Paul Irvine

James Rosenfeld [*]

We examine the impact of selling Monthly Income Preferred Stock (MIPS) on the common share prices of the issuing firms. We find that issuing MIPS to retire preferred stock raises the value of the firm, and that government policy can significantly affect the present value of the tax savings. Using proceeds to retire bank loans negatively impacts common share value. This negative response is larger for MIPS users with lower credit ratings on their senior debt. These findings support the view that banks perform a valuable monitoring service, which, if removed, can invoke an adverse market reaction.

In October 1993, Goldman Sachs introduced the financial world to a hybrid security called monthly income preferred stock (MIPS), which is structured to give the issuer the best of both worlds. MIPS is deeply subordinated debt that is repackaged as preferred stock. [1] The rating agencies consider MIPS tantamount to preferred stock. So far, the Internal Revenue Service has allowed the issuer to deduct preferred dividends as interest for tax purposes. This means that issuing MIPS enables the firm to increase its equity base at an after-tax cost nearly equal to that of long-term debt. Because of these features, MIPS now dominates the market for traditional perpetual preferred equity, presently accounting for over 70% of all new preferred issues.

To create MIPS, a special-purpose subsidiary of the parent firm sells the preferred stock to the public and then lends the proceeds to the parent. The parent's interest payments to the subsidiary are deductible as interest and are used by the subsidiary to pay its preferred dividends to the buyers of the securities. However, the interest income received by the subsidiary is not taxable income, because it is organized as a tax-free entity.

The creation of MIPS and its related products allows us to investigate a number of theoretical issues, including the question of whether a firm's capital structure affects share value. Using MIPS we can test the plausibility of the static trade-off theory of taxes and bankruptcy and, to a lesser degree, the agency theory version, both of which predict that firms have an optimal capital structure.

Because MIPS is equity with an after-tax cost similar to that of risky debt, issuing this security and using the proceeds to retire outstanding preferred stock enables us to isolate the impact of tax savings on common share value without the confounding influence of financial distress costs. We hypothesize that redeeming straight preferred stock should lower the firm's weighted average cost of capital and raise firm value by the present value of tax savings. Further, in the absence of agency considerations, using the proceeds to retire debt should lower the firm's present value of financial distress costs. Because the firm retains the tax savings in place, using MIPS should also increase the value of the firm. To test these hypotheses, we conduct an event study on 185 public corporations that issued MIPS between October 1, 1993 and December 31, 1998.

The paper is structured as follows. Section I briefly describes the static trade-off theory of corporate financing behavior and offers predictions of the common stock response based on the firm's use of proceeds. Section II provides a description and brief history of this interesting security and its various hybrids. Section III details the sample selection process and financial characteristics of MIPS users. Sections IV and V present the results of our event-study and regression tests. The final section summarizes our key findings.

I. Static Trade-off Theory of Capital Structure Behavior

There are basically two versions of the static trade-off theory, both of which predict that firms have an optimal capital structure. The first (and traditional) version postulates that a firm's capital structure is determined by the trade-off between the tax benefits of debt and the expected costs of financial distress. A firm achieves the optimal mix of debt and equity when the marginal present value of tax savings equals the marginal present value of financial distress costs.

The second version, agency theory, specifies a trade-off between the agency costs of equity (also referred to as the agency costs of managerial discretion) and the agency costs of debt. The equity costs occur when management's objectives are not fully aligned with those of shareholders. The debt costs are caused by restrictions, such as covenants in bond indentures, that prevent management from investing in some positive net-present-value (NPV) projects. The optimal debt-to-equity ratio is one that minimizes total agency costs, which is when the marginal agency costs of the debt equal the marginal agency costs of equity. Agency theory posits that firms have an optimal capital structure even in a world of no taxes or bankruptcy.

Due to the unique characteristics of MIPS, we can examine the plausibility of these theories by focusing on the market's reaction to the issuance of this security. The market's response depends primarily on the firm's stated use of the proceeds. Since the firm's capital structure is essentially unchanged when MIPS is issued to retire straight preferred stock, this use of funds is probably the cleanest test of the traditional static trade-off theory, which predicts an increase in common stock value in response to this event due to the resulting tax savings, with no offsetting change in financial distress costs.

When proceeds are used to retire debt, the impact on firm value is less clear. According to the tax-bankruptcy trade-off theory, retiring debt with MIPS should reduce the present value of financial distress costs without sacrificing the tax savings in place, and thereby increase firm value. However, replacing debt with MIPS is also a leverage-reducing action. Prior studies have shown that this strategy usually has a negative impact on common stock prices (see Masulis, 1980 and 1983; Eckbo, 1986; and Mikkelson and Partch, 1986).

Agency theory provides us with one explanation for this phenomenon. Replacing debt with MIPS gives managers greater flexibility, because preferred dividends, unlike bond coupon payments, can be suspended without forcing bankruptcy. In short, if the marginal increase in equity agency costs exceeds the marginal decrease in debt agency costs, the expected benefits of a reduction in financial distress costs will be reduced or even nullified.

Similar and stronger arguments apply to those cases in which firms use proceeds to retire bank debt. Prior research (James, 1987) shows that shareholders respond favorably to corporate announcements of new bank lending agreements, but unfavorably to the removal of bank loans. Fama (1985) attributes an unfavorable response to the removal of inside debt, which he defines as a loan to a party with access to corporate financial information not normally available to outside lenders, such as the holders of publicly traded debt. Thus, banks perform a valuable monitoring service that benefits shareholders. If this service were removed, it might negate the advantages of using MIPS.

In those cases in which firms use proceeds for general corporate purposes, it is not clear what effect, if any, this action will have on common stock value. Linn and Pinegar (1988) find that issuing straight preferred equity for general corporate purposes has no noticeable effect on common share prices. Because MIPS gives the issuer an added tax advantage, their finding suggests that common stock prices will react positively. On the other hand, Jung, Kim, and Stulz (1996) argue that issuing common equity to raise cash will have a negative (negligible) impact on the firm's common stock with weak (strong) growth prospects. Their finding suggests that the impact on the common stock of MIPS issuers also depends on its investment opportunities.

Based on the firm's stated use of proceeds, Table I shows the predicted effect of issuing this security on the firm and its common stock. Because there are no offsetting costs, using MIPS proceeds to retire straight preferred stock should raise common stock value by (at most) the firm's present value of tax savings (PVTS).

However, the use of MIPS proceeds for general corporate purposes produces both positive and negative benefits. On the positive side, there should be an increase in the firm's PVTS (assuming that proceeds are not eventually used to retire debt). On the negative side, there should be an increase in the agency costs of equity. The importance of the agency cost can also depend on the investment opportunities of the issuing firms.

Using proceeds to retire long-term debt reduces the present value of financial distress costs (PVFDC), which should have a positive influence on share prices. This use of funds also will decrease the agency costs of debt (a positive influence on share value), but increase the agency costs of equity (a negative influence on share value). Thus, the expected change in share prices is indeterminant.

Using proceeds to retire bank debt produces the same effects as retiring long-term debt, but has one additional negative effect, a reduction in bank monitoring activity. Again, we cannot predict the net impact on share prices.

II. Overview of MIPS

Pioneered by Goldman Sachs, the first MIPS offering was sold by Texaco in October 1993 through an offshore, tax-sheltered subsidiary in the Turks and Caicos Islands. The proceeds from the issue were lent to the parent firm in the form of a deeply subordinated intercompany loan with economic terms similar to those of the preferred stock. The parent made a tax-deductible interest payment to the subsidiary, which used the proceeds to pay dividends to the preferred shareholders. The parent also guaranteed the redemption value of both the preferred securities and dividends. Because the tax-sheltered subsidiary paid no taxes, the interest paid to the subsidiary was not taxable.

MIPS offerings have shown a substantial growth from just two issues in 1993 to 94 in 1998. In August 1994, the SEC allowed MIPS to be issued by special-purpose US subsidiaries (i.e., a limited partnership or a limited liability company) or by trusts, rather than only offshore subsidiaries, which simplified the legal work. This also led to the creation of MIPS hybrids, including TOPrs (trust-originated preferred security), QUIPS (quarterly income preferred security), QUICS (quarterly income capital securities), and QUIDS (quarterly income debt securities). As of December 31, 1998, over $64 billion of MIPS and its hybrids have been sold to the public. [2]

Although MIPS is deeply subordinated debt, the rating agencies consider this security and its various hybrids as equity instruments. Some of the features of these securities that rating agency analyses consider are deferability of payment, years to maturity, coverage ratios, the intercompany loan's degree of subordination, convertibility to equity, and tax risk. Standard and Poors view the length of payment deferability as an especially critical issue. Also, although the initial MIPS issues provided deferral for 18 months, the current standard is now five years. The security's maturity is also an important feature: the permanence of MIPS issues in the capital structure is determined by the maturity of the underlying intercompany loan.

Because MIPS issuers are allowed to deduct dividends for tax purposes, the IRS has ruled that investors in MIPS are not entitled to the 70% dividend exclusion that has been available to corporate investors in equity issues. It is because of this 70% tax exclusion that corporate investors have dominated the market for conventional preferred stock. In contrast, MIPS investors usually are not corporations. [3] To make the security attractive to potential investors, the MIPS dividend yield is often higher than the prevailing yield on conventional preferred issues. For example, when MIPS proceeds are used to retire outstanding preferred stock, the mean MIPS dividend yield is 8.23%, compared to 7.71% for the redeemed issues. This difference, 0.52%, is statistically significant at the 0.05 level.

Flotation costs on MIPS are also quite expensive. For our sample of 185 issuing firms, the mean underwriting commission was 3.2% of gross proceeds compared to 1.4% for conventional preferred stock (See Brigham, 1995, for a survey of flotation costs).

When Goldman Sachs first launched MIPS, it was unclear whether the IRS would continue to allow dividends in MIPS to be tax deductible and whether existing issues would be exempt from subsequent changes in tax law. The uncertain tax status is reflected in a "Tax Event Warning" given in the typical MIPS prospectus. One such warning appears in the October 26, 1994, prospectus supplement for MCN Michigan's 9 3/4% Series A Cumulative Preferred MIPS issue:

In the opinion of a nationally recognized independent tax counsel...there is more than an insubstantial risk that (i) MCN Michigan is subject to federal income tax with respect to interest received on the Series A Subordinated Debt Securities, (ii) interest payable to MCN Michigan on the Series A Subordinated Debt Securities will not be deductible by MCN for federal income tax purposes or (iii) MCN Michigan to subject to more than a de minimis amount of other taxes, duties or other governmental charges.

Should such a tax event occur, the parent company was authorized to redeem the outstanding MIPS issues and liquidate the MIPS corporate subsidiary. An issuer that could not redeem the MIPS faced the possibility of having to pay a large MIPS dividend without an accompanying tax advantage.

The events of December 7, 1995, confirmed the wisdom of including the tax event exemption in the MIPS prospectus. New tax provisions contained in the Clinton administration's 1997 budget proposal would have denied interest deductions for all securities with a term to maturity of more than 20 years that were not shown on the issuing firm's balance sheet. The proposal chilled the MIPS market, particularly because there was some uncertainty as to whether the proposals could be applied retroactively. However, on April 1, 1996, the chairmen of the House Ways and Means Committee and the Senate Finance Committee issued a joint statement in which they vowed to stop the Treasury from blocking certain types of financial transactions merely by issuing tax proposals with immediate effective dates. Tax lawyers concluded that, even in the unlikely event that Congress enacted the administration proposals in the future, the changes would not be retroactive. Thus, firms could now issue MIPS in the certainty that any changes in tax policy would not apply to securities issued before enactment of the legislation.

However, despite the apparent resolution of the uncertainty surrounding the tax status of MIPS, prospectuses issued after April 1, 1996 still carry a tax event warning, reflecting the possibility that the government might eventually disallow the tax deductibility of MIPS dividends. The political events surrounding the administration's proposals and the subsequent assurance by Congress of the MIPS tax status suggests that a reduction in the uncertainty surrounding the tax status of MIPS occurred between December 1995 and March 1996. This reduction is consistent with an increase in the expected present value of the tax shields associated with MIPS issues filed after this period.

III. Sample Selection and Financial Characteristics of MIPS Issuers

The New Products Department of Goldman Sachs provided us with a listing of all MIPS sold between October 1, 1993 and December 31, 1998. The initial sample consists of 309 issues. We were unable to locate public information on 85 issuers who are private subsidiaries of other corporations. Financial information on 222 of the remaining 224 firms was available on Compustat.

We eliminated 14 firms from the sample because the issues were either the second or third issues from an initial shelf registration. Including these issues would have double-or triple-counted the effects associated with the original shelf registration filing. [4] We eliminated 16 firms because we could not find the SEC filing date or use-of-proceeds information for these issues. We dropped five Canadian and two British firms because they had no common stock returns on the CRSP database. We gathered data on the SEC filing date, issue size, and use of proceeds directly from the prospectuses on the SEC's EDGAR database. If the prospectus was missing from the EDGAR database, the information was obtained from Security Data Corporation's database of preferred offerings. To ensure that the SEC filing date was the first public mention of the offering, we conducted a search of the DJNR. In 120 cases, the issue was mentioned in the DJNR, but always concurrently or subsequent to the SEC filing date. The final sample co nsists of 185 firms that filed MIPS issues prior to January 1, 1999.

Table II compares selected financial statistics of the sample firms with a set of comparable firms. We define a comparable firm as the single firm that has the same four-digit SIC code and is closest in size to the sample firm, as measured by the market value of its common equity. [5] In addition, we separate the full sample into subsamples of industrials and public utilities. We want to determine if there are common features among firms issuing MIPS that might influence the market's response to this event.

We focus our attention on factors corresponding to firm size, debt service ability, liquidity, and capital structure. The mean total asset value for the 185 sample firms is $26.47 billion compared to $19.42 billion for the comparable firms. This difference is significant at the 0.01 level.

We note that the mean interest coverage ratio for the sample firms is significantly lower than the comparable firms. This difference is more pronounced for the utilities with a mean ratio of 6.63 compared to the comparable group's 10.25, a difference statistically significant at the 0.01 level. The table also shows that the sample firms have lower current ratios (0.97 versus 1.12), a difference that is marginally significant at the 0.10 level.

As a group, the industrials and utilities have a significantly higher (0.05 level) total debt to assets ratio and a higher long-term debt to total assets ratio, also significant at the 0.05 level. For both ratios, the results are being driven primarily by the utilities.

The fact that the sample firms tend to be below-average in debt servicing capability, less liquid, and more heavily leveraged than their counterparts suggests that the pursuit of greater financial flexibility could be one of the motivating factors behind their decision to issue MIPS. However, issuing MIPS can also provide managers with more free cash flow, thus exposing shareholders to the risk of managerial opportunism, and higher agency costs of equity might influence the market's reaction to a MIPS offering.

IV. Event-Study Tests

We compute abnormal returns using the CRSP Excess Returns Tape, which ranks stocks in the CRSP daily file according to the magnitude of their market-model betas. We determine daily abnormal returns for each stock by subtracting the average daily return recorded by the stocks in its beta decile from the security's daily raw return. We compute cumulative abnormal returns (CARs) over a two-day period. Day 0 is the date that the sample firm filed a MIPS issue with the SEC. As mentioned above, a notice of the filing is often available in the Dow Jones News Service the next day, so we include Day +1 in the calculation of our abnormal return measure. We base statistical tests on averages of standardized abnormal returns to control for heteroskedasticity. [6]

Table III presents the two-day CAR for the registration of a MIPS issue with the SEC. The two-day CAR for all MIPS issuers is -0.12%, not significantly different from zero at the 0.10 level (t-statistic = -0.91).

In Panel A of Table III, we show that using part or all of the proceeds to redeem outstanding preferred stock (n=41) produces a positive abnormal return of just 0.45% with a t-statistic of 1.98 that is not statistically significant at the 0.05 level. There are several possible explanations for this result. First, the size of the MIPS issues might not be large enough to produce economically significant tax savings. Second, since 12 of the 41 firms in our sample used a portion of the proceeds for other purposes, confounding agency effects might mitigate the event's influence on the stock price of these firms. Further, because these 12 firms could have used part of the proceeds to retire debt, which would reduce the tax savings, we could have overestimated the magnitude of their PVTS. Another explanation arises from the uncertainty surrounding the tax status of MIPS prior to April 1,1996. If investors thought that the government might later disallow the tax deductibility of MIPS dividends, then Equation (1), wh ich assumes a permanent stream of tax savings, would overstate the expected PVTS for these securities.

To investigate these issues, we first compute the PVTS as follows:

PVTS = MIPS[([k.sub.mips] * [t.sub.c]) - ([k.sub.mips] - [k.sub.pf])]/[k.sub.mips] (1)

where:

MIPS = offering value of MIPS issue

[k.sub.mips] = pre-tax yield of MIPS issue

[t.sub.c] = marginal corporate tax rate [7]

[k.sub.pf] = pre-tax yield of retired preferred issue [8]

For the 29 firms where preferred stock redemption is the sole use of funds, the mean (median) issue size is $264.6 ($125) million with a corresponding mean (median) PVTS of $63.8 ($35.2) million. This PVTS represents 2.24% (1.26%) of the firms' market value of common equity.

Panel B of Table III shows that the impact from tax savings is larger for these 29 firms, with a two-day CAR of 0.83%. This CAR is statistically significant at the 0.01 level. [9] For the 15 issues filed before April 1, 1996, the two-day CAR is just 0.24% and not statistically significant (t-statistic = 0.70).

In contrast, the CAR for the 14 issues filed after this date is a positive 1.45%, and significant at the 0.01 level (t-statistic = 6.25). Thirteen of these 14 issues have a positive CAR, a percentage that is significant at the 0.01 level in a generalized sign test. [10] As a percentage of common equity, the mean PVTS for this subsample is 1.92%, or 0.47 percentage points above the mean CAR of 1.45%. This differential could be due to continued uncertainty in some investor's minds over the future tax status of existing issues. Tax event warnings were still included in MIPS filings after April 1, 1996, evidence that the tax uncertainty surrounding these securities had not completely disappeared. The Pearson correlation coefficient between the issues' CARs and corresponding PVTS is 0.47 (p-value = 0.06). Despite the small sample size, these results suggest that tax uncertainty reduced the expected PVTS of MIPS issued prior to April 1, 1996.

Returning to Panel A, when MIPS proceeds are used for debt reduction, the subsample's (n=77) two-day CAR is -0.36%, with a corresponding t-statistic of-1.66. Since agency theory hypothesizes that the market's reaction to the retirement of bank debt is different from its reaction to the reduction of other types of debt, we separate this subsample into two groups, as shown in Panel C. When proceeds are used to retire nonbank debt (either commercial paper or long-term debt), the abnormal return is 0.32%, with a t-statistic of 1.35. Since issuing MIPS to retire debt should reduce financial distress costs without sacrificing the tax savings in place, the static trade-off theory led us to anticipate a stronger market response. This result could be due to a confounding agency effect that dampens the market's reaction. Prior research (Masulis, 1980 and 1983, Eckbo, 1986, and Mikkelson and Partch, 1986) shows that leverage-decreasing actions, such as the replacement of debt with either preferred or common equity, has a negative impact on firm value. Substituting guaranteed coupon payments with preferred dividends, which can be suspended for up to 20 quarters, provides managers with greater financial flexibility that could be used to benefit themselves at the expense of shareholders. A negative reaction by shareholders to this agency cost could offset the positive benefits of a reduction in expected financial distress costs. Our sample firms are more heavily leveraged with lower debt servicing capability than comparable firms, which suggests that greater financial flexibility is one of the motivating factors behind the decision to sell MIPS. According to Jung et al. (1996), greater financial flexibility, coupled with limited growth opportunities for the issuing firm, should raise the agency costs of equity, thus reducing the (expected) positive price response to this event.

Panel C of Table III shows that for the 26 firms that retire bank debt, the two-day CAR is -1.33% with a corresponding t-statistic of -3.99. In Table I, the predicted price response to this event was indeterminate. However, the negative market reaction in Table III is consistent with James (1987), Lummer and McConnell (1989), and Hull and Moellenbemdt (1994) who find that firms that retire bank loans experience a significantly negative two-day CAR. Fama (1985) proposes an explanation for this result. Fama argues that banks have access to corporate information that is not available to other market participants. Consequently, the removal of the bank's monitoring activities is detrimental to other security holders, who must now incur the monitoring costs formerly borne by the banks. [11]

Panel A of Table III shows that when proceeds are used for general corporate purposes, the two-day CAR is -0.25%, and not statistically significant (t-statistic = -0.94). It appears that raising capital using MIPS does not have the negative impact of issuing seasoned common equity.

V. Regression Tests

In our next group of tests, we provide additional evidence on those factors responsible for the abnormal stock returns experienced by the sample firms. We regress the sample issues' two-day CARs on several independent control variables, including issue size (defined as the dollar size of the MIPS filing divided by total firm assets), firm size (defined as the log of total firm assets), and a dummy variable for public utilities.

The issue size control variable is included to ensure that a few very large MIPS issues do not drive our univariate results. Including a firm size variable as a control ensures that a few smaller firms are not driving our results. We include the utilities dummy variable as a broad control for industry effects.

Table IV shows the results of four regressions. The first three regressions analyze the impact of MIPS filings based on the three most commonly stated uses of funds: preferred stock redemption, debt reduction, and general corporate purposes. We also construct a final regression that uses the full sample of observations.

In addition to the control variables, we include variables for each particular use of funds. For example, Column A of Table IV, "Preferred Redemption," includes two variables equal to the calculated present value of tax shields from redemption of existing preferred stock. We create these two independent variables based on whether the issue is filed before or after April 1, 1996 (BEFORE PVTS and AFTER PVTS).

The results of this regression show that none of the control variables are statistically significant. This result suggests that the market's somewhat neutral response to the preferred redemption cannot be attributed to either the size of the MIPS offering or to firm size or industry. This result also holds true for the BEFORE PVTS variable, which shows that preferred redemptions filed before April 1, 1996 have an insignificant impact on share value.

However, we note that the AFTER PVTS variable is statistically significant with a coefficient of 0.85, suggesting that each dollar of PVTS raised firm value by $0.85. This finding is consistent with our event study results.

It is possible that a decline in issue costs over time could explain the AFTER PVTS results. Lower issue costs could be due to either lower dividend yields, lower underwriting fees, or both. To investigate these possibilities, we first compute the mean dividend yield for both the pre-April 1, 1996 and post-April 1, 1996 groups. The yields for both groups were very similar. The BEFORE group records an average yield of 8.28%, versus 8.19% for the AFTER group, a decline of just 0.09%. We also find that flotation costs are consistent over the full sample period. For the BEFORE group, mean underwriting costs were 3.1863%, versus an almost identical 3.1827% for the AFTER group. This stability in fees is probably because Goldman Sachs, the market leader, initially set their fees at 3.15% and this price was matched by Merrill Lynch, their major competitor.

We note that the 14 preferred redemption issues filed after April 1, 1996 record a two-day CAR of 1.45%, compared to a mean PVTS of 1.92%, or 75.5% of the total value of the PVTS. The fact that the AFTER PVTS coefficient is less than one supports the view that some investor uncertainty over the tax status of MIPS continues to exist even for issues sold after April 1, 1996.

The second model in Table IV, Column B, examines the common stock reaction to MIPS issuers who stated that they intended to use the proceeds to reduce debt. We form four subsamples of firms from the 77 firms that used MIPS to retire debt. We separate these 77 firms according to whether they retire bank debt or nonbank debt, and whether they have strong senior debt ratings (above BBB+) or weak senior debt ratings. [12]

Our motivation for examining bond ratings is based on a study by Gilson anti Warner (1996), who find that selling junk bonds to pay down bank debt results in negative stock returns for the issuing firms. Gilson and Warner contend that junk bonds convey negative information about the firm's prospects. Therefore, we examine whether there is an association between the market's responses to the retirement of bank debt and the credit rating of the issuing firm's senior debt. Although not reported in Table III, we find that for the 15 firms with debt rated A- or higher, the CAR is -0.71% (t-statistic = 1.31) versus -1.83% (t-statistic 5.12) for the 17 firms with BBB+ or lower ratings. This difference is statistically significant at the 0.05 level.

We isolate abnormal returns for three of the four subsamples using dummy variables. [13] The coefficient on BANKSTRONG in Column B, Table IV, shows the impact of a MIPS filing for firms with strong ratings that retired bank debt. The coefficient on this variable is -1.18% with a t-statistic of -2.15.

The negative impact from retiring bank debt is even more pronounced for firms with bonds rated BBB+ or lower (BANKWEAK). The coefficient on this variable is -2.14% with a t-statistic of -3.88. This means that these MIPS issuers have abnormal returns that are 2.14 percentage points lower than the other debt retirees, showing that removal of bank monitoring is more severe for firms in weak financial condition. [14] This result suggests that research on the value of bank monitoring of corporate activities should take into account the financial condition of the borrower. However the significantly negative impact of weak ratings does not apply to firms with weak ratings that retire nonbank debt. DEBTWEAK is a dummy variable for issuers with BBB+ ratings or lower that retire nonbank debt. We find that the coefficient on this variable is insignificantly negative.

Our sample contains 59 MIPS issuers that use all of the proceeds for general corporate purposes. Jung, Kim, and Stulz (1996) argue that cash raised from an equity offering can create considerable agency problems. They find that the negative impact of common stock offers increases for firms with weak growth prospects (low market-to-book ratios). If the cash raised for general corporate purposes from MIPS issues creates similar agency problems, then the sample firms' market-to-book ratios should explain the announcement-day impact for this subsample of MIPS issuers.

To determine if issuing MIPS has agency cost implications for the common stock, we examine the association between the issuing firms' market-to-book ratio (MKTBOOK), as described in Jung, Kim, and Stulz (1996), and our common stocks' abnormal returns. We present a regression using all 59 firms that used proceeds for general corporate purposes in Column C.

The coefficient on the market-to-book ratio is negative and insignificant. Although this result is inconsistent with the Jung, Kim, and Stulz (1996) results for common equity issues, it is consistent with their results for subordinated debt issues. In their study, a firm's market-to-book ratio has no effect on the common stocks' abnormal returns surrounding subordinated debt issues. We conclude that in terms of agency costs, the market sees the act of issuing MIPS as being closer to issuing debt rather than issuing common stock. This result is consistent with the relatively large and frequent dividend payments required from the issuer, and the right of the MIPS investors to force redemption if dividends are unpaid for a specified period (generally 20 quarters).

In Table IV, Column D, we present a composite regression using the entire sample (including eight other uses of funds firms). The coefficients are generally consistent with the results in Columns (A)-(C). The only difference in significance is for BANKSTRONG, where the coefficient becomes marginally less significant, with a p-value of 0.06.

The regressions presented in Table IV appear to be well specified and robust. Using White's (1980) general test, all four regressions fail to reject the null hypothesis of homoskedasticity. An analysis of the residuals indicates that the conclusions are robust to exclusion of outliers. Examination of the Cook's D statistic, the DFFITS statistic, and the studentized residuals produces three observations that are potentially influential observations. Further examination reveals that two observations are influential because the abnormal returns were relatively large (-4.7% and -6%). Including these two observations increases the mean squared error of the regression and tends to reduce the overall explanatory power of the regressions. Nevertheless, we do not exclude them, because these are not exceptionally large values for two-day abnormal returns. The third influential observation is a company in financial distress (S & P rating of BB). As a result, the size of the MIPS issue, and resulting PVTS, is an unusual ly large fraction of the market value of the firms' common stock. We exclude this observation from the analysis. If we were to include it, it would reduce the already insignificant coefficient on BEFORE PVTS.

VI. Conclusion

The unique nature of MIPS lets us examine the predictions from two prominent theories of corporate capital structure, agency theory and the traditional static trade-off theory. We find that both theories are needed to interpret correctly the market's positive or negative response to the filing of a MIPS issue. Which theory (if either) dominates depends on the issuer's use of proceeds.

The static trade-off model predicts that increasing the present value of tax savings without increasing financial distress costs will raise firm value. Therefore, we hypothesize that using MIPS proceeds to redeem outstanding preferred stock will result in a positive market response. When we focus on those securities issued after April 1, 1996, when it became clear that any future changes in tax policy would not affect the tax deductibility of MIPS, MIPS issues show positive and significant abnormal returns. We interpret this finding as both a confirmation of the traditional static trade-off theory and evidence that government policy can create uncertainty regarding the efficacy of new financial products.

When proceeds are used to retire long-term debt, the market reaction is again positive, but not statistically significant. Static trade-off theory predicts that reducing costs of financial distress without losing tax benefits will raise share prices. However, MIPS issuers are more heavily leveraged and have lower debt servicing capability than comparable firms, suggesting that the motivation for this security is to achieve a higher level of financial flexibility. This action, in turn, exposes shareholders to higher agency costs of equity that offset lower financial distress costs.

In contrast, when proceeds from a MIPS issue are used to retire bank debt, agency costs appear to dominate the benefits from a reduction in financial distress costs. The significantly negative stock price reaction is consistent with several studies that find that issuing long-term debt to retire bank debt has a negative impact on share prices. We also find that these negative returns are more severe for firms with relatively low credit ratings, which suggests that future research on the value of bank monitoring should consider the creditworthiness of the issuing firm.

Our results further support the argument that banks serve a valuable monitoring function on corporate activities. If bank monitoring is removed, it will have a negative influence on firm value.

(*.) Paul Irvine is an Assistant Professor and James Rosenfeld is an Associate Professor at Emory University.

(1.) Other examples of recently developed preferred security innovations include convertible exchangeable preferred stack (see Cowan, 1999) and auction rate preferred stock (see Alderson and Fraser, 1993).

This paper has benefited greatly from helpful suggestions by George Benston, Dave Blackwell, Sherry Jarrell, Omesh Kini, Bill Megginson, Shehzad Mian, Nicholas Valerio, the participants at workshops at Emory University and University of Georgia, and Rajish Narayanan, our discussant at the European Finance Association Meetings. We thank Ron Harris, Emma Lindstrom, and Catrina Deese for their assistance, and the Interactive Data Corporation and John Graham for providing us with marginal corporate tax rates. Special thanks goes to Andrew Feldschrieber and the analysts at Goldman Sachs for providing us with valuable sample information. Finally, we would like to thank the Editors and the two anonymous referees for their valuable comments and suggestions.

(2.) See Khanna and McConnell (1998) for a detailed summary of MIPS issue frequency.

(3.) Although data are not available to determine a precise percentage breakdown between corporate and noncorporate purchases of conventional preferred stock and MIPS, we base these statements on private conversations with investment bankers from Goldman Sachs, Merrill Lynch, and Salomon Bros.

(4.) Following Bhagat, Marr, and Thompson (1985), we consider shelf offerings registered under Rule 415 equivalent to ordinary offerings.

(5.) Tests using comparable firm groups of 3, 5, 7, and 10 firms closest in size to the sample firms produced similar results to those reported in Table II.

(6.) A description of the method for computing t-statistics for cumulative abnormal returns can be found in Linn and Pinegar (1988), pages 165-168.

(7.) As a proxy for the marginal corporate tax rate for the sample issues, we use the technique described in Graham (1993, 1996). In most cases, our sample firms' marginal tax rate was 35.0 percent, which is the marginal tax rate for firms earning over S18,333,333 before taxes. Houston and Houston (1990) contend that firms selling conventional preferred stock, whose dividends are not tax-deductible, will have below-average expected tax rates. We, in turn, would argue that firms that issue MIPS, whose dividends are tax-deductible, would be in higher expected tax brackets.

(8.) Because dividends on the retired preferred issue are not tax-deductible; its pre-tax yield is identical to its aftertax yield.

(9.) In a similar test, Khanna and McConnell (1998), report a three-day CAR of 1.61% (t-statistic = 2.14). However, their sample has only six observations.

(10.) Cowan (1992) contends that the generalized sign test is better specified in event studies than the conventional sign test.

(11.) "In an interesting twist, Booth (1992) reports that closely held (private) firms pay 33 basis points more in bank loan spreads relative to firms with publicly traded equity and rated public debt. These findings support the cross-monitoring hypothesis, which suggests that monitoring by public equity-holders and bond-rating agencies reduces bank monitoring costs. Johnson's (1998) finding that, all else equal, firms with bank debt are more highly levered also supports the cross-monitoring hypothesis.

(12.) We obtained Standard and Poor's senior debt ratings for 167 firms from the Compustat database (150 firms) and the Standard and Poor's Bond Guide (17 firms).

(13.) The fourth group, issuers with strong ratings that are nonbank debt, is not represented by a dummy variable. We wish to avoid the dummy variable trap.

(14.) When we split this regression into before and after April 1, 1996 subsamples, we find that the coefficients on BANKWEAK and BANKSTRONG become slightly less negative in the AFTER sample relative to the BEFORE sample. This result, though not reported in the tables, is consistent with the expected PVTS benefit increasing after April 1, 1996.

References

Alderson, M. and D. Fraser, 1993, "Financial Innovations and Excesses Revisited: The Case of Auction Rate Preferred Stock," Financial Management 22(2), 76-93.

Bhagat, S., W. Marr and R. Thompson, 1985, "The Rule 415 Experiment: Equity Markets," Journal of Finance 40(5), 1385-1401.

Booth, J., 1992, "Contract Costs, Bank Loans, and the Cross-Monitoring Hypothesis," Journal of Financial Economics 31(1), 25-41.

Brigham, E., 1995, Fundamentals of Financial Management, 7th ed., Dryden Press, Fort Worth, TX.

Cowan, A.R., 1992, "Nonparametric Event Study Tests," Review of Quantitative Finance and Accounting 2(December), 343-395.

Cowan, A.R., 1999, "Tax Options, Clienteles, and Adverse Selection: The Case of Convertible Exchangeable Preferred Stock," Financial Management 28(2), 15-31.

Diamond, D., 1991, "Monitoring and Reputation: The Choice Between Bank Loans and Directly Placed Debt," Journal of Political Economy 99(4), 688-721.

Eckbo, E., 1986, "Valuation Effects of Corporate Debt Offerings," Journal of Financial Economics 15(1/2), 119-152.

Fama, E., 1985, "What's Different About Banks?" Journal of Monetary Economics 15(1), 29-36.

Gilson, S. and J. Warner, 1996, "Junk Bonds, Bank Debt, and Financial Flexibility," University of Rochester Working Paper.

Graham, J.R., 1993, "Proxies for the Marginal Tax Rate," Journal of Financial Economics 42(2), 187-221.

Graham, J.R., 1998, "Simulated Tax Rates," available online from www.duke.edu/[sim]jgraham/.

Houston, A.L. and C.O. Houston, 1990, "Financing With Preferred Stock," Financial Management 19(3), 42-54.

Hull, R. and R. Moellenberndt, 1994, "Bank Debt Reduction Announcements and Negative Signalling," Financial Management 23(2), 31-42.

James, C., 1987, "Some Evidence on the Uniqueness of Bank Loans," Journal of Financial Economics 19(2), 217-235.

Johnson, S., 1998, "The Effect of Bank Debt on Optimal Capital Structure," Financial Management 27(1), 47-56.

Jung, K., Y. Kim, and R.M. Stulz, 1996, "Timing, Investment Opportunities, Managerial Discretion, and the Security Issue Decision," Journal of Financial Economics 42(2), 159-185.

Khanna, A. and J.J. McConnell, 1998, "MIPS, QUIPS and TOPrs: Old Wine in New Bottles," Journal of Applied Corporate Finance 11(1), 39-44.

Linn, S. and J.M. Pinegar, 1988, "The Effect of Issuing Preferred Stock on Common and Preferred Stockholder Wealth," Journal of Financial Economics 22(1), 155-184.

Lummer, S. and J. McConnell, 1989, "Further Evidence on the Bank Lending Process and the Capitalmarket Response to Bank Loan Agreement," Journal of Financial Economics 25(2), 99-122.

Masulis, R., 1980, "The Effect of Capital Structure Change on Security Prices," Journal of Financial Economics 8(2), 139-178.

Masulis, R., 1983, "The Impact of Capital Structure Change on Firm Value: Some Estimates," Journal of Finance 38(1), 107-125.

Mikkelson, W.H. and M.M., Partch, 1986, "Valuation Effects of Security Offerings and the Issuance Process," Journal of Financial Economics 15(1/2), 31-60.

White, H., 1980, "A Heteroscedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroscedasticity" Econometrica 48(4), 817-838.

COPYRIGHT 2000 Financial Management Association

COPYRIGHT 2000 Gale Group