Department of Economics Uppsala University
Economics C/Thesis Work, 15 c
Authors: Victor Hansen & Erik Lindholm-Röjestål Supervisor: Lars Forsberg
Term and Year: Autumn 2018
Evaluating the Viability of Merger Arbitrage in
Nordic Equities
Abstract
This thesis aims to examine whether a merger arbitrage strategy is able to generate market neutral alpha in the Nordic region. Similar studies of merger arbitrage strategies in both the US and Australian market find market neutral alpha. To investigate the viability of such a strategy, we developed a
“Merger arbitrage portfolio” which invests in 55 deals during 2003-2017 in the Nordic equity capital market. Our findings provide strong support that a merger arbitrage strategy is market neutral, even in times of financial turmoil. An excess return is recorded, however, when estimating the portfolio with the Market Model we find no statistically significant alpha. The results are affected by large outliers. We conclude that our version of the merger arbitrage strategy is not an optimal investment in terms of its Sharpe Ratio, compared to an index using a similar strategy and the stock market.
Key words
Merger arbitrage, Nordic equity market, Mergers & Acquisitions, Risk arbitrage, Cash mergers
TABLE OF CONTENTS
Glossary 4
1. Introduction 5
1.1. Background 5
1.2. Previous studies 7
1.2.1. Summary of previous studies 7
1.2.2. Baker & Savasoglu (2002) 8
1.2.3. Bhagat, Brickley & Loewenstein (1987) 9
1.2.4. Jetley & Ji (2010) 9
1.2.5. Maheswaran & Yeoh (2005) 10
1.2.6. Mitchell & Pulvino (2001) 11
1.3. Merger arbitrage hedge funds and indices 12
2. Theory 14
2.1. The merger arbitrage strategy 14
2.1.1. Cash mergers 14
2.1.2. Stock mergers 15
2.1.3. Portfolio weighting 15
2.1.4. Factors that affect returns 16
2.2. Capital Asset Pricing Model - CAPM 17
2.3. The Market Model 18
2.4. Sharpe Ratio 18
3. Data 19
3.1. Mergers and acquisitions data 19
3.1.1. Parameters used in Eikon 19
3.2. Deal sample 20
3.2.1. Descriptive statistics 22
3.3. Financial data 22
3.3.1. Indices 22
4. Method 23
4.1. Initiating and closing positions 23
4.1.1. Multiple offers 24
4.1.2. Cash position 25
4.2. Constructing the equal-weighted MAP 25
4.3. Calculating returns 27
4.4. Further assumptions 29
4.5. Evaluating the MAP’s viability 30
4.6. Static model methodology 30
5. Results 31
5.1. Descriptive statistics 31
5.2. Static model 38
5.3. Main results 39
6. Conclusion 41
Reference list 44
Appendix 46
Glossary
Alpha: return that differentiates from what is estimated using an asset pricing model, i.e. CAPM). Also known as Jensen’s Alpha
Arbitrage: profiting from a market imbalance without taking risk Arbitrageur: an individual or a firm exploiting arbitrage opportunities Abnormal return: same meaning as alpha
Assets under management (AUM): the amount of money managed by a fund
Beta: measure of systemic risk in comparison to a benchmark or an entire market, used in CAPM Cash merger: a tender offer made entirely in cash
CAPM: the “Capital Asset Pricing Model”
ECM: Equity Capital Market
Excess return: an individual asset’s return that exceeds the risk-free return M&A: Mergers and acquisitions
Market Model: the empirical version of CAPM, also known as the Single-Index Model Market Neutral: an asset or a strategy that is not affected by systematic risk
MAP: abbreviation for the “Merger Arbitrage Portfolio” (described in this paper)
Merger arbitrage: exploiting the merger arbitrage spread, not arbitrage in its actual meaning since there is a risk of deal withdrawal. Also known as risk arbitrage
Merger arbitrage spread: the difference between a stock’s current price and the bid price revealed in the tender offer
Risk arbitrage: same meaning as “merger arbitrage”
S&P Merger Arbitrage Index: an index created by S&P Dow Jones Indices with the goal of tracking the performance of a global merger arbitrage strategy
Sharpe Ratio: a commonly used measurement of risk-adjusted return
SIXRX: SIX Return Index (Sweden), used by us as a proxy for the entire Nordic ECM SSVX 1M: 1 month Swedish Treasury bill (T-bill)
Sweetened deal: a raised offer to acquire the outstanding shares in a company
Systematic risk: a.k.a. market risk, risk that affects the entire market and not a only a specific security Tender offer: an offer to purchase all or some of the shareholder’s shares in a company
Volatility: a measurement of risk which is measured as standard deviation and commonly written as a the Greek small sigma-sign in literature and formulas
1. Introduction
1.1. Background
In a world where a total of 8 trillion dollars’ worth of bonds trade with a negative yield (Bloomberg, 2018) and where equity prices have rallied to new all-time-highs, investors are in need of an alternative source of return. Desperate in their search for yield, investors have introduced “TINA” or “There is no alternative” as their mantra, resorting to riskier investments and allocating more assets in to equities.
Could arbitrage strategies perhaps be the solution to their headaches? Practiced in different ways by hedge funds for decades (Wyser-Pratte, 2009), arbitrage has continued to be the key to shattering risk-adjusted return metrics. In the classic book Risk Arbitrage, originally released in 1971, a member of the New York Stock Exchange is quoted, stating “The big money makers of Wall Street often mask their expertise in mystery, and among the most mysterious is a cliquish band of specialists known as
arbitrageurs. On the Street, they are a peculiar group apart, noted for their ability to spot instantly tiny profits that can be jockeyed into big ones” (Wyser-Pratte, 2009, p.4).
The strategy of merger arbitrage, or risk arbitrage as it is also known, takes advantage of the difference between the stock price after the deal is announced and its final acquisition price, this difference is known as “the merger arbitrage spread”. Merger arbitrage is not an arbitrage strategy in its true meaning, due to the risk of the offer being withdrawn, resulting in a loss far greater than the deal’s expected profit. Merger arbitrage is a commonly used strategy in the equity hedge fund universe. In the US vast amounts of research regarding the viability of merger arbitrage has been conducted, where for instance Mitchell & Pulvino (2001), Baker & Savasoglu (2002) and Jetley & Ji (2010) are some of the most prominent studies. The Nordic region has not received the same amount of attention. Previous studies in the field indicates that the generated return is mostly uncorrelated with major stock market indices, which is what AIF’s (Alternative Investment Funds) are often striving for. According to Jetley & Ji (2010) the arbitrage spread has continuously been shrinking since 1990, and possibly prior. Multiple previous studies, such as Mitchell & Pulvino (2001), Baker & Savasoglu (2002), and Maheswaran & Yeoh (2005) find abnormal returns using a merger arbitrage strategy.
With the limited amount of research in the Nordic region, we intend to build on the findings of previous articles and to evaluate the viability of the merger arbitrage strategy in the Nordic equity capital market (ECM). This leads to our research question, does a merger arbitrage strategy in the Nordic ECM generate market neutral alpha?
To answer our research question, we begin by collecting a dataset containing 175 deals from Thomson Reuters Eikon (2018). We continue by excluding a large part of the deals in the dataset, mostly due to having the wrong deal financing format or the lack of an arbitrage opportunity. We create an
equal-weight merger arbitrage portfolio (further abbreviated as “MAP”), which initiates positions in 55 deals and holds until the offers are either completed or withdrawn. Subsequently, we construct a daily time series of returns including each deal, and calculate the MAP’s return. The MAP’s Sharpe Ratio, a measure of risk-adjusted return, is compared to our benchmark the S&P Merger Arbitrage Index, and the SIXRX, to gain further understanding of its viability. Lastly, a regression analysis is performed, to test if the MAP can generate a market neutral alpha.
We have chosen to limit this paper to studying cash mergers where the target’s stock is listed in the Nordic ECM, during the period of 2003-2017. Furthermore, we restrict our analysis to deals larger than 300 MUSD. Our analysis does not take liquidity constraints, currency fluctuations, and transaction costs into account, a further elaboration on our reasoning for this can be found in section 4.
Our analysis provides a strong support, in line with previous studies in other markets, that a merger arbitrage strategy is market neutral in the Nordic region, defined as a statistically significant beta close to zero. We find an excess return, but no statistically significant abnormal return. The MAP is
determined as being an unattractive investment for a portfolio manager, with a Sharpe Ratio lower than both the SIXRX and our benchmark. We can conclude that probable factors influencing our results are the lack of bids and several large outliers.
The thesis has the following layout: section 1 continues with a description of previous related studies as well as real funds and indices engaged in merger arbitrage. Section 2 describes our theoretical
framework and the general merger arbitrage strategy. Section 3 describes our data collection process, followed by a description of our methodology and assumptions in section 4. Section 5 describes the results of this thesis, followed by our conclusion in section 6.
1.2. Previous studies
1.2.1. Summary of previous studies
Multiple studies have shown that a positive alpha can be generated by the use of a merger arbitrage strategy. Monthly alpha generated by others in academia with similar strategies vary from 0,74%
(Mitchell & Pulvino, 2001), 0,84% (Baker & Savasoglu, 2002) and 1,14% (Maheswaran & Yeoh, 2005) for their CAPM comparisons. In terms of market neutrality, Maheswaran & Yeoh (2005) and Baker &
Savasoglu (2002) findings indicated that the strategy is mostly market neutral. In contrast, Mitchell &
Pulvino (2001) found that the beta rises to 0,5 in times of a market downturn.
Jetley & Ji (2010) found that the arbitrage spread has shrunk during 1990-2007 in the US. Baker &
Savasoglu (2002) provided statistics regarding the impact of different variables on the probability of successfully completing an acquisition, where among other things firm size, takeover premium, completion risk and measurements of idiosyncratic risk proved to be significant factors. The overall probability is concluded to be above 83% for successfully completing an acquisition, however, Branch &
Yang (2003) estimated the success rate of all cash mergers to be 87%.
Jetley & Ji (2010) found that the impact of transaction costs on the merger arbitrage spread is negligible, however, when transaction costs are added in Maheswaran & Yeoh’s (2005) work, the positive alpha which they previously found becomes statistically insignificant. The alpha generated in Mitchell &
Pulvino (2001) research suffered a decline of 0,29% per month by including transaction costs. On the contrary, Baker & Savasoglu (2002) provided some evidence that transaction costs may not be the main determinant in deciding how large a reasonable spread should be. Conclusively, there seems to be mixed findings regarding the effects of transaction costs.
The methodology employed in the different papers vary a bit but is similar at its core. The method used in this paper is the most similar to Maheswaran & Yeoh (2005) who analysed the strategy’s performance in the Australian market.
An interesting way of defining the characteristics of a stock subject to a M&A proposal is described by Bhagat, Brickley & Loewenstein (1987), who argued that the stock functions similarly to writing a put option after a tender offer is announced. Bhagat, Brickley & Loewenstein (1987) also found that there is a statistically significant drop in volatility after the announcement of a deal.
The samples previously used vary from 4750 stock and cash mergers (Mitchell & Pulvino, 2001) to 193 cash mergers (Maheswaran & Yeoh, 2005), all much larger than the dataset used in this paper. Mitchell
& Pulvino (2001) used the longest dataset, 1963-1998. Most other studies have used datasets ranging from 10 to 20 years. The latest year incorporated in a previous study which we refer to is 2007 by Jetley
& Ji (2010).
1.2.2. Baker & Savasoglu (2002)
Baker & Savasoglu (2002) focused on 30-day abnormal return, since deal lengths vary, and used a cross-sectional analysis. They constructed risk arbitrage positions for 1901 cash and stock merger acquisitions offers between 1981 and 1996 out of which 1335 were pure cash mergers. The paper also tried to predict merger success and what influences it. The probability that the original offer is
completed is estimated to be roughly 83%.
Baker & Savasoglu (2002) provided quantitative evidence regarding the profitability of the merger arbitrage strategy in the US market during 1981-1996. The authors concluded that a risk arbitrage portfolio, including both pure cash mergers and a mixed offer consisting of stock and cash, grants an average risk-adjusted return of 0,60% to 0,90% each month. The portfolio also beats the market outright by 0,30% per month. Some evidence is provided against transaction costs being the main determinant of returns. They also found evidence that undiversified investors sell to avoid “completion risk” i.e. that the tender offer will be withdrawn, and that the arbitrageurs requires a premium for bearing the risk. Baker
& Savasoglu (2002) also found that when capital allocated in risk arbitrage falls, subsequent returns rise.
Another interesting finding is how Baker & Savasoglu (2002) predicted whether a merger outcome would be successful (completed) or withdrawn, which was the greatest risk to the merger arbitrage portfolio’s return. If the acquirer was defined as hostile, the likelihood of an accepted offer is affected negatively. Their findings indicated that acquirer attitude explained 8% of the variation in the outcome.
In terms of market capitalization, a higher number in the target company decreased the likelihood of success. The effect of a higher market capitalization on the likelihood of a successful offer correlated positively for the acquirer. It seems from their findings that larger firms had an easier time acquiring
smaller firms. Interestingly, the authors found that a higher premium does not in a statistically significant way increase the odds of a successful offer.
1.2.3. Bhagat, Brickley & Loewenstein (1987)
Bhagat, Brickley & Loewenstein (1987) studied 295 tender offers made in cash between 1962 and 1980 in the US. They restricted their analysis so that if a firm receives multiple bids during the offer period, the analysis is focused on the first bid only. The paper used tools provided by option-pricing theory to examine wealth effects of interfirm cash mergers.
In their paper, Bhagat, Brickley & Loewenstein (1987) found that, on average, the underlying stock price increased 11,16% following an announcement of a tender offer. It is concluded that a tender offer is similar to a put option in terms of risk. Both the beta and standard deviation of the common stock subject to a tender offer showed a statistically significant drop following the announcement. During the offer period, CAPM would due to this drop predict a lower expected return, although in reality a significant positive abnormal return is found. They argued that this may be due to the likelihood that new information about the target stock may be announced by the bidder, who is expected to have found synergies, which unlocked additional value.
1.2.4. Jetley & Ji (2010)
Jetley & Ji (2010) examined M&A deals between 1990 and 2007 in the US, both completed and failed.
The authors compared the median arbitrage spreads over the first 90 trading days in three different six-year periods. They also examined three reasons for the decline in the arbitrage spread and subsequently the aggregate alpha of merger arbitrage hedge funds. These are:
1) a reduction in transaction costs related to risk arbitrage
2) capacity constraints over time (more money chasing a limited number of deals) 3) a reduction in risks associated with risk arbitrage
Jetley & Ji (2010) found that the median arbitrage spread is steadily shrinking with each trading day. The median arbitrage spread has decreased by 4 percentage points since 2002 in those tender offers which later proved to be successful. The median first-day arbitrage spread for successful deals declined from 6,39% in 1990-1995 to 1,91% in 2002-2007.
The authors found that a change (a decline) in transaction costs is not expected to impact returns (and thus alpha related to a merger arbitrage investment strategy). Hedge funds pursuing a merger arbitrage strategy have seen large net inflows since the year of 2000. Jetley & Ji (2010) states that this results in a significantly larger proportion of the outstanding shares being owned by hedge funds during the tender offers’ acceptance periods, which they see as a probable cause for the decline in merger arbitrage spreads. Completion risk is the largest factor in determining the arbitrage spread. The losses from deal failures have declined since 1990. This may be due to lower bid premiums, which on average declined from 45% for the years of 1996-2001 to 36% in 2002-2007.
1.2.5. Maheswaran & Yeoh (2005)
Maheswaran & Yeoh (2005) used a sample of 193 M&A bids in the Australian stock market between 1991 and 2000. With this, they constructed daily time series of returns on both value-weighted and equal-weighted portfolios and benchmarked it against CAPM and Fama-French three-factor model.
Active deals per month in the authors’ Australian merger arbitrage portfolio ranged from 1 to 14.
Maheswaran & Yeoh (2005) also compounded daily returns for their merger arbitrage portfolio to its monthly frequency, for a total number of 112 months. The authors’ portfolio took a position two days after the announcement of the tender offer, in order to make certain that abnormal returns which often occur on the announcement day are excluded from the arbitrageur’s portfolio.
The study found no evidence of a non-linear relationship between the returns of a risk arbitrage portfolio and the market index, i.e. the strategy is market neutral for the Australian stock market. The authors also found that profitability is severely constrained by transaction costs. Comparing their portfolio against CAPM without transaction costs, Maheswaran & Yeoh (2005) found a significant alpha for the equal-weighted portfolio of 1,14% per month, after adjusting for transaction costs this alpha decreased to 0,76%. For the Fama-French 3-factor-model they found that alpha decreased from 1,20%
to 0,82% when including transaction costs. Maheswaran & Yeoh (2005) found that annual abnormal return over the market portfolio for this investment strategy is between 2,52% to 6,52% per year, and the standard deviation is greater than the market portfolio.
1.2.6. Mitchell & Pulvino (2001)
Mitchell & Pulvino (2001) analysed 4750 mergers between 1963 and 1998 to characterize risk and return in risk arbitrage. Furthermore, the authors constructed two different portfolios and measured the return. First a calendar-time value-weighted average of returns to individual mergers, ignoring practical limitations such as transaction costs. The second portfolio mimicked the returns of a theoretical risk arbitrage index and thus included practical obstacles like transaction costs.
In their research, Mitchell & Pulvino (2001) found that a merger arbitrage strategy, in flat and
appreciating markets, generated a return of 0,5% per month (6,20% annually) greater than the risk-free rate with a market beta value of 0,1232. However, when the stock market decreases by 4% or more their market beta increases to 0,5.
Some key results found by Mitchell & Pulvino (2001) are that studies prior to theirs have not shown regarding the impact of market performance on merger arbitrage returns. When ignoring transaction costs, they found a statistically significant alpha (assuming a linear asset pricing model) of 0,74% per month or 9,25% annualized. When the authors took transaction costs into consideration this alpha declined to 0,29% per month or 3,54% annually. In most market environments Mitchell & Pulvino’s (2001) portfolio generated a modest return, their findings however indicated that a large negative return may occur in a less favourable general market climate.
Mitchell & Pulvino (2001) claim that an alpha connected to a risk arbitrage strategy cannot be measured using a standard asset pricing model since a tender offer creates a situation similar to the “writing (of an) uncovered index put options”, which asset pricing models like CAPM doesn’t capture the actual risk-return ratio for. According to their research, adjustment for non-linearity has not been done in most studies prior to theirs, showing how the return may vary with general market climate. Linearity
assumptions can be problematic since the return on a merger arbitrage strategy may vary in some degree with the general market climate (i.e. a bull- or bear trend).
1.3. Merger arbitrage hedge funds and indices
As outlined in the previous studies-segment, research on the profitability of the merger arbitrage strategy has been conducted for a long time. There has also been multiple funds using various merger arbitrage strategies, with the Vivaldi Merger Arbitragefund (Vivaldi Asset Management, 2018) perhaps being the most noteworthy one due to its performance (Eikon, 2018), with a total return of
approximately 450% between their inception in 2000 and the end of 2017. Another example of an early (also founded in 2000) and successful hedge fund using this strategy is The Arbitrage Fund (The
Arbitrage Funds, 2018). There are as of December 2018 as many as 39 funds that report their performance to the Barclays’ Merger Arbitrage Index (BarclayHedge, 2019). With at least 39 funds competing for similar deals, arbitrage opportunities are harder to come by due to the increased competition, per the results of Jetley & Ji (2010) that illustrates the shrinking merger arbitrage spread.
Hedge funds engaged in merger arbitrage use a variety of different strategies, as an example, the Vivaldi Merger Arbitrage Fund defines their strategy as follows: “The strategy seeks to take advantage of the return opportunity presented by the natural deal spread that emerges after the announcement of a merger or acquisition. Our manager employs a research-driven process focusing on predominantly North American transactions with more well-defined regulatory or financing risk. The primary goal is to look for the best risk-adjusted deals for the portfolio, focusing on strategic combinations of solidly performing targets by well-financed acquirers. The strategy is focused on running concentrated in the best ideas with a preference for shorter-dated transactions. The portfolio managers continuously analyse and monitor pending transactions for all the elements of potential risk, including antitrust, deal terms, financing and shareholder approval“ (Vivaldi Asset Management, 2018). Vivaldi’s strategy is similar to the strategy that we have used in this thesis, apart from Vivaldi taking on a more dynamic approach by employing a team of managers in order to analyse the quality of the specific bids, instead of using a strict mechanic strategy such as ours, which we further discuss in sections 2.1.3. and 5.1.
The S&P Merger Arbitrage Index tracks the performance of merger arbitrage as a strategy. It will serve as our empirical benchmark since it uses a similar strategy to ours. The index takes all forms of offers into consideration (different forms are described in section 2), not just pure cash mergers. A maximum of 40 deals can be included at a time, and only a maximum of 3% of total assets are allowed in each deal (S&P Dow Jones Indices, 2018). The index screens deals from 22 countries which provides developed markets with deliverable settlements, and includes Sweden, Norway, Denmark, Finland and the US.
Furthermore, the index only includes deals above 500 MUSD. The S&P Merger Arbitrage Index holds the US 3-month T-bill instead of a pure cash position for excess liquidity. Furthermore, the arbitrage spread must be greater than 2% in order for the index to include the deal. The Domestic Currency Return (DCR) methodology is used to calculate the index in order to disregard currency effects. The index was
incepted on the 31st of December 2005 (S&P Dow Jones Indices, 2018), which means that it differs a bit from the MAP’s starting period. Total return for the benchmark since its inception amounts to 32,7% up to the last day of our time period (Eikon, 2018).
2. Theory
In this section we describe the merger arbitrage strategy in general, various types of deals, as well as the theoretical models that will be used to test our research question.
2.1. The merger arbitrage strategy
The core idea of merger arbitrage is to exploit the spread that normally arises between a stock’s price and the bid price in a tender offer. However, multiple definitions of what merger arbitrage exactly is seem to exist. These differences stem mainly from the type of deal that the arbitrageur exploits to make a profit. In this paper we have chosen to focus on deals where 100% of the consideration is paid in cash by the acquirer. By doing this, we have excluded stock and hybrid mergers. In the book “Risk Arbitrage”
by Guy P. Wyser-Pratte (2009, p.21), he explains the core idea of merger arbitrage as “an arbitrageur is not an investor in the formal sense of the word: (i.e. he is not normally buying or selling securities because of their investment value). He is, however, committing capital to the “deal” (the merger, tender offer, recapitalization, etc.) rather than to the particular security. He must thus take a position in the deal in such a way that he is at risk of the deal, and not at the risk of the market”. What Wyser-Pratte (2009) describes here is a market neutral strategy, where an arbitrageur is only concerned about the
probability of a deal being accepted, not the movements of the market.
2.1.1. Cash mergers
As previously mentioned, we have restricted our analysis to cash mergers in this paper. This type of deal occurs when Company X makes an offer to acquire all outstanding shares of Company Y for a specific price per share, and the entire deal is paid in cash by the acquirer (Wyser-Pratte, 2009). Typically, the offer is made at a premium to the closing price of the day before the announcement day, causing a sharp increase in the stock price of Company Y to a level closer to the offer price on the announcement day (Block, 2006). The stock price appreciates dramatically due to shareholders expecting that they will receive the stated offer price for their shares. A difference in the trading price and the offer price may also arise due to shareholders having to commit capital to a deal for a period of time, losing out on opportunity cost (compensation for not receiving interest on their capital, known as “the time value of money” (Chen, 2018)). The opportunity for the arbitrageur arises when there is a difference in the trading price of the stock and the offer price. This difference normally stems from the fact that there is a small probability of the offer being withdrawn, causing the stock to fall back down to the previous
pre-offer level (Block, 2006). The size of the spread varies with the perceived probability of deal failure, which is influenced by a variety of factors. The arbitrageur buys shares in the target company, and awaits deal completion. The profit from a typical deal is the difference between the price paid to acquire the shares and the offer price.
Another way to profit from a cash merger would be to short sell a stock which is trading at a price above the offer price, similar to the way an arbitrageur would buy shares if the stock was trading below the offer price. Generally, stocks in active deals only trade above the offer price if the market expects there to be a sweetened or competing offer, which is a real threat when short selling.
2.1.2. Stock mergers
Another common type of merger is the stock merger, also called a stock-swap transaction. A stock merger is a deal where Company X uses its own shares to acquire all outstanding shares of company Y, usually at a fixed rate (Wyser-Pratte, 2009). For example, in 2003 Vestas Wind Systems (VWS) made an offer for all outstanding shares in NEG Micon A/S (NEG) at a fixed rate of 1 share in VWS per share in NEG (Eikon, 2018). As one share in NEG now can be traded for one share in VWS, their stock prices should theoretically be the same. Any difference in the stock price of these two companies would therefore be an opportunity for the arbitrageur. If, for example, the VWS share is trading at 95 DKK and the NEG share is trading at 90 DKK, the arbitrageur would sell short 1 VWS share for every 1 NEG share purchased, and await the completion of the deal. The arbitrageur’s gross profit would be the difference between the two shares’ prices. Occasionally, the consideration in a merger is paid in both cash and in shares, this type of deal is called a hybrid merger (Wyser-Pratte, 2009).
2.1.3. Portfolio weighting
In this paper we have decided to use an equal-weighted portfolio, as described in section 4.2, to analyse the performance of a merger arbitrage strategy in the Nordic ECM. However, a merger arbitrageur could use a variety of other portfolio weighting-strategies, such as a value-weighted portfolio, or having a professional portfolio manager use fundamental analysis to decide on where the greatest opportunities exist. A value-weighted portfolio would invest a share of total assets in each deal that is proportional to the difference in market capitalization of the target companies, as described in Maheswaran & Yeoh (2005). In practice, it seems like actual merger arbitrage hedge funds commonly use fundamental analysis to determine portfolio weights, such as the previously mentioned Vivaldi Merger Arbitrage Fund (Vivaldi Asset Management, 2018) or The Arbitrage Fund (The Arbitrage Funds, 2018). The S&P Merger
Arbitrage Index uses yet another portfolio weighting strategy, where they invest a fixed percentage of total assets in each deal, in their case 3% (S&P Dow Jones Indices, 2018).
2.1.4. Factors that affect returns
There are a variety of factors that affects the profitability of the merger arbitrageur. The main factors/events that affect the profitability of a deal for the arbitrageur are the following
Opportunities
- the spread between the share price and the offer price - a raised competing offer in an active deal
- a sweetened (raised) offer from the same acquirer in an active deal
- the duration of the deal (shorter deals are preferred since the profit is received earlier) Neutral
- portfolio-weighting of the individual deal Downside risks
- withdrawn offer, likely to result in a substantial loss - an amended (lowered) offer from the same acquirer
The portfolio manager of an actively managed merger arbitrage fund has to determine the probability of each of these events happening, in order to calculate the expected value of a given deal and determine if the fund should commit to a certain deal.
For the managers to estimate the probability of the events, they could rely on historical data. As an example, Branch & Yang (2003) estimates the success rate of all cash mergers to be 87%, they also estimate the median duration for all types of tender offers to be between 101-120 days. For our data, we found the success rate of all cash mergers included in our portfolio to be 96,4%. The mean duration for an active deal in our portfolio is 109 calendar days and the median number of calendar days is 82.
2.2. Capital Asset Pricing Model - CAPM
CAPM (Capital Asset Pricing Model) is a theoretical model that is used to predict the expected returns of a specific risk asset, and how this return is related to the asset’s systematic risk. The CAPM-formula is written as
,
(1)where E(R )i is the expected return of a specific risk asset, i, Rf is the risk-free rate (in this paper defined as the 1 month Swedish T-bill, ”SSVX 1M”), E(R )m is the expected return of a market portfolio (in this paper defined as SIX Return Index, “SIXRX”). E(R ) m − Rf is therefore the market risk premium (Harrington, 1987). βi is the beta of the risk asset. The beta of a specific asset describes its riskiness, and is defined as
(2)
The covariance between the return of a specific asset and the return of the market, divided by the variance of the return of the market (Kenton, 2018). By this definition, the market’s beta is 1. In CAPM, beta represents risk. An asset that is more risky (more volatile) than the market as a whole has a beta above 1, and an asset that is less risky has a beta below 1. Using this logic, an asset with a beta of 2, should according to CAPM be twice as volatile as the market. An individual stock’s beta measures its level of systematic risk in relation to its prior performance (Kenton, 2018).
In conclusion, CAPM predicts that the return of a risk asset is composed of two parts. The level of the risk-free rate, as well as the asset’s riskiness (beta), multiplied by the market’s risk premium. The next segment will explain the Market Model, and how this model relates to CAPM.
2.3. The Market Model
Using the CAPM-foundation, the Market Model is constructed. The Market Model uses historical data instead of future predictions of returns like the CAPM does, it can therefore be used as the empirical version of CAPM. The Market Model states that the returns of an asset is linearly related to the returns of the market (Harrington, 1987). The Market Model (risk premium version) can be written as
(Harrington, 1987, p.104)
,
(3)where αi, or Jensen’s alpha (Jensen, 1968) (henceforth referred to as “alpha”), is the intercept of the Market Model, and can be interpreted as the abnormal return of a specific asset, meaning returns beyond what the Market Model predicts can be explained from the asset’s beta and the market risk premium (Harrington, 1987).
2.4. Sharpe Ratio
The Sharpe Ratio (Sharpe, 1966) measures the risk-adjusted returns of a portfolio. The ex-post version of the Sharpe Ratio can be written as
,
(4)where Ri is the return of an individual asset or portfolio, Rf is the risk-free rate of return (SSVX 1M for the SIXRX and MAP, and the 1 month US T-bill for our benchmark) and σiis the volatility of the asset or the portfolio. The original, ex-ante, Sharpe Ratio uses expected returns and volatility, instead of
historical data (Sharpe, 1966). A higher Sharpe Ratio is preferred, as an investor seeks maximum returns with minimum volatility.
3. Data
This section starts with a description of our mergers and acquisitions data and ends with an overview of our financial data.
3.1. Mergers and acquisitions data
Mergers and acquisitions (M&A) data and price history has been gathered from Thomson Reuters Eikon (2018). The daily price history used for each ongoing deal when constructing the MAP is quasi-daily and not mark-to-market (see section 4.2. for a more thorough reasoning).
3.1.1. Parameters used in Eikon
To identify the deals that we use in our analysis, we imposed the following restrictions
- Form of transaction: “Acquisition of Majority Assets” and “Acquisition of Remaining Interest”
The deal has to be an offer for the remaining shares in the company, or an acquisition of a majority of the share capital. This excludes deals where the acquirer only buys, for example, 10% of the company, which does not lead to the start of a mandatory tender offer.
- Target Nation: “Sweden”, “Denmark”, “Norway”, “Finland”
The analysis is restricted to the Nordic countries, excluding Iceland.
- Target public status: “Public”
Only acquisitions of publicly listed companies are considered.
- Announcement Date Between: “2003-01-01” and “2017-12-13”
We chose 2003-2017 due to the variety of market climates that this period contains, from just after the end of the “Dot-com Bubble” (2003) to the last full year prior to this study. We believe this period gives a good understanding of how a merger arbitrage strategy in the Nordic countries would perform, including both favourable and unfavourable market climates, with 2 booms (2003-2007, 2011-present) and 2 busts (2007-2009, 2011).
- Deal status: “Completed”, “Withdrawn”
We wanted to restrict our search to deals that have been either completed or withdrawn, to avoid deals that are currently pending approval.
- Deal Size (MUSD): “Greater than 300”
“Micro-caps” are excluded from the analysis, thereby limiting our analysis to deals which values the target companies above 300 MUSD, which is The U.S. Securities and Exchange Commission’s (2013) definition of a micro-cap. We imposed this restriction in order to make our strategy more practically feasible, as a hedge fund using this strategy might have significant issues with illiquid trading in
micro-cap-stocks, which would result in them not being able to acquire the necessary amount of shares without having significant market impact.
3.2. Deal sample
Figure 1 - Distribution of deals
Figure 1 illustrates how the observations in our original sample are distributed.
Our data consists of 175 observations which we obtained by using the parameters described in section 3.1.1. After a closer look at the specifics of each offer, only 55 cash mergers remain. Of the remaining observations, the majority (42) is completed without further complications, while 11 are subject to an event. Eight of these are sweetened and three companies receive at least one competing offer before later being completed. Two offers in which we have a position are withdrawn.
The deals which we have excluded fit in to two main categories: The first category is “No arbitrage opportunity”. These 47 deals are deals in which the offer fits our specified strategy (as described in section 4.1.), but the closing price on the announcement day (where we initiate our position) is higher than the bid price. The second one is “Miscellaneous”. Most of the observations in this category are excluded due to having the wrong deal type (as described by our methodology), of which most are stock mergers. We have also excluded some of these deals due to obscurity in the deal terms or lack of price data. For example, a company being acquired multiple times during our period, which removes price history from the first period as a public company. In total, this amounts to 73 offers.
Figure 2 - Announced deals per year
Illustrates the number of deals in our dataset with an arbitrage opportunity announced each year.
Figure 2 displays the distribution for each year of the announced deals that we have included in the MAP. The years with the highest number of deals are 2003, 2006, 2007 and 2014. For four years there is only one announced tender offer, we discuss the implications of this further in section 5.
3.2.1. Descriptive statistics
The mean return, 3,20%, is moderately higher than the median return, 1,26%, which is partially due to substantial outliers. The standard deviation of 10,48%, which is rather large compared to the mean return. The highest return in a specific deal is 47,14%, and the lowest is -39,4%. The number of days needed to complete a tender offer varies substantially, ranging from 2 to 477 days. The mean duration of a deal is 109 days, and the median is 82 days.
Table 1 - A selection of descriptive statistics
Illustrates summary statistics for the dataset used in constructing the MAP.
3.3. Financial data
All data for comparing our returns to what can be explained by CAPM has been obtained from the Swedish House of Finance Research Data Center (n.d.). This provides us with daily, monthly and annual data on SIXRX, as well as data on the yield of SSVX 1M. We have used data from the Swedish ECM as a proxy for the entire Nordic region. The US 1 month T-bill is used when calculating the benchmark’s excess return. Historical data on the yield of the US 1 month T-bill is retrieved from the Kenneth R.
French Data Library (2019). As the Swedish House of Finance-dataset only contains observations up until the 2nd of January 2017, and our analysis includes the entirety of 2017, we obtained daily price data for the SIXRX and SSVX 1M from Thomson Reuters Eikon (2018) for remaining dates during 2017. During the selected period, 2003-2017 there are 3760 trading days.
3.3.1. Indices
Daily, monthly and annual price history for the S&P Merger Arbitrage Index, which we refer to as our benchmark index, has been gathered from Thomson Reuters Eikon (2018). Information regarding the index is collected from its methodology-sheet (S&P Dow Jones Indices, 2018). The benchmark’s characteristics is more thoroughly described in section 1.3.
4. Method
This section outlines the methodology used and begins with a detailed specification on which type of deals the MAP invests in, and how the MAP reacts to certain events during the holding period. This is followed by a description of how the MAP is constructed, how returns are calculated and a discussion on simplifying assumptions. Lastly, a brief description of an alternative, artificial, version of the strategy.
4.1. Initiating and closing positions
As mentioned in section 2.1.1, we chose only to invest in cash mergers. This is a simplifying assumption, as an analysis of stock mergers (section 2.1.2.) would be much more complex, requiring us to take short positions in the acquirer's stock.
We initiate our position in the closing call auction on the first day of trading following the announcement. In contrast, Maheswaran & Yeoh (2005) initiate their position two days after the
announcement of the tender offer, in order to make certain that abnormal returns which often occur on the announcement day is excluded from their portfolio. We have manually made sure that no positions are taken prior to the announcement, to avoid capturing the full bid premium in the MAP, and not just the merger arbitrage spread. The announcement date we use is the date on which the tender offer is publicly announced.
We use the price in the first day of trading’s closing call as our buying price, due to a large portion of the trading taking place in that specific closing call, which would make it more likely for us to obtain the required volume of shares. After initiating the position, we hold it until the offer is either completed or withdrawn. We assume that we receive the offer price for our shares at the date of announcement of completion. In some cases it might not be possible to sell the required volume at this date, at the specified offer price, due to poor liquidity, but we make the assumption that we can manage to sell our shares at the offer price at this date. We make this assumption as, in theory, it would make sense for the market price to be equal to the offer price at this point in time due to the risk of deal failure being very close to zero. In the case of a withdrawn offer, we assume that we close our position (selling the previously acquired shares) in the closing call at the closing price of the date of the withdrawal. This is done due to the deal in question no longer being active, and thus not relevant to the merger arbitrage
strategy. If we were to accept the offer and sell them to the acquirer instead of in the market, the proceeds would be received at a somewhat later date.
If the stock is trading at a price higher than the offer price when we would initiate our position, we do not initiate a position, as there is no merger arbitrage opportunity at this point in time. This
phenomenon is most likely due to the market expecting a future competing (or sweetened offer), but is unique for every acquisition. According to how our methodology is defined, no position is taken and the offer is therefore not taken into consideration. Furthermore, the strategy is restricted to only initiating long positions, which means that we never short sell a stock trading above the offer price, as described in 2.1.1.
If the company in an active deal where we have initiated a position pays a dividend during the holding period, the assumption is that we receive this dividend at the completion date by adjusting the bid price to include the dividend. This means that all of the potential payoff/loss from a deal is received at the completion/withdrawal date. This assumption is primarily made for simplicity reasons. We do not expect this assumption to have any material effect on our results, apart from a small negative effect on our returns since we do not receive the proceeds from the dividends at an earlier date.
An example of a tender offer included in our data is the acquisition of Cardo AB, with Assa Abloy AB as the acquirer, announced in December 2010 (Eikon, 2018). On the day prior to the announcement day, the stock closed at 283 SEK. Assa Abloy’s offer was an all-cash offer for 420 SEK per share, resulting in a bid premium of 47,9%. The MAP initiates its position in the announcement day closing call at 418.5 SEK.
The deal is completed in March 2011. The merger arbitrage spread and the MAP’s profit was 0,36%.
The strategy uses no forms of leverage or hedging.
4.1.1. Multiple offers
If there is an active offer for a given stock which we have invested in, and an additional, more attractive, offer is received in the specific deal, either from the same acquirer or a competitor, we hold our position and instead accept the new offer at its completion. In the case of the withdrawal of all offers, we sell our shares as previously described. The raised offer results in a larger portion of our funds being invested in the raised offer. As we always hold an equal-weighted MAP (further described in section 4.2.), we
rebalance the portfolio by selling some of the shares in the company that has received a higher offer, and buying shares in the other active deals, in order to hold an equal-weighted MAP.
If we do not hold a position in an active deal, due to there being no arbitrage opportunity in the first deal and a new or raised offer is received, the new offer is evaluated to see if there is an arbitrage opportunity. If there is a positive merger arbitrage spread in the raised offer, a position is initiated in the same way as described in 4.1. Bhagat, Brickley & Loewenstein (1987) restricts their analysis so that if a stock receives multiple bids during the offer period, the analysis is focused on the first bid only. Our study takes all offers into consideration, a raised (competing) offer may cause a positive outlier benefiting the MAP.
4.1.2. Cash position
As we only initiate a position in 55 deals in total over a period of 15 years, we are often left without any deals to invest in, this occurs in 1338 trading days out of a total of 3760 trading days and corresponds to 35,6% of the time period. As this would likely reflect the reality of a Nordic merger arbitrageur, a strategy on how to deal with cash positions is of importance. We have chosen to hold SSVX 1M (when it yields above 0%) when no deals are active. The S&P Merger Arbitrage Index holds the US 3-month T-bill instead of a pure cash position (S&P Dow Jones Indices, 2018).
4.2. Constructing the equal-weighted MAP
The MAP is an equal-weighted portfolio of merger arbitrage positions. We decided on this approach partly due to its simplicity, and partly due to findings from previous studies. For example, in
Maheswaran & Yeoh (2005) an equal-weighted portfolio generated a drastically higher return than their value-weighted portfolio. We believe, although provide no evidence that this difference might arise due to there being larger arbitrage opportunities in smaller companies. Baker & Savasoglu (2002) stated that a possible explanation for the disparity could be that generally both the funding and the acquirer is more credible in tender offers for larger companies, which lowers the merger arbitrage spread.
The equal-weighted MAP is constructed by dividing the total portfolio into equally sized positions, varying by the number of active deals at any given point in time. For example, we hold 100% of our portfolio in the same stock if only one deal is active, and 50% of our portfolio in two different stocks if two deals are active simultaneously, and so forth.
The portfolio weight for an active deal at a given point in time is given by
,
(5)where i = a specific active deal in the MAP, t = a specific day in the daily time series, Wit= portfolio weight in the MAP of deal i on day t, Nt=Total number of active deals on day t.
When a deal is active, and another deal is announced during the holding period of the first deal, we sell shares in the first active deal to buy shares in the second active deal, to keep a balanced equal-weighted portfolio. When rebalancing, our selling price of shares in the first active deal is the same as the price as we paid to acquire the shares, except in the case where an offer has been raised. In the case of a raised offer on the first active deal, either by sweetening (raised offer from same acquirer) or a competing offer (higher offer from a new acquirer), the selling price on the first active deal is assumed to be the price in the closing call on the announcement day of the raise. This means that we assume that the stock trades at the same price every day, unless a raised offer is announced, or the offer is completed or withdrawn. This is a simplifying assumption, but is from our own experience very close to the truth. We make this assumption as most extreme price movements leading up to the completion or withdrawal are due to low liquidity, and the most volume is traded very close to the initiation price or the deal price.
This means that our daily price data regarding ongoing deals is quasi-daily and not actual daily
mark-to-market data. As Jetley & Ji (2010) showed, the price of the stock in a deal generally converges to the offer price as the offer is nearing its completion. If Jetley & Ji’s (2010) findings hold for the Nordic stock markets, and we use the assumption described above (all trading in the stock occurs at the price in the closing call on the announcement date), some of this converging effect is generally lost when selling at a lower price than the likely true market price.
4.3. Calculating returns
We start by calculating the return of each individual deal. This is done by utilising the following equation, where we adjust the final offer price for any eventual dividends during the period
,
(6)where Rdeal= return on a specific deal included in the MAP, Pclose= the price per share that we receive for our shares when closing our position (through selling after completion or withdrawal, as specified in section 4.1.), Pinit =the price per share that we pay to initiate our position, and D = dividend during the acceptance period of the tender offer.
In order to construct our daily time series of portfolio returns and calculate the monthly returns we use the methodology of Maheswaran & Yeoh (2005). We believe that their methodology fits our purpose, for a number of reasons. First of all, we have a very detailed dataset, with extensive data for each observation, which allows us to manually analyse every deal and from this construct a daily time series over our entire period. Another key reason for using Maheswaran & Yeoh’s (2005) methodology is its ease of interpretability and accuracy. By creating a daily time series we can illustrate the MAP’s
performance over different time periods and create easy to interpret charts, which can be compared to the performance of indices and other assets.
We also considered using a standard event study methodology such as the one described in Tuan et al.
(2007), although discarded the idea after concluding that it would be less suitable for evaluating the viability of merger arbitrage. We believe that an event-study using a fixed time-period is not ideal due to the fact that the duration until completion heavily varies. In order to evaluate merger arbitrage in the Nordic ECM, we wanted to use a methodology that holds the active deal until completion or withdrawal, as this is easy to replicate and captures the full merger arbitrage spread, which makes the MAP more realistic. Furthermore, an event-study might be less realistic due to very low liquidity in between the period of the announcement and preceding the completion, not allowing an arbitrageur to sell his shares at the end of the arbitrary time period.
After calculating the return for each deal, we construct a daily time series including all of our deals. The daily return for an active deal in our portfolio is given by
,
(7)where i = a specific active deal in the MAP, t = a specific day in the daily time series, Rit =daily return for a specific deal, Pit= closing call share price for a deal on day t, Pi,t−1= closing call share price for a deal on day t-1 and Dit = dividend received from a specific deal on a particular day. (Once again, we assume that all eventual dividends are received on the completion date by adjusting the offer price accordingly.)
Therefore, the return of the MAP for any given day where we have an active position is
,
(8)where RMAP ,t= return of the MAP on day t, Wit = portfolio weight in the MAP of deal i on day t, Nt= total number of active deals on day t.
When there are no active deals (so that Nt = 0), The MAP invests in SSVX 1M. As such, the daily return of the MAP when there are no active deals equals the daily return of SSVX 1M on that specific day.
Lastly, we compound our daily time series for each month to get the monthly returns. This is done by utilising the following equation
,
(9)where RMAP , j =the return in a specific month of the MAP, j = specific month and M = number of trading days in month j.
4.4. Further assumptions
We do not take currency fluctuations into account when calculating the returns, even though we invest in multiple markets, with four different currencies. We do not expect this to have any significant impact on the results in either direction, as the positive and negative effects should mostly cancel out each other over time. This would also deviate from our goal, to evaluate the viability of merger arbitrage as a strategy in the Nordic ECM, we believe that currency effects might cause misleading results if included.
Another key simplification is the assumption that there are no transaction costs. Over the last decades, direct transaction costs have been steadily decreasing (Goldstein et al., 2009). As an example, Goldstein et al. (2009) finds that that “institutional investor in the US average cents-per-share commission” has decreased from around 12 cents per share to just under 5 cents per share in 2004. From our expertise, the same trend can be observed in the Nordic ECM. We believe that this trend has continued, mainly driven by internet brokers, fierce competition and algorithmic trading. Section 1.2.1. contains a comprehensive picture of how previous studies have treated the question of transaction costs. The study containing the most recent period, 1990-2007, from Jetley & Ji (2010) finds that the impact of transaction costs on the merger arbitrage spread is negligible. Other studies, primarily using much earlier periods of study, finds a somewhat mixed picture on the importance of taking transaction costs into account in a merger arbitrage strategy. If we were to include transaction costs it would impact the portfolio returns negatively, but less so than previous, much earlier, studies. However, this should be taken into account when considering our results going forward.
4.5. Evaluating the MAP’s viability
After having constructed our portfolio, as described above, and calculated its cumulative return, we perform a regression analysis to compare its return to what can be explained by CAPM. Using the Market Model (described in section 2.3) as the empirical version of CAPM (Harrington, 1987), we perform a linear regression analysis to estimate the MAP’s alpha and beta, our OLS (Ordinary Least Squares) estimators of the Market Model. In our OLS-estimation, the intercept is interpreted as the portfolio’s alpha, and the coefficient (slope) is the beta. Alpha is therefore the return above the risk-free rate of our portfolio when the market risk premium is 0 (Harrington, 1987). ei is defined as the error term. For a specification of our linear regression model, the Market Model, see section 2.3.
We evaluate our portfolio over three different sub periods, as well as the entire period. By doing this we hope to gain a better understanding of the MAP’s risk-adjusted return in different market climates. The periods we have chosen to study are period A (2003-2007), period B (2008-2011), period C (2012-2017) and the entire period (2003-2017). Period A and period C are sub periods with stronger market climates, whereas period B contains the financial crisis of 2008-2009 as well as the European debt crisis of 2011.
We perform a regression analysis in each of our chosen sub periods.
To further evaluate the MAP’s viability, we also study its Sharpe Ratio (Sharpe, 1966), as described in section 2.4, over different periods which is then compared to our chosen benchmark index and the SIXRX. We have chosen the Sharpe Ratio as our main way of measuring the MAP’s risk-adjusted returns.
4.6. Static model methodology
The static strategy is a different, although highly unrealistic, approach where the portfolio is fully exposed to each deal. In the static model all returns are multiplied with each other and the total return equals the product of all returns. This means that the static strategy never has to weight the portfolio differently in order to include every offer available at a given point in time. The strategy is unrealistic due to the fact that multiple offers can, and do, occur concurrently which forces an action and limits the exposure towards each specific deal in a more realistic portfolio, as the MAP is an example of. We have decided to include and make some comparisons against the static strategy as it more clearly illustrates the impact of each deal regardless of how the portfolio happens to be weighted at the specific time.
5. Results
This section starts with a selection of descriptive statistics showing our results graphically and quantitatively, followed by the results for the static model, and ending with our main results.
5.1. Descriptive statistics
As Table 2 shows, the portfolio's Sharpe Ratio is considerably lower than the SIXRX. Our benchmark, the S&P Merger Arbitrage index, consistently offers a slightly higher Sharpe Ratio than the MAP, although still only roughly half of what is achieved by the SIXRX. When studying the monthly data, we find that the MAP’s returns has a correlation of 17,81% (not included in Table 2) with the SIXRX. This low correlation indicates that the MAP’s return is mostly market neutral, which is further discussed in section 5.3.
Table 2 - Benchmarking the MAP
Illustrates key statistics of our MAP in comparison to the SIXRX and the S&P Merger Arbitrage index.
The mean return of the MAP is considerably lower than the SIXRX, though approximately twice as high compared to the return of our chosen benchmark. Neither the benchmark nor the MAP can outperform the SIXRX in terms of absolute and risk-adjusted return. However, both the daily and annual data have a slightly lower standard deviation for the MAP compared to the SIXRX, albeit much higher than our benchmark. In contrast Maheswaran & Yeoh’s (2005) portfolio found a slightly higher volatility for their portfolio compared to the stock market. A key difference between our benchmark and the MAP is that the benchmark invests a maximum of 3% of their assets in a particular deal, whereas we can invest up to 100% of our portfolio in one deal. This may be one of the key reasons for the benchmark achieving a lower volatility than the MAP.
If the largest positive and negative outlier were to be excluded from the MAP - mean return would rise very slightly to 0,545% per month
- standard deviation would fall to 2,48 %, roughly a 50% decrease
- the Sharpe Ratio would drastically increase, reaching 0,22, from the original (0,079) - total compounded return would increase to 152,34 %
These effects occur even though only 2 out of 55 tender offers are removed. This illustrates the effect of large outliers on the dataset and the MAP’s performance.
To conclude the comparison, both the standard deviation and mean return is lower for the benchmark compared to our version of the MAP, suggesting that a higher return can be achieved at the cost of a (unproportionally) higher volatility within the general strategy. The MAP’s Sharpe Ratio is consistently the lowest.
Figure 3 - Individual deal return
Illustration of each individual return in our dataset.
As our boxplot depicted in Figure 3 illustrates, a number of large outliers are included in our dataset.
These includes two withdrawn offers and five observations where the returns have been staggeringly high for this kind of investment strategy. The data is fairly normal except for these outliers, yielding a small positive return. Most of our deals earn a return between 0 to 9% with a median return of 1,26%.
When analysing our data, we found some deals where there were obvious concerns with the seriosity of the acquirer and where an actively managed fund might not have invested. In the case of the Opera Software deal of 2016, the market clearly showed little confidence in the deal, as shares of Opera were trading 32% below the offer price on the announcement day. The market ultimately proved to be correct when the offer was withdrawn (Eikon, 2018), causing the MAP to lose 17,1% of its current value.
The deal yielding the highest return was Seadrill’s offer for Eastern Drilling ASA, where we initiated our position at 91,75 NOK and sold our shares for 135 NOK, giving us a return of 47,14%. In this case, the initial offer price was 92, and the raised offer price was 135 (Eikon, 2018).
The largest drawdown in a single deal is the withdrawn offer for Cision AB in 2008. In this deal, the offer price was 20 SEK (Eikon, 2018), and we initiated our position at 19,9 SEK. The small arbitrage spread indicated that the market had a high confidence in the deal’s completion, suggesting that a possible
raise could be announced. This was not the case, after the withdrawal of the offer we sold our shares for 12,06 SEK, which resulted in a loss of 39,4% for the entire MAP.
Figure 4 - Portfolio comparison
Illustrates the compound returns of the MAP, the SIXRX and SSVX 1M. The y-axis shows logarithmic returns and the x-axis shows number of months (180 months in total).
Each index starts at 100 and its respective return during the time period of 180 months is depicted in Figure 4. By observing the above figure it can easily be concluded that the SIXRX generated the highest total return during the period. The MAP does however earn a higher return than SSVX 1M, at a
significantly higher volatility.
Figure 5 - Portfolio return (left) and active deals (right)
Illustrates the MAP’s compound return (left y-axis) and the number of active deals (right y-axis) at the specific time and the x-axis shows the number of trading days in our time period.
Figure 5 shows how the MAP’s compound returns are trending upwards, except for the two large drawdowns in 2008 and 2016 respectively, where the strategy also happened to be “all-in” in the specific tender offer at the time of its withdrawal. We had a mixed exposure by the time of our largest profits, ranging from 25% to 100%, causing the positive return from large outliers to be less impactful in our equal-weighted portfolio.
As shown by the red line in Figure 5, there are multiple times where either zero or just one bid is active at a time in the portfolio. This lack of offers to invest in is not favourable given the fact that both the mean and median deal return is positive. The red line also shows the randomness and high variance in the number of active deals in which we are invested. For example, around day 3000, we are suddenly invested in 7 active deals. Generally, we are exposed to 1 or 2 bids.
Table 3 - Number of offers active
* means that the portfolio is exposed to SSVX 1M
Table 3 illustrates how many tender offers are concurrently active in our portfolio on an aggregated level. Out of our 3760 trading day sample, SSVX 1M is held for 1338 days since there are no active deals, which amounts to 35,6% of the total number of trading days. For 100 trading days, we hold 5 or more positions, showing the uneven distribution of our deals. For example, Maheswaran & Yeoh’s (2005) portfolio had 1-14 deals active at any given point in time.
As most deals generate a positive return, we believe that the MAP’s return would benefit from having more spread out deals, and not having long periods with no deals. If we could be fully exposed during all of our deals, the MAP would have had a much greater return, as shown by our static model results in the upcoming section 5.2.
Figure 6 - Deal duration
Figure 6 shows the number of calendar days each deal is active and when, we call this the duration.
As can be seen in Figure 6, the deal with the longest duration is General Electric’s acquisition of Arcam, which lasted 477 calendar days before its completion, due to the hedge fund Elliott Management blocking the deal and demanding a raised offer (Reuters, 2017). The MAP competes with this kind of activist funds. In this case the resulting raised offer benefits the MAP, on the other hand, if more investors were to use the merger arbitrage strategy, demand would rise and no hedge fund seeking to build a corner would get the shares necessary, hurting the return of an arbitrageur.
The deal with the shortest duration was TDC’s acquisition of Song Networks, which only lasted 2 calendar days before completion, due to it being a rival offer in a bidding war between TDC and Tele2 (Eikon, 2018). As Figure 6 shows, there was a large gap between the spring of 2009 and the ending of 2010 where the MAP did not commit capital to any deals, causing the MAP to own the SSVX 1M for the entire period. This gap might be due to the aftermath of the financial crisis, where companies were less inclined to make acquisitions.
5.2. Static model
Table 4 shows the results of our static model described in section 4.6. In reality, as stated in section 5.1, multiple offers may be ongoing concurrently. The static methodology is an unrealistic approach, but offers a glance of what the merger arbitrage strategy is capable of.
Table 4 - Descriptive statistics static model
Table 4 illustrates key statistics if our dataset would be applied in a static model instead of the MAP.
“Total return” depicts the result if every individual return would be multiplied with each other in consecutive chronological order. “Years active” uses the sum of each tender offer duration (calendar days) and translates it into full years. Noteworthy, this period is longer than our total actual period of 15 years. The total return corresponds to a CAGR (Cumulative Annual Growth Rate) of 9,22%. “𝛔 days”
highlights the standard deviation of the duration, which we consider as rather high when compared to the mean duration of 109 days.
