Documento de Trabajo 03/05
An Agency Model for Trade Credit Policy
Rafael Bastos
*
Julio Pindado
Universidad de Salamanca
Abstract
This article proposes an agency model to explain the trade credit offer to clients. Our
model is based on the existence of asymmetric information between sellers and buyers, which
results in the appearance of two phenomena known as adverse selection and moral hazard.
The former has already been explored by other authors, but not the latter, i.e., the possibility
of the buyer not paying the provider. The results obtained indicate that days of sales
outstanding of firms are positively related to adverse selection and negatively related to moral
hazard. In order to test the moral hazard hypothesis, we use three variables: variable cost,
demand elasticity and bad debts. Variable cost and demand elasticity present the expected
relation, but bad debts only presents the negative expected relation at low levels, which
suggests that when a firm presents high levels of bad debts the risk of the portfolio of clients
is also high. In this case, the clients are more likely to present a low liquidity situation and
consequently do not take advantage of the use of cash discounts. Traditional models are also
tested and compared with the proposed model. We did not find evidence to support tax theory
or to support the operational argument of transaction cost theory. We find weak evidence to
support the liquidity theory, while the asymmetric information theory was confirmed. A
comparison between the agency model proposed and traditional models concluded that the
Agency model reached better results in the explanation of the subject of study.
Keywords: Trade Credit, Asymmetric Information, Adverse Selection, Moral Hazard
Code JEL: G30
Universidad de Salamanca
Dpt. Administración y Economía de la Empresa
Salamanca, E3700
Spain
Tel.: +34 923 294400 ext. 3506
Fax: +34 923 294715
E-mail: pindado@usal.es
*
Corresponding author: Julio Pindado, The authors are grateful to an anonymous referee for
helpful comments and suggestions. Pindado thanks the research agency of the Spanish
Government, DGI (Project BEC2001-1851) and the Junta de Castilla y Leon (Project SA
033/02) for financial support.
1
1. INTRODUCTION
Trade credit is a very important source of financing for companies.
Although it is an old practice, it is not completely understood. Numerous
theories have been proposed to explain its existence and use, but none of them
can provide a complete explanation of the topic. While some of the models are
more consistent in the case of certain industries or categories of products others
work better in a financially constrained environment.
Four types of explanation can be enumerated as follows: a model based on
transactions costs arguments was proposed by Schwartz (1974); financial
models were first suggested by Emery (1984); a tax based model was suggested
by Brick and Fung (1984); and an asymmetric information model was suggested
by Smith (1987).
This paper focuses on explaining trade credit proposing an agency model,
based on the agency problem described by Jensen and Meckling (1976). We
consider the relation between a firm and its clients an agency relation from
which two observable facts arise: adverse selection and moral hazard. Adverse
selection happens when there is ex-ante asymmetric information between buyers
and sellers. In this case buyers do not know ex-ante the characteristics and
quality of the goods that are being acquired. The moral hazard is the possibility
of clients not paying, since there is ex-post asymmetric information.
Using a sample of UK manufacturing firms, we test the proposition that
trade credit is explained by a trade-off between these two phenomena. Our
results support the positive relation between adverse selection and trade credit
and the negative relation between moral hazard and trade credit, the first relation
is also consistent with asymmetric information theory. This result supports the
agency model suggested.
Other authors have used the adverse selection problem to explain trade
credit, but none of them have included the moral hazard. This is the main
contribution of our research.
This paper is organized as follows. Section 2 describes the state of the art
of the topic studied and summarizes the main theories and research on trade
credit. Section 3 proposes an agency model and Section 4 describes the data and
methodology used. Section 5 shows our interpretations for the results using
univariate and multivariate tests to contrast the agency model and also test
traditional models. In the last section we conclude and suggest some further
research studies.
2. STATE OF THE ART
Trade credit is one of the oldest forms of corporate financing and it is still
very important nowadays, it refers to financing provided by a seller to its buyer
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
2
(Wei and Zee, 1997). To understand the concept of trade credit it is important to
know the range of alternative credit arrangements that can occur in trade.
Depending on credit policy, payment can be made at different times. It can
occur before delivery, on delivery or after delivery. In the last case, the seller
can offer discounts for prompt payment or not, depending on trade
arrangements. When payment does not occur before delivery or at delivery time,
trade credit is being extended and the seller assumes the credit risk. Otherwise,
trade credit is not being offered and the buyer assumes the risk. Figure 1 can
better illustrate these situations.
Figure 1: Different payment times implied by credit policy.
Payment Before
Delivery
Cash on
Delivery
Delivery
Postpayment
Trade Credit
Payment Due if
Discount Taken
Payment Due
on Net Date
Total Credit Period
Credit Period if
Discount Taken
Note: Adapted from Ng, Smith and Smith (1999)
Although trade credit is a very useful source of resources for different kinds of
firms, its explanation is not very clear yet, as pointed out by Long, Malitz and Ravid
(1993). In the last three decades, many theories and models have appeared to
explain trade credit. Most of these theories rely on market imperfections such as the
existence of tax, transactions costs and asymmetric information, as shown in Figure
2. In the rest of this section we are going to detail each of those theories.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
3
Figure 2.: Trade Credit Theories
2.1. Existence of tax
The decision to accept or not accept a credit depends on other possibilities
of credit. A buyer must compare different alternatives to make sure which is the
best choice. In trade between a seller and a buyer, a post payment can be
offered, but it is not free, there is an implicit or an explicit interest rate included
in the final price. Therefore, to find the best source of financing, the buyer must
check out the real borrowing cost in other financing providers.
Brick and Fung (1984) suggest that tax must be considered in order to
compare trade credit cost with other alternatives of financing. The main reason
for this is that if buyers and sellers are in different tax brackets, they have
different borrowing costs. This is true, since interests are tax deductible. An
argument to reinforce this idea is that by offering trade credit, sellers can offer
discounts for cash payment that are tax deductible as well.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
4
The authors’ hypothesis is that companies in a higher tax bracket tend to
offer more credit than those in lower tax brackets; otherwise, buyers could
borrow more cheaply directly from a financial institution and will prefer to buy
at a lower cash price. Another conclusion is that firms allocated to a given
industry and placed in a tax bracket below industry average cannot profit from
offering trade credit. Brick and Fung (1984) even suggest that firms cannot both
use and offer trade credit.
2.2. Transactions Cost Theory
First developed by Schwartz (1974), this theory conjectures that suppliers
may have an advantage over traditional lenders in checking the real financial
situation or the credit worthiness of their clients. Suppliers also have a better
ability to monitor and force repayment of the credit. All these superiorities may
give suppliers a cost advantage when compared with financial institutions. Three
sources of cost advantages were cataloged by Petersen and Rajan (1997):
advantage in information acquisition, in controlling the buyer and in salvaging
value from existing assets.
The first source can be justified by the fact that sellers can get information
about buyers faster and at lower cost because it is obtained in a normal course of
business. That is, the frequency and the amount of the buyer’s orders give
suppliers a notion of the client’s situation; the buyer’s use of discounts for early
payment may serve to alert the supplier of a weakening in the credit-worthiness
of the buyer, and sellers usually visit customers more often than lenders. Smith,
in his model (1987), concludes that in a two-part credit with high interest rate,
those buyers that do not choose to take advantage of the discount can be
interpreted as high risks, because they may be having financial difficulties.
Depending on penalties for later payers, simple net terms can produce a similar
sign.
The second source can be seen as the power of the seller to threaten buyers.
In other words, in some cases there are just a few alternative suppliers for the
needed product and, consequently, buyers have very restricted choice. In this
case, suppliers can threaten to cut off future supplies if they note a reduction in
the chances of repayment. Compared with providers, banks and other lenders do
not have the same threat force. This advantage can become stronger when
buyers only represent a small part of the provider’s sales or the provider is part
of a network and future community sanctions can be made, not only by one
seller, but by a group, which makes this threat much stronger (Kandori, 1992).
This hypothesis was empirically confirmed by McMillan and Woodruff (1999).
Another interesting finding in this strand of literature was made by Petersen and
Rajan (1997), whose empirical results suggest that debtors are less willing to
repay a distressed seller. Their argument is that threats of cutting off future
supplies, made by a supplier with financial problems, are not so credible.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
5
The seller’s ability to salvage values from existing assets is the third source
of cost advantages. In the case of buyer default, the seller can seize the goods
that are supplied, of course, financial institutions can reclaim the firm’s assets as
well. The difference between them is that as companies trading are very often
from the same industry, the supplier already has a network to sell the goods and
consequently repossessing and resale costs would be lower. Two interesting
approaches related to this cost advantage were made by Mian and Smith (1992)
and Petersen and Rajan (1997). The former posit that the more durable the
goods, the better collateral they provide and the greater the credit offered by the
suppliers. The latter point out that the extent to which the customers transform
the product is also very important. The less they are transformed, the easier it
will be for the provider to repossess and sell the asset using the same channel.
An important reflection must be made about goods value relativeness. Ng,
Smith and Smith (1999) consider that the value of a product is relative to
different institutions, i.e., if a product has more value as a collateral to a seller
than to financial institutions, the seller may have cost advantage in recuperating
the product and selling it again. In this situation, providers tend to offer cheaper
credit than financial institutions because of the reduction of the credit risk.
Another study related to transaction cost theory was made by Emery
(1987), and hypothesizes that there is a positive relation between demand
variability and credit offered. The argument to support this idea is that as
demand fluctuates, a company has two traditional reactions: production or price
adjustment. The author believes that both are very costly and a better decision
could be taken. The seller could change trade credit terms according to demand.
Terms can be relaxed when demand drops and be tightened when demand
increases. In this case, trade credit can be seen as an operational tool. Long,
Malitz and Ravid (1993) obtained empirical evidence supporting this hypothesis.
Their results show that firms with variable demand extend more trade credit than
firms with stable demand.
2.3. Liquidity Theory
This theory suggests that credit rationed companies use more trade credit
than those with normal access to financial intermediaries. The central point of
this idea is that when there is a restricted monetary policy, the offer of trade
credit can make up for the reduction of the credit offer from financial
intermediaries. In accordance with this view, large firms, presenting good
liquidity or better access to capital markets can finance those constrained by the
policy.
Many approaches have tried to obtain empirical evidence to support this
theory. Nielsen (1999) using small firms as a proxy for credit rationed
companies found that in monetary contraction they react by borrowing more
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
6
from their suppliers. As a result, trade credit tends to be less used in countries
where companies have good relations with banks.
As liquid firms are less likely to demand trade credit and more likely to offer
it, a negative relation between buyer’s access to other sources of financing and
trade credit use is expected, as Petersen and Rajan (1997) find. Another
expected negative association, also found by these authors is between proxies
for the strength of bank relationships and trade credit demand.
2.4. Asymmetric information theory
Sellers, usually, do not know the real credit-worthiness of their buyers and
also buyers do not have knowledge about goods quality. To solve the first
problem, Smith (1987) suggests a model where sellers offer two-part credit
terms because they can recognize potential defaults faster than financial
intermediaries. And, as presented in Section 2.2, there are many other arguments
proposing that providers have costs advantages in acquiring knowledge about a
buyer’s financial situation. But, what about the second problem? Smith also
proposed that with asymmetric information about product quality, sellers offer
trade credit to permit buyers to verify product quality before payment.
The reason that leads suppliers to extend this credit is that they have an
immense interest in a customer’s success, since they expect the client to buy
more goods and service from them in the future. Although success of the buyer
is important, the quality of the product sold is also crucial, and could determine
new purchases. As a consequence, companies very often offer money-back
guarantees as warranties as well.
Trade credit has some advantages when compared with money-back
guarantees and warranties. First, in a case of money-back or warranties, if the
seller is not in business any more, the buyer can be damaged. Second, when
payment is made at the time of sale, a client, seeking the advantages of the
money-back system, may try to convince the seller that the quality of the
product is not as promised.
As pointed out by Smith (1987), one of the major proposals of trade credit
is to allow clients to assess product quality prior to payment, although this is not
true for some categories of product. Therefore, this theory works better in some
industries whose product quality is unknown at the moment of purchase. If this
idea is correct, trade credit will be higher in the trade of products where quality
is indefinite at a prior moment and the purchase is not frequent. Otherwise, trade
credit will be lower in trades of perishable items where acquisition is very
frequent.
Many researchers have studied trade credit from this point of view, see for
instance: Lee and Stowe (1993), Long Malitz and Ravid (1993), Deloof and
Jegers (1996) and Wei and Zee (1997). In general, the main results of these
authors are: i) Small firms tend to offer more trade credit than large firms, since
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
7
small firms still have to establish their reputation about product quality. ii) Firms
with longer production cycles prolong their collection period, since they produce
high-quality goods. iii) Firms selling products whose quality is difficult to
measure extend long credit periods because customers must have enough time to
assess quality. iv) Sellers of low quality goods may try to pass them off as highquality goods. In this case, as the cost of extending trade credit increases, these
companies will have less incentive to cheat on the information on quality.
2.5. What these models do not explain or why models are not complete
Although many theories have attempted, through different ways, to explain
the existence of trade credit, they cannot provide a complete explanation of the
topic. While some of the models are more consistent with the case of certain
industries or categories of products others work better in a financially
constrained environment. According to Frank and Maksimovic (1998) “the
existing theories show effects that may be important in specific circumstances,
but they do not capture what seems to be central for explaining the wide-spread
use of trade credit and the empirical patterns of its use”. Let us check theory by
theory and identify some inconsistencies or situations not very well explained.
Tax theory suggests that companies in high tax brackets tend to offer credit
to those in low tax brackets. Many research studies have found empirical
evidence to support this, but this explanation does not seem to be enough since it
can not explain trade credit between companies situated in the same tax bracket.
If trade credit is an operational tool and exists to minimize transaction costs
as in Ferris (1981), a reduction in the level of trade credit use would be expected
since many improvements in transaction technologies have taken place. This
reduction has not been observed.
Liquidity theory supposes that more constrained companies use more trade
credit than those with easier access to financial intermediaries. This may be an
explanation, but once more, does not seem to be enough since it does not explain
why companies not financially constrained also use trade credit.
In the presence of asymmetric information, trade credit is offered to allow
clients to check the real quality of the goods before payment, but traditional
models only consider this phenomenon, called adverse selection, and forget
another one that is the risk of goods not being paid for, the moral hazard.
3. AN AGENCY MODEL FOR TRADE CREDIT POLICY
The theories presented above are not able to explain the central point about
why some products and services do not have credit or should be paid in cash.
Because of this we proposed an agency model to explain trade credit from the
supplier’s point of view.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
8
From the agency problem proposed by Jensen and Meckling (1976), we
can consider the relation between a firm and its clients an agency relation from
which two observable facts arise: adverse selection and moral hazard. Adverse
selection occurs when there is ex-ante asymmetric information between
providers and buyers. In this case, clients do not know ex-ante the characteristics
and quality of the goods that are being acquired. As described in Section 2.4,
this phenomenon has already been studied by other authors, but not the moral
hazard, which consists of the possibility of the contractual relationship not being
carried out by the client, facilitated by the ex-post asymmetric information. The
worst consequence of this is to generate bad debts.
The main difference between this agency model and other models that
attempt to explain trade credit assuming the existence of asymmetric
information is the inclusion of moral hazard. Thus we define a function to
explain trade credit as follows: DSO=φ(ϖ,µ), where DSO is the day of sales
outstanding, ϖ is the cost of adverse selection and µ the moral hazard. From this
function we expect ϖ to be positively related to DSO and µ to be negatively
related.
We compute a proxy for days of sales outstanding (DSO) as the natural
logarithm of the average collection period measured by accounts receivable
divided by daily sales.
To proxy the adverse selection phenomenon, we find some variables in the
literature that have been used for this purpose. Representing technical industries
we construct a dummy variable SEC which takes the value of one if the goods
sold need a significant time to verify their quality level, and zero otherwise. To
classify each company in our sample, we follow Titman and Wessels (1988)
defining technical industries as those in SIC 3400-39991; we expect SEC to be
positively related to DSO variable. To classify firms whose products are easy to
observe (perishable), we follow Long, Malitz and Ravid (1993) and construct
another dummy variable PER which takes the value of one if the company’s
products are considered perishable, those in SIC code between 2000 and 2199
(food and agricultural), and zero otherwise. We expect this variable to be
negatively related to DSO. To proxy firm reputation we use the variable SIZE,
measured as the natural logarithm of a firm’s total assets; we expect a negative
correlation between SIZE and DSO. With the aim of proxying the fixed assets of
the firms we calculate FIX as 1-(Fixed Assets/Total Assets). This variable has
been used by Prowse (1990) who suggests that the greater the proportion of
fixed assets in a firm, the smaller the asymmetric information problem, thus we
expect it to be directly related to DSO. Another alternative proxy for the adverse
selection is EBIT (a measure for supplier results); the bigger it is, the smaller the
consequences of adverse selection to the buyer. The EBIT variable is calculated
by dividing earnings before interest and taxes for total assets; this variable is
1
Long, Malitz and Ravid (1993) used the same classification.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
9
expected to be negatively correlated to DSO. The last variable we use to proxy
for the adverse selection is DPA2 which consists of the days to pay accounts
receivable. This variable is used because high quality firms may increase the
cost of trade credit by financing their purchases. The expected relation with
DSO is positive.
The above described adverse selection phenomenon gives rise to our first
hypothesis:
Hypothesis 1: The greater the presence of adverse selection the greater the
trade credit offer. Therefore we expect days of sales outstanding to be directly
related to days to pay accounts payable and those sectors producing high
quality goods and inversely related to firms earnings, firm size, fixed assets and
those sectors producing perishable goods.
As regards proxy moral hazard we suggest three different variables: VCO
(variable cost); BDE (bad debts provision); and DEL (demand elasticity). We
base our suggestion of those proxies on Oh (1976). In accordance with this
author, the decision to change a current credit policy to a new one depends on
whether the marginal profitability is greater than or equal to the associated
marginal opportunity cost. But which benefits and costs are related to trade
credit and must be compared? When a firm decides to extend its credit policy it
is probably expecting a sales increase and hence will assume new costs such as
collection costs, bad debt losses and variable costs. Thus, the credit policy
decision involves a tradeoff between profits on the marginal credit sales and the
marginal costs.
According to trade credit literature, the more extensive the credit offer, the
greater the amount of bad debt losses the seller will support. To mitigate this
problem, companies have incentives to offer bigger cash payment discounts.
These discounts must be attractive enough to convince even the risky buyer to
pay sooner, because if this occurs, it may reduce the possibility of bad debt
losses since it restricts the time available to buyers to develop more problems.
We proxy for the bad debt using the balance data provision-other3, which
includes provision for bad debts and losses, divided by total assets. This variable
is expected to be negatively related to DSO.
When trade credit is extended, if the buyer does not pay, the provider will
support all costs generated by the new sales. Therefore, the damage caused by
2
We take this variable as calculated by Compustat Global Vantage as follows: accounts
payable divided by the sum of inventories plus cost of goods sold plus depreciation and
amortization minus the previous value for inventories minus depreciation and amortization.
This total is divided by 360.
3
This item also includes contingencies, provision for employee profit sharing, provision for
fluctuation and valuation of investments, provision for foreign exchange loss and provision
for taxes.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
10
the moral hazard will be greater, since a buyer’s default is the worst
consequence. For those reasons we expect that the greater the variable cost, the
more rigid the credit policy will be. To proxy for this we use the cost of goods
sold divided by total assets. We expect VCO to be inversely related to DSO.
The last factor we use to proxy for the moral hazard phenomenon is
demand elasticity. According to Borde and Mcarty (1998), an early payment
discount is actually a price reduction and if buyers are price elastic, cash
discounts may generate greater demand for the firm’s products. Following their
argument, and the concept of time value of money, a long period of payment can
be considered a price reduction as well, and in this case it can generate
aggregated demand. This revenue coming from the new clients resultant from a
less rigid credit policy can compensate or mitigate some risks assumed by the
provider such as the bad debts and the variable costs. Therefore, the bad
consequences of moral hazard will decrease when a firm’s clients are demand
elastic. We proxy for DEL using the differences between sales as a proportion of
total assets in yeart and sales as a proportion of total assets in yeart-1, divided by
the differences between the average collection period in yeart and average
collection period in yeart-1. We expect DEL to be negatively related to DSO.
Equation 1 shows our measure for the moral hazard as a function of three
components.
MORALHAZARD = φ ( DEL,VCO, BDE )
(1)
Our second hypothesis is based on the above described moral hazard
phenomenon.
Hypothesis 2: The greater the moral hazard consequences, the shorter the
trade credit offer will be. Thus, we expect days of sales outstanding to be
positively related to demand elasticity and negatively related to variable cost
and bad debts provision.
Using these two hypotheses we can test our model, suggesting that trade
credit can be explained by an agency model considering a trade-off between
adverse selection costs and moral hazard consequences. We can express this
model as follow:
DSOi = β0 + β1DELi + β2VCOi + β3BDEi + β4 SECi + β5 PERi + β6 SIZEi + β7 FIXi + β8 EBITi + β9 DPAi + ε i
where all the variables are displayed in Table 1.
[INSERT TABLE 1 ABOUT HERE]
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
(2)
11
4. DATA AND METHODOLOGY
Our sample was first composed of 348 active manufacturing companies in
the UK (SIC 2000 through 3999) from the Compustat Global Vantage database
for the three-year period ending in 2002. We decided to restrict our sample to
manufacturing firms because it is in this category where trade credit is produced
in its habitual form. To reduce the variability all variables are averaged over the
three-year period. We excluded from our sample 9 outlying observations and
another 45 were lost because of missing values. This reduced our sample to 294
observations. The firms in our sample are industry dispersed, i.e., the sample
does not appear to be more concentrated in any industry. Producers of
perishables only represent 9% of our sample and high quality producers
represent 44%. Table 2 shows the distribution between industries for our
original sample.
[INSERT TABLE 2 ABOUT HERE]
Descriptive data for the variables used in our analysis and Pearson´s
Correlation are in Table 3 and 4, respectively. We can see that firms in our
sample are relatively large, with average total assets of over £1.9 million. The
days of sales outstanding (DSO) is around 76 days and the median is 72.42 days.
Our analyses show that firms in our sample have on average more than
20% of their assets represented by accounts receivable. Some of them have more
than 50% of their total assets represented by accounts receivable.
[INSERT TABLE 3 ABOUT HERE]
[INSERT TABLE 4 ABOUT HERE]
5. RESULTS
To contrast the two hypotheses related to our proposed agency model and
find some evidence on trade credit theory, we will first develop a univariate test
to compare results with those Long, Malitz and Ravid (1993) found for USA
companies and also check the behavior of our new variables included in the
model and then we will try to find more conclusive evidence using a
multivariate method. In order to test traditional models we also included other
variables.
5.1. Univariate test
We use a cross-sectional t-test of differences between firms with days of
sales outstanding below the median (72.42 days) and those above the median.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
12
We also use the Levene test4 to contrast the equality of the variances. Except for
SIZE, EBIT, DPA and DEL variables, the t-test shows that, for the agency
model proposed, all differences are significant at the 5% level or less (see Table
5). Although the BDE variable shows significant differences between those two
groups, results were the opposite to what was expected. That is, firms from the
group High DSO showed greater values. Consequently, evidence is mixed to
support hypotheses 1 and 2.
[INSERT TABLE 5 ABOUT HERE]
Companies offering trade credit above the median of our sample show
higher levels of FIX; this supports the asymmetric information proposition that
the consequences of adverse selection are mitigated by the existence of fixed
assets. Thus, buyers will be less affected by adverse selection costs.
Our analyses are consistent with the findings of Long, Malitz and Ravid
(1993). The outcomes of the two dummy variables included in t-test contrast are
as expected. As a result, High-Tech producers usually extend more credit for
buyers to allow them to check the real quality of the goods before payment and
producers of perishable and easy-to-observe goods do not.
Differences between the two groups are not significant for the variable
EBIT, thus we do not confirm the proposed idea which suggests that the greater
the earnings of the provider, the smaller the cost of adverse selection to the
buyer. For the same reason, we do not confirm the supposition that large firms,
which probably have a good reputation, extend less credit than smaller firms that
probably have not yet built up their reputation.
Finally, as we have projected, the variable costs present differences
between the two groups. The VCO, one of our measurements of the moral
hazard phenomenon, on average, is statistically different between companies
that extend large credit terms and those that only extend short terms of credit.
The High DSO group has smaller levels of VCO indicating that the bigger the
variable cost, the greater the consequences of moral hazard and the shorter the
credit terms will be.
TD and CVS, variables included in the test in order to contrast traditional
models, were not significant.
According to univariate t-test results, evidence partially confirms product
quality theory of trade credit but does not support liquidity and tax theory.
Neither do we have enough evidence to make any inference about cost
transaction theory. Results are also inconclusive about the relevance of the
moral hazard influence in trade credit decision.
4
We assume equal variances for those variables with a significance value above 0.05.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
13
5.2. Results of multivariate test
To confirm our findings with a greater level of reliance or even to find
more evidence to support the proposed agency theory and then confirm the
relevance of moral hazard, we use multiple regression through the ordinary least
square method. Table 6 shows regression results.
[INSERT TABLE 6 ABOUT HERE]
Our regression results present an adjusted R2 of 0.356. The picture that
emerges from Table 6 is quite similar to the one from the univariate test,
although there are some differences. The principal difference is that by using the
multivariate test SIZE, EBIT, DPA and the DEL variables are now significant.
Those differences give more support to our proposed agency model.
The negative and significant relation between SIZE and DSO supports the
asymmetric information theory based on product quality guarantee argument
and is not consistent with liquidity theory which proposes that firms of known
liquidity, with large access to the financial market, should extend trade credit to
those more constrained. Instead, we find that smaller companies, with their
reputation not yet constructed, extend larger trade credit to allow clients to
check product quality and then build their status as high quality firms.
Once more supporting the asymmetric information theory, the variable FIX
shows a positive relation with DSO. According to the construction of the
variable, the greater its value, the smaller the proportion of fixed assets for a
company. Thus, trade relations between buyers and providers with a high
proportion of fixed assets present smaller problems of adverse selection.
The results for the SIC dummy variable are also as expected. Firms
categorized as high quality producers extend more credit than those categorized
as low quality producers. Therefore, companies whose products are considered
high quality offer a period of credit great enough to permit their clients to check
the real quality of goods before payment.
The PER dummy variable, as predicted, is negatively related to DSO.
Again, results are consistent with product quality theory. Therefore, firms whose
products are classified as easy-to-observe quality (perishable goods) offer
shorter terms of credit than others.
Contrary to the results for the univariate test, the EBIT variable is now
significant showing the expected relation with DSO. The more firms earn the
shorter is the credit offered. These results confirm our evidence supporting the
product quality argument that the greater the seller’s earnings, the weaker the
consequences of adverse selection to buyers.
Analyzing the same EBIT coefficient we do not find support to accept
liquidity and tax theory. According to liquidity theory, firms presenting a
comfortable financial situation offer credit to those presenting an opposite
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
14
situation. We understand that highly profitable companies have a good financial
situation and our results show evidence supporting a negative, not a positive
relation, between EBIT and DSO. To reject the tax hypothesis we must interpret
Brick and Fung’s (1984) proposition that companies in higher tax brackets offer
credit to those in lower tax brackets. Otherwise, buyers could borrow with a
lower after-tax cost than sellers and then, buy cash. This argument works only in
the case of the existence of seller earnings, because this ensures a smaller aftertax cost for the debt. As our results show that the greater the EBIT the shorter
the DSO variable, we have evidence to reject the tax proposition.
Although EBIT and SIZE results are not consistent with liquidity theory,
outcome for the DPA variable is consistent with liquidity theory, because it
predicts that credit offer is inverse-related to credit used. If the main argument of
the liquidity theory is that companies offer trade credit because they have wide
access to the financial market and at a lower cost, they should not need to use
trade credit. Therefore, trade credit extended and trade credit used are mutually
exclusive.
According to our findings, the greater the demand elasticity, the smaller the
consequences of the moral hazard borne by the provider and the less risky will
be the trade credit offered by a firm. Thus, firms with a high level of demand
elasticity will extend more trade credit.
Regression results are in accordance with the arguments presented in
Section 3, which predict that the variable costs negatively influence the length of
the trade credit. Thus, sellers with a high level of variable costs will be more
strongly affected in case of buyer default than those with lower levels and
consequently will extend less credit.
The unexpected results for the BDE variable may be related to the poor
liquidity situation of risk clients. According to our previous argument, firms
whose clients are considered “high risk”, i.e. presenting high levels of BDE,
should raise cash discounts offered to attract risk clients to pay soon and thus
avoid bad debts. Unfortunately, these risk clients, probably because of their
insolvency situation, can not make use of discounts offered and then pay the
total value on the net date. To solve this problem we run another regression to
check if there is a determinate level of bad debts that make trade credit offered
show a different sign.
To run this regression we constructed a dummy variable that takes the
value of 1 if a company is placed above the 66.66 percentile for the BDE
variable and zero otherwise. The new model is expressed as follow:
DSOi = β0 + β1DELi + β2VCOi + β3BDEi + β4 BDE2i + ε t
(3)
where BDE2 is the BDE variable multiplied by the dummy variable constructed.
Results are reported in Table 7.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
15
[INSERT TABLE 7 ABOUT HERE]
Our results can now confirm hypothesis 2 that BDE and DSO are inversely
related. We even find evidence to support our previous supposition that when
clients present a very high level of risk, seller’s efforts to diminish the average
collection period are never enough and in this case a positive relation between
trade credit and bad debts arises. For the other variables representing the moral
hazard, coefficients keep their expected sign and significance.
Overall, our results totally confirm the product quality guarantee argument
and consequently the positive influence of the adverse selection phenomenon on
trade credit offered and also confirm the moral hazard argument. Therefore, we
confirm hypothesis 1, except for the DPA variable, and even confirm hypothesis
2. We also find evidence to reject tax theory and mixed results for the liquidity
theory.
5.3. Testing traditional models
To strengthen our findings regarding traditional theories and also to
compare our findings using the agency model proposed, we run five more
regressions.
To test the tax theory, which predicts that sellers offer trade credit because
they have access to funds at a lower after-tax cost than buyers, we use TD, a
measurement for total debt. If sellers have a high level of debt, they are less
likely to obtain another loan with tax deductible interest and thus reduce its
borrowing cost. TD is the sum of long term debt and debt in current liabilities as
a percentage of total asset averaged over the three year period. We expect TD to
be negatively related to DSO. We also use EBIT as an independent variable
since if a seller does not show earnings its tax will not be deductible. Equation 4
shows the tax model and the results are reported in Table 8.
DSOi = β0 + β1EBITi + β2TDi + ε i
(4)
[INSERT TABLE 8 ABOUT HERE]
Although this model only shows an adjusted R2 of 0.014 and the result for
the EBIT variable is not significant, its outcome shows that the coefficient for
the TD variable is significant at 5% level, thus supporting the argument that
companies showing high levels of debt tend to offer less credit. As a
consequence, this result supports the tax theory.
Liquidity theory is tested using a model composed of three explicative
variables: SIZE, EBIT and DPA. Equation 5 shows this model.
DSOi = β0 + β1SIZEi + β2 EBITi + β3 DPAi + ε i
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
(5)
16
As this theory suggest that firms with wider access to the financial market
and showing more liquidity tend to extend credit to those with restricted access
to the financial market and also predicts that firms will either purchase on credit
or sell on credit, but not both, DSO is expected to be directly related to SIZE and
EBIT and inversely related to DPA. Results are reported in Table 9.
[INSERT TABLE 9 ABOUT HERE]
Results for this test offer mixed support for the liquidity theory. As the
result shows, on the one hand, the coefficient for EBIT is not significant and for
SIZE it shows a negative sign, which does not support the liquidity theory but
gives support to the product quality argument of the asymmetric information
theory. On the other hand, the coefficient for the DPA variable is negative as
predicted by the theory. Therefore firms tend to offer trade credit or buy on
credit but not both and large firms tend to extend less credit than small firms.
To test the operational argument of transactional cost theory we regress the
DSO using as an explanatory variable the CVS (coefficient of variation of
sales). This theory predicts that trade credit is important only when demand
variability exists. CVS is calculated as the standard deviation of total revenue
over the four year period ending in 2002. We expect DSO to be positively
related to CVS. The transaction cost model is shown in equation 6 and the
results are given in Table 10.
DSOi = β0 + β1CVSi + ε i
(6)
[INSERT TABLE 10 ABOUT HERE]
The regression result for this model shows that CVS is negatively related to
DSO. This relation does not support the operational argument of the cost
transaction theory that trade credit is important only when demand fluctuates.
To test the asymmetric information theory based on the argument that as
clients do not know ex-ante goods quality and thus trade credit must be offered
to allow them to check product quality before payment, we regress DSO using
all those variables used in the agency model to proxy the adverse selection
phenomenon without including those variables used to proxy the moral hazard.
Equation 7 shows the regression model and Table 11 reports our results.
DSOi = β0 + β1SECi + β2 PERi + β3SIZEi + β4 FIXi + β5 EBITi + β6 DPAi + ε i
(7)
[INSERT TABLE 11 ABOUT HERE]
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
17
Except for SIZE and DPA, the results for all the variables included in the
model support the product quality guarantee argument. Thereby, firms from very
specific industries that produce high quality goods tend to offer more credit than
those from industries of perishable goods. Also, small firms and those
presenting a small proportion of fixed assets in their balance sheets tend to offer
more credit than larger firms with high proportion of fixed assets. Results for
these variables are very similar to those already found in our agency model.
As these test shows, although some of these models were supported while
others were not or have presented only mixed support, their explanatory power
was not as strong as the agency model proposed. While the agency model
presents an adjusted R2 of 0.356, the traditional model with the higher adjusted
R2 was the one testing the asymmetric information theory with the value of
0.224. Thereby we run one last regression including all the variables used in
these traditional models to compare results against the agency model. Equation 8
shows the regression model and Table 12 reports the result.
DSOi = β + β1SECi + β2 PERi + β3SIZEi + β4 FIXi + β5 EBITi + β6 DPAi + β7CVSi + β8TDi + ε i (8)
0
[INSERT TABLE 12 ABOUT HERE]
Especially in three points, the results reported in Table 12 lead us to the
same conclusions we already have. First, the agency model can explain more
than other traditional models, since again the adjusted R2 is not as great as that
presented in Table 6. Second, evidence again supports the asymmetric
information theory. Third, again the liquidity theory only has weak support.
The main difference between this test and those testing each theory
separately changes the conclusion about tax theory. If when we tested the tax
theory, the TD variable was negatively related to DSO as expected, in this
regression it is not significant even at a level of 10%. Therefore, these new
results do not support the tax argument.
6. CONCLUSIONS
This paper proposes an agency model to explain trade credit offered. Based
on the agency problem suggested by Jensen and Meckling (1976), we consider
that in a relation between a seller and a buyer two phenomena are observable:
adverse selection and moral hazard. The former exists when, because of
asymmetric information, buyers do not know ex-ante characteristics or quality
of the product being acquired. The latter is the possibility of the contractual
relationship not being carried out and the worst consequence is the appearance
of bad debts. This paper also checks some traditional theories such as liquidity,
tax and transaction cost. To contrast these models a sample of UK
manufacturing companies is used.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
18
According to the results, traditional models such as liquidity, tax and
transaction cost do not offer a complete explanation for the trade credit
extended. We did not find evidence to support tax and transaction cost theories.
For the liquidity theory, results were only mixed. The product quality argument
is valid but not sufficient and the moral hazard phenomenon is useful to
complement the model. Therefore, our agency model based on a trade off
between adverse selection and moral hazard offers a better explanation since it
considers aspects related to trade credit risk such as the variable cost assumed by
the provider, the demand elasticity and bad debts.
Our model suggests and finds evidence that high quality, smaller and less
profitable firms, which have less fixed assets, are more likely to extend credit
because their relationship with clients presents more adverse selection problems.
We also find that firms with high levels of variable cost and low levels of
demand elasticity extend less credit because they support higher levels of moral
hazard consequences.
We also found evidence that the greater the bad debts for a company the
shorter the average collection period will be. This occurs because sellers tend to
offer more attractive discounts for cash payments or will reduce credit terms
when clients’ risk increases. Furthermore, we also found evidence that when bad
debts are very high, sellers’ efforts to diminish the average collection period and
trade risk are never sufficient, since their clients probably do not present
sufficient liquidity to take advantage of cash discounts or to pay up to date.
Concluding, as adverse selection is already a recognized phenomenon
related to trade credit, moral hazard has just started to be such. Thus, further
research can be carried out to reinforce this model. Better proxies for moral
hazard can be established for the future and research from the buyer’s point of
view should be done in order to contrast these ideas, since buyers’ risk is easy to
measure using traditional models.
BIBLIOGRAPHY
Biais, B. and Gollier C. (1997): “Trade Credit and Credit Rationing”, Review of
Financial Studies, 10:4, pp. 903–937.
Borde, S. F. and McCarty, D. E. (1998): “Determining the Cash Discount in the
Firm’s Credit Policy: An Evaluation”, Journal of Financial and Strategic
Decisions, 11:2, pp. 41-49.
Brick, I. E. and Fung, W. K. H. (1984): “Taxes and the Theory of Trade Debt”,
Journal of Finance, 39:4, pp. 1169–1175.
Brennan, M. J.; Maksimovic, V. and Zechner, J. (1988): “Vendor Financing”,
Journal of Finance, 43:5, pp. 1127–1141.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
19
Deloof, M. and Jegers, M. (1996): “Trade Credit, Product Quality, and
Intragroup Trade: Some European Evidence”, Financial Management,
25:3, pp. 33-43.
Deloof, M. and Jegers, M. (1999): “Trade Credit, Corporate Groups, and the
Financing of Belgian Firms”, Journal of Business Finance & Accounting,
26:7/8, pp. 945-967.
Emery, G.W. (1984): “A Pure Financial Explanation for Trade Credit”, Journal
of Financial and Quantitative Analysis, 19:3, pp. 271-285.
Emery, G. W. (1987): “An Optimal Financial Response to Variable Demand”,
Journal of Financial and Quantitative Analysis, 22:2, pp. 209–225.
Ferris, J. S. (1981): “A Transactions Theory of Trade Credit Use”, Quarterly
Journal of Economics, 96:2, pp. 243–270.
Frank, M. and Maksimovic, V. (1998): “Trade Credit, Collateral, and Adverse
Selection”, Unpublished Manuscript, University of Maryland.
Hill, N.C. and Riener, K.D. (1979): “Determining the Cash Discount in the
Firm’s Credit Policy”, Financial Management, 8:1, pp. 68-73.
Jensen, M. C. And Meckling, W.H. (1976): “Theory of the Firm: Managerial
Behavior Agency Costs and Ownership Structure”, Journal of Financial
Economics, 3:4, pp. 305-360.
Kandori, M. (1992): “Social Norms and Community Enforcement”, Review of
Economic Studies, 59:1, pp. 63-80.
Lee, Y. W. and Stowe, J. D (1993): “Product Risk, Asymmetric Information,
and Trade Credit”, Journal of Financial and Quantitative Analysis, 28:2,
pp. 285–300.
Long, M. S.; Malitz I. B. and Ravid, S. A. (1993): “Trade Credit, Quality
Guarantees, and Product Marketability”, Financial Management, 22:4, pp.
117–127.
McMillan, J. and Woodruff, C. (1999): “Interfirm Relationships and Informal
Credit in Vietnam”, The Quarterly Journal of Economics, 114:4, pp. 12851320.
Meltzer, A. H. (1960): “Mercantile Credit, Monetary Policy, and Size of Firms”,
Review of Economics and Statistics, 42:4, pp. 429-437.
Mian, S. L. and Smith C. W. (1992): “Accounts Receivables Management
Policy: Theory and Evidence”, Journal of Finance, 47:1, pp. 169–200.
Ng, C. K.; Smith J. K. and Smith, R. L. (1999): “Evidence on the Determinants
of Credit Terms used in Interfirm Trade”, Journal of Finance, 54:3, pp.
1109–1129.
Nilsen, J. (1994): “Trade Credit and the Bank Lending Channel of Monetary
Policy Transmission”, working paper, Princeton University.
Oh, J. S. (1976): “Opportunity Cost in the Evaluation of Investment in Accounts
Receivable”, Financial Management, 5:2, pp. 32-36.
Petersen, M. A. and Rajan, R. G. (1997): “Trade Credit: Theories and
Evidence”, Review of Financial Studies, 10:3, pp. 661–691.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
20
Prowse, S. D. (1990): “Institutional Investment Patterns and Corporate Financial
Behavior in the US and Japan”, Journal of Financial Economics, 27:1, pp.
43-66.
Schwartz, R. A. and Whitcomb, D. K. (1979). “The Trade Credit Decision”, in J.
L. Bicksler (ed), Handbook of Financial Economics, North-Holland,
Amsterdam.
Schwartz, R. A. (1974): “An Economic Model of Trade Credit”, Journal of
Financial and Quantitative Analysis, 9:3, pp. 643-657.
Smith, J. K. (1987): “Trade Credit and Informational Asymmetry”, Journal of
Finance, 42:4, pp. 863–872.
Titman, S. and Wessels, R. (1988): “The Determinants of Capital Structure
Choice”, Journal of Finance, 43:1, pp. 1-19.
Wei, P. and Zee, S. M. L. (1997): “Trade Credit as Quality Signal: An
International Comparison”, Managerial Finance, 23:4, pp. 63-72.
Wilner, B. S. (2000): “The Exploitation of Relationships in Financial Distress:
The Case of Trade Credit”, Journal of Finance, 55:1, pp. 153–178
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
21
APPENDIX
Table 1: Expected Sign
Phenomenon
Adverse Selection
Adverse Selection
Adverse Selection
Adverse Selection
Adverse Selection
Adverse Selection
Moral Hazard
Moral Hazard
Moral Hazard
Variables
SEC
PER
SIZE
FIX
EBIT
DPA
DEL
VCO
BDE
Expected Sign
+
+
+
+
-
Table 2: Industry Distribution
Sic
20
21
22
INDUSTRY
Food and Kindred Products
Tobacco Products
Textile Mill Products
Apparel and Other Finished Products Made from Fabrics and
23
Similar Materials
25
Furniture and Fixtures
26
Paper and Allied Products
27
Printing, Publishing, and Allied Industries
28
Chemicals and Allied Products
29
Petroleum Refining and Related Industries
30
Rubber and Miscellaneous Plastic Products
31
Leather and Leather Products
32
Stone, Clay, Glass, and Concrete Products
33
Primary Metal Industries
Manufactured Metal Products, Except Machinery and Transport
34
Equipment
35 Industrial and Commercial Machinery and Computer Equipment
Electronic and Other Electrical Equipment and Components,
36
Except Computer Equipment
37
Transport Equipment
Measuring, Analyzing, and Controlling Instruments;
38 Photographic, Medical and Optical Goods; Watches and Clocks
39
Miscellaneous Manufacturing Industries
Total
-
Nº.
29
3
11
%
8%
1%
3%
13
7
11
27
45
3
12
3
21
11
4%
2%
3%
8%
13%
1%
3%
1%
6%
3%
10
30
3%
9%
50
19
14%
6%
32 9%
11 3%
348 100%
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
22
Table 3: Descriptive Statistics
DSO
SIZE
EBIT
FIX VCO DEL BDESECPER DPA TD CVS ACCTS
N
294
294
294 294 294 294 294 294 294
294 294 294
294
Average
76,76 1804,91 155,84
,39 ,72 -,06 ,02 ,46 ,08 431,03 ,23 ,27
,21
Stand. Dev.. 35,40 9598,38 1116,92
,85 1,07 ,53 ,02 ,50 ,27 2789,03 ,29 1,32
,10
Minimum
8,17
,19 -3030,25-13,01 ,00-4,80 ,00 ,00 ,00
,12 ,00 ,01
,01
Median
72,42
113,55
4,80
,47 ,62 -,01 ,00 ,00 ,00
62,79 ,19 ,14
,20
Maximum 295,14 148073,67 15439,00 1,0016,68 4,33 ,14 1,00 1,0042267,834,21 22,54
,53
Notes:
DSO stands for the days of sales outstanding averaged over the three-year period.
SIZE represents the three-year period average of firms total assets .
EBIT represents the three-year period average of firms EBIT.
FIX is the fixed assets proportion on firms total assets averaged for the three-year period.
VCO represents the three-year period average for cost of goods sold divided by total assets.
DEL stands for the three-year period average for the differences between sales as a proportion of total assets in
year x and sales as a proportion of total assets in year x-1, divided by the differences between average collection
period in year x and average collection period in year x-1.
BDE is the three-year period average of the data provision-other divided by total assets.
SEC is a dummy variable which takes the value of 1 if goods sold need a significant time to verify their quality
and zero otherwise.
PER stands for a dummy variable which takes the value of 1 if goods sold are perishable or easy to observe
quality and zero otherwise.
DPA represents the days to pay accounts payable averaged over the three-year period.
TD represents the sum of long term debt and debt in current liabilities as percentage of total asset averaged over
the three year period.
CVS is the standard deviation of total revenue over the four year period ending in 2002.
ACCTS stands for the accounts receivable as a proportion of total asset averaged over the three year period.
Table 4: Pearson´s Correlation
LNDSO
DEL
VCO
BDE
SEC
PER
SIZE
FIX
EBIT
DPA
Pearson´s Correlation
Sig. (2-tailed)
Pearson´s Correlation
Sig. (2-tailed)
Pearson´s Correlation
Sig. (2-tailed)
Pearson´s Correlation
Sig. (2-tailed)
Pearson´s Correlation
Sig. (2-tailed)
Pearson´s Correlation
Sig. (2-tailed)
Pearson´s Correlation
Sig. (2-tailed)
Pearson´s Correlation
Sig. (2-tailed)
Pearson´s Correlation
Sig. (2-tailed)
Pearson´s Correlation
Sig. (2-tailed)
LNDSO DEL
1 ,190
, ,001
,190
1
,001
,
-,236 -,300
,000 ,000
,126 ,016
,031 ,787
,276 ,144
,000 ,013
-,367 ,030
,000 ,607
-,153 ,090
,008 ,122
,181 ,233
,002 ,000
,028 ,242
,636 ,000
-,134 ,012
,022 ,842
VCO BDE SEC
-,236 ,126 ,276
,000 ,031 ,000
-,300 ,016 ,144
,000 ,787 ,013
1 -,026 -,072
, ,663 ,221
-,026
1 ,149
,663
, ,010
-,072 ,149
1
,221 ,010
,
-,038 -,129 -,267
,516 ,027 ,000
-,262 ,327 -,121
,000 ,000 ,037
-,833 ,056 ,146
,000 ,335 ,013
-,837 ,022 ,041
,000 ,713 ,481
-,092 ,037 ,017
,117 ,529 ,776
PER
-,367
,000
,030
,607
-,038
,516
-,129
,027
-,267
,000
1
,
,186
,001
-,047
,426
,047
,423
,112
,054
SIZE
-,153
,008
,090
,122
-,262
,000
,327
,000
-,121
,037
,186
,001
1
,
,201
,001
,265
,000
,127
,029
FIX
,181
,002
,233
,000
-,833
,000
,056
,335
,146
,013
-,047
,426
,201
,001
1
,
,880
,000
-,012
,835
EBIT
,028
,636
,242
,000
-,837
,000
,022
,713
,041
,481
,047
,423
,265
,000
,880
,000
1
,
,015
,797
Notes:
** Significant at the 0.01 level (bilateral).
* Significant at the 0.05 level (bilateral).
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
DPA
-,134
,022
,012
,842
-,092
,117
,037
,529
,017
,776
,112
,054
,127
,029
-,012
,835
,015
,797
1
,
23
Table 5: Test of independent samples
Variables Low DSO High DSO
SIZE
FIX
EBIT
SEC
PER
VCO
DEL
BDE
DPA
TD
CVS
5,2668
,2907
-,0745
,36
,14
,9136
-,0966
,0116
636,9437
,2385
,3349
4,9155
,4960
-,0700
,55
,01
,5375
-,0308
,0191
227,8906
,2142
,2085
T-Statistic of
Difference
-1,524
2,075**
,036
3,218***
-4,251***
-3,049***
1,059
2,707***
-1,251
-,719
-,821
Notes:
** Significant at the 0.05 level.
*** Significant at the 0.01 level.
a
Low DSO is defined as DSO less than the median of 72.42 days.
b
High DSO is defined as DSO greater than the median of 72.42 days.
Table 6: Multivariate Tests of the Agency Model
Independent Variables
Coefficient
Constant
DEL
VCO
BDE
SEC
PER
SIZE
FIX
EBIT
DPA
T-Statistic
4,527
0.103
-0.283
2.241
0.101
-0.464
-3.623E-02
0.107
-0.289
-2.318E-05
Adjusted R square: 0.356
F-Statistic: 19.024
Notes:
* Significant at the 0.10 level
** Significant at the 0.05 level.
*** Significant at the 0.01 level.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
52,533***
2.362**
-6.747***
2.241**
2.101**
-5.280***
-2.827***
1.757*
-6.033***
-2.868***
24
Table 7: Multivariate Tests of the Moral Hazard
Independent Variables
Coefficient
Constant
VCO
DEL
BDE
BDE2
T-Statistic
4.293
-8.539E-02
0.112
-14.212
16.188
109.388***
-3.327***
2.172**
-1.703*
1.996**
Adjusted R square: 0.085
F-Statistic: 7.829
Notes:
* Significant at the 0.10 level
** Significant at the 0.05 level.
*** Significant at the 0.01 level.
Table 8: Multivariate Tests of the Tax Theory
Independent Variables
Coefficient
Constant
TD
EBIT
T-Statistic
4.304
-0.356
-6.210E-02
104.048***
-2.408**
-1.561
Adjusted R square: 0.014
F-Statistic: 3.014
Notes:
* Significant at the 0.10 level
** Significant at the 0.05 level.
*** Significant at the 0.01 level.
Table 9: Multivariate Tests of the Liquidity Theory
Independent Variables
Coefficient
Constant
SIZE
EBIT
DPA
T-Statistic
4.428
-3.724E-02
3.096E-02
-1.921E-05
Adjusted R square: 0.032
F-Statistic: 4.188
Notes:
* Significant at the 0.10 level
** Significant at the 0.05 level.
*** Significant at the 0.01 level.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
57.414***
-2.625***
1.196
-1.979**
25
Table 10: Multivariate Tests of the Transaction Costs Theory
Independent Variables
Coefficient
Constant
CVS
T-Statistic
4.238
-3.779E-02
152.972***
-1.835*
Adjusted R square: 0.008
F-Statistic: 3.368
Notes:
* Significant at the 0.10 level
** Significant at the 0.05 level.
*** Significant at the 0.01 level.
Table 11: Multivariate Tests of the Asymmetric Information Theory
Independent Variables
Coefficient
Constant
SEC
PER
SIZE
FIX
EBIT
DPA
T-Statistic
4.178
0.132
-0.459
-1.760E-02
0.276
-0.169
-1.421E-05
52.221***
2.570**
-4.821***
-1.359
4.484***
-3.447***
-1.623
Adjusted R square: 0.224
F-Statistic: 15.059
Notes:
* Significant at the 0.10 level
** Significant at the 0.05 level.
*** Significant at the 0.01 level.
Table 12: Multivariate Tests of the Asymmetric Information, Transaction
Costs, Liquidity and Tax Theories.
Independent Variables
Coefficient
Constant
SEC
PER
SIZE
FIX
EBIT
DPA
CVS
TD
T-Statistic
4.276
0.149
-0.446
-2.538E-02
0.184
-0.505
-1.515E-05
-0.376
0.209
Adjusted R square: 0.275
F-Statistic: 14.889
Notes:
* Significant at the 0.10 level
** Significant at the 0.05 level.
*** Significant at the 0.01 level.
DOCUMENTOS DE TRABAJO “NUEVAS TENDENCIAS EN DIRECCIÓN DE EMPRESAS” DT 03/05
dtecadem@eco.uva.es www.uva.es/empresa
View publication stats
52.597***
2.994***
-4.798***
-1.929*
2.885***
-5.671***
-1.790*
-4.725***
1.455