The relationship between bank credit and trade credit

Bank loans have been identified as an important factor in both developed and developing countries 16. It is offered after creditworthiness evaluation based on the borrower's past performance. However, some level of information asymmetry exists between corporate insiders and outsiders, which limits banks' capacity to evaluate firms’ creditworthiness 5, 17 correctly. Firms plagued by information asymmetry often have difficulty obtaining bank loans, especially during financial crises.

This lack of bank credit may induce firms to consider trade credit as a substitute source of financing18, 19, 20. In contractionary periods, suppliers generally have fewer difficulties facing information asymmetry due to the intimate knowledge about a regular buyer and the ability to repossess and redeploy the goods sold 21. Hence, suppliers with low financing costs could provide the much-needed trade credit to financially limited purchasers to nurture a long-term relationship 12, 22. This phenomenon can be referred to as the “redistribution” effect.

Numerous studies have investigated the benefits of trade credit. One of the primary advantages of trade credit for purchasers is that the granted period buys them time to evaluate the product's quality13, 23. The buyer can decline payments if the faulty products result in decreased transaction costs. Obtaining trade credit with favourable terms and conditions helps lower overall borrowing costs24, 25. Furthermore, firms can match their payments to suppliers with customer, eliminating the gap between cash inflow and cash outflow. As a result, the cost of managing inventory would be reduced. Financially constrained firms typically use trade credit to address excessive costs and the unavailability of capital market funding arising from asymmetric information. Importantly, trade credit is shown to help enable the survival of firms in financial crises 22, 26, 27.

However, if firms do not make use of the early discount facility, trade credit can be an expensive source of financing28, 29. High cost is the reason why firms prefer short-term bank loans, and only when firms cannot obtain more bank loans will they resort to trade credit, creating what is called a “pecking order” in the choice of financing sources.

Determinants of trade payables

The impact of financial constraint on the link between bank credit and accounts payable

A firm's accessibility to bank credit is affected by its size and asset tangibility. Firms with more tangible assets may have greater access to external funds 53. Large firms tend to have higher creditworthiness and better access to capital markets compared to small firms41, 42. Size and tangible assets can be regarded as factors that help firms benefit more from bank credit, thus increasing the levels of short-term debt. Therefore, firms with higher levels of size and asset tangibility should have better access to bank loans at better terms and conditions, thus being less financially constrained and reducing the need for trade credit.

The hypothesis is represented as follows:

Hypothesis 7: Financial constraints tend to increase the negative effect of bank credit on on accounts payable.

Empirical model

The research employs the following baseline model to evaluate the hypotheses from H1 to H6:

Paya = β + βSize + βShortdebt + βLongdebt + βCash + βInventory + βSalegr + a + Ɛ

The research employs the following model to evaluate the hypothesis from H7:

Paya = β + βSize + βShortdebt + βLongdebt + βCash + βInventory + βSalegr + βFC*Shortdebt + βFC*Longdebt + βFC*Cash + a + Ɛ

Where: Paya is the dependent variable, measured as the ratio of trade payables to total assets34, 37. Size is the logarithm value of total assets34, 54. Shortdebt is the main variable of interest, measured as the ratio of bank loans of less than one year to total assets. Longdebt is another main variable of interest, measured as the ratio of bank loans of more than one year to total assets34, 55. Cash is the proxy for cash holdings, measured as the ratio of cash holdings to total assets37. Inventory is the variable representing the level of inventories, measured as the ratio of total inventories to total assets49, 45. Salegr is the annual growth rate of revenue34, 55. FC is the financial constraint variable, proxied by size and asset tangibility. The interaction variables formed between FC and bank credit, and cash are included to evaluate the hypothesis H7. a is the individual effect, and Ɛ is the residual.

Research sample and estimation strategies

The research employs a panel dataset covering 590 firms listed in Vietnam from 2010 to 2022. The financial data are retrieved from the Thomson Refinitiv database. We remove firms with fewer than three years of observation due to extreme values are likely to be attached to these cases. The final data comprises 3,658 firm-year observations.

We employ panel data estimation methods, including fixed effects model (FEM) and random effects model (REM). As panel data have individual effects, these methods are more appropriate than Ordinary Least Squares (OLS). To further ensure the robustness of the research findings, we employ random effects with industry dummies to control for the characteristics of the industry on the tendency to use trade payables 45. Finally, we try to address potential endogeneity issues emanating from the two-way relationship between the dependent and independent variables, in this case the choice between trade and bank credit could be simultaneously determined56. All the models have been tested for the existence of problematic multicollinearity through the Variance Inflation Factor test 57. All the VIF values are lower than 4, indicating that the models are not subject to high level of multicollinearity.

Descriptive statistics and correlation matrix

Table 1 presents the descriptive values of the variables in the model. Paya on average accounts for approximately one tenth of the total assets. Shortdebt’s mean value is to somewhat similar to that of Paya, indicating that the two sources of financing might play equal roles in corporate capital structure. Longdebt has a higher mean of 13.85 percent. Compared to Luu & Nguyen 55, who examined listed firms in Vietnam from 2011 to 2019, we have a similar value of Paya, but a smaller value of Shortdebt and a higher value of Longdebt. This could be due to the effect of Covid-19 that makes long-term lending much more risky for banks.

Cash is also close to Paya, with hortdebt and Longdebt values. Inventory, on average, accounts for one-fifth of the total assets. Salegr is not favorable, with a negative value of 0.2 percent. This could be due to the effect of Covid-19 outbreak that negatively affects the performance of firms in Vietnam.

Table 2 presents the pairwise correlation coefficients of variables in the model. We can see that Shortdebt is positively linked to Paya, while Longdebt is negatively related to Paya. Cash is negatively related to Paya, suggesting a substitution effect rather than the argument that firms prepare cash to pay vendors. Large firms tend to use less vendor financing, supporting substitution effect. As firms have more inventory, they use more trade financing to support the associated costs. Overall, we can see evidence to support the dominating substitution effect, except for the positive correlation between Paya and Shortdebt. However, it is crucial to note that the correlation coefficients refer only to the association between two variables, without considering the other covariates, and this can easily lead to biases in estimating coefficients. Therefore, it is important to proceed with multivariate regressions to verify the hypotheses established above.

Regression results and discussion

Table 3 displays regression results using the Fixed Effects model (FEM), Random Effects Model (REM), REM with endogenous treatment (REM_endo), and REM with industry dummies. As previously discussed, we make an effort to address the potential endogeneity issue emanating from the two-way relationship between the dependent and independent variables, in this case the choice between trade and bank credit could be simultaneously determined 56. We use the one-period lead value of Paya (Paya), rather than the current value of Paya, as the dependent variable.

From Table 3, Shortdebt is generally negatively and significantly related to Paya, suggesting that firms tend to consider the two sources of financing as substitutes. Trade credit is more expensive than bank credit17, 55, 58. Therefore, firms with easy access to bank loans rely less on trade credit to save funding costs for short-term needs34, in line with the hypothesis 1b.

It is interesting to see that even though Vietnam is a young financial market with few financing options, firms still want to use less vendor credit, if they can use more short-term bank loans. This result also nullifies the diversification motive established in the research by Tsuruta30 and Kestens et al. 32. Therefore, in Vietnam, firms tend to view bank credit and vendor credit as substitutes, rather than complements, and ignore the insurance premium effect.

With regard to Longdebt, we also find negative and significant coefficients in all four columns. Even though debt of longer maturity is not meant to support short-term financing needs, there is evidence of some substitution effect between long-term bank loans and accounts payable, supporting the hypothesis 2b. This result is in line with the study of Luu & Nguyen 55 in Vietnam. The findings for Shortdebt and Longdebt variables suggest that even though Vietnam is a market plagued by information asymmetry and low institutional quality, commercial banks, through their extensive networks and huge volume of credit granted, can squeeze information effectively.

For the Cash variable, we also witness a negative relationship between cash holdings and accounts payable, in line with the hypothesis 3. This result negates the view that firms prepare cash to pay trade credit as in Wu et al. 36 and Abdulla et al. 37. Meanwhile, Chaieb 38 suggests that abundant cash helps lower the cost of debt, and while we document that there is a negative linkage between bank loans and trade credit, it is natural to expect that firms prioritize more debt in their capital structure and hoard more cash to reduce the cost of debt.

For Inventory and Salegr, these two variables are positively relate to Paya, consistent with the hypotheses 5 and 6. Caglayan et al. 48 find a positive correlation between trade payables and inventories, indicating that firms tend to rely on vendor financing to fund inventories. Cunat 49 identifies a positive relationship between inventories and account payables, arguing that firms with higher inventories tend to have higher trade payables as inventories can serve as collateral. Garcia-Teruel & Martinez-Solano 34 and Petersen & Rajan 29 provide a theoretical perspective, suggesting that firms with growth opportunities tend to gain more financing from suppliers, resulting in a positive correlation between sales growth and accounts payable. Growth opportunities are quite intangible and usually cannot serve as collaterals; as a result, firms might have to resort to the support from their suppliers, rather than banks to fund their expansion.

Table 4 presents the regression results for Model 2, i.e., Model 1 with interaction variables between financial constraints (FC) and bank credit. First, we use Size to guage the level of financial constraint. Compared to Model (2) suggested in Section 3, we remove the individual variables Size, Shortdebt and Longdebt, because the Variance Inflation Factor test indicates that including of these variables leads to serious multicollinearity among the regressors.

The results align with Table 3 for Cash, Inventory and Salegr. Specifically, Inventory and Salegr are positively related to Paya, indicating that firms with higher inventories and revenue growth rates tend to rely on vendor financing. On the other hand, if firms exhibit a stronger preference for bank credit than trade credit, they could hold more cash to reduce the cost of debt, leading to a negative correlation between cash holdings and accounts payable.

Importantly, the interaction variables (Size*bank credit) have negative and significant coefficients. Larger firms tend to experience more substitution effects between bank credit and trade credit. This provides evidence supporting the hypothesis H7: for firms that are not less financially constrained, they will seek more bank credit, rather than trade credit. Increasing bank credit, especially short-term debt, could lead to bankruptcy risk; however, this risk is less problematic for large firms. At the same time, a higher level of bank credit might indicate that firms can access official financing at a favorable conditions. Previous studies also confirm that firms generally prefer bank credit, and only if bank credit is limited will they switch to trade credit55.

To complete the analysis, Table 5 presents the regression results for Model 2, with asset tangibility being used to indicate the level of financial constraint. Tang is measured as the ratio of the net value of property, plant and equipment to total assets. We find that the results are similar to those in Table 4. This again confirms the validity of the hypothesis H7: for firms that are not less financially constrained, they will seek more bank credit, rather than trade credit.