Intervention impact of Tax Reform Act on the business failure process.

Choudhury, Askar H.

ABSTRACT

This paper investigates the impact of the intervention of Tax Reform Act on the business failure momentum. The data covers the period January 1967 through December 1986 and divided into pre-and post-event periods for both large and small business failures. We employ intervention analysis with transfer function modeling for the full data set and maximum likelihood time-series regression on the pre- and post-event periods. After controlling for the new business formations, we find the Tax Reform Act is instrumental in extending the memory of business failure momentum and amplifying the domino effect. These results also echoed in the intervention analysis. However, the impact of the intervention of Tax Reform Act is found to be more pronounced for large businesses than for small businesses.

INTRODUCTION

Business failure is generally viewed as an exogenous factor. Overall perception is that bankruptcy is a condition created by external factors that are beyond the control of the firms. Bankruptcy Reform Act of 1978 may be viewed as one of these external factors. Subsequently (early and mid 1980's), many firms sought to avoid the bankruptcy procedure by privately resolving conflicts among themselves. Between 1980-1986, 91 of the 192 (or 47%) defaulting NYSE and ASE companies were reorganized privately (Jensen, 1999). There are numerous motivations that can be attributed to these private workouts. In addition to the 40% continuity requirement that reflects a liberalization compared to 50% rule governing taxable acquisitions; avoidance of bankruptcy costs (legal and others), loss of tax carryforwards (in case of liquidation), decrease in value of the firm due to negative market perception. Shrieves & Stevens (1979) in their paper viewed these similar factors as a rationale for private workout arrangements. Jensen (1999) argues the popularity of private workout arrangements in the early 1980s was a natural market response to the high costs and time delays imposed by the bankruptcy procedure.

The objective of this paper is to analyze the effect of the Tax Reform Act on business failure process. We hypothesize, by encouraging private workout arrangements; the Tax Reform Act of 1978 enhanced the impact of the externalities of business failures, what has been characterized as a "domino effect" (see, Campbell & Choudhury, 2002).

Our sample consists of monthly observations of the number of business failure obtained from Dun and Bradstreet Corporation. This sample covers the period of January 1967 through December 1986. After dividing the sample observations into pre- and post-event periods, we examine the intervention effect of the Tax Reform Act on the business failure momentum for both large and small firms. We control for the new business incorporations and due to the presence of autocorrelation, maximum likelihood estimation method is used. Pre-event period providing a benchmark, we find the Tax Reform Act is instrumental in extending the memory of business failure and amplifying the domino effect. This suggests that the Tax Reform Act have impacted firms to accelerate private workout process by providing economic incentive. Our results contribute to the literature by documenting the constructive externalities of business failure and associating alternative recontracting procedures with dissimilarity in business failure momentum.

Following section summarizes the related literature on business failure. In the third section we discuss our data selection and research methodology. Results of our analyses are discussed in section four and we summarize our findings in section five.

RELATED LITERATURE

Bankruptcy issues and its impact on the capital market have been studied by many researchers (Baxter, 1967; Stiglitz, 1972; Kraus & Litzenberger, 1978; Scott, 1976). One of the most continuing issues in the bankruptcy literature concerns the efficiency of corporate bankruptcy. Many scholars consider bankruptcy, particularly bankruptcy reorganization process, an inefficient method and should be eliminated (e.g. Roe, 1983; Baird, 1986; Jackson, 1986; Wruck, 1990; Bradley & Rosenzweig, 1992). In bankruptcy procedure a judge determines valuation and parcels out interests. As a result, absolute priority rule is frequently violated, and deadweight economic costs are incurred (Jackson & Scott, 1989; Wruck, 1990; Baird, 1986). White (1989) concludes, "The U.S. bankruptcy system, rather than helping the economy move toward long-run efficiency, in fact appears to delay the movement of resources to higher value uses". Bulow & Shoven (1978) perceived that Chapter 11 happens only because of disagreement between the concerned parties.

The primary criticisms of the bankruptcy procedure involve the high costs and time delays it imposes on bankrupt firms (Bradley & Rosenzweig, 1992). Altman (1984) has presented a model to estimate the expected bankruptcy costs (both direct and indirect costs) on the basis of actual profits and expected profits. For large industrial firms, Weiss (1990) found direct administrative costs, such as legal fees and court costs; averaged 2.8 percent of total asset book value at the fiscal year-end prior to bankruptcy and the average time spent in Chapter 11 was 2.5 years. For small firms, the time spent in bankruptcy procedure is shorter but the direct bankruptcy costs are proportionally higher. Campbell (1997) found closely held firms spent on average 1.3 years in Chapter 11 and direct bankruptcy costs averaged 8.5 percent of total asset book value at the start of the proceeding. Moreover, assets values usually decline dramatically while a firm is in bankruptcy procedure. In contrast, the available evidence suggests the direct costs of private workout arrangements are only about 10 percent of those in a Chapter 11 proceeding of comparable size (Gilson et al., 1990). In addition to higher direct costs, bankruptcy reorganization also imposes substantial indirect costs on the debtor firm. Indirect costs include lost sales, lost profits, the inability to obtain credit from suppliers, and lost investment opportunities (Titman, 1984). Quantifying these indirect costs is difficult; however, in many bankruptcy proceedings the indirect costs are likely to exceed the direct costs. Jensen (1999) observed that a private workout commonly takes only a few months to negotiate and costs much less than Chapter 11, views the private workout arrangement as a natural market response to inefficiency.

Market studies suggest private workout arrangements do enhance firm value relative to bankruptcy reorganizations. Pastena & Ruland (1986) provide statistical evidence that distressed firms with high ownership concentration being systematically better off if their firm's debt is restructured privately. Belker, Franks & Torous (1999) report once the result of a workout attempt is known, the returns to shareholders are greater for firms which successfully complete a workout, than for firms entering bankruptcy procedure.

Traditional view of business failure is an exogenous event brought on by certain internal and external factors (e.g. bad management and high interest rates) that have rendered the debtor unable to meet its obligations. This view ignores the interdependence among firms through their contractual relationships and the constructive externalities of the failure process, what Campbell & Choudhury (2002) termed as domino effect. Consequently, market value of competitors may depreciate and cause accelerated failure process to others (Lang & Stulz 1992). Society has an interest in understanding the domino effect and helping otherwise viable businesses survive the disruption. These ideas are based on theories that business failure is a dynamic process of several events, rather than a single (or few) static event. Moreover, the traditional view ignores differences in the failure processes of large and small firms. Hambrick & D'Aveni (1988) found large bankrupt firms showed signs of relative weakness very early, as far back as ten years before failure, and they characterize the large firm failure process as a long protracted downward spiral. On the other hand, small firm failure often found to be abrupt and catastrophic as observed by Venkataraman et al. (1990).

DATA AND RESEARCH METHODOLOGY

The sample period is a twenty year window with 240 continuous monthly data. The event date, 1978, is the date the Tax Reform Act of 1978 went into effect. The Bankruptcy Code of 1978, made major changes in bankruptcy procedure. For example, under the former Bankruptcy Act of 1938 (the Chandler Act) there were different reorganization procedures for different types of firms. Chapter 11 of the Bankruptcy Code combines Chapters X, XI, and XII of the old Bankruptcy Act into a single procedure for business reorganization. Such major changes in reorganization procedures could impact business failure process, specifically to those firms that are financially distressed. To test the intervention effect of this event on business failures, we divide our sample into two periods: the pre-event period January 1967 through December 1978 (144 monthly observations) and the post-event period January 1979 through December 1986 (96 monthly observations). Since prior research has indicated the failure processes of large and small firms differ, we analyze large and small firms separately.

Table 1 presents summary statistics for the pre- and post-event periods. A "failure" is defined as, "a concern that is involved in a court proceeding or voluntary action that is likely to end in a loss to creditors" (Dun and Bradstreet's measures of failures). All industrial and commercial enterprises petitioned into the Federal Bankruptcy Courts are included as business failures. Also included are: 1) concerns forced out of business through actions in the state courts such as foreclosures, executions, and attachments with insufficient assets to cover all claims; 2) concerns involved in court actions such as receiverships, reorganizations, or arrangements; 3) voluntary discontinuations with a known loss to creditors; and 4) voluntary out of court compromises with creditors. In other words, the number of business failures is broadly defined to include private workout arrangements, state court actions, and federal bankruptcy proceedings. A small business is defined as a concern having less than $100,000 in current liabilities; a large business is defined as a concern having more than $100,000 in current liabilities. Current liabilities include all accounts and notes payable, whether secured or unsecured, known to be held by banks, officers, affiliated companies, suppliers, or the Government.

Table 1 shows the average number of small business failures rose dramatically over the twenty years study period. From January 1967 through December 1978, the pre-event period, small business failures averaged 580 per month, while from January 1979 through December 1986, the post-event period small business failures averaged 1298 per month. The average number of large business failures also rose over the two periods: for the pre-event period the number of large business failures averaged 231 per month, while for the post-event period the number of large business failures averaged 1444 per month. Finally, Table 1 also presents the summary statistics for the number of new business incorporations. For the pre-event period the number of new business incorporations averaged 26,446 per month; for the post-event period the number of new business incorporations averaged 49,905 per month.

We hypothesize that the intervention impact of the Tax Reform Act resulted in an elevated change in business failure momentum. To test our hypothesis we perform two separate analyses. First, we perform an intervention analysis for the event period using transfer function modeling to observe the direction of the effect of the Tax Reform Act and its magnitude. If there is a significant impact of the Tax Reform Act on business failure, and the Tax Reform Act enhances the constructive externalities of business failure process then the coefficient of the indicator variable (TaxLaw_78) should be large and positive. Second, we use time-series regression to examine the magnitude and trend of business failures over the pre- and post-event periods to observe the acceleration/deceleration of the momentum of the process. Specifically, we regress the number of business failures on a proxy for business failure momentum in both the pre- and post-event periods. The proxy variable, MOMENTUM, is a constant growth series beginning at one and growing by one each month. If the Tax Reform Act contributes to boost business failure momentum, then the coefficient for MOMENTUM should be larger in magnitude and positive in the post-event period compared to pre-event period.

In an effort to better disentangle the effects of business failure momentum from expanding business activity, regression model includes a control variable measuring the number of new business incorporations. Additionally, Durbin-Watson statistic on ordinary least squares (OLS) estimates indicated the presence of positive autocorrelation. One major consequence of autocorrelated errors (or residuals) when applying ordinary least squares is the formula variance [[sigma].sup.2] of the (X'[X.sup.-1]) of the OLS estimator is seriously underestimated (see Choudhury, 1994), which affects statistical inference. Where X represents the matrix of independent variables and [[sigma.sup.2] is the error variance.

Durbin-Watson statistic is not valid for error processes other than the first order (see Harvey, 1981; pp. 209-210) process. Therefore, we evaluated the autocorrelation function (ACF) and partial autocorrelation function (PACF) of the OLS regression residuals using SAS procedure PROC ARIMA (see SAS/ETS User's Guide, 1993). This allowed the observance of the degree of autocorrelation and the identification of the order of the model that sufficiently described the autocorrelation. After evaluating the ACF and PACF, the residuals model was identified as second order autoregressive model (1-[[phi].sub.1]B-[[phi].sub.2][B.sup.2]) [v.sub.t] = [[epsilon.sub.t] (see Box, Jenkins, & Reinsel, 1994). The final specification of the regression model is of the following form for large and small firm failures:

LGFAI[L.sub.t] = [[beta.sub.0] + [[beta].sub.1]MOMENTU[M.sub.t] + [[beta.sub.2]NEWBU[S.sub.t] + [v.sub.t] (1) and [v.sub.t] = [[phi].sub.1][v.sub.t-1] + [[phi].sub.2][v.sub.t-2] + [[epsilon].sub.t]

SMFAI[L.sub.t] = [[beta.sub.0] + [[beta.sub.1]MOMENTU[M.sub.t] + [[beta.sub.2]NEWBU[S.sub.t] + [v.sub.t] (2) and [v.sub.t] = [[phi.sub.1][v.sub.t-1] + [[phi].sub.2][v.sub.t-2] + [[epsilon].sub.t]

Where: MOMENTUM = a series starting at 1 and growing at a constant amount B=1 each time period; NEWBUS = the number of new business formations.

Maximum likelihood estimation method was used instead of two step generalized least squares to estimate the regression parameters in equations (1) and (2). Maximum likelihood estimation is preferable over two step generalized least squares, because of its capability to estimate both regression and autoregressive parameters simultaneously. Moreover, maximum likelihood estimation accounts for the determinant of the variance-covariance matrix in its objective function (likelihood function). In general, the likelihood function of a regression model with autocorrelated errors has the following form:

L([beta][theta][[sigma].sup.2]) = - n/2ln ([[sigma].sup.2])-1/2ln|[OMEGA]|-(Y - X[beta]'[[OMEGA].sup.1](Y - X[beta])/2[[sigma].sup.2]) (3)

where,

Y- vector of response variable (number of failures),

X - matrix of independent variables (MOMENTUM, NEWBUS, and Intercept),

[beta] - vector of regression parameters,

[theta] - vector of autoregressive parameters,

[[sigma].sup.2] - error variance,

[OMEGA] - variance-covariance matrix of autocorrelated regression errors.

For further discussion on different estimation methods and the likelihood function, see Choudhury et al. (1999); also see SAS/ETS User's Guide, 1993 for expressions of the likelihood function.

To estimate the direction of the effects and magnitude of the Tax Reform Act, intervention model is employed (see Box & Tiao, 1975). There are two common types of deterministic input variables that have been found useful to represent the impact of intervention events on a time series data. Both of these are indicator variables taking only 1 and 0 to indicate the occurrence and nonoccurrence of intervention. For our analysis, we use step function rather than pulse function, which is given as,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4).

The final specification of the intervention model that we have found for our analysis is of the following form for large and small firm failures:

LGFAI[L.sub.t] = [mu] + [[omega].sub.1][S.sup.1978.sub.t] + (1-[[theta].sub.1]B-[[theta].sub.2][B.sup.2]) (1-THETA][B.sup.12])[[epsilon].sub.t] (5)

SMFAI[L.sub.t] = [mu] + [[omega].sub.1][S.sup.1978.sub.t] + (1-[[theta].sub.1]B-[[theta].sub.2][B.sup.2]) (1-[THETA][B.sup.12])[[epsilon].sub.t] (6)

where [[theta].sub.1] and [[theta].sub.2] are regular moving average parameters and [THETA] is denoted for seasonal (monthly)moving average parameter. Maximum likelihood estimation is used to estimate these intervention models.

EMPIRICAL RESULTS

We report the results of our empirical analysis investigating the intervention effect of the Tax Reform Act of 1978 on business failures. First, we test the intervention effect using transfer function model. Intervention analysis of the event study has been reported in Table 2 for the period of January 1967 to December 1986 using step function. The estimated coefficient of the intervention indicator variable (TaxLaw_78) for the pre- and post-event period is statistically significant and positive for both large and small businesses. The magnitude of the estimated coefficient is substantial for both large and small business failures. However, the extent of the estimated coefficient is greater for the large firms compared to small firms.

This leads us to test the business failure momentum on a separate regression in order to gain insight into the force and its magnitude behind the domino effect. Campbell & Choudhury (2002) found the cumulative lagged effects of past business failures are significantly correlated with current business failures. These cumulative lagged effects usually have long memory characteristics. Choudhury & Campbell (2004) found that on average they stay statistically significant for about 24 months.

Table 3 reports the regression results for the January 1967 through December 1978 pre-event period. The estimated coefficient for business failure momentum (MOMENTUM) is statistically significant for both large and small businesses but positive for large firms and negative for small firms. However, the magnitude of the estimated coefficient is small for both large and small business failures. The control variable for new business formations, NEWBUS, is not significant.

In contrast, the regression results reported in Table 4 for the post-event period, January 1979 through December 1986, show the estimated coefficient for business failure momentum (MOMENTUM) is statistically significant for both large and small firms. Moreover, the magnitude of the estimated coefficient is large for both firms. Thus, for large businesses, if time is increased by one month (i.e., one month into the future), the number of business failures increases by 25 firms. Similarly, for small businesses, if time is increased by one month, the number of business failures increases by 30 firms.

In light of the previous results presented in Table 3, the Table 4 results suggest the Tax Reform Act has accelerated the domino effect by escalating the momentum of business failures. The estimated coefficients for the control variable NEWBUS are not significant either for the large firm or the small firm regressions. Overall, these results suggest the amplification of business failure momentum is a consequence of the Tax Reform Act of 1978. The impact is more pronounced for large businesses than for small businesses; however, in both cases the effect is clearly visible.

SUMMARY AND CONCLUSIONS

After controlling for increases in new business formations, we find strong evidence that the Tax Reform Act of 1978 is associated with expansion of the memory for business failure process and thereby strengthening the domino effect of business failure momentum. Intervention analysis of the event study also confirms the similar outcome. These results suggest the initiation of Tax Reform Act may have provided many firms with economic incentive to private workout arrangement rather than to attempt to restructure under the bankruptcy procedure.

Results of this study are consistent with the hypothesis that uncertainty in policy implementation combined with inefficient bankruptcy procedure generates a natural market response to private workout arrangement. Business failure by definition implies that firms are economically inefficient to continue to operate in the same form and this tax reform event probably enhanced their financial efficiency by accelerating the conversion of their resources into more efficient utilization. These findings contribute to the literature by documenting the constructive externalities of business failures and its association with business failure momentum.

REFERENCES

Altman, E. (1984). A Further Empirical Investigation of the Bankruptcy Cost Question. Journal of Finance, 39(4), 1067-1089.

Baird, D. (1986). The uneasy case for corporate reorganization. Journal of Legal Studies, 15, 127-147.

Baird, D. & R. Picker (1991). A simple noncooperative bargaining model of corporate reorganizations. Journal of Legal Studies, 20, 311-349.

Baxter, N. (1967). Leverage, Risk of Ruin and the Cost of Capital. Journal of Finance, 22, 395-403.

Belker, B., J. Franks, & W. Torous (1999). Are stockholders better off when debt is restructured privately. In E.I. Altman (Ed.), Bankruptcy and Distressed Restructurings: Analytical Issues and Investment Opportunities, (pp. 391-400). Washington, D.C.: Beard Books.

Box, G., G. Jenkins & G. Reinsel (1994). Time Series Analysis: Forecasting and Control. Englewood Cliffs: Prentice-Hall.

Bradley, M. & M. Rosenzweig (1992). The untenable case for Chapter 11. Yale Law Journal, 101, 1043-1095.

Bullow, J. & J. Shoven (1978). The Bankruptcy Decision. The Bell Journal of Economics, (Autumn), 437-456.

Campbell, S. (1997). An Investigation of the Direct Cost of Bankruptcy Reorganization for Closely Held Firms. Journal of Small Business Management, 35(3), 21-29.

Campbell, S. & A. Choudhury (2002). An empirical analysis of the business failure process for large and small firms. Academy of Entrepreneurship Journal, 8, 107-120.

Choudhury, A. (1994). Untransformed first observation problem in regression model with moving average process. Communications in Statistics: Theory and Methods, 23(10), 2927-2937.

Choudhury, A., R. Hubata & R. St. Louis (1999). Understanding Time-Series Regression Estimators. The American Statistician, 53(4), 342-348.

Choudhury, A. & S. Campbell (2004). The Effects of the Tax Reform Act of 1986 on Business Failure Momentum. Academy of Accounting and Financial Studies Journal, 8(1), 77-88.

Dunn and Bradstreet. (1998). Business Failure Record. New York, NY: Dunn and Bradstreet.

Gilson, S., J. Kose, & L. Lang (1990). Troubled debt restructurings: An empirical study of private reorganization of companies in default. Journal of Financial Economics, 27, 315-354.

Hambrick, D. & R. D'Aveni. (1988). Large corporate failure as downward spirals. Administrative Science Quarterly, 33, 1-23.

Harvey, A.C. (1981). The Econometric Analysis of Time Series. London: Philip Allan.

Jackson, T. (1986). The Logic and Limits of Bankruptcy Law. Cambridge, MA. Harvard University Press.

Jensen, M. (1999). Corporate control and the politics of finance. In E.I. Altman (Ed.), Bankruptcy and Distressed Restructurings: Analytical Issues and Investment Opportunities, (pp. 3-43). Washington, D.C.: Beard Books.

Kraus, A. & R. Litzenberger (1978). A state-preference model of optimal financial leverage. Journal of Finance, 33, 911-922.

Lang, L.H.P. & R.M. Stulz (1992). Contagion and Competitive Intra-Industry Effects of Bankruptcy Announcements. Journal of Financial Economics, 32, 45-60.

Pastena, V. & W. Ruland (1986). The Merger / Bankruptcy Alternative. The Accounting Review, 61(2), 288-301.

Roe, M. (1983). Bankruptcy and debt: A new model for corporate reorganization. Columbia Law Review, 83, 528-602.

SAS/ETS User's Guide. 1993. SAS Institute, Inc, Cary, North Carolina.

Scott, J. (1976). A theory of optimal capital structures. The Bell Journal of Economics and Management Science, 33-54.

Shrieves, R.E. & D.L. Stevens (1979). Bankruptcy Avoidance as a Motive for Merger. Journal of Financial and Quantitative Analysis, 14(3), 501-515.

Stiglitz, J. (1972). Some Aspects of the Pure Theory of Corporate Finance: Bankruptcies and Takeovers. The Bell Journal of Economics and Management, (Autumn), 458-482.

Titman, S. (1984). The effect of capital structure on a firm's liquidation decision. Journal of Financial Economics, 13, 137-152.

Venkataraman, S., A. Van de Ven, J. Buckeye & R. Hudson (1990). Starting up in a turbulent environment: A process model of failure among firms with high customer dependence. Journal of Business Venturing, 5, 277-295

Weiss, L. (1990). Bankruptcy resolution: Direct costs and violation of priority of claims. Journal of Financial Economics, 27, 285-314.

White, M. (1989). The corporate bankruptcy decision. Journal of Economic Perspectives, 3, 129-151.

Wruck, K. (1990). Financial distress, reorganization, and organizational efficiency. Journal of Financial Economics, 27(2), 419-444.

Askar H. Choudhury, Illinois State University

Table 1: Summary Statistics for Large and Small Firm Failures for the
Periods January 1967-December 1978 and January 1979-December 1986
(Monthly Data) (a)

 Monthly Standard
Variables (b) Period 19-- Means Deviations

SMFAIL 67-78 579.56 147.09
 79-86 1297.15 922.04
LGFAIL 67-78 230.59 60.88
 79-86 1443.51 995.59
NEWBUS 67-78 26445.90 7075.41
 79-86 49905.17 5861.38

Variables (b) Period 19-- Minimums Maximums

SMFAIL 67-78 244.00 1003.00
 79-86 242.00 3952.00
LGFAIL 67-78 96.00 446.00
 79-86 254.00 4145.00
NEWBUS 67-78 2135.00 42605.00
 79-86 27234.00 68087.00

(a) Small firms have less than $100,000 in current liabilities; large
firms have more than $100,000 in current liabilities. A failure is
defined as, "a concern that is involved in a court proceeding or
voluntary action that is likely to end in a loss to creditors." Source:
Dun & Bradstreet, Inc.

(b) Variable Definitions:

SMFAIL = number of small firm failures;

LGFAIL = number of large firm failures;

NEWBUS = number of new business incorporations.

Table 2: Intervention Analysis on 1978 Tax Reform Act of Large and
Small Firm Failures for the Period January 1967-December 1986 (Monthly
Data) (a) : Maximum Likelihood Estimates.

 Large Firm Failures Small Firm Failures
Independent (corrected for (corrected for
Variables (b) autocorrelation (d)) autocorrelation (e))

Intercept 324.20 (c) 668.0243
 (4.08) *** (8.52) ***
TaxLaw_78 970.7415 639.1512
 (8.20) *** (5.67) ***
MA-1 -0.4849 -0.5422
 (-8.62) *** (-10.94) ***
MA-2 -0.5581 -0.6465
 (-9.77) *** (-12.88) ***
MA-12 -0.4663 -0.5141
 (-6.74) *** (-5.34) ***

(a) Small firms have less than $100,000 in current liabilities; large
firms have more than $100,000 in current liabilities. A failure is
defined as, "a concern that is involved in a court proceeding or
voluntary action that is likely to end in a loss to creditors." Source:
Dun & Bradstreet, Inc.

(d) Variable Definitions:

TaxLaw_78 = an indicator variable coded 0 for t [less than or equal
to] 1978 and 1 for t >1978 time period.

(c) The t-statistics reported in parenthesis are significant at ten
(*), five (**), and one (***) percent levels.

(d) The time series part of the intervention model was identified as,
[v.sub.t] = (1 - [[theta].sub.1] B - [[theta].sub.2]
[B.sup.2])(1 - [THETA] - [B.sup.12]) [[epsilon].sub.t] and then the
structural parameters and time series parameters were estimated
simultaneously using maximum likelihood estimation method in SAS.

Both regular and seasonal moving average parameters are significant
at the one (***) percent level.

(e) The time series part of the intervention model was identified as,
[v.sub.t] = (1 - [[theta].sub.1] B - [[theta].sub.2]
[B.sup.2])(1 - [THETA] - [B.sup.12]) [[epsilon].sub.t] and then the
structural parameters and time series parameters were estimated
simultaneously using maximum likelihood estimation method in SAS.

Both regular and seasonal moving average parameters are significant at
the one (***) percent level.

Table 3: Regression Results for Number of Large and Small Firm Failures
for the Period January 1967-December 1978 (Monthly Data) (a) : Maximum
Likelihood Estimates.

 Large Firm Failures Small Firm Failures
Independent (corrected for (corrected for
Variables (b) autocorrelation (d)) autocorrelation (e))

Intercept 91.1804 (c) 1110.00
 (2.70) *** (11.85)
MOMENTUM 0.8666 -2.8562
 (3.37) *** (-4.92) ***
NEWBUS -0.00025 -0.0017
 (-0.22) (-1.18)
R-Squared 0.51 0.83
Durbin-Watson 2.19 1.92

(a) Small firms have less than $100,000 in current liabilities; large
firms have more than $100,000 in current liabilities. A failure is
defined as, "a concern that is involved in a court proceeding or
voluntary action that is likely to end in a loss to creditors." Source:
Dun & Bradstreet, Inc.

(b) Variable Definitions:

MOMENTUM = a series starting at 1 and growing at a constant amount
B = 1 each time period;

NEWBUS = the number of new business formations;

(c) The t-statistics reported in parenthesis are significant at ten
(*), five (**), and one (***) percent levels.

(d) The regression residuals model was identified as,
(1 - [[phi].sub.1] B - [[phi].sub.2] [B.sup.2]) [V.sub.t] =
[[epsilon].sub.t] and the estimated first and second order
autoregressive (AR) parameters from SAS were, (1 + 0.29 B + 0.28
[B.sup.2]) [v.sub.t] = [[epsilon].sub.t].

(3.42) *** (3.50) ***

Where t-statistics for autoregressive parameters are reported in
parentheses and they are both significant at the one (***) percent
level.

(e) The regression residuals model was identified as,
(1 - [[phi].sub.1] B - [[phi].sub.2] [B.sup.2]) [v.sub.t] =
[[epsilon].sub.t] and the estimated first and second order
autoregressive (AR) parameters from SAS were, (1 + 0.64 B + 0.15
[B.sup.2]) [v.sub.t] = [[epsilon].sub.t].

(7.67) *** (1.72) *

Where t-statistics for autoregressive parameters are reported in
parentheses and they are both significant at the one (***) percent
level.

Table 4: Regression Results for Number of Large and Small Firm Failures
for the Period January 1979-December 1986 (Monthly Data) (a) : Maximum
Likelihood Estimates

 Large Firm Failures Small Firm Failures
Independent (corrected for (corrected for
Variables (b) autocorrelation (d)) autocorrelation (e))

Intercept -5728.00 (c) -7197.00
 (-2.86) *** (-5.85) ***
MOMENTUM 25.17 29.46
 (3.39) *** (6.29) ***
NEWBUS -0.0028 0.00075
 (-0.21) (0.08)
R-Squared 0.84 0.90
Durbin-Watson 1.96 2.17

(a) Small firms have less than $100,000 in current liabilities; large
firms have more than $100,000 in current liabilities. A failure is
defined as, "a concern that is involved in a court proceeding or
voluntary action that is likely to end in a loss to creditors." Source:
Dun & Bradstreet, Inc.

(b) Variable Definitions:

MOMENTUM = a series starting at 1 and growing at a constant amount
B = 1 each time period;

NEWBUS = the number of new business formations;

(c) The t-statistics reported in parenthesis are significant at ten
(*), five (**), and one (***) percent levels.

(d) The regression residuals model was identified as,
(1 - [[phi].sub.1] B - [[phi].sub.2] [B.sup.2]) [v.sub.t] =
[[epsilon].sub.t] and the estimated first and second order
autoregressive (AR) parameters from SAS were, (1 + 0.45 B + 0.37
[B.sup.2]) [v.sub.t] = [[epsilon].sub.t].

(4.58) *** (3.72) ***

Where t-statistics for autoregressive parameters are reported in
parentheses and they are both significant at the one (***) percent
level.

(e) The regression residuals model was identified as,
(1 - [[phi].sub.1] B - [[phi].sub.2] [B.sup.2]) [v.sub.t] =
[[epsilon].sub.t] and the estimated first and second order
autoregressive (AR) parameters from SAS were, (1 + 0.34 B + 0.44
[B.sup.2]) [v.sub.t] = [[epsilon].sub.t].

(3.62) *** (4.60) ***

Where t-statistics for autoregressive parameters are reported in
parentheses and they are both significant at the one (***) percent
level.