SSRN-id967386

7/29/2019 SSRN-id967386

1/41Electronic copy of this paper is available at: http://ssrn.com/abstract=967386

Bank Financing as an Incentive for Earnings Management in

Business Start-ups

By

Nancy Huyghebaert

Heidi Vander Bauwhede

Marleen Willekens

Katholieke Universiteit Leuven

The authors thank Christof Beuselinck and Marc Deloof for useful comments on an earlier draft of this paper.

Corresponding author:

Nancy Huyghebaert, Katholieke Universiteit Leuven, Department of Accountancy, Finance and Insurance,

Naamsestraat 69, 3000 Leuven, Belgium; tel: 00 32 16 326 737, fax: 00 32 16 326 732, e-mail:

[email protected]

7/29/2019 SSRN-id967386

2/41Electronic copy of this paper is available at: http://ssrn.com/abstract=967386

Bank Financing as an Incentive for Earnings Management in

Business Start-ups

By

Nancy Huyghebaert

Heidi Vander Bauwhede

Marleen Willekens

Katholieke Universiteit Leuven

AbstractIn this paper, we investigate whether business start-ups in need for external financing manage

their earnings in the years prior to obtaining a first bank loan. Newly established firms typically

face valuable growth opportunities whereas their external financing sources usually are limited

to bank loans and trade credit. Due to lack of track record, information asymmetries between

entrepreneurs and potential financiers tend to be large. Business start-ups, as a result, may

manage their earnings upwards when applying for a first bank loan, to influence the lending

decisions of banks. We use a unique sample of Belgian start-up firms to test this hypothesis.

Earnings management behavior is captured through two measures of current accruals: trade

accruals and non-cash working capital accruals. Our multivariate analyses indicate that, aftercontrolling for elements that affect the normal level of accruals, business start-ups indeed have

significantly increased levels of current accruals and thus earnings in the years preceding a first

bank loan. However, we find no evidence that bank lending decisions are actually influenced by

this earnings management behavior.

JEL: G21, G32, G33

7/29/2019 SSRN-id967386

3/41

1

I. Introduction

Newly established firms typically face valuable growth opportunities, which need to be financed.

In this paper, we investigate whether earnings management by such firms if any is inspired

by the need for external financing. For start-ups in traditional industries, the external financing

sources are usually limited and mainly consist of bank loans and trade credit.1

Start-up firms

incentives to manage earnings upwards in order to influence the lending decisions of suppliers

likely are limited. The evidence in Ng et al. (1999), for example, suggests that suppliers do not

actively collect information on a firms creditworthiness before granting trade credit, as the

supply of trade credit is largely industry determined and varies little across firms within the same

industry. And, as trade credit is generally a short-term financing source, suppliers can react

quickly to newly obtained bad information by refusing to roll over the trade credit, which

protects them against adverse selection and moral hazard problems. So, suppliers likely base

their lending decisions on other considerations than analyzing information included in the annual

accounts.2

These ideas are confirmed by Sercu et al. (2003), who report that privately held

companies do not seem to target suppliers when managing their earnings.

For bank lending, the situation is different and, as a result, banks may carefully screen the

financial statements of start-up clients for the following reasons. First, banks do not only lend

larger amounts as compared to suppliers, but also extend loans with longer maturities, which

makes them more vulnerable to information and incentive problems. Furthermore, when a

debtor goes bankrupt, banks being specialists in the evaluation of creditworthiness can be

held liable by other creditors, for instance suppliers, for having given a too optimistic signal on

the firms credit quality. In the context of newly established firms, failure rates are typically

1Berger and Udell (1998) discuss the sources of financing firms can access according to their age. Typically,

venture capital is only available for firms in specific industries, and in Continental Europe, venture capitalists

largely finance firms in the growth rather than the start-up stage. For start-ups in traditional industries, Huyghebaert

(2006) and Huyghebaert and Van de Gucht (2007) show more explicitly that bank debt and trade credit are the main

financing sources.2

Wilner (2000), for example, points out the role of the supplier implicit equity stake, i.e. the rents that suppliers can

earn on future sales of their product to the client firm to which they extend trade credit. Because of this implicit

equity stake, suppliers may be willing to extend credit even to client firms that face potentially high adverse

selection and risk shifting problems. Huyghebaert (2006) and Huyghebaert and Van de Gucht (2007) find empirical

support for these ideas.

7/29/2019 SSRN-id967386

4/41

2

high in the first few years after start-up. Dun & Bradstreet (1994), for instance, document that

approximately 50% of all firms that failed in 1993 did so during the first five years of their

existence. Later studies report similar statistics (see, for example, Huyghebaert and Van de

Gucht, 2004). Such high failure rates likely urge banks to careful lending decisions.

Newly created firms have a limited track record and no established relationships with

banks, especially in the case of a first loan application. Examining annual accounts when

available may then help banks to determine the firms assets that can serve as collateral and its

future cash flow generation, to gauge the firms debt capacity. When start-ups are aware (or

assume) that a careful financial viability assessment will be done based on company accounts, an

incentive is created for them to manage their earnings upwards in the year(s) before receiving a

first bank loan. Survival of start-up firms often depends upon obtaining the necessary financial

resources to finance assets and operations,3

which creates a further incentive to manage earnings

upwards when in need of bank financing. Furthermore, as newly established firms typically

have no or few taxable earnings (see, for example, Laitinen, 1994; Huyghebaert, 2006), they

have no incentive to manage their earnings downwards. Due to the limited size of most firms,

company accounts issued by business start-ups also remain unaudited, which actually provides

opportunities for undetected earnings management.

In this paper, we examine whether business start-ups manage their earnings upwards in

the years before receiving afirstbank loan. Given that information asymmetries between start-

up firms and banks are largest at this point in time and as such firms have not yet built a

reputation for servicing their debt well, we expect earnings management to be particularly likely

in this case. One way to manage earnings is the use of discretionary accounting accruals to raise

reported earnings relative to the actual cash flows.4

In this paper, we focus on earnings

management through currentaccruals rather than total accruals because management of current

3Holtz-Eakin et al. (1994), for example, find that entrepreneurs whose financial constraints are reduced after

receiving an (exogenous) inheritance face significantly increased survival chances.4 Accruals consist of a non-discretionary or normal component that changes with the firms level of operating

activities, and a discretionary or abnormal component that is the result of earnings management.

7/29/2019 SSRN-id967386

5/41

3

accrual accounts is less visible than management of non-current accrual accounts (see also

Guenther, 1994; Teoh et al., 1998a). Current accruals relate to those short-term assets (such as

accounts receivable and inventories) and liabilities (such as accounts payable) supporting the

day-to-day operations of the firm.

Using data on a unique sample of Belgian business start-ups that published annual

accounts before receiving a first bank loan, we find that current accruals are significantly higher

in the years preceding a first bank loan as compared to the years thereafter, ceteris paribus.

These findings are consistent with accruals being managed upwards before obtaining a first bank

loan. The results further indicate that firms that are short of cash and with limited tangible assets

have significantly higher current accruals, ceteris paribus. We interpret the latter relations as

reflecting that more financially constrained business start-ups are more inclined to manage their

earnings upwards, ceteris paribus. Indeed, firms that lack cash may highly need the bank

financing, but when the amount of assets that can be pledged as collateral is limited, they may

expect banks to be less willing to lend.

As a supplementary analysis to this paper, we also examine whether start-up accruals,

which may at least partly result from earnings management, influence the bank lending decision,

but find no corroborative evidence for such a relation. Rather, we document that economic

variables shape the banks credit-granting decision in the context of business start-ups. More

specifically, start-up firms with significant financing needs resulting from growth opportunities

are more likely to obtain bank debt, ceteris paribus. Internal cash generation (current

profitability) also increases the likelihood of bank lending but, consistent with the pecking order

model of capital structure, firms with accumulated cash reserves resulting from pastprofitability

are less likely to borrow from banks. Finally, banks tend to lend more eagerly to firms with

higher tangible fixed assets whereas firm risk (activity risk nd financial risk) negatively affects

the bank lending decision. Overall, our results could reflect that 1) banks carefully examine

company accounts before lending, and are not being misled by start-up earnings management

7/29/2019 SSRN-id967386

6/41

4

when deciding on granting a loan, nor do banks penalize companies for the information risk

resulting from potential earnings management or 2) banks do not attentively scrutinize company

accounts in the case of business start-ups. Ravid and Spiegel (1997), for example, argue that the

relatively small size of start-up loans and the complexity associated with screening and

monitoring of these firms renders such activities cost ineffective from the point of view of banks

(see also Huyghebaert and Van de Gucht, 2007).

The remainder of this paper is organized as follows. In Section II, we develop our main

hypotheses based on prior earnings management studies. Also, we link start-up accrual accounts

to subsequent bank lending decisions. In Section III, we present our research design and

empirical models. Section IV reports the results of our empirical analyses on earnings

management by business start-ups in the period around their first bank loan. Section V

concludes this paper.

II. Literature and hypotheses

The accounting and finance literature has extensively studied incentives for and constraints on

earnings management. Incentives for earnings management, for example, stem from the role of

accounting information in specific financial contracts, such as bonus compensation plans or debt

covenants, or in assessing firm performance and value. By managing the reported income figure,

managers can artificially meet bonus targets, avoid debt covenant violations, or influence the

price of a firms stock. Various studies find evidence consistent with these hypotheses (see, for

example, Healy (1985) for the bonus plan hypothesis, DeFond and Jiambalvo (1994) and Jaggi

and Lee (2002) for the debt covenant hypothesis). The literature has also elaborated on

mechanisms that constrain firms earnings management behavior. The quality of the external

auditor (see Francis et al., 1999; Becker et al., 1998), large institutional shareholders (Chung et

al., 2002), and investor protection (Leuz et al., 2003) are examples of constraining mechanisms.

Some studies have examined earnings management prior to or around a particular event, such as

7/29/2019 SSRN-id967386

7/41

5

initial public offerings (Aharony et al., 1993; Friedlan, 1994; Teoh et al., 1998a), seasoned

equity offerings (Rangan, 1998; Teoh et al., 1998b), acquisitions (Erickson and Wang 1999;

Heron and Lie, 2002) or venture capital financing (Beuselinck et al., 2003). Most of these

studies find clear evidence of upward earnings management prior to or around the studied event.

Our paper is the first study on earnings management around bank lending decisions. In

particular, we examine earnings management by business start-ups around the time of their first

bank loan. Typical for newly established ventures is that they are privately held and ownership

is not separated from firm management. Hence, these firms have no incentive to manage their

earnings to influence stock prices and/or managerial compensation. In addition, there are other

relevant differences between start-up and mature firms, relating to tax incentives and factors that

constrain earnings management. In contrast to other privately held firms, business start-ups have

no incentive to decrease their earnings for tax reasons (see Sercu et al., 2003), for they have only

few taxable earnings (e.g., Laitinen, 1994; Huyghebaert, 2006). Furthermore, start-up firms are

not subject to the scrutiny of high-quality external auditors or large institutional shareholders.

The reason is that newly established enterprises usually do not exceed the size criteria that

trigger a mandatory external audit,5

and are not being financed by large institutional investors,

but by one or more entrepreneurs.

Business start-ups, however, are in constant need for new funds to finance their

investment opportunities. Persson (2004), for example, documents that the size of surviving

start-up firms has doubled eight years after their establishment (see also Audretsch, 1995). Not

surprisingly, start-up survival often depends upon being able to secure sufficient external

financing to initiate investment projects (e.g., Holtz-Eakin et al., 1994; Persson, 2004). Besides

supplier credit, bank debt is a main source of financing for these firms (e.g., Berger and Udell,

1998; Huyghebaert, 2006; Huyghebaert and Van de Gucht, 2007).

5In Belgium, the country from which we draw our sample, companies are required to appoint a statutory auditor if

they employ more than one hundred people, or if two of the following size criteria are met: a) total assets exceed

3,125,000; b) turnover exceeds 6,250,000; and c) more than 50 people are employed.

7/29/2019 SSRN-id967386

8/41

6

Although banks are generally considered to have inside information about the (mature)

firms they lend to, this is less true in a start-up context. Indeed, as these firms have not yet

developed a relationship with a house bank and as their track record is short, information

asymmetries between entrepreneurs and banks tend to be large. Furthermore, since we look at

the period around afirstdebt grant, the type of firms we study have not yet been able to build a

valuable reputation for servicing their debt well, thereby reducing potential agency problems

(e.g., Diamond, 1989). These information and incentive problems cannot be ignored given the

high failure risk that business start-ups face. Hence, accounting information may be a welcome

source to assess the creditworthiness of loan applicants in the context of business start-ups and

first-time loans. Since net income is positively related to profitability and solvency (that is, to

the extent that earnings are retained within the firm), start-ups have clear incentives to manage

their earnings numbers upwards in order to positively affect the lending decision by banks. Prior

research on quoted companies shows that earnings contain value-relevant information, in

addition to the information contained in cash flows (the main difference between the two

measures being accruals) (see, for example, Bowen et al., 1987), that accruals improve the

ability of earnings to measure firm performance as measured by stock returns (e.g., Dechow,

1994), and that accrual-based earnings better predict future operating cash flows than current

operating cash flows (Dechow et al., 1998). In addition, Subramanyam (1996) reports that the

part of accruals that is the result of earnings management (i.e. the discretionary or abnormal

accruals6) is priced by the market and predicts future profitability. Based on the arguments

above, we test the following hypothesis:

HYPOTHESIS 1: Earnings management by business start-ups is larger in the years before

they raise a first bank loan than in the years afterwards.

6 The other part of accruals, the normal or non-discretionary part, is the part that changes with the firms level of

operating activities (cfr. infra).

7/29/2019 SSRN-id967386

9/41

7

Given the role of financial constraints in the context of business start-ups, we expect that

especially firms with difficult access to bank loans may have an incentive to manage their

earnings upwards. On the one hand, companies that have limited cash available on their balance

sheet have no alternative than to raise external financing to embark on investment projects.

Consistent with this idea, de Haan and Hinloopen (2003) show that current liquidity is

significantly negatively related to the probability of raising bank debt whereas profitability a

measure of future liquidity is not significant. On the other hand, business start-ups with few

tangible assets may find it difficult to obtain a bank loan, as they have insufficient assets that can

be pledged as collateral for this bank debt (see, for example, Degryse and Van Cayseele, 2000;

Lopez Iturriaga, 2005).

HYPOTHESIS 2: Business start-ups that are financially constrained will manage their

earnings upwards.

As a supplementary analysis to examining the incentives for earnings management in business

start-ups, we also wish to determine whether bank lending decisions are actually influencedby

higher reported earnings as a result of higher accruals numbers. Hence, we wish to establish

whether firms that report higher accruals, and thus higher earnings, are more likely to obtain a

bank loan, and thus may succeed in their attempts to secure bank financing through accruals

management. Finding support for such a relation would also mean that banks can be misledby

earnings management, at least in the start-up context, where prior banking relationships are

lacking and information asymmetries are particularly extensive. For other sources of external

financing, in particular equity, the literature has offered some interesting insights regarding the

question whether firms can influence investor perceptions, and obtain funds at a lower cost.

Friedlan (1994), Dechow et al. (1996), Teoh et al. (1998b) and Rangan (1998), for example,

conclude that firms indeedsucceed in manipulating their stock price, and increase the proceeds

7/29/2019 SSRN-id967386

10/41

8

from initial or seasoned equity offerings. In the long run, however, the market sees through

earnings management and stock prices are corrected downwards again.

The above discussion results in the following hypothesis:

HYPOTHESIS 3A: Business start-ups that increase their earnings through the use of accrual

accounts are more likely to obtain a bank loan.

Alternatively, banks may be aware that increased levels of accruals could reflect opportunistic

earnings management. Indeed, accruals consist of a normal (i.e. non-discretionary) and an

abnormal (or discretionary) part. While only the abnormal part reflects opportunistic earnings

management, just total accruals are observable in practice. Hence, banks do not know whether

high values of accruals are due to some real underlying economic event or caused by

opportunistic earnings management. So, they may consider high accruals as an additional risk

factor, on top of the firms failure risk, and be reluctant to lend to firms with relatively high

accruals. To test whether banks associate high levels of accruals with increased risk, and adjust

their credit-granting decisions, we posit the following hypothesis:

HYPOTHESIS 3B: Business start-ups are less likely to obtain a bank loan the higher the level

of their accruals accounts.

III. Design and models

In this paper, we examine upwardearnings management by business start-ups prior to receiving

afirstbank loan. We do this by testing whether current accruals are higher in the years before a

first bank loan than in other years, ceteris paribus. As in Han and Wang (1998), we do not use a

separate model to divide current accruals into a discretionary and a non-discretionary

7/29/2019 SSRN-id967386

11/41

9

component, but rather directly control for the factors that explain the non-discretionary or normal

part of current accruals and other factors that may confound our analysis.

Accruals include a non-discretionary or normal component that changes with the firms

level of operating activities, and a discretionary or abnormal component that is the result of

earnings management. Firms may, for example, overstate revenues and accounts receivable or

understate the write-down for obsolete inventories in an attempt to increase their earnings.

While earnings management researchers are particularly interested in this abnormal or

discretionary part of current accruals, only total current accruals are observable. Previous

studies therefore have typically used a two-step procedure, which involves first calculating the

discretionary or abnormal accruals from the accruals of firms in the same industry and year, and

next estimating a model to explain these discretionary or abnormal accruals. However, this

procedure is not feasible in a start-up context, where motives and opportunities to manage

earnings are largely different as compared to more established firms (see also Section I).

Furthermore, in the case of newly established ventures, it is often difficult to identify a

comparable firm in the corresponding industry and year. According to Schumpeter (1934), an

entrepreneur is an innovator who implements changes within markets, such as introducing new

products or products of better quality, uses new methods of production, opens a new market, etc.

Hence, an entrepreneurial firm often has no peer company. Finally, in any context (business

start-up or not), a two-step procedure introduces unnecessary measurement error in the estimates

of discretionary accruals. In particular, when the average comparable firm also manages its

earnings, the calculated discretionary accruals of other firms will not be correctly estimated.

This reduces the power of subsequent earnings management tests (see, for example, Dechow et

al., 1995).

Therefore, we investigate earnings management in business start-ups by testing whether

current accruals are higher in the years before a firstbank loan, after controlling forfactors that

may explain the non-discretionary or normal part of the accruals and factors that may confound

7/29/2019 SSRN-id967386

12/41

10

our analysis. To that end, we estimate the following model using a one-way random effects

panel data estimation technique:7, 8

CURXit = 0 + 1 BANKit + 2 CASHit + 3 PPEit + 4VAit + 5NOCFit

+ 6 LAGCURXit + 7 LNAGEit + 8 LNSIZEit + 9 NOCFit + it (1)

where:

CURX = CUR1 (trade accruals) or CUR2 (non-cash working capital accruals)

BANK = BANKDUM or BANKCAT

BANKDUM = Dummy variable that equals one in the year(s) before obtaining the

first bank loan and zero otherwise.

BANKCAT = Indicator variable that equals one in the year(s) preceding the first

bank loan, zero in the year of this bank loan and minus one in the year(s)thereafter.

CASH = Net cash and cash equivalents in year (t1) / lagged total assets

PPE = Property, plant and equipment in year (t1) / lagged total assets

VA = Change in value added from (t1) to t / lagged total assets

NOCF = Change in net operating cash flow from (t1) to t / lagged total assetsLAGCURX = Lagged current accruals, using definition 1 (trade accruals) or definition

2 (non-cash working capital accruals)

LNAGE = Natural log (1 + years since start-up)

LNSIZE = Natural log (total assets) in year (t1)

NOCF = Net operating cash flow in year (t1) / lagged total assets

The dependent variable in this model is current accruals. Current accruals relate to the

short-term assets and liabilities that support a firms day-to-day operations. We prefer to

examine current accruals over total accruals (i.e. the sum of current and non-current accruals,

such as, for example, depreciation)9

because the management of current accrual accounts is less

7 OLS estimation may produce biased and inconsistent results owing to its failure to control for time-invariant firm-specific heterogeneity. This problem will occur when the disturbance term incorporates time-invariant omitted

factors that are contemporaneously correlated with the models explanatory variables. Hence, we estimated the

models by means of a random effects panel data estimation technique. Alternatively, we estimated the model using

OLS, and obtained qualitatively similar results. These results are not reported, but can be obtained from the authors

upon request.8

The Hausman statistic does not reject the random effects specification in favor of a fixed effects model. In

addition, conclusions are not affected under a two-way random effects estimation technique.9

Total accruals can be separated along two dimensions. First, total accruals are the sum of current and non-current

accruals. Second, total accruals can also be seen as the sum of normal or non-discretionary accruals and abnormal

or discretionary accruals. As total accruals, its current and non-current parts can be further partitioned in a normal

(non-discretionary) and abnormal (discretionary) part. Hence, total accruals can then also be seen as the sum of four

7/29/2019 SSRN-id967386

13/41

11

visible than the management of non-current accrual accounts (see also Guenther, 1994; Teoh et

al., 1998a). The level of current accrual accounts such as accounts receivable and inventories

is to a certain extent influenced by managements subjective and discretionary estimates. For

instance, the provisions for bad debt and for obsolete inventories are to a large extent based on

judgment. If done opportunistically, this is difficult to detect and therefore often goes unnoticed.

In contrast, management of non-current items, such as a change in the depreciation methods of

fixed assets, is more difficult to hide from stakeholders. The reason is that Belgian GAAP

requires that any such changes be duly reported and motivated in the notes to the financial

statements, also for the small firms in our sample that may file abbreviated accounts. This

makes the latter kind of earnings management more prone to observation by stakeholders. An

inspection of the notes did not reveal that the start-up companies in our sample changed their

depreciation method in the years after start-up.

We use two distinct measures for current accruals trade accruals and non-cash working

capital accruals to check whether our results are not idiosyncratic to the choice of current

accruals measure. Trade accruals are calculated from trade-related accounts, including

inventories, accounts receivable and accounts payable accounts. Non-cash working capital

accruals also include other short-term liabilities, such as taxes and wages payable and social

security payments.

Trade accruals (CUR1) =

[ inventories + accounts receivable + accrued assets accounts payable

accrued liabilities] / lagged total assets (2)

Non-cash working capital accruals (CUR2) =[( current assets cash and cash equivalents) ( current liabilities

short-term financial debt)] / lagged total assets (3)

separate parts: normal current accruals, abnormal current accruals, normal non-current accruals, and abnormal non-

current accruals.

7/29/2019 SSRN-id967386

14/41

12

The test variable for the first hypothesis (BANK) can take either of two forms, to let the

data decide on the exact specification of earnings management in business start-ups. First,

BANKDUM is a dummy variable that equals one in the year(s) before obtaining afirstbank loan

and zero otherwise. Second, BANKCAT is an indicator variable that equals one in the year(s)

preceding the first bank loan, zero in the year of this bank loan, and minus one in the year(s)

thereafter. This paper focuses on bank loans instead of bank credit lines; the latter is very short-

term debt in nature and involves smaller amounts of financing, which makes it less subject to

information and incentive problems. BANKDUM then captures whether earnings management

is higher during the years preceding the first bank loan, compared to all other years.

BANKCAT, however, additionally captures whether earnings management is lower or decreases

after the bank loan is granted. Since we hypothesize firms to manage their earnings upwards in

the years prior to their first bank loan, we expect to find a positive coefficient on both test

variables and interpret such coefficients as evidence of upward earnings management before

obtaining a first bank loan.

We further include our test variables for the second hypothesis, i.e. variables that may

capture earnings management incentives in a business start-up context. Firms with less available

cash tend to be more financially constrained and thus should be more likely to manage their

earnings upwards to obtain bank debt. We include CASH, measured as the ratio of net cash and

cash equivalents in the previous year to lagged total assets, to control for this effect and expect a

negative coefficient on this variable. Firms with no or only few tangible assets that can be

pledged as collateral have less chance to obtain a bank loan, ceteris paribus. Hence, these firms

may be more likely to resort to earnings management to influence bank perceptions of their

creditworthiness (borrower quality). We proxy the collateral value of assets by the level of

prior-year property, plant and equipment (PPE) relative to lagged total assets and expect a

negative coefficient on this variable.

7/29/2019 SSRN-id967386

15/41

13

The first set of control variables that we include are factors associated with the non-

discretionary or normal part of current accruals. First, we include VA the change in value

added as compared to the previous year to control for the change in current accruals that is

attributable to changes in the firms operating activities, i.e. firm growth.10

Note that the

expected sign on the coefficient for VA is not a priori clear, as the change in value added is

associated with both asset (e.g., inventories and accounts receivable) and liability components

(e.g., accounts payable) of current accruals. Hence, the net effect of changes in these accounts

cannot be predicted (see also Jones (1991), p. 213). As in Dechow (1994), we also include the

change in net operating cash flow (NOCF)11 as a second control variable. The rationale is that

changes in cash flows contain temporary components that are reversed over time, and the role of

accruals is to match cash disbursements and cash receipts that are associated with the same

economic event in order to obtain a performance measure that better captures current firm

performance (Dechow, 1994).12

Since some accruals also reverse over time, prior-year (i.e.

lagged) accruals may contain information with respect to current-year accruals. We therefore

also include lagged current accruals (LAGCURX) in the model and expect a negative sign on this

variable. All these control variables are scaled by lagged total assets to take potential

heteroscedasticity problems into account (see also Jones, 1991).

Finally, we include a number of control variables that may confound our analysis of

earnings management in a business start-up context. First, firm age (LNAGE), calculated as the

natural logarithm of (1 + years since start-up), is used to control for the level of information

10 Prior studies typically proxy changes in a firms level of operating activities by means of changes in its sales (see,

e.g., Jones, 1991; Dechow et al., 1995). Unfortunately, revenue figures are not available for all start-up firms in

Belgium. The reason is that enterprises classifying as a small firm are allowed to file abbreviated financialstatements, and the latter do not generally include revenues. Yet, these firms have to report value added, which is

calculated as sales minus the cost of goods sold. Changes in value added and changes in revenues are likely to

capture the same information when the trend in the cost of goods sold follows the trend in revenues.11

Net operating cash flow is defined as earnings before interest, taxes, depreciation and amortization (EBITDA)

minus the change in non-cash working capital.12

For example, assume that in period t firms sell on cash and on credit. Both types of sales determine the firms

operating performance in period t, i.e. the time of sale. However, cash flows from operations of period t only

capture the effect of the cash sales, and not of the credit sales. The impact of the credit sales in period t will be

reflected in the cash flow from operations in a later period, i.e. in the period when the customer actually pays. By

contrast, earnings of period t capture the effect of both cash and credit sales, and is therefore said to better capture

current firm performance.

7/29/2019 SSRN-id967386

16/41

14

asymmetries between the start-up firm and its external financiers. Since the level of information

asymmetries and hence the likelihood that earnings management goes undetected decreases as

firms grow older, we expect to find a negative coefficient on LNAGE (e.g., Richardson, 2000).

Consistent with the literature (e.g., Becker et al., 1998, Young 1998), we introduce the natural

log of lagged total assets (LNSIZE) to control for the potential effect of firm size on accounting

choices. Earlier research has conjectured and found a negative coefficient on this variable, based

on the argument that larger firms are more visible. This point has been made in the context of

publicly quoted companies, where larger firms invite more analyst coverage. In the case of

privately held business start-ups, however, firm size is certainly not a proxy for investor interest.

So, we include firm size, but are unsure about its sign. We measure this variable by lagged total

assets (instead of this-year total assets) since the latter are influenced by earnings management in

the current year. Finally, prior studies have shown that tests of earnings management may be

mis-specified for firms with extreme financial performance (e.g., Dechow et al., 1995). We

include net cash flow from operations (NOCF) to control for this effect. Given the results

reported by Dechow et al. (1995), we expect a negative coefficient on this variable. The model

and the predicted direction of the effects are summarized in Table 1.

**************

insert Table 1

**************

IV. Data and results

IV.A. Sample description

To test the theoretical predictions in the previous section, we need financial information from

business start-ups as of start-up. Little research has been done on newly established

entrepreneurial firms, simply because the data are not readily available. For the USA, the

Federal Reserve Boards National Survey of Small Business Finances (NSSBF) provides

financial information on 4,637 privately held firms, but Ang et al. (2000) report that mean firm

7/29/2019 SSRN-id967386

17/41

15

age in this database is 17.6 years. As a result, NSSBF is not representative for start-up firms.

Furthermore, the database does not include panel data (Petersen and Rajan, 1997).

We decided to use Belgian start-up data to test our model for the following reasons.

Belgium is an interesting environment for research on business start-ups, because all limited

liability firms, i.e. corporations except for financial institutions, insurance companies,

exchange brokers and hospitals are mandated to file annual accounts with the National Bank as

of the moment of start-up. In 2002, nearly 270,000 corporations filed financial statements with

the Belgian National Bank, covering more than 75% of GNP. Overall, the accounting principles

in Belgium (Belgian GAAP) are to a large extent comparable to those adopted in the Anglo-

Saxon world (see, for instance, Deloof and Jegers, 1999). The first time a Belgian firm registers

with the tax authorities, it receives a unique and chronologically accorded Value Added Tax

number. This VAT number allowed us to identify newly established firms and their financial

statements as of start-up in the database of the National Bank.13

So, the first year of data in our

database truly represents the firms start-up year. And another interesting feature is that Belgian

business start-ups are required to publish an abstract from their foundation charter in the

Government Newspaper (Staatsblad) shortly after start-up. This abstract contains information

on the firms ownership at start-up.

We identified 652 limited liability firms (corporations) that were founded in 1992 in

manufacturing. This industry was selected because of the larger scale of its operations, at least

when compared to retailers, wholesalers or service firms. Entrepreneurs in manufacturing

therefore are more likely to lack the personal financial resources to fully finance the firms assets

and operations during the first few years after start-up. To be included in the sample, companies

had to report their industry code, i.e. the European NACE code, at the four-digit level. All firms

in the sample report only one four-digit NACE code and hence are narrowly focused.

13This database is commercialized by Bureau Van Dijk Electronic Publishing.

7/29/2019 SSRN-id967386

18/41

16

To make sure that only firms that are true first-time business start-ups were included in

the sample, we subsequently removed all firms that were not entrepreneurial start-ups. From the

contents in the foundation charter as published in the Government Newspaper, true business

start-ups could be distinguished from newly established subsidiaries of existing firms, split-ups,

spin-offs, etc. Through follow-up phone calls, we also identified firms that were established

through the incorporation of a previously self-employed activity, and removed them from the

sample. These screening criteria reduced the sample to 328 true business start-ups, which are

examined during 19922002. 48 firms in this sample never borrowed any funds from banks and

hence are not the main focus of our study. The reason is that we cannot determine the moment

of theirfirstbank loan. We observe a lot of bankruptcies and voluntary liquidations in this small

sample (12 companies were liquidated voluntarily and 11 were liquidated following a

bankruptcy procedure). Of the remaining firms, 79 raised a first bank loan afterhaving filed

their annual accounts with the National Bank. On average, these firms raise their first bank loan

after 4.29 years (median of three years). Finally, 201 corporations already raised bank debt in

the start-up year. While for the latter 201 firms we can determine the moment of theirfirstbank

loan, it is also the case that bank lending decisions were not based upon publicly available annual

accounts.14

Hence, we will separate our analyses by first estimating the models on the

subsample of 79 event firms and then testing the robustness of our results when also including

the data on the 201 start-ups who got bank financing in the start-up year. Table 2 describes the

industry distribution of the 79 event firms, based on their two-digit NACE code. Industries that

are highly represented include the paper, printing and publishing industry (21 firms); the food,

drink and tobacco industry (10 firms); and the footwear and clothing industry (8 firms). Besides,

we also report the industry distribution for the 201 firms that already raised bank debt in the

start-up year.

14We recognize that in their lending decisions, banks may still use information on earnings and cash flows from the

financial plan, which has to be submitted as part of the business plan before a corporation can be founded (see, for

example, Vanhoutte and Sels, 2005). Yet, the earnings and cash flow information in this financial plan typically is

less extensive than the one in a firms (even abbreviated) annual accounts.

7/29/2019 SSRN-id967386

19/41

17

**************

insert Table 2

**************

We collected financial statement data from the first accounting year up to and including

2002, resulting in 691 and 1670 firm-year observations for the 79 event firms and 201 other

start-up firms, respectively. By the end of 2002, 15 event firms (18.99%) had discontinued their

operations: 10 due to bankruptcy and 5 firms were liquidated voluntarily.15

As a result, the panel

data set is unbalanced.

Table 3 reports summary statistics for the 79 event firms in the start-up year and

compares this information with the sample of 201 enterprises that already raised bank debt in the

start-up year, using a non-parametric Wilcoxon test. The average event firm employs two

persons in the start-up year and its total assets amount to 299,754 (median of 59,755). Firm

start-up size is significantly smaller when compared to the sample of enterprises that already

raised bank debt in the start-up year. The 79 event firms also have a significantly higher ratio of

cash and marketable securities to total assets (average of 13.36% and median of 10.13%),

suggesting a lower need for bank financing in the start-up year, ceteris paribus. Overall, their

ratios of inventories and accounts receivable relative to total assets are not significantly different

from those of firms that obtained bank debt already in the start-up year. Yet, the event firms

have a significantly smaller ratio of tangible fixed assets to total assets (average of 25.39% and

median of 15.44%), which suggests a link between the availability of assets that can be pledged

as collateral and access to bank loans. Not surprisingly, the event firms indeed have a

significantly smaller debt ratio, but do not differ in terms of trade credit used and profitability, as

measured by net operating cash flow to total assets and net income to total assets.

**************

insert Table 3

**************

15 These percentages are comparable to those in the subsample of 201 firms that already raised bank debt in the start-

up year: 39 firms (19.40%) discontinued their operations by the end of 2002, of which 27 because of bankruptcy.

7/29/2019 SSRN-id967386

20/41

18

IV.B. Earnings management results

In Table 4, we report descriptive statistics on the dependent (Panel A) and explanatory (Panel B)

variables in our model. These numbers are calculated from the sample of firms that everraised a

bank loan. Given the presence of extreme values in the data, we winsorized the variables at the

595%. In other words, these percentiles replace the corresponding extreme values. From Panel

A, it is clear that trade accruals on average represent 3.72% of lagged total assets (median of

1.50%) whereas non-cash working capital accruals average 3.50% (median of 2.00%). The

correlation between these two measures of current accruals amounts to 0.7042. Panel B presents

descriptive statistics for the test and control variables in our model, which are also winsorized.

A Pearson correlation matrix for the explanatory variables is provided in Appendix A. As the

maximum correlation coefficient amounts to 0.3087 (NOCF CASH), multicollinearity is

unlikely to be a problem in our study.

**************

insert Table 4

**************

In Tables 5 and 6, we report the results of our regression analyses. Table 5 relates to the

subsample of 79 business start-ups that raised no bank loan in the start-up year, but did so before

the age of ten (691 firm-year observations). Besides including the 79 event firms, Table 6 also

incorporates the data on the 201 start-ups that actually obtained a first bank loan in the start-up

year (i.e. a total of 280 firms and 2361 firm-year observations).16

In each Table, Panel A

contains the results when trade accruals (CUR1) is the dependent variable whereas non-cash

working capital accruals (CUR2) is the dependent variable in Panel B. Both panels include the

results of three models. The first model is the base model, i.e. the model excluding the test

16As a robustness check, we also did the analyses in Tables 5 and 6 when including the 48 firms that neverobtained

a bank loan (for these firms, BANKDUM and BANKCAT always equal one). This robustness check assumes that

these firms wanted to borrow from banks but were refused. We find that our results are robust under this alternative

specification although the models explanatory power is always somewhat lower. These findings are not reported,

but can be obtained from the authors upon request.

7/29/2019 SSRN-id967386

21/41

19

variables BANK, CASH and PPE. The two other models only differ regarding the specification

of the test variable for the first hypothesis: Model 2 uses BANKDUM whereas Model 3 uses

BANKCAT. From Tables 5 and 6, it is clear that all models are significant and have substantial

explanatory power. Although explanatory power cannot be easily compared from one study to

another, our results show that, in comparison to earlier earnings management studies, we can

explain a relatively large portion of current accruals for first-time business start-ups.

Furthermore, in both tables the trade accruals models (Panel A) have higher explanatory power

than the non-cash working capital accruals models (Panel B).

Table 5 shows that the test variables BANKDUM and BANKCAT have a positive sign,

as predicted, and are significant at the 10% level in all model specifications. In Table 6, the

coefficient estimates on the test variables for the first hypothesis are also positive and significant,

mostly at the 1% level. These results strongly support our hypothesis that business start-ups

manage their earnings upwards in the years before obtaining a first bank loan, as the proportion

of current accruals is significantly larger in years before than after a first bank loan, controlling

for factors that may explain the normal part of current accruals and other factors that may

confound our analysis of earnings management in a start-up context. Yet, we find no

corroborative evidence that earnings are managed downwards after obtaining a first bank loan as

the explanatory power of the current accruals models is only slightly higher when using the

BANKCAT specification of the test variable. Possibly, the reversal in accrual accounts is

already captured by including the lagged current accruals variable, which is highly significant in

all models.

Start-up firms with more cash on their balance sheet have both significantly lower trade

accruals and significantly lower non-cash working capital accruals. In addition, companies that

have more tangible fixed assets have significantly lower current accruals. These results are

consistent with the idea in hypothesis 2 that more financially constrained business start-ups are

inclined to manage their earnings upwards, ceteris paribus. Indeed, firms that lack cash may

7/29/2019 SSRN-id967386

22/41

20

highly need the bank financing but when the amount of assets that can be pledged as collateral is

limited, they may expect that banks are less willing to lend. Hence, as indicated by our findings,

these firms resort to earnings management in order to influence bank lending behavior. When

we include an interaction term between BANKDUM on the one hand and CASH and PPE,

respectively, on the other, we find that these interaction variables have a negative but not

significant sign (p-values around 0.20). These interaction terms do become significant when

removing CASH and PPE from the model, but this alternative specification has lower

explanatory power (not reported). In addition, BANKDUM retains its statistically significant

positive parameter estimate. Overall, these results suggest that financially constrained business

start-ups manage their earnings upwards, and that receiving a first bank loan does not fully

eliminate financial constraints.

Tables 5 and 6 further show that the variables explaining the non-discretionary or normal

current accruals, namely the change in value added, the change in net operating cash flow and

lagged current accruals, are highly significant, with signs in the expected directions. This is also

the case for most of the other control variables. The only exception is LNAGE, whose positive

sign is mostly not significantly different from zero.17

A positive sign on LNAGE, which

becomes significant only in the non-cash working capital accruals model (Panel B of Tables 5

and 6) could indicate that business start-ups are expanding their personnel at a smaller rate than

their cash flows as they grow older, such that non-trade current liabilities (wages and social

security payments) increase more slowly. Alternatively, as LNAGE is significant only when

including BANKCAT, it might be that a decrease in non-cash working capital accruals after the

loan is granted happens in a non-linear way. In short, the results on the control variables suggest

that we adequately control for the factors known to determine the non-discretionary (or normal)

part of current accruals and the factors that may confound our earnings management tests in a

17 When LNAGE is removed from the models in Tables 5 and 6, we find that the results and conclusions on the

other variables remain valid.

7/29/2019 SSRN-id967386

23/41

21

start-up context, thereby supporting the idea that the results on the test variables for our first and

second hypotheses indeed stem from discretionary accruals management.

*****************

insert Tables 56

*****************

IV.C. Additional analysis: resolution of the bank lending decision

As the results in the previous section show that business start-ups manage their earnings upwards

in the years prior to receiving afirstbank loan, an interesting follow-up question is whether this

earnings management affects the lending decisions of banks. That is, do banks actually take the

information from current accruals into account when deciding on granting credit to business

start-ups and, if so, do they value higher earnings that are the result of increased current accruals

or, alternatively, do they consider higher accruals as an additional risk factor? In this section, we

expand our analysis by investigating the factors that are related to the bank lending decision in

the context of a first loan to business start-ups. In particular, we examine whether higher current

accruals in the year(s) before obtaining a bank loan impacts bank lending decisions.

In Table 7, we report the results of a multivariate logistic regression analysis using the

sample of 127 firms that did not obtain a bank loan in the start-up year; indeed, only for those

firms, current accruals may have affected the decision of banks to provide a first loan. We

follow these firms from start-up until and including the year of their first bank loan.

Alternatively, when these firms never raised a bank loan, we follow them to 2002 or to the year

of their liquidation. This results in 623 firm-year observations. The dependent variable

LENDING equals one for the year in which a firstbank loan was obtained and zero otherwise.

The test variable in this analysis is a measure of current accruals: trade accruals in Panel A, and

non-cash working capital accruals in Panel B. Model 1 reports the results when using prior-year

current accruals as the test variable. We introduce alternative specifications for the test variable

in subsequent models. Model 2 calculates the test variable as a two-year average of current

7/29/2019 SSRN-id967386

24/41

22

accruals prior to obtaining the bank loan.18

Finally, the test variable in Model 3 is measured as

lagged current accruals minus the firm-level time series average of this variable over the

sampling period.

A number of control variables are included in the model to capture elements that are

related to the demand and supply for bank financing (see, for example, de Haan and Hinloopen,

2003; Beyan and Danbolt, 2004; Huyghebaert and Van de Gucht, 2007). Growth opportunities

are measured by the average growth rate in total assets of business start-ups in the corresponding

four-digit NACE industry during the studied window. As business start-ups have only limited

access to external financial resources, we expect a positive coefficient on this variable, reflecting

a higher demand for bank financing. Simultaneously, banks may be more willing to lend to

start-ups in high-growth industries, to develop a valuable lending relationship. To take into

account that firms with access to internal financing have a smaller demand for external (bank)

financing, we include net operating cash flow/total assets and net cash and cash equivalents/total

assets and expect a negative coefficient on these variables. These control variables are

calculated from the previous-year financial statements. In addition, we control for the fact that

firms with more tangible assets (PPE/total assets) likely find it easier to borrow from banks, as

they can pledge these assets as collateral. By contrast, banks may be reluctant to lend to high-

risk firms (measured by the failure rate of earlier business start-ups in the corresponding four-

digit NACE industry during 19881991). For a given level of activity risk, the start-ups capital

structure may further enlarge its bankruptcy risk. Hence, firms with a higher debt ratio (current

liabilities/total assets) may find it difficult to borrow from banks. Finally, we control for firm

age (natural log (1 + years since start-up)) and firm size (natural log (total assets)). When older

and larger firms face less information asymmetries, they may have better access to bank debt.

From inspection of Table 7, it is clear that none of the test variables are significantly

associated with the bank lending decision. Higher current accruals do not incite banks to lend

18 When two-year historical data are not available, we use one year of historical data to calculate the test variable in

order to preserve the sample size.

7/29/2019 SSRN-id967386

25/41

23

more eagerly nor do they deprive start-up firms access to bank financing. So, this evidence

suggests that any start-up earnings management prior to obtaining a bank loan does not affect the

banks lending decision. Overall, these results could reflect that 1) banks carefully examine

company accounts before lending, but are not being misled by any start-up earnings management

when deciding on granting a loan, nor do banks penalize companies for the information risk

resulting from potential earnings management or 2) banks do not attentively scrutinize company

accounts in the case of business start-ups. Ravid and Spiegel (1997), for example, argue that the

relatively small size of start-up loans and the complexity associated with screening and

monitoring of these firms renders such activities cost ineffective from the point of view of banks

(see also Huyghebaert and Van de Gucht, 2007).

Rather, we do find that economic variables drive the bank lending decision in the context

of business start-ups. More specifically, start-up firms that face significant financing needs

resulting from growth opportunities are more likely to raise and obtain bank debt, ceteris paribus.

This relation suggests that banks are interested in developing a long-term relationship with start-

ups in high-growth industries, possibly with the intention of reaping future hold-up or location

rents (see, for instance, Degryse and Ongena, 2002, 2005; Huyghebaert and Van de Gucht,

2007). Internal cash generation (current profitability) also increases the likelihood of bank

lending but, consistent with the pecking order model of capital structure, firms with accumulated

cash reserves resulting from pastprofitability are less likely to borrow from banks. The latter

results are consistent with the findings of de Haan and Hinloopen (2003), for example. Yet, the

results in column 1 indicate that banks are less likely to lend to start-up firms with more tangible

fixed assets, which is inconsistent with our priors. Huyghebaert and Van de Gucht (2007) also

document a negative relation between asset tangibility and the fraction of debt that consists of

bank loans for newly established ventures. They argue that banks are more likely to liquidate

firms with a high liquidation value following a default on their loans (see also Hart, 1995;

Manove et al., 2001). This issue of default and liquidation is relatively important in a start-up

7/29/2019 SSRN-id967386

26/41

24

context, especially for firms that face high default risk. Hence, risky firms with highly tangible

assets may abstain from bank financing in order to reduce the likelihood of premature liquidation

following default (see Huyghebaert et al., 2007). To take this argument into account, we re-

estimated Model 1 after including an interaction term between PPE/total assets and failure risk

(Model 1a) and find support for this conjecture. More specifically, the simple term measuring

tangible fixed assets has a positive and significant parameter estimate whereas that of the

interaction term is significantly negative. This allows us to conclude that banks tend to lend

more eagerly to firms with larger tangible fixed assets, but these firms are reluctant to raise bank

debt when their ex-ante likelihood of failure is higher, ceteris paribus.

The results further indicate that banks are less likely to ever lend to firms in industries

with high start-up failure risk. Besides activity risk (industry failure risk), financial risk also

significantly negatively affects the bank lending decision. Finally, we find that older firms are

less likely to raise a bank loan whereas firm size has a significantly positive impact on the

likelihood of borrowing from banks, ceteris paribus.

**************

insert Table 7

**************

V. Conclusions

This paper investigates earnings management by business start-ups around the time of theirfirst

bank loan. We argue that newly established firms have strong incentives to manage their

earnings numbers upwards to influence the lending decisions of banks. The reason is that their

survival often depends upon obtaining the necessary financial resources to finance assets and

operations. As these firms have only a limited track record and as they have not yet developed a

relationship with a house bank in the case of a first bank loan application, information

asymmetries between entrepreneurs and banks tend to be large. This makes it also difficult for

banks to distinguish between higher accruals resulting from some real underlying economic

7/29/2019 SSRN-id967386

27/41

25

event and higher accruals resulting from earnings management. This holds especially true in the

case ofcurrentaccrual accounts, which relate to the firms day-to-day operations.

Using a unique sample of 328 newly established ventures in the manufacturing industry

for a period up to ten years after start-up, we test whether current accruals are larger in the years

before these firms obtain their first bank loan than in subsequent years. Consistent with this

hypothesis, we find that current accruals of business start-ups are indeed significantly higher in

the years prior to obtaining their first bank loan, ceteris paribus. This result is robust to two

alternative specifications of the current accruals measure: trade accruals and non-cash working

capital accruals. In addition, we find that financially constrained business start-ups, i.e. firms

that are short of cash and with limited tangible assets, have significantly higher current accruals,

ceteris paribus. However, we find no corroborative evidence that bank lending decisions are

influenced by higher earnings resulting from higher current accruals. Rather, we document that

economic variables shape the banks credit-granting decision in the context of business start-ups.

These results could reflect that 1) banks carefully examine company accounts, but are not being

misled by start-up earnings management, nor do they penalize the information risk resulting

from potential earnings management behavior or 2) banks do not attentively scrutinize company

accounts before lending in the case of business start-ups. Our study, however, is unable to

discriminate between these two alternative explanations.

Overall, the results of our study may be of interest to start-ups firms, banks and

regulators. First, to start-ups the results indicate that earnings management is not successful to

influence the lending decisions of banks. Rather, the firms growth opportunities, its

profitability, the tangibility of its assets and the risk of its activities and financial structure are

important considerations in the bank lending decision Second, to banks they suggest that start-

up firms are managing their earnings figures when applying for a first bank loan. And third, to

regulators our findings suggest that start-up firms use the flexibility available in GAAP. The

results further indicate that the market for bank debt is not negatively affected by this potentially

7/29/2019 SSRN-id967386

28/41

26

opportunistic behavior. Finally, start-up firms that use earnings management are not being

deprived from access to bank loans, as they are not penalized for the information risk resulting

from potential opportunistic earnings management.

7/29/2019 SSRN-id967386

29/41

27

Table 1: Summary of testable predictions

Predicted Sign

Test variables:

BANKDUM (H1) +

BANKCAT (H1) +

CASH (H2)

PPE (H2)

Control variables:

Change in value added from (t1) to t / lagged total assets (VA) +/

Change in net operating cash flow from (t1) to t / lagged total assets (NOCF)

Lagged current accruals (LAGCUR)

Natural log (1 + years since start-up) (LNAGE)

Natural log (total assets) in year (t1) (LNSIZE) +/

Net operating cash flow in year (t1) / lagged total assets (NOCF)

BANKDUM = Dummy variable that equals one in the year(s) before obtaining the first bank loan and

zero otherwise.BANKCAT = Indicator variable that equals one in the year(s) preceding the first bank loan, zero in the

year of this bank loan and minus one in the year(s) thereafter.

CASH = Net cash and cash equivalents in year (t1) / lagged total assetsPPE = Property, plant and equipment in year (t1) / lagged total assets

7/29/2019 SSRN-id967386

30/41

Table 2: Industry distribution of start-ups

This table displays the industry distribution of the start-up firms, based on their two-digit NACE industry code. All sample firms a

operations in the manufacturing industry in 1992. The sample is constructed from the Belgian National Bank database. Based on the fo

only true entrepreneurial start-ups are retained. We make a distinction between the firms that raised a first bank loan after(79 firms) and

the firms that raised their first bank loan after the start-up year as the event firms.

NACE code Description

22 Production and preliminary processing of metals

23 Extraction of minerals other than metalliferous and energy-producing minerals; peat extraction

24 Manufacture of non-metallic mineral products

25 Chemical industry

31 Manufacture of metal articles (except for mechanical, electrical and instrument engineering and vehicles)

32 Mechanical engineering

34 Electrical engineering

36 Manufacture of other means of transport

37 Instrument engineering

41/42 Food, drink and tobacco industry

43 Textile industry

44 Leather and leather goods industry (except footwear and clothing)

45 Footwear and clothing industry

46 Timber and wooden furniture industries

47 Manufacture of paper and paper products; printing and publishing 48 Processing of rubber and plastics

49 Other manufacturing industries

TOTAL

7/29/2019 SSRN-id967386

31/41

Table 3: Characteristics of the start-up firms

This table provides descriptive statistics for the sample of 79 event firms that raised bank debt after the first start-up year and 201 firms

year. All firms are incorporated in Belgium and start their operations in the manufacturing industry in 1992. The sample is construct

Based on the foundation charter and follow-up phone calls, only true entrepreneurial start-ups were retained. The descriptive characteristi

first accounting year. We also report thep-value of a non-parametric Wilcoxon test that compares the two subsamples.

N=79 N=2

mean median std. dev mean med

FIRM SIZE

Number of employees 2 1 3.5984 3.1841

Total assets () 229,754 59,755 968,889 287,085 1

ASSET STRUCTURE

Cash and cash equivalents/total assets 0.1336 0.1013 0.2285 0.0391

Inventories/total assets 0.1068 0.0359 0.1460 0.1014

Accounts receivable/total assets 0.2366 0.1739 0.2126 0.2159

Property, plant and equipment/total assets 0.2539 0.1544 0.2387 0.4729 FINANCIAL STRUCTURE

Total debt and current liabilities/total assets 0.5691 0.5601 0.3809 0.8327

Bank debt/total assets 0 0 0 0.3150

Trade credit/total assets 0.2466 0.1804 0.2377 0.2097

PROFITABILITY

Net operating cash flow/total assets 0.0670 0.0826 0.5105 0.1255

Net income/total assets -0.0803 0.0062 0.4617 -0.0509 -

7/29/2019 SSRN-id967386

32/41

30

Table 4: Descriptive statistics on dependent and explanatory variables

Panel A: Dependent variables

mean median min max std. dev

Trade accruals (CUR1) 0.0372 0.0150 -0.3059 0.5204 0.1880

Non-cash WC accruals (CUR2) 0.0350 0.0200 -0.4215 0.5898 0.2200

CUR1 = Trade accruals = [ inventories + accounts receivable + accrued assets accounts payable

accrued liabilities ] / lagged total assets.

CUR2 = Non-cash working capital accruals = [( current assets cash and cash equivalents) ( current

liabilities ST financial debt)] / lagged total assets.

Panel B: Explanatory variables

mean median min max std. dev

CASH 0.0549 0.0319 -0.2840 0.5430 0.2211

PPE 0.3629 0.3187 0 0.9667 0.2842

VA 0.0342 0.0183 -0.4052 0.5657 0.2218

NOCF -0.3797 -0.3402 -6.7827 5.2982 2.4687

LAGCUR1 0.0263 0.0144 -0.2836 0.4000 0.1626

LAGCUR2 0.0170 0.0186 -0.3649 0.3811 0.1790

LNAGE 1.7556 1.7918 0 2.3979 0.5377

LNSIZE 8.8623 8.8345 6.7788 13.5944 1.3424

NOCF 0.1497 0.1542 -0.3821 0.6096 0.2358

CASH = Net cash and cash equivalents in year (t1) / lagged total assetsPPE = Property, plant and equipment in year (t1) / lagged total assets

VA = Change in value added from (t1) to t / lagged total assets

NOCF = Change in net operating cash flow from (t1) to t / lagged total assetsLAGCURX = Lagged current accruals

LNAGE = Natural log (1 + years since start-up)LNSIZE = Natural log (total assets) in year (t1)

NOCF = Net operating cash flow in year (t1) / lagged total assets

7/29/2019 SSRN-id967386

33/41

31

Table 5: The determinants of current accruals

This table presents one-way random effects regression results explaining trade accruals (Panel A) and non-cash working

capital accruals (Panel B). The first model is the base model. The test variable BANKDUM in the second model equals

one in the year(s) before obtaining the first bank loan and zero otherwise. The test variable BANKCAT in the third model

equals one in the year(s) preceding the first bank loan, zero in the year of this bank loan and minus one in the year(s)

afterwards. The sample includes data on 79 event firms that raised bank debt afterthe first start-up year (691 firm-year

observations).

Panel A: Trade accruals Pred. sign One-way random effects regression results

(1) (2) (3)

Intercept ? 0.0672 0.0388 0.0442

Test variables:

BANK BANKDUM (H1)

BANKCAT (H1)

+

+

0.0238*

0.0194**

CASH (H2) 0.1117*** 0.1137***

PPE (H2) 0.0460** 0.0463**

Control variables:

Change in value added / total assets +/ 0.1241*** 0.1438*** 0.1433***

Change in net operating CF / total assets 0.0087*** 0.0093*** 0.0092***Lagged current accruals (LAGCUR1) 0.0237** 0.0294** 0.0309***

Natural log (1 + years since start-up) 0.0009 0.0135 0.0217

Natural log (total assets) +/ 0.0116 0.0070 0.0074

Net operating CF / total assets 0.0749*** 0.0662*** 0.0644***

Buse R-square 32.20% 35.38% 35.67%

Panel B: Non-cash WC accruals Pred. sign One-way random effects regression results

(1) (2) (3)

Intercept ? 0.0521 0.0767 0.0599

Test variables:

BANK BANKDUM (H1)BANKCAT (H1)

++

0.0936***0.0525***

CASH (H2) 0.1305*** 0.1401***

PPE (H2) 0.0960*** 0.1038***

Control variables:

Change in value added / total assets +/ 0.0706*** 0.0966*** 0.0976***

Change in net operating CF / total assets 0.0082*** 0.0087** 0.0087**

Lagged current accruals (LAGCUR2) 0.0198*** 0.0263*** 0.0273***

Natural log (1 + years since start-up) 0.0119 0.0464 0.0606*

Natural log (total assets) +/ 0.0068 0.0040 0.0043

Net operating CF / total assets 0.0785*** 0.0602*** 0.0592***Buse R-square 21.81% 28.68% 29.22%

* indicates significance at the 10% level, ** at the 5% level and *** at the 1% level.

7/29/2019 SSRN-id967386

34/41

32

Table 6: The determinants of current accruals

This table presents one-way random effects regression results explaining trade accruals (Panel A) and non-cash working

capital accruals (Panel B). The first model is the base model. The test variable BANKDUM in the second model equals

one in the year(s) before obtaining the first bank loan and zero otherwise. The test variable BANKCAT in the third model

equals one in the year(s) preceding the first bank loan, zero in the year of this bank loan and minus one in the year(s)

afterwards. The sample includes data on 280 firms that raised bank debt in or afterthe first start-up year (2361 firm-year

observations).

Panel A: Trade accruals Pred. sign One-way random effects regression results

(1) (2) (3)

Intercept ? 0.0232 0.0056 0.0085

Test variables:

BANK BANKDUM (H1)

BANKCAT (H1)

+

+

0.0343**

0.0208***

CASH (H2) 0.0894*** 0.0908***

PPE (H2) 0.0383*** 0.0382***

Control variables:

Change in value added / total assets +/ 0.1326*** 0.1460*** 0.1453***

Change in net operating CF / total assets 0.0102*** 0.0100*** 0.0099***Lagged current accruals (LAGCUR1) 0.0295*** 0.0369*** 0.0380***

Natural log (1 + years since start-up) 0.0006 0.0074 0.0089

Natural log (total assets) +/ 0.0059 0.0040 0.0043

Net operating CF / total assets 0.0824*** 0.0739*** 0.0730***

Buse R-square 33.93% 35.97% 36.09%

Panel B: Non-cash WC accruals Pred. sign One-way random effects regression results

(1) (2) (3)

Intercept ? 0.0366 0.0252 0.0085

Test variables:

BANK BANKDUM (H1)BANKCAT (H1)

++

0.1548***0.0768***

CASH (H2) 0.1674*** 0.1738***

PPE (H2) 0.0856*** 0.0899***

Control variables:

Change in value added / total assets +/ 0.0824*** 0.1088*** 0.1091***

Change in net operating CF / total assets 0.0092*** 0.0084*** 0.0083***

Lagged current accruals (LAGCUR2) 0.0139*** 0.0221*** 0.0224***

Natural log (1 + years since start-up) 0.0217 0.0376 0.0419**

Natural log (total assets) +/ 0.0030 0.0013 0.0016

Net operating CF / total assets 0.0822*** 0.0570*** 0.0571***Buse R-square 21.61% 29.62% 29.76%

* indicates significance at the 10% level, ** at the 5% level and *** at the 1% level

7/29/2019 SSRN-id967386

35/41

33

Table 7: The determinants of the bank lending decision

This table presents logit regression results explaining the likelihood of obtaining a first bank loan. The test variable is

trade accruals in Panel A and non-cash working capital accruals in Panel B, respectively. Model 1 (and 1a) uses lagged

current accruals as the test variable. In Model 2, the test variable is calculated as a two-year average of current accruals

prior to obtaining the bank loan. Finally, the test variable in Model 3 is calculated as lagged current accruals minus the

firm-level time series average of this variable over the sampling period. The sample includes data on 127 firms that raised

no bank debt in the start-up year (623 firm-year observations).

Panel A: Trade accruals Pred. sign Logit regression results

(1) (1a) (2) (3)

Intercept ? 0.7138 0.0874 0.0750 0.3259

Test variable +/ 0.2272 0.1998 0.1178 0.0585

Growth opportunities + 0.9984*** 0.9581*** 0.8971*** 0.9168***

Net operating CF / total assets 0.7859* 0.6265 0.7423* 0.7613*

Net cash and equivalents / total assets 1.7924* 1.6398* 1.4493* 1.4672*

PPE / total assets + 1.9701** 2.0954** 2.0804** 2.1781**

PPE / total assets * Failure risk 2.2257** 2.1517** 2.2062**

Failure risk 0.1065** 0.1095** 0.1102** 0.1129**

Current liabilities / total assets 1.0345** 1.3531** 1.3333** 1.3551**

Natural log (1 + years since start-up) + 1.4094** 1.2029** 1.2017** 1.1648**

Natural log (total assets) + 0.2183* 0.2552** 0.2428** 0.2568**

Nagelkerke R-square 22.41% 27.15% 26.98% 27.07%

Panel B: Non-cash WC accruals Pred. sign Logit regression results

(1) (1a) (2) (3)

Intercept ? 0.7113 0.1034 0.0706 0.1307

Test variable +/ 0.3496 0.3212 0.0554 0.0923

Growth opportunities + 1.0454*** 1.0045*** 0.9140*** 0.9350***

Net operating CF / total assets 0.8170* 0.6415 0.7021* 0.6991*

Net cash and equivalents / total assets 1.9305* 1.7767* 1.5479* 1.5817*PPE / total assets + 2.0903** 2.1161** 2.0940** 2.1278**

PPE / total assets * Failure risk 2.3086** 2.1953** 2.2170**

Failure risk 0.1099** 0.1117** 0.1097** 0.1100**

Current liabilities / total assets 1.0469** 1.3846** 1.3497** 1.3312**

Natural log (1 + years since start-up) + 1.4032** 1.1985** 1.1883** 1.1995**

Natural log (total assets) + 0.2241* 0.2625** 0.2459** 0.2527**

Nagelkerke R-square 22.86% 27.57% 26.94% 27.02%

* indicates significance at the 10% level, ** at the 5% level and *** at the 1% level.

7/29/2019 SSRN-id967386

36/41

Appendix A: Correlation matrix for the explanatory variables

CASH PPE VA NOCF LAGCUR1 LAGCUR2 LNAGE

CASH 1.0000

PPE 0.0314 1.0000

VA 0.0762*** 0.1572*** 1.0000

NOCF 0.0469** 0.1039*** 0.0596*** 1.0000

LAGCUR1 0.0189 0.0335 0.0519** 0.0029 1.0000

LAGCUR2 0.0233 0.0465** 0.0795*** 0.0531** 0.6786*** 1.0000

LNAGE 0.0643*** 0.1156*** 0.1238*** 0.0031 0.0478** 0.0027 1.0000

LNSIZE 0.1701*** 0.0729*** 0.0544** 0.0010 0.0389* 0.0244 0.1361*

NOCF 0.3087*** 0.2063*** 0.1251*** 0.2820*** 0.1624*** 0.1960*** 0.0558*

7/29/2019 SSRN-id967386

37/41

7/29/2019 SSRN-id967386

38/41

36

Dechow, P.M. , Kothari, S.P. and R.L. Watts, 1998, The Relation between Earnings and Cash

Flows,Journal of Accounting and Economics 25 (2), 133-168.

DeFond, M.L. and J. Jiambalvo, 1994, Debt Covenant Violations and Manipulation of

Accruals,Journal of Accounting and Economics 17(1-2), 145-176.

Degryse, H. and S. Ongena, 2002, Bank-Firm Relationships and International Banking Markets,

International Journal of the Economics of Business 9(3), 401-417.

Degryse, H. and S. Ongena, 2005, Distance, Lending Relationships, and Competition, Journal of

Finance 60(1), 231-266.

Degryse, H. and P. Van Cayseele, 2000, Relationship Lending within a Bank-Based System:

Evidence from European Small Business Data, Journal of Financial Intermediation 9(1),

90-109.

de Haan, L. and J. Hinloopen, 2003, Preference Hierarchies for Internal Finance, Bank Loans,

Bond, and Share Issues: Evidence for Dutch Firms, Journal of Empirical Finance 10(3),

661-681.

Deloof, M. and M. Jegers, 1999, Trade Credit, Corporate Groups, and the Financing of Belgian

Firms,Journal of Business Finance & Accounting 25(7-8), 945-966.

Diamond, D.W., 1989, Reputation Acquisition in Debt Markets,Journal of Political Economy

97(4), 828-862.

Dun & Bradstreet, 1994, The Failure Record,Annual publication.

Erickson, M. and S. Wang, 1999, Earnings Management by Acquiring Firms in Stock for Stock

Mergers,Journal of Accounting and Economics 27(2), 149-176.

Francis, J.R., Maydew E.L. and H.C. Sparks, 1999, The Role of Big Six Auditors in the

Credible Reporting of Accruals,Auditing: A Journal of Practice and Theory 18(2), 17-34.

Friedlan, J.M., 1994, Accounting Choices of Issuers of Initial Public Offerings, Contemporary

Accounting Research 11(1), 1-31.

7/29/2019 SSRN-id967386

39/41

37

Guenther, D.A., 1994, Earnings Management in Response to Corporate Tax Rate Changes:

Evidence from the 1986 Tax Reform Act, The Accounting Review 69(1), 230-243.

Han, J.C.Y. and S.W. Wang, 1998, Political Costs and Earnings Management of Oil Companies

During the 1990 Persian Gulf Crisis, The Accounting Review 73(1), 103-117.

Hart, O., 1995, Firm Contracts and Financial Structure, Clarendon Press, Oxford.

Healy, P.M., 1985, The Effect of Bonus Schemes on Accounting Decisions, Journal of

Accounting and Economics 7(1-3), 85-107.

Heron, R. and E. Lie, 2002, Operating Performance and the Method of Payment in Takeovers,

Journal of Financial and Quantitative Analysis 37(1), 137-155.

Holtz-Eakin, D., Joulfaian, D. and H.S. Rosen, 1994, Sticking it out: Entrepreneurial Survival

and Liquidity Constraints,Journal of Political Economy 102(1), 53-75.

Huyghebaert, N., 2006, On the Determinants and Dynamics of Trade Credit Use. Empirical

Evidence from Business Start-ups,Journal of Business Finance & Accounting 33(1-2), 305-

328.

Huyghebaert, N. and L.M. Van de Gucht, 2004, Incumbent Strategic Behavior in Financial

Markets and the Exit of Entrepreneurial Start-ups, Strategic Management Journal 25(7),

669-688.

Huyghebaert, N. and L.M. Van de Gucht, 2007, Determinants of Financial Structure. New

Insights from Business Start-ups,European Financial Management13(1), 101-133.

Jaggi, B. and P. Lee, 2002, Earnings Management Response to Debt Covenant Violations and

Debt Restructuring,Journal of Accounting, Auditing and Finance 17(4), 295-324.

Jones, J.J., 1991, Earnings Management During Import Relief Investigations, Journal of

Accounting Research 29(2), 193-228.

Laitinen, E.K., 1994, Traditional versus Operating Cash Flow in Failure Prediction, Journal of

Business Finance & Accounting 21(2), 195-217.

7/29/2019 SSRN-id967386

40/41

38

Leuz, C. Nanda D.J. and P. Wysocki, 2003, "Investor Protection and Earnings Management: An

International Comparison,"Journal of Financial Economics 69(3), 505-527.

Lopez Iturriaga, F.J., 2005, Debt Ownership Structure and Legal System: An International

Analysis,Applied Economics 37(3), 355-365.

Manove, M., Padilla, J. and M. Pagano, 2001, Collateral versus Project Screening: A Model of

Lazy Banks, The Rand Journal of Economics 32(4), 726-744.

Ng, C.K., Smith, J.K. and R.L. Smith, 1999, Evidence on the Determinants of Credit Terms

Used in Interfirm Trade,Journal of Finance 54(3), 1109-1129.

Persson, H., 2004, The Survival and Growth of New Establishments in Sweden, 19871995,

Small Business Economics 23(5), 423-440.

Petersen, M.A. and R.G. Rajan, 1997, Trade Credit: Theories and Evidence, Review of

Financial Studies 10(4), 661-691.

Rangan, S., 1998, Earnings Management and the Performance of Seasoned Equity Offerings,

Journal of Financial Economics 50(1), 101-122.

Ravid, S.A. and M. Spiegel, 1997, Optimal Financing Contracts for a Start-up with Unlimited

Operating Discretion,Journal of Financial and Quantitative Analysis 32(3), 269-286.

Richardson, V.J., 2000, Information Asymmetry and Earnings Management: Some Evidence,

Review of Quantitative Finance and Accounting 15(4), 325-347.

Schumpeter, J.A., 1934, The Theory of Economic Development, Cambridge: Harvard

University Press.

Sercu, P., Vander Bauwhede, H. and M. Willekens, 2003, How Inside are House Banks

Really? Evidence from Earnings Management in Belgium, working paper K.U. Leuven.

Subramanyam, K.R., 1996, The Pricing of Discretionary Accruals, Journal of Accounting and

Economics 22(1-3), 249-281.

Teoh, S.H., Welch, I. and T.J. Wong, 1998a, Earnings Management and the Long-Run

Performance of Initial Public Offerings,Journal of Finance 53(6), 1935-1974.

7/29/2019 SSRN-id967386

41/41

Teoh, S.H., Welch, I. and T.J. Wong, 1998b, Earnings Management and the Underperformance

of Seasoned Equity Offerings,Journal of Financial Economics 50(1), 63-99.

Vanhoutte, C. and L. Sels, 2006, An Inquiry into the Impact of Pre-Start-up Business Planning

on a New Business Ventures Performance, working paper K.U.Leuven.

Young, S., 1998, The Determinants of Managerial Accounting Policy Choice: Further Evidence

for the U.K.,Accounting and Business Research 28(2), 131-143.

Wilner, B.S., 2000, The Exploitation of Relationships in Financial Distress: The Case of Trade

Credit,Journal of Finance 55(1), 153-178.