Form of Ownership and Financial Constraints:Panel Data Evidence

Empirica 27: 175–192, 2000.
© 2000 Kluwer Academic Publishers. Printed in the Netherlands.
Form of Ownership and Financial Constraints:
Panel Data Evidence From Flow of Funds and
Investment Equations
1 Boston College; 2 University of Torino, and CERIS-CNR, Via Avogadro 8, 10121 Torino, Italy
(E-Mail: [email protected])
Abstract. This paper analyzes the effects of the form of ownership on the substitutability between
internal and external sources of finance in Italy. In particular, we test whether financial constraints are
more severe for independent firms compared to members of large national business groups and subsidiaries of foreign multinational corporations. The results obtained from flow of funds and investment
equations estimated for a panel of Italian companies imply that independent firms face more severe
financial constraints. In fact, not only members of national groups and subsidiaries of multinational
corporations find it easier to substitute cash flow with external finance when the former falls but they
do not display excess sensitivity to cash flow and debt in their investment decisions.
Key words: Investment, financing constraints, ownership
JEL classification: E22, E44, G32
I. Introduction
Italian private firms can be classified, in terms of form of ownership, either as
independent firms or as members of larger organizations, such as national business groups or foreign multinational corporations. Business groups are a pervasive
form of organization in several countries, including Italy. They exist in a variety
of types, ranging from hierarchical groups with a pyramidal structure to associative groups pursuing their common interest through a more informal system of
co-ordinated decision making. In this paper we plan to analyze the role of group
membership in alleviating capital market imperfections faced by firms in Italy. In
this perspective, business groups and indeed multinational corporations can be seen
as organizational forms that allow a mitigation of the information and contract
enforcement problems that arise in accessing external financial resources. On the
one hand, business groups allow the formation of an internal capital market that
? We would like to thank K. Baum, D. Mueller, S. Titman, a very constructive referee and seminar
participants at the Universities of Bologna and Modena for useful comments and suggestions and D.
Margon and S. Zelli for their help in assembling the data set.
may partially replace the capital allocation function of the external market. A group
can pool funds from different affiliates and reallocate them to the most profitable
uses.1 On the other hand business groups may also improve access to external
capital markets. In some countries (e.g., Japan and Germany) groups are organically linked with banks that play an important role in financing, monitoring, and
co-ordinating activities of member firms. The association with banks can be seen
as another way to minimize information problems and to more closely align the
incentives of borrowers and lenders. Multinational corporations play a similar role
in creating an internal capital market and in facilitating their subsidiaries’ access to
external funds. When trying to obtain external finance, subsidiaries are very likely
to benefit from the financial strength, reputation, geographical and (often) product
diversification of the parent company.
In Italy most private firms are owned and controlled by families. Sometimes
control is exercised over several separate companies through complex pyramidal
organizational structures (hierarchical business groups) that allow a retention of
control rights, while at the same time minimizing financial requirements to do so.2
In any case, controlling coalitions tend to own a large fraction of the shares of the
company.3 National business groups operate in a context of underdeveloped capital
markets, in which bank loans are the most common form of external finance.4 Contrary to the experience of other bank-based countries, such as Japan and Germany,
the role of banks in equity financing is marginal and it is unusual for bankers to sit
on the boards of directors of industrial firms or to play an active role in influencing
industrial firms’ strategic decisions. It is however true that, partly for historical
reasons, large business groups have special informal relations with national financial institutions. Some of the latter are considered to play an important role not
only in financing but also in acting as exclusive clubs where mutual shareholdings
are organized and, more generally, decisions on corporate control are taken.5
It is common wisdom in Italy that members of large national business groups are
likely to face more favorable lending terms than independent firms. The existence
of these informal links to financial institutions, together with the more diversified nature of business groups is perceived as a distinctive advantage in obtaining
external funds for affiliated firms. Moreover, most firms quoted in the national
stock market are members of the largest business groups, which enhances their
advantage in obtaining funds from security markets.6 Subsidiaries of foreign multinational corporations are also likely to enjoy favorable relations with domestic
banking institutions, although perhaps not as good as affiliates to large national
groups. Moreover they have a distinctive advantage in accessing international
capital markets, either directly or indirectly through the parent company.
No systematic investigation has been conducted to date on the effect of the
form of ownership on the substitutability between internal and external finance.
The purpose of this paper is to fill this gap and to test whether financial constraints
are more severe for independent firms vis-à-vis affiliated firms, and whether there
are differences between members of national groups and subsidiaries of foreign
multinational corporations. One standard approach to assessing the substitutability
of internal and external sources has been to investigate the sensitivity of investment
to cash flow for different categories of firms.7 In our paper we address this issue by
investigating the role of cash flow in both flow of funds and investment equations.
The advantage of this approach is twofold. First, the effect of the availability of
internal finance on external financing decisions contains important information on
the degree of substitutability between different sources of funds that should be
exploited. Second, the joint empirical analysis of both decisions acts as a stringent
consistency check on the conclusions one reaches on the existence and severity of
financial constraints for different firms.
In our empirical work we make use of a novel dataset of Italian firms constructed at CERIS by merging balance sheet information on firms, published yearly
by Mediobanca, with qualitative information on firms’ form of ownership taken
directly from company reports that permits us to classify firms as members of
a national business group, subsidiaries of a foreign multinational corporation, or
other national firms. Section 2 of the paper discusses the nature of the data set
and summarizes the relevant descriptive statistics. Section 3 presents an empirical
analysis of the determinants of firms’ (change in) debt decisions, whereas in section 4 the effects of financial variables on fixed investment are tested. Concluding
remarks are provided in Section 5.
II. Descriptive Statistics on Cash Flow and Financial Debt
In this section we provide both a brief description of the unbalanced sample of
firms used in this paper and some descriptive evidence on the evolution of the
financial variables used in our study. Real and financial data are available for
1229 firms over the period 1977–1990. The number of consecutive observations
for each firm ranges from one to 14. In each year, firms are allocated to one of
three categories: members of large national business groups, foreign subsidiaries,
or other national firms.8 Firms are classified as affiliates of large national groups
if they are controlled, directly or indirectly, in the relevant year by the following
18 groups: Agnelli – Fiat, De Benedetti – Cir, Ferruzzi – Montedison, Fininvest
– Mondadori, Pesenti – Italmobiliare, Pirelli, Barilla, Benetton, Cartiere Burgo,
Falck, Ferrero, Gft, Lucchini, Marzotto, Merloni, Miroglio, Parmalat, and Smi.
These groups represented the ’core’ of the private national industrial sector in the
eighties and most of them have been ranked in the top positions in terms of size
since the first incomplete list of groups was published by Mediobanca in 1983.
Furthermore, these are the only private groups with a consolidated turnover greater
than 1,000 billion Lira in 1990. Firms are classified as foreign subsidiaries if the
parent company is foreign, and as other national firms when they do not satisfy
the requirements to be included in the first two categories. This category contains
mainly independent companies, but firms affiliated to smaller and younger business
groups are also included. We have grouped these two types of firms together for two
Table I. Descriptive statistics on size (number of employees)
Number of firm-year
Only firms with more than 3 consecutive observations
Full sample
Members of large national business groups 2349.5
Foreign subsidiaries
Other national firms
Full sample
Full sample
Members of large national business group
Foreign subsidiaries
Other national firms
reasons. First, information on the organizational structure of the smaller business
groups is not very rich (especially in the first years of our sample period) and the
decision to allocate some firms to a given business group (especially in the case
of indirect control) would have been rather arbitrary. Second, the smaller business
groups are more similar to the independent firms in our sample than to the large
business groups in terms of size and diversification.
As can be seen from Table I, out of a total of 7633 firm-year observations, 1489
pertain to large national groups, 2462 to subsidiaries of multinationals and 3682 to
firms not associated with either. The average number of employees is 1127 (1044
after excluding firms with fewer than 4 consecutive observations). However, this
figure hides significant differences among our three sub-samples. In fact, the average size of firms that are members of large national business groups (2603 or 2350
employees, depending on the sample used) is much bigger than the average size
of subsidiaries of multinationals (1057 or 1024 employees) and of other domestic
companies (577 or 561 employees).
In the descriptive analysis that follows, we discuss the evolution of cash flow
and indebtedness for the firms in each of the three categories. For both variables
we compute and plot the median (Q2),the first decile (D1), the first quartile (Q1),
the third quartile (Q3) and the ninth decile (D9).9
We start our analysis by focusing on the dynamics of internal finance over time.
Pre-dividend cash flow divided by total assets (computed as the sum of the replacement value of fixed assets and the accounting value of gross working capital)
is used as a proxy for internal finance.10 Pre-dividend cash flow is computed by
subtracting the sum of total labor costs, interest paid and taxes from value added.
In Figures 1a, 1b and 1c percentiles are plotted for the sub-samples of members
of large national business groups, affiliates to foreign multinationals, and other
national firms respectively. The data show that internal finance moves procyclically for all firms. It declines in the 1981–1982 recessionary period that follows
the second oil shock and the tightening of monetary policy. Then, following the
economic recovery which started in 1983, we observe a steady increase in internal
finance up to 1987. Finally, cash flow declines again in the most recent years,
partly anticipating the recession that occurred at the beginning of the nineties.
The comparison between our three figures does not suggest any striking difference
in either the dynamics or the levels of cash flow, even if firms affiliated to large
national business groups seem on the whole less profitable than the others (this
is particularly true for the lower tail of the distribution). The fact that firms that
do not belong to national or foreign groups are at least as profitable as the other
firms in the sample is a useful result since it suggests that differences that we may
discover in their debt and investment choices are not due to independent firms’
poor economic performance.
Figures 2a to 2c highlight some interesting aspects of the dynamics of total
financial debt divided by total assets for the three sub-samples of companies.11 Two
remarks are in order. First, in Italy bank loans constitute the bulk of financial debt.
Although our sample does not contain separate information for bank and non-bank
debt, additional aggregate information on 779 large firms and 191 smaller firms
suggest that bank loans represent 79.2% of the total for the former and 85.4%
for the latter. Second, for affiliated firms the data do not allow us to distinguish
the portion of debt obtained from the parent company or from other members of
the group. There is a sense in which this is not fundamental for the object of our
investigation. In fact belonging to a group relaxes financial constraints for member
firms both because it creates an internal capital market and because it enhances the
access to funds external to the group. Moreover, the additional Mediobanca data
referred to above suggest that intra-group financial debt represents an average of
only 13.2% of total debt, so that debt external to the group constitutes the vast
majority of financial debt.12 This means that the flow of funds equations that we
estimate very likely provide information about the substitutability between internal
funds and financial resources external not only to the individual firm, but also to
the group (national or foreign multinational).
It is apparent from Figure 2 that members of both large national business groups
and other domestic companies are characterized by a higher leverage compared to
the sub-sample of affiliates to foreign multinationals. There are two interesting
differences in the dynamics of leverage that distinguish non-affiliated firms from
the rest. First, we observe at the beginning of the 1981-82 recession an increase
in the leverage of the median member of large national groups. Such an increase
occurs for foreign subsidiaries (below the 3rd quartile) as well, but does not occur
for non-affiliated firms. A possible interpretation is that, in the face of monetary
tightening and recession, firms that are members of larger organizations can make
up the shortfall in cash flow with access to external funds. This is consistent with
the flight to quality hypothesis (Bernanke, Gertler and Gilchrist, 1996), whereby
Figure 1. Pre dividend cash flow/total capital at replacement value: C/K.
Figure 2. Total financial debt/total capital at replacement value: B/K.
in bad times investors concentrate their funding on those firms with lower agency
costs due to asymmetric information and contract enforcement problems. Affiliated
firms with low debt are prime candidates for external funding. Second, following
the recovery (1983–1988), we observe a steady decline in leverage for affiliates
to national groups and (although less pronounced) for subsidiaries of foreign multinationals whereas this trend is much less pronounced or absent in the sample of
independent firms. A possible explanation for this finding is that affiliated firms
were able to issue new shares in the years of recovery, which were used partly to
repay debt. Unfortunately our data set does not contain information on new share
issues. Another possible explanation is that the degree of centralization of financial
management increased in business groups over the eighties, possibly inducing a
reduction in the level of debt of affiliated companies.
III. Flow of Funds Equations and the Relation between Internal and
External Finance
In the presence of asymmetric information between insiders and outside investors
and contract enforcement problems, internal and external finance are not perfect
substitutes in the sense that firms must pay a premium to obtain outside funds or,
in some cases, they may be completely rationed.13 Abstracting from tax considerations, firms have a preference to finance investment internally, then with debt. Only
as debt becomes riskier, do firms finally issue equity. This is what Myers (1984)
calls the “pecking order” theory of financing.14 On the basis of this argument we
would expect a negative association between the change in debt and internal finance
for given investment opportunities. Since the severity of asymmetric information
and contract enforcement problems is likely to vary across firms, the degree of
substitutability between internal and external sources is also expected to differ
across firms. In the context of a (change in) debt equation this means that we would
expect a negative and larger in absolute value coefficient on cash flow (used here as
a proxy for internal finance) for firms less affected by capital market imperfections.
This can be most easily understood if we assume that cash flow decreases, while
the expected profitability of investment remains constant. If a firm has to pay a
premium for borrowing, it will replace cash flow with debt, but less than one for
one. The greater the premium is, the smaller the increase in debt will be.
However, matters are more complicated than that. First, a change in cash flow
may change expectations on future profitability and shift the demand for funds
schedule. Second, a change in cash flow, if it is at least partly observed by outside investors and if it is thought to provide information about industry wide
trends, may lead to a revised valuation of collateralizable assets. For instance a
positive cash flow shock could increase the value of collateralizable assets and
lead to a decreased premium on debt. Both these effects may weaken or even
reverse the negative association between cash flow and change in debt. It has also
been suggested (Jensen and Meckling, 1976) that debt helps mitigate the conflict
between managers and shareholders due to the former is bearing the full cost but is
not capturing the entire gain from profit enhancement activities. In this situation
managers may have an incentive to consume perquisites and invest less effort.
Jensen (1986) suggests that debt constitutes a commitment to pay out cash, limits
managers’ discretion, and reduces the agency costs associated with the managersshareholders conflict (the “free cash flow” hypothesis).15 Also for this reason, one
may observe a positive association between internal finance and the flow of debt
(when cash flow is high the benefits of debt are also high), for a given level of
investment opportunities. However, this last argument is unlikely to be important
for independent Italian firms. Not only are most companies not quoted but top
management positions are very often filled by members of the family owning the
company, thus reducing agency costs. This problem is likely to be more severe
for large national business groups and, particularly, for foreign multinationals. It is
in fact in these larger organizations where the standard agency problems between
shareholders and managers are more likely to occur.
In the light of the above discussion the relationship between cash flow and financial debt is essentially an empirical matter. Summarizing, asymmetric information
considerations would lead to a negative relationship, given the state of expectations
and the severity of agency problems between managers and shareholders. The
negative association should be stronger the greater the substitutability between
internal and external sources. In the absence of managers – shareholders agency
problems, we can test for cross-firm differences in substitutability by allowing the
coefficients to differ in flow of funds equations between firms that are expected to
suffer less (affiliated firms) or more (independent firms) from information problems. Although expectational considerations may mitigate or reverse the negative
relationship between debt and cash flow, there is no obvious reason why this importance should vary systematically across firms. However, as already mentioned,
it is possible that differences in the cash flow coefficient may also reflect the agency
problems between managers and shareholders. Since such problems are likely to be
more important when cash flow increases than when cash flow decreases, we will
allow the cash flow coefficient to differ, depending which of these two cases occurs.
The coefficient is more likely to be negative when cash flow decreases because in
this case the pecking order argument is more likely to dominate. Note that if we find
in this case a coefficient which is negative and larger in absolute value for firms that
are group members (or subsidiaries of multinationals), that is a clear indication that
there is greater substitutability between internal and external sources of finance for
those firms, compared to non-affiliated firms. Agency problems between managers
and shareholders are surely more important for affiliates of national or foreign
multinational groups than for independent firms. This would tend to make the cash
flow coefficients less negative.
In this section we provide some econometric evidence on the relation between
internal finance and changes in debt outstanding.16 We use the following estimation
strategy. As a benchmark, we start from a very simple equation where the ratio of
the flow of total financial debt to total assets, (1B/K)t is explained in terms of
the lagged debt to total assets ratio, (B/K)t −1 , the ratio between cash flow and
total assets, (C/K)t , and the contemporaneous and lagged changes in real sales,
1 log Yt and 1 log Yt −1 . In this section we have defined cash flow gross of interest
payments to avoid introducing a spurious negative correlation between debt and
cash flow, which could result if the latter was defined net of interest payments.
Using appropriate firm type dummies, all the coefficients are allowed to differ
between affiliates to large national groups, foreign multinationals and independent
firms. We then interact (C/K)t with a dummy variable, Dt which equals one if the
ratio between cash flow and total assets increases between time t − 1 and t, and
zero otherwise.
To eliminate time invariant firm specific characteristics that affect capital structure choices we estimate all the equations in first differences. To allow for the
endogeneity of the regressors, estimation is carried out by the Generalized Method
of Moments technique, using appropriately lagged variables as instruments.17 Cash
flow and sales growth are likely to be correlated with the error term, as well as the
dummy variables used to define the regime of increasing (decreasing) cash flow.
Assuming that the idiosyncratic component of the error is serially uncorrelated in
the level equations, an error is generated with a moving average structure of order
one in the equations in differences, so that once-lagged variables are also correlated
with the error term. However, values of the regressors lagged twice or more will be
legitimate instruments. To check the validity of the assumptions embedded in our
model we calculate and report tests on both first and second order serial correlation
on the residuals (M1 and M2 respectively) as well as the Hansen/Sargan test of the
correlation of the instruments with the error term as a general test of mispecification. We also include three sets of year dummies (one for each sub-sample of firms)
in all the equations, to allow for time effects common to each group of firms. Year
dummies can capture, among other things, changes in expectations about demand
or changes in the interest rate and in tax parameters that are common to all firms in
each sub-sample.
In Table II we report the estimates of the specification of the flow of funds
equation that does not allow for asymmetry in the cash flow coefficients. The
Hansen/Sargan test does not suggest gross forms of mispecification. The M1 and
M2 tests suggest that the error term has a moving average structure of order one, as
one would expect in the differenced form of the equation, when the idiosyncratic
component of the error term in the level equation is serially uncorrelated. Both tests
suggest that variables lagged twice or more are legitimate instruments.18
The coefficient on lagged leverage is negative for all the sub-samples of firms.
This implies that the change in debt to capital ratio is negatively related to the
initial degree of leverage, as one would expect since more highly indebted firms
face greater risks of bankruptcy and greater agency problems. More importantly
for the purpose of this paper, the estimated cash flow coefficient is negative and
significant for both firms affiliated to large national business groups and subsidi-
Table II. Flow of funds equation: basic model; dependent variable: (1B/K)t ; sample period:
1980–1990; GMM estimates in first differences
Members of large national
business groups
(B/K)t −1
1 log Yt
1 log Yt −1
Sargan test
Foreign subsidiaries
Other national firms
1 Instrument list: All included variables lagged twice and three times.
2 Sub-sample specific time dummies included.
aries of foreign multinationals (−0.289 and −0.273). It is in contrast positive and
significant (0.178) for independent firms. These results are consistent with a smaller degree of substitutability between internal and external finance for independent
firms compared to affiliated firms.19 Finally, the overall effect of a firm’s growth
rate is negative for subsidiaries of foreign multinationals and positive for the other
two sub-samples of firms. A negative effect is likely to capture the fact that there is
more potential for managers to invest in value decreasing projects when there is a
lack of growth opportunities. In this situation, the benefits of debt in bonding managers are higher. Moreover, it is reasonable to assume that this particular agency
problem is more severe for foreign multinationals than for nationally owned firms.
This is certainly true in the case of non-affiliated firms where management and
ownership basically coincide, but it is also true (relative to foreign subsidiaries) for
members of large national groups. As a result, the role of sales growth in capturing
greater actual investment and hence a greater need for finance dominates in these
In the context of a flow of funds equation, a more rigorous test of the asymmetric
information hypothesis is to focus only on the cases where a decrease in cash flow
has occurred. In fact, for given investment opportunities it is when cash flow falls
that firms are more in need of external financial sources. Furthermore, a decrease in
cash flow, if observed by creditors, is likely to enhance agency problems between
borrowers and lenders. For this purpose, in Table III, the cash flow coefficient is
allowed to differ depending whether cash flow increases (Dt = 1) or decreases (Dt
= 0). Basically, all our previous results are confirmed. The coefficients on cash
flow are negative and significant in both regimes for the samples of both affiliates
to large national groups and subsidiaries of foreign multinationals. In contrast,
the coefficients are positive (and significant only in the “increasing” regime) for
independent firms.
Table III. Flow of funds equation: model with asymmetric effect of cash flow; dependent variable:
(1B/K)t ; sample period: 1980–1990; GMM estimates in first differences
Members of large national
business groups
(B/K)t −1
Dt (C/K)t
(1 − Dt )(C/K)t
1 log Yt
1 log Yt −1
Sargan test
Foreign subsidiaries
Other national firms
1 D = 1 if (C/K) > (C/K)
t −1 ; Dt = 0 otherwise.
2 Instrument list: All included variables lagged twice and three times.
3 Sub-sample specific time dummies included.
IV. The Effects of Financial Factors on Investment
In the previous section we found evidence that is consistent with the idea that
external finance is a very imperfect substitute for internal finance for independent
firms while the degree of substitutability is higher for affiliated firms. The next step
is to test whether imperfect substitutability has an impact on firms’ real policies.
We estimate a simple accelerator model of company investment20 with the ratio of
investment to fixed capital stock, (I /K F )t as the dependent variable.21 In addition
to the lagged investment rate and the contemporary and lagged changes in real
sales, we also include the ratio of cash flow to fixed capital stock, (C/K F )t and
the ratio of total debt to fixed capital stock, (B/K F )t as regressors. The cash flow
variable in this section has been defined net of interest payments, as in most recent
empirical papers on investment.22 As suggested by Fazzari, Hubbard and Petersen
(1988) differences in the size of the cash flow coefficients provide information on
the importance of liquidity constraints. This approach has been used to assess the
effects of group membership in Japan by Hoshi, Kashyap and Scharfstein (1991).
They found that firms that are members of an industrial/financial group are less
sensitive to cash flow fluctuations.
Obviously, a significant positive cash flow effect on investment does not necessarily reflect the presence of financing constraints, but may simply depend upon
the fact that cash flow conveys information on expected profitability. However, as
already mentioned, we minimize the risk of misinterpreting our empirical results
by focusing on differences in the coefficients on cash flow among sub-samples of
firms. In this case, if differences are found it is rather implausible to attribute them
to differences in expectations formation.23
Table IV. Investment equation: model with debt; dependent variable: (I /K F )t ; sample
period: 1980–1990; GMM estimates in first differences
Members of large national
business groups
(I /K F )t −1
(C/K F )t
1 log Yt
1 log Yt −1
(B/K F )t
Sargan test
Foreign subsidiaries
Other national firms
1 Instrument list: All included variables lagged twice and three times.
2 Sub-sample specific time dummies included.
As in the previous section, all estimates are carried out in first differences to
eliminate firm specific characteristics, using GMM estimation techniques to allow
for the endogeneity of the regressors. In all the reported equations three sets of
year dummies are included, to allow time specific effects to differ between the
three sub-samples of firms.
We start from a specification which includes the lagged dependent variable,
(I /K F )t −1 , the contemporaneous ratios of cash flow to fixed capital, the contemporaneous and lagged rates of change in output and the ratio between total
financial debt and fixed capital (Table IV). The inclusion of this variable is meant to
capture the concept that the premium on external finance is an increasing function
of leverage. We would therefore expect a negative relationship between the degree
of leverage and investment. What is more important in our framework however, is
that the coefficient should be more negative for those types of firms characterized
by more severe asymmetric information or contract enforcement problems.
The coefficients on cash flow for non-affiliated and affiliated national firms are
both positive and significant but the former is ten times bigger than the latter. The
coefficient for multinationals does not differ significantly from zero.24 This indicates that the availability of internal funds for firms that are either group members
or subsidiaries of foreign multinationals is less important for investment decisions,
since their group association makes it easier to tap the external capital market, as
well as the capital market internal to the group.25 Also, the coefficient of leverage
is negative, large in absolute value and significant for non-affiliated firms. For instance, when (B/K F )t increases from the first quartile (0.265) to the third quartile
(0.875), the investment to fixed capital ratio decreases by 0.018, which corresponds to a 14.6% decrease relative to the average value over the period (0.123).
The coefficient on leverage is not significantly different from zero for subsidiar-
ies of multinational corporations. Although we can reject the hypothesis that the
coefficient is zero for firms affiliated to large national groups, it is so minuscule
(−0.0003) that it is not economically significant. Finally, as expected, the coefficients on the rates of growth in sales are positive and significant in most cases.
Interestingly the sales coefficients are larger for multinational firms, suggesting
that they respond more strongly to demand stimuli, whereas the point estimates for
affiliated and independent national firms are remarkably similar.
To check the robustness of these results we reran the equation in Table IV, after
including lagged cash flow. We have also performed other experiments by allowing
more dynamics in our basic equation with the introduction of twice lagged regressors for all the variables in the model. We do not report these results for reasons
of space. The basic conclusions do not change and, in any case, the coefficients on
the additional lags do not differ significantly from zero. We have also replaced the
rates of change in output either with the levels of output or with the ratios of the
change in output to fixed capital, and we have also added the rates of change in
industry output to the equation. Our basic findings were not altered in any of these
alternative specifications.
Finally, as we have done for the flow of funds equations, we allow the cash
flow coefficient to vary depending whether cash flow increases (Dt = 1) or decreases (Dt = 0). The dummy, Dt can be interpreted as a discrete and firm specific
indicator of business cycle conditions, signalling respectively their improvement
or worsening.26 If asymmetric information or contract enforcement problems are
more likely to occur in bad times, we would expect a positive and greater in absolute value for the coefficient on cash flow when Dt = 0. As can be seen from Table
V, this indeed occurs for independent firms that are neither members of national
large groups nor foreign affiliates. In fact, the coefficients for independent companies are not only both significantly greater in absolute value than the coefficients
for the other types of firms but the coefficient when cash flow decreases is 2.7
times greater (and significantly so) than the coefficient for the cases when cash
flow increases (0.41 versus 0.15). The effect of cash flow is very small in both
regimes (although statistically different from zero) for subsidiaries of multinational
corporations and members of domestic groups.27
V. Conclusions
In this paper we provide an integrated approach to the analysis of capital market
imperfections at the firm level by analyzing the role of cash flow both in investment
and in (change in) debt equations. We apply our methodology to a large unbalanced
panel of Italian private companies, categorized by form of ownership. We find
strong empirical evidence from the flow of funds equations to support the hypothesis that being part of a national business group or being affiliated to a foreign
multinational alleviates capital market imperfections. Domestic firms that are not
Table V. Investment equation: model with debt and asymmetric effect of cash flow; dependent
variable: (I /K F )t ; sample period: 1980–1990; GMM estimates in first differences
Members of large national
business groups
(I /K F )t −1
Dt (C/K F )t
(1 − Dt )(C/K F )t
1 log Yt
1 log Yt −1
(B/K F )t
Sargan test
Foreign subsidiaries
Other national firms
1 D = 1 if (C/K F ) > (C/K F )
t −1 ; Dt = 0 otherwise.
2 Instrument list: All included variables lagged twice and three times.
3 Sub-sample specific time dummies included.
part of large national business groups have more difficulty in substituting internal
with external finance.
The different degree of substitutability between cash flow and debt has implications on firms’ real choices. In fact we find that investment decisions of
non-affiliated firms are much more sensitive to the availability of cash flow, confirming the crucial role that internal finance plays for them. Moreover, there is
evidence of an interesting asymmetry, in the sense that the effect of cash flow
is greater when the latter decreases than when it increases. Finally, members of
domestic groups and subsidiaries of multinational corporations show little or no
excess sensitivity to cash flow. Thus, the results obtained from the investment
equations are consistent with those from the debt equations and provide a stringent
consistency check on the conclusions based on the standard investment equations
The overall pattern of results emphasizes the problems of the financial system in
making external finance accessible to non-affiliated firms throughout the eighties.
Moreover, it is also possible that our results actually underestimate the problem
since firms that make our sample of non-affiliated companies represent the upper
tail of the size distribution of the total population of non-affiliated firms and are
less likely to face unfavorable lending terms than do even smaller firms. Finally,
the evidence we have presented lends support to the idea that business cycle shocks
may have important distributional consequences across various types of firms,
characterized in our case by different forms of ownership.
1. Williamson (1975) discusses this issue in the context of conglomerates. For a theoretical analysis
of costs and benefits of internal versus external capital markets in allocating financial funds see
Gertner, Scharfstein and Stein (1994).
2. In the early eighties new groups were set up and existing groups were expanded by splitting
single companies in several legally independent units because of fiscal benefits.
3. In 1990 the three main shareholders of quoted and unquoted joint stock companies owned on
average respectively 71% and 91% of total equity (Cannari, Marchese and Pagnini, 1993).
4. See Mayer (1990) for comparative evidence on financing patterns across countries.
5. Mediobanca, by far the most important investment bank in Italy, is defined by the economic
press as the “salon of Italian capitalism”.
6. Notice that the stock market plays on the whole a very limited role in Italy. For instance the ratio
of the market value of quoted companies to GDP in 1991 was 102.3% in the UK, 61.4% in the
US, but only 14.8% in Italy (FIBV – Fèdèration Internationale Bourses des Valeurs – Statistics,
1992). However, in the second half of the eighties some of the business groups members were
able to use the stock market as a source of funds. See also Section 2.
7. See Fazzari, Hubbard and Petersen (1988) for a seminal contribution in this area. See also Hoshi,
Kashyap and Scharfstein (1991) for an analysis of the implications of group membership in
8. A detailed data appendix is available from request from the authors. In the econometric estimates
firms with less than 4 consecutive observations have been excluded.
9. The figures that follow are obtained using the unbalanced dataset. In order to check that our
descriptive evidence is not contaminated by changes in the sample composition we have also
calculated the percentiles for a balanced sample of firms. The results are very similar and are not
reported here.
10. Since we compute our measure of cash flow by subtracting nominal interests, it incorporates the
component of interests which represents an advance on loan repayment. Since the data set does
not contain information on dividend payments, we cannot calculate retentions.
11. Trade debt is not included in total financial debt.
12. See Mediobanca, Dati Cumulativi di 1790 Società Italiane (1992). Transfer of financial resources
between associated firms could also occur through transfer prices. However there is no way,
using our data, to quantify the importance of this channel.
13. On rationing, see Stiglitz and Weiss (1981).
14. See also Myers and Majluf (1984).
15. See also Stulz (1990) for a formal model of financial structure based on the disciplinary role of
debt, in which debt payments reduce free cash flow.
16. The empirical literature on the determination of capital structure is vast. See for instance Titman
and Wessel (1988) and the extensive references in Harris and Raviv (1991) for the US. There
are few papers on Italy with the exceptions of Bonato and Faini (1990), and Faini, Galli and
Giannini (1991). A good survey on both theoretical and institutional aspects in the Italian context
is Bonato, Hamaui and Ratti (1993).
17. See Arellano and Bond (1988, 1991).
18. These comments also apply to the equations in Tables III through VI.
19. The cash flow coefficient for independent firms is significantly different at the conventional
statistical level from those for both multinationals and affiliated firms.
20. The importance of capital market imperfections for investment decisions can be investigated using a different approach based on the Euler equation for the capital stock. See Whited (1992) for
panel data evidence for the US, Bond and Meghir (1994) for the UK, and Galeotti, Schiantarelli
and Jaramillo (1994), and Rondi, Sembenelli and Zanetti (1994) for Italy.
Note that the median value of (I /K F )t is 0.104 for the sample of affiliates to large national
groups, 0.100 for the sample of foreign subsidiaries, and 0.102 for the other national firms.
The results are very similar if cash flow is defined gross of interest payments.
Gilchrist and Himmelberg (1995) present evidence that the mechanism of expectation formation
does not differ significantly among US firms.
The cash flow coefficient for non-affiliated firms is significantly different, at conventional levels,
from the one for members of national business groups (t = 8.15) and for foreign subsidiaries (t
= 8.60).
It is well known (see Carpenter (1995) and Kathuria and Mueller (1995)) that a positive cash
flow coefficient in investment equations is consistent with both the asymmetric information and
the managerial discretion hypothesis. For the reasons stated above affiliated firms are in principle
more likely to suffer from agency problems between shareholders and managers. The fact that
we find a positive coefficient only for independent firms suggests that we can safely interpret
our results as evidence in favor of the existence of capital market imperfections in Italy.
The hypothesis that financial constraints are more likely to affect fixed investment decisions in a
recession is tested for US firms in Gertler and Hubbard (1988) and Oliner and Rudebusch (1994)
using aggregate indicators of business cycle conditions.
The hypothesis that the cash flow coefficients are different in the two regimes cannot be rejected
at conventional levels for these two types of firms, but this is of little economic significance,
given the small size of the coefficients.
Arellano, M., and Bond, S.R. (1988) ‘Dynamic Panel Data Estimation Using DPD – A Guide for
Users’, Institute for Fiscal Studies Working Paper, 88/15, London.
Arellano, M. and Bond, S.R. (1991) ‘Some Tests of Specification for Panel Data: Monte Carlo
Evidence and an Application to Employment Equations’, Review of Economic Studies 58,
Bernanke, B., Gertler, M., and Gilchrist, S. (1996) ‘The Financial Accelerator and the Flight to
Quality’, Review of Economics and Statistics 78, 1–15.
Bonato, L. and Faini, R. (1990) ‘Empirical Determinants of Corporate Debt Decisions: Some
Evidence for Italy’, Mimeo.
Bonato, L., Hamaui, R., and Ratti, R. (1993) ‘Come spiegare la struttura finanziaria delle imprese
italiane’, Politica Economica 9, 49–103.
Bond, S.R. and Meghir, C. (1994) ‘Dynamic Investment Models and the Firm’s Financial Policy’,
Review of Economic Studies 61, 197–222.
Cannari, L., Marchese, G., and Pagnini, M. (1993) ‘Forma giuridica, quotazione e struttura proprietaria delle imprese Italiane: prime evidenze comparate’, Temi di Discussione, Bank of Italy, p.
Carpenter, R. (1995) ‘Finance Constraints or Free Cash Flow’, Empirica 22, 185–209.
Faini, R., Galli, G., and Giannini, C. (1992) ‘Finance and Development: The Case of Southern Italy’,
Temi di Discussione, Bank of Italy, p. 170.
Fazzari, S.M., Hubbard, R.G., and Petersen, B.C. (1988) ‘Financing Constraints and Corporate
Investment’, Brooking Papers on Economic Activity, pp. 141–195.
Galeotti, M., Schiantarelli, F., and Jaramillo, F. (1994) ‘Investment Decisions and the Role of Debt,
Liquid Assets, and Cash-Flow, Evidence form Italian Panel Data’, Applied Financial Economics
4, 121–132.
Gertler, M. and Hubbard, R.G. (1988) ‘Financial Factors in Business Fluctuations in Manufacturing’,
in Financial Market Volatility, Federal Reserve Bank of Kansas City.
Gertner, R.H., Scharfstein, D.S., and Stein, J.C. (1994) ‘Internal versus External Capital Markets’,
Quarterly Journal of Economics 109, 1211–1230.
Gilchrist, S. and Himmelberg, C. (1995) ‘Evidence on the Role of Cash Flow for Investment’, Journal
of Monetary Economics 36, 541–572.
Harris, M., and Raviv, A. (1991) ‘The Theory of Capital Structure’, Journal of Finance 46, 292–355.
Hoshi, T., Kashyap, A., and Scharfstein, D.S. (1991) ‘Corporate Structure, Liquidity and Investment:
Evidence from Japanese Industrial Groups’, Quarterly Journal of Economics 106, 33–60.
Kathuria, R. and Mueller, D. (1995) ‘Investment and Cash Flow: Asymmetric Information or
Managerial Discretion’, Empirica 22, 211–234.
Jensen, M.C. (1986) ‘Agency Costs of Free Cash Flow, Corporate Finance and Takeovers’, American
Economic Review. Papers and Proceedings 76, 323–329.
Jensen, M.C. and Meckling, W. (1976) ‘Theory of the Firm: Managerial Behavior, Agency Cost, and
Ownership Structure’, Journal of Financial Economics 3, 306–360.
Mayer, C. (1990) ‘Financial Systems, Corporate Finance and Economic Development’, in R.G. Hubbard (ed.), Asymmetric Information, Corporate Finance and Investment, University of Chicago
Press, pp. 307–332.
Myers, S.C. (1984) ‘The Capital Structure Puzzle’, Journal of Finance 39, 575–592.
Myers, S.C. and Majluf, N.S. (1984) ‘Corporate Financing and Investment Decisions when Firms
Have Information that Investors do not Have’, Journal of Financial Economics 13, 187–221.
Oliner, S.D, and Rudebusch, G.D. (1994) ‘Is there a Broad Credit Channel for Monetary Policy’,
Board of Governors of the Federal Reserve System, Economic Activity Section Working Paper
Series, p. 146.
Rondi, L., Sembenelli, A. and Zanetti, G. (1994) ‘Is Excess Sensitivity of Investment to Financial
Factors Constant across Firms? Evidence from Panel Data on Italian Companies’, Journal of
Empirical Finance 3/4, 365–383.
Stiglitz, J.E. and Weiss, A. (1981) ‘Credit Rationing in Markets with Imperfect Information’,
American Economic Review 71, 393–410.
Titman, S. and Wessels. R. (1988) ‘The Determinants of Capital Structure Choice’, Journal of
Finance 43, 1–19.
Whited, T. (1992) ‘Debt, Liquidity Constraints and Corporate Investment, Evidence from Panel
Data’, Journal of Finance 47, 1425–1460.
Williamson, O. E. (1975) Markets and Hierarchies: Analysis and Antitrust Implications, Free Press,
New York.