Financial innovations and excesses revisited: the case of auction rate preferred stock.

The surge in financial innovation that occurred during the 1980s produced a family of securities that were designed to lower the cost of preferred stock financing by widening the availability of the Internal Revenue Code Section 246(c) dividends-received deduction. Consistent with the properties of optimal security design developed by Allen and Gale |2~, the objective of this effort was to enhance the marketability of preferred equities to the tax clientele that values them most, corporate purchasers. In a competitive market, the tax advantage of preferred equity to those purchasers is reflected in a lower pre-tax dividend yield, causing the cost of preferred equity to fall below the cost of debt financing for companies with low marginal tax rates.

Central to the success of this effort was the development of a feature that reduced share price volatility so that variable rate preferred could effectively serve as a substitute for money market investments. The first generation product to obtain acceptance was adjustable rate preferred stock, a security with a dividend rate that is adjusted quarterly to reflect changes in money market yields. Although it enjoyed a high degree of popularity early on, investor demand for adjustable rate preferred stock weakened when experience revealed that the shares retained a degree of price volatility that was incompatible with the objectives of corporate cash managers (Winger et al |13~).

The disappointing experience of adjustable rate preferred shares led to the development of improved dividend adjustment features in a second-generation product, auction rate preferred stock.(1) In addition to resetting the dividend more frequently (usually every 49 days), the refinements to the dividend adjustment procedure for auction rate preferred provide for the dividend yield to be determined in a dutch auction, thereby allowing it to fully adjust to movements in interest rates and to adapt to changes in credit risk premia within a limited range.

The dutch auction process begins when the auction agent takes orders from both existing holders and from potential holders through designated broker-dealers. The supply of available shares at a particular auction is determined by the number of shares offered for sale by existing shareholders; the new dividend yield is established by identifying the lowest rate at which those shares can be sold. This dividend rate serves a dual purpose; it restores the market value of the shares to the auction exchange price (the liquidation preference amount) and represents the cost of preferred equity to the issuer over the next seven weeks.

Existing shareholders who wish to retain their shares may choose to do so on a noncompetitive basis at the new dividend yield or may bid a rate at or above which they are willing to hold their shares. In the event that all holders elect to retain their shares, the dividend yield is automatically set at a prespecified floor that usually equals 58% of the 60-day AA commercial paper rate.(2) Current holders who place a bid must nevertheless sell their shares if the new dividend yield falls below the rate which they specified.

Bids from potential purchasers are not accepted if they exceed a prespecified maximum dividend rate which typically ranges between 110% and 150% of the 60-day AA commercial paper rate. As a consequence, existing shareholders who elect to sell their shares will be able to divest completely only if the number of acceptable bids is sufficient to cover the number of available shares. If the number of shares demanded is insufficient, the new dividend yield is set at the maximum dividend rate. In that event, the auction is said to have failed, and existing holders electing to sell will only be allowed a partial divestiture at the auction on a pro rata basis, to the extent that demand allows.

The cost of preferred equity following an auction failure is defined by the prespecified maximum dividend rate. Because the maximum dividend rate is less than the market clearing dividend yield, the market value of a failed auction rate issue is less than the auction exchange price. The holder suffers a capital loss on the auction rate shares, because the dividend yield cannot rise by the amount necessary to restore the market value to the liquidation preference amount. This condition highlights an important characteristic of auction rate preferred stock: existing holders who wish to divest have no recourse to the issuer in the event that the number of shares demanded at a dividend yield below the stated maximum rate is insufficient to cover the available supply.(3)

Initial growth of the auction rate preferred stock market was so rapid that the total dollar value of new auction rate issues exceeded the amount of convertible preferred stock and fixed rate preferred issued in 1987 (Houston and Houston |6~). This growth occurred in apparent response to the significant financing advantage to auction rate issues, which typically carry a lower dividend yield than comparable fixed rate shares. In spite of this apparent yield advantage, however, the total dollar amount of auction rate preferred stock outstanding peaked in 1988 and has declined since.

Previous research by Alderson, Brown, and Lummer |1~ demonstrated the effectiveness of the emerging market for auction rate preferred in allocating capital from fully taxable companies to issuers with low marginal tax rates. Here, we examine the factors influencing issue and redemption activity in a more seasoned version of that market, with the objective of explaining why it has atrophied. The issue is important because it reflects on whether auction rate preferred stock is a viable financial innovation that enhances the completeness of the market, or whether it more closely resembles a "speculative balloon" slated for virtual extinction.(4) A linkage between the contraction of the auction rate preferred market and systematic factors, such as a reduction in the tax advantage of the security or an indication that the auction process was unreliable, would provide reason to question its sustaining innovative features. On the other hand, if the changes in the market are primarily attributable to the replacement of low-quality issues by higher quality shares, recent changes in the market could be viewed as transitory, thereby suggesting that auction rate preferred stock constitutes a legitimate financial innovation that contributes to the completion of the capital markets.

The approach here includes an empirical analysis of the redemption decision along with a summary of the results to a mail survey of 51 companies that have called their auction rate preferred stock for redemption. Our analysis suggests that the contraction of the market for auction rate preferred stock occurred in response to a strong clientele preference for low-risk issues and the enactment of both thrift reform and tax legislation which encouraged the retirement of auction rate shares sold through bankruptcy-remote subsidiaries. The survey responses indicate that the reduction in the dividends-received deduction, volatility in the time-series of dividend yields, and the risk of auction failure have influenced only a small proportion of the firms that have chosen to redeem their shares. Independent of the tax and regulatory changes, much of the redemption activity in the auction rate preferred market appears to reflect the exit of high-risk shares. Since those shares are being replaced to some extent by low-risk issues, it might be appropriate to conclude that auction rate preferred stock can meet the needs of a very narrowly defined investor clientele for high-quality preferred stock. However, the supply of high-quality, low-tax-rate firms is naturally limited, and perhaps especially so following the changes imposed by the Tax Reform Act of 1986. As a consequence, the design features of auction rate preferred, while sound, may be tailored to serve a market that is too small to be of consequence.

I. Factors Affecting Issue and Redemption Activity

The decision to issue auction rate preferred stock or to redeem an outstanding issue depends partially upon whether the security is the lowest cost source of capital on a risk-adjusted basis. This section describes the various economic factors that may have influenced the cost of auction rate preferred capital, and derives empirical proxies for firms' incentives to issue or redeem the security.

A. Natural Selection Among Risk Profiles

A well-known risk associated with auction rate preferred stock is the "failure" of an auction. Auction failure occurs when the market clearing dividend yield exceeds the prespecified rate ceiling for a particular issue, causing the auction exchange price of the security (typically set equal to $100,000 per share) to exceed its market value. Under those circumstances, the auction "fails" because potential buyers are unwilling to purchase a security of impaired value at the fixed exchange price.(5) Auction failure operationalizes the realization of preferred equity risk by providing for states of nature in which the required return will exceed the constraints imposed on the dividend yield. The purpose of this design is to satisfy the Section 246(c) "at risk" requirement; as a consequence, the dividends-received deduction and the potential for auction failure are inseparable: holders of money market preferred stock cannot simultaneously maintain a position in variable-rate preferred equity, take advantage of the dividends-received deduction, and avoid the risk of auction failure.(6)

By design, the principal clientele within the market for auction rate preferred stock is comprised of corporate cash managers attempting to maximize the after-tax yield on liquid reserves.(7) Because auction failure corresponds to a loss of invested principal and a drastic reduction in liquidity, it would be reasonable to expect these highly risk-averse corporate cash managers to avoid issues that they perceived to be likely candidates for failure.(8) (This abbreviation is reinforced by the absence of any interest-rate-driven capital gains potential for the holder of these variable rate instruments.) Owing to the nature of a dutch auction, the fear of an imminent failure can become a self-fulfilling prophesy; supply irreconcilably exceeds demand as existing shareholders tender their shares for sale and potential shareholders do not submit bids.

The aversion of corporate investors to auction failure is perhaps typified by the recent experience of Citicorp, which saw the dividend yield on its auction rate preferred increase from 8.776% to 13% in a series of auctions between October 17 and November 1, 1990. Press reports characterized the increase as "a strong sign that investors are nervous about the securities of the largest banking company in the U.S." (Wall Street Journal, October 23, 1990, p. 4).(9) Citicorp first responded to the increase in dividend yields by unilaterally increasing the maximum dividend rate. When that change failed to calm investor fears, the bank announced that it would redeem its auction rate preferred shares and dividend yields fell sharply at subsequent auctions.

We hypothesize that the contraction of the market for auction rate preferred stock, in part, reflects the outcome of a natural selection process by which risk-averse corporate cash managers, virtually the only participants on the demand side of this segmented market, flee the high-risk sector. The withdrawal of their demand increases the yield on high-risk issues and leads issuers of high-risk DARPS to redeem their shares. To test this hypothesis, we relate the firm's choice with regard to auction rate preferred financing (reset or redeem) to a set of variables that directly or indirectly measure cross-sectional differences in their risk. The obvious choice for measuring this risk is the dividend yield itself, because it allows for a direct examination of the relationship between the redemption decision and the cost of auction rate preferred capital. Because redemptions occur across time, however, the use of the dividend yield in isolation would ignore the fact that the time-series of dividend rates is influenced by both short-term interest rates and changes in the tax law. We therefore employ as the measure of the cost of auction rate preferred capital the ratio of the dividend yield to the contemporaneous 60-day AA commercial paper rate, adjusted to reflect the concurrent tax advantage to preferred stock. For firms that redeemed their issues, the tax-adjusted dividend yield ratio is measured at the latest auction available just prior to the redemption date; for the nonredeeming companies, the ratio is the latest available in the first quarter of 1991.

Redemption behavior represents only one side of the natural selection hypothesis, however. Under the behavior characterized by that theory, the market not only discriminates against high-risk issues but selects in favor of low-risk shares. To explore the full symmetry of the theory, we also examine the characteristics of firms that have issued new auction rate preferred shares concurrent with the surge in redemptions (which began in 1988). In this test, the timing of the issue (before 1988 or after 1987) is also related to the variables that measure cross-sectional differences in risk.

B. Regulatory Changes

1. Thrift Reform

Following the first thrift auction rate preferred issue in November 1985 (a $75 million issue by City Federal Savings Bank), thrifts became major sellers of this type of security, with over $4 billion outstanding by April 1987 (representing 42% of the total market).(10) A number of factors account for the attractiveness of this type of financial instrument to thrifts.

Issuing auction rate preferred is an attractive financing alternative for non-tax-paying entities, a characteristic that certainly applied to many large thrifts throughout the 1980s. Particularly relevant to thrifts, however, is the ability to lessen funding costs even further by segregating marketable securities (mostly mortgage-backed securities) in a special, limited-purpose subsidiary and raise funds at or near an Aaa rate. Also, thrifts are able to use these securities as a form of overall asset/liability management.

Most thrift issues are sold through limited-purpose subsidiaries, corporations formed solely for the purpose of selling the auction rate preferred and managing the assets that support it. In order to avoid existing limitations on the amount of assets that may be devoted to any one use, the Federal Home Loan Bank Board referred to these organizations as "nonsubsidiary" subsidiaries funded with "noninvestment" investments.(11) The assets of the subsidiary (at least 55% of which must be GNMAs, FNMAs, or FHLMCs) must substantially exceed (in some cases by as much as 70%) the value of the auction rate preferred.(12) The securities became highly rated because of their over-collateralization and also due to the willingness of the thrift regulator (the Federal Home Loan Bank Board prior to 1989) to recognize the "corporate veil" in the event of the failure of the parent.(13) This allowed thrifts to use special-purpose subsidiaries to conduct tax arbitrage by offsetting taxable earnings from the subsidiary against the operating losses of the parent.

We hypothesize that a confluence of factors gravitated against thrift participation in the auction rate preferred market beginning in 1989 and that these factors contributed to the observed contraction. The test of this hypothesis relates a classification variable that identifies auction rate preferred shares sold through limited-purpose subsidiaries to both the redemption and issue timing decisions. The thrift crisis itself affected the market receptivity to any securities from this industry, and resulted in larger credit risk premiums for thrift issues. Moreover, the passage of thrift reform legislation in 1989 and of the Revenue Reconciliation Act of 1990 dramatically affected portfolio management practices at thrifts.

Perhaps most importantly, the reform bill dramatically increased the capital requirements at thrifts by requiring them to meet the same capital requirements as FDIC insured banks by June 1, 1991 (six percent as compared to a previous loosely enforced standard of three percent).(14) For most thrifts, with little or no earnings, and unable to access the capital market for equity, shrinking total assets was the only feasible way to meet the new capital requirements and the sale of the mortgage-backed securities in the limited-purpose subsidiaries (and the redemption of the auction rate preferred issued by these subsidiaries) became the most practical way to shrink. At the same time, since the thrift reform legislation abolished the existing thrift regulator (the Federal Home Loan Bank Board) and replaced it with a new (and tougher) regulator (the Office of Thrift Supervision), there was some concern about the continued existence of the "corporate veil" at these limited-purpose subsidiaries, referred to as "regulatory call" risk.(15)

The Revenue Reconciliation Act of 1990 also reduced the attractiveness of special-purpose subsidiary financing by imposing limitations on the consolidation of parent net operating losses to offset income earned on the investment of auction rate preferred proceeds. Because these limitations apply to auction rate preferred stock issued after November 17, 1989, they serve to limit the tax arbitrage opportunities available from selling new bankruptcy-remote issues.

2. Bank Capital Requirements

Changing capital requirements in the banking industry may also have played a role in the redemption decision of banking organizations. On December 29, 1989, the Federal Reserve Board issued for public comment a proposal to limit the proportion of perpetual preferred stock to 25% of Tier I capital.(16) This proposal was subsequently adopted (along with the other component of risk-based capital) late in 1990.

Dividend yields on auction rate preferreds in the banking sector rose dramatically during late 1990 in apparent response to weak industry earnings. We hypothesize that the change in Tier 1 capital requirements reduced the nontax benefits of employing auction rate preferred as a source of capital, thus exacerbating the incentive to redeem issues within the banking sector in favor of issuing lower cost securities.(17)

C. Other Factors

Van Horne |11~ defines a legitimate financial innovation as a security which makes the financial markets more efficient and/or more complete. In his analysis of financial innovation and excesses, he attributes the development of adjustable rate preferreds to both tax law changes and to a demand for products that reduce the risk associated with volatile interest rates, and observes that:

The various devices recently conceived and sold as financial innovations are not bubbles, although some of the same excesses are present ... A balloon might be a better metaphor for certain financial promotions. It is blown up to be sure, but not to the extent that it pops. The eventual deflation is less abrupt.

|An~ example of a balloon was the 1982 to 1984 explosion of instruments tailored so the corporate investor could take advantage of the 85 percent tax exemption of dividend income ... Never have financial promoters shown greater imagination; they lifted the tax avoidance industry to a new level of illusion. As might be expected, promoter fees were high, but initial demand proved extraordinary. For adjustable rate, preferred stocks investor excesses were particularly evident. At the peak in February of 1983, an ARPS security was priced with an adjustable return nearly 500 basis points below the yield on the appropriate benchmark, Treasury security. At that point, the balloon quickly deflated, and the negative differential declined. The 1984 Tax Act further deflated this balloon, but the opportunities have not been extinguished.

One implication of Van Horne's analysis is that the contraction of the market for auction rate preferred has been caused by a mitigation of the conditions that spawned its development. Here, we hypothesize that the market has been adversely affected by (i) a reduction in the corporate dividends-received-deduction that decreased the tax subsidy to auction rate preferred stock, and (ii) a reduction in the availability of nondebt tax shields (in the form of depreciation write-offs and investment tax credits) that increased the availability of the tax subsidy to debt financing. The combined effect of these changes would appear to have increased the cost of auction rate preferred capital on both a relative and absolute basis, providing an additional catalyst for the contraction of the market.

Exhibit 1. Comparison of the Tax Advantage of Auction Rate
Preferred Stock Relative to a Money Market Instrument Under
Original and Revised Tax Laws

                                            Yield
                    Auction     Money       Advantage
                     Rate       Market    (basis points)

Pre-tax yield        8.00%      10.0%          200
After-tax yield
original tax laws    7.45%(a)   5.4%(b)        205
After-tax yield
revised tax laws     7.18%(c)   6.6%(d)         58

Notes:

The corporate tax rate under original tax laws is 46% versus
34% under the revised tax laws. Corporations were allowed to
exclude 85% of dividends received from taxable income under the
original tax laws; under the revised laws the
dividends-received deduction is 70%.

a 8% - 8% (1-0.85) (0.46).

b 10% - 10% (0.46).

c 8% - 8% (1 -0.70) (0.34).

d 10% - 10% (0.34).

At the time of its introduction, corporations were allowed to exclude 85% of their dividend income from taxation at a maximum rate of 46%. Since 1986, however, the value of this tax break to corporate investors has been curtailed. The Tax Reform Act of 1986 lowered the exclusion to 80% against a reduced corporate tax rate of 34%; a further reduction in the exclusion to 70% became effective for dividends received after December 31, 1987.

Exhibit 1 demonstrates the extent to which the reduction in both corporate tax rates and the dividend exclusion reduced the tax advantages of holding auction rate preferred as a substitute for taxable money market investments. A money market instrument yielding ten percent returned 5.4% on an after-tax basis under the tax regime existing at the time auction rate preferred stock was introduced. An auction rate preferred issue trading at 80% of the money market yield would have returned 7.45% after-tax, providing a 205 basis point yield advantage.

The yield advantage under the same pre-tax yields has dropped to 58 basis points under the revised tax law. The narrower spread results from the lower corporate tax rate which raised the after-tax money market yield to 6.6% and the reduced dividend exclusion which lowered the after-tax auction rate yield to 7.18%. In a Scholes and Wolfson |10~ framework, investors would respond to the increased explicit tax on preferred stock by demanding a higher pre-tax dividend yield, in order to restore the previous risk-adjusted, after-tax return spread between auction rate preferred and fully taxable securities. The result would be an increase in the cost of preferred equity financing.

The Tax Reform Act of 1986 also eliminated the investment tax credit and lengthened the period over which corporate assets could be depreciated for tax purposes. These changes reduced the amount of nondebt tax shields available to offset taxable corporate income, potentially increasing the effective marginal corporate tax rate. Auerbach |3~ has shown that the reduction in nondebt tax shields more than offset the impact of the decreased statutory corporate tax rate, resulting in a net increase in the effective marginal tax rates of industrial companies. All other things equal, an increased marginal corporate tax rate would increase the relative cost of preferred equity by lowering the absolute cost of debt financing.

In summary, a confluence of changes in the tax code would appear to have increased the relative cost of auction rate preferred stock by (i) decreasing the implicit tax subsidy on preferred dividends and (ii) increasing the explicit tax subsidy to debt financing. Independent of the reduced tax benefits, however, the high interest rates that helped to spawn the initial demand for auction rate preferred have since abated, thereby reducing rewards to investing in auction rate shares. To demonstrate, recall that the pre-tax yields shown in Exhibit 1, which are representative of yields prevailing when auction rate preferred stock was introduced (August 1984), implied an after-tax yield advantage of 205 basis points. Given the secular decline in interest rates since that time, the yield advantage to auction rate preferred has been reduced in absolute value: at pre-tax yields of eight percent and 6.4 on money market and auction rate instruments, respectively, the after-tax yield on auction rate preferred currently exceeds the after-tax return to money market securities by only 47 basis points. The difference is even smaller for lower pre-tax yield spreads.

Van Horne's analysis also seems to imply, however, that the benefits of auction rate preferred stock to both purchasers and issuers were initially oversold. Such would be the case if (i) corporate investors overestimated the benefits of the dividends-received deduction relative to the incremental risk associated with preferred equity and/or (ii) corporate issuers underestimated the risk of rising dividend rates. Unfortunately, we cannot directly estimate the relative increase in the cost of auction rate preferred, nor the degree of disappointment experienced by corporations that redeemed their shares. Instead, we conducted a mail survey of corporations that redeemed auction rate preferred shares in an attempt to identify additional causes for the surge in redemption activity over the period 1988-1991.

II. Empirical Analysis

A. Data and Sample Selection

The Dutch Auction Rate Preferred Stock History Quarterly, published by Salomon Brothers, served as the point of origin for the empirical analysis. An assortment of documents published between March 1988 and March 1991 were used to identify 103 firms that issued at least one series of publicly traded auction rate preferred stock between the third quarter of 1984 through the first quarter of 1991. The same quarterly histories also reported the names of companies that have redeemed at least one series of auction rate preferred, along with the respective series of shares and associated redemption dates.

Each issue of the DARPS Quarterly also contains all or part of the time-series of auction dates and dividend yields for each issue outstanding at the publication date, along with the ratio of each dividend yield to the contemporaneous 60-day AA commercial paper rate (hereafter CP percentage). From the assortment of documents available, we were able to identify the dividend yield and CP percentage for each issue at the time of the initial offering. Because the time-series of quarterly histories was uniformly distributed over the 1984-1991 period, it was also possible to reconstruct at least a portion of the time series of dividend yields for each issue. For the subsample of companies that redeemed their auction rate preferred shares, this time-series provided the last available dividend yield and associated CP percentage prior to the redemption date. It also served as the source for the dividend yields resulting for the latest available auction for each respective company that had not redeemed its issues as of the end of the first quarter of 1991. Separately, industry classifications, parent-subsidiary relationships and the identity of limited-purpose subsidiary issues were also obtained for each issue by searching the Moody's Bond Survey and the Compact Disclosure Data Base.

Exhibit 2 summarizes select attributes of issue and redemption activity by year in the market for auction rate preferred stock over the period spanning the third quarter of 1984 through the first quarter of 1991. The reversal in new issue activity is evident from Panel A of the exhibit. Over the first three full calendar years (1985-1987) of its existence, $13 billion of auction rate preferred stock was TABULAR DATA OMITTED issued by 73 companies, with over one-half (38) originating from limited-purpose finance subsidiaries. In contrast, less than $5 billion of DARPS was issued in the last three full years (1988-1990) by 25 companies, only one of which was a limited-purpose subsidiary.

Consistent with the findings of Linn and Pinegar |8~, the population of auction rate issues displays a strong concentration of firms in the financial sector. A breakdown of the 103 issuing companies by industry classification in Panel B shows that fully one-half of the companies issuing auction rate preferred stock were either banks, thrifts or credit companies; the remaining portion of the sample was split evenly between nonbank financial institutions, industrial companies and utilities.

Panel C of Exhibit 2 details the extent of redemption activity in the market for publicly traded auction rate preferred stock. The first redemptions occurred in 1987 when three companies exercised the call feature of their respective issues. Between the first quarter of 1987 and the second quarter of 1991, 61 of the total 103 companies redeemed all or a portion of their issues. The aggregate liquidation preference value of the redeemed issues was just under $10 billion, which represents 54% of the total dollar value issued over the sample period. Redemption activity was quite strong among bankruptcy-remote issues, which represented over one-half of the companies that redeemed auction rate shares. The impact of redemption activity on the available supply of limited-purpose subsidiary issues is striking; of the 41 limited-purpose subsidiaries that issued DARPS over the period 1984-1988, 34 had redeemed by the first quarter of 1991.

When compared to the sample as a whole, redeemed issues have been smaller as measured by the liquidation preference amount, and slightly younger in terms of the time elapsed since their original issue. As shown in Panel D of the exhibit, the average (median) book value of the redeemed issues was $161 ($100) million, versus $202 ($150) million for the issues still outstanding. Redeemed shares were called at approximately 3.1 years from the time that they were issued, which was a full year sooner than both the average and median ages of the shares still outstanding.

The redeemed shares also appear to have exhibited a different risk profile at the time they were called than shares that remained outstanding, as measured by the average initial dividend yield percentage at issue and the final dividend percentage at redemption for the sample firms, which is presented, by year, in Panels A and C of Exhibit 2, respectively. These dividend percentages measure the ratio of the auction rate dividend yield to the contemporaneous 60-day AA commercial paper rate. A comparison by year shows that the average yield ratio for redeeming issues in any particular year has exceeded the ratio for newly issued shares in the same year by 600 to 1,000 basis points.

B. Results from Logit Regression

1. Analysis of the Redemption Decision

A logit regression model was employed to examine some of the factors affecting redemption activity in the market for auction rate preferred stock. In the analysis reported here, the cost of auction rate preferred equity is measured by the ratio of the annualized dividend yield determined at the latest auction to the contemporaneous 60-day AA commercial paper rate ("CP percentage"). This particular method of measurement controls for the impact of fluctuating short-term interest rates on auction rate yields and is consistent with observed industry practice.

Under the tax laws in effect in 1984, interest on commercial paper was taxed at a rate of 46% and dividends on auction rate preferred stock were taxed at an effective rate of 6.9% (reflecting the fact that only 15% of dividends received were subject to taxation at the corporate level). The after-tax returns to both taxable money market securities and tax-favored auction rate preferred stock were equal when the auction rate preferred dividend yield was set at 58% of the commercial paper rate. The effective tax rate on preferred dividends initially fell to 6.8% when the corporate tax rate and dividends-received deduction were reduced to 34% and 80%, respectively, but increased to 10.2% following a decrease in the exclusion for dividends received after 1987. Accounting for these changes in explicit tax rates leaves the after-tax returns to both taxable money market securities and tax-favored auction rate preferred stock equal when the auction rate preferred dividend yield is 74% of the commercial paper rate.

It is reasonable to expect that the time series of CP percentages for any particular issue would be influenced by changes in both the perceived risk characteristics of the shares and the explicit tax levied on their dividends. To control for changes in the CP percentage brought about by tax changes, we constructed an adjusted CP percentage variable which measures the ratio of the observed CP percentage to the no-arbitrage CP percentage implied in a risk-neutral framework.

Exhibit 3 contains the results from the logit regressions relating firm characteristics to the decision to redeem an TABULAR DATA OMITTED issue of auction rate preferred stock. For each model, the coefficient estimates from the logit regression are displayed along with their respective asymptotic (large sample) p-values, the p-value for the model and the pseudo R-square measure (which is based upon the differences between the actual frequencies in each subgroup and the estimated frequencies obtained from the regression model). Overall, the estimated coefficients are consistent with the hypotheses that redemption activity in the auction rate market was encouraged by both investor aversion to high-risk issues and by the thrift reform process.

For redeemed issues, the cost of preferred equity was measured by the last available CP percentage just prior to redemption; for issues still outstanding, it was estimated by the latest available CP percentage in the first quarter of 1991. The first set of coefficient estimates (model 1) in Exhibit 3 indicate that the decision to redeem was more likely the higher the cost of preferred equity. The estimated coefficient for the final dividend percentage was 0.031, with an asymptotic p-value of 0.048. Similar results were obtained in the second model, where the cost of preferred equity was measured by the adjusted final dividend percentage.

To test for the influence of the thrift reform process on the redemption decision, we included a dummy variable to identify limited purpose subsidiaries in the logit analysis (model 3). The dummy variable was assigned a value of one if Moody's Bond Survey or other sources of information such as the Wall Street Journal indicated that the shares were issued by a bankruptcy-remote, limited-purpose subsidiary created specifically for the purpose of issuing auction rate preferred stock. Separate consideration of the limited-purpose subsidiary issues was necessitated by the fact that their collateralized structure generally commanded a lower dividend yield than comparable (Aaa-rated) non-credit-enhanced issues.

As shown in Exhibit 3, each of the coefficient estimates in this expanded model are significant at the five percent level. The estimated coefficients were also significant when the adjusted dividend yield at issue was employed to measure the cost of auction rate equity (model 4). This result is particularly intriguing, because it appears to indicate that the market has selected against issues that were identified as high risk at the time they were sold. The results from both specifications provide evidence consistent with the hypothesis that the decision to redeem was also more likely for issues which were affected by the thrift reform process.

2. Analysis of Issue Timing

The evidence in Exhibit 3 is consistent with the natural selection hypothesis because it suggests that the market selects against higher risk firms by requiring greater dividend yields. We searched for additional evidence on this TABULAR DATA OMITTED natural selection process by examining the relationship between industry affiliation and the timing of the issue. The results show that the market has selected against shares issued by banks, thrifts and credit institutions, and in favor of issues sold by nonbank financial institutions and utilities.(18)

The set of tests reported here seek to determine whether the risk profile of firms that first issued auction rate preferred stock differ from the risk profile of companies that have sold shares more recently. To conduct this portion of the analysis, the entire sample period was partitioned into two subintervals of approximately equal length: the first spans the years 1984 through 1987; the second subinterval covers the years 1988 to 1991. We then relate the timing of each auction rate preferred issue (first or second subinterval) to a set of variables that directly or indirectly measure cross-sectional differences in their risk.

The subintervals are distinct in that issues clearly outnumbered redemptions over the first, while redemption activity dominated new issues over the second. If firms that issued early (1984-1987) were characterized by lower-credit quality than firms that have issued more recently, redemption activity in the market for auction rate preferred would appear to be explained in part by a turnover between low- and high-risk issues.

Exhibit 4 contains the results from the logit regressions relating firm characteristics to the timing of the auction rate preferred issues. The first set of coefficient estimates (model 1) in Exhibit 4 indicate that an issue was more likely to have been floated in the first subinterval, the higher the cost of preferred equity, which was measured by the unadjusted initial dividend percentage. The estimated coefficient for the cost of preferred equity was 0.144, with an asymptotic p-value of 0.000. Again, qualitatively similar results were obtained in the second model, where the cost of preferred equity was measured by the initial dividend percentage adjusted for tax changes.

To evaluate the relationship between thrift financing and the timing of an issue, we included the dummy variable that identifies limited purpose subsidiaries in the logit analysis (model 3). As before, separate consideration of the limited-purpose subsidiary issues was necessitated by the fact that their collateralized structure generally commanded a lower dividend yield than comparable (Aaa-rated) non-credit-enhanced issues. As shown in Exhibit 4, both of the coefficient estimates in this expanded model are positive and significant at the five percent level.

Additional insight into the behavior of the auction rate preferred stock market was obtained by examining the risk profiles of the firms which issued in the first and second subintervals. To conduct this portion of the analysis, we examined the information on the respective lines of business for each of the 103 companies in our sample. Each company was then classified into one of three groups (strong, neutral or weak) based upon the current financial condition of its principal industry. Companies comprising the "weak" sector were banks, thrift institutions or credit companies. A company was assigned to the "strong" sector if it operated in the nonbank financial area (such as an insurance company) or if it were a regulated utility. Neutral companies were primarily industrial concerns.

An analysis of frequency distributions shows that the contraction of the auction rate preferred market has been driven by a reversal of issue activity by "weak" sector firms. Of the 54 issues classified as weak, 47 issued their shares in the first subinterval. In contrast, the 32 of 103 issues classified as solid originated on a more evenly distributed basis: 14 were issued over the period 1984-1987 and 18 were initially offered during the 1988-1991 period.

III. Survey Evidence

To obtain additional perspective on the factors influencing redemption activity in the market for auction rate preferred stock, a mail survey was sent in July 1991 (and a follow-up survey in September 1991) to the corporate treasurer of each of the firms in our sample that redeemed an issue of auction rate preferred stock prior to that time. A total of 51 survey questionnaires were mailed (two addresses were unavailable and six envelopes were returned to us marked "addressee unknown"); 24 completed forms were returned, resulting in a response rate of 47%. The maximum response rate was reduced due to the fact that most of the 18 savings and loans that redeemed auction rate preferred were no longer in existence at the time of the mailing. In fact, eliminating the nonresponding thrifts from the sample produced an effective response rate of 65% (24 of 37).

The purpose of the questionnaire was to enrich the statistical analysis by providing information on the perceptions of financial decision-makers concerning their experiences with the entire process of using DARPS as a financing vehicle. The specific questions posed in the survey are given in Exhibit 5. Respondents were asked to react to each statement on the basis of a six point scale ranging from strongly agree (1) to strongly disagree (6). The frequency of each response is also reported in the exhibit. In the discussion that follows, we classify the responses into one of three categories: agreed (1 or 2); neutral (3 or 4); and disagreed (5 or 6).

The results of the survey suggest that the respondents initially were attracted to the DARPs market due to its lower cost and moreover, that the yield established in the dividend reset process was "fair." For example, asked to agree or disagree with the statement that "the principal reason for initially issuing DARPs was the ability to raise funds more cheaply than other sources" (Question 1), 16 of the 24 respondents agreed (a mean response of 2.417). Similarly, asked to agree or disagree with the statement "the auction reset procedure provided a 'fair' dividend yield" (Question 2), 16 of the respondents agreed (a mean of 2.409).

Questions 3 and 4 sought to determine the extent to which the decision to redeem was influenced by firm-specific factors that would influence the relative cost of auction rate preferred stock financing. For example, preferred stock, in general, is an attractive source of financing for non-tax-paying corporations when the pre-tax dividend yield is lower than the after-tax cost of borrowing. The differential reflects the tax advantage accorded preferred dividends by the dividends-received deduction. A firm might want to redeem an issue of auction rate preferred if its tax position has changed such that the dividend yield exceeds the after-tax cost of borrowing. Similarly, the decision to redeem might also be optimal if a deterioration in credit quality caused the dividend yields at auction to rise.

The responses indicate that tax position and credit risk played an important part in the redemption decision among approximately 40% of the responding companies. To explore the importance of changes in the tax position of the issuing firms, we asked respondents to agree or disagree with the statement that "the changing tax position of my firm reduced the attractiveness of DARPs." The mean response to this question was 3.667, but the responses were bimodal: ten of the 24 respondents agreed, while 14 of 24 disagreed. The survey also asked the respondents to react to the statement that "the increasing yield on DARPS due to rising credit risk was the principal reason for redeeming the security." Again, while the mean was 3.522, the responses were bimodal in nature: ten firms agreed, three expressed neutrality, and ten disagreed.

The next set of questions (5 through 9) attempted to explore the extent to which potential shortcomings in the design of auction rate preferred stock influenced the decision to redeem. These questions separately searched for the relative importance, if any, of (i) a constraining dividend ceiling; (ii) the reduced dividend exclusion; (iii) the risk of auction failure; (iv) the intertemporal volatility of TABULAR DATA OMITTED yields; and/or (v) a decline in demand among the investment community. In summary, the responses indicate that only the risk of auction failure was considered relatively important.

Question 5 asked for a response to the statement that "a dividend ceiling that was too low played a significant role in the decision to redeem." Of 22 firms responding to that statement, two agreed, two were neutral and 18 disagreed. Question 6 probed the importance of the reduction in the dividend exclusion with the following statement: "The reduction in the dividends-received deduction made our DARPS issue prohibitively expensive." Perhaps surprisingly, only one firm agreed with this statement, while 14 of 23 respondents disagreed and eight were neutral.

Survey participants were also asked whether the redemption decision was influenced by the volatile nature of dividend yields from one auction to the next and/or a decline in investor interest which increased the relative cost of auction rate preferred capital. Of the 22 firms that chose to respond to these particular inquiries, a slim majority indicated that either factor was unimportant, while the remaining respondents were split between agreement and neutrality.

Perhaps the most interesting responses were observed for our question regarding the importance of the risk of auction failure. Nine of 22 firms agreed with the statement that "the risk of auction failure played a significant role in the decision to redeem," while ten disagreed and three expressed neutrality. To our minds, the indicated importance of auction failure to issuers is surprising, because it represents the point at which any further deterioration in credit quality occurs at the sole expense of the preferred stockholders. One might speculate that the importance of an auction failure is derived from the high-profile nature of the event as an adverse signal of credit quality.

The final three questions focused on issues that were most relevant to thrifts and other financing firms, such as the use of DARPs for asset-liability management, the usefulness of special purpose subsidiaries and the importance of the thrift reform legislation. Not surprisingly, given the demise of most of the thrifts in the redemption sample, and the preponderance of nonfinancial firms among the respondents, most of the survey respondents disagreed to strongly disagreed with the importance of these factors.

IV. Conclusion

Auction rate preferred stock appears to offer a significant financing advantage relative to fixed rate preferreds, yet the size of the auction rate market has contracted since 1987. This study has sought to determine why that contraction occurred and the extent to which it reflects on the viability of auction rate preferred stock in corporate capital structures.

The design features of auction rate preferred stock suggest an exclusive suitability for transferring capital from fully taxable, risk-averse cash managers to low-risk corporations with low marginal tax rates. The empirical evidence indicates that much of the observed redemption activity can be explained by the exit of issuers that experienced declines in credit quality, making their shares unsuitable for the cash management clientele. Although a steady stream of new, high-quality issues have been underwritten concurrent with those redemptions, the market has witnessed a net reduction in size. The analysis here attributes the shrinkage to a confluence of tax and regulatory changes that drastically reduced the number of corporations for whom auction rate preferred would represent an optimal source of capital.

From a tax standpoint, a reduction in the dividends-received deduction increased the explicit tax on preferred equity, thereby increasing the pre-tax dividend yield required by corporate purchasers. In conjunction with a simultaneous reduction in the availability of nondebt tax shields, this change increased the cost of preferred equity relative to debt financing. Survey evidence suggests that a portion of the redemptions in our sample can be attributed to these changes.

Another apparent catalyst of market atrophication has been a specific set of regulatory changes affecting financial institutions that issued auction rate preferred stock through special-purpose subsidiaries. Chief among these changes was the effective elimination of collateralization arrangements that enhanced the appeal of auction rate preferred to the cash management clientele. The importance of credit-enhancement to the sustentation of the market would appear to be critical.

Finally, and perhaps most importantly, we also conjecture that much of the decline in the market for auction rate preferred stock is due to the fact that the high interest rates that helped to spawn its development have since abated, thereby reducing the rewards to both issuing and investing in auction rate shares.

1 Here we refer to this particular security by its generic name, dutch auction rate preferred stock (DARPS), or merely as auction rate preferred. Depending on the investment banking firm that underwrites the issue, DARPS is also referred to as short-term auction rate stock (STARS), dutch auction rate transferrable securities (DARTS), market auction preferred stock (MAPS) or auction market preferred stock (AMPS).

2 This is the typical percentage that defines the floor on issues sold prior to 1988 and expresses the ratio between the preferred dividend yield and the commercial paper yield that equates the after-tax yield on both issues to corporate investors under a 46% corporate tax rate and an 85% dividends-received deduction.

3 A more complete explanation of the mechanics of the dutch auction rate preferred market is provided by Alderson, Brown, and Lummer |1~.

4 Johnston and McConnell |7~ report on the "virtual extinction" of the GNMA futures contract that, at one point, was one of the most widely traded futures contracts.

5 Holders of money market preferred can apparently divest their positions at a discount in a thinly traded secondary market following the failure of an auction.

6 McCabe, Rassenti, and Smith |9~ discuss the disadvantages of a dutch auction market: (i) it is inaccessible except at the time of the auction; (ii) no bid, offer, contract or price information is available until the results of the auction are announced; and (iii) there is a transaction uncertainty because a submitted bid (offer) may be too low (high) to execute inside the supply-demand cross. They appear to attribute the September 1988 failure of the Kroger reset auction to transaction uncertainty. In reality, the failure of the Kroger auction was caused by a recapitalization plan which increased the required dividend yield in excess of the ceiling.

7 See Chambliss |5~.

8 This is clearly spelled out in the February 16, 1986 issue of Pensions and Investments Age:

"With some exceptions, issuers of auction rate preferred must carry high-quality credit ratings," said Edward R. Chambliss, vice president at Salomon. "Corporate cash managers, the prime buyers of the security, insist on quality and liquidity in their portfolios."

9 A number of significant negative news reports concerning Citicorp took place late in 1990, producing drastic effects on the prices of both the equity and debt of the nation's largest banking organization. Following company announcements of sharp declines in second quarter earnings (Wall Street Journal, October 17, 1990, p. 17) and of a reduction in staff levels (Wall Street Journal, October 31, 1990, p. 4), Standard & Poor's lowered its credit rating of Citicorp's debt securities (Wall Street Journal, November 6, 1990, p. 8). Shortly thereafter, Citicorp disclosed that delinquencies on its $30 billion mortgage loan portfolio surged to $1.1 billion, a delinquency rate that was more than double the national average (American Banker, November 24, 1990, p. 1). Citicorp saved the worst for last, however, when it reported that it expected to record a loss of up to $400 million in the fourth quarter, reduced its dividend by four percent and announced further staff reductions (Wall Street Journal, December 19, 1990, p. 3; American Banker, December 19, 1990, p. 1).

10 Weinstein |12~.

11 "FHLBB Ponders Rules on Financing Ideas," Savings Institution, September 1984, pp. 132-133.

12 "FHLBB Ponders Rules on Financing Ideas," Savings Institution, September 1984, p. 17.

13 "FHLBB Resolution Bars Consolidation of Finance Subsidiary's Preferred Stock," BNA Banking Reporter, September 19, 1988, p. 484.

14 See Barth, Benston, and Wiest |4~.

15 We interviewed the former controller of a failed savings and loan association that was in our sample. He confirmed that holders of bankruptcy-remote auction rate issues were acutely aware that there was not an absolute assurance of bankruptcy protection and that this imposed a substantial yield premium on issuers relative to dividend yields on non-credit-enhanced AAA-rated issues. The uncertainty prompted the FHLBB to adopt a resolution (No. 88-911) which became effective August 31, 1988, assuring investors that it would not use its power to take the assets pledged to secure the preferred equity in the event that it placed the parent in receivership or conservatorship. However, the resolution also listed a number of conditions necessary to prevent the preferred stock of the thrift subsidiary from being consolidated into the parent thrift. Source: BNA Banking Reporter, September 19, 1990, p. 484.

16 Press Release, Board of Governors of the Federal Reserve System, December 29, 1989. This limitation on the use of perpetual preferred stock is more restrictive than that specified in the Basle Accord which does not explicitly limit the proportion of Tier I capital accounted for by perpetual preferred stock.

17 In order to more fully explore the impact of this regulation, an analysis was made of the percentage of Tier I capital accounted for by perpetual preferred stock by banking organizations in the sample as of December 31, 1989. Based upon the data provided in the 10-K's, it appears that only one of the sample banks exceeded the 25% threshold.

18 We examined the Moody's Bond Survey to see whether any of the redeeming firms in our sample issued other securities around the time that the auction rate preferred shares were called for redemption. Of the 61 redeeming firms, we were able to identify at least 19 that floated other securities within one year around the redemption date.

References

1. M. Alderson, K. Brown, and S. Lummer, "Dutch Auction Rate Preferred Stock," Financial Management (Summer 1987), pp. 68-73.

2. F. Allen and D. Gale, "Optimal Security Design," Review of Financial Studies (Fall 1988), pp. 229-263.

3. A. Auerbach, "The Tax Reform Act of 1986 and The Cost of Capital," Journal of Economic Perspectives (Spring 1987), pp. 73-86.

4. J.R. Barth, G.J. Benston, and P.R. Wiest, "The Financial Institution Reform, Recovery and Enforcement Act of 1989: Description, Effect, and Implication," Issues in Bank Regulation (Winter 1990), pp. 3-11.

5. E. Chambliss, "Introduction to Auction-Rate Preferred Stocks," Salomon Brothers Inc., May 1985.

6. A. Houston and C. Houston, "Financing with Preferred Stock," Financial Management (Autumn 1990), pp. 42-54.

7. E. Johnston and J. McConnell, "Requiem for a Market: An Analysis of the Rise and Fall of a Financial Futures Contract," Review of Financial Studies (Vol. 2, No. 1, 1989), pp. 1-23.

8. S. Linn and M. Pinegar, "The Effect of Issuing Preferred Stock on Common and Preferred Stockholder Wealth," Journal of Financial Economics (Vol. 22, No. 1, 1988), pp. 155-184.

9. K. McCabe, S. Rassenti, and V. Smith, "Auction Institutional Design: Theory and Behavior of Simultaneous Multiple-Unit Generalizations of the Dutch and English Auctions," American Economic Review (December 1990), pp. 1276-1283.

10. M. Scholes and M. Wolfson, Taxes and Business Strategy, Englewood, NJ, Prentice Hall, 1992.

11. J. Van Horne, "Of Financial Innovations and Excesses," Journal of Finance (July 1985), pp. 621-631.

12. R.M. Weinstein, "Thrift Issuance of Auction Rate Preferred Stock," Salomon Brothers Inc., Mortgage Finance, May 1987, pp. 2, 3.

13. B.J. Winger, C.R. Chen, J.D. Martin, J.W. Petty, and S.C. Hayden, "Adjustable Rate Preferred Stock," Financial Management (Spring 1986), pp. 48-57.

Michael J. Alderson is an Assistant Professor of Finance at the University of Missouri - St. Louis, St. Louis, Missouri, and Donald R. Fraser is a Professor of Finance at Texas A&M University, College Station, Texas.