Skip to main content
Log in

Financial and economic determinants of firm default

  • Regular Article
  • Published:
Journal of Evolutionary Economics Aims and scope Submit manuscript

Abstract

This paper investigates the relevance of financial and economic variables as determinants of firm default. Our analysis covers a large sample of medium-sized limited liability firms. Since default might lead, through bankruptcy or radical restructuring, to firm’s exit, our work also relates to previous contributions on industrial demography. Using non parametric tests we assess to what extent defaulting firms differ from the non-defaulting group. Bootstrap probit regressions confirm that economic variables, in addition to standard financial indicators, play both a long and short term effect. Our findings are robust with respect to the inclusion of Distance to Default and risk ratings among the regressors.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7

Similar content being viewed by others

Notes

  1. Bankruptcy procedures might procrastinate for years before they conclude and the exiting firm is eventually deleted from the register. In Italy, for instance, it took on average more than 7 years, in 2004, for bankruptcy procedures to get to an end (ISTAT 2006).

  2. Such conditions might slightly vary, depending on the internal regulations of every single bank. Since we consider here only default events collected by a single bank, we confront ourselves with a homogeneous definition.

  3. According to the Italian civil law there is a subjective and an objective requisite for accessing bankruptcy law. The former is the professional nature of the activity and the latter is insolvency. The subjects entitled to appeal for the application of bankruptcy law are the firm itself, the bank or other creditors.

  4. The evidence suggests that the subjects involved do not always find it convenient to apply for bankruptcy law (Shrieves and Stevens 1979; Gilson et al. 1990; Gilson 1997; Hotckiss et al. 2008).

  5. Starting from similar considerations Grunert et al. (2005) propose an augmented version of a standard financial model of default prediction which also includes two “soft” non-financial characteristics (managerial quality and market position) among the regressors.

  6. The same message is consistent with broad sense neoclassical models of firm-industry dynamics (see, for instance, Jovanovic 1982; Ericson and Pakes 1995; Melitz 2003), as well with models originating from the evolutionary tradition (see Winter 1971; Nelson and Winter 1982).

  7. A huge empirical literature has highlighted the positive effect exerted on survival by the technological characteristics of the firms, like R&D expenditures or patents (see Agarwal and Audretsch 2001, for a review). Unfortunately we lack the necessary data to include these further dimensions in our analysis (see details in Section 2).

  8. The identity of firms has not be disclosed to us. The matching procedure was performed directly by the Bank.

  9. Zmijevski (1984) analyzes this point in depth. Notice however that default events tend to be over-represented in the samples typically employed in that literature, an opposite situation as compared to the problem we must face here.

  10. At the same time these cuts allow to focus the study only on firms displaying at least a minimal level of structure and operation.

  11. Sectors are defined according to the NACE (Rev.1.1) industrial classification Nomenclature génerale des Activités économiques dans les Communautés Europénnes, which is the standard at European level, and perfectly matches, at the 2-Digit level, with the International Standard Industrial Classification, ISIC.

  12. This is typically the case with credit ratings issued by international agencies (see the “prototype risk rating system” described in Crouhy et al. 2001). This tendency has been more recently confirmed by the effect of the provisions of the Basel II process, encouraging banks and financial institutions to introduce ratings-based internal systems of risk assessment which consider a broad and multidimensional evaluation of their exposure (see BIS 2001).

  13. See Duffie et al. (2007), for recent advances in financial literature, and the works cited therein for a review of duration models based on Distance to Default. Crosbie and Bohn (2003) offer an extensive introduction to Moody’s KMV model, which is also based on Distance to Default theory.

  14. The short horizon of our analysis implies that capital intensity of each firm is basically fixed over the sample period.

  15. Here, as well as in the following, estimates are performed applying an Epanenchnikov kernel, and the bandwidth is set following the “optimal rules” suggested in Section 3.4 Silverman (1986) .

  16. Under the further assumption that the two compared distributions are symmetric, testing H 0 is equivalent to testing for equality of medians between possibly heteroskedastic samples. This is what is usually referred to as the Fligner–Policello test (see Bottazzi et al. 2008, for details).

  17. This is standard in controlled experiments. Consider for instance that you want to test if a given drug is effective. You treat a group of people for one month and then compare the result with an untreated group. Suppose you find significant differences, and therefore conclude that the drug is actually effective. Now if somebody in the control group had some doses of the drug, this of course testify in favour of the effectiveness of the drug, not against it: those control subjects who were not in contact with the drug were different enough to suggest an effective treatment. Coming back to our problem: if we find significant differences comparing the characteristics of defaulters and the control group of non-defaulters, then these differences would be even more significant if we could eliminate defaulters from the control group.

  18. Recall that 1998 is excluded simply because growth rates cannot be computed for that year.

  19. Several refinements of the bootstrap estimates of confidence intervals are discussed in the literature, most notably the BC a and ABC corrections. These methods require an estimate of the bias, which we can only obtain by performing a “first step” probit regression on the overall original sample. This is however exactly what we want to avoid, in order to overcome under-sampling of defaulting firms. Alternatively, one could try to estimate the bias by re-sampling from each random sample. This second order bootstrap seems to us unnecessary due to the relatively large size of the sample considered.

  20. Also notice that the main results persist if we perform bootstrap estimate of linear probability or logit models, and do not change if we take the number of employees as a proxy for size.

  21. Minimizing the overall number of errors is equivalent to maximize the total sum of correctly predicted observations. The weighting is instead introduced to address the specific characteristics of our exercise. True 0’s are indeed much more frequent than true 1’s, simply because default rates in each bootstrapped sample equal the population-wide frequencies presented in Table 1.

  22. The application of the bootstrap to compute model performance measures is particularly important. Zmijevski (1984) indeed shows that classification and prediction errors of the defaulting group are generally overstated without an appropriate treatment of the “choice-based sample” problem.

  23. Statistical irrelevance of sectoral dynamics motivate the exclusion of 2-Digit dummies from the exercise.

  24. Notice that firms’ “ability” to improve their rating does not depend on the exit of better firms from the sample. The matrices are indeed computed taking all the firms which are still in the sample in the last year, when default is measured, and then tracing back their credit rating history. The findings reported in Bottazzi et al. (2008) show that a similar intertemporal behavior in the CeBi index is also appearing when a different division of firms into rating classes is chosen.

  25. Once again, statistical irrelevance of sectoral dynamics motivate the exclusion of 2-Digit dummies from the models.

References

  • Agarwal R, Audretsch DB (2001) Does entry size matter? The impact of the life cycle and technology on firm survival. J Ind Econ 49(1):21–43

    Article  Google Scholar 

  • Agarwal R, Gort M (1996) The evolution of markets and entry, exit and survival of firms. Rev Econ Stat 78:489–498

    Article  Google Scholar 

  • Altman EI (1968) Financial ratios, discriminant analysis and the prediction of corporate bankruptcy. J Finance 23(4):589–609

    Article  Google Scholar 

  • Altman EI, Saunders A (1998) Credit risk measurement: developments over the last 20 years. J Bank Financ 21(11–12):1721–1742

    Google Scholar 

  • Audretsch DB, Santarelli E, Vivarelli M (1999) Start-up size and industrial dynamics: some evidence from Italian manufacturing. Int J Ind Organ 17(7):965–983

    Article  Google Scholar 

  • Beaver WH (1966) Financial ratios as predictor of failure. J Acc Res 4:71–111. Empirical research in accounting: selected studies 1966

    Article  Google Scholar 

  • Bharath ST, Shumway T (2008) Forecasting default with the merton distance to default model. Rev Financ Stud 21:1339–1369

    Article  Google Scholar 

  • BIS (2001) The internal ratings-based approach. Supporting document to the new basel capital accord. Basel Committee on Banking Supervision

  • BIS (2006) Basel II: international convergence of capital measurement and capital standard. Technical report, Basel Committee on Banking Supervision

  • Black F, Scholes M (1973) The pricing of options and corporate liabilities. J Polit Econ 81:637–654

    Article  Google Scholar 

  • Bottazzi G, Secchi A, Tamagni F (2008) Productivity, profitability and financial performance. Ind Corp Change 17(4):711–751

    Article  Google Scholar 

  • Bottazzi G, Tamagni F (2011) Big and fragile: when size does not shield from default. Appl Econ Lett (forthcoming)

  • Brier GW (1950) Verification of forecasts expressed in terms of probability. Mon Weather Rev 78(1):1–3

    Article  Google Scholar 

  • Carey M, Hrycay M (2001) Parameterizing credit risk models with rating data. J Bank Financ 25(1):197–270

    Article  Google Scholar 

  • Cefis E, Marsili O (2007) Going, going, gone. innovation and exit in manufacturing firms. Research paper, Erasmus Research Institute of Management (ERIM)

  • Cosci S, Meliciani V (2002) Multiple banking relationships: evidence from the Italian experience. Manch Sch 70(0):37–54

    Article  Google Scholar 

  • Cosslet SR (1993) Estimation from endogenously stratified samples. In: Maddala GS, Rao CR, Vinod HD (eds) Handbook of statistics. North Holland, Amsterdam, pp 1–43

    Google Scholar 

  • Crosbie P, Bohn J (2003) Modelling default risk. Kmv technical document, Moody’s

  • Crouhy M, Galai D, Mark R (2000) A comparative analysis of current credit risk models. J Bank Financ 24(1–2):59–117

    Article  Google Scholar 

  • Crouhy M, Galai D, Mark R (2001) Prototype risk rating system. J Bank Financ 25(1):47–95

    Article  Google Scholar 

  • Disney R, Haskel J, Heden Y (2003) Entry, exit and establishment survival in UK manufacturing. J Ind Econ 51(1):91–112

    Article  Google Scholar 

  • Duffie D, Saita L, Wang K (2007) Multi-period corporate default prediction with stochastic covariates. J Fin Econ 83:635–665

    Article  Google Scholar 

  • Dunne T, Roberts MJ, Samuelson L (1988) Patterns of firm entry and exit in us manufacturing industries. Rand J Econ 19:495–515

    Article  Google Scholar 

  • Efron B, Tibshirani RJ (1993) An introduction to the Bootstrap. Chapman & Hall, London

    Google Scholar 

  • Ericson R, Pakes A (1995) Markov-perfect industry dynamics: a framework for empirical work. Rev Econ Stud 62(1):53–82

    Article  Google Scholar 

  • Esteve-Pèrez S, Sanchis-Llopis A, Sanchis-Llopis J (2010) A competing risks analysis of firms. Empir Econ 38:281–304

    Article  Google Scholar 

  • Fligner MA, Policello GE (1981) Robust rank procedures for the Behrens–Fisher problem. J Am Stat Assoc 76(373):141–206

    Google Scholar 

  • Foglia A, Laviola S, Marullo-Reedtz P (1998) Multiple banking relationships and the fragility of corporate borrowers. J Bank Financ 22(10–11):1441–1456

    Article  Google Scholar 

  • Gilson SC (1997) Transactions costs and capital structure choice: evidence from financially distressed firms. J Finance 52(1):161–196

    Article  Google Scholar 

  • Gilson SC, John K, Lang LHP (1990) Troubled debt restructurings: an empirical study of private reorganization of firms in default. J Financ Econ 27(2):315–353

    Article  Google Scholar 

  • Grunert J, Norden L, Weber M (2005) The role of non-financial factors in internal credit ratings. J Bank Financ 29(2):509–531

    Article  Google Scholar 

  • Honjo Y (2000) Business failure of new firms: an empirical analysis using a multiplicative hazards model. Int J Ind Organ 18(4):557–574

    Article  Google Scholar 

  • Hotckiss ES, Kose J, Mooradian RM, Thorburn KS (2008) Handbook of empirical corporate finance, vol II, Chapter. Bankruptcy and the resolution of financial distress. Elsevier, Amsterdam

    Google Scholar 

  • Imbens GW (1992) An efficient method of moments estimator for discrete choice models with choice-based sampling. Econometrica 60(5):1187–1214

    Article  Google Scholar 

  • ISTAT (2006) Statistiche giudiziarie civili. Annuari, ISTAT

  • Jovanovic B (1982) Selection and the evolution of industry. Econometrica 50(3):649–70

    Article  Google Scholar 

  • Manski CF, Lerman SR (1977) The estimation of choice probabilities from choice based samples. Econometrica 45(8):1977–1988

    Article  Google Scholar 

  • Manski CF, McFadden D (eds) (1981) Structural analysis of discrete data with econometric applications. MIT Press, Cambridge

    Google Scholar 

  • Mata J, Portugal P (1994) Life duration of new firms. J Ind Econ 42:227–246

    Article  Google Scholar 

  • Melitz MJ (2003) The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica 71(6):1695–1725

    Article  Google Scholar 

  • Merton RC (1974) On the pricing of corporate debt: the risk structure of interest rates. J Finance 29:449–470

    Article  Google Scholar 

  • Nelson RR, Winter SG (1982) An evolutionary theory of economic change. The Belknap Press of Harvard University Press, Cambridge

    Google Scholar 

  • Ongena S, Smith DC (2000) What determines the number of bank relationships? Cross-country evidence. J Financ Intermed 9(1):26–56

    Article  Google Scholar 

  • Schary MA (1991) The probability of exit. RAND J Econ 22(3):339–353

    Article  Google Scholar 

  • Shrieves RE, Stevens DL (1979) Bankruptcy avoidance as a motive for merger. J Financ Quant Anal 14(3):501–515

    Article  Google Scholar 

  • Silverman BW (1986) Density estimation for statistics and data analysis. Chapman & Hall/CRC, London

    Google Scholar 

  • Winter SG (1971) Satisficing, selection, and the innovating remnant. Q J Econ 85(2):237–261

    Article  Google Scholar 

  • Zmijevski ME (1984) Methodological issues related to the estimation of financial distress prediction. J Acc Res 22:59–82. Supplement 1984

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Giulio Bottazzi.

Additional information

We would like to thank the attendants of the “Revolving Doors” conference for useful suggestions on a preliminary version of the paper. The final version benefits from the insightful hints of Elena Cefis and two anonymous referees. The research leading to these results has received funding from the European Community’s Seventh Framework Programme (FP7/2007-2013) under Socio-economic Sciences and Humanities, grant agreement n° 217466.

Appendix: Costruction of book distance to default

Appendix: Costruction of book distance to default

As explained in the text, we start from the naive DD of Bharath and Shumway (2008). For each firm, this is defined as

$$ \label{eq:naive_DD} \mbox{naive DD} = \frac{\ln [(E + F)/F] + \left( r_{i, t-1} - 0.5 \; \mbox{naive} \; {\sigma^{2}_{V}}\right) T }{ \mbox{naive}\; {\sigma_V} \sqrt{T}} , $$
(12)

where E is market value of equity, F is face value of debt, T is the time-to-maturity assuming each firm has issued just one bond maturing in T periods, r i, t − 1 is the firm’s stock return over the previous year, and naive σ V is the volatility of the value of the firms, computed as

$$ \label{eq:s_V} \mbox{naive } {\sigma_V} = \frac{E }{E+F } {\sigma_E} + \frac{F }{E+F } (0.05 + 0.25{\sigma_E}) , $$
(13)

with the last term in parenthesis being an estimate of the volatility of firm debt

$$ \label{eq:s_D} \mbox{naive } {\sigma_F} = 0.05 + 0.25{\sigma_E} . $$
(14)

This default predictor involves a computationally easier estimate of the underlying value of a firm as compared to numerical solution of Black-Scholes-Merton’s equations, and its predictive power of default has been found to be comparable to that of the original DD measure.

To make this definition operational in the context of our dataset, where most of the firms are not publicly traded, we make the following choices, based on available accounting book variables. First, we place the time of computation in 2002, the last year before default is measured in our data, so that T = 1. Second, E is proxied with the sum of annual income after taxes plus face value of outstanding shares, which we define Book Equity, BE. This is simply the denominator of our measure of Leverage, and therefore we can compute it by

$$ \label{eq:our_E} BE = \mbox{Total Assets} / \mbox{Leverage} . $$
(15)

Third, since Total Assets equals the sum of BE plus the stock of outstanding debt, due to Italian accounting practices, we can proxy F via

$$ \label{eq:our_F} D = \mbox{Total Assets} - BE . $$
(16)

Fourth, in place of r i, t − 1, we take the time series average of firm i’s growth rates of Book Equity, μ BE , computed over the years before default occurs (1999–2002). This smoothing is done to incorporate all available past information, which is what the naive DD assumes to be entirely captured by stock returns over the previous year, due to efficient and fully informed stock markets. Fifth, in place of the approximation of debt volatility contained in Eq. 14, we directly compute the volatility of D, as the standard deviation of the growth rates of D over the years 1999–2002. Finally, the same is done for σ BE , the volatility of Book Equity. Therefore, our “accounting book version” of the naive DD becomes

$$ \label{eq:DD_our} \mbox{BookDD} = \frac{\ln [(BE_{2002} + D_{2002})/D_{2002}] + \left( {\mu_{BE}} - 0.5 \; \mbox{naive} \; {\sigma^{2}_{V}} \right) }{ \mbox{naive}\; {\sigma_V}} , $$
(17)

with

$$ \label{eq:s_V_our} \mbox{naive } {\sigma_V} = \frac{BE }{BE + D } {\sigma_{BE}} + \frac{D }{BE+D }{\sigma_D} . $$
(18)

Rights and permissions

Reprints and permissions

About this article

Cite this article

Bottazzi, G., Grazzi, M., Secchi, A. et al. Financial and economic determinants of firm default. J Evol Econ 21, 373–406 (2011). https://doi.org/10.1007/s00191-011-0224-6

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00191-011-0224-6

Keywords

JEL Classification

Navigation