

the more likely it will be that it is above the median times interest earned for the SMEs investigated. This

would be the case if takes younger businesses some time to achieve their profitability potential and/or, as

suggested by some findings reported earlier, younger concerns tend to be more heavily laden with debt.

Conclusions and Recommendations

The key findings from this research into industry influences on the financing behaviour of Australian SMEs

included in the BLS CURF can be summarised as follows:

• Industry does appear to have a consistent statistically significant influence upon short-term debt vs

total funding, particularly for the construction and wholesale trade industries and possibly for

cultural and recreational services. Null hypothesis H0A is therefore rejected.

• Industry does appear to have a consistent statistically significant influence upon long-term debt vs

total funding, particularly for the manufacturing, retail trade and transport and storage industries and

possibly for the wholesale trade and finance and insurance industries. Null hypothesis H0B is

therefore rejected.
20

• Industry does appear to have a consistent statistically significant influence upon total debt vs total

funding, particularly for the construction, retail trade and transport and storage industries and

possibly for cultural and recreational services. Null hypothesis H0C is therefore rejected.

• Industry does appear to have a consistent statistically significant influence upon times interest

earned, particularly for the mining and retail trade industries. Null hypothesis H0D is therefore

rejected.

Overall, the research findings reported in the paper provide substantial empirical evidence that cross-

industry differences in financing behaviour do exist even after controlling for other relevant influences on

SME financing choices such as enterprise size, business age, profitability, growth, asset structure and risk.

The key suggestion is that industry does not simply proxy for one or more of these other factors, but is an

important influence in its own right.

Although incidental to the main purpose of the study, this research has also revealed (inter alia) the

following consistent statistically significant influences upon SME financing behaviour:

• Enterprise size in terms of number of employees is positively related to long-term debt vs total

funding, suggesting that larger businesses are more inclined and/or better able to access long-term

debt financing.

• Business age is negatively related to short-term debt vs total funding, and positively related to long-

term debt vs total funding. In other words, as businesses grow older they place less relative reliance

upon short-term debt and more upon long-term debt. Overall, business age appears to be negatively

related to total debt vs total funding, suggesting that accumulated profits mean less reliance upon

debt funding of whatever term.

• Asset structure which is negatively related to short-term debt vs total funding, and positively related

to long-term debt vs total funding. In other words, the higher the proportion of fixed assets held the

lower will be the dependence on short-term debt and the greater the use made of long-term debt.

Overall, asset structure appears to be negatively related to total debt vs total funding, suggesting that

the more capital-intensive the business the greater the dependence on equity financing.

Notably, this research has also found few, if any, statistically significant influences upon SME financial

choices for profitability, growth and risk. The correspondence of these outcomes with those of prior
21

research, reviewed earlier in the paper, increases confidence that this study’s finding on industry’s

influence upon SME financing behaviour is not simply a methodological artefact.

Beyond the inevitable bounds introduced by the broad data and methodological choices made, this

research has been limited by its reliance on essentially a finance paradigm. The study shares this orientation

with the recent research of Gibson (2001, 2002), Holmes and Cassar (2001) and Cassar and Holmes (2001,

2003) using the same datafile. Romano et al. (2000) is a recent example of an SME capital structure study

employing a broader strategic management paradigm. Apart from identifying some of the independent/

control variables employed in the present research, the literature review undertaken by Romano et al.

(2000) includes consideration of such diverse influences as culture, entrepreneurial characteristics, family

considerations, owner-manager experience and preferences, business goals, life-cycle issues, preferred

ownership structures, attitudes to indpendence and control, business planning, perceived risk and attitudes

towards personal risk. Unfortunately, the BLS CURF lacks data to operationalise most of these influences.

Broadening the analysis with more comprehensive information on owner-managers and their SMEs

provides an opportunity for further research in the area.

The main implication of this research for scholars and policy-makers concerned with SMEs is

clearly the need to regard industry as an important independent influence upon financing behaviour. There

are evidently effects arising from the fundamental nature of industries that require better understanding

before a reliable prescriptive position on SME financing can be reached. What these effects are cannot

really be ascertained using the research data and methods employed in this study, which give a relatively

superficial perspective on the matter. A need for more in-depth qualitative investigation is indicated.

Notes
1.
In regression modelling of a k-level multinomial or polytomous independent variable like industry, the
default treatment in computerised statistical packages is to recode the variable into k-1 indicator or dummy
variables. The coefficients for the new variables then represent the effect of each category compared to a
reference category, usually the first or the last category. This is referred to as simple contrasting. With a
variable like industry, this treatment results in a more-or-less arbitrary choice of one particular industry as
the reference category. In this research it was considered more appropriate to use deviation contrasting, as a
result of which the regression coefficients indicate the effect of each category compared to the average
effect for all categories. Thus, each industry is benchmarked for its effect against the overall average effect
for all industries.
22

References

Aldrich, J.H. and Nelson, F.D., 1984, Linear Probability, Logit, and Probit Models, Sage Publications:

Beverly Hills, California.

Ang, J.S., 1992, ‘On the Theory of Finance for Privately Held Firms’, The Journal of Small Business

Finance, 1(3), pp. 185-203.

Balakrishnan, S. and Fox, I., 1993, ‘Asset Specificity, Firm Heterogeneity and Capital Structure’, Strategic

Management Journal, 14(1), pp. 3-16.

Bennett, M. and Donnelly, R., 1993, ‘The Determinants of Capital Structure: Some UK Evidence’, British

Accounting Review, 25(1), pp. 43-59.

Bradley, M., Jarrell, G.A., and Han Kim, E., 1984, ‘On the Existence of an Optimal Capital Structure:

Theory and Evidence’, The Journal of Finance, 39(3), pp. 857-878.

Cassar, G. and Holmes, 2001, ‘Capital Structure and Financing of SMEs: Australian Evidence’, paper

presented to the Annual Conference of the Accounting Association of Australia and New Zealand,

Auckland, New Zealand.

Cassar, G. and Holmes, 2003, ‘Capital Structure and Financing of SMEs: Australian Evidence’, Accounting

and Finance, 43(2), pp. 123-147

Chittenden, F., Hall, G. and Hutchinson, P., 1996, ‘Small Firm Growth, Access to Capital Markets and

Financial Structure: Review of Issues and an Empirical Investigation’, Small Business Economics, 8 (1),

pp. 59-67.

Gibson, B., 2001, ‘Financial Structure in Australian Small Firms’, paper presented to the Annual

Conference of the Small Enterprise Association of Australia and New Zealand, Wellington, New

Zealand, September.

Gibson, B., 2002, ‘Clusters of Financial Structure in Australian Small Firms’, Small Enterprise Research:

The Journal of SEAANZ, 10(1), pp. 59-74.

Hair, J.F., Anderson, R.E., Tatham, R.L. and Black, W.C., 1995, Multivariate Data Analysis, 4th edn,

Prentice-Hall: Englewood Cliffs, New Jersey.

Hall, G., Hutchinson, P. and Michaelas, N., 2000, ‘Industry Effects on the Determinants of Unquoted

SMEs’ Capital Structure’, International Journal of the Economics of Business, 7 (3), pp. 297-312.
23

Harris, M. and Raviv, A., 1991, ‘The Theory of Capital Structure’, The Journal of Finance, 46(1), pp. 297-

355.

Holmes, S. and Cassar, G., 2001, ‘Capital Structure and Financing of SMEs: Australian Evidence’.

Working Paper, Department of Accounting and Finance, The University of Newcastle, Newcastle, New

South Wales.

Holmes, S., Hutchinson, P., Forsaith, D., Gibson, G. and McMahon, R., 2003, Small Enterprise Finance,

John Wiley & Sons Australia: Brisbane, Queensland.

Hutchinson, P., Hall, G. and Michaelas, N., 1998, ‘The Determinants of Capital Structure for Micro, Small

and Medium-Sized Enterprises’, Proceedings of the 43rd Annual World Conference of the International

Council for Small Business, Nanyang Technological University, Singapore, pp. 1-9.

Jordan, J., Lowe, J. and Taylor, P., 1998, ‘Strategy and Financial Policy in UK Small Firms’, Journal of

Business Finance and Accounting, 25 (1-2), pp. 1-27.

Lopez-Garcia, J. and Aybar-Arias, C., 2000, ‘An Empirical Approach to the Financial Behaviour of Small

and Medium Sized Companies’, Small Business Economics, 14(1), pp. 55-63.

Michaelas, N., Chittenden, F. and Poutziouris, P., 1999, ‘Financial Policy and Capital Structure Choice in

U.K. SMEs: Empirical Evidence From Company Panel Data’, Small Business Economics, 12 (2), pp.

113-130.

Myers, S.C., 1984, ‘The Capital Structure Puzzle’, The Journal of Finance, 39 (3), pp. 575-592.

Peirson, G., Bird, R., Brown, R. and Howard, P., 1995, Business Finance, 6th edn, McGraw-Hill Book

Company: Sydney, New South Wales.

Pettit, R.R. and Singer, R.F., 1985, ‘Small Business Finance: A Research Agenda’, Financial Management,

14 (3), pp. 47-60.

Romano, C.A., Tanewski, G.A. and Smyrnois, K.X., 2001, ‘Capital Structure Decision Making: A Model

for Family Business’, Journal of Business Venturing, 16 (3), pp. 285-310.

Van Auken, H.E. and Neeley, L., 1996, ‘Evidence of Bootstrap Financing Among Start-Up Firms’, The

Journal of Entrepreneurial and Small Business Finance, 5(3) , pp. 235-249.

Van der Wijst, D. and Thurik, R., 1993, ‘Determinants of Small Firm Debt Ratios: An Analysis of Retail

Panel Data’, Small Business Economics, 5 (1), pp. 55-65.


24

Weston, J.F. and Brigham, E.F., 1981, Managerial Finance, 7th edn, Dryden Press: Hinsdale, Illinois.
25

Table 1: Selected Statistics for Dependent Variables

1994-95 1995-96 1996-97 1997-98


Dependent Variable Statistics (n=7,703) (n=4,814) (n=4,790) (n=4,754)

Short-term debt Median across all industries 41.7 44.2 40.2 38.9
vs total funding (per cent)

Statistical significance in 0.000 0.000 0.000 0.000


Kruskal-Wallis test for
differences across industries

Long-term debt Median across all industries 5.3 5.6 8.3 8.8
vs total funding (per cent)

Statistical significance in 0.000 0.000 0.000 0.000


Kruskal-Wallis test for
differences across industries

Total debt Median across all industries 71.3 71.9 71.7 71.4
vs total funding (per cent)

Statistical significance in 0.000 0.000 0.000 0.000


Kruskal-Wallis test for
differences across industries

Times interest earned Median across all industries 5.5 4.5 4.6 5.2
(times)

Statistical significance in 0.000 0.000 0.000 0.000


Kruskal-Wallis test for
differences across industries
26

Table 2: Coefficient Details in Logistic Regression Modelling of


Short-Term Debt vs Total Funding

1994-95 1995-96 1996-97 1997-98


Independent Variable (n=7,703) (n=4,814) (n=4,790) (n=3,465)
3/3 significant 3/3 significant 3/3 significant 3/3 significant
Industry:
overall** overall** overall** overall**
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Mining
0/3 significant 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 negative 3/3 positive 3/3 negative
Manufacturing
3/3 significant* 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Construction
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Wholesale Trade
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Retail Trade
3/3 significant** 0/3 significant 3/3 significant* 0/3 significant
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Accommodation, Cafés and Restaurants
3/3 significant** 0/3 significant 3/3 significant** 0/3 significant
3/3 positive 3/3 negative 3/3 positive 3/3 positive
Transport and Storage
0/3 significant 0/3 significant 3/3 significant* 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Finance and Insurance
3/3 significant** 3/3 significant** 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Property and Business Services
3/3 significant* 0/3 significant 3/3 significant* 0/3 significant
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Cultural and Recreational Services
3/3 significant** 3/3 significant** 3/3 significant** 0/3 significant
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Personal and Other Services
0/3 significant 3/3 significant** 3/3 significant* 0/3 significant

Enterprise Size:
1/1 positive 1/1 positive 1/1 positive 1/1 positive
Number of Employees
1/1 significant** 0/1 significant 1/1 significant** 0/1 significant
1/1 positive 1/1 positive 1/1 positive 1/1 positive
Annual Sales
1/1 significant** 0/1 significant 0/1 significant 1/1 significant*
1/1 positive 1/1 positive 1/1 positive 1/1 positive
Total Assets
0/1 significant 0/1 significant 0/1 significant 0/1 significant
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Business Age
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**
3/3 positive 3/3 positive 3/3 negative 3/3 negative
Return on Total Assets
0/3 significant 0/3 significant 3/3 significant** 3/3 significant**

*, ** Indicate statistical significance at the 5 and 1 per cent levels respectively.


Note: 3/3 means three of three logistic regression models. Bold italics for an independent variable indicates
consistent statistically significant relationships.

. . . cont.
27

Table 2 (cont.): Coefficient Details in Logistic Regression Modelling of


Short-Term Debt vs Total Funding

1994-95 1995-96 1996-97 1997-98


Independent Variable (n=7,703) (n=4,814) (n=4,790) (n=3,465)

Growth:
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Employment Growth
2/3 significant* 0/3 significant 0/3 significant 1/3 significant*
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Sales Growth
0/3 significant 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive
Asset Growth n.a. 0/3 significant 0/3 significant 0/3 significant
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Asset Structure
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**
3/3 positive
Risk n.a. n.a. n.a. 0/3 significant

*, ** Indicate statistical significance at the 5 and 1 per cent levels respectively.


Note: 3/3 means three of three logistic regression models. Bold italics for an independent variable indicates
consistent statistically significant relationships.
28

Table 3: Coefficient Details in Logistic Regression Modelling of


Long-Term Debt vs Total Funding

1994-95 1995-96 1996-97 1997-98


Independent Variable (n=7,703) (n=4,814) (n=4,790) (n=3,465)
3/3 significant 3/3 significant 3/3 significant 3/3 significant
Industry:
overall** overall** overall** overall**
3/3 positive 3/3 positive 3/3 negative 3/3 positive
Mining
0/3 significant 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Manufacturing
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**
3/3 negative 3/3 positive 3/3 negative 3/3 negative
Construction
0/3 significant 1/3 significant* 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 2/3 negative
Wholesale Trade
3/3 significant** 3/3 significant** 2/3 significant** 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Retail Trade
3/3 significant* 3/3 significant** 3/3 significant** 2/3 significant*
3/3 negative 3/3 negative 3/3 positive 3/3 negative
Accommodation, Cafés and Restaurants
0/3 significant 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Transport and Storage
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Finance and Insurance
3/3 significant** 3/3 significant** 3/3 significant** 0/3 significant
3/3 negative 3/3 positive 3/3 positive 3/3 negative
Property and Business Services
0/3 significant 0/3 significant 0/3 significant 0/3 significant
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Cultural and Recreational Services
0/3 significant 3/3 significant* 3/3 significant** 0/3 significant
3/3 negative 3/3 positive 3/3 positive 2/3 positive
Personal and Other Services
0/3 significant 0/3 significant 0/3 significant 0/3 significant

Enterprise Size:
1/1 positive 1/1 positive 1/1 positive 1/1 positive
Number of Employees
1/1 significant** 1/1 significant** 1/1 significant** 1/1 significant**
1/1 positive 1/1 positive 1/1 positive 1/1 positive
Annual Sales
1/1 significant* 0/1 significant 0/1 significant 0/1 significant
1/1 positive 1/1 positive 1/1 positive 1/1 positive
Total Assets
0/1 significant 0/1 significant 0/1 significant 0/1 significant
3/3 positive 3/3 positive 3/3 negative 3/3 negative
Business Age
3/3 significant* 2/3 significant** 0/3 significant 1/3 significant*
3/3 positive 3/3 negative 3/3 negative 3/3 positive
Return on Total Assets
0/3 significant 3/3 significant** 3/3 significant** 0/3 significant

*, ** Indicate statistical significance at the 5 and 1 per cent levels respectively.


Note: 3/3 means three of three logistic regression models. Bold italics for an independent variable indicates
consistent statistically significant relationships.

. . . cont.
29

Table 3 (cont.): Coefficient Details in Logistic Regression Modelling of


Long-Term Debt vs Total Funding

1994-95 1995-96 1996-97 1997-98


Independent Variable (n=7,703) (n=4,814) (n=4,790) (n=3,465)

Growth:
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Employment Growth
0/3 significant 2/3 significant* 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Sales Growth
0/3 significant 2/3 significant* 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive
Asset Growth n.a. 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Asset Structure
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**
3/3 positive
Risk n.a. n.a. n.a. 0/3 significant

*, ** Indicate statistical significance at the 5 and 1 per cent levels respectively.


Note: 3/3 means three of three logistic regression models. Bold italics for an independent variable indicates
consistent statistically significant relationships.
30

Table 4: Coefficient Details in Logistic Regression Modelling of


Total Debt vs Total Funding

1994-95 1995-96 1996-97 1997-98


Independent Variable (n=7,703) (n=4,814) (n=4,790) (n=3,465)
3/3 significant 3/3 significant 3/3 significant 3/3 significant
Industry:
overall** overall** overall** overall**
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Mining
3/3 significant** 3/3 significant* 0/3 significant 0/3 significant
3/3 positive 3/3 negative 3/3 negative 3/3 negative
Manufacturing
0/3 significant 0/3 significant 3/3 significant* 2/3 significant*
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Construction
3/3 significant* 3/3 significant** 3/3 significant* 3/3 significant*
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Wholesale Trade
3/3 significant** 0/3 significant 0/3 significant 3/3 significant**
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Retail Trade
3/3 significant** 3/3 significant* 3/3 significant** 3/3 significant*
3/3 negative 3/3 negative 3/3 negative 3/3 positive
Accommodation, Cafés and Restaurants
3/3 significant* 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Transport and Storage
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**
3/3 positive 3/3 negative 3/3 negative 3/3 positive
Finance and Insurance
0/3 significant 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 2/3 negative
Property and Business Services
3/3 significant* 0/3 significant 0/3 significant 0/3 significant
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Cultural and Recreational Services
3/3 significant** 3/3 significant** 3/3 significant** 0/3 significant
3/3 negative 3/3 positive 3/3 positive 3/3 negative
Personal and Other Services
0/3 significant 0/3 significant 0/3 significant 0/3 significant

Enterprise Size:
1/1 positive 1/1 negative 1/1 negative 1/1 negative
Number of Employees
0/1 significant 0/1 significant 0/1 significant 0/1 significant
1/1 positive 1/1 positive 1/1 positive 1/1 positive
Annual Sales
1/1 significant** 0/1 significant 0/1 significant 0/1 significant
1/1 positive 1/1 positive 1/1 positive 1/1 positive
Total Assets
0/1 significant 0/1 significant 0/1 significant 0/1 significant
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Business Age
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**
3/3 positive 3/3 positive 3/3 negative 3/3 positive
Return on Total Assets
0/3 significant 0/3 significant 3/3 significant** 3/3 significant**

*, ** Indicate statistical significance at the 5 and 1 per cent levels respectively.


Note: 3/3 means three of three logistic regression models. Bold italics for an independent variable indicates
consistent statistically significant relationships.

. . . cont.
31

Table 4 (cont.): Coefficient Details in Logistic Regression Modelling of


Total Debt vs Total Funding

1994-95 1995-96 1996-97 1997-98


Independent Variable (n=7,703) (n=4,814) (n=4,790) (n=3,465)

Growth:
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Employment Growth
3/3 significant** 3/3 significant** 3/3 significant** 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Sales Growth
0/3 significant 0/3 significant 3/3 significant** 0/3 significant
3/3 positive 3/3 positive 3/3 positive
Asset Growth n.a. 0/3 significant 0/3 significant 0/3 significant
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Asset Structure
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**
3/3 positive
Risk n.a. n.a. n.a. 0/3 significant

*, ** Indicate statistical significance at the 5 and 1 per cent levels respectively.


Note: 3/3 means three of three logistic regression models. Bold italics for an independent variable indicates
consistent statistically significant relationships.
32

Table 5: Coefficient Details in Logistic Regression Modelling of


Times Interest Earned

1994-95 1995-96 1996-97 1997-98


Independent Variable (n=7,703) (n=4,814) (n=4,790) (n=3,465)
3/3 significant 3/3 significant 3/3 significant 3/3 significant
Industry:
overall** overall** overall** overall**
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Mining
3/3 significant** 3/3 significant* 3/3 significant* 3/3 significant*
3/3 negative 3/3 negative 3/3 positive 3/3 negative
Manufacturing
0/3 significant 0/3 significant 1/3 significant* 0/3 significant
3/3 positive 3/3 negative 3/3 positive 3/3 negative
Construction
0/3 significant 0/3 significant 0/3 significant 0/3 significant
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Wholesale Trade
3/3 significant** 0/3 significant 0/3 significant 3/3 significant*
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Retail Trade
3/3 significant** 3/3 significant* 3/3 significant** 3/3 significant*
3/3 negative 3/3 negative 3/3 negative 3/3 positive
Accommodation, Cafés and Restaurants
1/3 significant* 0/3 significant 0/3 significant 2/3 significant*
3/3 positive 3/3 negative 3/3 negative 3/3 negative
Transport and Storage
3/3 significant* 0/3 significant 0/3 significant 3/3 significant*
3/3 negative 3/3 positive 3/3 negative 3/3 positive
Finance and Insurance
0/3 significant 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 2/3 positive
Property and Business Services
0/3 significant 2/3 significant* 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Cultural and Recreational Services
0/3 significant 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 negative 3/3 negative 3/3 negative
Personal and Other Services
0/3 significant 0/3 significant 0/3 significant 0/3 significant

Enterprise Size:
1/1 positive 1/1 negative 1/1 negative 1/1 negative
Number of Employees
0/1 significant 0/1 significant 0/1 significant 0/1 significant
1/1 positive 1/1 positive 1/1 positive 1/1 positive
Annual Sales
0/1 significant 1/1 significant* 1/1 significant* 0/1 significant
1/1 positive 1/1 positive 1/1 positive 1/1 positive
Total Assets
0/1 significant 0/1 significant 0/1 significant 0/1 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Business Age
0/3 significant 3/3 significant* 3/3 significant* 3/3 significant**
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Return on Total Assets
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**

*, ** Indicate statistical significance at the 5 and 1 per cent levels respectively.


Note: 3/3 means three of three logistic regression models. Bold italics for an independent variable indicates
consistent statistically significant relationships.

. . . cont.
33

Table 5 (cont.): Coefficient Details in Logistic Regression Modelling of


Times Interest Earned

1994-95 1995-96 1996-97 1997-98


Independent Variable (n=7,703) (n=4,814) (n=4,790) (n=3,465)

Growth:
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Employment Growth
0/3 significant 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive 3/3 positive
Sales Growth
0/3 significant 0/3 significant 0/3 significant 0/3 significant
3/3 positive 3/3 positive 3/3 positive
Asset Growth n.a. 0/3 significant 0/3 significant 0/3 significant
3/3 negative 3/3 negative 3/3 negative 3/3 negative
Asset Structure
3/3 significant** 3/3 significant** 3/3 significant** 3/3 significant**
3/3 positive
Risk n.a. n.a. n.a. 0/3 significant

*, ** Indicate statistical significance at the 5 and 1 per cent levels respectively.


Note: 3/3 means three of three logistic regression models. Bold italics for an independent variable indicates
consistent statistically significant relationships.
34

Figure 1: Analytical Model for the Research

Enterprise
Characteristics: Financing
Behaviour:
• Industry
• Short-term debt vs
• Enterprise size
Influence total funding
• Business age
• Long-term debt vs
• Profitability total funding
• Growth • Total debt vs total
funding
• Asset structure
• Times interest earned
• Risk
35

Appendix: Logistic Regression Modelling

The generalized form of the multivariate logistic regression model for the case of a dichotomous dependent

variable with values 0 or 1 and continuous independent variables can be expressed as follows:

ln[ π /(1 − π)] = φ + β1x1 + . . . + βn xn + u

where π = probability that the value of the dichotomous dependent variable, y, equals 1

x1, . . . , xn = independent variables

φ = constant

β1, . . . , βn = coefficients

u = stochastic disturbance term representing that part of ln[π/(1-π)]

which is unexplained by the independent variables

Note that the left hand side of the equation is not the dependent variable, y, itself; but the so-called ‘log

odds’ or ‘logit’ of y. Where an independent variable is categorical rather than continuous, two treatments

are possible. The variable can possibly be dealt with as if it is continuous. Alternatively, indicator (design

or dummy or contrast) variables may be created and coded as 0 or 1 for all but one category (usually the

last); and coefficients are estimated for each of these indicator variables. The latter treatment is more

common for polytomous independent variables whether they are nominal or ordinal. It is usually

recommended that dichotomous independent variables are treated as if they are continuous.

Você também pode gostar