1
Privatizing European Firms
“Does privatizing improve operating performance of firms throughout the European continent between 1990 and 2008”
Paul Sonneveld
Business studies Intensive Program – Amsterdam Business School 6143792
2 Inhoudsopgave
1. Introduction ... 3
2. Literature Research ... 5
3. Problem Statement and hypothesis ... 10
4. Data sample ... 12
5. Methodology ... 15
6. Results ... 20
6.1 Univariate tests ... 21
6.2 Subsample Year of privatization ... 25
6.3 Subsample Country of origin ... 29
6.4 Multivariatieanalyse ... 32
7. Discussion & Conclussion ... 35
References ... 38
3 1. Introduction
The first European government to denationalize its industry, was the English government Thatcher in the seventies. Currently it is a resource for governments all over the world to open up markets and sectors. In the period from 1977 to 1999, 2459 deals in 121 countries were reported (Bortolotti et al., 2003). Currently governments around the world have privatized more than $ 1.25 trillion. A quite substantial amount if we compare this to the market capitalization of the Financial Times Global 500 Index, which contains the 500 largest companies in the world, of $ 6.6 trillion. (Megginson, 2005)
According to Bortolotti et al. (2003) governments privatize mainly when there is a lot of public debt and when there is a right-wing government in office. Kaya et al. (2012) state that also the type of industry in a country determines the government’s decision to privatize. But what motivates governments to denationalize their public companies? According to Arocena et al. (2012) governments have different reasons, but the main reason is to improve the operating performance of public companies1. Research has shown that there are benefits for companies to change from a national company to a privately held company ((D’Souza & Megginson, 1999 en Megginson en Netter, 2001). But if we have to believe the newspapers, products delivered by former public firms such as water, electricity, gas and public transport are getting more expensive by the day now that they are privately held. For example the price of water in Portugal has risen 400% after privatization2, and research done in England by ‘campain for better transport’ has shown that train fairs are 3.5 to 9.7 times more expensive on their (English) privately held companies than on the public companies in other European countries.
These statements contradict most of the research done on the topic, research done in the eighties show that public companies had significantly higher costs and had worse
1 http://libertarian.nl/vrijbrief-archief/2004/06/wat-is-privatiseren-en-waarom/ 2
4
operating performance in comparison with privately held firms (Mueller 1989, Vining and Boardman, 1992). D’Souza & Megginson (1999) argue that the benefits of privatization are the same for companies that were privatized in the eighties and those that were privatized in the nineties. Their research done on companies that were privatized between 1990 and 1996 also show significantly performance gains after privatization, although a small decrease in performance gains was detected.
Times have changed after the eighties and the nineties as well as the demands of governments on the performance on public companies. This study has researched if 81 European companies that are privatized between 1990 and 2008 still have performance gains after privatization and to what extent these gains have changed compared to the results of Megginson et al. (1994) and D’Souza et al. (1999).
Although the research of D’Souza & Megginson (1999) still shows a significant improvement of performance with companies that are privatized in the nineties, this research hasn’t found any evidence of improved profitability and efficiency. On other variables studied like debt ratio, payout of dividend, capital investments and employment the same results have been found. Although research has shown efficiency gains after privatization there is a growing debate about those gains. Like this research Estache & Rossi (2002) also found that privatized firms didn’t have a significantly higher level of efficiency compared to public companies. According to their study public firms also have improved their efficiency in order to survive in the market, Estache & Rossi (2002) call this the “catching-up effect”. Because of this effect it could be that the proposed efficiency gains of privatization are not that great or are not there at all. When comparing the results of this research to that of Megginson et al. (1994), D’Souza et al. (1999) and later research the conclusion can be drawn that privatization isn’t a good method any more if the incentive for governments is to improve profitability and efficiency.
5
The rest of this paper will cover the following topics: Chapter 2 will address relevant research about privatization. The problem statement, hypothesis and data sample will be discussed in chapter 3 and 4. The methodology and research predictions are covered in Chapter 5. In Chapter 6 the results of the sample and the sub-samples are presented. Chapter 7 holds the discussion and conclusion.
2. Literature Research
The incentive in the nineties for governments to privatize their state owned enterprises (SOE) was to make them more efficient. The specific objectives of governments to privatize are studied by Price Waterhouse (1989). According to that study all governments aim at the same objectives which are: reduce government interference in the economy, raise money for the state, promote increased efficiency, the opportunity to introduce competitors in to the market, promote wider share ownership and expose SOE to market discipline. According to Boycko (1996) SOE have to cope with contradictory objectives such as “providing employment and
address political objectives which negatively affect their business”. Subsidies that SOE
receive contribute to the fact that managers of SOE lack incentives to improve their efficiency, because continuity of their business is guaranteed. The research of Dewenter and Malatesta (2001) shows that the efficiency of SOE is significantly less in comparison to privately owned enterprises (POE). Although research (Sun and Tong, 2003) shows that privatized firms improve their efficiency, Boycko et al. (1996) only predict an improvement in efficiency if a SOE will be privatized, and only under the condition that the full ownership of a SOE goes to private equity. Paudyal et al. (1998) also highlight that it’s important that a large enough piece of the public firm is sold, otherwise there is a change that the government is interfering with the business which has a negative influence on the performance of the firm.
6
According to Boardman and Vining (1989) SOE always perform worse than private firms in the same sector. Megginson & Netter (2001) and Djankov & Murrell (2002) also conclude in their studies that there is improvement in profitability after privatization. But how can this increase in profitability be explained. Is this due to better efficiency of companies or are there other causes? Anderson et al. (1997) state that privatization will initially lead to higher price, but also give a wider variety of products to the consumer. Although many consumers will doubt that higher prices have benefits for them, Ugaz & Price (2003) state that the level of consumer wellbeing rises with privatization of firms. The research of Megginson & Netter (2001) and Djankov & Murrell (2002) state that not higher prices cause better performance of private firms, but better efficiency. The privatization of SOE leads to an open market where companies can compete with each other. Ramamurti (1997) and Newbery & Pollitt (1997) find that competition improves the performance of companies, because it drives a company to create better efficiency and profitability to get a market share. The main advantage of privatization is that there will be a more competitive environment between firms because of new entrees. This could eventually lead to lower prices for the consumer (Anderson et al; 1997). This effect has been acknowledged by D'Souza and Megginson (1999), they state that if the performance gains are related to better efficiency of a company, privatization is better for the consumer.
According to Anderson et al. (1997) SOE have a controlling factor in keeping retail prices at a certain level. A SOE is not driven by maximizing profit, but want to create a higher social level. This will lead to lower prices set by the public firm. If a POE wants to be competitive in this market it has to be efficient to be profitable and survive in that business. But does privatization always guarantees a better future for the companies involved? Not always, the study of Mikkelson, Partch, and Shah (1997) shows bad results after privatization. They found that median operating incomes fall with 81% comparing the year before
7
privatization to the five years after privatization. And their results also show that the nature of the new owner doesn’t affect the outcome of privatization. Recently Wang (2005) also found evidence that across all industries the operating performance of POE declines after privatization. Research of Ritter (1991; 1995) shows that privatized firms not always perform better, they even conclude an underperformance of privatized firms. This conclusion is supported by research of Grompers & Lerner (2003) and Eckbo & Norli (2005). Research done by Donghui et al. (2007) also found no evidence of better performance of POE that are privatized in the period between 1992-2000. An example can be given in the United States where the government deregulated the transportation system. According to Winston (1998) this caused higher cost and a decrease of performance. An explanation for the underperformance of privatized firms is given by Loughran & Ritter (1995), they state in their research that investors are too optimistic about the future of privatized firms. Governments take full advantage of this by timing their public offerings when the outlook of those firms is good.
Anderson et al. (1997) argue that privatization of SOE is the best option when the outlook and profitability of a company are good. A SOE is profitable when the consumer has only one desire or choice of product. The low price set by a public firm will generate demand in quantity which leads to profits (Anderson et al; 1997). In this sense the growth in variety of products caused by privatization is undesired. Only a profitable SOE with ridiculously high cost is worth well privatizing (Anderson et al; 1997). In the research of D’Souza et al. (2007) there is another view on the process before privatization. They state that restructuring before privatization leads to better and more efficient firms after the privatization. D’Souza et al. (2007) also conclude that the ownership before privatization has an influence on the performance after the privatization. D’Souza et al. (2007) state that ownership is: ‟the most
8
research of Sun & Tong (2002). Both find that if a firm is fully owned by the government before privatization the gains of the privatization process will be higher. The change in ownership brings another advantage to companies: ‟help to redefine the firm's objectives and
the manager's incentives‟ (D’Souza et al; 2007). However Mulherin & Boone (2000) find it
is better to restructure after privatization. According to their research firms that are privatized follow new strategies and adapt to the changing environment of the economy, this means that it’s more logical to restructure after the privatization. Barberis et al (1996) also acknowledge this relation between performance and ownership, however according to their research “a
fresh injection of human capital is also a critical factor to the success of corporate restructuring’’
But ownership is not the only important factor with privatization, the market environment in which a firm operates has a large effect on the success of privatization (Cato, 2011). The economic state of a country will also play an important role in the success of privatization. According to Djankov & Murell (2002) privatization has greater effects in developing countries. D’Souza et al. (2007) also find that the gains are greater for firms in developing countries. Mainly because the economic effects of a country play an important role in the results of a company.
Megginson et al. (1994) have studied the effect of privatization multiple times, his studies show that privatization “yields significant performance improvements”. Although his first study on privatization wasn’t the first on the subject of privatization, it was the first that had a large multinational and multi industry sample. The sample consisted out of 61 companies that were privatized before 1990, and according to their own opinion contains the “most economically important privatizations of recent years”. Megginson et al. (1994) state that the goal of governments privatization programs is to improve operating and financial performance. In order to achieve that goal, governments expect that privatization will lead to
9
an increase of a firms: ‘profitability’, ‘operating efficiency’, ‘capital investments’, ‘output’ and ‘dividend payout’ and a decrease of the ‘debt ratio’ and ‘total employment’. The theoretical model on privatization of Boycko, Schliefer, and Vishny (1983) support these predictions although they state that the output of privatized firm will decrease. Megginson et al. (1994) study if the prediction as proposed by the governments occurs. They used a matched sample for a period of seven years. To test the predictions they calculated a pre privatization mean of three years before privatization and a post privatization mean of three years after privatization of each variable, the year of privatization itself was excluded from the sample. A Wilcoxon signed-rank test was used in order to test whether the computed means differ significantly from each other. Megginson et al. (1994) found that companies that are privatized before 1990 have “strong performance improvements” after privatization. All prediction expected by the government where met, except the decrease of employment, Megginson et al. (1994) even found an increase in total employment.
D’Souza and Megginson (1999) wonder if the effect of privatization have remained the same in the nineties in comparison to the research of Megginson et al. (1994). In order to test if any changes have occurred they use the same methodology and testable predictions as used in the study of Megginson et al. (1994). The data sample consists of 85 companies that were privatized between 1990 and 1996. The results show that firms still have a significant increase in profitability, output, operating efficiency and dividend payments. However the results are less significant as in comparison to the study of Megginson et al. (1994). The total employment of privatized companies also increased significantly in the study of Megginson et al. (1994) where D’Souza (1999) found a significant decline in total employment.
More recently the counterfactual has been introduced into the research of privatization. The counterfactual studies the effect what would be the consequences if none of the SOE where privatized. Research in the electricity prices in Britain show that they would have been
10
lower if the electricity firms weren’t privatized (Branston, 2003). The counterfactual is also studied by Newberry & Pollit (1997) and they conclude that privatization only has led to greater profit margins for firms and not necessarily to lower prices for the consumer. In order to study the counterfactual both use the same regression in which revenue in year t is explained by ‘after tax profits’, ‘costs’ and ‘transfers to the government’ in the same year. Although it is important to realize of what would have happend if SOE didn’t privatize, this research will not provide a counterfactual. According to Shirley et al. (2000) many counterfactuals are based on optimal firm behavior and these frameworks highlight the shortcomings of privatization. Many governments already demand more efficiency of their own offices and companies, so the advantage of privatizing is lower, in many cases this is not fully addressed in the counterfactual. (Shirley et al. 2000). Although Kikeri et al. (2004) recognize the importance of the counterfactual in research on privatization, they state most are ‟crystal ball gazing”.
3. Problem Statement and hypothesis
As described there still is a lot of debate about the effect of privatization on operating efficiency. Millward and Parker (1983) found no evidence in their research that privatization leads to better performance. However the study of Megginson et al. (1994) shows a significant performance boost caused by better efficiency. Later research by D’Souza & Megginson (1999) still show significant performance gains after privatization, although they are lower in comparison to the prior research of Megginson et al. (1994). The aim of this research is to study the effects of privatization on companies that are privatized between 1990 and 2008 in comparison to the two studies done by Megginson.
11
In order to study the effect of privatization and be able to compare it to the studies of Megginson et al. (1994) and D’Souza & Megginson (1999) the same hypotheses will be tested. This leads to the following hypothesis, privatization leads to:
1) An increase in the profitability of a company;
The theoretical framework of Boycko, Shleifer and Vishny (1993) on privatization outcomes states that privatization leads to an increase in profitability.
2) Higher operating efficiency of a company;
According to Price Waterhouse (1989) the higher operating efficiency is one of the reasons why governments want to privatize their SOE. Also Djankov & Murrel (2002) and Sun & Tong (2003) found that privatization leads to higher operating efficiency.
3) More capital investments by companies;
The theoretical model of Boycko (1993) shows that companies increase their capital investments after privatization. Also the results of Boubakri and Cosset (1998) show a significant increase in capital investments after privatization.
4) An increase in the output of a company;
There are different opinions about what happens with the output of company after privatization. The model of Boycko et al. (1993) describes a decrease in output after privatization, however Megginson et al. (1994) state that governments expect an increase in output after privatization. Because this study uses the same methodology as that of Megginson et al. (1994) we adopt the same hypothesis.
5) Lower employment;
According to Boycko (1996) SOE have an objective to create employment. If this statement is correct it would lead to lower employment after privatization because POE aim at maximum profit, not at providing employment.
12 6) A decrease in the leverage of companies;
The study of Bradley et al. (1984) show that after privatization companies have to cope with capital markets which lead to higher costs of debt and because of that, they will reduce their total amount of debt.
7) Higher dividend payments.
According to Megginson et al. (1994) there are no arguments to support this hypothesis. However in their research they state that it’s logical that dividend payments will increase because private equity holders normally demand dividend and governments don’t.
Although research has been performed on the topic, none use a data sample which contains companies which are privatized before and after the millennium. As described by D’Souza & Megginson (1999) the year of privatization could have an influence on the results of privatization. This effect will also be researched in a separate sub analysis by comparing the results of privatization before and after the millennium with the same hypothesis. According to D’Souza (2007) the country of origin has an influence on the outcome of privatization. Although the research of D’Souza (2007) is recent a separate sub analysis will be performed in order to test whether this effect occurs in our sample.
4. Data sample
The sample for this research primarily comes out of financial databases. To ensure a large enough sample, there has been made use of a number of databases. The first step was to select privatized companies. Zephyr is a database which contains specified data about change in ownership and for example privatization deals, in the period of 1997 to 2007 already more than 300.000 ‘change in ownership deals’ were announced in this database. This database has been used to select the companies for this sample. To ensure that the sample contains
13
companies which could measure the objective of this research there has been made use of search criteria’s such as: the change of ownership (is privatization), country of origin of the companies involved (the continent Europe) and the privatization has taken place after 1990. Any financials and government agencies were excluded from the search. This provides a list of 151 companies. To ensure that the correct financial information for each company was collected from other databases, there has been made use of ISIN/SEDOL numbers. This number is unique for each company and can be used to retrieve information about a company in databases. However not all the companies that were on the created list of Zephyr contained an ISIN or SEDOL number, those where deleted as well as double entries. After that procedure there remained 109 companies.
The data will be used to compute several ratios in the pre- and post privatization period, Megginson et al. (1994) use three years for both periods. Other research on the topic use periods of 2 up to 20 years after the year of privatization. This means that in general a three years post privatization period is long enough. The aim for this research is, if the databases contains enough information, to have a pre privatization period of 3 years and a post privatization period of 3 years.
Two databases have been used to search for the company financials. First Datastream has been used where the following financials where retrieved: sales, net income, total assets, total debt, return on equity, total employment, capital expenditures and cash dividends. Because Datastream didn’t contain information about all the selected companies the missing companies where looked up in Wharton WRDS with the same ‘company financials’. Although those two databases provide access to most of the financial database, not all companies where found or didn’t contain the complete series of -3 to +3 of privatization. To ensure a large enough sample the pre privatization period has been reduced to a one year
14
period. After removing those from the sample there remained 81 companies which are examined. In table 1 an overview of the sample can be seen.
Table 1: Sample overview
This table gives an overview how the sample is divided. The three left columns give an overview how many companies are privatized in each year. The right columns show how many companies of a country are represented in the sample.
Year of privatization Country of origin
Year Frequency Precent Country Frequency Precent
1990 2 2,5 Austria 5 6,2 1991 3 3,7 Swiss 1 1,2 1992 3 3,7 Denmark 3 3,7 1993 3 3,7 Germany 7 8,6 1994 9 11,1 Spain 1 1,2 1995 4 4,9 Estonia 2 2,5 1996 1 1,2 Finland 3 3,7 1997 2 2,5 France 11 13,6 1998 7 8,6 Great Brittain 7 8,6 1999 12 14,8 Greece 2 2,5 2000 7 8,6 Hungary 3 3,7 2001 7 8,6 Ireland 2 2,5 2002 2 2,5 Italia 12 14,8 2003 2 2,5 Latvia 1 1,2 2004 1 1,2 Netherlands 4 4,9 2005 6 7,4 Norway 2 2,5 2006 5 6,2 Poland 5 6,2 2007 3 3,7 Portugal 3 3,7 2008 2 2,5 Sweden 4 4,9 Turkey 3 3,7
15 5. Methodology
The aim for this research is to find out if privatization still has the great positive effects over the last two decades as mentioned by Megginson et al. (1994) and D’ Souza & Megginson (1999). To ensure that a comparison can be made between the results of those researches, the same variables that have been used in those studies will be used for this research. However our sample doesn’t contain data to test the hypothesis about the ‘output’ of a company. Although this is unfortunate this will not influence the overall results, the first three predictions contain the most valuable information for this research.
The hypothesis that remain contain different financial variables, those variables have to be computed before the hypothesis can be tested. The following variables will be used in order to test the hypothesis:
Profitability: Measures if a company makes a return on the invested capital, for this research
the three main profitability ratios will be used: return on sales (net income/total sales), return on assets (net income/total assets), return on equity (net income/total equity)
Operating Efficiency: Governments incentive to privatize is to make SOE more efficient, to
measure if this occurs, the ratios: sales efficiency (Sales/Total employment) and net income efficiency (Net income/Total employment) will be used.
Capital Expenditure: The capital expenditures of a company is the amount of money they
spend to invest in the future. The Capital Expenditure ratio gives an indication how much a company spend in relation to their sales (capex/sales) or assets (capex/total assets).
Employment: Boycko et al. (1996) state that SOE have to provide employment, it is likely
that the number of employees a companies has, will change after privatization. In order to measure this the total number of employees of a firm will be measured.
16
Leverage: The leverage of a company is measured by the ratio of debt (total debt/total assets),
this ratio gives an indication about the financial risk a company has. Because the risk of a SOE differs from that of a POE, there is a chance that the leverage of a company will change after privatization
Dividends: The ratio to measure the payout of dividends is: dividends/sales. Because a POE
has to deal with shareholders there is a probability that the dividend policy of companies changes after privatization.
In order to test the hypothesis, the change between the variables of pre and post privatization periods will be tested. Because the sample contains data of three years post privatization, it is necessary to compute means of each ratio before testing. The means are computed by adding up the data of the three years and divided by three. If any entry missed a year the data of that entry was divided by three minus the number of years that are missing. Because the sample only contained a one year pre privatization period these data could be used directly. The year of privatization has been left out during testing because this year contains both results of the SOE and the POE. The inflation rate of each country could give a bias in the date, To ensure a good comparison the complete sample has been corrected for inflation. Each entry has been corrected with its own country’s inflation rate with the year of privatization as a base. After computing the means and the inflation correction, there will be made use of a non-parametric statistical hypothesis (Wilcoxon-test). This test takes the mean values of the pre- and post privatization period and tests if the mean difference is zero with the following regression:
17 Prediction complete sample
Because the reason for governments to privatize their national industries is to improve the operating performance of companies and the study of D’Souza & Megginson (1999) shows a significant increase in profitably after privatization, the prediction is that ‘profitability’ of a company will increase after privatization. As the profitability of a company increases because of privatization, the expectation is that ‘operating efficiency’ as well as ‘capital investment’ will increase. This prediction is in line with the results of Boubakri and Cosset (1998). The general assumption is that privatization leads to better efficiency, because of this statement the total employment of a company will be lower because companies can provide their service with fewer staff. Hower if a significant decrease of employment is visible one can argue to what extent computers and automation have contributed to this decrease. The study of Megginson et al. (1998) show a significantly decrease of leverage, the prediction is that this sample will give the same conclusion. Boubakri and Cosset (1998) conclude in their results that the payout of dividends increased significantly. The prediction is that this research will give the same results.
Subsample: Year of Privatization
D’Souza & Megginson (1999) state in the conclusion of their research that the results of privatization in the nineties differ from those of their research done on companies that were privatized in the eighties. But although they mention it, they haven’t performed any statistical test to prove their statement. The subsample will test if: The year of privatization has an effect
on the results of privatization? In line with the research of D’Souza & Megginson (1999) and
Donghui et al. (2007) the prediction is that companies that were privatized late will experience less effects of privatization.
18
In order to test this prediction the sample will be divided into two groups, the companies that are privatized before 2000 (47 companies) and those that are privatized in 2000 or later (34 companies). After the sample is divided the same six prediction will be tested in each group with the Wilcoxon-test. Any significant change of the results between the two groups or between the complete sample is reported as an effect because of the difference in ‘year of privatization’.
Subsample: Country of origin.
The study of D’Souza (2007) shows that the effect of privatization in developed and transition countries are different. In order to test this effect the sample will be divided into three groups, although testing each individual country would normally be preferred, the sample doesn’t contain enough cases of each country to do so (see table 1). In order to distinguish the different countries they have been divided in three equal groups to region, Western Europe. South-Europe and North/East Europe. Testing if the companies country of origin has an influence on the effects of privatization a one way ANOVA will be performed. This test compares the means of the available ratios of two or more groups on the basis of population variance. With the ANOVA all the hypothesis of the full sample will be tested if they differ among the three selected regions.. Although the ANOVA can determine if privatization has a significantly different effect on the prediction between the three groups, it cannot determine between which groups the prediction differs. If the ANOVA shows any significant difference between the groups it is necessary to perform an extra test. This extra test is a Tukey’s HSD which uses data from the performed ANOVA, and calculates the distance between the groups. The outcome of that calculation is called the Honest Significant Difference and represents the distance that must occur if two groups significantly differ from each other.
19
Because a transition country already has a large effect on the efficiency and performance of companies the expectation is that privatization has more influence in developed countries. This means that the region ‘Western Europe’ is expected to show a significant change of efficiency after privatization.
Subsample: Multivariate analysis.
The univariate analysis provides a good inside whether the different variables are significant across the sample. But they don’t control whether a relationship between the variables can result in a different outcome. To ensure that the relationship between variables doesn’t change the outcome of the results the following regression model will be employed:
The regression will be used to test five different depended variables ( . At first the
operating performance variables: ‘return on sales’, ‘return on equity’ and ‘return on assets’ will be used as depended variables. The same regression will be used to measure the effect on the payout ratio and debt ratio.
The independent variable ‘year of privatization’ stands for the year in which a company has been privatized. The research of Okoli (2003) also uses this variable in their regression on privatization because the year of privatization can have an influence on the results. The variable ‘year of privatization’ has been divided into four groups, namely: (1) 1990 – 1994, (2) 1995 – 1999, (3) 2000 – 2004, (4) 2005 – 2008. In order to use this variable there are three dummy variables (for group 1 to 3) which is one if a company belongs to the group otherwise zero.
20
The variable ‘firm size’ has been adopted from the method out of the research of Dewenter et al. (2001). They use this variable in order to clarify whether differences occur between small and large companies, in order to rank the cases there have been chosen for sales. To incorporate sales in the regression the sales a company has in the year of privatization will be used. The private dummy categorizes the sample into two groups privatized or not privatized. The dummy variable contains zero for a public company and one for a private company. The dummy ‘number of years private’ takes the value one if a data entry is from two or more years after privatization, otherwise it is zero. In the sample there are four main industries indentified, Industrials, Power, Telecommunication & Transportation and Consumer Products. For each industry, except consumer products, a separate dummy has been made, it takes the value one if an entry belongs to the corresponding industry otherwise zero. The data consist of paired samples, a pair consists out of data pre privatization, and post privatization which are separate variables in the sample. In order to run the regression the original sample has been modified. Each year in the paired sample entry has been transformed to a separate entry. The private dummy separates the pre and post privatization data, the dummy ‘number of years private’ indentifies if data belongs to the beginning or the end of the post privatization period. This means that after the modification every company has four entries, one before privatization and three after privatization.
6. Results
In this chapter the results of the research are being presented, the eight predictions will be discussed separately and there will be given an explanation if a difference occurs between the prediction and actual results. The total sample consisted of 81 companies from 20 different
21
European countries. First the results of the six predictions of the full sample are discussed, later on the subsamples and the multivariate analysis.
6.1 Univariate tests
P1) Profitability
To test the prediction of profitability the ratios: return on equity, return on sales and return on assets are used. The results in table 2 show that the mean of ROE drops from 15,4% post privatization to 13.9% after privatization, the Wilcoxon test shows that there is no significant evidence of a better ROE after privatization. The ROA also shows no significant difference, there is even no noticeable change in the mean ROA post and pre privatization (6%). However the return on sales ratio gives information about how much return a company makes out of standard operations, that makes it the ratio which contains the most valuable information. But the ROS ratio also shows a non significant change and the mean scores of ROS are lower after privatization than before. However the mean sales rises post privatization significantly, this implies that sales increases post privatization but that it doesn’t lead to more net income. So although Ramamurti (1997) and Newbery & Pollitt (1997) claim an improvement in profitability in their research, no evidence is found in this sample. D’Souza et al. (1997) and Megginson et al. (1994) also found significant higher profitability after privatization in their studies. Because a recent study of Wang (2005) also found a decline in operating performance after privatization it is plausible that the effects of privatization are changing.
P2) Operating Efficiency
From the studied theories it can be assumed that operation efficiency will increase after privatization. Out of the sample more than 92% of the companies has improved their sales
22
efficiency significantly after privatization (table 2). The income efficiency also shows a significant change, the income efficiency of the sample drops with 3.76% after privatization. Although these results aren’t what was predicted, still a remarkable 62,8% of the companies in the sample change in the predicted direction. That the efficiency of companies would rise is supported by research of Megginson & Netter (2001) and Djankov & Murrell (2002) on the topic. However they conclude an overall increase in operating efficiency, because the companies in our sample didn’t improve their net income efficiency, the prediction that the operating efficiency of privatized firms will rise can’t fully be supported. However there are different theories about the foundation of this improved sales efficiency. Is the higher sales efficiency the result of reducing the amount of employment or do privatized companies improve their sales. An extra paired t-test has been performed on the variable ‘sales’, that test shows that the increase in sales is significant (table 2) . To conclude which factor provides the higher ‘sales efficiency’ the employment factor has to be implemented. This will be discussed later on in this paragraph.
P3) Capital Expenditure
The ratios which measured capital expenditures give an indication about the investments in the future by a company. Expected was that the ratios of capital expenditures are higher post privatization, because privatized companies have to invest more in the future to at least keep the same market share. The data in table 2 shows that there is a decrease in both capital expenditure ratios, the sales ratio drops with 12% and the asset ratio with 6% after privatization, however both ratios aren’t significant. Unlike what was expected there is no significant rise in capital expenditures after privatization.
23
Table 2: Results of univariate test for the full sample of privatized firms
This table presents the results of all univariate tests over the full sample. The table contains for each test the number of usable observations, the mean values of the entries before and after privatization, the mean change of the value after privatization versus before privatization. The Wilcoxon signed rank test has been employed to test for significance in the mean change, the Z-statistics of those tests are presented in column 5 of this table. The last column of this table represents the percentage of firms that change in the direction as predicted in paragraph 5. All data used to compute the ratios for the Wilcoxon test have been corrected for inflation with the privatization year as base.
Variables N
Mean before privatization
Mean after
privatization Mean change
Z statistics or significance of proportion of change Percentage of firms that changed as predicted Profitability
Return on Equity (ROE) 59 0,1546 0,1396 -0,02 -1,449 0,440
Return on Sales (ROS) 81 0,1063 0,0892 -0,02 -0,148 0,519
Return on Assets (ROA) 76 0,0644 0,0623 0,00 -0,016 0,579
Efficiency Sales efficiency 78 1143 1718 575 -6,841*** 0,923 Income efficiency 78 133 128 -5,00 -2,812*** 0,628 Sales 81 11259547 14296000 3036453 -3.717 Capital Expenditures Capex / Sales 80 0,18 0,15 -0,03 -0,384 0,500 Capex / Assets 75 0,095 0,089 -0,01 -644 0,533 Debt Ratio of debt 75 0,27 0,25 -0,02 -1,758** 0,586 Payout Payout of dividends 81 0,0337 0,0429 0,01 -4,375*** 0,703 Employment Total employment 78 30805 28201 -2604 -0,142 0,500
***, **, * Significant at the 1, 5, and 10 percent levels. ) t-test
24
Megginson et al. (1994) research on companies privatized in the eighties found a significant increase in both capital ratios. However the research done by D’Souza et al. (1997) on companies privatized in the nineties found no significant change in the capital expenditure ratios. According to them the reason is that total sales and assets have increased faster than the capital expenditures.
P4) Employment
The sample shows an 8 percent (table 2) decrease of employment, however this decrease isn’t significant. This is quite remarkable because the general assumption is that total employment drops after privatization. And although a decrease of employment is found the general assumption that total employment drops must be fed by the cognitive dissonance of the public, because this research found no significant evidence for it. The results are in line with the latest research of D’Souza et al. (1997) because their research also found a insignificant decrease of 3.6%. But although Megginson et al. (1994) found a significant increase of employment after privatization in the eighties. Because over time the effects on total employment after privatization changes in a negative way, it could be that future research will find a significant decrease in employment after privatization.
The employment level also leads back to the operating efficiency ratio, because the total employment of companies doesn’t drop significantly after privatization the rises in operating efficiency is fully caused by the significant rise in ‘sales’.
P5) Leverage
The leverage of a firm is measured by the ratio of debt companies has. Expected was that after privatization the companies would get a lower ratio of debt. The mean ratio of debt drops by 7 percent (table 2), and 58 percent of the companies has a lower ratio of debt after privatization. This change in debt ratio is significant at the 10% significance level, this means that the
25
expectation was met. Compared to the study of Megginson et al. (1994) and D’Souza et al. (1999) where 70% and 67% of the companies had a significantly lower debt ratio, our results show a lower faction of companies that realize a lower debt ratio. However the decline in mean debt ratio found in our sample is higher than was found in the mentioned researches (2.43% and 6%).
P6) Dividends:
The expectation was that ratio of dividend payout would be higher after privatization. The sample in table 2 shows a significant rise in the mean payout ratio by a remarkably 27 percent. Although still almost 30 percent in the sample has a payout ratio that is lower or the same after privatization. The higher payout ratio is remarkable, although the higher payout ratio was expected private companies often only payout a small amount to their shareholders. The rest will be used to make investment, however our results show that this statement is incorrect and that shareholders demand a large enough return on their investments. This results are completely in line compared to the results found in the research of Megginson et al. (1994) and D’Souza (1997).
6.2 Subsample Year of privatization
The sample consists of companies that are privatized between 1990 to 2008. It could be possible that the results of privatization are different for companies that are privatized in 1990 than those that are privatized in 2008. It is unknown if there are any differences in the privatization results between companies that are privatized early and late. The prediction is that early adapters achieve higher results after privatizing than the late group. The sample is divided into two groups to investigate the subject. The sample consisted of companies that are
26
privatized in the nineties (56% of the sample) and those that are privatized after 2000. The results for this subsample can be seen in table 3.
The return on sales ratio still shows no significance change when splitting the sample into two groups. However for the return on equity ratio there can be seen a difference in the sample between companies in the two groups. A significant decrease of 26% of the return on equity is found in the group ‘after 2000’. In the return on asset ratio there is a change noticeable between the two different groups. In the ‘before 2000 group’ there is a significant positive change of 37% where the mean of the ‘after 2000’ group drops insignificantly with 11,1%.The two ratios of operating efficiency are significant for the whole sample, however change can be detected when dividing the sample into two groups. Although sales efficiency still remains significant in the two groups the change in mean rank is far greater in the group that has been privatized ‘after 2000’ namely 240%. In the group ‘before 2000’ the change in mean rank was only 15%. But remarkable is that the change in income efficiency is significant for the full sample and the group privatized ‘before 2000’ but not for the ‘after 2000 group’ separately. This is remarkable because the same phenomenon isn’t visible at the sales efficiency ratio. As can be seen by the growth in sales it isn’t that enormous after privatization in the ‘after 2000’ group. This means that public companies already acquired a large market share before privatization in the 21ste century.
The total employment of a company has a relation to sales-, and income efficiency, although total employment decreased in the total sample it isn’t significant. In the separate groups there still is no significant change.
27 ***, **, * Significant at the 1, 5, and 10 percent levels.
Table 3: Results of univariate test with the sample in two groups.
In this table the results are presented of the univariate test that has been performed on a divided sample. The sample contains companies which are privatized between 1990 to 2008. The sample was divided into two groups namely, companies which are privatized before 2000 and those which are privatized in 2000 and later. The univariate test has been performed on each part of the sample separately. The table contains for each test the mean values of the entries before and after privatization, the mean change of the value after privatization versus before privatization. The Wilcoxon signed rank test have been employed to test for significance in the mean change, the Z-statistics of those test are presented in column 4 & 8 of this table. All data used to computed the
ratios for the Wilcoxon test have been corrected for inflation with the privatization year as base.
Privatised before 2000 Privatised after 2000
Variables
Mean before privatization
Mean after
privatization Mean change
Z statistics or significance of proportion of change Mean before privatization Mean after
privatization Mean change
Z statistics or significance of
proportion of change
Profitability
Return on Equity (ROE) 0,109 0,131 0,022 -0,118 0,205 0,150 -0,055 -2,14**
Return on Sales (ROS) 0,088 0,089 0,001 -1,229 0,13 0,088 -0,042 -1,376
Return on Assets (ROA) 0,0448 0,0615 0,0167 -1,976** 0,087 0,063 -0,024 -1,933*
Efficiency Sales efficiency 1330 2113 783 -5,009*** 875 1183 308,000 -4,731*** Income efficiency 166 148 -18 -3,032*** 85 99 14,000 -0,598 Capital Expenditures Capex / Sales 0,178 0,17 -0,008 -0,781 0,1827 0,1419 -0,041 -1,581 Capex / Assets 0,0958 0,0994 0,0036 -0,447 0,0944 0,0762 -0,018 -1,428 Debt Ratio of debt 0,2863 0,2607 -0,0256 -2,057** 0,25 0,23 -0,020 -0,344 Payout Payout of dividends 0,0103 0,0385 0,0282 -4,948*** 0,0645 0,486 0,422 -1,197 Employment Total employment 28023 26573 -1450 -1,098 34804 30402 -4402,000 -1,103
28
Looking at the debt ratio of the two groups there is a remarkable difference. In the full sample there is no significant change. But in the group ‘before 2000’ a significant change is visible, the mean ratio of debt drops with 8.9%. In the group ‘after 2000’ there is no significant change, however the mean ratio in the group still drops with 5.7%.
The capital expenditures of companies differ slightly for companies that are privatized ‘before 2000’ from those that are privatized ‘after 2000’. But in both groups the change is insignificant. In the group ‘before 2000’ the mean of both capital expenditures ratio remains the same. The ‘after 2000’ group shows a slight drop in the mean ratios, although this change is insignificant.
In the full sample there is a significant rise in the payout ratio of companies, dividing the sample in the two groups gives a different result. In the group that is privatized ‘before 2000’ the mean ratio of payout rises significantly. The ratio of payout in the group that is privatized ‘after 2000’ drops insignificantly.
But does the period that a company gets privatized has an influence on the outcome of privatization? Megginson et al. (1994) found significant evidence on almost every hypothesis that privatization leads to better performance on companies that are privatized before 1990. The review of that study by D’Souza et al. (1999) on the period 1990-1996 also found significant evidence, however a slight decline was observed. A research on a privatization sample between 1992-2000 (Donghui et al., 2007) already found no evidence that leads to significantly better performing enterprises after privatization. Dividing the sample shows that the ‘after 2000 group’ had lower means of profitability and both capital expenditures ratios are lower. Although no real significant evidence was found by the separate univariate test, it is plausible that the ‘after 2000 group’ has lower benefits from privatization.
29 6.3 Subsample Country of origin
A factor that could have an influence on the results of privatization of a company is the country where it is active (Estrin and Perotin, 1991 and D’Souza et al. 2007). In addition to analyzing the full sample and difference between the year of privatization, the firm’s country of origin also have been analyzed if it could have an influence on the performance after privatization. The sample consists of companies from 20 different countries, these have been divided into three different groups which represent a part of Europe and/or have the same economic status namely: West Europe, South Europe and North/East Europe. This have been tested by a one way ANOVA (table 4), which leads to the conclusion that only income efficiency differs significantly between the three groups at a 95% confidential interval. In order to find out between which groups the income efficiency differs a Tukey HSD test have been performed. The result (table 5) shows that West Europe and North/East Europe differ significantly from each other. The prediction was that privatization has more effect in developed countries. This prediction was not met with the sample, only one variable has been found that significantly differs between the three groups, and looking at the results of the Tukey HSD test, Western Europe has a lower income efficiency after privatization in comparison to North/East Europe. If a 10% confidential interval was held, then the return on sales and Sales Efficiency would also be significant. But both variables show that western Europe has less benefits of privatization. It could be that, because our sample only consists of companies from Europe, they do not differ enough to detect any significant difference between the three groups.
30
Table 4: Results of the One Way ANOVA between three regions
In order to test if there is any significant difference on the outcome of privatization between regions in Europe a one way ANOVA has been performed. The sample has been divided in three equal groups: West Europe, South
Europe and North/East Europe. The table contains the following information, sum of squares, degrease of freedom, the mean square and the F value. The ANOVA uses the mean square to calculate the F-value.
Variables
Sum of Squares Degrease of
freedom Mean Square
F-value of mean square
Profitability
ROS Between Groups 0,034 2 0,017 2,713*
Within Groups 0,484 77 0,006
ROA Between Groups 0,011 2 0,006 1,595
Within Groups 0,277 77 0,004
ROE Between Groups 2,458 2 1,229 0,007
Within Groups 13414,427 75 178,859
Efficiency
Sales efficiency Between Groups 3,12E+08 2 1,56E+08 2,837*
Within Groups 4,18E+09 76 5,50E+07
Income efficiency Between Groups 1451601,22 2 725800,61 4,031**
Within Groups 1,37E+07 76 180041,879
Capital Expenditures
Capex / Sales Between Groups 0,018 2 0,009 0,359
Within Groups 1,9 77 0,025
Capex / Assets Between Groups 0,005 2 0,003 0,818
Within Groups 0,244 77 0,003
Debt
Ratio of debt Between Groups 0,029 2 0,014 0,513
Within Groups 2,14 77 0,028
Payout
Payout of dividends Between Groups 0,013 2 0,006 2,636
Within Groups 0,188 77 0,002
Employment
Total employment Between Groups 9,97E+09 2 4,99E+09 1,954
Within Groups 1,94E+11 76 2,55E+09
31
Table 5: Results of the Post Hoc Tuckey HSD test
This table shows the results of the Tukey HSD test, this is a post hoc test of the one way ANOVA. Where the ANOVA only can point out if there is a difference amongst groups, the Tukey HSD test can point out which groups differ. In order to do that the test calculates the mean difference between groups, which can be found in the first column. If the mean difference is
large enough there is a significance difference (last column) also the standard error can be found in the table.
Variable (I) Continent (J) Continent
Mean Difference
(I-J) Std. Error p value
Return on Sales South Europe West Europe 0,0203 0,02132 0,609
North/East Europe -0,03208 0,02156 0,302
West Europe South Europe -0,0203 0,02132 0,609
North/East Europe -0,05238 0,02267 0,06*
Return on Assets South Europe West Europe -0,0009 0,01612 0,998
North/East Europe -0,02653 0,01631 0,241
West Europe South Europe 0,0009 0,01612 0,998
North/East Europe -0,02562 0,01714 0,299
Return on Equity South Europe West Europe -0,01608 3,595 1
North/East Europe -0,40134 3,72822 0,994
West Europe South Europe 0,01608 3,595 1
North/East Europe -0,38527 3,90952 0,995
Sales efficiency South Europe West Europe 503,71662 1993,86271 0,965 North/East Europe -4126,07697 2041,28763 0,114 West Europe South Europe -503,71662 1993,86271 0,965 North/East Europe -4629,79359 2143,08198 0,085* Income efficiency South Europe West Europe 57,27437 114,0591 0,87
North/East Europe -268,11469 116,77204 0,062*
West Europe South Europe -57,27437 114,0591 0,87
North/East Europe -325,38905 122,5952 0,026**
Capex / Sales South Europe West Europe 0,0052 0,04223 0,992
North/East Europe -0,02979 0,04271 0,766
West Europe South Europe -0,0052 0,04223 0,992
North/East Europe -0,03499 0,0449 0,717
Capex / Assets South Europe West Europe -0,00709 0,01513 0,886
North/East Europe -0,01951 0,01531 0,414
West Europe South Europe 0,00709 0,01513 0,886
North/East Europe -0,01242 0,01609 0,721
Debt ratio South Europe West Europe 0,03793 0,04481 0,675
North/East Europe 0,03956 0,04533 0,659
West Europe South Europe -0,03793 0,04481 0,675
North/East Europe 0,00163 0,04764 0,999
Payout of dividends South Europe West Europe 0,02846 0,01329 0,088*
North/East Europe 0,00271 0,01344 0,978
West Europe South Europe -0,02846 0,01329 0,088*
North/East Europe -0,02575 0,01413 0,169
Total employment South Europe West Europe -9668,2972 13578,77036 0,757 North/East Europe 18752,11033 13901,74751 0,373 West Europe South Europe 9668,2972 13578,77036 0,757 North/East Europe 28420,40754 14594,99592 0,133 ***, **, * Significant at the 1, 5, and 10 percent levels
32 6.4 Multivariatie analysis
Although the univariate test has provided an insight in the different variables, the possible relation among the variable hasn’t been tested. The regression that has been performed tests whether the relationship between ‘firm size’, ‘the year of privatization’, ‘privatized yes/no’, ‘the time a company is privatized’ and ‘the industry’ effect the outcome of the tested variables. The tested variables are: return on sales, return on equity, return on assets, debt ratio and the payout ratio. The results of the regression on the performance indicators are reported in table 6. The outcome of the regression shows that return on sales and debt ratio are significant at the 5 percent level and the payout ratio of dividends is significant at the 1 percent level.
Out of the individual coefficients we can conclude that ‘firm size’ which is measured by the sales of a company doesn’t affect the outcome of privatization, none of the five variables tested shows a significant difference at that coefficient. It is also quite remarkable that the dummy ‘private yes/no’ isn’t significant in any of the five variables. However this is in line with the research of Mikkelson, Partch, and Shah (1997) which also found no increase in performance after privatization. The dummy ‘year from privatization’ is used to indicate whether the data entry was two years from privatization or more. No significant relation between the dummy and the five tested variables could be found. This could imply that firms don’t gain multiple years from privatization, but that it only generates a one-time benefit. These results are in line with our results and those of the research of Anderson et al. (1997) Their research found that the benefits of privatization emerge five years after privatization. The ‘year of privatization’ does show significant results. Three dummy variables have been used in order to test whether the ‘year of privatization’ has an effect on the five tested variables. The first dummy variable contains companies that are privatized between 1990 until
33
1994, the second between 1995 and 1999, and the third between 2000 and 2004. The first dummy is significant for all variables expect return on equity. Only return on assets, debt ratio and the payout ratio are significant by the second dummy. The third dummy shows that return on sales as well as the debt and payout ratio are significant. It is remarkable that the debt ratio and payout ratio are significant across all dummies. The univariate test between the groups ‘before 2000’ and ‘after 2000’ test shows that the ‘before 2000’ group has a significant change in payout ratio, but the ‘after 2000’ group doesn’t. However the univariate test of the complete sample give the same results as the regression. But based on the results of the coefficient there is no strong evidence that the year of privatization has a large effect on the results of privatization, at least not in our sample.
Looking at the industry dummies only the industry dummy that contains the ‘power’ companies show a small significant change for the return on equity. The industry dummy ‘telecommunication and transportation’ is significant at the 5 percent level for return on sales and at the 10 percent level for the payout ratio. However these results don’t lead to the conclusion that the industry has a large effect on the outcome of privatization.
Research of Barber and Lyon (1996) state that large firms benefit more from privatization, our results doesn’t comply with that, however Sug and Tong (2003) also argue that large firms benefit more from privatization? According to their study large firms have more market power, but also have to deal with too many employees who interfere negatively with the performance of a company. Although the univariate test has relatively similar results, it still is surprising to see that the dummy ‘private yes/no’ isn’t significant at any of the five tested variables. These results are in line with those of Grompers & Lerner, (2003) and Eckbo & Norli, (2005) which also found no evidence of better performance after privatization.
34
Table 6: Results of the multivariate analysis
To test whether there are any relations between the different variables which could have an effect on the overall outcome of the research a multivariate test has been performed. With this test five variables has been tested, the return on equity, return on sales, return on assets, the ratio of debt and the payout ratio of dividends. The table contains the mean square and the F statistics of the complete regression. The second part of the table contains the t value of the independent coefficients used in the regression. The 'dummy private yes/no' is used to indicate if the entry of a firm is from the pre or post privatization period. The dummy year from privatization is one if the data entry was from two or more years after privatization, otherwise zero. The dummy year of private 1,2,3 were used to categorize the entry in the year a company is privatized. The dummy gets the value one if a company was privatized between 1990 - 1994, otherwise zero. The other dummy groups where used to categorizes the groups 1995 - 1999 and 2000 - 2004. The last column shows the sales of a company which was used to measure the influence of ‘firm size’. The other three columns are industry dummies. If a company belongs to one of the three groups it gets the value one, otherwise zero.
Regression Coefficients Mean Square F Dummy private yes/no Dummy year from privatization Dummy year of private 1 Dummy year of private 2 Dummy year of private 3 Industry dummy: Power Industry dummy: Industrial Industry dummy: Telecomunication & Transportation
Sales (Firm Size)
ROE 263,934 1,01 0,028 -0,877 -0,741 -1,18 -0,202 -1,917* -1,486 -1,381 0,91 ROS ,029 2,270** -,858 -,177 -2,319** -1,378 -2,074** 1,659* ,932 2,377** -1447 ROA ,005 ,858 ,024 -,385 -1,693* -1,840* -,959 ,108 1,037 1,178 ,057 DEBT RATIO ,075 1,979** ,000 -,696 1,999** 2,879*** 2,825*** ,448 -,203 1,093 -1,149 PAYOUT RATIO ,018 3,071*** ,643 ,195 -4,098*** -3,109*** -3,364*** 1,485 ,836 1,885* -,822
35 7. Discussion & Conclusion
This research studied the effect of privatization on the performance of companies in Europe that were privatized between 1990 and 2008. The results show that the profitability of privatized firms in the sample didn’t improve, the ratios ROS, ROA, ROE all showed an insignificant decrease. Nevertheless the efficiency of privatized companies did partially improve significantly. Sales efficiency shows an 50% increase after privatization with a remarkable 92% of the companies in the sample that changed in the way that was predicted. That the net income efficiency shows a significant decline could also have a relation to the profitability of privatized companies. However it is surprising that 62% of the firms did increase their income efficiency. Although the expectation was that capital expenditures would increase after privatization, the results show a small insignificant decline after privatization. This research found that 50% of the companies have lower employment after privatization, although a decline was expected the result is insignificant. As was expected the ratio of debt significantly declined after privatization by 7.4% . The payout of dividend shows a significant increase of 27% after privatization which is in line with the prediction.
But to what extent does these results differ from the previous research done? This research made use of the same theoretical model as the researches of Megginson et al. (1994) and D’Souza et al. (1999) The research of Megginson et al. (1994) found that companies that are privatized before 1990 increased their: profitability, employment, efficiency, capital investments and dividend payments significantly. Furthermore they found a significant decline in the leverage of a firm. D’Souza et al. (1999) found similar results in their sample of privatized firms between 1990 and 1996, except that employment and capital investments show a insignificant decline. In comparison to this research the effect of privatization has changed. No evidence is found that the profitability of a company will increase after privatization. Our research only found partial evidence that the efficiency of a company will
