Development of Prices Paid For Unquoted Companies in
Relation to Prices Paid For Quoted Companies in Europe.
Investigating the Private Company Discount
MSc Business Administration Specialization Finance Profile: Corporate Financial Management
Rijksuniversiteit Groningen Author: Martijn Nuis Student ID: 1303775
1stSupervisor: Dr. ing. N. Brunia 2ndSupervisor: Drs. M.E. Helmantel
Development of Prices Paid For Unquoted Companies in
Relation to Prices Paid For Quoted Companies in Europe.
Investigating the Private Company Discount
Abstract
Preface
1 INTRODUCTION ... 1
1.1 Research Problem ... 2
1.2 Relevance... 3
1.3 Structure... 4
2. Background / Literature... 5
2.1 Valuation of unquoted companies... 5
2.1.1 Liquidity ... 5
2.1.2 Risk Factors ... 6
2.2 Influence of private company discount on valuation methods... 7
2.3 Valuation methods ... 7
2.3.1 Description and literature of the multiple method ... 7
2.3.1.1 Conclusion ... 9
2.3.2 Description and literature of the Discounted Cash Flow method ... 10
2.3.2.1 Conclusion ... 10
2.3.3 Relation Multiples/DCF ... 11
3. Background / Methods of research into the valuation of unquoted companies ... 12
3.1 Developments/BDO Stoy Hayward ... 12
3.1.1 Comments and conclusion... 13
3.2 Approaches in calculating the small firm discount ... 13
3.2.1 IPO Approach ... 14
3.2.2 Restricted Stock Approach ... 15
3.2.3 Acquisition approach... 15
3.2.4 Conclusion ... 15
4. Data & Methodology ... 16
4.1 Descriptive statistics ... 16
4.2 Descriptive statistics of the dataset ... 18
4.3 Research with real enterprise values ... 19
4.3.1 Descriptive statistics and first analysis ... 19
4.3.2 Conclusion ... 20
4.4 Research with estimated enterprise values... 21
4.4.1 Description of the independent variables ... 21
4.4.2 Methodology... 21
4.4.3 Results ... 24
4.4.4 Conclusions ... 27
4.5 Eastern and Western Europe separated ... 27
4.6 Cross checks on the dataset... 29
5. Reflection... 30
5.1 Conclusions... 30
5.2 Suggestions for further study ... 30
References ... 32
1 INTRODUCTION
”The merger and acquisition market did not slow down, but strongly changed character.”(Het
Financieele Dagblad, January 14th 2008). The total value of all deals in the Netherlands in 2007 was half the value of the top years in the Merger and Acquisition (M&A) market. However, the total number of deals completed was undiminished, averaging about 100 deals each month. The decreasing value of all transactions actually completed as well as the stable number of transactions are results of two developments. The first development is the decreasing number of high value deals involving large high value acquired companies. Second development is the increasing number of low value deals involving small target companies. Does this change in character have its influence on the valuation of companies?
The valuation of unquoted companies is challenging due to limited available information. The non-existence of a ready market to sell shares of unquoted companies means that the standard techniques for estimating risk parameters, necessary in valuation, are not feasible and therefore cannot be used for unquoted companies. The discount on unquoted companies is demanded by investors if they are unable to sell their shares within a given time period. Kooli et al [2003] argue that the lack of liquidity is costly to investors, not only because of cash flow considerations, though it can also cause them to miss opportunities to rebalance their portfolio and allocate capital to alternative assets.
the value of quoted companies reduced by a private company discount, it is interesting to know if the private company discount changed over the last 8 years. In this study we will analyze if the private company discount did change over the last 8 years. If it did change, the nowadays applied private company discount, should be adjusted such that valuing a company results in the most accurate market price of a stand alone company.
1.1 Research Problem
In this study we will investigate the private company discount and whether the private company discount became smaller over the last 8 years in Eastern and Western Europe.
The key question is:
Is there a development in the difference between relative prices paid for unquoted companies and prices paid for quoted companies in Eastern and Western Europe from 2000 through 2007?
Sub questions are:
1. Does the private company discount exist in Eastern and Western Europe from 2000 through 2007?
2. How much did prices of unquoted companies and quoted companies change over the last 8 years? 3. Should the applied private company discount, as used in valuation, on unquoted companies be
lower today than it was 8 years ago?
Answers to these questions should help us determine the proper discount for an unquoted company.
To investigate relative prices of quoted companies and unquoted companies three multiples will be analyzed:
1. EBIT multiple: The Enterprise Value to Earnings Before Interest and Taxes (EV/EBIT);
2. EBITDA multiple: The Enterprise Value to Earnings before Interest Taxes Depreciation and
Amortization (EV/EBITDA);
3. Sales multiple: The Enterprise Value to Sales (EV/Sales).
valuation multiple. Companies with the closest multiples are identified as peer firms. In our study it will be investigated if the difference of a company being quoted or unquoted has influence on its multiples. A regression analysis will be carried out to analyze the influence. Furthermore, guided by valuation theory, five other value drivers will be investigated by using the regression model as proposed by Bhojraj and Lee [2002].
1.2 Relevance
There are several reliable valuation methods (e.g. Discounted Cash Flow method, multiples method, Adjusted Present Value [APV]). Each method has its advantages and disadvantages and each is best in different situations. Valuing an unquoted company is different from valuing a quoted company. An unquoted company deserves a discount compared to a quoted company because of the illiquidity of its shares (among others: Koeplin et al. [2000], Kooli et al. [2003], Rijken et al. [1999]). Additionally, small firms, which are mostly unquoted, bear extra unsystematic risk compared to quoted companies. For example, in case of a small number of managers, if a chief manager passes away, the firm will have difficulty surviving (Veeger [2005]). In general, unsystematic risk is not value relevant, since investors can diversify their portfolio to overcome this sort of risk. Real life situations at BDO CampsObers Corporate Finance B.V. show that the average investor in unquoted companies does not diversify his portfolio. Therefore, the required rate of return on equity applied in valuation is higher for unquoted companies than for quoted companies. It also depends on the characteristics of the investor. Eduard [1999] also mentions this argument for a higher required rate of return for unquoted companies. Practice at BDO CampsObers Corporate Finance B.V. shows that if the higher required rate of return on equity for unquoted companies is not applied, the calculated stand-alone value (going concern, without synergy advantages etc.) does get close to the stand-alone market price. The market is always right: if a price cannot be realized in the market, it is not the market price.
in the pricing of companies as described above. This is especially important for consultants involved in the valuation of unquoted companies.
At BDO CampsObers Corporate Finance B.V. more unquoted companies are being sold over the last few years. Most of this activity is the result of the “baby-boom”-- the large group of people born just after World War II. These people reach the age of retirement, and for those having their own company it is time to sell their company. It is expected that the high amount of unquoted companies sold by people born in the baby boom will continue for the coming years as more people reach the age of retirement. It will therefore be to the advantage of corporate finance advisors to learn what the current accurate private company discount is.
This study investigates the private company discount and tries to determine whether the gap between the price paid for unquoted companies and the price paid for quoted companies has narrowed in Eastern and Western Europe over the last 8 years. If so, the private company discount now applied on unquoted companies used in valuation methods must be adjusted accordingly.
1.3 Structure
2. Background / Literature
This chapter will focus on valuation of unquoted companies. It will be explained what the difference is between valuing a quoted company and an unquoted company as also why there is a difference in treatment. Furthermore two commonly used valuation methods are described. The relation between the two valuation methods is discussed in the last section of this chapter.
2.1 Valuation of unquoted companies
The small firm effect and size effect are closely related subjects in this study (see, for example, Chan et al. [1985], Jegadeesh [1992], Cheung [1999], Heijden [1999], Jonk [2002]). Studies investigating the small firm effect or size effect investigate the influence of the size of a company on for example enterprise value. Since the average value of a quoted company is higher than the average value of an unquoted company, obviously the small firm effect should also be found in the difference of a firm being quoted or unquoted. The private company discount is caused by two factors: the liquidity of company shares and the company risk. These factors will be described in this section.
2.1.1 Liquidity
Liquidity refers to how quickly an asset can be converted into cash without the owner incurring substantial transaction costs or price concessions (Bajaj et al. 2001). Liquidity is one of the two most important factors involving the private company discount.
Some studies (for example Steenbeek and Vliet [2005]) investigating the small firm effect are based on the differences between small listed companies and large listed companies. Despite the fact they only investigate data of listed companies, these studies are closely related to the private company discount. Although shares of unquoted companies are less marketable than shares of small quoted companies, (because of the absence of a stock exchange for unquoted companies), shares of small quoted companies are also less marketable than shares of large quoted companies. The difference in marketability of shares of quoted companies is a result of less trade in shares of small quoted companies (Veeger, 2005). Shares of large quoted companies are traded and priced continuously. The absence of continuous pricing causes the price of a share to be inaccurate since not all available information is represented (Roll, 1981). The price paid in the last trade, which could be several months before, does not reflect information from after that transaction date.
when the asset’s value does not fluctuate that much). Lower liquidity generates higher potential opportunity costs which will cause the demand for a higher discount. For example, if an asset is worth less than expected, opportunity costs are high. To compensate this uncertainty, investors require higher discounts. Second, and similar to the first argument, the more difficult it is to estimate the value of an asset, the less liquid the asset will be. An investor will not pay a high price for an asset if its value is unknown. Therefore, the investors require an extra discount. Thirdly, Bajaj et al. [2002] mention an assets substitutability as a factor likely influencing an asset’s liquidity. A lack in the existence of close substitutes of an asset will cause a lack in its liquidity. Fourth, the period of a restriction to sell an asset is of influence on its liquidity. All else being equal, the longer the duration of liquidity restrictions, the higher the demanded discount will be. Last, the larger the amount of an asset being sold, the less liquid the asset will be and the lower its value will be. This last effect, according to Bajaj et al. [2001], probably arises by two different causes. First, the higher the amount sold, the less marketable the remaining market for the asset will be. Second, if the transaction size increases, the pool of potential investors will decrease because of capital constraints and the willingness to diversify a portfolio.
Besides the factors described by Bajaj et al. [2001], the trade of shares of unquoted companies is more expensive than the trade of shares in quoted companies (Eduard, 1999). The trade of shares of unquoted companies bear higher transaction costs because sellers and buyers find each other more difficult than in the market of quoted companies. Most investors in unquoted companies are other companies acquiring an unquoted company. In most cases these buyers spend much time (and money) to find the right acquisition (BDO CampsObers Corporate Finance B.V., Eduard, 1999). Furthermore, the availability of necessary information is less for unquoted companies than for quoted companies. This makes valuing the target company more time consuming and expensive (Banz, 1985). If it is not possible to obtain all relevant information, investors are only willing to invest in the company if they earn a higher return (Banz 1985).
2.1.2 Risk Factors
enterprise value. Also according to Isberg and Thies [1992] and Jegadeesh [1992] investors require higher returns for higher risks, even for unsystematic risks.
2.2 Influence of private company discount on valuation methods
Valuation methods are not applied differently on unquoted companies than on quoted companies. Although, in for example the discounted cash flow method, the reference in calculating the value of an unquoted company, is a quoted company. The rate at which future operating cash flows are discounted for a specific quoted company is increased with an additional (small-)firm specific premium. As a result the unquoted company is valued lower than the quoted company (everything else being equal). Using the multiple method, a lower multiple will be used to value an unquoted company than for valuing a quoted company. In the next section the mentioned valuation methods are described in more detail.
2.3 Valuation methods
This study uses the multiple method and the discounted cash flow method to investigate the private company discount. This section starts with a description of the multiple method, followed by a description of the discounted cash flow method. It finishes with a description of the relation of the two methods.
2.3.1 Description and literature of the multiple method
A multiple is a ratio used to value a company. For example, Finnerty and Emery [2000] describe this comparable valuation technique as estimating a company’s value by multiplying a ratio derived from a set of comparable companies times the company’s earnings before interest taxes depreciation and amortization (EBITDA), earnings before interest and taxes (EBIT), revenue, or any other performance measure. Liu et al. [2002] find that the comparable company method is a very reliable method for valuing shares of listed companies, especially when based on forward looking multiples. Forward looking multiples are multiples based on expected future performance measures.
earlier times, when less information about unquoted companies was available. An unquoted company can also be valued by multiplying its performance measure with the corresponding trading multiple of a comparable quoted company. However in theory a quoted company is not comparable with an unquoted company since at least the discount for illiquidity of its shares is different.
Over and above the problem of which is the appropriate set of comparable companies, the other problems of the valuation of unquoted companies (as stated before) arise: the illiquidity of their shares (among others: Koeplin et al. [2000], Kooli et al.[2003], Rijken et al. [1999]) and the on average higher unsystematic risk that unquoted companies bear.
Where BDO Stoy Hayward [2007] uses the net profit multiple to analyze the private company discount, this study will investigate the development of the EBIT, EBITDA and Sales multiples. Table 2-1, presented hereafter, describes the valuation ratios used.
Variable Description
Enterprise value The enterprise value is calculated by multiplying the number of actual target shares outstanding by the offering price and then adding the book value of total debt and debt equivalents before subtracting marketable securities. Book value of total debt, debt equivalents and marketable securities are based on the most current financial information prior to the completion of the transaction. Transactions of which the most current financial information is older than 1 year are excluded from the sample because of possible changes in the figures.
EBIT multiple Ratio of enterprise value to EBIT where EBIT is defined as earnigs before interest and taxes over 12 months ending on the date of the most current financial information prior to the transaction. The most current financial information is at maximum 1 year old.
EBITDA multiple Ratio of enterprise value to EBITDA where EBITDA is defined as earnigs before interest and taxes depreciation and amortization over 12 months ending on the date of the most current financial information prior to the transaction. The most current financial information is at maximum 1 year old.
Sales multiple Ratio of enterprise value to sales for the 12 months ending on the date over 12 months ending on the date of the most current financial information prior to the transaction. The most current financial information is at maximum 1 year old.
Table 2-1 Description of valuation ratios used
EBITDA multiples are used, and we do not compare values discounted by the WACC, denominators are barely influenced by tax rates and capital structure.
Other possible measures are the sales multiple and the book multiple. The reason why at times the sales multiple is used, is that an acquirer looking for the expansion of its sales could be interested in the price paid per additional Euro of sales. If the acquirer’s profit per additional sales is higher than the price paid per additional sales, this is at first sight a profitable expansion. If the company can take over the sales only, without anything else of the company, and integrate it in the own production because of over capacity, the transaction generates a higher NPV. The book multiple (book to market ratio) shows how much the acquirer will pay for every Euro of invested capital in the target.
The quarterly executed study by BDO Stoy Hayward [2007] shows that transaction multiples (net profit multiples) of unquoted companies (including all industries except for banking) are lower than trading multiples of quoted companies. Besides that, Pagano et al. [1998] show that for decennia shareholders holding a portfolio of shares of small companies earn higher returns on average than shareholders of large companies. Higher required rates of return on equity lead to higher weighted average cost of capital. In the hereafter described DCF method, enterprise value is calculated by discounting future operating cash flows by the weighted average cost of capital. Ceteris Paribus, higher costs of capital do lead to lower enterprise values. Higher costs of capital mean that the company is liable to higher opportunity costs of capital. Since a profitability index is calculated as the discounted (at cost of capital) future operating cash flows divided by the initial investment, higher costs of capital lead to a lower profitability index. Since small companies bear higher costs of capital small companies are valued relatively low in comparison with large companies. The phenomenon of lower transaction multiples of unquoted companies in comparison to transaction- (and trading) multiples of quoted companies is known as the small-firm effect.
2.3.1.1 Conclusion
2.3.2 Description and literature of the Discounted Cash Flow method
Although the multiple method is relatively easy in use and is able to give reliable results if accurately applied, methods in which future operating cash flows are discounted at the corresponding discount rate have also showed their relevance in valuing companies (Koller et al., 2005). Since the value of a company in the DCF method is represented by future operating cash flows, operating cash flows must be forecasted. Future operating cash flows are discounted to find the enterprise value. To use any discount method, the accurate discount rate has to be found. The discount rate is calculated as the weighted average cost of capital of the optimal structure of debt and equity of the company. The cost of debt is the percentage at which a healthy company is able to borrow money. This rate can be obtained by adding a spread to the risk free rate. The risk free rate is roughly equal to the interest rate of a 10 year government bond. The spread depends on the economical situation although, is the same for every company in the same economical area. The calculation of the cost of equity is more complicated. The cost of equity should be equal to the return an investor gets investing his money in an asset with similar risk and similar expected return. To calculate the required rate of return on equity for quoted companies, for example the Capital Asset Pricing Model (CAPM) or the Arbitrage Pricing Theory (APT) can be used. These models can be applied on quoted companies since they require company specific risk factors (betas). Betas represent the systematic risk of a company’s equity in comparison to the market as a whole. About unquoted companies Bowman and Bush [2006 p. 1] say:
“Comparable company analyses provides a reasonably accurate estimate of beta, when the comparable companies are similar in size to the private company”. However, most unquoted
companies are smaller than quoted companies. Therefore, it is in most situations not possible to make this reasonably accurate estimation of an unquoted companies’beta.
2.3.2.1 Conclusion
2.3.3 Relation Multiples/DCF
3. Background / Methods of research into the valuation of unquoted companies
In this chapter different methods to investigate the private company discount will be described. It starts with a description of a quarterly executed study on the private company discount in Britain. Then three approaches that are used in existing literature are described.
3.1 Developments/BDO Stoy Hayward
Figure 3-1 shows the P/E ratios of transactions of unquoted target companies and quoted target companies in Britain over the last 7 years as calculated by BDO Stoy Hayward [2007]. BDO Stoy Hayward publishes the Private Company Price Index every quarter of a year. The PCPI is calculated as the arithmetic mean of the Price/Earnings ratios of deals where sufficient information (e.g. transaction price, earnings) has been disclosed. Earnings are defined as net after tax profits. Reasons for the price changes between unquoted and quoted companies are not investigated in their publication but, the following reasons are proposed: rise of private equity funds raised, declining investor sentiment with respect to the quoted company markets, the increase in competition in the unquoted company market, the availability of professional advice to unquoted company vendors, increasing sophistication of financing instruments and investor protections with unquoted companies and the recent credit crunch (BDO Stoy Hayward, Private Company Price Index, 2007).
0 5 10 15 20 25 20 01 20 02 20 03 20 04 20 05 20 06 20 07 Q uoted c ompanies P /E Unquoted c ompanies P /E L ineair (Q uoted c ompanies P /E ) L ineair (Unquoted c ompanies P /E )
Figure 3-1 Private Company Price Index Q1 2001 –Q3 2007. Source: BDO Stoy Hayward
-2 0 2 4 6 8 10 12 2002 2003 2004 2005 2006 2007 gap L ineair (gap)
Figure 3-2 Gap between P/E ratio’s of quoted- and unquoted companies
As far as we know there are no other studies in which the development over time of multiples of unquoted and quoted target companies is described. Neither are there, as far as we know, any published studies that investigate the private company discount in Eastern and Western Europe. However, many empirical studies investigated the price difference between unquoted and quoted companies in the United States (among others: Bajaj et al. [2001], Koeplin et al. [2000], Rijken et al. [1999]).
3.1.1 Comments and conclusion
The study of BDO Stoy Hayward investigates net profit multiples. As explained earlier, net profit multiples are dependent on capital structure. Interest costs do not belong to operating costs since the capital structure of a single company can be restructured without changing operations. Also a single investor can diversify its own portfolio becoming independent of the capital structure of the company. Therefore, enterprise value is also independent of capital structure. Taking into account these arguments, the net profit multiple is not the most representative multiple to analyze enterprise values.
Furthermore, although it seems that the gap between the multiples of quoted and unquoted companies in BDO Stoy Hayward’s study is declining, their findings are not tested on significance.
3.2 Approaches in calculating the small firm discount
Approach Description Pitfalls
IPO approach 1. Large discount;
2. Investors are insiders;
3. Only succeeded IPO's are investigated.
Restricted stock approach 1. Discount is attributable to many factors;
2.
Acquisition approach 1.
2. Synergie advantages are included in the transaction price.
Comparison of stock prices before and after an IPO.
Comparison of two types of shares in the same firm where the one is marketable and the other is not.
Comparison of transaction prices of two firms where one firm is quoted and the other is unquoted.
Discount may represent compensation for future services offered.
In unquoted companies, the manager often also is the owner / seller. The seller could be compensated in another way than by transaction price;
Table 3-3 Description approaches
3.2.1 IPO Approach
In the Initial Public Offering (IPO) approach, the price of the same stock before and after an IPO is compared. John Emory1 and Willamette Management Associates2 conducted series of IPO studies (respectively between 1980 and 1997 and between 1975 and 1992) (Bajaj et al. [2001]). Their approach will not be used in this study because of three pitfalls. Bajaj et al. [2001] discerned the following three pitfalls:
The first pitfall is that the discount found in an IPO approach is implausibly large if applied as a private company discount in the valuation of unquoted companies. An average discount found in IPO approaches of 46% means that an investor earns an 85% rate of return (46%/[100%-46%]=85%). If the investor buys his shares six months prior the IPO, he can earn a 242% rate of return on a yearly base3. The second pitfall mentioned by Bajaj et al. [2001] is the type of the transaction. Transactions before an IPO are likely to be done by other investors than transaction done at the time of the IPO. Investors investing their money in shares prior an IPO probably are insiders who know the company well. They invest in the company with the intention to go public in the near future. The discount they actually earn when going public could be seen as a service costs for e.g. management services and / or capital provisions (in case the investor is a venture capitalist). The third pitfall, also mentioned by Koeplin et al. [2000] and Heijden [1999], is that the IPO approach only investigates companies which succeeded in going public. Companies which perform badly will not succeed in going public. Selecting companies by this method results in a higher liquidity discount than a discount caused only by illiquidity.
1http://www.emoryco.com/
2http://www.willamette.com/
3An investor investing 100 Euros owns 185 Euros if he sells his shares after 6 months just after the IPO.
3.2.2 Restricted Stock Approach
The second approach, which will not be used in our study, is the restricted stock approach. The restricted stock approach makes a contemporaneous comparison of the prices of two shares in the same company, where one share is marketable and the other is not. The discount is attributable to several factors; liquidity is only one. Bajaj et al. [2001] argue that the discount regarding restricted stock may represent compensation for investors for future services offered (e.g. the investor also is a manager). Hertzel and Smith [1993] also mention this pitfall and find evidence that restricted stock discounts reflect a fee for expected services from investors. Furthermore, Kooli et al. [2003] agree with the arguments of Hertzel and Smith [1993].
3.2.3 Acquisition approach
The third approach is the acquisition approach, which is used in this study. It compares transaction prices of two companies, where one is a quoted company (marketable shares) and the other an unquoted company (non-marketable shares). This method avoids the pitfalls of both the IPO approach and the restricted stock approach, but it is subject to another flaw. Unquoted companies are mostly held by owners who also are the senior managers. If these owners are compensated in any other way than by the transaction price (for example, by being paid an above-market salary if they keep working after the transaction) the transaction price is lower compared to an otherwise comparable quoted company (Bajaj et al. [2001]). Another problem is that an acquisition price could be based on synergy advantages. In this situation, the stand alone value of the target company is lower than the transaction value.
3.2.4 Conclusion
4.
Data & Methodology
This chapter describes the selection criteria for the deals investigated. It also describes the statistics of the resulting datasets and the methodology used for the investigation. The analyses will be carried out using real enterprise values and using estimated enterprise values.
4.1 Descriptive statistics
To investigate the private company discount by use of the acquisition approach, we use data of completed deals. The required historical deal data will be obtained from the databases Amadeus top 250.000 (hereafter: Amadeus) and Zephyr of publisher Bureau Van Dijk. Amadeus provides financial data of the largest 250.000 European companies. Zephyr provides available information of many announced, pending, rumored, withdrawn and completed deals. All required information for this research is supplied by these databases, except for corporate tax rates. To find the corporate tax rates corresponding to the target company, the yearly survey of accountancy firm KPMG “KPMG’s
Corporate Tax Rate Survey”is used. In addition the website www.worldwide-tax.com is used in order
to find tax rates missing in the survey of KPMG.
The criteria used for the deal selection are the following:
Deal is completed between January 1st
of 2000 and January 1stof 2008; Deals in which at least 50% of the target’s shares is sold;
Target is an Eastern Europe or Western Europe based company4
; A positive (estimated) enterprise value is known.
Deals in the financial sector are not included since companies in this sector are predominately asset based. An enterprise value multiple is therefore not meaningful for the financial sector. This study will try to explain the multiples using the following formula:
g WACC ROIC g T EVmultiple 1 14The following countries are included: Albania, Andorra, Armenia, Austria, Belarus, Belgium,
Where:
EV multiple = Enterprise Value divided by a value driver (EBIT, EBITDA or Sales) T = Corporate Tax Rate
g = Expected future growth
ROIC = Expected future Return On Invested Capital WACC = Weighted Average Cost of Capital
Although the right hand side of this expression is the EBITA multiple, we will use it to try to explain the EBIT, EBITDA and Sales multiple.
Koller et al [2005] recommend to make some adjustments to the variables to give a better estimation of the enterprise value multiples. One of the recommendations is to use forward looking estimates instead of trailing data. Since our databases do not provide forward looking estimates, we use trailing data. Furthermore Koller et al [2005] recommend to choose comparables with similar prospects for ROIC and growth. The dataset used for our analysis is too small to compare only those prospects. Since we are unable to make the adjustments recommended, results could be less significant than when the adjustments were made.
All required variables collected and described in table 4-1. The relation between the variables is described more extensively in section 4.3.
Variable Description
Enterprise value The enterprise value is calculated by multiplying the number of actual target shares outstanding by the offering price and then adding the book value of total debt and debt equivalents and subtracting marketable securities. EBIT Earnings Before Interest and Taxes
EBITDA Earnings Before Interest Taxes Depreciation and Amortization
Sales Sales
Growth Geometric average yearly growth of the value driver over the last 5 years if year 1's up to year 5's values are available and positive. If not, year 2's value is used instead of year 1's value, etc. It is expected that this variable is postively correlated with the multiples.
ROIC Return On Invested Capital. To calculate the ROIC the following variables are needed: Corporate tax rate, EBIT, Working capital and Fixed assets. The corporate tax rate is the tax rate in the target's country in the year of the transaction. Average yearly EBIT of the 5 years prior to the transaction. Average year-end working capital of the 5 years prior to the transaction. Average year-end fixed assets of the 5 years prior to the transaction. ROIC is calculated as
ROIC is expected to be positively correlated with the multiples.
Book leverage Book value of debt divided by the book value of equity. Both values for the 12 months ending on the date of the most current (not older than 1 year) financial information prior to the transaction. Although Gebhardt et al. (2001) find that book leverage is not significant in explaining cost of capital, this variable is included to capture elements not captured by other variables. It is expected that book leverage is negatively correlated with all multiples.
Taxes The corporate tax rate is the tax rate in the target's country in the year of the transaction. Taxes are expected to be negatively correlated with the multiples.
Dummy variable Dummy variable which takes value 1 if the company is quoted and 0 if the company is unquoted. Expectations are that quoted companies are valued higher, so the dummy variable is expected to be postively correlated with the multiples.
Dependent variables
Independent variables
Each deal and each company is provided with a unique identification number. To find relevant information which is not supplied by Zephyr, the identification numbers of the target companies are imported in Amadeus. Amadeus does not supply the required information of all target companies since it only contains data of the 250.000 largest European companies. Deals involving a target company of which no information is available in Amadeus are excluded from the dataset. The resulting deals include at least one of the relevant variables of the target company or the deal.
4.2 Descriptive statistics of the dataset
. We end up with three datasets. One represents the EBIT multiple, one represents the EBITDA multiple and one represents the sales multiple. The descriptive statistics of these three datasets are presented in the tables below.
Mean Median Maximum Minimum Std. Deviation Skewness Kurtosis Jarque - Bera Probability Observations Pooled Quoted 13.3 11.4 56.4 0.1 8.7 1.6 6.1 620.9 0.0 753 Unquoted 17.6 15.0 61.2 0.2 10.7 0.9 3.2 281.7 0.0 1,938 2000 Quoted 13.2 11.1 49.5 2.4 8.6 2.0 7.6 247.5 0.0 157 Unquoted 17.7 16.3 44.3 4.5 9.7 1.1 3.8 6.4 0.0 29 2001 Quoted 12.6 9.5 51.3 0.3 10.4 1.7 5.6 60.1 0.0 79 Unquoted 17.1 16.1 43.7 3.7 9.5 1.1 4.0 5.3 0.1 23 2002 Quoted 10.3 8.4 56.4 0.1 8.8 1.9 7.6 139.4 0.0 97 Unquoted 17.9 14.5 48.1 0.2 11.4 0.9 3.0 16.4 0.0 129 2003 Quoted 11.4 9.5 41.0 0.2 7.6 1.6 6.2 75.4 0.0 85 Unquoted 16.2 13.5 61.2 0.2 10.2 1.0 3.5 51.0 0.0 260 2004 Quoted 13.1 10.3 44.3 0.3 9.1 1.5 5.1 32.7 0.0 60 Unquoted 18.3 15.2 59.6 0.9 11.2 1.0 3.1 59.8 0.0 367 2005 Quoted 15.1 12.7 44.2 2.0 7.9 1.3 4.8 35.8 0.0 90 Unquoted 17.9 15.7 48.7 0.8 10.9 0.8 2.9 54.9 0.0 505 2006 Quoted 15.1 13.8 47.4 2.6 8.0 1.7 6.4 91.8 0.0 97 Unquoted 17.8 15.2 53.4 0.5 10.8 0.9 3.1 69.7 0.0 533 2007 Quoted 15.0 13.1 43.4 3.9 8.0 1.7 6.3 81.1 0.0 88 Unquoted 16.4 13.8 49.6 2.0 9.0 1.3 5.2 43.6 0.0 92 Table 4-2: Descriptive statistics EBIT multiple
Mean Median Maximum Minimum Std. Deviation Skewness Kurtosis Jarque - Bera Probability Observations Quoted 2.0 0.9 53.4 0.0 5.5 14.0 267.5 2,549,663.0 0.0 864 Unquoted 2.7 1.4 44.9 0.1 4.3 4.6 30.4 42,503.5 0.0 1,217 Quoted 2.1 0.8 53.4 0.0 5.1 7.2 64.8 32,739.7 0.0 193 Unquoted 2.5 1.4 14.0 0.1 3.5 2.3 7.7 36.8 0.0 20 Quoted 2.2 0.9 27.1 0.0 3.9 3.9 20.1 1,652.7 0.0 112 Unquoted 2.1 1.4 7.8 0.3 2.4 1.6 4.0 5.0 0.1 11 Quoted 1.4 0.7 21.5 0.7 2.3 6.0 49.5 12,028.0 0.0 125 Unquoted 2.3 1.2 28.9 0.1 4.1 4.5 25.5 2,062.6 0.0 84
Quoted n.a n.a n.a n.a n.a n.a n.a n.a n.a n.a
Unquoted 2.5 1.3 35.7 0.2 4.3 4.6 28.0 4,969.2 0.0 168 Quoted 1.0 0.7 6.3 0.0 1.1 2.3 9.5 242.3 0.0 90 Unquoted 2.3 1.5 17.6 0.1 2.9 2.8 11.6 1,008.9 0.0 228 Quoted 1.8 1.0 23.6 0.1 2.9 5.6 38.6 6,603.6 0.0 114 Unquoted 2.5 1.4 23.3 0.1 3.4 3.5 17.4 3,533.8 0.0 331 Quoted 1.7 1.1 8.5 0.1 1.7 2.1 7.4 183.0 0.0 119 Unquoted 3.4 1.5 44.9 0.1 5.9 4.1 22.3 6,125.8 0.0 334 Quoted 2.5 1.2 46.0 0.1 5.3 6.0 44.8 8,746.2 0.0 111 Unquoted 2.3 1.1 18.4 0.3 3.5 3.6 15.8 369.1 0.0 41
Table 4-4: Descriptive statistics Sales multiple
The datasets are tested for normality according to the Jarque-Bera test which is explained as:
4 32 2 K S S N Bera JarqueThe Jarque-Bera test is based on the kurtosis and skewness of the sample. The null hypothesis of the Jarque-Bera test assumes normality. As can be seen in the tables 4-2 to 4-4 the null hypothesis can be rejected in almost all cases on a 5% significance level. All datasets are not normally divided.
4.3 Research with real enterprise values
In this section an analysis of multiples calculated by use of real enterprise values is carried out. If an enterprise value is unknown, because the transaction price is never published, Zephyr calculates an ‘estimated enterprise value’. Obviously, these estimated enterprise values do never exactly match real values. The number of deals in which the transaction price is known, is 10% of the number of deals in which the transaction price is calculated.
4.3.1 Descriptive statistics and first analysis
criteria and with a known enterprise value, sales are known. 146 deals contain a quoted target and 924 deals contain an unquoted target. The table below presents the descriptive statistics for the pooled datasets.
Mean Median Maximum Minimum Std. Deviation Skewness Kurtosis Jarque - Bera Probability Observations EBIT Quoted 18.8 23.5 55.3 0.5 23.0 3.6 19.4 1,435.6 0.0 107 Unquoted 25.0 15.6 59.9 0.2 27.6 2.7 10.8 2,286.3 0.0 681 EBITDA Quoted 15.0 10.7 91.7 0.3 15.2 2.7 11.4 466.8 0.0 112 Unquoted 33.9 10.9 46.4 0.2 14.3 2.6 11.0 2,393.3 0.0 650 Sales Quoted 3.1 1.4 43.8 0.0 5.5 4.1 24.8 3,249.9 0.0 146 Unquoted 5.6 1.2 44.1 0.1 20.0 7.3 64.0 149,352.9 0.0 924
Table 4-5: Descriptive statistics of deals with a known enterprise value
Table 4-5 shows that the dataset with real enterprise values also is not normally distributed. To test if there is a significant difference between the multiples of quoted companies and those of unquoted companies we use the Kruskal-Wallis test. The Kruskal-Wallis is used for non normal distributed data. Additionally we took 10 samples of 10% of the quoted companies and 10% of the unquoted companies. To test if there is a significant difference between the multiples of quoted and unquoted companies in these samples the independent Sample T-test is used. Both test show insignificant differences on a 95% confidence level between the multiples of quoted and unquoted companies (see table 1 in the appendix).
To make the dataset suitable to execute the analysis of the private company discount with all variables as described in table 4-1, much more deals have to be eliminated. Since all variables of each deal must be known to let the deal be part of the dataset, very limited deals remain. In splitting up the database by year, some years do not include more then 2 quoted target companies. To be able to execute a representative analysis, more quoted target companies are required.
4.3.2 Conclusion
In testing on significant differences using the Kruskal-Wallis test and the independent sample T-test, none of the differences are statistically significant. Therefore it cannot be concluded that the private company discount does or does not exist in Eastern and Western Europe from 2000 through 2007.
4.4 Research with estimated enterprise values
Since the databases provide more deals with an estimated enterprise value then with a real enterprise value, there is no limitation of too less deals with sufficient information if we use estimated values. Estimated enterprise values are calculated by Bureau Van Dijk and defined as5: “Equity value
(made up to 100%) + Total debt + Preferred Stock –Cash and cash equivalents.”Equity value is
defined as: “If in the source data about the deal, the value is stated as the equity value then this value
will be the same as the deal value. If in the source data we are not told whether the deal value is the equity or enterprise value, then this value will be the value for a given number of shares, where we know the number of shares and the (estimated) price per share.”The deal value defined as: “The consideration paid for actual stake acquired.”
Values based on financials more then one year old are excluded from the analysis. In this section the analysis will be carried out with estimated enterprise values.
4.4.1 Description of the independent variables
The tables 2 to 4 presented in the appendix provide information on the mean and median of the independent variables used in the annual estimation regressions of the EBIT, EBITDA and Sales multiple. The tables show that in the earlier years there were more quoted target companies known whereas this turns around in 2002 and further. Companies with negative equity book values are eliminated from the collected data. Furthermore, deals in which one of the required variables is missing are also eliminated. Additionally, according to Rijken et al [1999], since we use these data for the ordinary least square regressions, the upper and lower 5% of all deals are eliminated ranked by the variable. In paragraph 4.4.3 the statistics will be analyzed and described more extensively.
4.4.2 Methodology
To find if there exists a significant difference between a multiple of a quoted and an unquoted company the methodology of Bhojraj and Lee [2002] will be used. Bhojraj and Lee [2002] develop a regression model which estimates a “warranted multiple”of a company. Subsequently they identify comparable companies, as those having the closest warranted multiple. In identifying the independent variables influencing the “warranted multiple”, Bhojraj and Lee [2002] are guided by valuation theory. Comparable companies selected by use of the “warranted multiple”offer sharp improvements over comparable companies selected on the basis of different techniques as for example industry membership in selecting comparable companies (Liu et al. [2002], Kim and Ritter [1999]).
5Source: https://zephyr-bvdep-com.proxy-ub.rug.nl/version-20081224/cgi/template.dll (Profile; Product User
Whereas Bhojraj and Lee [2002] estimate the regression model for the price to book multiple and the enterprise value to sales multiple, this study will exclude the book multiple and focus on the enterprise value to EBIT, enterprise value to EBITDA multiple and the enterprise value to sales multiple. According to Koller et al. [2005], in theory, enterprise value to EBITA (Earnings Before Interest Taxes and Amortization) is driven by tax, Return On Invested Capital (ROIC), growth and the Weighted Average Cost of Capital (WACC). In practice Koller et al. [2005] find that differences in corporate tax rates and differences in cost of capital are less important than differences in growth and profitability prospects. Since the dataset does not give information about profitability prospects, we use the following corresponding theoretical formula:
g WACC ROIC g T EBITA EV 1 1 Where: EV = Enterprise ValueEBITA = Earnings Before Interest Taxes and Amortization T = Corporate Tax Rate
g = Growth
ROIC = Return On Invested Capital
WACC = Weighted Average Cost of Capital
The reason why EBIT and EBITDA levels are used instead of EBITA is that EBIT and EBITDA levels are available in the databases in contrast to EBITA levels. To explain the enterprise value to EBIT multiple and enterprise value to EBITDA multiple, the same variables will be used except for the WACC, since it is not possible to calculate the WACC for all companies in the dataset. To calculate the WACC we need to know the target level of debt to equity. According to Koller et al. [2005] book value of debt reasonably approximates the current market value in most cases, although market value of equity must be determined using a multiples approach or through frequent DCF calculations. Since this study involves a high amount of data which makes it impossible to calculate each company’s WACC individually, the book leverage will be used as an alternative independent variable. As an alternative variable book leverage captures elements possibly not captured by other variables (Gebhardt et al. [2001]).
The enterprise value is explained by the multiple as:
it it t it m X
Where:
EVit= The Enterprise Value of company i at time t
Xit= The value driver (EBIT, EBITDA or sales) of company i at time t
mt= The multiple on the value driver estimated by the regression model of a company at time t εit= Pricing error
To estimate mt, the following regression models, reflecting the large sample relation between the
enterprise value to EBIT or EBITDA multiples and the independent variables, will be used:
it it it it it lev T ROIC g EBIT EV 4 3 2 1 0 and it it it it it lev T ROIC g EBITDA EV 4 3 2 1 0 Where: μ0 = Intercept
git = Average growth of the value driver over the last 5 years of company i at time t (for the
deals in 2000 last 4 years, because of availability)
levit = Last known book leverage before the deal was completed (at maximum one year old,
otherwise the deal is eliminated from the dataset), calculated as the long term debt divided by book value of equity of company i at time t. Although Gebhardt et al. [2001] find that book leverage is not significant in explaining cost of capital, this variable is included for completeness, in case it captures elements of cross sectional risk not captured by other variables
Tit = Corporate Tax rate of company i at time t
ROICit = Return On Invested Capital of company i at time t which is calculated as
ital WorkingCap s FixedAsset xEBIT T 1 εit = Error termunquoted on the multiples, a dummy variable (dumit) will be introduced as an extra independent
variable. The regression will be adjusted to:
it it it
it it
it lev T ROIC dum
g EBIT EV 5 4 3 2 1 0 and it it it it it
it lev T ROIC dum
g EBITDA EV 0 1 2 3 4 5
Where the additional term dumitis the dummy variable of company i at time t which is 1 if the target
company is quoted and 0 if the target company is unquoted.
In addition to the regression models mentioned above, the sales multiple will be investigated. The following regression model will be analyzed to see if the sales multiple can be estimated by the same variables as the variables analyzed in the EBIT and EBITDA multiple regressions.
it it it
it it
it lev T ROIC dum
g Sales EV 0 1 2 3 4 5
Sales are unaffected by any costs in a company. Sales data therefore are not directly related to (operating) profits. Since enterprise values in general increase if the (operating) profits increase, it is expected that the regression analysis of the sales multiple show least significant results. Although, if a takeover is realized because of synergy effects (e.g. most operating costs are fixed) a sales multiple is meaningful. Therefore, the sales multiple also is analyzed in this study.
4.4.3 Results
The tables presented in this subsection show the results from the annual estimation regression estimated by the corresponding formula. Each analysis is described hereafter.
EBIT Multiple
Table 4-6 presents the results for the estimated enterprise value to EBIT multiple. Corresponding regression model is defined as:
it it it
it it
it lev T ROIC dum
Year Intercept g lev tax roic dum R-sq Pooled Coefficient 14.05 0.22 1.59 10.37 -2.63 -4.85 0.09 P-value 0.00 0.13 0.00 0.00 0.00 0.00 2000 Coefficient 11.50 -0.14 1.90 15.44 -0.43 -4.72 0.09 P-value 0.00 0.82 0.03 0.12 0.85 0.01 2001 Coefficient 21.56 1.18 2.25 -20.20 -2.86 -3.59 0.11 P-value 0.00 0.23 0.03 0.32 0.28 0.14 2002 Coefficient 18.24 -0.26 1.72 -0.54 -3.23 -8.31 0.16 P-value 0.00 0.58 0.01 0.96 0.05 0.00 2003 Coefficient 12.09 0.05 1.91 11.35 -2.85 -4.81 0.10 P-value 0.00 0.92 0.00 0.22 0.04 0.00 2004 Coefficient 10.33 0.94 1.40 26.00 -4.52 -7.16 0.11 P-value 0.00 0.03 0.01 0.01 0.00 0.00 2005 Coefficient 14.77 0.15 1.60 9.75 -3.80 -3.30 0.08 P-value 0.00 0.60 0.00 0.18 0.00 0.01 2006 Coefficient 12.98 0.25 1.37 14.01 -1.44 -3.12 0.06 P-value 0.00 0.46 0.00 0.09 0.04 0.01 2007 Coefficient 5.88 0.19 1.12 37.11 -3.55 -2.40 0.08 P-value 0.27 0.68 0.06 0.03 0.02 0.08
Table 4-6 Annual estimation regressions for EV/EBIT multiple
Results show that the variance in the multiple is for a small part explained by this regression model. An R-squared of 8,8% for the pooled dataset, means that the multiple cannot be estimated accurately by use of this regression model if a constant intercept is assumed. Although, even if a target specific intercept is assumed, results do barely improve. This is not very remarkable since only 5% of the target companies out of the dataset are sold more than once between 2000 and 2007. Table 5 in the appendix shows the R-squared values of the regression analysis with random effects. All other values do not change.
The dummy variable shows the opposite sign from expectations, although we already observed and mentioned this in the summary statistics of section 4.2. According to expectations, growth is positively related to the explained multiple. Book leverage, however, which was expected to be negatively correlated with the multiple, is positively correlated with the multiple. The coefficients of the tax rate show very different values with high p-values. The tax rate seems not to be a correct variable to estimate the multiples of this dataset. Also ROIC shows values we did not expect.
EBITDA multiple
The next table, table 4-7, presents the regression results for the estimated enterprise value to EBITDA multiple. Its variables are estimated by the following regression model:
it it it
it it
it lev T ROIC dum
Year Intercept g lev tax roic dum R-sq Pooled Coefficient 12.29 0.03 0.50 -4.63 4.08 -3.97 0.05 P-value 0.00 0.54 0.00 0.00 0.29 0.00 2000 Coefficient 15.72 -0.08 -0.89 -4.48 -2.98 -3.48 0.03 P-value 0.00 0.94 0.29 0.64 0.23 0.08 2001 Coefficient 11.02 0.61 0.37 -4.71 8.54 -3.38 0.05 P-value 0.18 0.60 0.81 0.13 0.72 0.27 2002 Coefficient 11.56 -0.50 1.28 -2.18 7.16 -4.58 0.02 P-value 0.16 0.27 0.37 0.50 0.78 0.06 2003 Coefficient 4.44 -0.38 0.84 -1.25 2.28 -4.15 0.10 P-value 0.09 0.03 0.16 0.34 0.00 0.00 2004 Coefficient 11.83 0.23 0.36 -1.35 7.25 -5.52 0.05 P-value 0.00 0.42 0.34 0.09 0.47 0.00 2005 Coefficient 14.89 0.05 0.12 -1.83 -2.29 -4.54 0.05 P-value 0.00 0.57 0.65 0.02 0.75 0.00 2006 Coefficient 16.64 0.12 1.22 -3.34 -9.90 -1.73 0.03 P-value 0.03 0.21 0.14 0.18 0.69 0.34 2007 Coefficient 5.72 0.05 0.37 -1.30 2.83 -2.32 0.03 P-value 0.30 0.85 0.33 0.34 0.12 0.09
Table 4-7 Annual estimation regressions for EV/EBITDA multiple
Since table 4-7 shows similar, insignificant results as the analysis of the EBIT multiple, the regression model shown above is not even an accurate estimator of the EBITDA multiple. The analysis shows low R-squared values.
Sales Multiple
The last estimated regression model is the estimation of the sales multiple. Table 4-8 shows the results from the ordinary least square regression of the following formula:
it it it
it it
it lev T ROIC dum
g Sales EV 5 4 3 2 1 0
Year Intercept g lev tax roic dum R-sq
Pooled Coefficient -5.37 -0.15 -2.44 3.42 -5.44 2.99 0.00 P-value 0.93 0.97 0.71 0.87 0.82 0.26 2000 Coefficient 3.30 0.07 0.14 -2.01 -1.39 -0.46 0.01 P-value 0.14 0.67 0.70 0.69 0.27 0.70 2001 Coefficient -2.70 -0.04 -0.51 1.51 -0.62 0.30 0.07 P-value 0.22 0.83 0.35 0.01 0.47 0.80 2002 Coefficient 1.13 0.04 0.16 3.19 -0.03 -1.00 0.03 P-value 0.38 0.63 0.37 0.39 0.92 0.02 2003 Coefficient 0.00 0.75 0.04 7.51 -1.66 0.00 0.82 P-value 0.99 0.12 0.88 0.11 0.23 0.00 2004 Coefficient 1.73 0.03 0.16 1.57 0.09 -1.39 0.06 P-value 0.02 0.67 0.18 0.46 0.65 0.00 2005 Coefficient 2.30 0.11 0.30 -0.36 -0.02 -0.63 0.03 P-value 0.00 0.51 0.00 0.87 0.93 0.07 2006 Coefficient 2.61 0.12 0.12 2.36 -1.12 -1.63 0.07 P-value 0.07 0.00 0.08 0.58 0.04 0.00 2007 Coefficient -8.92 -1.84 -1.05 3.14 -4.16 2.71 0.01 P-value 0.50 0.82 0.53 0.45 0.80 0.50
Most remarkable finding in the analysis of the sales multiple is the relatively high R-squared value in the estimated regression from 2003. This analysis is only carried out with unquoted companies, since in 2003 there is no deal completed involving a quoted target company which suffices all requirements. It seems if the independent variables of quoted companies are different from those of unquoted companies. If the independent variables of both groups are combined, the multiple is not accurately explained by these variables. If they are not, at least for unquoted companies the multiple is explained by the independent variables for 82%.
4.4.4 Conclusions
The results of the private company discount analyses based on estimated enterprise values, do not show the expected private company discount. The proposed regression models are unable to accurately estimate the multiples using the analyzed variables. The only estimated regression explaining a relatively large part of the variation in the multiple, is the analysis of the sales multiple in 2003. Since this analysis did not include any quoted target companies, it seems that the independent variables of quoted target companies are different from those of unquoted target companies. Although, even in this specific analysis, coefficients do still not all have the expected signs. It can be concluded that the variables used, cannot explain the multiples of this dataset on a significant level.
4.5 Eastern and Western Europe separated
To find a possible explanation for the unexpected results, a short analysis of the median multiples in Eastern Europe compared with the median multiples in Western Europe is carried out. Results are shown in table 4-9. The analysis shows that higher multiples for unquoted companies, are mostly generated by Eastern European countries. This possibly disturbs results. For example: if an unquoted Hungarian target is relatively worth more than an unquoted Dutch target, the average multiple in the dataset increases compared to the average Dutch multiple. Since the dataset contains a relatively small amount of quoted targets, this difference has its influence on the results.
Median
Quoted Unquoted Difference Quoted Unquoted Difference Quoted Unquoted Difference
Western Europe 15.47 14.76 0.71 3.17 3.43 -0.26 1.33 1.17 0.16 Eastern Europe 10.99 15.53 -4.54 6.12 7.21 -1.09 0.86 1.11 -0.25 Difference 4.48 -0.78 5.25 -2.95 -3.78 0.83 0.47 0.06 0.41
Significance 0.00 0.56 0.00 0.00 0.13 0.56
EBITDA multiple Sales Multiple EBIT multiple
Table 4-9: Split up of Eastern and Western Europe
Since the dataset is not normally divided, significance is tested using the Kruskal-Wallis test. Figures in bold show significant differences (on a 5% significance level) between the multiples in Eastern and Western Europe.
Additionally a regression analysis is carried out on these two variables. The following expressions are used: it it W E it u q dum dum EBITDA EV / 2 / 1 0 it it W E it u q dum dum EBITDA EV / 2 / 1 0 it it W E it u q dum dum Sales EV / 2 / 1 0 Where: μ0 = Intercept it u q dum /
= dummy variable of company i at time t which is 1 if the target company is quoted and 0 if the target company is unquoted
it W E
dum /
= dummy variable of company i at time t which is 1 if the target company is Western Europe based and 0 if the target company is Eastern Europe based
εit = Error term
Results of this test are shown in table 4-10 below:
Intercept dumq/u dumE/W
R-sq EBIT Coefficient 14.31 0.78 1.16 0.00 P-Value 0.00 0.05 0.00 EBITDA Coefficient 14.45 1.98 -2.87 0.00 P-Value 0.00 0.05 0.00 Sales Coefficient 2.48 0.09 0.14 0.00 P-Value 0.00 0.53 0.35
Table 4-10: Regression analysis split up Eastern and Western Europe
EBITDA multiples significant differences in the multiple of a company being Eastern Europe or Western Europe based.
4.6 Cross checks on the dataset
5. Reflection 5.1 Conclusions
In investigating the private company discount in Eastern and Western Europe from 2000 through 2007, by use of the acquisition approach, we did not find this discount on a significant level. We tested if the EBIT, EBITDA and Sales multiples could be estimated by five variables including a variable representing if the target company was quoted or unquoted. Results showed that the multiples cannot be estimated accurately using Eastern and Western European deal data from 2000 through 2007 using the variables we analyzed.
The used dataset included deals in which the target company was based in Eastern or Western Europe at the time the deal was completed. This broad geographical area is chosen to be able to collect enough data in order to investigate the difference of the private company discount over time. A plausible explanation of the fact that the private company discount is not found using these data, is that the markets for mergers and acquisition in the different countries included in the dataset have different characteristics. After a comparison of multiples of Eastern Europe based target companies with the multiples of Western Europe based target companies, results show that there is a difference between the private company discounts of these areas. In investigating the whole geographical area at once, these differences disturb results.
One of the limitations of this study is the limited amount of deals of which an enterprise value is known. It is not possible to carry out a detailed analysis of the private company discount using real enterprise values. Using estimated enterprise values is less reliable since these estimations are different from real values.
Since the private company discount is not found in the analysis, the question if there is a development in the private company discount from 2000 through 2007 can be answered negatively. Results show that a trend in the private company discount is not found. Results of this study also show that there is no general private company discount which is applicable in all investigated countries. In general we can conclude that the private company discount should not be applied and that it should not have been applied 8 years ago. Furthermore, the investigated data show that prices of quoted and unquoted companies varied over the last 8 years. Therefore, we are unable to answer the question How
much did prices of unquoted companies and quoted companies change over the last 8 years?
5.2 Suggestions for further study
the private company discount is different in Eastern European countries from Western European countries. Further research could be done by selecting countries or markets with comparable characteristics and execute the same study on deals completed in these markets. Most challenging then is to collect enough data. If enough data can be collected to only investigate Western Europe, the development of the private company discount possibly can be analyzed.
References
Bajaj M., Denis D.J., Ferris S.P., Sarin A., 2001. Firm Value and Marketability Discounts. Journal of
Corporation Law, Volume 27 Number 1, 2001, pp. 89-115
Banz, Rolf, 1981. The Relationship Between Return and Market Value of Common Stocks. Journal
of Financial Economics
BDO Stoy Hayward, 2007. Private Company Price Index Q3, 2007
Bhojraj, Sanjeev and Lee, Charles M. C., 2002. Who Is My Peer? A Valuation-Based Approach to the Selection of Comparable Firms. Journal of Accounting Research, May2002, Vol. 40 Issue 2, p407-439 Bowman, Robert G.and Bush, Susan R. Using Comparable Companies to Estimate the Betas of Private Companies. Journal of Applied Finance, Fall/Winter2006, Vol. 16 Issue 2, p71-81, 11p Chan, K.C., N.F. Chen and D.A. Hsieh, 1985. “An exploratory investigation of the firm size effect”,
Journal of Financial economics, vol. 14, p. 451 –471
Cheung, Joe, 1999. A Probability Based Approach to Estimating Costs of Capital for Small Business.
Small Business Economics, Jun99, Vol. 12 Issue 4, p331
Eduard, Duncan, 1999. De rendementseis op aandelen van ondernemingen in het midden- en kleinbedrijf. BDO CampsObers Corporate Finance B.V. (Unpublished)
Finnerty, J.D., Emery, D.R., 2000. The Value of Corporate Control and the Comparable Method of Valuation. Financial Management, Spring2004, Vol. 33 Issue 1, p91-99
Gebhardt, W.R., Lee, C.M.C., Swaminathan, B, 2001. Toward an Implied Cost of Capital. Journal of
Accounting Research, Jun2001, Vol. 39 Issue 1, p135-176, 42p
Heijden, W.T. van der, 1999. Is Big Beautiful? De invloed van small firm effecten op de internationale waarde van kleine ondernemingen. (Unpublished)
Hertzel M., Smith R.L., 1993. Market Discounts and Shareholder Gains for Placing Equity Privately.
Journal of Finance, Jun93, Vol. 48 Issue 2, p459-485
Holterman, Wim, 2002. Transparantie en de marktwaarde van de aandelen van een onderneming.
Maandblad voor Accountancy en Bedrijfseconomie, Bladnummer 12, 2002
Ibbotson, Roger G., Kaplan, Paul D., Peterson, James D, 1997. Estimates of Small-Stock Betas Are Much Too Low. Journal of Portfolio Management, Vol. 23 Issue 4, p104-111
Isberg, S.C. and C.F. Thies, 1992. Small firm mutual funds; additional evidence on the small firm effect, Small business economics, Vol.4, p.211-219
Jegadeesh, Narasimhan, 1992. Does Market Risk Really Explain the Size Effect? Journal of Financial
& Quantitative Analysis, Sep92, Vol. 27 Issue 3, p337
Kaplan S.N., and Ruback R.S., 1996. The market pricing of cash flow forecasts.: Discounted cash flow vs. the method of “comparables”. Journal of Applied Corporate Finance, vol. 9, pp. 45 –60 Kim, Moonchul, Ritter, Jay R., 1999. Valuing IPOs. Journal of Financial Economics, Sep99, Vol. 53 Issue 3, p409-437
Koeplin J, Sarin A, Shapiro, A.C., 2000. The Private Company Discount. Journal of Applied
Corporate Finance, vol. 12, No. 4
Koller, T., Goedhart, M., Wessels, D., 2005. Valuation: Measuring and Managing the Value of
Companies. Fourth Edition. Wiley
Kooli, Maher, Kortas, Mohamed, L'Her, Jean-Francois, 2003. A New Examination of the Private Company Discount: The Acquisition Approach. Journal of Private Equity, Summer2003, Vol. 6 Issue 3, p48
KPMG, 2008. KPMG’s Corporate Tax Rate Survey, 2007.
Liu J., Nissim D., Thomas J. 2002. Equity valuation using multiples. Journal of Accounting Research, vol. 40, pp. 135 –172
Pagano, Marco, Panetta, Fabjo, Zingales, Luigi, 1998. Why do companies go public? An empirical analysis. Journal of Finance, Feb98, Vol. 53 Issue 1, p27, 38p
Rijken, Herbert A., Booij, Menno C., Buckley, Adrian, 1999. Valuation differences between quoted and unquoted companies - empirical evidence from the UK. European Journal of Finance, Sep99, Vol. 5 Issue 3, p256-275
Roll, Richard, 1981. A Possible Explanation of the Small Firm Effect. Journal of Finance, Sep81, Vol. 36 Issue 4, p879-888
Steenbeek O.W., Vliet M.A., 2005. Waarde en prijs van Nederlandse beursgenoteerde ondernemingen (unpublished)
Tirole, Jean, 2006. The Theory of Corporate Finance. Princeton and Oxford: Princeton University Press
https://zephyr-bvdep-com.proxy-ub.rug.nl/version-20081224/cgi/template.dll (30-12-2008)
Newspapers
