niif para pymes

Journalof BANKING & ELSEVIER

Journal of Banking & Finance 20 (1996) 151-164

FINANCE

Wealth effects of asset securitization Larry J. Lockwood a, * Ronald C. Rutherford b Martin J. Herrera c a Finance and Decision Science Department, Texas Christian University, P.O. Box 32868, Fort Worth, TX 76129, USA Department of Finance and Real Estate, The University of Texas at Arlington, Arlington, TX 76019, USA c Colonia Jardines Universidad, Zapopan, Jalisco 45l 10, Mexico

Received 15 October 1993; accepted 15 August 1994

Abstract

This paper examines changes in wealth for finns that securitize assets. Findings are industry specific with wealth increase for finance companies, with no wealth change for industrial companies and automobile companies, and with wealth loss for banks. Further examination, however, reveals that excess returns for banks are significantly related to financial slack in the quarter preceding the securitization announcement. Findings indicate that strong banks experience wealth gain while weak banks experience wealth loss upon the announcement of asset securitization. JEL classification: G14; G21 Keywords: Asset backed security; Banks; Excess return; Finance companies; Financial slack

* Corresponding author. Phone: 817-921-7420, Fax: 817-921-7227. 0378-4266/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved SSDI 0 3 7 8 - 4 2 6 6 ( 9 4 ) 0 0 1 0 1 - 4

152

L.I. Lockwood et al. /Journal of Banking & Finance 20 (1996) 151-164

1. Introduction

The securitization of assets is a recent innovation in which principal and interest payments on loans are rebundled and sold as new securities. The origin of asset backed securities (ABS) can be traced to the mid 1970s, when the Government National Mortgage Association (GNMA) pass-throughs were developed. ABS were introduced by First Boston in 1985. By 1992, the cumulative volume of ABS reached approximately $200 billion (Asset Sales Report, 1985-1993). This paper examines several propositions related to asset securitization. First, we test the proposition that securitization leads to wealth effects for shareholders of issuing firms. There are many potential benefits to the issuer. For example, Hess and Smith (1988), Zweig (1988), Greenbaum and Thakor (1987), and Pavel and Phillis (1987) suggest that securitization provides a means to reduce risk, to diversify portfolios, and to fund new assets and operations. Donahoo and Shaffer (1991) suggest that depository institutions securitize to reduce reserve and capital requirements. Rosenthal and Ocampo (1988) suggest that securitization usually offers lower cost financing for the firm by separating the credit risk of securitized assets from the risk of the firm, regardless of the firm's credit ratings. Securitization also offers the firm expanded borrowing capacity, freeing the firm to pursue additional positive NPV projects. ABS also provides off-balancesheet financing. Companies are required to report as liabilities only that part of the ABS that they guarantee (e.g., through credit enhancement). Thus, firms are able to convert assets to cash quickly and to remove liabilities arising from assets acquired through business expansion (Bemstein, 1993, p. 213). The issuer also receives fee income for servicing the ABS (Kopff and Lent, 1990, p. 157). The cash inflow from the ABS issue can be used to retire existing debt, which, in turn, reduces interest expense, increases reported earnings, and increases stockholder equity. The financial benefits offered by securitization, therefore, suggest that there are positive wealth effects for shareholders of issuing firms. Second, we examine the proposition that wealth effects from securitization differ by industry. Securities will be classified into four broad groups using SIC codes; banks and thrift institutions, finance companies, automobile companies, and other industrial firms. Wealth effects will be compared across groups. There are reasons to expect wealth differentials by industry. In particular, Greenbaum and Thakor (1987) argue that banks may securitize their best assets and retain their poorer quality assets because banks can transfer their liabilities to the FDIC. Thus, banks have incentive to securitize their better assets because of the protection offered by the FDIC and because of underpriced FDIC insurance. The authors contend that these subsidies may increase securitization by banks and may lead to a deterioration in bank asset quality. Moreover, to be treated as an asset sale to reduce capital requirements for banks, assets must be sold strictly on a non-recourse basis. The non-recourse requirement leads to the use of a credit enhancement that may take the form of overcollateralization or securitization of best

L..I. Lockwood et al. /Journal of Banking & Finance 20 (1996) 151-164

153

quality assets. 1 Thus, according to the asset quality deterioration proposition, banks should experience wealth loss at the time of ABS announcement. Third, we test the proposition that wealth effects of the ABS differ on the basis of financial slack status of issuing firms. We hypothesize that the ABS announcement of firms with little financial slack will be viewed less favourably than the ABS announcements of firms with superior financial slack. The market may view low financial slack as indicative of an eroded capital base. Thus, the sale of assets by low slack firms may be viewed by the market as confirmation of financial distress. Alternatively, Schwarcz (1993, p. 15) explains that it is unlikely that a letter of credit will be acceptable as credit enhancement for an ABS issue from a weak bank. Thus, the bank will be required to provide costlier means of enhancement (e.g., cash collateral). ABS also may be associated with the origination of new loans to maintain a desired loan portfolio size, which may require the issuing bank to spend additional resources. 2 The bank may choose, instead, to shrink the size of its loan portfolio. Either possibility is bad news to investors. Therefore, according to the financial slack proposition, banks with low financial slack will experience wealth loss at the time of the ABS announcement. Fourth, we test the proposition that wealth effects differ on the basis of type of asset being securitized. We test for differential wealth effects of the securitization of auto loans, credit card receivables, trade and lease receivables. Last, we test the proposition that the ABS issue leads to a change in market and interest rate risk of the issuing firms. For example, Zweig (1989) argues that securitization may lead to a decrease in firm risk through reduced leverage and to earnings smoothing through a timing of receivable sales. In contrast, the potential negative effects of ABS for weak banks (asset deterioration, decrease in loan portfolio size) are likely to lead to a decrease in the quality and stability of earnings (see Bernstein and Siegel, 1979), resulting in an increase in risk. We find that the effects of securitization are industry specific. Initially, we find wealth loss for banks, no wealth change for industrial companies, and wealth gain for finance companies. Further examination, however, reveals that wealth change for banks is positively and significantly related to financial slack in the quarter preceding the announcement. After segmenting the bank sample on financial slack, we find that strong banks experience wealth gain and that weak banks experience wealth loss at the time of ABS announcement. We also find that the type of asset being securitized has no effect on excess returns. Finally, we find that securitiza-

Zweig (1989) reports that in some cases, overcollateralization of lease and trade receivables may reach 20 percent. 2 Sinkey (1992, pp. 694-695) explains that once sound banks have the necessary resources for originating loans (e.g. capital, loan officers, reputation, customer relationships, computer equipment, etc.), the marginal cost of producing new loans usually is quite low. Weak banks, however, are unlikely to possess the resources for originating new loans at low marginal cost.

154

LJ. Lockwood et al. /Journal of Banking & Finance 20 (1996) 151-164

tion is associated with marginal reductions in market risk a n d / o r interest rate risk for strong banks, finance companies, and automobile companies and with increases in risk for weak banks. The remainder of this paper consists of three sections. Data and methodology are described in the next section. Results are presented in Section 3 and concluding and summarizing remarks are made in Section 4.

2. Data and methodology A sample of 397 public offerings of securitized assets issued by 39 firms over the period January 1985 through December 1992 is obtained from the A s s e t S a l e s R e p o r t for 1985 through 1993. The report provides detailed daily information of asset-backed securities including offering date, seller, amount of the issue, collateral, and yield. To be included in the sample, each firm must: (a) have traded on the New York, American, or NASDAQ exchanges during the period 1984-1992, (b) have no missing returns during the period, and (c) have no contemporaneous announcements reported in T h e W a l l S t r e e t J o u r n a l during the event window. A final sample of 294 public offerings is obtained. 3 The sample consists of 121 ABS issues by banks, 48 by finance companies, 65 by automobile companies, and 60 by other (non-auto) industrial companies. The sample is represented by the securitization of auto loans (n = 109), credit card receivables (n = 91), and trade and lease receivables (n = 94). The size of the offerings range from $43 million to $3,050 million with an average of $533.7 million. The S & P rating on the securities ranges from A + to AAA. We proxy financial slack for the quarter preceding the ABS announcement as capital surplus plus retained earnings. Quarterly capital surplus and retained earnings data are taken from the 1993 COMPUSTAT CD-ROM database. We derive a relative firm slack variable as follows. First, we divide slack for the firm by the firm's market value (price per share times number of outstanding shares of common stock) for the quarter preceding the ABS announcement to derive a size-adjusted slack measure for the firm (si). Second, we repeat the procedttrea~-or the industry to which the firm belongs to derive a size-adjusted slack measure for the industry (Si). Industry slack is the sum of the slack variables for all firms in the industry and industry size is the sum of the market values of all firms in the industry. Third, we define the ratio R S i = s i / S i as the relative slack measure for firm i.

3 The final sample of 294 public offerings of ABS was determined by searching The Wall Street Journal and The New York Times indices for contemporaneousevents during the 21 day event window. From an initial sampleof 397 ABS offerings, 87 cases were excluded for contemporaneousannouncements. An additional 16 cases were excluded due to missing returns during the event period. Dates of the announcementswere verified using the Moody's Bond Record and The Wall Street Journal when possible.

L.J. Lockwoodet al. /Journal of Banking & Finance 20 (1996) 151-164

155

T o e x a m i n e the w e a l t h effects o f the A B S a n n o u n c e m e n t , s t a n d a r d e v e n t s t u d y m e t h o d o l o g y is e m p l o y e d (see B r o w n a n d W a r n e r , 1985). D a i l y r e t u r n s for the 2 9 4 e v e n t s are t a k e n f r o m the 1993 D a i l y C R S P ( C e n t e r for R e s e a r c h in S e c u r i t y P r i c e s ) files for the N Y S E - A M E X a n d the O T C m a r k e t s . T h e N Y S E - A M E X e q u a l l y - w e i g h t e d i n d e x m a i n t a i n e d b y C R S P is u s e d to p r o x y the m a r k e t , T h e p a r a m e t e r s o f the m a r k e t m o d e l are e s t i m a t e d u s i n g daily r e t u r n s o v e r d a y s - 111 to - 11 r e l a t i v e to the a n n o u n c e m e n t day ( t = 0),

R j t = OLj + fljRmt + ejt ,

t

=

--

111 ..... - 11

(1)

w h e r e R jr is the r e t u r n o n s e c u r i t y j for day t, R,, t is the r e t u r n o n the e q u a l l y - w e i g h t e d m a r k e t i n d e x for day t, a n d e j, is the r a n d o m d i s t u r b a n c e for s e c u r i t y j, day t. T h e f o l l o w i n g 21 d a y s ( - 10 t h r o u g h + 10) are d e s i g n a t e d as the e v e n t period. 4 T o e x a m i n e the null h y p o t h e s i s that the m e a n e x c e s s r e t u r n a c r o s s the n e v e n t s e q u a l s zero, the f o l l o w i n g test statistics for e x c e s s r e t u r n for e v e n t day t, ERpt, a n d for c u m u l a t i v e e x c e s s r e t u r n s o v e r e v e n t d a y s TI t h r o u g h T2, CERv(T1,T2), are used,

Ea ,/S(eRp,),

(2)

CER e( v l , T 2 ) / S ( CERp( T I , T 2 ) ),

(3)

ERpt= ~ ( R i , - ( ~ i + f l i R m t ) ) / n ,

(4)

where

i=l

s(eRp,) =

,/99

,

(5)

t= T2

CERp(T1,T2) = E ERot,

(6)

t=Tl

4 Peterson (1989, p. 38) states that the choice of the length of the event window is arbitrary, although previous studies and institutional factors may influence the choice. The event window ( - 10, + 10) was selected in order to insure that no other announcements (such as earnings reports, dividend announcements, etc.) were reported that might impact the share price around the announcement date. We report results for both day - 1 and day 0 because some of the announcements were made while the market was still open and were not reported in The Wall Street Journal until the following day. Smith (1986) also reports CERs over days - 1,0 in his review of capital acquisition articles. Alternative event periods were used and the results were not sensitive to the definition of the event period. For example, using an estimation period of days - 1 1 1 through - 2 , cumulative excess returns over days - 1 , 0 are as follows: 0.17% (t-statistic = 0.84) for all firms, -0.58% (t-statistic = -2.13) for banks, 2.82% (t-statistic = 2.36) for finance companies, 0.01% (t-statistic = 0.03) for auto firms, and -0.26% (t-statistic = - 1.01) for other firms. These results are essentially the same as when the event window is - 10 to + 10, as reported in Table 1.

156

L.J. Loclavood et al. /Journal of Banking & Finance 20 (1996) 151-164 S(CERp(T1,T2))=[T2-T1

+ 1]I/2S(ERpt).

(7)

Assuming that the ERpt are serially independent, identically distributed, and normal, the test statistics in (2) and (3) are distributed Student-t with 99 degrees of freedom. Notice that the portfolio approach used to compute test statistics accounts for cross-sectional correlation in the idiosyncratic excess returns, ey,. To determine the effect of securitization on the wealth of stockholders of the issuing firms, excess returns are computed and examined for the entire sample and on subsampies grouped by industry. Differential effects based on financial slack and type of ABS will be assessed using a cross-sectional OLS regression of the C E R ~ ( - 1,0), C E R i ( - 1,0) = Y0 + "Y1RSi + y2D2~ + T3D3i + ei i = 1,2,...,n

(8)

where R S i is the relative financial slack of the firm as of the quarter preceding the ABS announcement; D2i equals one if the ABS is backed by auto loans, zero otherwise, and D3i equals one if the ABS is backed by trade or lease receivables, zero otherwise; and ei is a random disturbance. Differential effects of unique characteristics of the issuers of ABS are assessed with the t-statistics on the ~/s in (8). For instance, after controlling for the type of issue, "Yl measures the sensitivity of the CER~ to the financial slack of the issuing firm. Meanwhile, Y2 measures the difference in CERs between ABS backed by auto loans versus those backed by credit cards, after controlling for slack effects of the issuing firm. Similarly, Y3 measures the difference in CERs between ABS backed by trade or lease receivables versus those backed by credit cards, after controlling for slack effects of the issuing firm. The test-statistics for the gammas in (8) are distributed Student-t with n - 4 degrees of freedom, where n is the number of ABS included in the regression. To examine the effect of securitization on the market risk and interest rate risk of issuing firms, the following regression is performed on each security,

Rit = a i + bliRmt + b 2 i I t + b 3 i R m t D t + b4iItD t + u t, t = - 1 1 1 , . . . , - 11,+ 11,...,+ 120

(9)

where Rit is the daily stock return, R,~ t is the return on a equally-weighted index, and I t is the daily interest rate variable orthogonalized with the market index; that is, I t is the residual for day t in the regression performed on days - 111 through - 11 and + 11 through + 120 of daily treasury bill rates on the market return. O t is a dummy variable equal to zero for days - 1 1 1 through - 1 1 and equal to one for days + 11 through + 120. The change during the post-event period (days + 11 through + 120) in systematic risk is measured by b3, while the change in interest rate risk is measured by b4.

L.J. Lockwood et al. /Journal of Banking & Finance 20 (1996) 151-164

157

The null hypothesis that the number of increases in risk equals the number of decreases in risk for the sample will be examined using the Fisher non-parametric sign test statistic (see Hollander and Wolfe, 1973, p. 40),

[ B - ( n / 2 ) ] / [ n / 4 ] W2,

(10)

where B is the number of observed risk increases (b3 > 0 for market risk and b4 > 0 for interest rate risk) or decreases (b3 < 0 for market risk and b4 < 0 for interest rate risk) observed in the sample, and n is the number of ABS examined. Under the null hypothesis, the test statistic tends toward a standard unit normal distribution. t-Tests also will be performed on estimates of b 3 and b4 to detect significant changes in market and interest rate risk, respectively. The t-ratios for b 3 and b4 will be assumed to follow Student-t distributions with 206 degrees of freedom. The percentage of significant changes will be compared with the percentage of rejections that could occur purely by chance. As explained by Brown and Warner (1980), the rejection rate itself should be treated as a Bernoulli trial. The upper bound for the chance rejection rate equals ot + [a(1 - a ) / n ] 1/2, where a is the significance level. Thus, support for risk change is found if the percentage of reported cases that reject the null hypothesis (of no risk change) exceeds the chance rejection rate.

3. Results Table 1 presents results on the wealth effects of securitization. Event days are defined relative to the day of announcement of the ABS in The Asset Sales Report. Portfolio excess returns, t-statistics, and percent of the ABS issues with positive excess returns for day - 1 are reported in rows 1-3, respectively. Results for day 0 and for interval ( - 1,0) follow. The number of ABS issues is presented in the last row. Results for the entire sample of 294 ABS are presented in the first column. Columns 2 - 5 present results broken down by industry; banks (n = 121), finance companies (n = 48), automobile companies (n = 65), and other (non-automobile) industrial firms (n = 60). t-Statistics for the excess returns are reported in parentheses. Results for the total sample of 294 ABS issues indicate that shareholders experience a wealth loss of 0.32% (t-statistic = - 2.21) on day - 1 and a wealth gain of 0.45% (t-statistic = 3.13) on day 0. The results, however, are not constant across industry. For example, there appears to be a significant negative reaction to ABS issued by banks. On average, shareholders of banks experience statistically significant wealth loss of 0.64% (t-statistic = - 2.23) over the period ( - 1,0). In contrast, shareholders of finance companies experience significant wealth gain of 2.79% (t-statistic = 2.34) over period ( - 1 , 0 ) . Shareholders of automobile and other industrial companies experience no significant change in wealth.

158

L.J. Lockwood et al. /Journal of Banking & Finance 20 (1996) 151-164

Table 1 Excess returns by industry for ABS announcements, 1985-1992. This table presents excess returns and t-statistics for firms issuing asset-backed securities during the 1985-1992 period. Excess returns are computed using the market model. The pre-event estimation period for each ABS event spans days - 1 1 1 through - 1 1 relative to the announcement day reported in The Asset Sales Report. Excess returns (in percent), t-statistics, and percentage of the sample with positive excess returns for day - 1 are reported in rows 1-3, respectively. Results for day 0 and for interval ( - 1, 0) follow. The t-statistics, reported in parentheses, follow student-t distributions with 99 degrees of freedom. The number of ABS per sample is reported in the last row. The first column presents results for the total sample. Columns 2 - 5 present results by industry for banks, finance companies, automobile companies, and other (non-auto) industrial companies, respectively. The final column (6. Sound Banks) presents results for the banks falling in the top two-thirds of the financial slack sample. The financial slack bank sample consists of the 68 banks for which COMPUSTAT data were available. Interval

(1) All

(2) Banks

(3) Finance

(4) Autos

(5) Other

(6) Sound banks

Day - 1 t-statistic % positive

-0.32 ( - 2.24 a) 51.0

-0.63 ( - 3.10 a) 45.5

0.25 (0.29) 58.3

-0.27 ( - 1.23) 43.7

-0.19 ( - 0.96) 43.7

-0.36 ( - 1.34) 44.4

Day 0 t-statistic % positive

0.45 (3.13 a) 53.7

- 0.01 (-0.04) 46.3

2.54 (3.01 a) 66.6

0.25 (1.16) 47.7

- 0.08 (-0.43) 47.7

0.65 (2.41 a) 71.1

Days ( - 1,0) t-statistic % positive

0.13 (0.63) 50.7

-0.64 ( - 2 . 2 3 a) 45.5

2.79 (2.34 a) 60.4

-0.01 (-0.05) 38.4

-0.28 (-0.98) 38.4

0.29 (0.75) 57.7

65

60

n

294

121

48

45

a The t-statistic is significant at the 0.05 level.

The results from Table 1 suggest that only specific industries realize wealth effects from securitization. 5, 6 Specifically, finance companies appear to benefit by using ABS as an alternative source of financing. In contrast, banks do not benefit. These initial findings are consistent with the argument of Greenbaum and Thakor (1987) that banks often securitize their best assets and retain their poorer quality assets. Further investigation, however, indicates that the wealth loss is concentrated among the weaker banks. These findings are discussed below.

5 Rosenthal and Ocampo (1988) suggest that securitization allows firms to obtain financing at lower rates that those of alternative sources. The argument is based on the separation of credit risk of the assets being securitized from that of the company. If the rating of the ABS is superior to that of the company, it is likely that more positive financing terms can be obtained. To verify this argument, for each company, we compared the rating of ABS issue to that of the last issue of unsecured long-term debt. In all cases except seven, the rating of the ABS was substantially higher than that of the firm's long-term debt. 6 For 19 events taking place during the last quarter of 1992, a post event period of 120 days is not attainable (the 1993 CRSP tape ends on Dec 31, 1992). In those cases, we use the maximum number of post-event data available. The minimum number of post announcement days is 67.

L.J. Lockwood et al. /Journal of Banking & Finance 20 (1996) 151-164

159

Table 2 Differential effects for banks and finance companies based on slack and type of ABS. This table presents the results of a regression of the 2-day CER i over days - 1 and 0 associated with ABS issues on RSi, the firm's relative capital surplus, and on 2 dummy variables that identify the type of asset being securitized: Di2 equals one for securities backed by auto loans, zero otherwise; Di3 equals one for trade and lease receivables, zero otherwise. Yl measures effects of RS of the issuing firm on its CER. Y2 measures CERi differentials between securities backed by auto loans versus credit cards, and Y3 measures CER i differentials between securities backed by trade or lease receivables versus credit cards. From the original sample of 121 banks and 48 finance companies during the period spanning January 1985 to December 1992, RS data is available for 68 banks and 31 finance companies. The t-statistics, presented in parenthesis, are distributed with 64 degrees of freedom for banks and with 27 degrees of freedom for finance companies. CERi = T0 + TIRSi + y2Di2 + "Y3Di3 + ~i i = 1,2,...,n Estimate

Banks

~/o ~ ¢/2 Y3 n

-0.0024 0.0027 -0.0112 -0.0075 68

Finance ( - 0.50) (2.11 a) ( - 1.46) ( - 1.09)

-0.0176 -0.0007 - 0.0118 0.0248 31

( - 1.21) (-0.42) (-0.41) (1.50)

a The t-statistic is significant at the 0.05 level.

Results of Eq. (8) are presented in Table 2. The regression measures differential wealth effects between the types of ABS issue after controlling for the financial slack of the issuing firm. To provide a contrast in the significant excess returns reported in Table 1, we perform the analysis on banks and finance companies alone. The sample size decreases because COMPUSTAT did not report capital surplus for all the firms in the quarter prior to the ABS issue. Findings reported in Table 2 indicate that CERs are significantly and positively related to relative slack for banks ('Yl = 0.0027, t-statistic = 2.11). Slack, however, has no effect on the CERs of finance companies (~q = -0.0007, t-statistic = -0.42). These findings support the hypothesis that the ABS announcement signals bad news for banks with little financial slack. The type of ABS has no effect on the CERs of either banks or finance companies. To examine the slack proposition further, the 68 banks from Table 2 are ranked on slack. CERs for the ( - 1,0) interval for the top third (n = 23) and bottom third (n = 23) in the sample are 0.825% (t-statistic = 2.08) and - 1 . 0 2 % (t-statistic = -2.84), respectively. These findings suggest that strong banks experience wealth gain while weak banks experience wealth loss at the time of ABS announcement. We also examine the excess returns for the 45 banks classified in the top two-thirds of the financial slack sample. These results are presented in the last column, labelled 'Sound Banks', in Table 1. The CER for the ( - 1,0) interval is 0.288% with a t-statistic equal to 0.75. The excess return and t-statistic for day 0 are 0.647% and 2.41, respectively. Thus, after deleting the weak banks, the ABS announcement day effect on the remaining bank sample is significantly positive.

160

L.J. Lockwood et al. /Journal of Banking & Finance 20 (1996) 151-164

Our findings indicate that only weak banks experience wealth loss at the time of ABS announcement. These findings are consistent with the hypotheses that weak banks must provide costlier means of credit enhancement, incur additional costs originating new loans, or experience decreases in loan portfolio sizes. Moreover, weaker banks perhaps are more likely to securitize their better assets leading to a deterioration of quality on their balance sheets. That is, the asset deterioration argument of Greenbaum and Thakor (1987) may apply to weak banks, but not to all banks. The effect of the ABS announcement on market and interest rate risk from Eq. (9) are presented in Table 3. The percentage of increases ( + ) and decreases ( - ) in market risk estimates are presented in the first row of the table. The sign test statistics are presented in the second row. The percentage of significant increases and decreases in market risk are presented in the third row. Results for changes in interest rate risk are presented in rows 4 - 6 of the table. The last row presents the upper bound for the percentage of positive and negative rejections that could occur purely by chance. Results are reported for the entire sample in the first column of the table. Results for banks, finance companies, automobile companies, and other non-auto industrial companies are presented in columns 2-5, respectively. Results for banks falling in the top two-thirds of the financial slack sample are presented in the last column. We predict that risk will decrease for all industry groups except for weak banks, which are more likely to suffer losses in asset quality, increases in future costs of loan issuance, a n d / o r loss in loan portfolio sizes, all leading to a deterioration of earnings quality and predictability. The results in Table 3 indicate that there are more increases in risk than decreases in risk for the entire sample. The results, however, differ by industry. For example, there is evidence of risk decrease for finance and automobile companies. For finance companies, change in market risk, measured by b3,^is negative for 70.8% of the sample, and change in interest rate risk, measured by b4, is negative for 62.5% of the sample. The sign test indicates that the net number of decreases in risk is statistically significant for market risk at the 0.05 level and for interest rate risk at the 0.10 level. Moreover, the percentage of significant decreases in market risk and in interest rate risk exceeds the chance rejection rate (8.33% versus 4.75% for market risk and 10.41% versus 4.75% for interest rate risk). There is similar support for market risk decrease for automobile companies, although the sign test is significant only at the 0.10 level. The percentage of significant declines in interest rate risk for automobiles also exceeds the chance rejection rate (10.41% versus 4.43%). On the other hand, the data in Table 3 indicate that risk rises after the ABS announcement for banks and other (non-auto) industrial firms. For banks, b3 > 0 for 75.2% (sign test statistic = 5.55) and b4 > 0 for 71% of the sample (sign test statistic = 4.63). Further examination, however, indicates that results differ for strong and weak banks. Results presented in the last column of Table 3 indicate that risk does not change for banks falling in the top two-thirds of the financial slack sample. Thus,

L.J. Lockwood et al. /Journal of Banking & Finance 20 (1996) 151-164

II

~ e~ ,-" +

o

~ "~

~

¢~

II

~'~

~ ._

o

~

.~

~

. . . .

._= ..~

~,~

II ~

II o

_

~

_ ~

~

.~

=.. ~

"

-.-

~ . ~

~

~ ~"'~ ~

~

=

t~ ~

~

'-= ~ ~ " ~

~

.~

~'=

~-~

~

~

~

~

~

.o '~

II

~

I'~

+~T

~

,¢--i-°

16l

162

L.J. Lockwood et al. /Journal of Banking & Finance 20 (1996) 151-164

increases in risk for banks are concentrated among the weaker banks. For example, after ranking banks on financial slack, we find that 61% of the 23 banks ranked in the top third have negative estimates of b 3 and 70% have negative estimates of b 4. The sign test statistics for 63 and 64 are - 1.04 and - 1.88, respectively. Using a 0.10 significance level, therefore, the evidence suggests that interest rate risk falls for strong banks after the ABS announcement. In contrast, for the banks ranked in the bottom third of the financial slack sample, 66% have positive b 3 estimates and 61% have positive b 4 estimates. The sign test statistics for b 3 and 64 are 1.46 and 1.04, respectively. While the sign test statistics are insignificant for both market and interest rate risk change, we find that the t-statistics for 63 are significantly positive for 4 banks and for b, for 5 banks (18% and 22% of the weak bank sample, respectively). Both percentages exceed the rate of positive rejections that could occur by chance (4.77%). These results are consistent with the earlier propositions of asset quality deterioration, rising future costs, or loan portfolio shrinkage for banks, all leading to a loss in quality and stability of earnings.

4. Summary and conclusions This paper examined the wealth effects of announcements of asset securitizations. Four major findings emerged. First, the effects of asset securitization were found to be industry specific. These initial tests indicated that finance companies realized wealth gain and banks realized wealth loss at the time of ABS announcement. Automobile and other industrial firms realized no change in wealth at the time of their ABS announcements. Second, the wealth change at the time of ABS announcement was found to be positively related to financial slack for banks. These findings indicated that banks that issue ABS at a time of capital weakness were viewed negatively by the market. To examine the slack proposition further, we compared the CERs of strong banks (high financial slack) with the CERs of weak banks (low financial slack). Strong banks experienced significant wealth gain, whereas weak banks experienced significant wealth loss. Third, we found that wealth effects of ABS were unaffected by the type of asset being securitized. There were no significant differential wealth effects across credit card, auto loans, and trade and lease receivables securitizations after controlling for the financial slack status of the issuing firms. Fourth, the evidence suggested that market and interest rate risk dropped after the ABS announcement for automobile and finance companies. The evidence also indicated that interest rate risk dropped after the ABS announcement for strong banks. In contrast, there were a significant number of increases in both market and interest rate risk after the ABS announcement for weak banks.

L.J~ Lockwood et al. /Journal of Banking & Finance 20 (1996) 151-164

163

Therefore, we conclude that the market perceives benefits accruing to finance companies and strong banks that securitize assets. For these firms, securitization may reduce the need for financing, offer fee income, and increase earnings even more in cases where securitization is used to retire existing debt. ABS also may lead to a decrease in risk through a decreased reliance on debt financing and through earnings smoothing by an appropriate timing of receivables sales. We also conclude that the market perceives costs accruing to weak banks that securitize assets. Securitization by weak banks may lead to overcollateralization, high marginal cost to originate new loans, and reduction in loan portfolio size. Weak banks also may be more likely to securitize their better assets in attempts to exploit underpriced FDIC insurance. The potential negatives also may have led to a deterioration in the quality and stability of reported earnings, increasing risk for weak banks that securitized assets.

Acknowledgements Much of the work on this paper was done while Professor Lockwood was at The University of Texas-Arlington.

References Asset Sales Report, 1985-1993, (American Banker, New York).

Bernstein, L.A., 1993, Financial statement analysis: theory, application, and interpretation, 5th edition. (Irwin, Homewood, IL). Bernstein, L.A., and J.G. Siegel, 1979, The concept of earnings quality, Financial Analysts Journal 35, 72-75. Brown, S.J. and J.B. Warner, 1985, Using daily stock returns, Journal of Financial Economics 14, 3-31. Brown, S.J. and J.B. Warner, 1980, Measuring security price information, Journal of Financial Economics 8, 205-258. Donahoo, K.K. and S. Shaffer, 1991, Capital requirements and the securitization decision, Quarterly Review of Economics and Business 4, 12-23. Greenbaum, S.I. and J.V. Thakor, 1987, Bank funding modes: securitization versus deposits, Journal of Banking and Finance 11,379-392. Hess, A. and C. Smith, 1988, Elements of mortgage securitization, Journal of Real Estate Finance and Economics 1,331-346. Hollander, M. and D.A. Wolfe, 1973, Nonparametric statistical methods, (John Wiley and Sons, New York). Kopff, G.J. and Lent J., 1990, Management challenges in the age of securitization, in J. Lederman, ed,, Handbook of asset-backed securities, (Simon and Schuster, Inc., New York) 153-176. Pavel, C.A. and D. Phillis, 1987, Why commercial banks sell loans: an empirical analysis, Economic Perspectives 11, 3-14. Peterson, P.P., 1989, Event studies: a review of issues and methodology, Quarterly Journal of Business and Economics 28, 36-66. Rosenthal, J.A. and J. Ocampo, 1988, Securitization of credit, (John Wiley and Sons, New York).

164

L~I. Lockwood et al. /Journal of Banking & Finance 20 (1996) 151-164

Schwarcz, S.L., 1993, Structured finance: a guide to the principles of asset securitization, 2rid edition, (Practicing Law Institute, New York). Sinkey, J.F., Jr., 1992, Commercial bank financial management in the financial-services industry, 4th edition, (MacMillan Publishing Company, New York). Smith, C., Jr., 1986, Investment banking and the capital acquisition process, Journal of Financial Economics 15, 3-29. Zweig, P.L., 1989, The asset securitization handbook, (Dow Jones and Company, Inc., Homewood, IL).