• No results found

Moderating effects on the diversification performance relationship in Chinese firms : the government's influence

N/A
N/A
Protected

Academic year: 2021

Share "Moderating effects on the diversification performance relationship in Chinese firms : the government's influence"

Copied!
37
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

MODERATING EFFECTS ON THE

DIVERSIFICATION PERFORMANCE

RELATIONSHIP IN CHINESE FIRMS

THE GOVERNMENT’S INFLUENCE

L. Kretzschmar (11132329) Amsterdam Business School Executive Program in Management Studies

Specialization: Strategy & Organization Academic year: 09/ 2016 – 03/ 2018 Supervisor: Dr. Christopher Williams

(2)

Contents Statement of Originality ... 3 Abbreviations ... 5 Abstract ... 6 1. Introduction ... 7 2. Literature Review... 9 2.1. Diversification Theory ... 9

2.2. Drivers for Diversification ... 10

2.3. State Ownership in China ... 10

2.4. Research model ... 12

3. Method ... 13

3.1. The Variables ... 13

3.1.1. Independent variable: diversification (X) ... 13

3.1.2. Dependent variable: performance (Y) ... 14

3.1.3. Moderator: State ownership (M) ... 15

3.1.4. Control variables ... 16 3.2. The Sample ... 17 3.3. Statistical Method ... 18 3.4. Descriptive ... 20 4. Results ... 21 4.1. Main analysis ... 21

4.2. First stage – Regression analyses including all firms ... 23

4.3. Second stage – Regression with partitioned data ... 26

(3)

5.1. Theoretical and practical implications ... 30

5.2. Limitations ... 31

5.3. Further research ... 31

References ... 33

(4)

Statement of Originality

This document is written by Student Lydia Kretzschmar who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is

(5)

List of tables and figures Tables

Table 1: Diversity Degree by distinct US SIC codes 14

Table 2: Shareholder type and assigned dummy variable 15

Table 3: Sample Characteristics 18

Table 4: Firm Performance of group 1 and 0 with increasing diversification 21 Table 5: Mean, Standard deviation and Correlations of Study Variables 22 Table 6: Performance Measures Descriptive statistics one-way ANOVA 22

Table 7: Regression Results for all Firms 23

Table 8: Regression Results for private firms 27

Table 9: Regression Results for state-controlled firms 28

Table 10: Industry classification 36

Figures

Figure 1: The inverted U-model (Palich, Cardinal, & Miller, 2000) 9 Figure 2: Interaction Effect of the Regression for all firms 25 Figure 3: Private firms/ curvilinear effect based on the regression equation 28

(6)

Abbreviations Abbreviation CC Corporate Company DV Dependent Variable EE Emerging Economy IV Independent Variable MM Million

KPI Key Performance Indicator

PE ratio Price Earnings ratio

ROCE Return on Capital Employed

USD Currency US Dollars

(7)

Abstract

This study reviews the classic theory of diversification and past attempts to

understand the implications on financial performance. My empirical evidences show that for state-influenced firms, diversification dispersion is negatively associated to financial performance. However, for private firms, this effect is not present for most performance measures. Based on financial data from a sample of China’s most successful companies, this study is the first to directly examine the moderating effects from ownership structures on the diversification and performance relationship. My results depict how important it is to consider the ownership type when diversification strategies in emerging market, where, such as in China, state ownership is still prevalent, are assessed. The results also reveal that firm size, industry, employee power, location, experience and innovation affect company performance in China. Explicitly, emerging markets offer the opportunity to observe the diversification in the light of the political system and the…. The paper demonstrates that previous

assumptions by (Palich, Cardinal, & Miller, 2000) on the positive relation of government’s political networking and functional experience to new venture

performance are proven wrong, especially with regards to the case of unique political environment.

Keywords: corporate governance, ownership structure, political interference, state-owned enterprise (SOE), diversification, performance, China

(8)

1. Introduction

In China’s transition economy, a firms’ ties with the government represent a unique type of managerial resource, because the government still controls significant portions of strategic factor resources and exerts considerable power over firms with regards to the approval of projects and allocation of resources (Li & Zhang, 2007). As a result, managers tend to maintain a ‘disproportionately greater contact’ with the government. A firm may benefit from these political ties, if the ownership structure entails a governmental stake in a firm.

In 2016 and 2017, the Chinese conglomerate HNA Holding Group Co. Limited (HNA) undertook several acquisitions that drew the attention from the media. Financial Times describes the HNA Group as aggressive dealmaker, because of their international buying spree (FT, 2017). It has shown impressive capabilities to leverage existing assets allowing them to fund the purchase of new ones. In Chinese, this process is also known as a “snake swallowing an elephant”. Because of numerous acquisitions, HNA’s corporate structure is very complex, and its total assets doubled last year to about 159 billion USD (Fickling, 2017). In other words, the privately-owned company that was founded in 2000 is now a bigger enterprise than Johnson & Johnson or Glencore Plc, 342 million USD and 78 billion USD respectively (reuters.com, 2018). The scale of HNA’s approach to equity-backed borrowing makes credit-rating agencies concerned about how the company can manage its leverage across the various businesses (South China Morning Post, 2017).

On the other hand, ZTE Corporation (ZTE) provides an example of a mostly state-owned company. ZTE is a Chinese multinational telecommunications equipment and systems company that operates in three business units: carrier networks, terminals and

telecommunication (zte.com, 2018). In 2017, ZTE has formed partnerships with over 20 operators to jointly advance the verification and testing of 5G technologies, thereby

(9)

accelerating 5G commercial deployments. The company does not undertake acquisitions outside its related businesses (Telecomlead, 2018), and financial performance remained profitable throughout 2017. Cases like HNA and ZTE provide evidence that the political degree of engagement in a company’s diversification strategy has effects on the

diversification-performance relationship. This study aims to examine how state ownership in Chinese enterprises moderates the diversification-performance relationship. It is assumed that the political degree of engagement in a company’s diversification strategy leads to a different performance compared to their privately-owned counterparts.

“The effect of state ownership on the diversification performance relationship in Chinese firms”

As for the listed firms in China, it is common that big shareholders enjoy priority of information, primarily because the board of directors tend to consult with the big

shareholders to discuss their strategy and to seek direct support from them (OECD, 2016). Where the state holds shares in firms, an influence may affect the diversification strategy and hence the firm´s performance. The following analysis shows government influence as

moderator of the diversification performance relationship. This research addresses this effect by answering the following research question: How does state ownership influence

diversification decisions and therefore the related firm’s business performance? To answer these research questions, the following hypotheses will be discussed:

Hypothesis 1:

The relationship between diversity of a corporate company and its financial performance is negative or curvilinear, so that an increasing diversification in unrelated businesses is associated with increasing performance at lower levels of diversification and decreasing performance at high diversification.

(10)

Hypothesis 2:

The curvilinear relationship between diversification and a firm´s performance is negatively moderated by governmental influence in such a way that it attenuates the positive effect at a lower degree of diversification and amplifies the negative effect at a higher degree of diversification.

In the following chapters the result of the research will be discussed. As a starting point, existing research up to date is reviewed, followed by an elaboration on the different variables of the hypotheses and the statistical method behind it. Thirdly, the outcome and implications of the research are being discussed. Finally, a conclusion, further research suggestions and its limitations are summarized in the last chapter.

2. Literature Review 2.1. Diversification Theory

Several studies have been conducted in the attempt to research the linkage between diversification and performance, and so far, no

consensus could be reached. It remains considerable disagreement about when and in what extent diversification leads to a better performance. A number of studies focus on performance differences between related and unrelated diversification, such as (Markides & Williamson, 1994). Palich, Cardinal & Miller (2000) derive three competing models of the diversification- performance relationship from

previous literature and assess the three models using data from 55 previous studies. They find support that moderate levels of diversification yield higher levels of performance compared

Figure 1: The inverted U-model (Palich, Cardinal, & Miller, 2000)

(11)

to either limited or extensive diversification. This translates into the curvilinear model, meaning that performance increases as firms shift from single-business strategies to related diversification, while it decreases when changing from related to unrelated diversification. The curvilinear or inverted-U model is illustrated in Error! Reference source not found..

Past literature examines the nature of the relationship between diversification and performance but does not examine moderating effects on the relationship. Moreover, previous literature does not take into consideration the political engagement as trigger for strategic diversification and the effect on performance compared to companies without political attachment.

2.2. Drivers for Diversification

Diversification is positively associated to performance for various market power advantages (Palich, Cardinal, & Miller, 2000). However, there are mixed results, in the context of predation. Predatory pricing or reciprocal buying can be seen as evidence for the performance diversification relationship. This explains why research has shifted away from the argument of pure market power as sole justification. Especially when it comes to conglomerates, value maximizing behavior can be seen (McCutcheon, 1991).

2.3. State Ownership in China

The U.S.- China Economic and Security Review Commission (USCC) annually reports the national security implications of the bilateral trade and economic relationship between the two countries. According to USCC, the state-owned enterprises in China consists of three groups (Szamosszegi & Kyle, 2011):

i. Fully state-owned enterprises and Supervision and Administration

Commission (SASAC) of the State Council and by SASACs of provincial, municipal, and county governments

(12)

ii. SOEs which are majority owners of enterprises and are not officially considered SOEs but are effectively controlled by their SOE owners iii. Entities, owned and controlled indirectly through SOE subsidiaries based

inside and outside of China. The actual size of this third group is unknown. Furthermore, the USCC states that the Chinese government plans a strategy for certain industries that they view as important. These include not only strategic industries, but also emerging industries. China’s SOEs benefit from several government preferences in China. Based on recent U.S. regulatory filings by SOE-owned entities, the SOEs and their subsidiaries benefit from preferred access to capital, lower market interest rates on loans from state-owned banks, advantageous tax treatment, policies that create a favorable competitive environment and large capital injections if needed (Szamosszegi & Kyle, 2011).

In 2000, Qi et al. (2000) analyzed a sample of Shanghai Stock Exchange-listed Chinese firms from the early 90’s and concluded that state equity ownership has a negative effect on the operating performance. Further studies from Ng et al. (2009) and Hess et al. (2010), who examine Chinese listed firms of the early 2000s confirm the findings by Qi et al. (2000). The authors find a convex relationship between state ownership and market

performance. However, Sun et al. (2002) examine a sample of Chinese listed firms in a similar period and conclude that state equity ownership has a concave relationship with the overall market performance. The authors argue that governmental support and business connections provided through state ownership add value and are necessary to vitalize

performance. This contradicting assumption in comparison to Qi et al. (2000) accentuates that state ownership may result in a negative performance of the firm, but under certain

circumstances allows for benefits that can be utilized for a firm’s benefit. The circumstances may be unique in China. Reddy, Xie & Huang (2016) analyze cross-border acquisitions by state-owned and private enterprises from the perspective of emerging economies. Their

(13)

research question investigates “what acquisition strategy do emerging economy (EE) enterprises follow for cross-border acquisitions in China and India over the last two

decades?”. Their hypothesis is that cross-border acquisitions enhance the competitiveness of EE firms by offering immediate ownership control over the target firm´s resources and their strategic assets such as advanced technology, branding and the distribution network.

Reddy, Xie & Huang (2016) find that Chinese SOEs, which complete more high-valuation deals, seek to invest in resource-scarce industries, and diversify their risk by targeting developed, emerging and developing countries. Also, Luo, Zhou, & Liu (2005) found that SOEs are not reported to be less entrepreneurial than international joint ventures.

The study suggests that overseas acquisitions by Chinese SOEs are motivated to secure scarce resources for their home market and expand into other markets through host country development. The intention is to acquire ownership control and strategic assets in diversified sectors, developed economies, thus to improve the overall firm´s market value, promise significant stock returns to shareholders and compete in their home market (Reddy, Xie, & Huang, 2016).

When state-owned enterprises enter the stock market, certain changes can be observed (OECD, 2016). Firstly, as a consequence of stock-market listings, the commercial orientation leads to an improved financial structure and to a low-cost financing channel. Secondly, the proportion of independent directors grows as a requirement in the governing provisions. The funds and system improvements caused by listings have improved the commercial objectives. Fourthly, listed companies provide a higher level of transparency (OECD, 2016).

2.4. Research model

Diversification is the independent variable (X), performance the dependent variable (Y). The relationship between the two is moderated by political involvement that is expressed

(14)

by state ownership. In other words, political ownership moderates the strength of this relation.

3. Method 3.1. The Variables

For the examination of the hypotheses the variables diversification, performance and state ownership are used. These variables are defined as follows:

3.1.1. Independent variable: diversification (X)

Diversification of the corporate company is defined as the spread across industries. For each CC and its numerous subsidiaries, the U.S. SIC core codes are compared. If the U.S. SIC core code of the subsidiary differs from the CC, it is counted as spread into one other (unrelated) business. The sum of all distinct U.S. SIC core codes of the subsidiaries is used to assign a degree of diversification ranking from 1 to 7, while 1 expresses no or little

diversification and 7 the highest degree of diversification. Table 1 shows the possible number of industries that a CC is spread through its subsidiaries and how many CCs show this spread. When the number of CC is below 100, then the assigned degree is grouped. The least spread

State Equity Share [M]:  Shareholder type ‘authority,

government or public’ >15% then the CC is [1] state-influenced, otherwise [0] private Control variables:  Firm size  Innovation (patents)  Top City  Experience (age)  Industry Sector

Business Performance [Y]:  Profit  ROCE  ROA  P/E ratio

H2

H1

Diversification across industries

[X]:

 1: zero or one other industry  2: 2 other industries

 […]

(15)

of zero occurs for 52 different companies that have no subsidiary in a different industry and the highest diversification is seen among five companies that own subsidiaries which are spread across nine distinct industries.

Table 1: Diversity Degree by distinct US SIC codes # distinct US SIC core codes

(spread of subsidiaries in other industries)

#companies Degree of diversity

0 52 1 1 342 2 566 2 3 594 3 4 516 4 5 395 5 6 219 6 7 120 7 8 36 9 5

3.1.2. Dependent variable: performance (Y)

A firm’s performance is measured through various financial key performance

indicators (KPIs). At first, the profit or loss before tax in USD from the financial year 2014 is used. This financial figure from the income statement shows the overall profitability in absolute numbers. Furthermore, I take a closer look on the operating ratios, such as ROCE, the return on capital employed, which illustrates how well a company is using equity and debt to generate a return. Operating ratios can be used to evaluate the firm’s return by comparing its income to its investments (Berk & DeMarzo, 2017). ROCE is the most useful measure in assessing the performance of the underlying business. Further, I include ROA, the return on assets, which includes interest expenses. Finally, a common valuation ratio, which is the price-earnings (PE) ratio is added. The PE ratio allows to gauge the market value of the firm (Berk & DeMarzo, 2017). The financial KPIs are available for the financial years 2013 till 2016. Hence, all dependent variables from four years were separately analyzed and the results for the years that were most outstanding are summarized in section 4.

(16)

3.1.3. Moderator: State ownership (M)

Extant research suggests that the type of ownership is critically linked to the strategic choices, especially when size and scope of the business are decided (Fernández & Nieto, 2006). Ownership type can influence corporate strategy and performance because it is related to different degrees of risk aversion (Thomsen & Pedersen, 2000). For simplification

purposes the ownership structure is expressed as a dummy variable [1] for government-controlled and [0] for private firms. The analysis of state influence is based on information from the Orbis database on ownership structures. The database provides a list of all

shareholders, the corresponding shareholder type and the direct share in percent in the corporate company. The shareholder types are divided into sixteen distinct categories to identify the influence of the government by the sum of direct share. Government influence is already assumed, when the sum of direct share by group 1 of the shareholder types exceeds a threshold of fifteen percent. For the ownership structures, it can be assumed that those remain generally quite stable over time and do not change in response to economic circumstances. The structural stability allows to regard ownership as exogenous variable (Thomsen & Pedersen, 2000).

Table 2: Shareholder type and assigned dummy variable

Shareholder type Private [0] or Governmental [1]

Companies

[0] Private equity firm

One or more named private person (s) or family (s) Venture capital

Employee / Manager / Director own property

Group of natural persons not named by name Hedge fund

Bond and investment funds / trust companies Bank

Insurance

Financial institution

Group of other non-named shareholders Authority / State / Government

[1] Public (in listed companies)

(17)

3.1.4. Control variables

Five control variables are entered when testing the hypothesized relationships. These control variables are widely believed to influence the relationship between diversification and performance.

I. Company Size

The number of employees of the CC is used as a measure for company size. By means of the number of employees, the companies an assigned to groups with less than 500

employees, 500 to 5,000 employees, 5,001 to 50,000 employees and over 50,000 employees. Previous literature has documented that the size of a company as an organizational attribute significantly impacts a firms’ strategic decision making and financial performance (Luo, Zhou, & Liu, 2005).

II. Experience Variable

The age of a firm can serve as indicator for experience. The variable age was

calculated by deducting the year of foundation of the CC from the current year (2018). Also, the age contributes on the practice of corporate entrepreneurship, which in turn results in superior performance (Luo, Zhou, & Liu, 2005).

III. Industry of CC

Industry classification is the main sector description by the Bureau van Dijk

investment database. The dataset contains nineteen distinct industry sectors of which the top three by the number of companies are

1. machinery, appliances, furniture & recycling, 2. chemistry, rubber, synthetics & non-metals and 3. other services.

(18)

The full list of industries and corresponding number of companies and total turnover is documented in Table 10 in the attachment.

IV. Location

The corporate headquarters are spread across 364 distinct cities in China, whereas the top three locations are Beijing counting 304 firms, Shanghai counting 261 firms and

Shenzhen counting 201 firms. All remaining cities are summarized under ‘others’ in the model. Anecdotally, the big governmental projects of Chinese cities intended to accelerate the transition of the economy through large investments into infrastructure, which lead to certain distinctions locally. One example for this are firms in Tianjin, which are more entrepreneurial than firms in other regions as Tianjin is one of the four centrally administered municipalities and the largest industrial and port city in Northern China (Zacharias & Yang, 2016). The urban plans were executed in hundreds of cities, although there is no doubt that Beijing, Shanghai, Guangzhou, and Shenzhen remain the most advanced (Zacharias & Yang, 2016).

V. Innovation

The number of patents a CC has registered will serve as indicator for the level of innovation. However, this variable is only filled for sixty percent of the firms and therefore assumptions are made cautiously.

3.2. The Sample

For this analysis, the dataset was retrieved from Orbis, a database by Moody’s Analytics Company that provides comprehensive information on both listed and unlisted companies worldwide (UvA, 2018). Orbis includes corporate and ownership structures, financial information, stock data and other information. Students of the University of Amsterdam receive free access to this database. For this sample very large and large, stock-listed and active companies in China that fall under the definition ‘corporate company’, were retrieved from the database. Thus, the dataset includes 2,845 companies and information on

(19)

the ownership structure, fully-owned subsidiaries, financial performance and control variables. Table 3 summarizes the sample characteristics.

Table 3: Sample Characteristics

N In Total 2,845 By Location Beijing 304 Shanghai 261 Shenzhen 201 Other Cities 2,079 By Size < 500 employees 563 500 – 5,000 1,735 5,001 – 50,000 486 > 50,000 21 By Industry

Machinery, Appliances, Furniture, Recycling

913 Chemistry, Rubber, Synthetics,

Non-metal products

565

Other services 357

Wholesale & Retail 179

Metal & Metal Products 176

Textile, Apparel, Leather 111

Food, Beverages, Tobacco 102

Other 442

By Age

< 10 years 88

Between 10 and 25 years 2,385

> 25 years 375 By Innovation < 10 patents 874 Between 10 and 100 422 Between 100 and 500 248 > 500 78 3.3. Statistical Method

For the statistical analyses the data set is analyzed in SPSS. As described earlier dummy variables where created for the variables diversification and state ownership. Assuming the presence of an inverted U-shaped relationship, the squared value of the

(20)

independent variable is computed and named X_DIV2. Further, I standardized the dependent and independent variables, as well as the control variables, so that they are centered around zero and have a standard deviation of 1. Descriptive statistics, skewness, kurtosis and normality tests were computed for all variables. The dependent variables (performance measures) are normally distributed. Positive kurtosis was found for profit/ loss in USD and it appeared that they have been caused by minor outliers. The outliers were examined to ensure no data entry or instrument errors were made. A normality test of the variable with exclusion of outliers showed a normal distribution. A One-way ANOVA was used to test hypotheses related to the differences of means between the populations companies with state-influence and private companies.

The moderating variable is assumed to change the direction or magnitude of the relationship between the predictor variable and the dependent variable (Baron & Kenny, 1986). For appropriately measuring the moderation hypotheses, the statistical analysis must test for the differential effects of the predictor variable on the dependent variable as a

function of the moderator (Baron & Kenny, 1986). Therefore, I will consider three models: In model 1, the control variables are added, which capture firm specific characteristics, such as experience by company age, the level of innovativeness by the number of patents that a firm holds and a dummy variable that specifies the industry of the CC are entered to the

equitation. In model 2, the dependent or predictor variables are entered to capture the direct effect of the relationship. And in model 3, the moderator variable is added to the equitation. The hypothesis is tested by adding the product of the moderator and the IV to the regression equation. Hence, when IV is denoted as X, the moderator as M and the outcome variable as Y, then Y is regressed on X, M and XM. The moderator effects are indicated by the

significant effect of XM while X and M are controlled (Baron & Kenny, 1986). The regression model is also tested for the quadratic effect which is determined by the

(21)

significance of the squared term of the independent variable ‘diversification’. After

partitioning the data into private and state-influenced, the quadratic effect of the IV is tested on both groups. Because I am testing a moderated quadratic relationship, it is the significance of the interaction between the squared term and the moderator that determines whether there is a moderated effect. The regression analyses were undertaken for four outcome variables: (1) profit/ loss before tax in million USD, (2) return on assets in percent, (3) return on capital employed in percent, and (4) price-earnings ratio.

3.4. Descriptive

The sample consists of 2,845 datasets of individual corporate companies, thereof 193 firms with governmental stake and 2,652 firms without relevant governmental stake. A first glance at the two groups provide an overview of the dataset.

As illustrated in Error! Reference source not found. the government-influenced firms [1] and private firms [0] differ in the selected performance measures depending on their degree of diversification. The average turnover for the financial year of 2015 is lower for [1] in all degrees of diversification than for [0] with the exception when [1] is more diversified into five industries. Similar pattern can be observed for the average profit, but here, group [1] generates higher profits compared to [0], if they are not diversified or again highly diversified into other industries. Hence, the overall impression is that private firms perform better on average. However, for private firms the increasing degree in diversification has a negative influence on some measures, such as the ROCE and ROA. For the PE ratio, some

diversification appears to be positive for the private firms. For the state firms, the average values show that some diversification is good, but high diversification is extremely negative. The further regression analysis gives more certainty on the relationship of the variables.

(22)

Table 4: Firm Performance of group 1 and 0 with increasing diversification1 KPI [means]2 Gov. or Private [1] [2] [3] [4] [5] [6] [7] TTL #Companies 1 52 51 36 31 12 6 5 193 0 342 515 558 485 383 213 156 2,652 Turnover in MM USD 1 288 568 187 325 2,081 167 749 471 0 740 621 579 493 510 615 843 600

Profit before tax in MM USD 1 284 80 23 23 103 27 148 118 0 109 44 42 46 48 51 78 55 ROCE in % 1 14.5 16.0 17.0 11.7 13.7 10.9 -0.9 14.3 0 11.4 7.8 7.0 7.9 5.4 6.3 7.0 7.5 ROA in % 1 8.6 8.4 8.3 5.3 8.6 4.7 0.9 7.7 0 5.8 5.3 3.9 3.7 3.5 4.1 3.2 4.3 P/E ratio 1 38 23 32 77 64 67 196 52 0 70 62 97 67 229 86 70 99 4. Results 4.1. Main analysis

The mean, standard deviation and correlations of the study variables are provided below. Table 5 shows that state influence is negatively related to the diversity degree and significant (.000, p < .01). Further, state influence is positively related to profits (.040, p < .05), as well as ROCE and ROA (.000, p < .01). Diversity is negatively related to ROCE, ROA and significant (.000, p<.01). Hence, the correlations show that being state-controlled has a positive effect on firm performance and diversity has a negative effect.

1 Diversity degree varies from [1] little or no diversification to [7] heavily diversified, spread in 7-9 different industries other than the industry of the corporate company (by U.S. SIC Core Code) 2 KPI: Key Performance Indicators are the rounded average values of the group from financial data of the last available year (2016)

(23)

Table 5: Mean, Standard deviation and Correlations of Study Variables N M SD 1 2 3 4 5 1. State Influence 2,845 .07 .252 2. Diversity Degree 2,845 3.44 1.70 - .121** 3. Profit [MM USD] 2,792 59.62 393.64 .040* -.032 4. ROCE [%] 1,909 7.93 16.90 .110** - .067** .094** 5. ROA [%] 2,824 4.46 8.42 .108** - .093** .019 .767** 6. P/E Ratio 1,946 96.65 967.56 -.010 -.023 -.014 -.125** -.065**

**. Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).

A one-way ANOVA is conducted for all financial measures profit, ROCE, ROA and PE ratio. The mean values for profit, ROA and ROCE of private companies are significantly lower compared to the state-influenced companies, while the PE ratio is larger. The ANOVA shows that the between groups effect for profit (.034, p < .05) and ROCE and ROA is significant (.000, p < .01). For the PE ratio, the between groups effect is not significant.

Table 6: Performance Measures Descriptive statistics one-way ANOVA

State Influence M SD N

Profit/ Loss in MM USD (2014)

[0] Private 55.37 301.47 2,603 [1] State influenced 118.08 1,019.25 189 Total 59.62 393.64 2,792 ROCE in % (2015) [0] Private 7.44 17.12 1,783 [1] State influenced 14.92 11.43 126 Total 7.93 16.90 1,909 ROA in % (2015) [0] Private 4.21 8.42 2,633 [1] State influenced 7.85 6.98 191 Total 4.46 8.42 2,824 P/E Ratio (2016) [0] Private 98.91 991.32 1,853 [1] State influenced 51.50 84.36 93 Total 96.65 967.56 1,946

I tested the hypotheses by conducting a multiple hierarchical regression analysis in two stages. In the first stage, a regression including all firms, both state-controlled and private, is run in SPSS comprising three models for each dependent variable. The outcome of the regression analysis that predicts the firm performance for the complete dataset is shown in

(24)

Table 7. In the second stage, the data is partitioned using the selection variable ‘ownership’. Ownership is binary and can be divided into state-influenced and private firms. Thus, if the dummy variable is equal to zero (private) and if the dummy equal to one (state-controlled), the regression was run again to test for diversity effects on performance solely for each group and each outcome variable.

4.2. First stage – Regression analyses including all firms Table 7: Regression Results for all Firms

Profit ROCE ROA P/E Ratio

M1 M2 M3 M1 M2 M3 M1 M2 M3 M1 M2 M3 Control variable Firm Size .439*** .440*** .441*** .053 .061* .063* -.011 -.009 -.008 -.074* -.075* -.074* Location -.060** -.061** -.064** -.041 -.039 -.043 -.055* -.058* -.061* .032 .029 .029 Age .185*** .193*** .194*** -.058* -.037 -.034 -.035 -.023 -.022 -.005 -.003 -.003 Innovation -.090*** -.091*** -.091*** -.017 -.022 -.022 .014 .012 .012 .020 .020 .020 Industry .014 .014 .017 -.010 -.008 -.005 -.026 -.027 -.025 -.017 -.018 -.020 Z-Div -.062** -.047* -.064* -.051* -.105*** -.093*** -.020 -.029 Z-State .068** .175*** .112*** .195*** .091*** .177*** -.015 -.091 Interaction Div x State -.125** -.097* -.101* .088 Div²x State -.066* -.061 .067 Fitness indices R² .230 .239 .242 .007 .025 .028 .005 .027 .029 .006 .007 .009 Adj. R² .228 .236 .238 .004 .021 .023 .002 .023 .025 .003 .002 .003 F 99.668 75.022 66.521 2.094 5.442 5.282 1.642 6.523 6.331 1.761 1.368 1.518 df 1670 1668 1667 1470 1468 1467 1673 1671 1670 1364 1362 1361 Δ R² .010 .003 .018 .003 .022 .003 .001 .002 Δ F 10.556 5.574 13.722 4.083 18.637 4.887 .388 2.556

Note 1: *p<.05; **p<.01; ***p<.001, one-tailed test. The entries in the table are standardized coefficients (βs).

In general, the regressions consist of three models: model 1 (M1) tells how successful the control variable can predict the performance variables. For profits/ losses after tax, R square is .230, meaning that firm size, location, age, innovation and industry account for 23 percent of the variation in profits/losses. For the other DV, R square accounts for less than 3 percent of the variation in ROCE, ROA and PE ratio. The adjusted R square values show that all models generalize well, since they are very close to the value of R square, which is ideal. I

(25)

also analyzed the Variance Inflation Factors (VIFs) to further ensure that there are no multi-collinearity issues. The VIFs are all below 3.7 for all models; hence they are below the limit of 10 proposed by (Myers, 1990). Further, I tested for the squared variable of diversity in another regression, but it is not significant and therefore does not confirm collinearity.

For the DV ‘profit’: the change statistics show that the change in R square is

significant. So, model 1 causes R square to change from zero to .230, and this change in the amount of variance explained, gives rise to an F-ratio of 99.668, which is significant with a probability less than .000. In model 2, in which the diversity level and the state influence (dummy) variable have been added as predictors, R square increases by .010, making the R square of this model .239. This increase yields an F-ratio of 75.022, which is significant (p < .000). Therefore, it can be concluded that the regression model overall predicts firm performance in terms of its profits significantly well. If the predictor variable ‘diversity’ is increased by one unit, then the model predicts that profits will decrease by 6.2 percent for every additional degree of diversity. However, profits will increase by 6.8 percent, if the firm is state-controlled. Moreover, in model 1 to 3 of the DV ‘profit’ the control variables location and innovation are negatively related and significant, while firm size and experience (age) are positively related and significant. Interestingly, diversity is negative and significant, while the direct effect of state ownership is positive and significant.

(26)

Figure 2: Interaction Effect of the Regression for all firms

In Figure 2 above, the two-way interaction effect for the binary moderator

‘ownership’, which is either controlled [1] or private [0] is illustrated. Clearly, the state-influenced firms achieve greater profits at low levels of diversity compared to the private firms. But the slope for the state-controlled line is steeper, which accentuates that higher levels of diversification for this group is even more negative than for the private firms.

For the DV ‘ROCE’: the change statistics show that the change in R square is significant for model 2 (p<.001) and model 3 (p<.05), but not model 1. The change of R square from model 1 to model 2 is 1.8 percent, and this change in the amount of variance explained gives rise to an F-ratio of 5.442, which is significant with a probability less

than .000. If the predictor variable ‘diversity’ is increased by one unit, then the model predicts that ROCE will decrease by 6.4 percent for every additional degree of diversity. However, ROCE will increase by 11.2 percent, if the firm is state-controlled. Moreover, in model 1 DV ‘ROCE’ the control variable age is negatively related and significant, while for model 2 and 3, firm size is positively related and significant. Again, the result on ROCE is the same as it was on profit when it comes to the direct effect of the predictor variables: diversity is

negative and significant, while the direct effect of state ownership is positive and significant. 78 79 80 81 82 83

Low diversity High diversity

P

rofit/

Loss befor

e tax [MM

USD]

(27)

The interaction effect of diversity and state influence on ROCE is negative and significant (p< .05). There is no relevant quadratic effect.

For the DV ‘ROA’: the change statistics show that the change in R square is

significant for model 2 (p<.001) and model 3 (p<.05). The change of R square from model 1 to model 2 is 2.2 percent, and this change in the amount of variance explained gives rise to an F-ratio of 6.523, which is significant with a probability less than .000. If the predictor

variable ‘diversity’ is increased by one unit, then the model predicts that ROA will decrease by 10.5 percent for every additional degree of diversity. However, ROA will increase by 9.1 percent, if the firm is state-controlled. Moreover, in model 1 to 3 of the DV ‘ROA’ the control variable location is negatively related and significant. Again, the result on ROA is the same as it was on profit and ROCE when it comes to the direct effect of the predictor variables: diversity is negative and significant (p< .001), while the direct effect of state ownership is positive and significant (p< .001). The interaction effect of diversity and state influence on ROA is negative and also significant (p< .05). There is no relevant quadratic effect.

For the DV ‘PE ratio’: the change statistics do not show a significant change in R from the models 1 to 3. Noteworthy is that the control variable “size” plays a role for the P/E ratio. Growing size affects the price-earnings ratio negative and significant with a probability <.05.

4.3. Second stage – Regression with partitioned data

The binary variable for ownership allows for easy partitioning of the data to take a closer look into the individual results of the groups ‘state-influenced’ and ‘private’ firms and their comparison. Again, model 1 includes the control variables, and now, model 2 is testing for the direct effect of diversity on the outcome variable, as well as diversity squared to test for quadratic effect. Table 8 provides the results for the ‘private’ group on all four dependent variables.

(28)

Table 8: Regression Results for private firms

Profits ROCE ROA P/E Ratio

M1 M2 M1 M2 M1 M2 M1 M2 Control variable Firm Size .421*** .420*** .063* .062* .001 .000 -.076* -.077* Location -.062** -.065** -.035 -.040 -.057* -.064* .032 .030 Age .024 .028 -.034 -.029 -.012 -.004 -.010 -.008 Innovation -.039 -.040 -.019 -.020 .012 .011 .022 .022 Industry .011 .009 -.002 -.004 -.017 -.022 -.019 -0.020 Z-Div -.043 -.051 -.094*** -.028 Z-Div² -.030 .132 -.085** -.028 Fitness indices R² .173 .175 .005 .008 .004 .012 .007 .008 Adj. R² .171 .172 .002 .004 .001 .009 .003 .003 F 65.760 55.455 1.520 1.848 1.181 3.276 1.812 1.682 df 1570 1569 1378 1377 1573 1572 1310 1309 Δ R² .002 .003 .009 .001 Δ F 3.422 3.474 13.701 1.030

Note 2: *p<.05; **p<.01; ***p<.001, one-tailed test. The entries are standardized coefficients (βs). Selecting only cases for which STATE_INFLUENCE =0.

The outcome for private firms on the dependent variables profits, ROCE and PE ratio do not show any effects from diversity. For the profitability, ROCE and the market value, the degree of diversification into other industries irrelevant. However, the outcome variable ROA shows the complete opposite. If the predictor variable ‘diversity’ is increased by one unit, then the model predicts that ROA will decrease by 9.4 percent for every additional degree of diversity. Diversity degree is negative and significant with a probability < .001. Furthermore, the control variables firm size and location effect the change in variance in profits and partly in ROCE, ROA and PE ratio. Firm size is positive and significant for the profitability

(p<.001) and return on capital employed (p<.05). However, the size of the firm effects the market value, when measures as PE ratio, negatively and significant (p<.076).

(29)

Table 9: Regression Results for state-controlled firms

Profits ROCE ROA P/E Ratio

M1 M2 M1 M2 M1 M2 M1 M2 Control variable Firm Size .646*** .639*** .017 .154 .003 -.008 -.123 -.066 Location -.040 -.091* .001 -.107 .072 -.014 -.036 -.018 Age .322*** .332*** -.220* -.117 -.230 -.212 .069 .046 Innovation -.062 -.063 -.177 -.249* -.157 -.160 .036 .058 Industry -.058 -.029 -.069 -.026 -.100 -.054 .020 -.074 Z-Div -.188*** -.403** -.313** .317* Z-Div² -.178*** -.370** -.297** .354** Fitness indices R² .801 .833 .088 .200 .107 .195 .008 .098 Adj. R² .790 .822 .035 .143 .059 .143 -.095 -.017 F 75.651 77.131 1.656 3.537 2.244 3.744 .076 .0853 df 94 93 86 85 94 93 48 47 Δ R² .032 .112 .088 .090 Δ F 17.626 11.891 10.151 4.706

Note 3:*p<.05; **p<.01; ***p<.001, one-tailed test. The entries are standardized coefficients (βs). Selecting only cases for which STATE_INFLUENCE =1.

In the regression for the state-influenced firms, less control variables can predict the profits/ losses after tax compared to the regression for all firms. For companies with state-control, only firm size and location account for 17.3 percent of the variation of profits. Model 2 shows that there is a direct effect of diversity, as well as a quadratic effect of diversity squared, which is significant with a probability of less than .05 percent. Following figure illustrates the effect.

Figure 3: Private firms/ curvilinear effect based on the regression equation 70 71 72 73 74 75

Low Diversification High Diversification

Profi ts before tax [MM U SD ]

(30)

Thus, the question arises whether the same curvilinear effect can be seen for the private firms. The following table provides the results of the second regression of stage two. Again, there are different results from the control variables. It becomes clear that for private firms, the firm size and age predicts the profits/ losses to a certain extent. However, the independent variable ‘diversity degree’ does not change anything in the profits and is not significant.

5. Discussion and conclusion

In hypothesis 1, I assumed an inverted-U shaped relationship between the predictor variables diversification and the firm performance, such that an increasing diversification in unrelated business is associated with increasing performance at lower levels of diversification and decreasing performance at high diversification. Hypothesis 1 is rejected, because the direct effects from diversification could not confirm the inverted U-shape model. In contrast to this, the effects show a convex trend that emphasizes that no or only little diversity reaches high performance levels in terms of profitability and operating returns (ROCE and ROA), while some diversity will negatively affect these measures. The effect turns around out at high levels of diversity and becomes slightly positive.

In hypothesis 2, I presumed that the curvilinear relationship between diversification and performance is negatively influenced by government influence in such way that it attenuates the positive effect at lower degree of diversification and amplifies the negative effect at higher degree of diversification. Hypothesis 2 is confirmed, because the interaction effect confirms the moderation effect and it also confirms that it is positive to be state-influenced at low levels of diversity as it enhances performance, but also amplifies the negative effect at high level of diversity (illustrated in Figure 2).

(31)

5.1. Theoretical and practical implications

This study adds to the theory on diversification performance relationship by

examining possible moderator affecting the relationship. My findings are contradicting to the Palich, Cardinal & Miller (2000) statement that some diversification effects performance positively, but only up to a certain point, after which, as diversification continues to increase, the performance drops. This research paper proves, if diversification increases, performance decreases instantly up to a certain point after which both diversification and performance increase together. This U-shaped curvilinear relationship can be observed for state-affiliated and private firms in China.

But it becomes evident that there is a main difference between private and state-influenced firms at low levels of diversity. For the ‘state’ group, the ownership influence has a and enhancing effect of the negative moderation. State-affiliated firms do not seem to benefit from their government linkages, instead, the ‘state’ group appears to be less

committed to an economic logic than private firms (Li & Xie, 2013). It appears that the effect is less strong for the private firms. Overall, it could be proven that state ownership plays a significant role in the relationship. When adding other control variables, it is evident that for private firms the location and firm size plays a role. It appears to have a positive effect, when a private firm grows numbers of employees and further, it has a negative effect, if the firm is located outside the top economic cities in China, which are Beijing, Shanghai and Shenzhen. In contrast to this, the location factor is not relevant for state-influenced firms. Their linkage to government and access to funds seems to be unaffected regardless of the networking environment that a major business city offers. However, state-affiliated firms respond

strongly to the control variable age and firm size. Hence, with growing experience in number of years and growing number of employees, a state-affiliated firm can reach higher profits. This may also result from their government linkage, because politics responds to well

(32)

established firms better, because of long-standing relationships and the common interest to grow the number of employed people. Since market dynamics and government intervention affect a firm’s strategic choices, the study suggests that acquisitions by Chinese enterprises are motivated to secure a stable market position through size and influence rather than building financially steady enterprises and allowing competition.

My findings also suggest that managers who do business with investors from emerging economies should bear in mind that state ownership may influence the investors’ strategic decisions. Further, the state-controlled Chinese firms are becoming more and more alike to their counterparts that are privately owned in terms of their performance.

5.2. Limitations

Currently, the data cannot fulfill the lack effect. To consider time deferred results in performance, it would be necessary to collect data that contains the information of

diversification from at least one, ideally two years ago. This provides more reliable

conclusion on the performance of a firm after changes in diversification. The dataset of 2,845 companies shows minor deficiencies, because of incompleteness of data fields. The

information on ownership structures and share is incomplete and lacks transparency. It appears that the number of private firms is significantly higher than the number of state-influenced firms for this sample.

5.3. Further research

The moderating variable for governmental control can be measured in different ways and once can argue that state ownership does not reflect the political attachment and therefore not the decision-making process of a firm when it comes to its diversification strategy, the information on their ownership structures is complex and not transparent. In many cases, ownership according to official sources of Chinese financial websites, show private stakeholder. However, some sources state that public entities are behind these private

(33)

stakeholders. For this reason, further research on each company may be done to verify the official sources and gain further information on the linkage to the politburo. Even though this is also not completely accurate, it can provide a better understanding of how strategic

decisions are influenced. Regarding the performance variable, market measures may be interesting to diversification research since these capture expectations of future returns from firm performance as opposed to past outcomes reflected in accounting-based measures.

(34)

References

Anderson, K., & Brooks, C. (2006). Decomposing the price-earnings ratio. Journal of Asset Management, 456–469.

Baron, R. M., & Kenny, D. (1986). The Moderator-Mediator Variable Distinction in Social Psychology Research. Journal of Personality and Social Psychology, 1173-1182. Berk, J., & DeMarzo, P. (2017). Corporate Finance. Harlow: Pearson Education Limited. Cendrowski, S. (2015, July 22). Fortune International. Retrieved from fortune.com:

http://fortune.com/2015/07/22/china-global-500-government-owned/

Fernández, Z., & Nieto, M. J. (2006). Impact of Ownership on the International Involvement of SMEs. Journal of International Business Studies, 340-351.

Fickling, D. (2017, June 06). Bloomberg Gadfly. Retrieved from www.bloomberg.com: https://www.bloomberg.com/gadfly/articles/2017-06-06/chinese-cash-magnet-hna-must-clear-the-murk-to-reach-its-goals

FT. (2017, June 02). FT.com. Retrieved from Financial Times:

https://www.ft.com/content/8acfe40e-410b-11e7-9d56-25f963e998b2

Hess, K., Gunasekarage, A., & Hovey, M. (2010). State-dominant and non-state dominant ownership concentration and firm performance: evidence from China. International Journal of Managerial Finance, 264-289.

Li, H., & Zhang, Y. (2007). The role of managers' political networking and functional experience in new venture performance: evidence from China's transition economy. Strategic Management Journal, 791-807.

Li, J., & Xie, Z. (2013). Examining the Cross-Border Acquisition Strategy of Chinese Companies: The Moderating Roles of State Ownership and Institutional Transition. Journal of Leadership & Organizational Studies, 436-447.

(35)

Luo, X., Zhou, L., & Liu, S. (2005). Entrepreneurial firms in the context of China's transition economy: an integrative frameork and empirical examination. Journal of Business Research, 277-284.

Markides, C. C., & Williamson, P. (1994). Related diversification, core competencies and corporate performance. Strategic Management Journal, 149-165.

McCutcheon, B. J. (1991). What caused conglomerate formation? An examination of managerial and internal capital market in the 1960s conglomerates. Doctoral dissertation.

Myers, R. (1990). Classical and modern regression with applications (2nd. ed.). Boston: Duxbury.

Ng, A. Y. (2009). Determinants of state equity ownership, and its effect on value/ performance: China's privatized firms. Pacific-Basin Finance Journal, 413-443. OECD. (2016). Broadening the Ownership of State-Owned Enterprises - A Comparision of

Governance Practices. Paris, France.

Palich, L. E., Cardinal, L. B., & Miller, C. C. (2000). Curvilinearity in the Diversification-Performance Linkage: An Examination of over Three Decades of Research. Strategic Management Journal, 155-174.

Qi, D., Wu, W., & Zhang, H. (2000). Shareholding structure and corporate performance of partially privatized firms: evidence from listed Chinese companies. Pacific-Basin Finance, 587-610.

Reddy, K., Xie, E., & Huang, Y. (2016). Cross-border acquisitions by state-owned and private enterprises: A perspective from emerging economies. Journal of Policy Modeling 38, 1147–1170.

reuters.com. (2018, March 20). Retrieved from

(36)

South China Morning Post. (2017, July 13). Retrieved from http://www.scmp.com: http://www.scmp.com/business/companies/article/2102540/behind-chinas-boldest-dealmaker-hna-are-billions-pledged-shares

Sun, Q., Tong, W., & Tong, J. (2002). How does government ownership affect firm performance? Journal of Business Finance and Accounting, 1-27.

Szamosszegi, A., & Kyle, C. (2011, October 26). U.S. - China Economic and Security Review Commission . Retrieved from uscc: https://www.uscc.gov/

Telecomlead. (2018, March 15). Retrieved from www.telecomlead.com:

http://www.telecomlead.com/telecom-equipment/how-zte-performed-in-2017-82965 Thomsen, S., & Pedersen, T. (2000). Ownership structure and economic performance in the

largest Eurpean companies. Strategic Management Journal, 689-705.

UvA, D. (2018, February 10). Databases UvA. Retrieved from Databases UvA Web site: https://databases.uba.uva.nl/en#labelO

Zacharias, J., & Yang, W. (2016). A short history of the Chinese Central Business District. Planning Perspectives, 611-633.

(37)

Attachments

Table 10: Industry classification

BvD Main Sector #Companie

s

Total Turnover [MM USD, 2016] * MACHINERY, APPLIANCES, FURNITURE,

RECYCLING

913 394,024

CHEMISTRY, RUBBER, SYNTHETICS, NON-METAL PRODUCTS

565 271,183

OTHER SERVICES 357 146,638

WHOLESALE & RETAIL 179 159,655

METAL & METAL PRODUCTS 176 130,462

TEXTILE, APPAREL, LEATHER 111 78,752

FOOD, BEVERAGES, TABACCO 102 56,513

PRIMARY SECTOR (FORESTRY, MINING, FISHING, AGRICULTURE)

81 33,146

CONSTRUCTION 62 70,867

GAS, WATER, ELECTRICITY 59 37,070

TRANSPORTATION 59 67,912

PUBLISHING & PRINTED MATERIAL 49 13,761

WOOD, CORK, PAPER 48 17,238

BANKS 28 158,572

POST & TELECOMMUNICATION 23 8,683

HOTELS & RESTAURANTS 18 1,792

EDUCATION & HEALTH 9 964

INSURANCE 5 84,435

Referenties

GERELATEERDE DOCUMENTEN

Maar daardoor weten ze vaak niet goed wat de software doet, kunnen deze niet wijzigen en ook niet voorspel- len hoe de software samenwerkt met andere auto-software. Laten we

The Royal Commission on Aboriginal People (RCAP) report; the Indian Residential Schools Settlement Agreement (IRSSA) made by residential school survivors against the government

The main findings of this study is that the asset tangibility, firm size, and future growth opportunities have significant and positive relationship with the

The main contribution is twofold: first, investigate the relationship between multiple dimensions of CSR and financial performance in the automotive industry; second,

As the results show mixed results with different environmental performance measurements, it implies that only some aspects (underlying variables) of the environmental

The present research aimed to show that changes in company performance are important antecedents of CEOs’ positive and negative emotion expressions, as found in

For example, from Ren and Liu (2005)’s study, even though there are different cultural backgrounds (from Table 1, the quite different cultural scores in collectivism/

It can be used as, a legitimacy tool, a means to influence people’s perceptions about a firm, an outcome and part of reputation risk management processes, a means that