Private Company Discount: Evidence of Acquisition Targets

(1)

Private Company Discount: Evidence of

Acquisition Targets

25-06-2020 Eldert Elzinga Student number: 2563428 e.elzinga.1@student.rug.nl University of Groningen Program: MSc Finance

Faculty of Economics and Business Economics Supervisor: dr. J.J. Bosma

Second Assessor: dr. W. Bessler

Abstract:

This paper reports discounts for private companies, between 13% and 30%, and subsidiaries, between 10% and 16%, compared to publicly traded companies based on acquisition multiples. The main contribution to existing literature is the use of alternative matching procedures to determine Private Company Discount. Also, where previous research has focused on the US, this paper uses a dataset of transactions in West-Europe between 2000 and 2020. Especially for private firms, acquisition discounts are lower when the company is larger. Furthermore, results indicate that the reason for the discounts can be explained, by the liquidity constraints private companies and subsidiaries face.

(2)

2

1 Introduction

Fundamental differences between private and public companies makes valuation of private companies difficult. The equity of a public company is traded in the market and the company is therefore also valued by the market. Private companies are not listed on an exchange and equity is not easily tradable. Fuller et al., (2002) find that for unlisted firms there is lower competition in bidding compared to listed firms, what could be a reason for a discount on the value of the unlisted firm. Also, practitioners such as Pratt and Grabowski (2014) argue that it is unreasonable to value private companies with the same multiple as public companies. Zanni (2015) states, that privately held companies often value their non-marketable activities, which should not have value at all and discounts have to be applied to correct for these non-marketable activities. This paper uses the transaction multiple approach to determine whether private companies trade at a discount compared to their public peers. This method was introduced by Koeplin et al. (2000). Private Company Discount is determined by matching a private company with a similar public company and comparing the acquisition multiples. The matching procedure used by Koeplin et al. (2000) has not changed over time. All acquisition studies use date, industry and size for matching private and public companies. This paper reviews whether other matching criteria impact the results of Private Company Discount calculation. In previous research the deal value is used as size measure. However, theory (Damodaran, 2016; Koller et al., 2015) suggest that deal value can be influenced by many factors and recommend measuring company size using revenues or assets of the company. For the univariate and multivariate analysis of this paper, revenues are used instead of deal value. Other matching criteria used in addition to date and industry are profitability and growth, which are stated as drivers of value by Koller et al. (2015).

(4)

4

the subject finds discounts for unlisted companies. Not only the question whether a Private Company Discount exist is important, but also what the exact height of the discount should be and which factors drive the height of the discount. The question whether private companies trade at a discount and which factors influence the Private Company Discount (PCD) is answered using univariate and multivariate analysis and supported by literature about the subject.

In contrast to previous research, which is concentrated around the United States, this paper uses a dataset of mergers and acquisitions of West-European target companies. Both Western Europe and the United States are regions in which economic prosperity is high. Large differences in results are not expected, since the economic state of both regions is comparable and the focus of this paper will not be whether differences exist. Still, the results of previous papers are compared to the results of this paper.

The biggest limitation of acquisition studies has been the limited availability of data. The dataset of Koeplin et al. (2000) consisted of 192 firms matched with one public peer company. Although the limited availability of data continues to be a problem, over time the availability of data has increased and the results can be stated with more power. The univariate analysis of this paper includes a range of 453 up to 1051 (dependent on the multiple) unlisted target firms for the sales matching procedure. Also, PCD is calculated using a portfolio of public peer companies instead of just one public peer.

This paper contributes to existing literature, firstly by evaluating alternative matching criteria in the process of determining Private Company Discount. The empirical research gives insights how results change when the methodology is altered. Also, the methodologies of previous research (Jaffe et al., 2019; Officer, 2007) are used on the dataset and the results are discussed. Secondly, by using a dataset of European target companies the existence of PCD is tested in a new environment. Thirdly, the impact of several variables on PCD is tested using the new sales matching procedure.

(5)

5

results of the empirical research and in the sixth section the conclusion and limitations of the paper are discussed.

2 Literature review

2.1 Theoretical Framework

In the theoretical framework the most important valuation models are discussed. Firstly, the Discounted Cash Flow Model (DCF), which is seen as the foundation for valuation (Damodaran, 2016). Secondly, the multiple approach of valuation. Thirdly, the main factors which affect private companies differently than public companies are identified and the hypothesis is made whether the private company should exist. In short, the private companies should trade at a discount compared to public companies. The main factors driving the discount are that private companies experience liquidity constraints, principal-agent problems, higher risk of default and their investors are often not well diversified.

2.1.1 DCF Valuation

The discounted cash flow model estimates the value of an asset by discounting its future expected cash flows (Damodaran, 2016; Koller et al., 2015). The general formula is as follows:

𝑉𝑎𝑙𝑢𝑒 = ∑ 𝐶𝐹𝑡 (1 + 𝑟)𝑡 𝑡=𝑛

𝑡=1 (1)

• 𝑡 is the period

• 𝑛 is the life of an asset

• 𝐶𝐹𝑡 is the Cash Flow in period t

• (1 + 𝑟)𝑡_{is the discount rate with dependent on time,}

(6)

6

company 2 over the years, with revenue growth and profits being equal, Company 1 will have higher cash flows, therefore a higher value. Koller et al. (2015) adjusts the general formula to the key value driver formula for determining company value:

𝑉𝑎𝑙𝑢𝑒 = 𝑁𝑂𝑃𝐿𝐴𝑇𝑡=1(1 − 𝑔 𝑅𝑂𝐼𝐶)

𝑊𝐴𝐶𝐶 − 𝑔 (2)

• 𝑁𝑂𝑃𝐿𝐴𝑇 is net operating profit less adjusted taxes. NOPLAT consists of profit excluding interest cost and only the operating part of the taxes are subtracted.

• 𝑔 is the growth rate of 𝑁𝑂𝑃𝐿𝐴𝑇 and 𝑅𝑂𝐼𝐶. • 𝑅𝑂𝐼𝐶 is the return on invested capital 𝑁𝑜𝑝𝑙𝑎𝑡

𝐼𝑛𝑣𝑒𝑠𝑡𝑒𝑑 𝐶𝑎𝑝𝑖𝑡𝑎𝑙. Invested capital contains the total amount the company has invested in the core operations of the firm, such as Property plant and equipment and net working capital.

• 𝑊𝐴𝐶𝐶 is the weighted average cost of capital (equity and debt), the discount rate. The key value driver formula (2) is an annuity, which assumes a constant ROIC, WACC and growth rate (g) and infinite life of a company. According to Koller et al. (2015) the advantage of using (2) is that it captures the core of valuation by including the key factors which drive the value of a company.

The general formula as well as the key value driver formula are based on expectations, expected future cash flows and expected growth rates. The DCF method therefore can be used when cash flows are positive or will become positive in the near future and expectations can be made with confidence. However, when this is not the case, it would be tricky to use the DCF method to determine company value (Damodaran, 2016). Also, certain characteristics and states of companies make DCF valuation more difficult. Examples are firms which are in financial distress, cyclically revenue and products with patents. Furthermore, when firms are involved in an acquisition or are unlisted, DCF valuation will be more difficult (Damodaran, 2016). For unlisted firms determining the risk factor is more difficult because analysis on historical prices cannot be performed, since the companies are not traded publicly. Therefore, comparison has to be made with a peer group of publicly traded companies (Damodaran, 2016).

(7)

7

2.1.2 Multiple Valuation

Discounted cash flow analysis is the most accurate way of valuating companies, assets and projects (Koller et al., 2015). While the focus of valuation mostly lies on DCF valuation, relative valuation is often used in praxis for valuing assets (Damodaran, 2016). The theory of multiple valuation is based on the assumption that similar assets trade at comparable prices. When buying a house, stocks, or an entire company, these assets are often compared to other assets on the market for accessing its values. For valuing companies, multiple analysis is also often used. When valuing a company using a DCF valuation a multiple analysis can be used as a benchmark (Koller et al., 2015). Examples of commonly used multiples are, Enterprise Value (EV) / Revenue, EV/EBITDA, EV/EBIT, Price/Earnings.

Koller et al. (2015) mentions four rules for valuing using multiples. Firstly, use the right multiples. Investors commonly refer to the Price/Earnings ratio to determine whether a company is under or overvalued. This multiple can be misleading, because it does not account for capital structure, and can include non-cash write offs. When company 1 has relatively high interest costs and non-cash write offs compared to company 2, the P/E is higher although company value can be equal. EV/EBITDA and EV/EBIT are other earnings-based multiples used for comparing company values. Both multiples also have their weaknesses. EBITDA does not include depreciation and even though depreciation is a non-cash write off, it often signals capital expenditures that will be made in the future. The drawback of using the EBIT multiple is that amortization is included in the calculation, which is a non-cash write off and can distort valuation. Another multiple commonly used in the EV/Revenues multiple. Using revenue multiples requires the assumption of comparable operating margins for peer companies. Secondly, Koller et al. (2015) states multiples have to be calculated in a concise manner. Thirdly, that peer group selection is particularly important. Peer group selection should not only be based on industry, but also on similar key value drivers, such as return on invested capital, long term growth, and profitability. Lastly, it is recommended to avoid using average or median multiples of an entire industry for analysis, because it understates effects, such as long-term growth and return on invested capital. If companies from the same region are used and they are operating in the same industry, these companies should have similar Weighted Average Cost of Capital (WACC) and tax-rates (Koller et al., 2015).

(8)

8

of equity of these companies, known as market capitalization, is determined by the market. The multiple can be derived using this information. Transaction Multiples are obtained from data of acquisitions of companies. Financial information about the deal can be used to calculate the multiples.

2.1.3 Private Company Valuation

Valuation of privately held firms differs from valuation of publicly held firm, because of several factors (Damodaran, 2016). Firstly, it is more difficult to acquire the financial data required for valuation of privately held firms. Publicly traded firms have to disclose financial information and have to comply with certain accounting standards. Private firms do not have to oblige to these rules and can have different accounting practices than other companies. Because the equity of private firms is not available in the market, liquidating the equity of a privately held firm can be more difficult and expensive (Damodaran, 2016). Furthermore, in private companies the management of the firm often holds relatively more equity of the firm than for publicly traded companies. This could result in business expenses be translated to the benefit of the manager, which could have a negative impact of the overall value of the firm. These differences have impacts on cash flow, expected growth rates and discount rates.

Cost of capital

(9)

9 Cash flows

Characteristics of private companies have an effect on cash flows. The method in which the manager is rewarded for the work performed, is indifferent to the manager (Damodaran, 2016). Dividends that are paid out to the manager are not part of the operating income used in valuation, however, because the payments are made because of the managers work, these should be added to the personnel costs. Owners/managers also tend to mix personal and business expenses, which has a negative impact on the value of the firm (Damodaran, 2016). Furthermore, Damodaran (2016) states that managers/owners of unlisted firms tend to stay on for a long time and for instance for a family business is a relative often the successor, which could affect the persistence of growth negatively. For a public firm it is normal that the CEO changes often and should always be acting in the best interest of the firm, whereas for private firms the manager can act more easily in its own best interests.

Liquidity

Liquidity can be a strong motivator for investors. Liquidity can be defined, as the amount of liquid assets in a company or market. When investors take equity positions in companies, they prefer to have the option to liquidate this position when needed (Damodaran, 2016). When holding equity of a publicly traded firm it is easy and cheap to liquidate that position, however for private firms this does not hold. Liquidating equity in a private firm can be difficult for investors and be more costly, which is also a reason that a cost of equity should be higher for private firms (Damodaran, 2016).

For most private companies obtaining and keeping liquidity in the company is difficult and has come with risk. For public companies, it is easier because of the shares being publicly traded in the market (Officer, 2007). Liquidity problems for a company can result in an unwanted sale and decrease the price. As a consequence, a liquidity discount should be applied for private firms relatively to public firms (Damodaran, 2016; Officer, 2007). When a large part of the assets of a private firm consist of cash and marketable securities, the company is relatively liquid and the discount should be lower in comparison to other private companies (Damodaran, (2016).

(10)

10

illiquidity discounts. Large companies will have a lower illiquidity discount, since the illiquidity discount can be seen as a percent of the total value of the firm.

Also, when liquidity in the market is low it can be more difficult for companies to obtain liquidity and decreases the price (Officer, 2007). In financial crisis illiquidity becomes a large problem, in which prices drop severely (Koller et al., 2015). This illiquidity problem is even higher for private firms and they trade at a higher discount than when liquidity in the market is higher. Also, when the possibility of going public and performing an IPO (Initial Public Offering) is high, the liquidity discount should be lower, since it is relatively easy to make the firm a more liquid firm. Theory suggest that certain characteristics differ between public and private firms and value between these firms should also differ, ceteris paribus. These factors have a negative influence on the value of companies, with illiquidity as most prominent factor. The negative effects on value are displayed in table 1.

2.2 Empirical Studies

In this section the results of several empirical studies researching the Private Company Discount and their results are presented. Table 3 displays an overview of the discounts found over the years for different methods. Only Jaffe et al. (2019) finds a premium for private companies. Other empirical studies find discounts ranging from 17% to 73% for private companies. The Theory and results of the restricted stock studies, IPO studies and option studies are discussed in Appendix A.

Table 1. Factors influencing private company valuation

This table summarized the effect of factors affecting private company valuation.

Factor Effect on Expected Sign

(11)

11

2.2.1 Acquisition / Multiples studies

Acquisition have been for the last 20 years a more popular method of research in explaining discounts between private and public transactions. Koeplin et al. (2000) state in their paper “The Private Company Discount” that with acquisition studies more than a marketability problem influences the difference between the private and the public transaction. The name is changed to Private Company Discount, which is also used in subsequent studies.

Valuation of private companies is considerably different in several aspects than valuation of public companies and the liquidity of stock is an key factor (Koeplin et al., 2000). This phenomenon is researched in Koeplin et al. (2000) by using the multiple approach of valuing businesses. Private company data was scarcer before the year 2000 than it is nowadays. For their research Koeplin et al. (2000) collected data from 84 domestic and 108 foreign private company deals. For every private transaction, a public peer was obtained. This company was

Table 3.

Table 3. summarizes the results of the empirical studies concerning Private Company Discount. The

overview includes the results of four methods and 10 papers. The percentages indicate the percentage discount found for a private company or subsidiary in comparison to public peer companies.

EV / Sales

EV / EBITDA EV / EBIT EV / Net

Income Average Acquisition Discount Restricted stock studies Maher (1976) 35% Wruck (1989) 18% Silber (1991) 30%

IPO studies Emory (2002) 47%

(12)

12

selected based on time-period, size, and industry classification. Only deals were used, where a controlling interest in a company was acquired. The Private Company Discount was calculated using the following formula:

Private Company Discount = 1 – (Private Company Multiple / Public Company Multiple)

Koeplin et al. (2000) reports Private Company Discounts of 18% for the EBITDA multiple and 31% for the EBIT multiple (table 3). Other methods of determining PCD are criticized, but also limitations of using multiple analysis are mentioned. The biggest limitations are the small sample size and that per private company only one peer company is selected.

Kooli et al. (2003) followed up on the research of Koeplin et al. (2000) and also used acquisition/multiple estimations to calculate the PCD or Discount Lack of Marketability (DLOM). Kooli et al. (2003) acknowledge the benefits and limitations of the study of Koeplin et al. (2000) and adjust accordingly. The sample size is enlarged and not one peer company is picked for every private company transaction, but a peer group of companies is selected. This reduces the problem that the characteristics of a single firm play a huge role. The methodology used is based on the work of Brav et al. (2000), who introduced a reference portfolio of companies. The total dataset included 331 private company transactions between 1995 and 2002. The results indicate median discounts of 17% for the EV/Sales multiple, 34% for the EV/NetIncome multiple (table 3). Despite using other multiples results are similar to Koeplin et al. (2000).

(13)

13

impact of information asymmetry is difficult and it is possible that it plays a larger role than observed (Officer, 2007). The results of the paper include average acquisition discounts, which is the average of four multiples. All the multiples are significantly different from zero and for private firms, discounts are present for all multiples excepts book value of equity. For subsidiaries, when averaging all multiples, the results indicate a median discount for around 30%, where discounts for private firms indicate a median discount of around 20% (table 3). Paglia and Harjoto, (2010) criticizes previous acquisition studies because of its small sample sizes and broad matching criteria. The total sample size of Koeplin et al. (2000) consisted of under 100 companies. Paglia and Harjoto (2010) attempts to tackle these problems by only selecting one public peer company, but with stricter selection criteria. In contrast to previous research Paglia and Harjoto (2010) uses 6-digit NAICS codes, which are more specific than the 2-digit SIC codes. Also, for selection a public peer company, comparable revenues, profitability of the target firm, deal characteristics, for example buyer type, transaction type and organization type are included. Paglia and Harjoto (2010) find when calculating the discounts significantly larger discounts than previous research has found. Median PCD of 73% for the sales multiple and 66% for the EBITDA are suggested. The study does not use transaction multiples for comparison, but trading multiples of public companies which stay public for at least 3 years. This methodology is motivated by the assumption that market values of companies are close to control values such as transaction values (Nath, 1990).

(14)

14

suggest using the average acquisition multiple premiums of 25% for private companies and discounts of 15% for subsidiary targets (table 3). Trimming the sample does not alter results. Conclusion of the literature

Theory suggests that many factors which play a role in the valuation of private companies negatively impact their values. Based on the negative impact of the characteristics and the results of empirical research on the subject, the expectation is that private companies and subsidiaries trade at a discount compared to similar public companies.

H1: Private companies and subsidiaries trade at a discount compared to public firms.

2.3 Explanatory variables

Theory and previous research also suggest factors influencing the PCD. The factors affecting PCD are divided into 2 categories: Target characteristics and Acquirer characteristics.

2.3.1 Target characteristics

Theory and empirical research identify certain factors affecting the height of the discount. Kooli et al. (2003) extend the research on PCD by testing the impact of target characteristics on the PCD. The results indicated that for larger firms and firms with higher growth, PCD was lower than for small and low growth firms. Also, is reported that when liquidity is high in the market as well as for the company itself, PCD is lower, because of a relatively low chance of financial distress. PCD was also higher for certain industries. Construction, manufacturing, wholesale, and retail trade private companies had a higher PCD than firms from other industries. Paglia and Harjoto (2010) find lower PCD for healthcare and transportation for the Sales multiple, lower PCD for transportation and construction for the EBITDA multiple and high discounts for information technology for both multiples. Furthermore, Paglia and Harjoto (2010) research the impact of different variables on the PCD. Results indicate that PCD is lower for larger firms and companies with higher profits.

(15)

15

for private companies. Cash payments should result in higher discounts because it provides immediate liquidity for the owners. Owners are expected to accept lower multiples when the acquisition is paid in cash (Officer, 2007).

H2: PCD will differ across industries. H3: Larger companies have lower PCD.

H4: Companies with higher profitability have lower PCD H5: Companies with higher growth rates have lower PCD

H6: Companies with a relatively high amounts of liquid assets will have lower PCD H7: PCD will be higher for all cash transactions

2.3.2 Acquirer characteristics

Not only the characteristics of the target firm play a role in the, but characteristic of the acquiring firm could also play a role in the deal value and the Private Company Discount. Jensen (1986) states that large public firms tend to overpay at takeovers because of agency problems. The so-called empire building problem arises when managers act in their best interest by engaging in acquisitions that are not in the best interest of the company. PCD is therefore expected to be lower when acquired by relatively large companies and public companies. Information asymmetry is mentioned to impact the value of private firms negatively (Damodaran, 2016; Officer, 2007). Officer (2007) states that information asymmetry is larger for acquiring firms, when the acquiring firm does not operate in the same industry as the target firm. Officer (2007) reports that PCD is twice as high when the acquiring firm does not operate in the same industry as the target firm. When acquiring firms acquires a target in the same industry PCD is expected to be lower.

H8: When the acquiring company is relatively large PCD is lower H9: When the acquiring company is publicly listed PCD is lower

(16)

16

3 Data and Methodology

This section describes the steps that have been taken to collect the data and the research methods used to test the hypotheses.

3.1 Data

To empirically test the hypotheses, data is collected from the Thomson Reuters Eikon database. This database covers deals data with specific information about the deals, targets, and acquirers. The first selection is made and described below.

• Deals with announcement dates between 01-01-2000 and 01-01-2020 are included, which accumulate to a timespan of 20 years.

• Only completed deals are included in the sample for private companies, subsidiaries and public companies.

• Only deals with deal type “mergers” and “acquisitions” are included. • Targets are companies from West-European countries.

• Only deals with a total value above $10 million are included.

• Deals are included in which the acquiring party has acquired a controlling interest, thus over 50% of the shares of the target firm.

The minimum deal value is set at $10 million, which is based on the assumption that above this threshold companies are publicly traded (Jaffe et al., 2019). Below the threshold of $10 million it is difficult to find peer companies for the private companies and subsidiaries. In accordance with previous research only deals are included in which a controlling interest is acquired. The selection of sample can be motivated, but the dataset is susceptible to selection bias. Earlier in the paper it is mentioned that private companies do not have to oblige to the same accounting and reporting rules as public companies, which results in a significant amount of missing data. One could argue that private companies and unlisted subsidiaries only report their detailed financials when the performance is well. Because detailed information is needed for the analysis, this could lead to selection bias.

3.2 Private Company Discount calculation

(17)

17

median of a portfolio of public peer companies to calculate PCD, similar to Officer (2007). The median is used to deal with outliers. The formula is as follows:

Private Company Disount = 1 − ( Private Company Multiple

Median Public Company Multiple) (3)

Multiples

Four multiples are used to determine PCD. • Deal Value (EV) / Sales

• Deal Value (EV) / EBITDA • Deal Value (EV) / EBIT

• Deal Value (EV) / Net Incomes

The nominator of all the multiples used is the deal value. The deal value is the total value, in US dollar, paid by the acquirer, excluding fees and expenses. The total value includes payments to common stock, common stock equivalents, preferred stock, debt, options, and assets, which are payed in the last half year before the announcement date. Liabilities are assumed and included when disclosed. Preferred stock is only included if the acquisition accumulates to 100% of the target company. Because debt and assets are included the deal value is comparable to enterprise value of the company.

The denominators used are one revenue based multiple and three earnings bases multiples. The denominators are from the financial statements of the companies the year prior to the acquisition. The revenue-based multiple is the EV/Sales multiple. The advantage of this multiple is that in contrast to the earnings basis multiples, it is independent from accounting procedures (Damodaran, 2012). This is important because most countries in the European Union have their own accounting policies. The disadvantage of the sales multiple, is that it does not control for differences in profitability, which is an important value driver for company value (Koller et al., 2015).

(18)

18

3.3 Standard Peer Group Selection

The standard peer group selection is based on the selection procedures of previous research (Jaffe et al., 2019; Koeplin et al., 2000; Kooli et al., 2003; Officer, 2007). The private companies and subsidiaries are matched using three selection criteria. The companies are matched on date, industry and size.

The first criterion is that the time between the announcement date of acquisition of the private company or subsidiary and the public peer company cannot exceed three years. Secondly, the TRBC economic sector classification is used for assigning the companies in 10 industry classes. The Thomson Reuters Business classification (TRBC) sector classification is globally relevant, market focused and revised every four years. Thirdly, only public deals with a deal value within 30% of the deal value of the company are selected. This is similar to Officer (2007) and Jaffe et al (2019).

3.4 Alternative Peer Group Selection

Peer group selection is key when analyzing multiples (Koller et al., 2015). Previous research uses deal value as a variable to match companies with similar size. However, deal value is not the best variable to measure the size of a company. Theory suggest that size should be measured by using revenues or the assets of the company (Damodaran, 2016). Key fundamentals when comparing companies include growth, profitability, return on operating capital and long term growth (Koller et al., 2015). Using these insights, three new methods are created for the selection the peer group of companies. All methods still include the industry and date criteria for matching public companies with private companies or subsidiaries.

Sales matching

(19)

19 Profitability Matching

Koller et al. (2015) gives in his key value driver formula return on operating capital as measure and also mentions that profitability is a key fundamental in company value. Because return on invested capital is not available as variable, profitability is used for the matching procedure. The profitability of each firm is calculated using the following formula:

𝑃𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦 = 𝑁𝑒𝑡 𝑃𝑟𝑜𝑓𝑖𝑡 𝐿𝑇𝑀

𝑁𝑒𝑡 𝑆𝑎𝑙𝑒𝑠 𝐿𝑇𝑀 (4)

LTM stands for the last twelve months before the announcement of the merger or acquisition. For the matching procedure the profitability of the matched firm has to be in a range between 50% more and 50% less than the private company or subsidiary.

Growth matching

The third alternative method used for matching companies is based on growth rates. Long term growth rates are usually similar for companies in the same industry. However, in the short-term some companies may have a competitive advantage over other companies in the same industry. Forecasted growth rates would tell more about company value, since the value is determined by the cash flows the company generates in the future. However, because these growth rates are not available, historical growth rates are used and the assumption is made that historical growth rates will be sustained in the near future. For the matching procedure, the matched company can have a growth rate of EBIT of 50% higher or lower than the private company or subsidiary.

3.5 Univariate analysis

When for each company the individual Private Company Discount (PCD) is calculated, this can be used for analysis. For each multiple the median PCD is calculated. When PCD > 0, the private company or subsidiary trades at a discount. When PCD = 0, there is no discount or premium and when PCD < 0, the private company or subsidiary trades at a premium. To test the statistical relationship of the Private Company Discount a t-test is used. Koeplin et al. (2000) used this in their paper and define it as the traditional method. The t-test determines if the median of the PCD is statistical different from zero for private companies and subsidiaries.

(20)

20

a t-test is used to test whether the coefficient is statistically different from zero. This method is used for the alternative peer group selection and the standard peer group selection.

Jaffe et al (2019) calculates average acquisition discount by dividing the mean of all the average acquisition multiples across all the private companies, by the mean of average acquisition multiple across the public control groups. In this paper the same technique is used, but than for median discount instead of average discounts. Officer (2007), as mentioned before, discards private company premiums over 100% and therefore uses one-sided truncation. Both these methods are mimicked using the new database and the standard peer group selection and the results are compared.

3.6 Multivariate analysis

The multivariate analysis should indicate which factors influence the height of the Private Company Discounts and to what extent. For the multivariate analysis, the sales matching procedure has been selected. Matching companies on sales controls better for size then on deal value but limits the amount of observations lost compared to profitability or growth matching. For each multiple a quantile regression is run including the explanatory and control variables, which are based on theory and previous empirical research. The motivation for using a quantile regression and not an Ordinary Least Squares (OLS) regression is based on characteristics of the dataset. As can be seen in the descriptive statistics in section 4, the dataset contains outliers which increases the variance of the dataset and the data is skewed. Where OLS assumes parametric distribution and constant variance, a quantile regression does not. A quantile regression is the better option, since it does not assume parametric distribution, constant variance and is robust to outliers.

The effect of the target characteristics and the acquirer characteristics (section 2.3.1, 2.3.2) are determined separately because the size of the target company, in terms of assets, is highly correlated with the size of the acquiring company. The variables are described in Appendix B. The equations that will be estimated reads:

𝑀𝑒𝑑𝑖𝑎𝑛𝑃𝐶𝐷 = 𝛼₀+ 𝛽₁𝐿𝑁𝐴𝑠𝑠𝑒𝑡𝑠 + 𝛽₂ 𝑃𝑟𝑜𝑓𝑖𝑡𝑎𝑏𝑖𝑙𝑖𝑡𝑦 + 𝛽₃𝐺𝑟𝑜𝑤𝑡ℎ + 𝛽4𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝛽5 𝐶𝑎𝑠ℎ𝑡𝑟𝑎𝑛𝑠𝑎𝑐𝑡𝑖𝑜𝑛

(21)

21

𝑀𝑒𝑑𝑖𝑎𝑛𝑃𝐶𝐷 = 𝛼₀+ 𝛽₁𝑃𝑢𝑏𝑙𝑖𝑐 𝑠𝑡𝑎𝑡𝑢𝑠 + 𝛽₂𝑆𝑎𝑚𝑒 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦 + 𝛽₃𝐿𝑁𝐴𝑠𝑠𝑒𝑡𝑠𝐴𝑐

(6) 𝑀𝑒𝑑𝑖𝑎𝑛𝑃𝐶𝐷 is the median of the individual Private Company Discounts. 𝛼₀ is the constant and 𝛽𝑖 are the coefficients of the variables. The variables are described in Appendix B. the quantile regressions do not assume parametric distribution and constant variance. Tests for heteroskedasticity cannot be performed and the assumption is made, that heteroskedasticity is not problematic for the analysis. Correlograms are reviewed to assess whether multicollinearity is a problem, (an example is displayed in appendix C table 1. The conclusion is made, that the independent variables are not correlated to the extent that it affects the results problematically. Robustness

To test whether the findings of the quantile regressions are robust to changes in the model, OLS regressions are run and their results compared to the standard model. Since the problem of the outliers and the skewness in the distribution of the dependent variable is still there, the dependent variable is winsorized. For both the private companies and subsidiaries the top and bottom 10% of the PCD is winsorized and modeled, as well as for the top and bottom 25%. The motivation behind winsorizing large parts of the sample, is that the premia have a larger impact than the discounts. As can be seen in the example in appendix C table 2, for the EBITDA multiple of private companies, the PCD at the 10th percentile is equal to -2.15, whereas the PCD at the 90th percentile is equal to 0.80. When the bottom and top 10% of the PCD is winsorized, the results are still biased towards a premium. The choice of winsorizing over trimming is made, since otherwise a large part of the sample has to be deleted. White tests are used to test for heteroskedasticity and because the null hypotheses of homoskedasticity were rejected, robust standard errors are used to ensure no inference problems.

(22)

22

4 Data and descriptive statistics

The first dataset contains 7590 transactions based on the specifications described in the methodology. This includes completed mergers and acquisitions, in which a controlling interest is acquired between 1-1-2000 and 1-1-2020 in western-Europe with a total value above 10 million as described in section 3.1. Since the discount only for private companies and subsidiaries is researched in this paper, joint ventures and governmental companies are excluded. This leaves 7491 transactions. As can be seen from table 4, only for public targets are for most companies the multiples available.

The descriptive statistics of the multiples used are displayed in table 5 on the next page. What stands out is the extremely high volatility in the results. These outliers have a strong effect on the mean values, which results in large differences between mean and medium values. Private companies have the highest median multiple for the Sales and EBITDA multiple, whereas the EBIT multiple is higher for public companies. The median NetIncome multiple is the highest for subsidiaries, however the other three multiples are the lowest. For testing the statistical differences, a non-parametric test is used, because the multiples do not follow normal distribution. The distribution of the multiples is skewed to the left side. The Mann Whitney U test, also called the Wilcoxon z test, is used to test equality of the means of two groups and tests if these groups are similar in terms of shape.

Table 4. Dataset specifications

This table contains data about the amount of transactions available in the raw dataset for every type of company and for every multiple.

Private Targets Subsidiary Targets

Public Targets Total

(23)

23

For privately held companies only for the Sales Multiple the mean and median are significantly different from the mean and median of the Public companies and for the EBITDA multiple the mean is statistically different from the mean of the public companies. This results in the expectation that for the sales multiple and the EBITDA multiple private companies trade on average at a premium compared to public companies. For subsidiaries only for the EBIT multiple the mean is statistically different from the mean of public companies. Therefore, for the EBIT multiple on average a premium is expected for subsidiaries.

Standard deviations are higher for private companies and subsidiaries compared to public companies. This indicates that outliers have a higher impact on private companies and subsidiaries than on public company multiples. When not controlling for this impact, this could bias the analysis towards a premium for subsidiaries and private companies.

Table 5. descriptive statistics multiples

This table displays the mean median and standard deviation for private companies, public companies and subsidiaries for every multiple. Furthermore, is tested whether the mean of the private companies and subsidiaries are significantly different from the mean of the public companies. Significance is tested using the Mann-Whitney U test and the Median test.

Deal value / Sales Deal value / EBITDA Deal Value / EBIT Deal Value / Net income Private Companies Mean 61.79 68.69 255.78 101.48 Median 1.78 11.14 15.30 22.73 Std. Dev. 553.20 904.25 2553.53 417.85 Subsidiaries Mean 487.03 35.31 730.64 621.63 Median 1.50 10.139 14.52 23.66 Std. Dev. 13112.65 133.44 10921.38 10595.17 Public Companies Mean 24.47 27.75 43.33 49.50 Median 1.575 10.66 16.69 23.11 Std. Dev. 599.27 157.62 171.94 144.44 Private companies Mann-Whitney U (Wilcoxon z) 3.485*** 1.898* -1.612 -0.104

(24)

24

5 Results

In this section the results of the empirical analysis are discussed. The results of the univariate display discounts for private companies and subsidiaries in comparison with public peer companies. Three methods of calculating Private Company Discount and the results are explained and compared. Also, the outcomes of using different matching procedures are reviewed.

5.1 Univariate analysis

The first row of table 6 displays the results of the standard sample. The private company discounts are calculated by dividing private company multiple with the median of the public company multiples (3). A quantile regression is performed to calculate the median PCD and test its significance. Jaffe et al. (2019), states that when calculating medians of medians this understates the overall result. Therefore, the PCD in the second row is calculated by dividing the private company multiple by every public peer and then running the quantile regression on all PCDs. The results of the third row display the result of the quantile regression after discarding the premia above 100%, which is similar to Officer (2007).

Table 6. Standard matching

This table contains the results of the univariate analysis for three different methods of calculating PCD. PCD is calculated for private companies and subsidiaries and separately for the four multiples. The standard method is calculating the mean of the PCDs of all companies. The second method using the methodology of Jaffe (2019) and the third using the methodology of Officer (2007). The first number indicates the median discount or premium, for instance -0.78 indicates a premium of 78%. The second number indicates the t-value and the third number the amount of observations. ***, **, * represent significance at the 1%, 5%, and 10% levels, respectively.

PCD Sales PCD EBITDA PCD EBIT PCD Net Income

Private Subsidiary Private Subsidiary Private Subsidiary Private Subsidiary

(25)

25

The results (table 6.) of the first two methods are as expected quite similar. For both methods, when looking at the PCD for the sales multiple, it displays a high premium for private companies of 78% and a lower but still significant premium for subsidiaries. The high premium can be explained on the fact that on average the private companies and subsidiaries are much smaller than their peers in terms of revenues. For the other multiples the results of the first two methods are also quite similar. The statistical significance is stronger for the method used by Jaffe et al. (2019). This could be explained, by the fact that the second method includes more observations, which increases the significance. However, using this method, the multiples of the private companies and subsidiaries are used more than once and not used equally often which results in bias of the results. The results of the standard method display a relatively large and significant premium for private companies for the EBITDA multiple. For subsidiaries also premiums are suggested when looking at the results.

(26)

26 Alternative Method

The results using the alternative matching criteria are displayed in table 7. The matching criteria are explained in section 3.4. The standard method of calculating PCD is used.

As can be seen in table 7, the results of the sales matching procedure, gives significant results for every multiple except the net income multiple for private companies and subsidiaries. As expected, the results differ from the previous method. The impact of controlling for company size using revenues is especially large for the sales multiple. The median PCD changes from minus 80% to plus 30%. For the EBITDA and EBIT multiple significant discounts are displayed in contrast to previous results. Part of the observations have been lost, because of lack of availability of the net sales of certain companies and because of the new matching criteria. Based on theory using sales as a matching criterion should be preferred above deal value matching and therefore the results indicate discounts for private companies and subsidiaries. The second row displays the result of the median PCD for matching on profitability. For private companies only a statistically significant discount is present for the EBIT multiple and for subsidiaries for the EBIT and EBITDA multiple. The rest of the results indicate no existence of a Private Company Discount. The results of the third method, for which matching is used on the basis of growth rates, give contradicting results. For subsidiaries premia should be payed when looking at the Sales Multiple, but discounts according to the EBITDA multiple. For

Table 7. Univariate analysis alternative matching

This table contains the results of the univariate analysis for four different matching procedures using the standard method to calculate PCD. PCD is calculated using (3) for private companies and subsidiaries and separately for the four multiples. The first number indicates the median discount or premium, for instance 0.30 indicates a discount of 30%. The second number the t-value and the third number the amount of observations. ***, **, * represent significance at the 1%, 5%, and 10% levels, respectively.

PCD Sales PCD EBITDA PCD EBIT PCD Net Income

Private Subsidiary Private Subsidiary Private Subsidiary Private Subsidiary

(27)

27

private companies weakly significant results indicate discounts, only for the EBIT and net income multiple. Also, growth rates are not available for a large portion of the companies and observations drop severely. When matching for sales and for profitability, displayed in the fourth row of table 7, results are comparable to the results when is matched for profitability only. The EBIT and EBITDA multiples support discounts for subsidiaries and the results for the EBIT multiple support discounts for private companies.

5.2 Multivariate analysis

In this section the results of the multivariate analysis are discussed. Firstly, the multivariate analysis of private companies (table 8) then of subsidiaries (table 9). The main factor affecting Private Company Discount for private companies, is the size of the firm. How larger the company how lower the median PCD will be. For subsidiaries, the profitability of the target firm has the most significant impact on the Private Company Discount. The results indicate that the median PCD increases when profitability of the target firm is higher.

Private companies

(28)

28

firm is larger the acquiring firm is larger as well. The constant is, except for the Net Income Multiple (1), significantly different from zero at the 1% level. The constant indicates a median discount for private companies when all coefficients of variables are zero. The Pseudo R-squared values are low. This indicates that the variables do not explain the variability of the median discount very well.

Subsidiaries

The results of the multivariate analysis for subsidiaries display different results that for private companies (table 9, on the next page). Most of the constants are not significant and only one constant is significant at the 1% level. Surprisingly and in contrast to what is predicted, higher profitability is the main driver of PCD for subsidiaries. The coefficients are statistically

Table 8. Multivariate analysis Private Companies

This table displays the results of the multivariate analysis performed on the Private Company Discount for private companies. For each multiple a quantile regression is run separately for target and acquirer characteristics. The first number displays the effect on the Median PCD of the explanatory variable. The second number is the t-statistic. ***, **, * represent significance at the 1%, 5%, and 10% levels, respectively. The variables are defined in appendix

(29)

29

significant at least at the 5% level and indicate, for every multiple, that when profitability is higher, PCD is higher as well. The amount of liquid asset in the company relative to sales, seems to have the predicted impact when looking at the Sales multiple. Most variables in the 4 models explaining the impact of acquirer characteristics on PCD are not significant. For the model of the sales multiple, when the acquiring companies is relatively large the median PCD is reduced. For the EBIT multiple there seems to be a positive relation between whether the acquiring firm is public and the height of the PCD. The expectation wat that when a firm is public, it tends to overpay, so this results in inconsistent to expectation. The Pseudo R-squared indicates that the model does not explain the variance of the median PCD very well..

Table 9. Multivariate analysis Subsidiaries

This table displays the results of the multivariate analysis performed on the Private Company Discount for private companies. For each multiple a quantile regression is run separately for target and acquirer characteristics. The first number displays the effect on the Median PCD of the explanatory variable. The second number is the t-statistic. ***, **, * represent significance at the 1%, 5%, and 10% levels, respectively. The variables are defined in appendix

(30)

30

5.3 Robustness

To test whether the findings of the quantile regressions are robust to changes in the model, OLS regressions are run and their results compared to the standard model. The results of the OLS regressions are displayed in Appendix D. For the most part the results of the OLS regressions match the results of the quantile regressions.

(31)

31

6 Conclusion

The aim of this paper was to determine whether a private company discount should be paid for private companies and subsidiaries compared to public peer companies. Results of univariate analysis, multivariate analysis and robustness check, suggest that private companies trade at a discount compared to public peer companies. For subsidiaries, univariate analysis also supports the existence of discounts compared to public companies.

(32)

32

7 References

Black, F., Jensen, M. C., & Scholes, M. (1972). The Capital Asset Pricing Model: Some Empirical Tests. Studies in the Theory of Capital Markets, 81(3), 79–121.

Brav, A., Geczy, C., & Gompers, P. A. (2000). Is the abnormal return following equity issuances anomalous. Journal of Financial Economics, 56, 209–249.

Damodaran, A. (2012). Equity Risk Premiums: Determinants, Estimation and Implications: The 2012 Edition.

Damodaran, A. (2016). Investment Valuation (Second Edition).

Emory, J. D. (2002). Discounts for Lack of Marketability: Emory Pre-IPO Discounts Studies 1980-2000.

Finnerty, J. D. (2018). Measuring the Discount for Lack of Marketability An Empirical Investigation.

Fuller, K., Netter, J., & Stegemoller, M. (2002). What Do Returns to Acquiring Firms Tell Us? Evidence from Firms That Make Many Acquisitions. The Journal of Finance, 57(4), 1963-1793.

Jaffe, J. F., Jindra, J., Pedersen, D. J., & Voetmann, T. (2019). Do Unlisted Targets Sell at Discounts? Journal of Financial and Quantitative Analysis, 54(3), 1371–1401.

https://doi.org/10.1017/S0022109018001060

Jensen, M. C. (1986). Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers.

The American Economic Review, 76(2), 323–329.

Koeplin, J., Sarin, A., & Shapiro, A. C. (2000). THE PRIVATE COMPANY DISCOUNT.

Journal of Applied Corporate Finance, 12(4), 94–101.

Koller, T., Goedhart, M., & Wessels, D. (2015). Valuation: Measuring and Managing the Value of Companies (5th Edition).

Kooli, M., Kortas, M., & L'Her, J.‑F. (2003). A New Examination of the Private Company Discount: The Acquisition Approach.

Longstaff, F. A. (1995). How Much Can Marketability Affect Security Values? The Journal

of Finance, 50(5), 1767–1774.

Longstaff, F. A. (2001). Optimal Portfolio Choice and the Valuation of Illiquid Securities.

The Review of Financial Studies, 14(2), 407–431.

Maher, J. (1976). Discounts for Lack of Marketability for Closely Held Business Interests.

The Tax Magazine, 54(9), 562–571.

Nath, E. W. (1990). Control premiums and minority interest discounts in private companies.

Business Valuation Review, 9.2, 39–46.

Officer, M. S. (2007). The price of corporate liquidity: Acquisition discounts for unlisted targets. Journal of Financial Economics, 83(3), 571–598.

(33)

33

Opler, T., Pinkowitz, L., Stulz, R., & Williamson, R. (1997). The Determinants and Implications of Corporate Cash Holdings.

Paglia, J. K., & Harjoto, M. (2010). The Discount for Lack of Marketability in Privately Owned Companies: A Multiples Approach. Journal of Business Valuation and Economic

Loss Analysis, 5(1). https://doi.org/10.2202/1932-9156.1089

Poulsen, A. B., & Stegemoller, M. (2008). Moving from Private to Public Ownership: Selling Out to Public Firms versus Initial Public Offerings. Financial Management, 81–101. Silber, W. L. (1991). Discounts on Restricted Stock: The Impact of Illiquidity on Stock

Prices. Financial Analysts Journal, 47(4), 60–64.

Wruck, K. H. (1989). Equity ownership concentration and firm value: Evidence from private equity financing.

(34)

34

8 Appendix

Appendix A addition literature

Option valuation

Option pricing models are normally used to access the values of options. In recent years option pricing theory has been used to value assets other than options. Damodaran (2016) states that the basis for this lies on the assumption that the value of an asset cannot surpass the discounted cash flows that this asset generates, if the cash flows are conditional on an event happening or not happening.

Options are valued using a binomial pricing model developed by Black et al. (1972). An asset can be valued as an option when the payoff of the asset can be seen as a function of the underlying asset. An asset can be valued as a call option, when the assets pays off a certain amount once it surpasses a certain threshold value, or as a put option, when a payoff is generated when the underlying value falls below a certain value (Damodaran, (2016). For example, a patent could be valued as a call option, because the owner can use a certain technology for a certain time, in which competitors cannot. When the net payoff of a patent over its lifetime is higher than the opportunity costs of developing the patent the option of using the patent will be exercised.

The equity of a firm could also be seen as a call option. The underlying asset is the market value of the firm and the face value of debt is the strike price, with the end date of the debt as expiration date (Damodaran, (2016). The limitations of using this method are, that when used for companies that are not publicly traded, the values have to be estimated, also unrealistic assumption have to be made, such as constant dividend yield and variance.

Option studies

(35)

35

methods. Also, liquidity is mentioned as an important influence on price discounts Longstaff (2001). More recent research by Finnerty (2018) stresses the importance of research in private company deals, because of the growth of private equity. In the paper, Measuring the Discount for Lack of Marketability, a DLOM model is developed using option pricing theory. Pre-IPO private equity deals are used as for development of the model. Finnerty (2018) suggests that for companies, which lack marketability, liquidity is also a problem.

The limitations of option studies are that in case of Longstaff (1995) only the upper bounds of the DLOM can be obtained and is built on unrealistic assumptions Paglia and Harjoto (2010), because by using lookback options, the timing of an investor should be perfect. Also is suggested, that more factors play a role by measuring the discounts of private companies. Restricted stock studies

Restricted stock consists of securities traded by publicly traded companies with a private like character. This stock is not registered with the SEC and can be sold privately. Comparable to private stock this restricted stock lacks marketability and is therefore used to estimate the Discount Lack of Marketability (DLOM) or PCD.

Maher (1976) researched the DLOM using data of mutual funds from 1969 till 1973. In his study he analyzed differences in priced paid for restricted stock in comparison to unrestricted stock. He first found an average discount of 35.43% and thought this was too high. Therefore, to overcome the problem of outliers, he eliminated the top and bottom 10%, which resulted in a discount of 34.73%. The conclusion of the paper was that the discount should be around 35%, which was in that time significantly higher than previous research suggested. Wruck (1989) analyzes private sales of equity and finds in her paper an average discount of 17,6% for private sales in comparison to public issues. Silber (1991) Follows up on the research on restricted stock studies and finds an average discount of more than 30% for restricted stock and stresses the importance of liquidity, by valuing common stock.

(36)

36 IPO studies

To overcome the problems with the restricted stock studies to measure the discount lack of marketability, research moved to IPO studies, Option studies, and Acquisition studies. An IPO is an Initial Public Offering. In an Initial Public Offering a private company raises capital by issuing its shares to the public. When a company executes an IPO, they must disclose the last three years of stock transactions, this offers potential for research.

(37)

37

Appendix B Variable definitions

Table B1. Explanatory variable list

Net Sales Net sales contain the primary source of revenue or total

revenues for the last 12 months ending on the date of the most recent financial information prior to the announcement of the transaction.

LnAssets Lnassets is the natural log of the total amount of assets the

target company, which is total amount of assets for the last 12 months ending on the date of the most recent financial information, prior to the announcement of the deal.

Profitability Profitability is the profitability of the target company. The

profitability is calculated by dividing net income before extraordinary items and preferred dividends by the net sales of the target company. Both variables come from the last financial information available prior to announcement.

Growth rates Growth is the 3-year EBIT growth rate of the company

before announcement date. Both variables come from the last financial information available prior to announcement.

Liquid assets The liquid assets variable is defined as the amount of liquid

assets the target company holds in the last 12 months up to announcement date of the deal relative to net sales. Liquid assets include cash and temporary investment vehicles for cash.

Cash transaction Cash transaction is a dummy variable which equals to 1

when the acquisition is all cash financed.

Public status Public status is a dummy variable which equals to 1 when

the acquiring firm is a public company

Same industry Same industry is a dummy variable which equals to 1 when

the acquiring firm is from the same industry as the

LnAssetsAc LnAssetsAC is the natural log of the total amount of assets

the acquiring companies. Total amount of assets is for the last 12 months ending on the date of the most recent financial information, prior to the announcement of the deal.

Table B2. TRBC Economic Sector Classifications

No. Industry

50 Energy

51 Basic Materials 52 Industrials

53 Cyclical Consumer Goods & Services 54 Non-Cyclical Consumer Goods & Services 55 Financials

56 Healthcare 57 Technology

(38)

38

Appendix C

Table C1. pairwise correlation matrix

Target Characteristics LnAssets Profitability Growth rates Liquid assets Cash transaction

LnAssets 1.00

Profitability -0.06 1.00

Growth rates -0.07 -0.01 1.00

Liquid assets 0.13 0.05 -0.03 1.00

Cash transaction -0.14 0.03 -0.07 -0.04 1.00

Acquirer characteristics Public status Same industry LnAssetsAc

Public status 1.00

Same industry -0.09 1.00

LnAssetsAc -0.09 0.06 1.00

Table C2. EBITDA PCD summary statistics

(39)

39

Appendix D

Table D1. Multivariate analysis Private Companies, winsorized mean 10%/90%

This table displays the results of the multivariate analysis performed on the Private Company Discount for private companies. For each multiple a regression is run separately for target and acquirer characteristics using robust standard errors. The first number displays the effect on the PCD of the explanatory variable. The number between parentheses is the t-statistic. ***, **, * represent significance at the 1%, 5%, and 10% levels, respectively. The variables are defined in Appendix B

(40)

40

Table D2. Multivariate analysis Private companies winsorized mean 25%/75%

This table displays the results of the multivariate analysis performed on the Private Company Discount for private companies. For each multiple a regression is run separately for target and acquirer characteristics using robust standard errors. The first number displays the effect on the PCD of the explanatory variable. The number between parentheses is the t-statistic. ***, **, * represent significance at the 1%, 5%, and 10% levels, respectively. The variables are defined in Appendix B

(41)

41

Table D3. Multivariate analysis Subsidiaries, winsorized mean 10%/90%

This table displays the results of the multivariate analysis performed on the Private Company Discount for subsidiaries. For each multiple a regression is run separately for target and acquirer characteristics using robust standard errors. The first number displays the effect on the PCD of the explanatory variable. The number between parentheses is the t-statistic. ***, **, * represent significance at the 1%, 5%, and 10% levels, respectively. The variables are defined in Appendix B

(42)

42

Table D4. Multivariate analysis Subsidiaries, winsorized mean 25%/75%

This table displays the results of the multivariate analysis performed on the Private Company Discount subsidiaries. For each multiple a regression is run separately for target and acquirer characteristics using robust standard errors. The first number displays the effect on the PCD of the explanatory variable. The number between parentheses is the t-statistic. ***, **, * represent significance at the 1%, 5%, and 10% levels, respectively. The variables are defined in Appendix B

Private Company Discount: Evidence of Acquisition Targets