Elsevier

Global Finance Journal

Volume 31, November 2016, Pages 1-17
Global Finance Journal

The curious case of converts

https://doi.org/10.1016/j.gfj.2016.11.002Get rights and content

Abstract

We document negative abnormal returns and abnormally high short selling in the trading days immediately before the private placements of U.S. convertible bonds. Issues experiencing greater post-placement short selling have more intense pre-placement short selling. In contrast, there are no pre-placement negative abnormal returns and less pre-placement abnormal short selling for issuers who also engage in share repurchases. Pre-placement findings are related to specific terms of the converts and related buybacks. While other potential explanations exist, the overall weight of the evidence suggests that the most plausible explanation is front-running.

Introduction

In June of 2007, four hedge funds and Deutsche Bank were fined $8.4 million by the Paris-based Autorite des Marches Financiers (AMF) after a trading probe involving the private placement of a 3-year, 8.25%-coupon, $1 billion face-value convertible bond issued by Vivendi Universal SA in November 2002. The hedge funds were blamed for using privileged information resulting from conversations with Deutsche Bank, and massively selling shares ahead of the placement (“front-running”). Vivendi shares fell 14% in the three trading days just prior to placement, and the company requested the AMF to investigate. While placement banks typically speak with potential investors to gauge demand before selling securities in the practice known as “wall crossing”, the banks are required by the AMF to inform buyers that they cannot trade on any privileged information, and the banks must store the dates and times of the conversations. Deutsche Bank destroyed tapes of those discussions. Remarkably, some of the hedge funds did not purchase the Vivendi issue.1

Similar trading restrictions apply in the U.S. private placement market including Regulation Fair Disclosure (“Reg FD”), which makes potential buyers temporary insiders upon their execution of a non-disclosure agreement (“NDA”), thereby precluding them from licitly trading on material information provided by the institutional sales and trading desks of the placement banks. Over the last decade the Securities and Exchange Commission (SEC) has filed insider trading complaints against several hedge funds from their alleged trading on material information gleaned during the wall-crossing period in the U.S. private placement market. One example is a $15.8 million settlement reached with Langley Partners involving the shorting of stock prior to the issuance of several private placements of common stock.2 Still, to our knowledge the SEC has never brought a complaint against any entity for wall-crossing (or other trading) abuse involving the private placement of a convertible bond (“converts” hereafter).

What is particularly interesting about convert issues – as opposed to other types of financing events, such as the private placement of common stock or the negotiation or renegotiation of a loan – is that the mere knowledge that the issuance of a convert is forthcoming is almost certainly sufficient information to safely profit. That is, the mere knowledge that a convert issue is in the pipeline may be regarded as “material information”, albeit not necessarily in a legal sense, because large profits can be made, with high probability, by shorting the issuer's stock prior to convert placement — regardless of the health (or lack thereof) of the issuing firm. As described in the Appendix A, the principal form of convertible bond arbitrage (“convert arb” hereafter) occurs when a hedge fund purchases the convert and reverse-engineers its straight-debt component through the trading of swaps, thus leaving the fund an isolated long position in the equity call option embedded in the convert. The arbitrageur then sells a real or synthetic call, either of which occasions the short-selling (and downward price pressure) of the issuer's stock. In the U.S., it is estimated that convert arb hedge funds purchase about 75% of convert issues (Mitchell, Pedersen, & Pulvino, 2007). Thus, the vast majority of convert issuers are subject to having their stock sold short, often intensely. Therefore, the mere knowledge that a convert is forthcoming is very likely to occasion profits by shorting the stock prior to convert placement. Put another way, one does not need to know any traditionally-defined “material information” to profitably front-run; probably more so than any other security issuance marketplace, the knowledge of issuance per se in the convert market is almost certainly highly valuable information. In addition, and per the aforementioned Vivendi case, one does not have to purchase the convert issue in order to profit from shorting the issuer's stock prior to convert placement. Under the current rules (Reg FD), it appears that the following scenario can (and does) take place: an underwriter contacts a potential buyer, stating that a firm is coming to market with a convert, and asking if the buyer is interested, with perhaps an accompanying NDA; the buyer can merely reply “no thank you” and then proceed to short the firm's stock, and profit.3

Front-running alarms regulators and other market participants because it erodes public trust in capital markets which in turn can raise the cost of capital, increase trading costs, and decrease economic growth (Bhattacharya & Daouk, 2002).4 This study provides the results of a systematic investigation of potential trading abuse in the U.S. convertible bond marketplace for the period 1997–2010. Our results are consistent with market participants actively trading prior to the placements of U.S. converts that are likely to have been sold to hedge funds.5 We document that the stocks of convert-issuing companies have significantly negative abnormal stock returns prior to the placements of their converts, provided that these companies do not engage in a related stock buyback. For our focal sample period of 2005 and 2006, a period for which we have reliable short selling data, and for the event window − 5 trading days to − 1 trading day immediately before placement, the cumulative abnormal stock return associated with convert issues (when the issuing company does not engage in a related stock repurchase) is − 1.60% (significant at the 1% critical level). Furthermore, these issuers' stocks experience significantly high levels of short selling (at the 0.1% critical level) for the same event period. On a daily average, the excess shares (relative to normal trading days) being shorted exceeds 0.25% of the total number of shares outstanding. These findings are more pronounced for the event window [− 2,− 1]. When we expand the sample period to 1997 through 2010, abnormal return results remain qualitatively similar.

We also document that pre-placement abnormally negative returns and short selling are more pronounced for the convert issues subject to a more intense level of post-placement convert arb. If there is front-running, the pre-announcement CAR (pre-announcement short selling) should be more negative (intense) the larger the issue size relative to shares outstanding; however, because only a small tranche of an issue may have been sold to arbitrageurs, post-issuance shorting intensity better captures convert arb activity than does issue size. Post-issuance shorting intensity also reflects the issue's Delta; an issue subject to arbitrage may have little value to pre-placement traders if Delta is low. Specifically, we show that for convert issues where post-issue stock shorting is particularly intense (top quintile sample of post-placement abnormal shorting), the average pre-placement abnormal stock return is − 2% for event window [− 5, − 1] and the level of pre-placement abnormal stock short selling is more than three times that of other convert issues. As discussed more within, this evidence helps to refute alternative explanations for the pre-placement findings, including the potential capability of hedge funds to predict forthcoming convert issues.6 For these findings to be explained by hedge fund's ability to predict convertible issuance, not only would the funds have to be able to predict which firms are more likely to issue converts and when a convert issue is going to occur, the funds also would have to be skilled at predicting which issuers are going to be subject to more intense stock shorting. It is extremely difficult to envision how hedge funds could predict shorting intensity; that would require the ability to predict the size of the issues, the Deltas of the issues, the amounts of the issues that are sold to convert arb hedge funds versus other convert buyers, and more.7 The documented relation between post-issue shorting intensity and pre-placement trading suggests that those trading prior to placement are aware of various specifications (valuable details) of the convert.

By contrast, for convert issues where the issuing firm repurchases some of its common stock outstanding (such a convert issue-stock buyback combination is known as a “happy meal”), the pre-placement abnormal stock return is not statistically significantly different from zero. In addition, for happy meals the level of pre-placement abnormal stock short selling is lower than that for non-combined converts. These findings complement the results described in the preceding paragraph and are consistent with the inference that unusual trading occurs in the convert pre-placement marketplace; after all, if the downward stock price pressure from short selling due to convert arb is anticipated to be offset by upward price pressure due to a related stock repurchase by the issuing firm, then traders aware of forthcoming happy meals are not as incentivized to assume a pre-placement short position in the issuer's stock, relative to non-combined convert offerings. The stock repurchase provides buy-side orders to match the arb-induced short-sale orders, thus offsetting the anticipated post-issue downward price pressure and eliminating the incentive to short the stock prior to convert placement.

Happy meals offer a rare opportunity to examine typically undetectable forms of potential trading abuse. There are two forms of trading upon privileged information that are nearly impossible to detect. One occurs when a trader who wants to buy a security is illicitly informed that he should not engage in the purchase, because bad news is forthcoming. The other occurs when a trader who wants to short a security is illicitly told that he should not, because good news is forthcoming. With both of these forms of abuse, no actual trading takes place and, thus, it is essentially impossible to detect the leakage and use of privileged information. Happy meals potentially allow us to inspect this non-trading phenomenon. Because non-happy meal private placement converts have more significantly negative pre-placement abnormal returns and more abnormal short sale volumes than happy meals, it seems reasonable to conclude that traders who would otherwise short the stock are not shorting happy meals due to the knowledge that the post-placement stock price is going to be supported via a stock buyback program. Possibly unscrupulous traders who are made aware that a convert issue is in the placement pipeline normally will short the issuer's stock, but these same traders may not short the stock if the convert issue is complemented by a stock repurchase. Our findings suggest that “non-trading abuse” may occur in the U.S. convertible bond private placement market. Indeed, we document that pre-placement anomalies wane as the following measure (“NET”) decreases: the difference between dollar amount of convert proceeds and the dollar amount of stock repurchases, scaled by market capitalization. This finding further supports the notion that the ability of hedge funds to predict forthcoming issues is not a likely explanation for our pre-placement findings. After all, for this happy-meal finding to be explained by hedge fund prowess, the funds would have to be able to accurately discern between issuers likely to engage in happy meals and issuers likely to engage in non-combined offerings. It is more likely that traders are being told that issues are either happy meals or non-combined, prior to convert placement.

As discussed more within, unusual pre-placement trading in the U.S. convert market is further supported by the results of robustness and endogeneity checks. Results persist when using an alternative definition of abnormal short selling, when changing the window of time that distinguishes happy meals from non-combined convert offerings, and when varying other relevant metrics. The “non-trading abuse” argument regarding happy meals is subject to an endogeneity problem, namely that the issuer's choice to design the offering as a happy meal may be non-random. Happy meals may be engineered to assist arbitrageurs in obtaining their required short positions. For issuers for whom it is difficult to engage in substantial open market short sales (because their stocks have low “float”), the lower pre-issuance short selling documented for happy meals is therefore also consistent with the reason for the existence of these packages, and thus might have nothing to do with trading abuse, i.e., front-running. This endogeneity problem makes the two inferences – non-trading abuse and non-random convert design – difficult to disentangle. To address endogeneity, we examine five firms that issued both happy meals and non-combined converts on two separate occasions over 2005–2006. This way, we isolate firm-specific factors and simply compare happy meals with non-combined issues. Similar to our cross-sectional results, we find that the differences between happy meals and non-combined issues persist within-firm.

At this time we impart four comments: First, we have no proverbial “smoking gun” proving that front-running occurs in this market. We have no emails or deposition testimony or the like. Thus we are careful to temper our inferences throughout. Most notably, we only suggest that front-running may be occurring. Other explanations exist. Front-running is, in our opinion, merely the most likely explanation, as suggested by the weight of the evidence. Second, if front-running is occurring, then we emphasize that it may not be hedge funds, at least entirely, that are engaging in such trading abuse. Insiders of issuing firms as well as the proprietary trading desks of the placement banks may be the front-runners. We do not assume that hedge funds actively participating in the private placements are necessarily the ones who are engaging in front-running. We do not establish, nor attempt to establish, the channel of information leakage. If front-running is occurring, then we leave it up to future researchers and regulators to establish this channel. Third, while we establish a correlation between pre-placement abnormal returns and pre-placement abnormal short selling, we do not establish abnormal shorting prior to stock price declines. Finally, we are confident that we have established the correct announcement and placement dates. In this market, announcement and placement dates typically occur simultaneously. Our day-0 abnormal return is significantly negative and a cross-check with Factiva news confirms this simultaneity.

The remainder of this paper is organized as follows. Section 2 describes related studies and expands on our contribution to the literature. In Section 3 we describe our data and sample construction process. In Section 4 we present our tests and principal empirical results based on our focal sample. In Section 5 we address alternative explanations for our inference of unusual pre-placement trading. Section 6 concludes with suggestions for future research.

Section snippets

Related literature

Recent academic research raises troubling concerns about the level of the playing field in U.S. capital markets. Five studies in particular have provided evidence that financial institutions, including dealer banks and hedge funds, may be engaging in trading based upon privileged information. Bodnakuk, Massa, and Siminov (2009) investigate the potential of privileged information leakage around merger and acquisition events and show that deal advisors who have a position in the target before the

Data and sample construction

We initially obtain information on convertible bond issues and common stock share repurchases made by U.S. publicly-listed firms for the calendar years 2005 and 2006.13

Empirical tests

We investigate the following four questions. Empirical evidence affirming these questions indicates the prospect of trading abuse:

  • Q1: Do non-happy meal private placement converts have abnormally negative pre-placement returns?

  • Q2: Do non-happy meal private placement converts have abnormally high pre-placement short selling volume?

  • Q3: For non-happy meal private placement converts, are any abnormally negative pre-placement returns more pronounced for converts subject to intense post-issuance short

Alternative explanations

There are other potentially plausible explanations for our main findings, and we caution that trading abuse is not demonstratively proven in this study. Indeed, we explicitly interpret our findings as merely consistent with trading abuse. Still, we believe that the weight of the evidence points to trading abuse as the most likely explanation.

An alternative explanation for our key findings of higher pre-placement short selling and more negative pre-placement abnormal stock returns for

Conclusion

Most private-label convertible bond issues are purchased, at least in part, by hedge funds engaging in convertible bond arbitrage. Unlike more traditional debt instruments, these private-placement borrowings occasion little if any due diligence by the lenders. The mechanics of the trading strategy entail the shorting of the issuing firm's stock, which in turn occasions downward price pressure on the stock. Thus, traders who are able to short the stock prior to most convertible bond placements

References (38)

  • N. Massoud et al.

    Do hedge funds trade on private information? Evidence from syndicated lending and short-selling

    Journal of Financial Economics

    (2011)
  • D. Mayers

    Why firms issue convertible bonds: The matching of financial and real investment options

    Journal of Financial Economics

    (1998)
  • J. Stein

    Convertible bonds as backdoor equity financing

    Journal of Financial Economics

    (1992)
  • V. Acharya et al.

    Insider trading in credit derivatives

    Journal of Financial Economics

    (2007)
  • H. Berkman et al.

    Hole in the wall: Informed short selling ahead of private placements

  • U. Bhattacharya et al.

    The world price of insider trading

    Journal of Finance

    (2002)
  • A. Bodnakuk et al.

    Investment banks as insiders and the market for corporate control

    Review of Financial Studies

    (2009)
  • M. Brennan et al.

    The case for convertibles

    Journal of Applied Corporate Finance

    (1988)
  • D. Brophy et al.

    Hedge funds as investors of last resort?

    Review of Financial Studies

    (2009)
  • Cited by (3)

    • Short selling patterns in cross-listed stocks

      2021, Global Finance Journal
      Citation Excerpt :

      Our final dataset covers a total of 3161 earnings announcements, of which 2661 are mandatory and 500 are voluntary. We test our hypotheses using event studies of short selling intensity (e.g. as in Li, Lin, & Tucker, 2016) on the A- and H-share markets around mandatory earnings announcements. To set up our event studies we first separate negative and positive earnings announcements using our ΔEPS measure from Eq. (1).

    The authors thank Jeffrey Coles and Jennifer Huang for their comments. All errors are our own.

    View full text