1
Master Thesis
The Private Company Discount on the European Market
By Johan Dahl
MSc. Finance
Faculty of Business and Economics
University of Groningen
Abstract
This paper investigates if an illiquidity discount exists on the European market using 127 private acquisitions matched with one or two public counterparts between 2002 and 2014. I find evidence
that supports the existence of this discount and that it is larger than the one on the US market. Furthermore I observe if company size, operating industry and the financial crisis have any impact on
the discount. I find that two of the variables, namely company size and industry, influence the discount. Larger companies and companies in the financial sector are acquired for a smaller discount
than small and non-financial companies.
Author: Johan Dahl Studentnr: S2739496
Email: Johan.m.dahl@gmail.com
2
1. Introduction
1.1 Background
In the modern economy as of today many companies are heavily involved in different types of mergers and acquisitions. It could be of the sort of buying a whole entity, acquire a division or buy assets. The practice involves a wide range of different operators, such as firms, banks and
institutional investors (Deepak et al. 1992). A common denominator of the M&A industry is the preference towards liquidity (Johnson, 1999), meaning that companies prefer assets that are easily sold and liquidated. Large public companies traded on an active market have higher share prices than similar companies on small over the counter markets (Damodaran, 2005), creating an illiquidity discount for the acquisition of private companies. Today the private company acquisitions account for a majority of the total M&A deals (Officer, 2007). This paper strives to investigate if an illiquidity discount exists on the European market, and its associated characteristics. A better understanding of this particular element of M&A could be useful for institutional investors, investment banks and regulators, participants that all have a significant influence on the practice.
According to Bekaert et al. (2013) the European financial integration has increased with the introduction of the euro. This integration and to some extent, standardization, of the European financial market implies that a sample collected from all over the continent would be rather comparable and not to biased from country effects.
The vast majority of the companies that exist within the euro zone is privately owned. Valuing these firms present numerous challenges since the availability of information is limited. Public companies are demanded by law to conclude a very detailed disclosure of their financial information.1 This demand of information revelation is not as present for an equivalent private company.
When there are no publically traded bonds and shares it is harder to assess the true value of the company. No prices can be observed on the market since no trade is being conducted with the firm’s assets. It is also not possible to value the firm by observing other private companies in other
industries since those too, lack properly disclosed information.
This deficiency of existing public trading of the company is also an indicator of the lesser liquidity associated with private businesses. The theorized discount would then be viewed as a result of this absence of information and liquidity.
The importance of understanding liquidity in assets can be illustrated by the Financial Crises we experienced recently. In July 2007 the investment bank Bear Sterns announced the disruption of two of its hedge funds. This was the starting point of a chain of events that would unfold to what we would later refer to as the Financial Crises. The distress was the result of a wide range of different factors colluding to create a crisis so big that it almost sunk the global financial system. However, the biggest single cause was the mistreating of assets on the banks’ balance sheets, (Zandi , 2009).
3
When the financial institutions realized that the assets they possessed were nowhere near as valuable and liquid as they first thought they tried to sell them off and buy assets of higher quality, an action that tumbled the price downwards, creating a spiral of even further price decreases were the banks eventually were only able to sell to “fire-sell-prices”, with more distress as a result (Hull, 2012).
The enormous consequences created by the miscalculation of the level of liquidity led to dramatic changes in the financial industry regarding loans (Ivashina & Scharfstein, 2010) and practices (Akbar et al. 2013). These consequences stress the importance of a profound understanding of assets liquidity and associated discounts. Acquisitions are something that have prevail as a strategy amongst companies during several centuries (Alberts & Varaiya, 1989) but have experienced some changes and declines in the wake of the crisis (Hufbauer & Skogvaard, 2011).
Numerous approaches are established in existing research to calculate the discount for illiquidity. Since the assets of the company can not be directly observed other methods have been developed to establish the discount. One way to measure the discount for illiquidity is to compare equivalent multiples derived from the Enterprise Value for a private and a public company in an acquisition. Koeplin et al. (2000).
The hypothesized difference in the multiples is the existing illiquidity discount. I will use this method to investigate whether this discount exist on the European market.
The massive impact that the Financial Crises had on the world economy motivates a sub objective that compare the allegedly existing illiquidity discount pre- and post- crises.
To further investigate the differences within the sample I will observe other possible determinants and their relationship to the discounts, such as size and industry effects. Evidence shows that size and discounts are highly correlated with each other. Comment (2012)
Schlingemann & Stulz (2002) find that companies prefer divesting assets from more liquid industries and Amhiud & Mendelson (1991) determine that companies pay higher prices for more liquid assets and securities.
In addition to the lack of liquidity associated with private companies there are other existing factors that might influence the discount. An observation of the discount itself together with its possible differences in relation to said additional variables creates the following four research questions.
- Does an illiquidity discount exist on the European market?
- Has the possible illiquidity discount changed after the Financial Crises? - Is the possible illiquidity discount dependent on size?
4
I will observe private and public company M&A deals from the zephyr database and compare different multiples based on the Enterprise Value to establish if there is a discount. The Enterprise Value in this context is defined as the price paid for the company. To see whether other
5
2. Literature Review
Numerous parameters might affect the private company discount. In this section I will describe and discuss these possible determinants of the theorized discount, further on I will describe the chosen approach for measuring the discount together with a brief description of other approaches that have been created to measure illiquidity discounts. Lastly the existing empiric results in the field will be discussed.
2.1 Theoretical background for the illiquidity discount
As previously stated I will investigate if an illiquidity discount exists for private company acquisitions on the European market. This possible discount is not only a consequence of the lack of liquidity but can be caused by other determinants as well. In order to explain the empirical results regarding the discount a theoretical framework for the different possible determinants is needed as a basis. This section will focus on the theories surrounding these determinants of the illiquidity discount separately.
The main reason for the existence of this discount is due to the lack of liquidity associated with private companies in comparison with publically traded businesses. The fact that investors prefer liquidity has been cemented for decades, Modigliani (1944) described this preference for liquidity, and its relationship with interest rates and money.
Evidence from widely spread sections of the financial market states that investors and buyers have a clear preference for liquidity in investments and acquisitions. Johnson (1999) points to liquidity as an important factor in decision making regarding Mergers and Acquisitions. Acharya & Pedersen (2005) create a model for asset pricing with respect to liquidity risk, they find evidence of both the
importance of liquidity in said pricing process and as a safe haven for investors in times of crises. 2.2 Empirical evidence of the determinants impact on the discount
The stock market in Europe shows a clear relationship between the liquidity and premiums, that is, market participants are ready to pay more for assets or companies with higher levels of liquidity associated with them. (Hamon & Jacquillat, 2002)
When distinguishing between decision making of investing in private or public entities or equity instruments liquidity seems to be an important issue, even the dominating factor, for the discount that is associated with the placements in privately owned companies (Hertzel & Smith 1993) The Financial Crises stirred the M&A markets globally, companies were inclined to liquidate assets but had problems with finding buyers since lending were restricted. This created an overall downturn in both acquisitions and capital flows. (Grave. et al. 2012)
When entities are being considered for an acquisition or merger the discount for private companies that are inflicted on the deal are larger for targets that are smaller in size. This correlation between size and the associated discount seem to be consistent where larger discounts in general are inflicted on smaller companies compared to comparable private counterparts of greater size.
6
When a company seeks to sell assets or branches of the entity they are more inclined to divest segments that operate on markets holding higher levels of liquidity. (Schlingemann & Stulz (2002). Firms that are in a position of buying firms, divisions or assets pay higher prices for those items that are more liquid. (Amhiud & Mendelson, 1991)
2.3 Methods for measuring the illiquidity discount
2.3.1 The acquisition approach
There are several different methods to measure and determine the illiquidity discount in companies. The acquisition method bases its estimated illiquidity discount on the hypothesized difference in trading price between a private and a public company. In doing so a private acquired company is matched with one or more equivalent public companies with similar size and industry, and the apparent differences in multiples are used as indicators of an existing illiquidity discount. The formula used to determine the discount is:
1-(private company multiple/public company multiple) (1)
The calculated value from this formula will be the discount in percentage that is paid for the private company compared to the more liquid public counterpart.
Multiples used for estimating the discount with this approach are usually based on the Enterprise Value (EV) divided with different sets of earnings, sales or book values.
The weaknesses surrounding this approach are straight forward. Koeplin et al. (2000) recognize the possible disparities of the results when comparing multiples from a wide range of countries and regions. Dissimilar accounting rules might result in different multiples which limit the validity of the comparison. Additional to this there might be specific characteristics associated with the firms being compared such as size, risk environment and growth rate that further hardens the discount
estimation.
Lie & Lie (2002) recommend adjusting multiples based on Enterprise Value to earnings or sales since large holdings of cash and cash equivalents within the company increase EV but leave earnings and sales unaffected, resulting in distorted multiples.
2.3.2 Other approaches
In addition to the above described acquisition approach other methods to measure the illiquidity discount have been established.
The restricted stock approach uses the difference between prices of restricted and common stock to calculate the discount.(Silber, 1991) Differences between share prices prior to and after an Initial Public Offering are also used as a benchmark for a possible illiquidity discount. (Emory et al. 2002) The two remaining approaches existing in this field of research focus on option pricing theory
7
2.4 Empirical Results
One of the earlier papers on the illiquidity discount using the acquisition approach was Koeplin et al. (2000) who collected data from 84 private companies and paired them with a public counterpart based on size and industry, determined by the Standard Industrial Code (SIC). If more than one corresponding public company existed the one with revenue closest to the targeted private entity was selected.
The authors used four different multiples divided in two separate sub categories. Enterprise Value (EV) to Earnings before interest and taxes (EBIT) and Earnings before interest, taxes, depreciation and amortization (EBITDA) were used to exclude any impact from the capital structure. Two similar companies with closely replicated operating cash flows may have large differences in the net profit because of different weights of debt respective to equity. A highly levered company might for example enjoy tax shields due to interest deductibility. By using ratios that exclude the impact of taxes and interests the comparability over the sample spectra increases.
The other two multiples used were EV to Book Value (BV) and EV to Sales. The EV/BV multiple describes how much of a unit that was paid for every unit of capital invested in the company. The reasoning behind the choice of the EV/Sales multiple lies in the, for companies, usefulness of knowing how much they pay for every additional unit of sales. Koeplin et al. (2000) motivates the choice of using these four multiples by emphasizing the common usage of them in Mergers & Acquisitions as a tool to measure the correctness of the price being paid.
The study concluded a trading discount from the different multiples ranging from 28 per cent to -2 per cent for domestic acquisitions and 34 per cent to 53 per cent for the foreign sample.
The paper written by Block (2007) closely resembled Koeplin et al. (2000) in their methodology, however the author extended the research by include a fifth multiple and an additional industry, EV to Earnings per share (EPS) and the financial industry. The paper observes 91 acquisitions on the US market from 1999 to 2006 and finds an average discount of 20-25 per cent for private companies. Block (2007) concludes that industries with more liquid assets trade for a smaller discount compared to those with more illiquid assets.
Instead of matching one private company to one public counterpart Kooli et al. (2003) use a somewhat different matching methodology than the two above mentioned papers. The authors criticize the matching process used in prior research and intend to avoid the, according to them, noisy and subjective procedure of choosing and matching a private company transaction to one corresponding public firm. They believe this procedure is victim to elements that might dilute the results. If the private company is matched with a public firm the calculated discount might not reflect the reality in a sufficiently satisfying way.
8
The authors use Price/Sales, Price/Earnings and Price/Cash Flow as multiples in their research and find a median discount between 17 per cent and 34 per cent.
Research on the restricted stock illiquidity discount has been conducted since the 70’s. Gelman (1972), Morony (1973), Maher (1976) and Trout (1977) all observed companies and transactions between 1968-1973 and determined the average discounts to lie in the interval between 33-35.4 per cent.
A slightly more recent study using the restricted stock approach was performed by Silber (1991) when he observed the illiquidity discounts on 69 companies between the years 1981 and 1988. Even though the range of his sample varied a lot his result of an average discount of 33,75 per cent were similar to the earlier papers using a similar approach.
One important conclusion derived from Silber (1991) were the notion that companies that were to be considered more healthy, with higher revenues and earnings, traded at a smaller discount. Instead of focusing on differences between restricted share prices and common share prices Emory et al. (2002) studied prices pre- and post-Initial Public Offerings. The IPOs were executed between the years 1980 and 2000 dispersed over nine different studies were the authors found a median discount of 47 per cent for the whole sample.
The study experienced criticism from Feldman (2005) who argued that a mark down have already taken place since most transactions of shares prior to an IPO are between insiders, hence the discount for lack of marketability is overstated due to an already existing discount.
A different approach using the mechanics of financial instruments, namely a put option, was designed by Longstaff (1995) in his paper when he measures the discount due to lack of
marketability. He assumed that the theorized put option holder had perfect market timing and the discount was then measured as the difference between the highest possible price during the holding period and the value when the security could be liquidated.
In line with existing option pricing theory relevant variables such as volatility of the earnings and length of the holding period were used in the model. Longstaff (1995) found that higher volatility and longer holding period increased the discount for illiquidity.
Longstaff’s (1995) findings of the discount when using two years as holding period and a volatility between 20-30% are fairly similar to earlier researched conducted on illiquidity discounts with other approaches.
Amihud & Mendelson (1986) identified the discount as the transaction cost in an immediate realisation of the asset or security. They observed monthly returns of securities on the New York Stock Exchange between 1961 and 1980.
9
3. Data and Methodology
This paper intends to measure the private company discount on the European market. I will do so by following the structure of Koeplin et al. (2000) and how they measure the illiquidity discount. This approach depends on multiples based on the Enterprise value of public and private companies, information that in many cases is available for the public. By using this approach I do not have to rely on data that might only be accessible for insiders, such as share prices pre IPO or the number of preferred stock in a private company. If a discount is proved it is presented in percentage which is easy to comprehend and put in context for a reader.
3.1 Data collection
The sample will be collected from the Zephyr database of Mergers & Acquisitions where both private and public company deals can be found. I will select acquisitions deals conducted within the
European market between 2002 and 2014. The reason to begin in 2002 is to avoid possible distortion of the sample created by the Dot-Com-Bubble that climaxed just at the turn of the millennium.
3.2 Matching process
The matching process is a crucial part in this thesis since it forms the foundation for the sampling of multiples and measurement of the discount. I will largely follow Koeplin et.al (2000), Kooli et al. (2003) and Block (2007) in how they match the private companies with a public counterpart that closely resembles the private company in terms of industry and size. I will select public companies with the same two UK Sic codes. In addition to this I will, if possible, add one public company to the match, thus creating a portfolio of sorts. This follows the portfolio selection established by Kooli et. al (2003). I will create an average multiple of the two public firms. This intends to decrease the possible noisy process of selecting one specific public entity to a private counterpart since the variation in important parameters such as size might be substantial. By creating an average matched multiple I hope to create a more similar match. I will make the assumption that two public companies, combined together, in general better resembles the conditions of its one private counterpart. The broader industries based on Zephyrs industry classification are the following six:
Manufacturing Financial Sector Services Retail Real Estate Other
3.3 Formula
To acquire the actual discount I will use the following formula:
10
3.4 Ratios & Multiples
I will use three different multiples to observe if a private company discount exists on the European market. The multiples are based on four different ratios that will be described more closely below.
Enterprise Value (EV): The EV of the company is defined as the price paid for the entire company, or
each share multiplied with the number of them outstanding (this is commonly referred to as the market capitalization) added with the debt, and subtracted with cash and marketable securities.
Earnings Before Interest and Taxes (EBIT): EBIT is the profit excluding taxes and interest, where the
latter is dependent on the capital structure.
Earnings Before Interest, Taxes, Depreciation and Amortizations (EBITDA): EBITDA is the profit
excluding taxes, interest, depreciations and amortizations, a ratio that also excludes the impact of the capital structure of the company together with non-cash deductibles.
Sales: Sales are computed as the total revenues of the company.
The multiples used for the calculation of the illiquidity discount will be EV/EBIT, EV/EBITDA and EV/Sales. The computations of these ratios are straight forward. It is just the EV divided by EBIT, EBITDA or sales respectively. All ratios are calculated by dividing the Enterprise Value that is given in Zephyr as the EV of the private and public company or companies at the time of the acquisition, with the EBIT, EBITDA or sales figures from the Financial Report the year prior. Kooli et al. (2003) discuss the problem in using the mean of the discounts as the main measurement for the results, the main critique is the large impact that extreme outliners have on the results. Matsa & Miller (2011) recommend removing the upper and lower one per cent tails of the sample in order to retrieve a more representative result. Kooli et al. (2003) acknowledge the same issue but use the median as the main measurement tool in order to exclude the impact of extreme outliers. Since Kooli et al. (2003) closer resemble the methodology and goal of my own research I will follow their recommendations and focus on the median. I will report the mean as well but will not put any emphasizes on it in the conclusions regarding the overall possible existent of the discount on the European market.
3.5 Hypothesis and testing of hypothesis
3.5.1 The illiquidity discount on the European market
To be able to test the hypotheses I will first decide if there actually exists a private company discount. I will do so by observing the whole sample and the possible differences between the matched private and public companies.
H0 – No discount on private company acquisitions exist
11 3.5.2 The illiquidity discount and size effects
There is a theorized relationship between size of the company and illiquidity discounts. I will test this possible relationship by comparing the mean and median of the discounts (if they exist) between different size groups of the sample. I will divide the sample in private companies with their counterpart having an Enterprise Deal Value bigger and smaller than 20 million euro respective.
H0 – The discount is the same for all sizes
H1 – The discount is larger for smaller firms
3.5.3 The illiquidity discount and the Financial Crises
The third hypothesis will be tested by comparing companies before and after the Financial Crises. The crises started year 2007 and unfolded during several years. Zandi (2009)
My time frame consists of 13 years in total, to get an even as possible separating of the sample but still have a clear dividing line of pre- and post-financial crisis I will select 2008 as the separating year. Hence, the period 2002-2007 is the pre-crisis category and 2008-2014 are the post-crisis category. I will observe the mean and median of both sub samples and compare to see if any apparent difference can be concluded.
H0 - The discount has not changed after the Financial Crises
H1 – The discount has changed after the Financial Crises
3.5.3 The illiquidity discount and industry
To test whether the possible discount differ between industries I will divide the acquisitions between companies active in the financial industry and companies in non-financial industries. The financial industry actively trades with liquid instruments and investments and could therefore be viewed as more liquid in general than other industries.
H0 – The discount is the same for all industries
H1 – The discount is smaller for the financial industry
3.6 Regression Analysis
The main objective with the study is to determine whether a private company discount exist on the European market. To further investigate the possible discounts with its different features a
12
This paper theorizes that private companies experience a discount in acquisitions due to the lack of available liquidity. However, as previously stated some other possible factors might affect this discount. These three determinants are the firm size, industry and financial crisis. In order to observe if this actually is the case or if the discount is mostly contributable to the lack of liquidity I will
perform a regression analysis together with dummy variables for the different determinants. The regression will be performed so that to see if these three determinants have any significant impact on the discount. If no significant influence on the discount were attributable to these three
determinants, it could be stated that the lack of liquidity is the only reason for the private company discount.
This regression will be constructed in such a way that each of the three determinants will be tested for the effect on the discount. The size determinant will be regressed with the natural logarithm of the Enterprise Value of the private targets to the calculated discount.
Financial crisis effects will be divided into pre- and post with dummies attach to the two sub-samples. I have divided all acquisitions in to six different industry sectors. These sectors are fairly broad
defined and include a wide range of companies. In order to be able to investigate my stated hypothesis that more liquid firms, in this case financial industry firms, trade for a smaller discount I will focus on distinguish that particular industry. I have made the assumption that firms in that industry in general are more liquid in their nature due to their frequent activities in the trading of liquid instruments. Because of the diversity among the companies within each of the other industry sectors I will not make assumptions of the level of liquidity in those. Instead I will group them together and deem them less liquid than the financial industry group. To test this in a regression model I will attach dummies to the two sub-samples of Financial Industry and Non-Financial Industry. The impact on the discount will be calculated as
ILLDIS=α+β1LogEVi+β2D1i+β3D2i+εi (2) Where α is the intercept and ILLDIS is the illiquidity discount for each acquisition. It is given in percentage and have a positive value if there is a discount, and a negative value if the private acquisition hade higher multiples than the public match. LogEV is the natural logarithm of each private companies Enterprise Value and D1 and D2 are dummy variables for industry and financial crisis.
3.7 Statistical Testing
13
14
4. Empirical Findings
This section presents the empirical findings from the research that I have conducted. The first section will describe the whole sample and its characteristics. This will be followed by a summary of the mean and median of the private company discount of the whole sample, followed by a closer look of the differences within the sample with regards to the financial crises, firm size and operating
industry. Lastly the results from the regression analysis will be displayed.
4.1 Matches of years and industries
The sample consists of 127 matches. That is 127 private companies matched with one or two public counterparts. For each match the discounts have been calculated and compiled to a broadening median and mean for each of the three multiples. The sample has been collected from six industries. 48 matches comes from Manufacturing, 23 from Services, 26 from the financial industry, 12 from retail, two from Real Estate and 16 from Other. Of the total sample of 127 private companies, 58 could be matched with two different public companies average multiples. Matches have been collected between the years 2002 and 2014.
Graph 1 and 2 contain an overview of the sample and its origins.
0 10 20 30 40 50
60 Graph 1. Matches between Industries
0 2 4 6 8 10 12 14 16 18 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
15
4.2 The illiquidity discount on the European Market
For the whole sample the private companies trade for a substantial discount. The calculated median of the discount for the EV/EBITDA multiple is 50.41 per cent, for EV/EBIT it is 55.80 per cent and EV/Sales is 57.58 per cent. This result, which holds statistical significance from every aspect with respect to multiples, shows clear support for the existence of a discount associated with private companies on the European market. According to my findings the illiquidity discount seems to be somewhat higher on the European market compared to the US where Koeplin et al. (2002), Block (2007), and Kooli et al. (2003) concluded a discount ranging from -2 per cent to 34 per cent, with a clustering around 20 and 30 per cent. The notion that the illiquidity discount seem to be higher in general on the European market holds when comparing additional empirical evidence using other approaches for measuring the illiquidity discount. Longstaff (1995), Silber (1991) and Emory (2001) concluded discounts on the US market around 20 and 30 per cent. The summary statistics for the whole sample is summarized in table 6.
EV/EBITDA EV/EBIT EV/Sales
Private Targets Median 8.50 12.18 1.19 Mean 17.40 26.13 2.78 Standard Deviation 55.69 65.27 4.66 Skewness 9.84 8.40 3.70 Kurtosis 103.03 81.19 15.99 Public Targets Median 17.74 27.56 2.81 Mean 44.70 85.56 7.99 Standard Deviation 99.43 338.21 16.04 Skewness 5.89 10.26 4.68 Kurtosis 42.89 111.11 26.88 Discounts Median 0.5041** 0.5580** 0.5758** Mean 0.6118*** 0.6946** 0.6254*** N Observations 127 127 127
More than one match 59 59 59
16
4.3 The illiquidity discount and size effects
The first sub question handled possible differences in the discount between smaller and larger firms. I stated a dividing line of 20 MEUR of Enterprise Value for the private firms, companies with an EV smaller than 20 MEUR would be considered to be small and companies with an EV larger than 20 MEUR are in this sample to be viewed as large entities. This gave be a sub-sample of 72 matches belonging to the small size group, and 55 matches attached to the large size group.
Small firms had a median discount of 61.16 per cent in the EV/EBITDA multiple, compared to large firms’ discount of 22.07 per cent. For the EV/EBIT multiple small firms had a median discount of 65.14 per cent meanwhile the corresponding discount for large firms were 23.76 per cent. Smaller firms traded at a median discount of 70.01 per cent when looking at the EV/Sales multiple, compared to the larger firms discount of 21.09 per cent. All results showed statistical significance, both within the sub-samples matches but also between the two sub-samples respective discounts. For all three multiples a clear distinction between the discount between small and large private firms are being displayed, demonstrating the fact that smaller firms’ in general trade for a bigger discount in Europe than their larger equivalents. See table 7 for the multiples and discounts for small and large targets.
4.4 The illiquidity discount and the Financial Crisis
My second sub question concerned the Financial Crises and its possible effects on the illiquidity discount. The crisis unfolded to its fullest extent during 2008 Zandi (2009) so to be able to capture the effects of the crisis as well as possible I divided my sample in two separate categories where 2008 worked as the first year of the “post financial crisis group”. My sample consisted of 62 matches prior to the fully developed consequences of the crisis, 65 matches where completed after the crisis.
Small Private Targets
Small Public Targets
Discounts Large Private
Targets
Large Pubic Targets
Discounts
Median Mean Median Mean Median Mean Median Mean Median Mean Median Mean
EV/EBITDA 6.93 19.57 17.84 52.46 0.6116** (**) 0.6270** 11.69 14.45 15.00 34.54 0.2207** (**) 0.581*** EV/EBIT 10.45 25.42 29.99 114.55 0.6514** (**) 0.7781** 17.15 27.06 22.50 47.61 0.2376** (**) 0.4317** EV/Sales 0.85 1.97 2.82 8.37 0.7001** (**) 0.7642*** 2.17 3.83 2.75 7.49 0.2109** (**) 0.4889** N obs 72 72 72 72 72 72 55 55 55 55 55 55
Table 7. Multiples and discounts of small and large targets. The sample has been divided in to four sub-samples within each multiple category depending on size and ownership with respective means and medians. The discounts have then been calculating by dividing the private means and medians of each size category to its public counterpart and subtract that result from 1. ***Corresponds to a statistical significance level of 0.01. **Corresponds to a significance level of 0.05. *Corresponds to a statistical significance level of 0.1. (**) is interpreted as that the median discount difference between the two sub-samples, small and large firms, are statistically significant at the 0.05 significance level as well.
17
The findings show that for the EV/EBITDA multiple the discount increased from 45.47 per cent before the crisis to 49.36 per cent afterwards. Opposite from the EV/EBITDA multiple the discount
decreased when using the EV/EBIT multiple, from 51.72 per cent pre-crisis down to 32.44 per cent post-crisis. However the effect was reversed yet again for the last multiple, EV/Sales, where the private company discount increased from 54.15 per cent pre-crisis to 62.88 percent after 2008. Unlike the size difference observations the changes are not consistent over all three multiples within this sample and the financial crisis does not seem to impact the discount in any given way. All mean and median discounts within the six different categories showed statistical significance up to an alpha level of 0.01, however the difference between the sub-samples failed to show statistical significance. See table 8 for the differences between targets before and after the crisis.
Private Targets Pre-Crisis
Public Targets Pre-Crisis
Discount Private Targets
Post-Crisis
Public Targets Post -Crisis
Discount
Median Mean Median Mean Median Mean Median Mean Median Mean Median Mean
EV/EBITDA 9.51 15.19 17.43 32.72 0.4547** 0.5358*** 8.31 19.42 16.41 56.12 0.4936** 0.6540**
EV/EBIT 15.47 25.93 32.03 48.80 0.5172** 0.4687*** 17.43 26.33 25.80 120.63 0.3244** 0.7818**
EV/Sales 1.20 2.93 2.62 7.91 0.5415** 0.6295** 1.16 2.63 3.13 8.06 0.6288** 0.6739*
N Obs 62 62 62 62 62 62 65 65 65 65 65 65
4.5 The illiquidity discount and industries
The main argument for the existence of the private company discount is the lack of liquidity associated with firms having that ownership. However there are serious limitations concerning the assessment of each company’s level of liquidity. Even though multiples are provided through the database Zephyr financial statements are not offered and no regulations force private companies to enable access to their income statements, balance sheets and cash flow statements for the public. This hardens the ability to evaluate the level of liquidity for each target company.
One way to perhaps sidestep this issue is to observe companies from different industries and compare the discounts. In order to make any assumptions and conclusions regarding the different industry discounts one would have to establish a sort of industry pecking order based on the
18
different levels of liquidity associated with each industry. Since this is not possible considering the information available I will focus on distinguishing one industry that could clearly be viewed as more liquid compared to the others, namely the financial industry. I will focus on the difference between this industry and all the others. I have made the assumption that companies in the financial industry in general have a higher level of liquidity following their practices of trading with highly liquid instruments and assets on a highly active market. In addition it would not be feasible to separate other industries than the financial since it would not be possible to discern if it was differences in liquidity that actually contributed to the differences in discounts.
However in order to provide an overview of the differences amongst industries the discounts for each sector will be disclosed.
The empirical findings show that financial firms experience a lower discount for both the EV/EBITDA multiple and the EV/EBIT multiple. The private company discount for financial firms associated to the EV/EBITDA multiple is 40.91 percent, the discount for non-financial firms are higher at 50.41 per cent. For EV/EBIT the difference is 5.06 per cent where private financial firms are bought at a discount of 33.83 per cent compared to non-financial firms that are being bought at a discount of 38.89 per cent. This supports Block (2007) findings that firms with more liquid assets trade for a smaller discount compared to other investigated companies.
When observing the EV/Sales multiple there is a slightly smaller difference associated to it compared to the two other multiples, financial companies trades for almost an equally big discount compared to the non-financial firms. The discount is 52.35 per cent for financial firms and 54.12 per cent for non-financial entities. All discounts but one, namely the mean of the EV/Sales multiple in the financial firm sub sample are statistical significant, no statistical significance between the discounts of the different sub-samples are observed.
Private Financial Targets
Public Financial Target
Discount Private
Non-Financial Targets
Public Non-Financial Targets
Discount
Median Mean Median Mean Median Mean Median Mean Median Mean Median Mean
EV/EBITDA 10.34 18.35 17.49 42.03 0.4091** 0.5634* 8.50 17.09 17.14 45.38 0.5041** 0.6233***
EV/EBIT 12.29 21.67 18.57 61.50 0.3383** 0.3166* 12.18 27.28 28.05 91.76 0.3889** 0.5054**
EV/Sales 2.23 6.38 4.68 8.28 0.5235** 0.2296 1.14 1.85 2.49 7.92 0.5412** 0.7663***
N Obs 26 26 26 26 26 26 101 101 101 101 101 101
19
A broader view of all six separate industries shows that they all are relative stable in the EV/EBITDA and EV/EBIT multiples with some outliers in the EV/Sales multiple. The manufacturing industry has a discount of 49.75 per cent for EV/EBITDA, 52.17 per cent for EV/EBIT and 52.35 per cent for EV/Sales. The Services industry clusters around 49 and 56 per cent and Retail 61 and 81 per cent. The Real Estate industry with only two samples has a median discount of 53 per cent for EV/EBITDA, 55 per cent for EV/EBIT and 81 per cent for EV/Sales. The industry marked as “Other” have a median discount spreading from 44 to 75 per cent. Most of the results for the discounts were statistical significant, the only industry showing no significance what so ever were “Real Estate”, this is attributable to the fact that only two acquisitions were derived from that category.
Industry EV/EBITDA EV/EBIT EV/Sales
Median Mean Median Mean Median Mean
Financial 40.91** 56.34* 33.83** 31.66* 52.35** 22.96 Manufacturing 49.75** 76.60*** 52.17** 64.33*** 52.12** 80.53** Services 49.57** 51.21*** 56.75** 61.82*** 53.51** 49.61** Retail 70.17** 79.98** 73.64** 83.61** 87.43** 93.01** Real Estate 53.38 53.38 55.21 55.21 81.41 81.41 Other 44.33** 29.02 54.23** 74.34 75.11** 68.11*
4.6 Regression Analysis
This section will present the results from the regression analysis conducted on the illiquidity discount with firm size, industry and pre- and post-financial crisis as explanatories. The dependent variable was the discount calculated from each matched private and public acquisition. This gave me 381 discounts in total divided in to 127 discounts separately for each of the three multiples. The three independent variables in the regression were the logged Enterprise Value of the private companies and dummies for the qualitative variables financial and non-financial industry and pre- and post- financial crisis.
The regression were conducted for each of the three multiples with the same explanatory variables regressed on the calculated illiquidity discount attributable to the private acquisitions. The results are in general consistent with the above observations of means and medians on the sample.
20
Variables Median Mean Min. Max. Std.Dev N observations
Dependent Variables
EV/EBITDA 0.4376 0.1785 -2.4432 0.9887 0.7919 127
EV/EBIT 0.4889 0.1409 -5.9508 0.9856 1.0585 127
EV/Sales 0.5564 0.0511 -13.3661 0.9966 1.6434 127
Independent Quantitative Variables
LOG(EVP) (In MEUR) 17500 133696 441.4 6601706 649276.4 127
Independent Qualitative Variables
Dummy Industry 127
Dummy Financial Crises 127
Enterprise Value, which is the representation for the size variable, show a statistical significant negative relationship with the discount over all three multiples, implying that firms with a higher enterprise value, and therefore having a larger company size, are acquired for a smaller discount compared to acquired firms with smaller enterprise value and size.
The regression output suggests that if the enterprise value was to increase with one per cent the illiquidity discount would decrease between 0.21 and 0.27 per cent for the three multiples. My sample private Enterprise Value is ranging from four hundred thousand euros to six billion euros with a median enterprise value of seventeen million five hundred thousand euros. Considering these large values and the wide range within the sample a 0.21 percent change in the discount for every one per cent change in Enterprise Value, the effect and impact could be seen as fairly significant. The size effects are therefore not only statistical significant but holds an economical significance to it as well. The only statistical significant coefficient for the industry variable where the EV/Sales multiple, with an estimation value of 1.0681. This positive estimation indicates that firms operating outside of the financial sector are acquired for a higher discount in at least the EV/sales multiple, compared to financial firms. This relationship was supported by the mean and median differences in discounts between the financial and non-financial sub-samples. The estimation in the EV/Sales multiple bear some economic significance, an increasing discount of 106 per cent is substantial considering the range, means and medians of the discounts.
Table 11.Summary statistics of the dependent and independent variables used in the regression. EV/EBITDA, EV/EBIT and EV/Sales are the three dependent variables, with a corresponding discount for each acquisition.LOG(EVP) is the independent quantitative variable, it is the logged Enterprise Value of each private acquisition with values given in hundreds of millions euro as a proxy for size. The logged value is used in order to counter the effect of asymmetry in the sample. UKSIC and YEAR OF
21
No clear relationship over all three multiples could be observed using the medians and means for discounts before and after the financial crisis respectively. Different multiples showed different effects, resulting in contradicting results. This is consistent with the coefficient estimations from the regression where the results for EV/EBITDA and EV/EBIT contradicts EV/Sales, no statistical
significance are reported in any of the three results. See table 12.
Illiquidity Discount (ILLDIS)
Discount EV/EBITDA Discount EV/EBIT Discount EV/Sales
Intercept 2.0489 2.2614 2.0675
Independent Variables
LOG(EVP) -0.2128*** -0.2483*** -0.2775***
UKSIC 0.1998 0.2623 1.0681*
YEAR OF ACQUIRING 0.1260 0.2219 -0.2646
Mean Dependent Variable 0.1785 0.1409 0.0511
S.D Dependent Variable 0.7919 1.0585 1.6434
S.E. of regression 0.7165 0.9822 1.5380
R-squared 0.2009 0.1595 0.1450
Adjusted R-squared 0.1814 0.1390 0.1242
N observations 127 127 127
22 4.6.1 Robustness Check
This model incorporates some robustness testing in its design since the discounts are tested from three different aspects through the three different multiples. I enlarged this check by dividing the whole sample in three time periods, period one consisted of all matches between 2002 and 2005, period two included years between 2006 and 2009 meanwhile period three included 2010 up to and including 2014. The financial crisis variable were removed from the model due to its dependence on years and deemed insignificant effect on the illiquidity discounts.
The size variable shows a negative relationship with the discounts over all three time periods and multiples. When observing the industry dummy variable it is positive, thus indicating the same relationship shown in the basic model as well. Even though there is some statistical insignificance in the model, especially for the Dummy variable in the first two time periods and the size variable for EV/EBIT in the first time period, the resulting output show more or less replicating effects of the basic model and the conclusions seem to be robust over time.
Illiqudity Discount (ILLDIS) 2002-2005 2006-2009 2010-2014 Discount EV/EBITDA Discount EV/EBIT Discount EV/Sales Discount EV/EBITDA Discount EV/EBIT Discount EV/Sales Discount EV/EBITDA Discount EV/EBIT Discount EV/Sales Intercept 1.6497 2.9401 1.3897 1.7586 2.0892 3.7749 2.5745 2.1412 1.0916 Independent Variable LOG(EVP) -0.1499** -0.2964 -0.1013* -0.1952** -0.2242*** -0.4055** -0.2551*** -0.2304*** -0.3204** UK SIC 0.1785 0.3256 0.0744 0.2155 0.0120 0.3244 0.3313 0.6487** 2.4878*
Mean dependent variable 0.3690 0.3880 0.4898 -0.0069 -0.1402 0.0019 0.2787 0.3280 -0.2506
S.D Dependent variable 0.6419 0.9028 0.7091 0.8388 1.2537 1.1406 0.8007 0.7693 2.5336
S.E. of regression 0.6079 0.7734 0.7107 0.8001 1.2374 0.9927 0.6637 0.6243 2.3186
R-squared 0.1590 0.3121 0.0583 0.1226 0.0619 0.2705 0.3490 0.3761 0.2066
Adjusted R-squared 0.1030 0.2662 -0.0045 0.0888 0.0260 0.2424 0.3128 0.3414 0.1625
N observations 33 33 33 55 55 55 39 39 39
23
24
5. Conclusions
This paper investigate whether an illiquidity discount exist on the European market. The summaries of the means and medians of the private company discount of the whole sample showed clear evidence of the existence of such an illiquidity discount among private companies on the European market. In addition the discount is apparently larger than the one found on the US market where discounts clustered around 20 and 40 percent. The results from Koeplin et.al (2000) which is the paper closest resembling mine in terms of methodology consistently showed lower discounts over all multiples and samples.
Further on I theorized that other determinants than just the pure lack of liquidity associated with private companies would have an effect on the discount. Namely firm size, operating industry and the recent financial crisis.
My hypothesized differences in discounts for small and large firms were supported by the empirical findings that showed a higher mean and median discount associated with smaller private companies, the difference in discounts between small and large companies were also statistically significant over all three multiples. This difference was further cemented by the regression analysis that produced negative statistically and economically significant coefficients for all three multiples, demonstrating a negative relationship between Enterprise Value and discounts. Smaller firms have a larger illiquidity discount.
Since the financial industry is to be viewed as more liquid than other industries and branches I developed a hypothesis claiming that financial firms would experience a smaller discount than other entities. I found evidence supporting this when observing the means and medians of the discounts within the sub-samples. The dummy variables in the regression output for all three multiples showed the same relationship, but unfortunately only one coefficient were statistically significant.
The mean and median discounts for the financial crisis sub sample contradicted each other over the multiples, showing no particular pattern in a discount increase or decrease after the crisis. This was supported by the regression that showed different and statistically insignificant relationships for the dummies in different multiples, failing to support my hypothesis that the discount changed in any particular way after the financial crisis.
This study is dependent on the availability of financial information provided in the selected database. This might lead to a skewed result since acquisitions that lack relevant information have to be
removed from the data set. A larger sample, perhaps obtained through the use of several databases and complemented with information direct from the acquired companies would enlarge the data set and improve the quality of the discount estimation and the regression model.
25
6. References
Acharya, V.V., Pedersen, L.H., 2005. Asset pricing with liquidity risk. The Journal of Financial Economics 77, 375-410.
Akbar, S., Rehman, S., & Ormrod, P., 2013. The Impact of Recent Financial Shocks on the Financing and Investment Policies of UK Private Firms. International Review of Financial Analysis 26, 59-70. Alberts, W.W., Varaiya, N.P., 1989. Assesing the profitability of growth by acquisition: A ‘premium recapture ‘ approach. International Journal of Industrial Organization 7, 133-149.
Amihud, Y. & Mendelson, H., 1986. Asset pricing and the Bid-ask Spread. Journal of financial Economics 17, 223-250.
Amihud, Y. & Mendelson, H., 1991. Liquidity, Asset Prices and Financia Policy. Journal of financial Economics 47, 56.
Bajaj, M., Dennis, D.J., Ferris, S.P., & Sarin, A., 2001. Firm Value and Marketability Discounts. Journal of Corporate Law 27, 89-115.
Bekaert, G., Harvey, C., Lundblad, C., Siegel, S., 2013. The European Union, the Euro, and equity market integration. Journal of Financial Economics 109, 583-603.
Black, F., Scholes, M., 1973. The Pricing of Options and Corporate Liabilities. The Journal of Political Economy 81, 637-654.
Block, S., 2007. The Liquidity Discount in Valuing Privately Owned Companies. Journal of Applied Finance 17, 33-40.
Comment, R., 2012. Revising the Illiqudity Discount for Private Companies: A New (and “Skeptical”) Restricted-Stock Study. Journa of Applied Corporate Finance 24, 80-91.
Damodaran, A., 2005. Marketability and Value: Measuring the Illiquidity Discount”, Stern school of Business, July 2005.
Dyl, E. & Jiang, G., 2008. Valuing Illiquid Common Stock. Financial analysts journal 64, 40-47. Emory, J.D., 1997. The Value of Marketability as Illustrated in Initial Public Offerings of Common Stock. Business Valuation Review 16, 123-131.
Feldman, S.F., (2005). Principles of Private Firm Valuation, Hoboken, NJ, John Wiley & Sons. Gelman, Milton., 1972 An Economist-Financial Analyst’s Approach to Valuing Stock in a Closely Held Company. Journal of Taxation June 1972, 353.
Grave, K., Demos, V., B, Y., 2012. The Global Financial Crisis and M&A. International Journal of Business and Management 7, 56-67.
26
Hertzel, M., Smith, R.L., 1993. Market Discounts and Shareholder Gains for Placing Equity Privately. The Journal of Finance 48, 459-485.
Hufbauer, G.C., Poulsen, L., 2011. Foreign Direct Investments in Times of Crisis. Peterson Institute for International Economics Working Paper 11, 2-19.
Hull, J.C., 2011. Options, Futures, And Other Derivatives. Harlow, Pearson Education Limited. International Accounting Standards. Downloaded 2015-03-03
http://www.iasplus.com/en/standards/ifrs/ifrs8s
Ivashina, V., Scharfstein, D., 2008. Bank lending during the financial crisis of 2008. Journal of Financial Economics97, 319-338.
Johnson, B.A., 1999. Quantitative Support for Discounts for Lack of Marketability. Business Valuation Review 16, 152-155.
Koeplin, J., Sarin, A., & Shapiro, A., 2000. The Private Company Discount. Journal of Applied Finance 12, 94-101.
Kooli, M., Kortas, M., & L´Her, J-F., 2003. A New Examination of the Private Company Discount: The Acquisition Approach. Institutional Investor Journals 6, 48-55.
Lie, E., and H. Lie., 2002. Multiple used to Estimate Corporate Value. Financial Analysts Journal 58, 44-54.
Longstaff, F.A., 1995. How Much Can Marketability Affect Security Values? Journal of Finance 50, 1767-1774.
Maher, J.M., 1976. Discounts for Lack of Marketability for Closely Held Business Interests. Taxes September 1976, 562-571.
Mann, H., Whitney, D., 1947 On a Test of Whether one of Two Random Variables is Stochastically Larger than the Other. Annals of Matchematical Statistics, 50-60.
Matsa., D.A. and Miller, A.R. 2012. A Female Style in Corporate Leadership Evidence from Quotas. American Economic Journal: Applied Economics 5, 136-169
Modogliani, F., 1944. Liquidity Preference and the Theory of Interest and Money. Econometrica 12, 45-88.
Moroney, Robert E., 1973. Most Courts Overvalue Closely Held Stocks. Taxes March 1973, 144-155.
Newbold, P., Carlson, L.W., Thorne, B., 2009. Statistics for Business and Economics. 7th edition. New York: Pearson.
Officer, Micah S., 2007. The price of corporate liquidity: Acquisition
discounts for unlisted targets,. Journal of Financial Economics 83, 571-598.
27
Schlingemann, F.P., Stulz, R.M., 2002. Divestitures and the liquidity of the market for corporate assets. Journal of Financial Economics 64, 117-144.
Silber, W.L., 1991. Discount on Restricted Stock: The impact on illiquidity on Stock Prices. Financial Analyst journal 47, 60-64.
Trout, Robert R., 1977. Estimation of the Discount Associated with the Transfer of Restricted Securities. Taxes June 1977, 381-85.
Vayanos, D., 1998. Transactions Costs and Asset Prices: A Dynamic Equilibrium Model. Journal of Financial Economic 11, 1-58.
