SELF-DECLARED FINANCIAL CONSTRAINTS: EMPIRICAL ANALYSIS FOR A SAMPLE OF SPANISH SMES

Transcription

1 143b SELF-DECLARED FINANCIAL CONSTRAINTS: EMPIRICAL ANALYSIS FOR A SAMPLE OF SPANISH SMES Ana Mol Gómez-Vázquez Ph. D. Student at the Universidad Politécnica de Cartagena Department of Financial Economics and Accounting Ginés Hernández Cánovas* Associate Professor at the Universidad Politécnica de Cartagena Department of Financial Economics and Accounting Topic: B) Finance Keywords: Financial constraints, SMEs, indebtedness, credit rationing

2 SELF-DECLARED FINANCING CONSTRAINTS: EMPIRICAL ANALYSIS FOR A SAMPLE OF SPANISH SMES Abstract This study uses a survey dataset of Spanish SMEs with detail information regarding the existence of credit rationing. The main advantage is that we base our analyses on statements of self-declared financial constraints given by the firm in the survey. We find that rationed firms were ex ante more indebted than firms that obtained financing from their bank. We confirm that financially constrained firms would result as significantly riskier if bank screening were based on credit risk indicators. And we show that credit rationing is a mix of efficient screening and discrimination from the lender based on firm specific characteristics. Resumen Este trabajo usa una muestra de Pymes españolas con información detallada sobre la existencia de racionamiento financiero. La principal ventaja es que basamos nuestros análisis en declaraciones propias de restricciones financieras dadas por las empresas a través de una encuesta. Encontramos que las empresas racionadas estaban más endeudadas con anterioridad a pedir el préstamo que aquellas empresas que obtuvieron la financiación del banco. Confirmamos que las empresas restringidas financieramente aparecerían como significativamente más arriesgadas si el análisis previo del banco se basara en indicadores de riesgo de crédito. Y mostramos que el racionamiento del crédito es el resultado de una mezcla de análisis previo y discriminación basada en características específicas de la empresa.

3 1. Introduction Are financing constraint under the extreme form of credit rationing the rational outcome of bank credit scoring processes based on financial statement indicators? How much additional environmental variables (geographical location, size, age and investment in research and development) matter in the credit rationing decisions? Were rationed firms relatively more indebted ex ante than the rest of the sample? We try to answer these questions by combining the traditional expected determinants of financing constraints in the specific literature with those identified as enhancing borrower risk in the bankruptcy literature. Since our sample is composed by non-listed firms, we focus on corporate credit risk modeling for privately held firms in order to choose the adequate credit scoring measure. Although firms with unlisted equity or debt represent a significant fraction of the corporate sector worldwide, research in this area has been hampered by the scarce availability of public data. This implies that, for privately held firms, accounting based credit scoring models have been mostly applied. We use the credit risk predictor developed by Domingo et al. (1996) for Spanish SMEs. The rest of the paper is organized as follows. Section 2 describes the data and methodology. Section 3 presents some descriptive statistics. Section 5 discusses the main results. Summary and conclusions are given in Section Data and Methodology 2.1. Data In our analyses we use a survey dataset of Spanish SMEs. The questionnaire, collected between April and June, 2009, includes questions about firms financing situation just after the Lehman Brother collapse. We complete our information with accounting data between 2008 and 2010 obtained from the database SABI, elaborated by Bureau Van Dijk Methodology In the following, the two logit model specifications used in our analyses are written in compact forms:

4 Raz( Fc) i 11 j= 1 6 = α + α Identity + γ FinancialStatement + ε (1) 0 j ij k= 1 k ik i 11 Raz( Fc) i = α 0 + α j Identityij + CreditScorei + ε i (2) j= 1 where the dependent variable Raz(Fc) is alternatively, one if the firm belongs to the desire-credit (DES), denied-credit (DEN) or rationed (RAT) subgroup and zero otherwise. These classifications are possible due to the information collected from the survey. The survey includes the following questions to identify the existence of financing restrictions in our sample: 1) has your company desired more credit from financial institutions during the past three months? In case of affirmative answer the following two questions were asked: 2) did the company pay a higher interest rate in order to obtain more credit? 3) did the company demand more credit without obtaining it? We classify as desire-cred firms answering affirmatively to question 1), rationed firms those answering affirmatively to questions 2) or 3), denied-credit firms those answering positively to question 3). The vectors of identity and financial statement variables include a set of measures generally used in previous literature as determinants of firms capital structure and the existence of financial restrictions (De Miguel and Pindado, 2001; Cessar and Holmes, 2003, Hall et al., 2004; Gaud et al., 2005; Degryse and Ongena, 2005; Hernández-Cánovas and Martínez-Solano, 2007, 2010; Brav, 2009). Our 11 identity variables include 8 dummies (Small, Young, Location4, Export, Research, Industry, Construction and Service) taking the value of one if the firm has the relevant characteristic and zero otherwise. Small is a dummy for firms with less than 50 employees, and Young is a dummy for firms whose age is under the 20 th percentile of the sample. We define three dummies for firms belonging to Industry, Construction and Service sector respectively. Export is a dummy for firms exporting part of their production while Research is a dummy for firms having a research and development department. Location4 is a dummy for firms operating in the city of Murcia, which is the capital of the Region. This variable tests how the firm s financial situation is affected by the distance between the firm and the main office that each bank has in the capital of the Region, which acts as headquarter for the Regional affairs. More distance might imply little knowledge about the firm and therefore more asymmetric information problems and more financial restricted. The remaining identity variables are Lbanks, calculated as the natural logarithm of one plus the number of banks the firms works with, LLength that equals to the natural logarithm of one plus the

5 number of years the firms has been working with its main bank, and Main which takes on the value one if more than 50% of debt financing comes from a financial institution and zero otherwise. Our vector of financial statement variables includes the following variables calculated using accounting data for the year 2008: ROS, ROI, ROE, LEV, COSTV and SALEE which measure respectively, the value of operating profits over net sales, operating profits over total assets, net earnings over net worth, firm leverage, interest payments over net sales and net sales over the number of workers. In our second model we substitute the vector of financial statement variables by our credit risk indicator (GL) to test whether this credit risk predictor have relevance if added to the vector of traditional determinants of financing constraints. We test the financial statement variables and credit risk indicator separately to avoid correlation problems. We build firm s credit risk indicator using the bankruptcy prediction model developed by García et al. (1995) for the Spanish SMEs. This multivariate model is calculated with the following ratios: QLIQ, quick assets over current liabilities; ATD, total asset to total debt; INT, total interest payments to sales; AMO, annual amortization to amortizable assets; EAR, earnings before taxes over total debt. This indicator, which evaluates the probability of the firm going bankrupted, is based on traditional financial failure prediction models (Altman, 1968 and Deakin, 1972 among others). This model has already been used as credit risk indicator for the Spanish market by Baixauli and Módica-Milo (2010). All estimates in different specifications are run at a constant number of observations to avoid that our results be driven by sample selection effects caused by missing variables. 3. Some descriptive findings on the sample We describe the characteristics of our sample by looking at firms by size classes in Table 1. Medium-sized firms are much older than micro firms (approximately 26 against 18 years). Firms are also generally bigger in the area dominated by the Capital of the Region (Murcia). As expected, medium-sized firms export (around 49.02% against 12.7%) and have a research department (25.49% against 3.17%) in a much higher proportion than micro firms. As for the activity sector, micro firms are concentrated in the service sector while medium-sized firms are in the industry sector.

6 When we look at bank-firm relationships we find that medium-sized firms have, on average, commercial relationships with 6 banks, while micro firms only with 3. As expected, the relationship with the borrower is longer in medium-sized firms compared to micro firms (20.29 and years respectively). Table 1. Descriptive features by firm size Micro Small Medium Industry Sector Construction Service Location4 (Capital) Research Export Age* s* Length* Observations Percentage values except * which are averages 3.1. Some descriptive findings on credit rationing and financing constraints Our three classifications for firms financial situation identify some potential differences in the intensity of financing constraints. Consider, in fact, that option 2) indicates the existence of a positive difference between demand and supply of credit in correspondence of the additional (demanded and refused) marginal unit of credit (and, therefore, a gap between the reservation price and the market price at that point), while option 3) does not necessarily imply it. Consider also that the set of the desire-credit firms obviously includes as a subset the group of denied-credit and rationed firms, but that many firms (91 out of 577 firms, which accounts for 16% of the sample) respond affirmatively to question 1) and not to questions 2) and 3). These firms may just be financially constrained (being offered additional finance at a price higher than the market rate which they may have refused), but not credit rationed, give the absence of positive answers to questions 2) and 3). As far as our definition of financing constraints gets tighter, the share of financially constrained firms becomes smaller: desire-credit firms are around 60.31% of the sample, rationed firms are around 38.3%, while denied-credit firms are almost 8.67%. In Table 2a we also find that firms belonging to the denied-credit group are smaller than the complementary sample (mean size of 29 employees against 36 of the control sample). Constrained firms (denied-credit and rationed firms) are younger and

7 maintain shorter bank relationships in both mean and median with a difference with respect to the complementary samples ranging between 1 and 5 years. Table 2a. Continuous identity characteristics Our sample includes 577 firms. Firms are grouped by intensity of financing constraints as defined in section 2.2. Size is the number of employees in Age is the number of years since the firm was created. s is the number of banks the firm works with. Length is the number of years the firm has been working with the main bank. All Desirecred Rationed Deniedcred Yes No Yes No Yes No Size Mean Median Sd Age Mean Median Sd s Mean Median Sd Length Mean Median Sd Obs Table 2b. Dummy identity variables All Desirecred Rationed Deniedcred Yes No Yes No Yes No Mircro Size Small Medium Location Main Export Research Industry Sector Construction Service Observations With regards to the credit rationing geographical breakdown, Table 2b shows that 28.2% of sample firms are located in Murcia (the Capital of the Region), and this share doesn t change at all when we consider the different subgroups. However, we do find variations in the composition of the sample when we look at other firm specific characteristics. In table 2b, we find that the proportion of firms in the overall sample (denied-credit subgroup) performing R&D investments is 33% (20%), exporting a fraction of their production is 13% (2%), belonging to the construction sector is 10% (20%) or being dependant on bank debt financing is 33% (38%). Therefore, firms that

8 have being denied a credit seem to be less involved in R&D and export activities, more orientated to the real estate business and more dependent on bank financing. Descriptive evidence provided in Table 3 also suggests that constrained firms underperform with respect to their complementary samples in terms of most of the indicators. Compared with the control sample, firms in the desire-credit group show lower ROS and ROI, whereas firms in the rationed and denied-credit group show lower ROI and productivity per worker. We observe also that firms in the three subgroups have a higher leverage and interest on net sales ratio than the complementary sample. Finally, all the three subgroups of firms have, compared to the whole sample, a lower GL value, which means that they have a higher credit risk. Table 3. Balance variables and credit risk indicator GL is the credit risk indicator. It goes from 0 to 100, with higher values indicating a better financial situation. All Desirecred Rationed Deniedcred Yes No Yes No Yes No ROS Mean Median Sd ROI Mean Median Sd ROE Mean Median Sd LEV Mean Median Sd COSTV Mean Median Sd SALEE Mean Median Sd GL Mean Median Sd Obs financing patterns in the sample In this section we also look at the situation of debt financing for firms in our sample between 2008 and Contrary to our expectations we find in Table 5 that financial leverage remains relatively unchanged during this period, with the amount of 1 In this analysis we keep only those firms having the relevant accounting information during the three years to avoid differences arising due to sample selection.

9 total (bank) debt representing around 63% (24%) of total assets. However, if we analyze the structure of indebtedness we observe that the proportion of short-term (bank) debt over total assets has decreased more than 5.5% (3%) whereas long-term (bank) debt has increased more than 4.5% (4%). Table 4. financing situation Panel A. Total sample Short Financial Leverage Short- Mean Mean Mean Panel B. Micro firms Short Financial Leverage Short- Mean Mean Mean Panel C. Small firms Short Financial Leverage Short- Observ. Observ. Observ. Mean Mean Mean Panel D. Medium-sized firms Short Financial Leverage Short- Mean Mean Mean Observ. Table 5, Panel B, C and D, repeat the analysis by sample size (micro, small and medium-sized firms), showing the same patterns than for the whole sample, with increases (reductions) in long (short) debt financing. Looking at debt levels across firm sizes we observe that micro firms are more indebted than firms of a larger size, whereas the proportion of bank debt is quite similar between groups.

10 4. Results We start our analyses by running our independent variables on the financial leverage ratio and the bank debt ratio in Table 5. Firms are more indebted when are young, work with more banks, maintain shorter bank relationships, are more dependent on bank debt financing, operate in the construction sector, show higher levels of ROS and COSTV, and lower values of ROI. When using the proportion of bank debt over total assets as dependent variable, the variables Location4 and ROE become significant with a negative and positive coefficient respectively. This indicates that firms in the capital of the Region and firms with higher ROE have reduced bank debt financing. However, compared to total debt, the use of bank debt does not seem to depend on age, ROS and the activity sector. Since debt ratios are jointly determined by demand and offer factors, the above results do not help us to shed light on the credit rationing situation. Instead, selfdeclared financial constraints reported by firms in our sample throughout the survey are a better way of analyzing the existence of credit rationing and its determinants. We do this in Tables 6 and 7. The question whether rationed firms were ex ante more indebted has undoubtedly a positive answer. Table 6 shows that firms belonging to the three subgroups have significantly higher leverage ratios. The inclusion of the credit risk indicator in Table 7 shows that financially constrained firms would result as significantly riskier if bank screening were based on the reported risk measures. According to our credit risk indicator a high level is associated with a good financial position of the firm, and this means that the negative sign we find in logit estimates is consistent with our interpretation of efficient screening. The interesting finding though, is that, after correcting for performance, indebtedness and risk measures in Tables 6 and 7, identity variables remain significant. Firm specific characteristics such as being in the construction sector, having a research and development department, the existence of foreign trading, age and the dependence on bank financing do influence the likelihood of experiencing of financial restrictions. Our interpretation is that credit rationing is a mix of efficient screening and discrimination from the lender.

11 Table 5. Analysis of total debt ratio and bank debt ratio OLS: Total debt ratio OLS: debt ratio Constant *** (0.0956) *** (0.0511) Small (0.0318) (0.0170) Young * (0.0371) (0.0198) Location (0.0254) * (0.0136) Export (0.0265) (0.1414) Research (0.0349) (0.0187) Lbanks ** (0.0283) *** (0.0151) Llength *** (0.0257) *** (0.0137) Main * (0.0241) *** (0.0129) Industry (0.0252) (0.0135) Construction * (0.0402) (0.0214) ROS *** (0.2072) (0.1108) ROI *** (0.0460) *** (0.0246)* ROE (0.0088) * (0.0047) COSTV ** (0.1193) * (0.0638) SALEE * (0.0000) (0.0000) Number of observations Adjusted R-squared F *, **, ***: Significant at the 10%, 5%, 1% level

12 Table 6: Analysis self-declared using financial statement variables Logit DES Logit RAT Logit DEN Constant (0.8629) (1.2764) (1.5845) Small (0.2641) (0.4094) (0.4743) Young (0.3070) ( (0.5328) Location (0.2121) (0.3091) (0.3721) Export (0.2190) ** (0.3388) (0.4262) Research (0.2923) (0.4115) * (1.0374) Lbanks (0.2471) (0.3608) (0.4237) Llength (0.2162) (0.3256) (0.3995) Main (0.2034) (0.2980) (0.3531) Industry (0.2072) (0.3205) (0.3852) Construction (0.3386) * (0.5742) * (0.4910) ROS (1.6956) (3.5636) (2.7812) ROI (1.0131) ** (2.0809) (0.9793) ROE (0.0753) (0.3963) (0.1172) LEV *** (0.4147) ** (0.6984) * (0.8663) COSTV (0.9589) (2.4332) (2.7934) SALEE (0.0000) (0.0000) (0.0000) N. Observations Pseudo R Chi *, **, ***: Significant at the 10%, 5%, 1% level

13 Table 7. Analysis self-declared using credit risk indicator Logit DES Logit RAT Logit DEN Constant (0.7853) * (1.2046) (1.4698) Small (0.2575) (0.3945) (0.4688) Young * (0.3029) (0.5156) (0.5301) Location (0.2066) (0.3009) (0.3730) Export (0.2110) * (0.3282) (0.4168) Research (0.2836) (0.4027) * (1.0401) Lbanks *** (0.2353) (0.3360) (0.3977) Llength ** (0.2094) (0.3197) (0.3912) Main *** (0.1992) (0.2828) (0.3374) Industry (0.2038) (0.3133) (0.3806) Construction (0.3265) * (0.5523) * (0.4779) GL *** (0.0028) *** (0.0044) *** (0.0051) N. Observations Pseudo R Chi *, **, ***: Significant at the 10%, 5%, 1% level The interpretation of the sign for some significant identity variables deserves further attention. Table 7, the positive coefficient on the variable Construction in models (2) and (3) indicates that the likelihood of being in the denied-credit or the rationed group increases for firms operating in the construction sector. In addition, credit rationing is more likely for firms exporting a fraction of their production, whereas the likelihood of being credit denied decreases for those firms undertaking research and development activities. As for model (1), the likelihood of being in the desire-credit group significantly increases for firms that are older, work with more banks, maintain shorter banking relationships and obtain more than 50% of their debt financing from banks. 5. Conclusions This study analyzes a survey dataset of Spanish SMEs with detail information regarding their financial situation. Knowing if the firms try to obtain additional credit and whether they obtain it or they have to paid a higher interest rate allows us to study the

14 existence of financial restrictions. The main contribution of this dataset is that we do not need to proxy for the existence of credit rationing because we have the firm s declaration regarding their difficulties to obtain financing. We focus our analyses trying to answer three main questions. Are financing constraint under the extreme form of credit rationing the rational outcome of bank credit scoring processes based on financial statement indicators? How much additional environmental variables (geographical location, size, age and investment in research and development) matter in the credit rationing decisions? Were rationed firms relatively more indebted ex ante than the rest of the sample? First, the question whether rationed firms were ex ante more indebted has undoubtedly a positive answer. Rationed and denied-credit firms have significantly higher leverage ratios. Second, we confirm that financially constrained firms would result as significantly riskier if bank screening were based on the reported risk measures. And third, the likelihood of experiencing financial constraints depends on firm specific characteristics such as activity sector, research and development activity, foreign trading, age and dependence on bank financing. Our interpretation is that credit rationing is a mix of efficient screening and discrimination from the lender. References Altman, E., 1968, Financial ratios, discriminant analysis and the prediction of corporate bankruptcy, The Journal of Finance 23(4): Baixauli, J.S. and A. Módica, 2010, The bias of unhealthy SMEs in bankruptcy prediction models, Journal of Small Business and Enterprise Development 17 (1): Baixauli, J.S., S. Alvarez and A. Módica, 2012, Combining structural models and accounting-based models for measuring credit risk in real estate companies, International Journal of Managerial Finance 8 (1): Brav, O., 2009, Access to capital, capital structure, and the funding of the firm, Journal of Finance 64, Cassar, G. and S. Holmes, 2003, Capital structure and financing of SMEs: Australian evidence, Accounting & Finance 43 (2), Deakin, E.B., 1972, A Discriminant Analysis as predictor of Business Failure, Journal of Accounting Research, spring, Degryse, H. and S. Ongena, 2005: Distance, lending relationships, and competition, The Journal of Finance 60 (1): DeMiguel, A. and J. Pindado, 2001, Determinants of the Capital Structure: New Evidence from Spanish Data, Journal of Corporate Finance 7, García Pérez de Lema, D., A. Arques and A. Calvo-Flores, 1995, Un modelo discriminante para evaluar el riesgo bancario en los créditos a empresas, Revista Española de Financiación y Contabilidad, 24 (82): Gaud, P., E. Jani, M. Hoesli, and A. Bender, 2005, The capital structure of Swiss companies: an empirical analysis using dynamic panel data, European Financial Management 11,

15 Hall, G. C., J. H. Patrick, and M. Nicos, 2004, Determinants of Capital Structures of European SMEs, Journal of Business Finance & Accounting 31, Hernández-Cánovas, G. and P. Martínez-Solano, 2007, Effect of the number of banking relationships on credit availability: Evidence from panel data of Spanish small firms, Small Business Economics 28: Hernández-Cánovas, G. and P. Martínez-Solano, 2010, Relationship lending and SME financing in the continental European bank-based system, Small Business Economics 34: