A BUSINESS SUCCESS VERSUS FAILURE PREDICTION MODEL FOR ENTREPRENEURS WITH 0-10 EMPLOYEES Robert N. Lussier Springfield College Joel Corman Suffolk University ABSTRACT This empirical survey study presents a success versus failure (SIF) prediction model (S/F = f [advisors, planning, education, minority business ownership, staffing, parents owned a business, record keeping and financial control, capital, industry experience, economic timing}). Using stepwise discriminant analysis, the predictor variables explained 54% of the variance in SIF. The model accurately predicted 75.00% of the surveyed 96 matched pairs firms as being successful or failed. The model will predict a group of businesses as successful or failed more accurately than random classification over 99% of the time (model p=. 0006). INTRODUCTION The important role of small business suggests that an understanding of why firms fail and succeed is crucial to the stability and health of the economy(Gaskill, Van Auken, and Manning 1993). Of major concern to any would-be entrepreneur is the chance of success for the proposed business. Success versus failure prediction research benefits entrepreneurs, those who assist, train and advise them, those who provide capital for their ventures, suppliers, and public policy makers (Altman 1983; Ballantin, Cleveland, and Koeller 1992; Cameron, Kim, and Whetten 1987; D'Aveni 1989; Dugan and Zavgren 1989; Koh and Killough 1990; Lussier l 995a&b, Pech and Alistair 1993; Storey, Keasey, Watson, and Wynarczyk 1987). There are many studies to better understand business success versus failure. However, as Gaskill, Van Auken, and Manning (1993) stated: there are many questions still to be resolved and warrant additional exploration ... previous studies do not provide a comprehensive or unified explanation for small firm failure ... comparisons are needed between successful and failed small business owners. Prior empirical studies of failure have concentrated almost exclusively on financial ratio data, though other studies offailure usually cite managerial variables as being critical (Scherr 1989). The usefulness of ratio-based business failure prediction models has been questioned (e.g., Alves 1978; Corman and Lussier 1991; Lussier and Corman 1995; Gilbert, Menon, and Schwartz 1990; Shelton 1986; Stockton 1989; Sommers and Koc 1987). For example, El- Zayaty (1986) found ratio models to be poor predictors of bankruptcy: of 132 businesses predicted to fail, only 5 were discontinued over a five-year period. These models had about a 21 97 percent Type II error rate. Storey et al. ( 1987) indicated that qualitative data can provide at least as good predictions as traditional financial ratios. Scherr ( 1989) recommended virgin research to search for links between management's ability and failure. This study is not based on financial ratios, but on quantitative and qualitative managerial factors contributing to success or failure. To date, the author has found only two majornonratioempirical studies (Cooper et al. 1990 + 1991 and Reynolds and Miller 1987 + 1989) similar to this study. However, the Reynolds and Miller model cannot predict failure of a business before it starts because the age of the business, and the first year sales are needed to predict failure. The major differences between this study and the Cooper et al. studies are: I. Their sample includes businesses one to three years old whereas this sample includes ages one to ten. 2. Their sample was limited to members of the National Federation oflndependent Business (NFIB); this sample is not. 3. They do not survey failed businesses; this study does. 4. Their sample does not include matched pairs whereas this design does. They surveyed firms conducting business; then after a year or longer, some of the firms failed. At that time, they compared the responses of the failed firms and the surviving firms to analyze the differenceswithouteverquestioningthe failures to ask them why they failed. This matched pairs design avoids comparing larger businesses to smaller ones, retailers to manufacturers or construction companies, and businesses from different locations. Purpose The purpose of this research is to develop and test a generic nonfinancial model that will predict young businesses with 0 - I 0 employees as successful or failed. This study adopts Dun & Bradstreet's(l 994: i) definitions of failure and discontinuance. Business failures are firms involved in court proceedings or voluntary actions involving losses to creditors. Chapter 7 and Chapter 11 companies are both considered failures due to loss to creditors. Chapter? companies liquidate whereas Chapter 11 companies restructure their debt and stay in business. Firms going out of business without loss to creditors are not considered business failures: they are discontinued businesses. To be considered a success the business must make at least industry average profits. A young firm is 0 to I 0 years old. METHODOLOGICAL DEVELOPMENT AND TESTING OF THE LITERATURE MODEL There is no generally accepted list of variables distinguishing business success from failure. Prior research has created discrepancies within the literature by reporting different variables as contributing factors to success or failure. The two most commonly stated distinguishing variables are capital and management experience. However, in 20 journal articles only 14 (70%) specifically state that these two variables contribute to success versus failure. To help clarify discrepancies, this study supports or does not support each of the major variables in the literature later in the implications section of this article. The model in this study was developed by including the fifteen major variables, identifioo in 20 journal articles, as contributing to success versus failure. See Table I for an explanation 22 __ ... ,,, of the 15 variables, and Table 2 for a comparison of the 20 studies that support, do not support, or do not mention each variable. Table I EXPLANATION OF SUCCESS VERSUS FAILURE VARIABLES I. Capital (capt). Businesses that start undercapitalized have a greater chance of failure than firms that start with adequate capital. 2. Record keeping and financialcontrol (rkfc). Businesses that do not keep updated and accurate records and do not use adequate financial controls have a greater chance of failure than firms that do. 3. Industry Experience (inex). Businesses managed by people without prior industry experience have a greater chance of failure than firms managed by people with prior industry experience. 4. Management Experience (maex). Businesses managed by people without prior management experience have a greater chance of failure than firms that are managed by people with prior management experience. 5Planning(plan). Businesses that do not develop specific business plans have a greater chance of failure than firms that do. 6Professiona1Advisors (prad). Businesses that do not use professional advisors have a greater chance of failure than firms using professional advisors. 7 Education (educ). People without any college education who start a business have a greater chance of failure than people with one or more years of college education. 8. Staffing (stall). Businesses that cannot attract and retain quality employees have a greater chance of failure than firms that can. 9. Product/Service Timing (psti). Businesses that select products/services that are too new or too old have a greater chance of failure than firms that select products/services that are in the growth stage. I 0. Economic Timing (ecti). Businesses that start during a recession have a greater chance of failure than firms that start during expansion periods. 11 Age (age). Younger people who start a business have a greater chance of failure than older people starting a business. l 2Partners (part). A business started by one person has a greater chance of failure than a firm started by more than one person. 13 Parents (pent). Business owners whose parents did not own a business have a greater chance of failure than owners whose parents did own a business. 14. Minority (mior). Minorities have a greater chance of failure than nonminorities. l 5Marketing(mrkt). Business owners without marketing skills have a greater chance of failure than owners with marketing skills. 23 Table 2 A COMPARISON OF VARIABLES IDENTIFIED IN THE L.ITERATURE AS FACTORS CONTRIBUTING TO BUSINESS SUCCESS VERSUS FAIL.URE Senior Independent variables Author capt rkfc inex manx plan grad educ staf psti ecti aae part pent mior mrkt Cooper 90 F B!::au2lda 112 E i3S::t.DQls;l11: 112 E E yesper E E E 14 10 10 2 0 N F E E E N 14 11 3 2 E 9 0 N F N E ' 3 N ~ 5 2 F N E E E ' 1 E 5 3 F F N N ~ E E 1 3 3 1 F 1 0 F f 2 0 N E 5 1 Total F Total N Tgtal - 4 10 9 ) 7 11 11 13 13 12 16 16 19 16 14 F supports variable as a factor contributing to failure N does not support variable a factor contributing to failure - does not mention variable as a contributing factor The Literature Model variables S/F = f (capital, record keeping and financial control, industry experience, management experience, planning, professional advisors, education, staffing, product/service timing, economic timing, age of owner, partners, parents owned a business, minority, marketing skills) Discriminant function analysis is used to predict group membership when the dependent variable is nominal/categorical. Therefore, discriminant analysis was used to predict the 96 sample businesses as successful or failed; based on the 15 independent variables. The stepwise procedure was run on all the variables in the model to identify those variables that do or do not distinguish success versus failure. The elimination of the variables without discriminant ability results in the development of the prediction model. DESCRIPTIVE STATISTICAL RESULTS The sample included approximately 20 percent from Connecticut, 5% Maine, 44% Massachusetts, 19% New Hampshire, 9% Rhode Island, and 6% Vermont. The mean age of 24 the failed and successful firms was 5.5 and 5 .8. Industry representation (as classified by Dun +Bradstreet) includes approximately: 2% agriculture, 14% construction, 17% finance, 10% manufacturing,22% retailing, 3% wholesale, 6% transportation and communication, and 25% services. For a comparisonofthe descriptive statistics, measurement of the 15 variables in the model, and correlation matrix see Table 3. Table 3 DESCRIPTIVE STATISTICS AND CORRELATION MATRIX Failed Failed Success Success Variables Mean/ s.d. Mean/ s.d. (Measurement method) Frequency 0/o Frequencyo/o I. Capital 4.09 1.52 4.58 1.37 (I adequate - 7 inadequate) 2. Record keeping and financial control 4.64 1.64 4.88 1.42 (I poor - 7 good) 3. Industry experience 9.20 8.26 6.74 7.15 (number of years) 4. Management experience 9.22 7.78 6.26 8.15 (number of years) 5. Planning 4.09 1.92 3.58 1.53 (I specific - 7 no plan) 6. Professional advise 3.71 1.71 4.52 1.75 (I used - 7 not used) 7. Education 15.84 3.28 14.54 2.53 (number of years) 8. Staffing 5.13 1.56 4.48 1.80 (I difficult - 7 easy) 9. Product/service timing 3.84 1.49 3.90 1.33 (I intro. - 7 decline) 10. Economic timing 4.07 1.79 4.40 1.69 (I recession - 7 expand) 11. Age ofowner 36.98 8.31 35.34 9.38 (number of years) 12. Partners 40% 34% (%with> I owner/part.) 13. Parents 33% 44% (% parents own a business) 14. Minority 7% 6o/o (o/o of minority owners) 15. Marketing 4.44 1.84 3.74 1.78 (I unskilled - 7 skilled 25 Table 3 Continued 2 3 4 5 6 7 8 9 10 II 12 13 14 2 ·.212' 3 ·.142 ·.08 4 .,34•• -.043 39•• 5 .24' -.37" ·.04 -.07 7 ·.23' .07 ·.21' .08 ·.00 ·.05 8 -.01 .07 .10 .06 .01 -.10 ·.05 9 .07 .16 .06 ·.09 ·.17 ·.18 ·.05 .03 10 -.03 ·.16 ·.02 .05 .13 .06 ·.16 ·.05 .00 11 ·.21' .05 .21' .52" ·.09 ·.12 -.02 .01 ·.10 .10 12 ·.19+ .16 .01 .06 ·.05 ·.09 .05 ·.10 .01 ·.I I .07 13 .06 .OJ .14 ·.12 .12 .03 ·.21' .02 ·.26' .05 ·.13 ·.02 14 -.04 .01 .15 .06 .01 -.II ·.06 ·.04 .II ·.01 ·.38".10 .07 15 ·.44" .15 .16 .49" ·.19+ ·.26' .15 .19 ·.09 ·.05 .29" .05 .01 .00 n = 45 failed, and 51 successful firms + p < .10 • p < .05 •• p < .01 SAMPLING METHODOLOGY The sample was limited to the six New England states--Connecticut, Maine, Massachusetts, New Hampshire, Rhode Island, and Vermont. A total of I 03 questionnaires were return, but seven were dropped from analysis due to missing data. Therefore, 96 usable questionnaires were analyzed: 45 from failed and 51 from successful firm owners. The combined response rate was 43%. This rate is high for a failure survey. 26 The population of failed businesses includes Chapter 7 and 11 companies. Due to difficulties in locating liquidated business owners, the failure sample frame was Chapter 11 companies. Lussier (I 995a&b) and Lussier and Corman ( 1995) used a sample of Chapter 11 companies. The rationale for the company size of 0 to I 0 employees is the fact that a very small percentage of businesses start with more than 6 employees (Cooper et al. 1990). The model was designed for young firms up to ten years old with up to ten employees. The failure sample was generated from the bankruptcy court records. The questionnaire was first mailed to each owner/CEO filing Chapter 11 during the most recent year. The questionnaire was then mailed to each failed respondent's successful company match. Matching was selectively based on size (number of employees O - I 0), age (all firms are I 0 years old or less), location (same state and city, or city close by), and industry (same Dun + Bradstreet classification) to ensure relevant comparisons. Each Chapter 11 company owner was asked to identify the company's major competitors. Estimating a 50% response rate from successful company matches, two competitors were selected and sent a questionnaire as the failed company's match. All 45 failed firms have at least one match. Six additional successful companies data were also included as a second match. RELIABILITY AND VALIDITY TEST OF THE SAMPLE Reliability of the Questionnaire The questionnaire was carefullydcvcloped through four pretests to increase reliability of the sample measurement instrument. One of the major concerns of the study was response rate. The questionnaire'slength was limited to increase the response rate; therefore, a trade-off was made. Rather than having several repeat questions, the questionnaire used one open-ended question to check reliability. There was only one nonreliable response (.005%); therefore, reliability is inferred. Chapter 11 firms are early representativesof closed businesses. Wood ( 1990) reported less than 5% of Chapter I I companies survive whereas Flynn ( 1989) reported a I 0 to 12% survival rate. In this study 19% of the respondents were in Chapter 7 proceedings. The responses of the Chapter 7 companies were compared to the Chapter 11 companies to determine differences. Of the questions testing the model, none were significantly different (p<.05). Therefore, I-testing implies that the dominantly Chapter 11 sample is a valid representation of failures. Sample Representation To ensure that the sample represents the population, a comparison was made of the sample failure frequency distributions to the failure population by state and industry. The population figures include Chapter 7 and Chapter 11 failures (Dun+ Bradstreet 1993). Using the chi-square test, there is no significant difference (p<.05). In other words, businesses in all six states from all industries are represented by about the same percentage in the sample as in the population which they represent. Percentage representation of the sample was listed above. 27 Nonresoonse Bias Nonresponse bias was minimized in this study by including initial nonrespondents in the sample, and by comparing statistically the initial nonrespondents' data to that of the initial respondents to ensure that there is no significantdifference.Approximately I 0% of the sample includes initial nonrespondents. Of the questions testing the model, no responses are significantly different (p<.05). The t-test and chi-square test results imply that the sample is not problematic due to nonresponse bias. In addition, the combined response rate of 43% is relatively high for failure studies. RESULTS AND DISCUSSION OF MODEL TESTING Testing the Literature Model to Pevelop the Startuo Business Success versus Failure Prediction Model To test the 15 independent variable literature model, stepwise discriminant analysis was run in order to predict the success or failure of 96 businesses. Because some of the variables are not significant, they are dropped from the Literature Model to form the Business Success Versus Failure Prediction Model. Below is a list of the variables retained and those excluded from the model after step I 0 of stepwise discriminant analysis. The variables are in rank order by ability to discriminate between success and failure. The Success versus Failure Prediction Model for Entrepreneurs with 0-10 Employees S/F = f (professional advisors, planning, education, minority business ownership, staffing, parents owned a business, record keeping and financial control, capital, industry experience, economic timing) The following variables were excluded from the model: management experience, product/service timing, age, partners, and marketing. Ten variables are kept in the model, and five are dropped from the model. For the standardized canonical discriminant function coefficients, classification results, and functional statistics of the model see Table 4. The standardizedcoefficients(Beta) are similar to the coefficients in regression. The standardizedcoefficientsare adjusted for unequal means and standard deviation. The absolute values of the standardized coefficients provide information concerning the relative importance of the predictor variables in the function. Therefore, they are presented in rank order. However, unlike regression, it is inappropriate to interpret the signs of the coefficients (Scherr 1989). The most appropriate evaluation statistic of the model is not the coefficients because they are rank ordered, and the signs should not be interpreted. Therefore classification accuracy,canonical correlation, and the significance of the model are better test of the validity and reliability of the model. 28 Table 4 Discriminant FUNCTION Analysis Results Model Standardized Canonical Discriminant Function Coefficients Variable Name (Ranked by coefficient discriminant ability) 6. Professional advisors 5. Planning 7. Education 14. Minority 8. Staffing 13. Parents owned business 2. Record keeping and financial control I. Capital 3. Industry experience I 0. Economic timing ••• p < .001 Classification Table Predicted Actual Group Failure Success Overall Grouped Cases Classified Discriminant Function Statistics Canonical correlation Eigenvalue Wilks' lambda Chi-square Model significance level Failure 67% (30) 18% (9) 29 Group Success 33% (15) 82% (42) .54 .42 .70 30.89 __.QQQQ Standardized Coefficients -.15••• .65*** .46*** -.37*** .3 t ••• .29••• -.29••• -.23••• .21••• -.22••• Percentage Correct 67.7% 82.4% 75.0% VaHdity and Reliability of the Model The research results imply content validity, empirical validity.construct vaHdity, and reliability. To address content vaHdity, this study based all questions on prior research. The model includes 15 variables stated in the Hterature as being factors contributing to success versus failure. There is no generally accepted list of variables/factors contributing to success or failure. Therefore, a criterion is estabHshedconcurrentlywith the model. To help ensure the reHability of the model, stepwise discriminant analysis was run to eHminate collinear variables, and those with little discriminatory power. It can be inferred that multicollinearity is not problematic. Empirical validity and reHability were tested three ways: I. Predictive ability of the model. The model predicted 67.7% of the failed sample, and 82.4% of the successful firms correctly for an overall accuracy rate of 75%. See Classification Table section of Table 4. If random guessing produces a 50% correct classification, then the model is 25% more reliable at classifying a specific business as successful or failed. 2. Ability of the 15 independent variables to explain the variance in successor failure The canonical correlation [simply the Pearson correlation between Y and X, similar to R in regression] was .54 indicating a correlation between predicted and observed group membership. This is reflective of the overall 75% correct classification. 3. Significanceofthe model. The chi-squareof30.89 was used to test the significance level of the model (p=.0006) [comparable to the F test in regression]. In other words, the model will reliably predict a group of businesses as failed or successful more accurately than random guessing over 99% of the time. Two other discriminant function statistics shown in Table 4 include: Eigenvalue .42 [the ratio of the between-groups to within-groups variabiHty] and Wilks lambda .70 [the ratio of within-group variability to the total variability). IMPLICATIONS Use and Limitations of the Model There are at least seven groups that can benefit from using the prediction model developed in this study: entrepreneurs, investors, lenders, suppliers, educators, consultants, and pubHc poHcy makers. Entrepreneurs can assess the probability of success or failure before starting a business. Investors and lenders may be able to avoid conducting business with firms with a high probability of failure. Suppliers may decline credit or limit it to high risk businesses. Educators and consultants can make these other groups aware of the model and help them to use it. Public policy makers can use the model when selecting assistance either to prevent failure or to promote further growth. When the seven groups use the model, it is important to reaHze that the model is not intended to replace existing default risk prediction techniques. It should be used in conjuncticn 30 with the present techniques used by investors, lenders, creditors, and other groups to avoid default. The model does not provide numerical guidelines for variables distinguishing success from failure. Judgment is needed when applying the model. For example, the variable "planning" states that people starting firms with a specific business plan have a greater chance of success than people without a specific plan. However, the model neither defines "specific" nor gives a sample business plan needed to be successful. The user must look at the list of variables in the model and subjectively assess the strengths of the business for each variable. Use judgment when a business is strong on some variables and weak on others. When judgment is needed, the other decision criteria previously used become increasingly important in avoiding default. Further Research Some unanswered questions require further research. Why are there inconsistencies within the literature and discrepancies between the literature and this study? Almost all of the major variables identified in the literature as factors contributing to success or failure have been rejected by one or more other existing studies; see Table 2. Management experience constitutes one major area of discrepancy between the literature and this study. Fourteen of the twenty articles support this variable as a distinguishing factor between success or failure. However, management experience was not a significant variable in this perdition model. In this study ten of the 15 variables were significant predictors of success or failure (p<.00 I). Of the 20 journal articles from which the variables were adapted, only Cooper et al. (1990 + 1991 ), and Reynolds and Miller ( 1987 + 1989) developed nonfinancial empirically tested predictive models. They each developed two models based on the same data set. The following bullet list compares each of the 15 variables, Tables I and 2, for these four models and this study's model (prediction model for 0-10 employees] to show discrepancies among studies. I. Capita/is the only tested significant variable in their four models and this 0-10 model. 2. Record keeping and financial control was a significant variable in the Reynolds models and this 0-10 model, and it was not a tested variable in the Cooper models. 3. Industryexperiencewas a significant variable in the Cooper 1991 model and this 0-10 employee model, but it was not significant in the Cooper 1990 model, and it was not a tested variable in the Reynolds models. 4. Management experience was not a significant variable in the Cooper models or this 0-10 model, and it was not tested with the Reynolds models. 5. Planning was a significant variable in Cooper 1990, both Reynolds models, and this 0-10 model, and it was not a tested variable in the Cooper 1991 model. 6. Professional advisors was significant in the Cooper models and this 0-10 model, and it was not a tested variable in the Reynolds models. 31 7. Education.vas significant in the Cooper 1991 model and this 0-10 model, but it was not significant in the Cooper 1990 model or the Reynolds and Miller 1989 model, and it was not a tested variable in the Reynolds 1987 model. 8. Staffinb"'as a significant variable in this 0-10 model, but it was not significant in the Reynolds models, and it was not a tested variable in the Cooper models. 9. Productlsl'TVice timing was a significant variable in the Cooper 1990 model and the Reynolds models, but not in the Cooper 1991 model or this 0-10 model. I 0. Economic timing was a significant variable in the Cooper 1990 modei and this 0-10 model, but not in the Cooper 1991 model, and the Reynolds models did not test it as a variable. 11. Age of owner was a significant variable in the Cooper 1990 model; but it was not significant in Cooper 1991, Reynolds and Miller 1989, or this 0-10 model; and it was not tested in the Reynolds 1987 model. 12. Having partners was a significant variable in the Cooper 1990 and Reynolds and Miller 1989 model, but not the Cooper 1991 or this 0-10 model, and the Reynolds 1987 model did not test this variable. 13. Parents owned a business was a significant variable in the Cooper 1991 model and this 0-10 model, and it was not tested by the Reynolds models or Cooper 1990 model. 14. Minority ownership was a significant variable in both Cooper models and this 0-10 model, and it was not a tested variable in either Reynolds models. 15. Marketing skills was not a significant variable in any of the models, it was rejected as significant in the Reynolds 1987 model and this 0-10 model, and the Cooper models and Reynolds and Miller 1989 model did not test it. To summarize, the only variable tested and accepted by all five models is capital. The variables tested by at least three models and not rejected by any as significant include: planning, record keeping and financial control, professional advisors, and minority ownership. Variables both supported and rejected as significant include: industry experience, education, staffing, product/service timing, economic timing, age of owner, and partners. The variables not supported as significant by any of the five models, and rejected by two or three include: marketing skills and management experience. With these discrepancies among studies, further predictive model testing is needed to provide reliability and validity. Conclusjon The purpose of this study has been achieved. The generic young nonfinancial business success versus failure prediction model for businesses with 0-10 employees developed in this study will reliably outperform random classification of a group of businesses as successful or 32 failed over 99% of the time (model p=.0006). The model will accurately predict the success or failure of a specific business 75% of the time. The ten variables in the model explained 54% of the variance of factors contributing to success or failure. 33 REFERENCES Altman, E. (1983), "Why Businesses Fail," Journal of Business Strategy (Spring), 15-21. Alves, J. ( 1978), "The Prediction of Small Business Failure. Utilizing Financial and Nonfinancial Data," Dissertation University of Massachusetts. Ballantin, J.W., F.W. Cleveland, and C.T. Koeller (1992), "Characterizing Profitable and Unprofitable Strategies in Small and Large Businesses," Journal of Small Business Management 30, 13-24. Barsley, G., and B.H. Kleiner ( 1990), "Small Business Management: Ensuring Your Client's Success," National Public Accountant 35, 30-33. Bruno, A., J. Leidecker, and J. Harder (1987), "Why Firms Fail," Business Horizons (March/April), 50-58. Cameron, K.S., M.V. Kim, and D.A. Whetten (1987), "Organizational Effects of Decline and · Turbulence," Administrative Quarterly 32, 222-240. Cooper, A., W. Dunkelberg, C. Woo, and W. Dennis (1990), New Business jn America: the firms+ Their Owners. Washington D.C.: The NFIB Foundation. Cooper, A., J. Gascon, and C. Woo (1991), "A Resource-Based Prediction ofNew Venture Survival and Growth," Proceedings Academy of Management (Summer). Corman, J, and R.N. Lussier (1991), "Reasons Businesses Fail in New England: A Survey Study," Business Journal 8 (1), 21-27. Crawford, G. ( 1974), "An Analysis of Management Factors that may Contribute to the Success or Failure of Selected Small Retailing," Dissertation University of Arkansas. D'Aveni, R. (1989), "The Aftermath of Organizational Decline: A Longitudinal Study of the Strategic and Managerial Characteristics of Declining Firms," Academy of Management Journal (September), 577-605. Dugan, M. and C. Zavgren (1989), "How a Bankruptcy Model could be Incorporated as an Analytical Procedure," CPA Journal (May), 64-65. Dun+ Bradstreet (1994), Business Failure Record. New York: Dun+ Bradstreet. El-Zayaty,A. ( 1986), "Business Failure Prediction: An Analysis ofType II Prediction Errors," Dissertation City University of New York. Flahvin, A. ( 1985). "Why Small Businesses Fail," Australian Accountant (October), 17-20. Flynn, E. (1989), Statistical Analysis of Chapter I !. Administrative Office of the United States Courts. Gaskill, L.R., H.E. Van Auken, and R.A. Manning (1993), "A Factor Analytic Study of the Perceived Causes of Small Business Failure," Journal of Small Business Management 31, 18-31. Gilbert, L, K. Menon, and K, Schwartz(l 990), "Predicting Bankruptcy for Firms in Financial Distress," Journal of Business Finance +Accounting (Spring), 161-171. Hoad, W. and P. Rosco (1964), Management Factors Contributing to the Success or Failure of New Small Manufactures. Ann Arbor: University of Michigan Press. Kennedy, C. (1985), "Thinking of Opening Your Own Business? Be Prepared!" Business Horizons. (September/October), 38-42. Koh, H. and L. Killough (1990), "The Use of Multiple Discriminant Analysis in the Assessment of the Going Concern Status of an Audit Client," Journal of Business Finance+ Accounting (Spring), 179-192. 34 . • • • ,, !' I.., · 'l- ,·I •·' ' . Lauzen, L. (1985), "Small Business Failures are Controllable," Comorate Accounting (Summer), 34-38. Lussier, R.N. (1995a). "A Nonfinancial Success Versus Failure Prediction Model for Young Firms," Journal of Small Business Management, 33 (January), 8-20. Lussier, R.N. (1995b). "Starup Business Advise from Business Owners to Would-Be Entrepreneurs," SAM Advanced Management Journal, 60 (Winter), I 0-13. Lussier, R.N. and J. Corman, "There are Few Differences Between Successful and Failed Small Businesses," Journal of Small Business Strategy. 6 (Spring), 21-33. McQueen, J. (1989), "The Causes and Lessons of Business Failure," Credit Management (October), 24-25. Pech, R.J., and M. Alistair(I 993), "Critical Factors for Consulting to Small Business," Journal of Management Consulting 7, 61-63. Reynolds, P. (1987), "New Firms: Societal Contribution versus Potential," Journal of Business Venturing (Summer), 231-246. Reynolds, P. and B. Miller (1989), "New Firm Survival: Analysis ofa Panel's Fourth Year," Ninth Annual Babson College Research Conference. Sage, G. (1993), "Entrepreneurship as an Economic Development Strategy," Economic Development Review 11, 66-67. Scherr, F.C. (1989), "Causality, Regression, Discriminant Analysis, and Research on Failure," Akron Busjness and Economic Review (Spring), 8-20. Shelton, M. (1986), "Prediction of Corporate Failure with an Expanded Variable Set Using Multiple Discriminant Analysis," Dissertation University of Houston. Sommers, W., and A. Koc (1987), "Why Most New Ventures Fail (and how others don't)," Management Review (September), 35-39. Stockton, R. (1989), "Symptoms of an Ailing Business, "Financial Manager (January/February), 14-21. Storey, D., K. Keasey, R. Watson and P. Wynarczyk (1987), Performance of Small Firms: Profits. Jobs. and Failure. London: Croom Helm. Thompson, R. (1988), "Business Plans: Myth and Reality,"Nation's Business(August), 16-23. Vesper, K. (1990), New Venture Strategies. Englewood Cliffs, NJ: Prentice Hall. Wight, C. (1985), "Business Failures: Early Diagnosis and Remedies," Australian Accountant (September), 30-39. Wood, D.L. (1989), "Why New Businesses Fail and how to Avoid Disaster," Cornorate Cashflow (August), 26-27. Wood, D.L. (1990), "Credit Technique: Who Gets Hurt by Bankruptcy?" Business Credit 92, 44. 35