Trade Credit, Bank Credit and Crisis: Some Empirical ...

Margin—The Journal of Applied Economic Research 9 : 4 (2015): 1–29 Sage Publications Los Angeles/London/New Delhi/Singapore/Washington DC DOI: 10.1177/0973801015596854

Trade Credit, Bank Credit and Crisis: Some Empirical Evidence for India? Saibal Ghosh Employing data on an extended sample of manufacturing firms during 1993–2012, the article studies the inter-linkage between trade credit and bank credit and how this interaction evolved during the crisis. Accordingly, we first discuss the relevance of a trade-credit channel in India and subsequently, explore the interplay between these two financing forms. The analysis provides some evidence in favour of a trade-credit channel in India. More importantly, the findings indicate that bank credit and trade credit are complementary, although there was a differential impact on accounts payable (demand) and accounts receivables (supply) during the crisis. Keywords: Trade Credit, Bank Credit, Crisis, Monetary Policy, India JEL Classification: G21, G32, P52

1. Introduction The issue of channels of transmission of monetary policy has engaged the attention of researchers and policy-makers alike. Broadly speaking, two major channels of monetary transmission have been highlighted: the money (or interest rate) channel and the credit channel. According to the money channel, monetary policy affects output through the interest rate. It is argued that a fall in money supply would raise the real interest rate, thereby increasing the cost of capital. The alternative to the money channel—the credit channel—contends that the transmission of monetary policy works through quantity. There are two views regarding how the credit Acknowledgements: I would like to thank the anonymous referee for helping reconstruct the previous draft, weeding out irrelevant discussions and providing insightful comments to ensure focused attention on the central theme. Needless to state, the views expressed and the approach pursued in the article reflects the personal opinion of the author.

The author is with the Department of Economic and Policy Research, Reserve Bank of India, Mumbai, email: [email protected]

2 Margin—The Journal of Applied Economic Research 9 : 4 (2015): 1–29

channel brings about a monetary policy transmission. The former observes that it is through the broad credit channel that money affects output, whereas the latter emphasises the importance of bank lending as a channel of monetary transmission. Proponents of the broad credit channel argue that a rise in the lending rate owing to shrinkage in loan supply increases not only the interest rates of banks but also the cost of external finance for firms. As a result, their retained profits are curtailed. This affects the value of collateral offered and, in turn, raises the cost of external finance. As a result, borrowing becomes expensive, lowering the demand for bank loans as well as other kinds of external finance. The search for a broad credit channel has led researchers to focus on the behaviour of different categories of firms during periods following a tight monetary policy. In fact, studies have focused on several firm-specific criteria, such as size, age, leverage and growth prospects. Indeed, several studies reported a strong association between financial variables and the activity of financially constrained firms (Carpenter et al., 1998; Fazzari et al., 1988; Gertler & Gilchrist, 1994; Guariglia & Mateut, 2006). One argument which has been advanced to explain why firms might exhibit a low sensitivity of investment to financial variables is that, particularly when bank lending is constrained (or more generally, there is a high premium on external finance), firms employ an alternative financing source—trade credit—to overcome liquidity shortages. Simply defined, trade credit (that is, accounts payable) is the amount of short-term loans provided by suppliers to their customers upon purchase of their products (Cunat & Garcia-Appendini, 2012). It is automatically created when the customers delay payment of bills to their suppliers. Trade credit is typically more expensive than bank credit, especially when customers do not use the early payment discount. Yet, according to Berger and Udell (1998), in 1993, roughly 16 per cent of the total assets of small US businesses were funded by trade credit. Similarly, Rajan and Zingales (1995) document that in 1991 funds loaned to customers represented 17.8 per cent of total assets for US firms, 22 per cent for UK firms and more than 25 per cent for European economies such as Italy, France and Germany. Thereafter, Kohler et al. (2000) document that 55 per cent of the total short-term credit received by UK firms during the period 1983–95 was in the form of trade credit. Against this backdrop, this article investigates the interaction between trade credit and bank credit, focusing on India as a case study. More specifically, we address two issues. First, is there evidence of a trade-credit channel of monetary policy in India? Second, and more importantly, what is the nature of the

Ghosh Trade Credit, Bank Credit and Crisis 3

relationship between trade credit and bank credit and how has this association panned out during the crisis? The rest of the analysis continues as follows. Section 2 critically reviews the available literature. In Section 3, we discuss the key variable of interest, the monetary policy variable. Section 4 details the database, followed by the empirical strategy, results and concluding remarks.

2. Literature Overview While there has been a significant volume of research on the determinants of trade credit, there is admittedly limited research on the interrelationship between trade credit and bank credit. Meltzer (1960) first introduced the idea of the trade-credit channel of monetary policy. Using data on US manufacturing firms, the evidence indicated that in response to a contractionary monetary policy, firms with relatively large cash balances increased the average length of time for which credit was extended, especially to firms that were credit-rationed. Following from this line of research, several authors investigated the relevance of the trade-credit channel. For example, Brechling and Lipsey (1963) found similar results in their study of 75 British firms. Thereafter, Schwartz (1974) observed that firms with a relatively low cost of financing borrow more from banks during periods of tight money in order to extend trade credit to downstream firms that encounter difficulties in accessing bank loans. According to de-Blasio (2003), monetary policy has effects on the real economy by reducing firms’ financial resources. In particular, during a tight monetary policy, firms with financial constraints will cut back on their inventory holdings. Moreover, since it is difficult for these firms to obtain bank loans, they will increase the use of the more expensive trade credit (Mateut et al., 2006). Kohler et al. (2000) indicate the existence of a trade-credit channel that mitigates the effect of the bank-credit channel of monetary policy transmission. Nilsen (2002) argues that the dependence of firms on trade credit increases during monetary contractions if they do not have access to markets for traded long-term securities or commercial paper. Using balance sheet data for UK manufacturing firms, Mateut et al (2006) indicate that while bank lending typically declines when monetary policy is tightened, this is complemented by increased trade credit issuances which even out the contractionary impact. Fisman and Love (2003) study the impact of trade credit on industry growth and find that industries with higher reliance on trade credit exhibit higher growth, especially in countries where financial institutions are less robust. Margin—The Journal of Applied Economic Research 9 : 4 (2015): 1–29


Focusing on UK firms for 1981–2000, Atanasova (2007) show that financially constrained firms use more trade credit during periods of monetary contraction. In contrast to studies for advanced economies, research on the tradecredit channel for emerging markets and developing economies (EMDEs) is limited. Employing cross-national data on several Asian economies around the time of the 1997 Asian financial crisis, Love et al. (2007) provide evidence that the use of trade credit increased during the crisis. More recently, Ozlu and Yalcin (2012) report the existence of a trade-credit channel for Turkish manufacturing firms. It is therefore possible that even during periods of tight monetary policy when bank loans are not forthcoming, financially constrained firms do not significantly lower their investments, since they can finance these with trade credit. As a result, trade-credit issuance can increase during periods of contractionary monetary policy because suppliers can monitor their clients more closely and can in fact, threaten to cut off future supplies or repossess goods to ensure repayment (Kohler et al., 2000; Petersen & Rajan, 1997). The presence of a trade-credit channel could therefore dampen the relationship between a firm’s real activities and financial variables and, in general, could weaken the credit channel of transmission of monetary policy. A related strand of the literature explores the interlinkage between trade credit and bank credit. Employing US data Appendini and Garriga (2011) find evidence that firms with high levels of pre-crisis liquidity extended more trade credit during the crisis. Around the same time, employing quarterly Compustat data from 2005–09, Huang (2011) finds that bank credit and trade credit supply are simultaneously determined and negatively related, whereas trade credit demand and bank loans are positively related. This meant a substitute/ complementary effect between bank credit and accounts payable/receivable. Utilising longitudinal data on Spanish small- and medium-sized enterprises (SMEs), Carbó-Valverde et al. (2015) show that the substitution between bank loans and trade credit is dependent on the degree of financing constraints. Studies on trade credit for India date back to the 1980s. In an early attempt, Menon (1977) examined the status of corporate liquidity for over 1500 large public and private companies from 1965–66 to 1974–75. Subsequently, Raj (1982) explored the relevance of trade credit for medium and large Indian firms in both the public and private sectors during 1971–79. Recent research provides evidence in support of an inventory management motive for trade credit (Vaidya, 2011). A common theme running through most studies is a rise in the proportion of trade credit in total external corporate funding during


periods of monetary contraction, presumably a reflection of the attempt by the corporate sector to dilute the severity of monetary policy. All this evidence has a bearing on the financing pattern of firms in India for several reasons. First, India is one of the largest developing countries with a predominantly bank-based financial system—bank asset comprises, on average, anywhere between 90–100 per cent of GDP during the last several years. In this context, it remains a moot question as to the link between trade credit and bank credit. Second, the importance of trade credit could generally be more pertinent in developing economies, such as India, because public capital markets are typically underdeveloped, the legal and regulatory infrastructure to protect creditors is in a state of evolution, and obtaining credit is often quite challenging. According to the World Bank database, India was ranked 142 (out of 189 countries) in its Ease of Doing Business in 2015 with its ranking being on the lower side for most sub-indicators. This raises the relevance of trade credit as an important source of funds for firms (Wilner, 2000). Indeed, evidence during 2001–05 provides strong evidence in support of the growing importance of trade credit: on average, it accounted for 11 per cent of total external finance for large Indian firms and this proportion was significantly higher for SMEs (Allen et al., 2012). And finally, unlike advanced economies where accounts receivables can be easily collateralised, in India, banks have been somewhat reluctant to lend against accounts receivables (RBI, 2000): bills discounted accounted for just over 1 per cent of total credit advanced by commercial banks in 2012. This institutional feature could exert a discernible impact on the use of trade credit by firms.

3. Monetary Policy Variable A crucial variable in our framework is the monetary policy indicator. As Berger and Bouwman (2009) observe, employing a direct measure of monetary policy (or, its change) could be misleading since it might contain anticipatory movements. In other words, movements in monetary policy might respond to information about future developments in the economy (which, in turn, could influence the supply of and demand for trade credit), making it harder to isolate the impact of monetary policy. To circumvent this possibility, we employ a coding process (see, for instance, Romer & Romer, 2004). In our case, the coding process reflects the twin effects of both a price (policy rate) and a quantity (cash reserve ratio, CRR) variable, since both these measures were widely employed during this period. Margin—The Journal of Applied Economic Research 9 : 4 (2015): 1–29


We code the variable on a scale ranging from −1 to +1, where higher values indicate a contractionary monetary policy.1 Coding the direction and magnitude of change instead of employing raw monetary policy numbers enables us to circumvent the anticipatory movements alluded to earlier. Accordingly, in case there is an increase in either the policy rate (‘repo rate’ from 2001 and ‘Bank Rate’ prior to that) and/or the CRR by more than 50 basis points (bps) between two consecutive months in a given year, it is coded as +1.2 Monetary policy in this case is deemed strongly contractionary. An increase of over 25 bps up to 50 bps in either or both these variables between two successive months is coded as 0.5. In that case, the monetary policy is deemed as medium contractionary. Monetary policy is deemed as mildly contractionary provided there has been an increase of up to 25 bps in either or both these variables, in which case it is coded as 0.25. The coding process is just the reverse in the event monetary policy is expansionary. Provided there is no change in any measure during the month, it is coded as zero. Therefore, the maximum value the index can assume in any month equals +2 (provided both the policy rate and CRR are increased by over 50 basis points over the previous month) and the minimum value equals −2 (provided both the policy rate and CRR are reduced by over 50 basis points over the previous month). The raw scores for a month are cumulated for the year as a whole to arrive at an aggregate index for the year. This cumulative score can range from −24 (loosening of both the policy rate and CRR across all months during the year) to +24 (tightening of both the policy rate and CRR across all months during the year). The summary statistics of the monetary policy variable indicates that the mean is −0.338 (with a variance of 4.97). The minimum and maximum values of the variable equal −4 and +6, respectively. Taken together, these numbers imply that, on average, monetary policy has been mildly expansionary for the entire period; there have also been years when monetary policy has been fairly expansionary and, likewise, years when it has been fairly contractionary. Employing these numbers, we construct a monetary policy dummy as follows. We scale all numbers by 24 (the maximum achievable value). A value greater than zero for a given year would then signify a tightening (or contractionary); the reverse would be the case in the event the value is less than zero. Monetary policy is deemed neutral in case the value of the index in a year equals zero. 1 Besley and Burgess (2004) employed a similar strategy to code labour regulations in Indian states during 1958–92. 2 Prior to the initiation of the repo rate as the policy rate in 2001, the bank rate used to be the monetary policy indicator (see RBI, 2001).


Figure 1 (A) provides a graphical representation of the years when tight monetary policy was pursued. It provides evidence in support of tight monetary policy for nine (out of the 20) years of the sample. When we look at policy activism (defined in this case as instances when changes occurred in key monetary policy variables) across months, the evidence appears to suggest greater policy activism during the second half of the year (October–March) compared to the first half (April–September); the average number of policy changes per month equals around seven (Figure 1 (B)). Figure 1(A) Years of Monetary Policy Tightening 1993–2012

Figure 1(B) Monetary Policy Activism by Month, 1993–2012

Source: Author’s calculations based on data from Handbook of Statistics on Indian Economy. Margin—The Journal of Applied Economic Research 9 : 4 (2015): 1–29


4. Database and Sample For the analysis, we employ the Prowess database of the Centre for Monitoring the Indian Economy (CMIE), a private think tank in India. Prowess is a firmlevel database, akin to the Compustat database for US firms and the Financial Analysis Made Easy (FAME) database for UK and Irish public and private limited companies. This database is being increasingly employed for firm-level analysis concerning issues like the impact of privatisation on profitability (Gupta, 2005), the impact of financial liberalisation in alleviating financing constraints (Ghosh, 2006), the linkage between trade liberalisation and productivity (Khandelwal & Topalova, 2011), and the financing pattern of corporate houses in the postliberalisation era (Allen et al., 2012). The present dataset contains financial information on over 10,000 companies, including services and construction companies. Non-listed firms are also part of the database, provided the average sum of sales and assets is at least `200 million (.US $3.5 million) as per the latest audited financial results. The database contains detailed information on firm financial performance culled out from their profit and loss accounts, balance sheets and stock price data, besides information on the ownership type of the firm. The selection of the sample proceeds in several stages. In the first step, we select all manufacturing firms for which information is available in the database for the period 1993-–2012. This provides us with over 11,000 firms. We subsequently delete several firms from the sample. First, since our focus is on trade credit, which one would expect firms to resort to when they reach some minimum maturity, we excluded 236 firms which had been in existence for less than five years. Second, we deleted merged firms and firms with less than three consecutive years of data. In the final stage, we excluded firms for which ownership data was not reported, providing a total of 8459 firms. The composition of the sample by ownership and industry group is set out in Table 1. Nearly 90 per cent of the firms belong to business groups or are private Indian firms. In terms of industry groups, food, textiles and chemicals are the most dominant, comprising nearly 40 per cent of the firms over the sample period. For these firms, we culled out information on all relevant variables including accounts receivables, accounts payables, sales, year of incorporation, bank borrowings and total assets. We also employed ownership classification for firms as provided in the database. The firms belong to four categories: government owned, business groups, foreign firms and private Indian firms. We employed dummies to control for firm ownership. To moderate the influence of outliers, all variables were Winsorized at 1 per cent at both ends of the sample. The Appendix provides the empirical definition of the variables, including data sources and summary statistics.


Table 1 Classification of Sample Firms by Ownership and Industry Groups Firms Panel A: By Ownership Business groups Private Indian Foreign State owned

Number

Percentage of Total

2,265 5,650 315 229

26.8 66.8 3.7 2.7

1,051 1,063 121 772 1,056 900 519 95 1,434 1,448

12.4 12.6 1.4 9.1 12.5 10.6 6.1 1.1 17.0 17.1

Panel B: By Industry Food Textiles Cement Electronics & electrical Chemicals/pharmaceuticals Metals & metal products Plastic & rubber Transport equipment Miscellaneous manufacturing Others Source: Author’s calculations.

Employing a classification akin to Allen et al. (2012), we categorise the financing sources for our sample firms during 1993–2012 (Table 2), segregated according to ownership (Panel A) and year (Panel B). Several features of Table 2 are of note. First, at around 28 per cent each, business groups and Indian private firms appear to have the maximum dependence on bank finance. The lower dependence of government firms is ‘in between’ those for Indian private and business group firms. This could arise if the share of government financing (loans and subsidies) in total borrowings is high, as pointed out by Gupta (2005). Second, foreign banks have the highest dependence on trade credit, consistent with Love and Martinez Peria (2005). At 11 per cent, their dependence on trade credit during the crisis appears to be far lower than that for the entire period which stood at 17 per cent.3 These findings emphasise the importance of trade credit for Indian firms across sub-periods and firm-ownership categories, especially for private firms. Third, the evidence appears to indicate a decline in the role of trade credit during the crisis, compared to an average of 17 per cent for the entire period, suggestive of possible substitutability between these financing types. The definition of the crisis period (2008–10) follows Eichengreen and Gupta (2013).

3

Margin—The Journal of Applied Economic Research 9 : 4 (2015): 1–29


Table 2 Financing Sources for Indian Manufacturing Firms, 1993–2012 (as percentage of total funding) Panel A: By Ownership Internal External Market Bank/FIs Non-market, non-bank Of which: trade credit Panel B: By year Internal External Market Bank/FIs Non-market, non-bank Of which: trade credit

Government

Business Group

Indian Private

Foreign

All

7.3 28.9 16.9 46.9 15.4

11.2 31.5 26.9 30.4 17.0

7.7 35.5 27.7 29.0 16.2

16.8 31.3 17.9 33.8 20.9

9.1 33.9 26.8 30.1 16.6

1993–2000 2001–08 2009–12 2008–10* 1993–2012 6.4 33.5 32.1 28.0 16.9

9.4 34.5 25.6 30.6 16.7

12.2 33.5 22.4 31.9 16.1

10.7 30.4 27.8 20.3 10.8

9.1 33.9 26.8 30.1 16.6

Source: Author’s calculations. Note: *Denotes crisis period; FIs—financial institutions.

5. Estimation Strategy We test two basic propositions: first, when monetary policy is tightened, how does the demand for and supply of trade credit vary for firms with different levels of financing constraints? Second, is there evidence of a trade-off between trade credit and bank credit and how does this behaviour evolve during crisis periods? In order to do this, we apply reduced-form models pertaining to the supply of and demand for trade credit. We estimate both sides separately, since data is available for only one side of the transaction, in contrast to bilateral data in which both parties in the transaction are known (see, for example, Klapper et al., 2011, for an application). Empirically, there are two aspects involved in examining the trade-credit channel: the categorisation of firms as constrained and unconstrained; and a measure of firm financial distress. Prior to outlining the empirical framework, we explain these two important variables. 1. Categorisation of Firms Based on Degree of Financing Constraints Economically, capital structure and financial characteristics affect the ability of firms to raise funds, and in turn influence their behaviour towards trade credit. Based on this argument, we categorise firms as unconstrained and constrained so as to obtain a more comprehensive assessment of their credit conditions and asymmetric informational problems, which are more likely to be manifest during periods of tight monetary policy.


Towards this end, we employ the WW index proposed by Whited and Wu (2006). The WW index is a linear combination of several indicators, such as solvency, profitability and business. Higher WW values indicate greater levels of financial constraint. We sort firms by their WW index and define the first quartile of firms as unconstrained and the last quartile of firms as constrained. 2. Categorisation of Firms Based on Financial Distress Our measure of firm financial distress is based on the MacKie-Mason (1990) specification that provides a formula (the MM-score) to measure the likelihood of financial distress. The MM-score aggregates firmlevel information on profitability, solvency and liquidity and is widely employed as an indicator of financial health. For the sample as a whole, we compute the MM-score for each firm–year combination and split them into quartiles, each quartile being represented by a dummy variable, the d(MM-score)n,ft–1, n = 1, 2, 3, 4. Firms in the topmost quartile are those which are least distressed and, therefore, have the lowest probability of being bankrupt. At the other end, the firms in the bottom-most quartile are the most distressed. Without loss of generality, it appears likely that the least distressed firms are those are most likely to have access to institutional finance. Additionally, as our previous discussion would suggest, the use of trade credit is also likely to differ across firm ownership. For instance, given possible support from their parent firms, it appears likely that foreign firms would exhibit higher dependence on trade credit. Similarly, given the implicit guarantee enjoyed, dependence by government firms on trade credit could also be higher (Bhole & Mahakud, 1984).

5.1 Supply of Trade Credit We employ the following specification to test the hypothesis of a trade credit channel from the standpoint of suppliers:

a

3 3 AR k = a o + | b n [d (MM score)]n,fi (t - 1) + | c n [d (MM score)]n, fit - 1 A fit n=1 n=1

*[DMYP ]t - 1 + { [DMYP ]t - 1 + | h k X k,fi (t - 1) k

+ } [d (OWN )]fit + h it + f fit

(1)

where f indexes firm, i indexes industry and t indexes year; and (AR/A) is the ratio of accounts receivable to total assets for firm f at time t and is the proxy for the supply of trade credit. Margin—The Journal of Applied Economic Research 9 : 4 (2015): 1–29


The variables on the RHS include the following: • [DMYP]: This is the dummy for the monetary policy indicator constructed along the lines discussed earlier. • [d-MMScore]: This represents a set of dummy variables categorising firms according to their extent of financial constraint as discussed earlier and is basically a proxy for the firm’s ability to access institutional finance. Thus, [d(MM-score)]1,f(t–1) represents the least distressed whereas [d(MMscore)]4,f(t–1) represents the most distressed firms (the control category). • [d(MM-score)]f*[DMYP]: This is the focus variable that captures the ability of the f th firm to access institutional finance during periods of tight monetary policy. Therefore, for the existence of a trade credit channel, cn > 0. Provided a trade credit channel exists, a tightening of monetary policy will mean that the quantity of trade credit demanded by constrained firms and the quantity of trade credit supplied by unconstrained firms should both increase. • Xk: The X’s are a vector of control variables. The control variables comprise observable firm-specific variables to capture other potential factors that may influence trade credit. For example, the log of assets (LTA) and log of age (LAGE) capture firm reputation. One can expect that smaller and younger firms may need to offer more trade credit due to quality or reputational concerns (Long et al., 1993). As a result, the estimated coefficient for these variables is expected to be negative. Following Deloof and Jegers (1999), we measure a firm’s growth opportunities by its Tobin’s Q ratio (TQ). In India, as in several other emerging markets, the computation of Tobin’s Q is rendered difficult primarily because a significant proportion of corporate debt is not actively traded. As a result, we compute a proxy for Tobin’s Q. The numerator is the sum of the market value of equity and book value of debt; the denominator is the book value of assets (see, for example, Khanna & Palepu, 1997). Firms may tend to sell more to boost their sales taking into account their higher future growth prospects. As a result, a positive estimated coefficient is expected for this variable. Cash balance is captured by the ratio of cash plus short-term securities to assets (Cash/A). One might expect a negative relationship between cash and the dependent variable since a firm may reduce its cash or equivalents to provide more receivables. In this case, the firm would be holding more receivables other than cash or equivalents. • [d[OWN)]: This is the ownership dummy indicating which category of firms exhibit greater supply response of trade credit. As discussed earlier, we consider four ownership categories: business group firms (BGF), Indian private firms (IPF), foreign firms (FF) and state-owned firms


(SOF), the last being the control category. We also include the relevant interaction terms of [d(MM-score)]f*[DMYP]t*[d(OWN)]f to ascertain the differential response of constrained versus unconstrained firms across ownership to a monetary tightening. • hjt controls for all factors that are industry specific (capital intensity, location, etc.) and industry-year specific (changes in policies, product demand, etc.). • f is random error. The estimation strategy proceeds as follows. We first estimate Equation (1) with fixed effects for all firms. Next, we classify firms according to quartiles based on financial constraints and compare the coefficient estimates. Finally, we estimate the possible trade-off between trade credit and bank credit for the entire sample, as also for constrained and unconstrained firms, separately. The rationale for the choice of fixed-effects regression is based on the Hausman test which supports this framework over the competing randomeffects model. Additionally, since the fixed-effects estimator permits covariance between firm-specific effects and the other regressors, which is likely to be the case, we prefer to use with this model. We also prefer to employ this specification over the censored regression model (for example, Tobit) because although our dependent variable is censored at the lower end, it is theoretically possible for it to exceed unity. We hypothesise that during periods of tight monetary policy, unconstrained firms that have better access to institutional loans will supply more trade credit. In contrast, constrained firms with better access to institutional loans will supply relatively less trade credit compared to unconstrained firms during these contractionary periods.

5.2 Demand for Trade Credit Akin to our previous specification, we employ the following reduced-form equation from the point of view of buyers:

a

3 3 AP k = m o + | b n [d (MM - score )]n, fi (t - 1) + | c n [d (MM - score )]n,fit - 1 A fit n=1 n=1

*[DMYP ]t - 1 + { [DMYP ]t - 1 + | h k X k, fi (t - 1) k

(2) + } [d (OWN )]fit + n it + ffit where on the LHS, (AP/A) is the ratio of accounts receivable to total assets for firm f at time t and is the proxy for trade credit demand. We employ Margin—The Journal of Applied Economic Research 9 : 4 (2015): 1–29


explanatory variables similar to those in Equation (1), although in this case some of the variables might control for different factors. We explore the trade credit channel by examining the coefficient of [d(MM-score)]f and its interaction with the monetary policy variable. Following from Myers and Majluf (1984), institutional loans stand at the top of the pecking order, since they are cheaper than trade credit when the implicit cost is taken into account.4 To the extent that least distressed firms have better access to institutional loans, their demand for trade credit will be the lowest. Additionally, it seems likely that the behaviour of constrained and unconstrained firms will vary according to fluctuations in the business cycle and the evolution of monetary policy. Firms that are constrained and have limited access to institutional loans will be hard-pressed for external finance, necessitating greater recourse to trade credit. Therefore, if a tighter monetary policy entails constrained firms to demand more trade credit, this would suggest that cn > 0. As earlier, we employ a set of control variables which includes LTA, LAGE and TQ as defined previously in Equation (1). In addition, we control for internal finance capacity with the ratio of cash flow to assets (CF/A), where cash flow is defined as net income plus depreciation. According to the pecking order theory, internally generated funds are higher in the pecking order compared to trade credit. As a result, the ability of firms to generate cash internally will decrease with their demand for trade credit. Following from the maturity matching hypothesis (Deloof & Jegers, 1999), firms finance short-term investment or assets with short-term credit. One can expect that firms demand more trade credit if they hold more shortterm assets. We control for this effect by adding accounts receivables (AR/A) and inventories (INV/A) to the model, with an expected positive sign on these variables. We also control for firms’ access to institutional/market finance by including an additional variable. More specifically, since banks are a predominant source of finance for firms, we include the bank debt-to-asset ratio (BkDEBT) 4 The implicit cost refers to the opportunity cost of not taking advantage of the early payments discount. For the most common trade credit terms ‘2/10 net 30’, the annualised implicit interest of trade credit is 43.9 per cent. To calculate the cost we employ the following expression: