Government Subsidisation and Shareholder Wealth Impact: Evidence from Malaysia

: This paper investigates the shareholder wealth impact of government investment in listed companies (and by extension, government subsidisation of those companies), using data from Malaysia. We distinguish two overlapping categories of government-related investors: those whose principal mission relates to economic policy and those whose principal mission relates to social policy. The methodology entails Ordinary Least Squares regressions. There are two dependent variables measuring management success at generating shareholder wealth: an intrinsic value surrogate and return on equity. The ﬁnal sample comprises 1732 company–year observations from the investigation period 2011–2014. The evidence indicates that companies subject to shareholder by a government-related investor with a social (economic) policy mission are more (less) successful at generating wealth than companies without any government shareholding at all. The ﬁndings indicate that for companies subject to ownership by government investors with a mission related to economic policy, government subsidies are wealth-enhancing, subject to diminishing marginal returns beyond a threshold level of government shareholding. The research design reﬂects adaptations to the Malaysian institutional setting via choice of control variables and usage of data from a leading Malaysian equity analyst.


Introduction
The purpose of this study is to investigate the shareholder wealth impact of government subsidisation of listed companies. Malaysia is a suitable setting for investigating this research question. The Malaysian Government has combined two different ideologies in formulating policies for economic growth and nation-building: laissez-faire economics and the developmental state. These policies have been directed towards two principal goals: eliminating the association between ethnicity and economic function and increasing overall national wealth. These policies have been implemented via close co-operation between the public and private sectors, characterised by government shareholding in and subsidisation of listed companies (Gomez 2009).
These government-related investors share the over-arching objective of facilitating the economic and social development of Malaysia. However, the investors are somewhat heterogeneous with respect to policy objectives, portfolio management styles and degree of share ownership in listed companies (Gomez and Micheaux 2017;Gomez et al. 2018). It follows that in Malaysia, listed companies vary cross-sectionally and temporally according to the extent to which they are subject to government ownership, type of government ownership and degree and nature of government subsidisation. Hence, our paper contributes to the extant literature on shareholder wealth impacts of government subsidisation (Chen and Wang 2004;Lee et al. 2014).
The results indicate that the shareholder wealth impact of being subject to government investment (and thus being subsidised) depends on the type of government-related institutional investor. Our findings suggests that companies subject to share ownership by a government investor with a social (economic) policy mission are more (less) successful at generating shareholder wealth than companies without any government shareholding at all. For companies subject to ownership by a government investor with an economic policy mission, shareholder wealth is positively associated with the degree of share ownership by these investors (and hence the extent to which the investee received government subsidisation and government input into their decision-making). However, for levels of government shareholding above a certain threshold, this effect experiences diminishing marginal returns. Considered jointly, the findings of this paper indicate that aggregate evidence that government ownership erodes shareholder wealth (Wang and Shailer 2018) may not hold universally, for at least two reasons. Firstly, the sign of association may differ according to the type of government-related shareholding. Secondly, the relation may be non-linear.
There are several novel features of this study. Most importantly, to the authors' knowledge, this is the first empirical paper to closely use the Boycko et al. (1996) model, as a lens for understanding the performance of entities subject to government input. Our research design reflects adaptations to the Malaysian setting. Some of our regression variables are constructed using data from Dynaquest Sendirian Berhad, a firm of Malaysian equity analysts. One of our control variables, the flag for the Nanyang company (an ethnic Chinese family company) status, reflects the prevalence of this type of family company in Malaysia (Sinnadurai 2018). A unique feature of our sample selection is examination of companies subject to government shareholding, in the period commencing three years after floating.
The remainder of the paper is structured as follows. Section 2 locates the study within the extant literature. Section 3 presents our regime for classifying governmentrelated institutional investors in Malaysia. Section 4 exposes the underlying theory and develops the research hypotheses. Section 5 explains the research methodology and data collection. Section 6 explains the sample selection. Sections 7 and 8, respectively, present the descriptive statistics and empirical results. Section 9 concludes.

Literature Review
This study informs the debate regarding the desirability of government investment in the private sector. Compelling arguments have been advanced to support the view that government share ownership of private sector companies results in erosion of shareholder wealth. Government share ownership means that listed companies are involved in implementing public policy. Corporate policy choices that maximize shareholder wealth may sometimes conflict with choices that achieve public policy objectives (Boycko et al. 1996). Furthermore, government share ownership may result in public sector officials, lacking business qualifications and acumen, becoming directors and executives of listed companies (Wang 2005). The "grabbing hand" theory postulates that government-related investors may also engage in wealth expropriation for political expediency. (e.g., they may divert corporate resources towards funding their election campaigns (Boubakri et al. 2020;Yu and Wang 2020)).
Conversely, sound arguments have been advanced, suggesting that government share ownership of private sector companies results in shareholder wealth enhancement. Government-related investors can facilitate access to more financial resources and capital investment projects (Mok and Hui 1998;Paudyal et al. 1998). This may reduce the need for the companies to hold liquid assets to service transactions, thereby enabling the companies to hold other assets with lower opportunity costs (Yu and Wang 2020). Government agents, represented on boards and executive teams of listed companies, may have enhanced authority to implement governance mechanisms to protect shareholders from wealth expropriation by management (Ang and Ding 2006). These advantages may be particularly pertinent for companies operating in under-developed regions, for which the government is implementing focused economic development programs (Yu and Wang 2020).
One stream of empirical studies, investigating which of these two viewpoints is stronger, has compared initial underpricing and long-run performance of privatization versus private sector Initial Public Offerings (IPOs). On balance, empirical evidence from the People's Republic of China, tends to suggest that statement ownership is either unrelated to privatisation IPO underpricing and performance, or that these variables are related inversely (Mok and Hui 1998;Qi et al. 2000;Wang 2005). Conversely, evidence suggests that Malaysian privatisation IPOs experience productivity and profitability increases after floating and do not suffer long-run underperformance (Paudyal et al. 1998). Differences in evidence between the People's Republic of China and Malaysia may reflect differences in the state ideologies underpinning the privatisations. The Chinese privatisations were part of a program of transitioning towards a market economy in a Communist state (Qi et al. 2000). Malaysian privatisations were conducted with the State objectives of the New Economic Policy (Gomez et al. 2018). All of these studies may have been marred by IPO-related phenomena (temporal clustering, initial underpricing and long-term underperfomance). Our study circumvents IPO anomalies via investigating performance in the period commencing (at least) three years after the float.
Other studies have also investigated the impact of government ownership on corporate performance, via lenses other than the share market aftermath of floating. For example, Amado et al. (2017) document evidence from Portugal that privatisation results in improved aggregate performance but did not result in improved production efficiency (a specific component of financial performance). Yu and Wang (2020), using data from the People's Republic of China, documented evidence that listed companies controlled by the government may have a lower value of cash, due to shareholder concern that government control may result in expropriation of shareholder wealth. Wang and Shailer (2018), in a meta-analysis, empirically evaluate the balance of evidence as suggesting that government ownership erodes, rather than enhances, corporate performance. We take the view that this aggregate evidence does not apply to all modes of government ownership, in Malaysia. Furthermore, Wang and Shailer (2018) suggest that further research could investigate this issue via modelling the dynamics of the performance-ownership relationship. Our study is a step towards implementing this suggestion by considering the Boycko et al. (1996) parameters.
A limitation of all of the aforementioned studies is their failure to acknowledge possible non-monotonicity in the relationship between shareholder wealth and state ownership. Boubakri et al. (2020) is a pioneering study that addresses this concern. Their evidence, using data from 53 different countries, is consistent with the relationship following an "inverted U" (i.e., at low levels of state ownership, this variable is positively associated with shareholder wealth. However, for levels of state ownership larger than a certain threshold, the costs outweigh the benefits and the relationship becomes negative). Our study builds on Boubakri et al. (2020) by using a theory that could identify variables that explain the location of this turning point.
The second stream of literature to which our paper contributes is literature on the antecedents and economic consequences of government subsidisation of private sector entities. The extant literature on this topic principally uses data from the People's Republic of China. This country requires government subsidies to be treated as a component of earnings and disclosed as such in the financial statements. Hence, the methodology of Easton and Harris (1991) and Ohlson (1995) has been used to investigate the value relevance of the subsidies. Evidence suggests that the share market regards government subsidies as a value-relevant component of earnings and that the value relevance of different types of subsidies differs systematically, according to firm characteristics and the nature of the subsidies. (Chen and Wang 2004;Lee et al. 2014). Yu and Wang (2020) document evidence that government subsidisation is value-enhancing, for companies operating in lesser developed regions of China, subject to special government economic development packages. The evidence also indicates subsidised companies engage in less upwards accruals-based earnings management, resulting in enhanced quality of earnings (He 2016). Unlike Chen and Wang (2004) and Lee et al. (2014), our study is not limited to observable and quantified subsidies reported in the financial statements.

Regime for Classifying Types of Government-Related Institutional Investors in Malaysia
Ownership of corporate Malaysia is dominated by Government-Linked Investment Companies (GLICs). These are investment companies linked to the Federal Government, that own shares in listed companies, for the purpose of implementing public policy. They also invest in several dozen non-listed companies and are significant investors in government and corporate bonds and property. The Federal Government is responsible for appointing members of the board and executives of the GLICs. There are seven GLICs in Malaysia; each with a different public mandate. Government-Linked Companies (GLCs) are the listed companies, subject to share ownership by the GLICs (Gomez et al. 2018, pp. 8-9). In the current study, the investors classified as "government-related institutional investors" principally (but not exclusively) comprise the GLICs. However, our classification regime does not treat the GLICs as one homogeneous category. Rather, each GLIC is allocated to one of two categories, based on the GLIC's policy mandate and investment style. Furthermore, both of our two categories include investors that are government-related but not GLICs.
Our classification regime recognises two overlapping categories of government-related institutional investors. The first type comprises companies subject to government shareholding because their output is compatible with economic policy, directed towards increasing the size of Gross Domestic Product (Fraser et al. 2006;Mitchell and Joseph 2010). The second category of government-related investors have social policy missions. "Social policy" is defined as policy directed towards re-distributing national wealth. The investors within this category typically have beneficiaries corresponding to sections of society targeted for financial assistance (Fraser et al. 2006).
Four sub-categories of government-related investors are classified as having an economic policy focus. The first sub-category consists of The Federal Ministry of Finance and entities under its budget jurisdiction. The Minister of Finance has the objective of increasing value and returns on Federal Government assets. The second sub-category comprises Khazanah Nasional Berhad, the investment arm of the Ministry of Finance and its controlled entities. The Federal Ministry of Finance and Khazanah Nasional Berhad are both GLICs. Both of these investors may obtain funds from government taxes and directly from the securities markets (Gomez et al. 2018, p. 65). The third sub-category of government investors, classified as having an economic policy focus, are the Development Financial Institutions. These are small government-owned financial institutions, regulated via statute. Specific socioeconomic functions are specified for each Development Financial Institution. There are seven Development Financial Institutions; most of them are focused on a particular economic sector. The sectors captured by these institutions include local manufacturing, agriculture, infrastructure, the maritime industry, technology and oil and gas, small savers, export-oriented industries and small and medium enterprises (Gomez et al. 2018, pp. 114-18). The final sub-category of government investors, classified as having an economic policy focus, are the State Economic Development Corporations. These are institutions, established in the individual states of Malaysia which function to promote economic development and Malay welfare within the particular state. 1 The State Economic Development Corporations achieve their objective via substantial shareholding in listed companies within the relevant state (Gomez et al. 2018, p. 98).
We recognise two sub-categories of government-related investors with a social policy focus. The first sub-category comprises the remaining five GLICs (and their related parties), not classified as having an economic policy focus. Unlike the Federal Ministry of Finance and Khazanah Nasional Berhad, the other five GLICs obtain funds directly from the investing public. Their policy mandates include retirement savings, supporting Muslims in undertaking their pilgrimage to Mecca and increasing the equity ownership of bumiputera (Malays and other indigenous Malaysians) (Gomez et al. 2018, p. 65). The second sub-category comprises investors associated with political parties. There are two principal investors in this sub-category. The first is United Malays National Organisation, a constituent of the Barisan Nasional ("National Alliance") party that has ruled Malaysia for most of its history (and the entirety of our investigation period). The other investor is Huaren Management Sendirian Berhad ("Ethnic Chinese Management Pty Ltd."). This is the investment arm of the Malaysian Chinese Association, another principal constituent of Barisan Nasional. Share ownership by these two investors is concentrated in the media sector (Gomez et al. 2018, p. 119). Empirical evidence from the era after these regulatory changes suggests that the regulations have been at least partially successful. Evidence indicates that Malaysian companies subject to substantial shareholding by a government investor with an economic policy mission enjoy higher implied abnormal earnings growth rates than other companies (Sinnadurai 2016). Both the GLICs and the Development Financial Institutions survived the Asian Financial Crises relatively undamaged. Evidence indicates that these investors implemented sound governance practices within the GLCs (and presumably other companies subject to government shareholding). The government bailout packages tended to be issued to private entrepreneurs and banks (Gomez et al. 2018, p. 116).
To the extent that these reforms has been successful, they would reduce the strength of mechanisms auguring for government shareholding, eroding (rather than enhancing) shareholder wealth. For example, if the boards of directors on GLCs increased their efficiency (following the GLC Transformation Program), there would be reduced scope for staff lacking business qualifications and experience to obtain senior management positions (Wang 2005). Furthermore, if corruption was successfully combatted, there would be reduced potential for the government to exercise a "grabbing hand", to expropriate wealth from other shareholders (Boubakri et al. 2020;Yu and Wang 2020).
We acknowledge that there is an element of the "grabbing hand" in Malaysia following these reforms. There have been some high-profile corporate scandals, including the Port Klang Free Zone Scandal, the National Feedlot Corporation Scandal and the 1Malaysia Development Berhad scandal. However, there has been a degree of accountability following these scandals. For example, the Port Klang Free Zone scandal was followed by a joint investigation conducted by the police and Malaysian Anti-Corruption Commission. A Federal audit, following the National Feedlot Corporation Scandal, exposed several business irregularities. There was high-level public dissatisfaction with the handling of the 1Malaysian Development Berhad scandal by the Najib Government. This was a key factor in why the coalition that had ruled Malaysia since independence, Barsian Nasional, was ousted in the fourteenth General Election in 2018 (Nadzri 2018). A possible factor limiting the extent to which the reforms have been successful may be that in Malaysia, regulators are partially "captured" by powerful businessmen and politicians (Vithiatharan and Gomez 2014). However, we maintain our position that any degree of accountability would reduce the extent to which the "grabbing hand" could function.
The investigation period of the current study coincides with the eras during and after these reforms. Hence, during our investigation period, mechanisms auguring for a positive association between shareholder wealth and government share ownership of listed companies are likely to be stronger than counter-mechanisms. In particular, the enhanced accountability would reduce scope for the government shareholders to engage in expropriation of wealth from the private sector shareholders of the GLCs and other companies subject to government shareholding (Boubakri et al. 2020;Yu and Wang 2020).
The first research hypothesis follows.
Hypothesis 1 (H1). In Malaysia, during our investigation period, companies subject to shareholding by (either type of) government-related investor are more successful at generating shareholder wealth than other companies.

Association between Shareholder Wealth and the Level of Shareholding by a Government-Related Investor with an Economic Policy Mission
There are at least two other reasons why the aggregate evidence that government share ownership erodes shareholder wealth in developing countries (Wang and Shailer 2018), might not generalise to Malaysia. Firstly, the relationship may be non-monotonic (Boubakri et al. 2020). Secondly, the relation may differ among types of governmentrelated investors. Our second hypothesis acknowledges these possibilities by isolating one category of government shareholding and considering the level of government ownership.
The second hypothesis is developed by applying the Boycko et al. (1996) model to Malaysia. The theory models the employment decision within a privatised statutory body. The theory assumes the statutory body has two shareholders: the manager, owning α of the shares, and the politician, owning the remaining (1 − α). The manager seeks to maximise shareholder wealth. The politician is an agent of the electorate, seeking to implement public policy. Both shareholders act to maximise the utility of their personal wealth. It is assumed that the politician pays a subsidy to the company, with a view to align corporate and public policy. If these two goals were successfully aligned, the consequence of privatisation improving profitability at the expense of job losses (Amado et al. 2017) would not be inevitable.
The Boycko et al. (1996) model accurately characterises Malaysian companies subject to shareholding by a government investor with a mission related to economic policy. Consistent with the Boycko et al. (1996), model, government investors with missions related to economic policy tend to be long-term shareholders, implementing public policy via governance of the company. 3 Companies with shareholding by government investors with missions related to economic policy also face a conflict of goals, responsible for both maximising shareholder wealth and achieving public policy goals. Conversely, government investors with social policy missions that are mostly public sector investment funds actively managed to maximise portfolio returns (Gomez et al. 2018, pp. 63-78). Hence, management of companies with this type of government ownership may not face the same degree of goal conflict.
The Boycko et al. (1996) model assumes that expenditure on employment within the statutory body, the variable,Ȇ, has two states, H (high) and L (low). If the decision is made to selectȆ = H, the public policy objectives are satisfied. The model analyses the situation in which it is possible for the politician to pay a subsidy of t to the privatised company, to entice the manager to chooseȆ = H. The two investors' utility functions are represented via Equations (1) and (2). where: U (politician) = wealth utility of the politician. U (manager) = wealth utility of the manager. q = the marginal direct benefit to the politician of spending one dollar on labour (q < 1). m = marginal cost to politician of spending one dollar on labour (e.g., lower income tax revenue) (m < 1). k = the marginal cost to the politician of making the subsidy (k < 1). α = the portion of the entity's shares owned by private sector shareholders. Hence, (1 − α) = the portion of the entity's shares owned by the politician. t = the amount of the subsidy paid to the company.
Equation (1) identifies several parameters affecting the politician's wealth. The first parameter, q, represents the marginal benefit of spending one extra dollar on labour. Components include obtaining more votes and less pressure from trade unions. The second parameter, m, captures the marginal cost of spending an extra dollar on labour, including reduced fiscal revenue from the income tax paid by the statutory body. (The employment costs would be an allowable deduction for the statutory body.) The impact of m on the politician's wealth is ameliorated by the fact that they are not the only shareholder. Equation (1) also shows a politician's wealth utility is diminished by payment of the subsidy. If they pay a subsidy of t, they have a claim over a portion of this subsidy, (1 − α) × t, as a shareholder. Hence, their net "out-of-pocket" is α × t. k is the marginal cost per dollar of subsidy spent, including transactions costs from issuing government securities to fund the subsidy and loss of wealth arising from voters being disenchanted at partly bearing the cost of the subsidy.
Equation (2) reveals that there are two components of the manager's wealth utility. The manager's wealth is negatively related toȆ. As a shareholder, the manager partially bears the reduction in shareholder wealth from employee expenses. However, the manager's wealth is increased by α × t, their claim over the subsidy, as a shareholder. The change in utilities of the politician and manager, from making the switch from choosingȆ = L to choosingȆ = H, may be obtained by substitutingȆ = H − L into Equations (1) and (2), respectively, generating Equations (3) and (4).
From economic theory, the equilibrium t, t*, occurring when the product of Equations (3) and (4) is maximised over t. It can be shown that It would be optimal to subsidise when t* > 0. This occurs when Inequality (6) is satisfied. q Inequality (6) represents the condition for it to be optimal to switch from choosing E = H toȆ = L, where the politician pays a subsidy. q, the incremental wealth accruing to politicians from implementing public policy imperatives, would be cross-sectionally high in Malaysia. It is very common for Malaysian politicians to serve as independent directors on boards of listed companies during their careers and after retirement from politics (Gomez et al. 2018, pp. 190, 194, 203). Their remuneration from these activities would be a principal component of q. m would also be cross-sectionally high in Malaysia. Corporate decisions that achieve public policy objectives, at the expense of shareholder wealth maximisation, are likely to entail a large opportunity cost, due to the statutory body forgoing investment options that maximise shareholder wealth.
In Malaysia, there are at least two mechanisms affecting k, the cost to the politician of making the subsidy. Both of these mechanisms are indirect. Firstly, the subsidy would result in a wealth transfer to the target members(s) of society from other members of society. The politician, as a member of either or both sections of society, would partially bear this wealth transfer. In cases where the politician belongs to the section of society making the transfer, this component of k would be positive. In cases where the politician belongs to the section of society receiving the wealth transfer, this component of k would be negative. The latter scenario frequently prevails in Malaysia and hence contributes to k being lower than q and m. Political costs would also affect k in Malaysia. Subsidisation of listed companies is frequently accompanied by public criticism (Lai 2012), potentially leading to the politician having lower chances of being elected and hence a lower expected present value of the remuneration.
Further application of Boycko et al. (1996) demonstrates the following.
where MC (politician) = the marginal cost to the politician from making a corporate investment inconsistent with shareholder maximisation and paying the subsidy Since (m − k) would be positive in Malaysia, an increase in the level of government shareholding would always increase the marginal cost to the politician of subsiding listed companies. There would be a threshold level of (1 − α) which, if exceeded, would result in the marginal cost to the politician exceeding the marginal benefit, causing Inequality (6) to no longer be satisfied. For levels of (1 − α) in this range, shareholder wealth would be maximised by choosingĚ = L, the choice that does not achieve the public policy objective. However, since management is compelled to chooseĚ = H, due to the input of the government-related investor, shareholders suffer wealth deterioration. This second research hypothesis follows.
Hypothesis 2 (H2). In Malaysia, for companies subject to shareholding by government investors with a mission related to economic policy, the association between shareholder wealth and the level of government share follows an "inverted U".

Basic Model for Testing the Research Hypotheses
Equations (8) and (9) present the basic models for testing the H1.
where all variables are defined in the Appendix A. β i , β* i , δ j and δ* j are regression parameters. ε i,t and ε* i,t are stochastic disturbance terms.
Equations (8) and (9) are estimated using the full sample of observations. Equation (9) recognises that the association between shareholder wealth and receipt of government subsidies may differ between our two categories of government-related investors. Equation (10) presents the basic model for testing H2, estimated using the sub-sample of observations subject to shareholding by government investors with a mission related to economic policy.
where all other variables are defined in the Appendix A. β' 0 , β' 1,j , β' 2,j and δ' j are regression parameters. ε' i,t is a stochastic disturbance term. Equations (8)-(10) address potential endogeneity between shareholder wealth and government investment. The dependent variables are observed in the year following the year when the independent variables of interest are observed. The control variables are observed in the same year as the independent variables of interest, except as noted otherwise.

Dependent Variables
Two indicators of shareholder wealth are employed. The first, DQ-Strength and Growth t+1 , is an intrinsic value surrogate. Our metric was calculated using stock ratings awarded to Malaysian listed companies by Dynaquest. DQ-Strength and Growth t+1 may be regarded as a reliable proxy, owing to the standing of the company within Malaysia. Dynaquest is the largest independent firm of investment consultants in Malaysia and the first investment house to be licensed as a corporate investment adviser (Dynaquest Sendirian Berhad 2021). Each stock rating is a number on a scale of zero to 12, with possible scores incrementing by 0.5. Naturally, higher ratings indicate more favourable assessments. Each rating is an aggregate of four sub-ratings of the following stock quality dimensions: dividend and earnings stability, financial strength, management strength and future growth of dividends and earnings. Each sub-rating is out of three. DQ-Strength and Growth t+1 is the sum of the sub-ratings for financial strength, management strength and future growth of earnings and dividends. The sub-ratings were obtained from Dynaquest. The second indicator of shareholder wealth is return on equity, calculated by Dynaquest. Observations were sourced from Dynaquest Sendirian Berhad (2012)(2013)(2014)(2015). Previous Malaysian studies have used different shareholder wealth metrics, including Tobin's Q (Lau and Tong 2008), return on assets and share market returns (Johnson and Mitton 2003;Mitchell and Joseph 2010) and cost of equity capital (Boubakri et al. 2012).
The use of an intrinsic value surrogate as one of our dependent variables contributes to addressing concerns about potential endogeneity. Equity value is the discounted sum of expected future earnings (Ohlson 2005). Hence, although government-related investors may choose to invest in companies because of their sound earning prospects, higher intrinsic value would be due to the share market anticipating higher future earnings, as a consequence of the government's decision.

Independent Variables of Interest
Companies subject to government share ownership, the relevant investors and their shareholdings were identified from Dynaquest Sendirian Berhad (2011)(2012)(2013)(2014). H1 would be supported if the coefficient of GovSHAny t is positive and significant in Equation (8) and at least one of (GovSHEcon t , GovSHSoc t ) in Equation (9) is positive and significant. Data for GovSHEcon t and GovSHSoc t were sourced from Dynaquest Sendirian Berhad (2011)(2012)(2013)(2014)(2015).
(1 − α) t represents the percentage of equity owned by government-related investors with an economic policy mission. The interaction terms in Equation (10), (1 − α) t × Q j,i,t (j = 2, 3, 4), permit the linear association between shareholder wealth and the level of government shareholding, for companies subject to shareholding by government investors with a mission related to economic policy, to vary among quartiles of (1 − α) t . The following three scenarios, if jointly upheld, would support H2. The first situation is a positive coefficient attaching to (1 − α) t , indicating that within the lowest quartile of government share ownership, there is a positive association between value and government shareholding. The second requirement is the coefficients of (1 − α) t and Q 3 × (1 − α) t summing to less than zero, indicating that in the third quartile of government share ownership, there is a negative association between value and government share ownership.
The third requirement, similar to the second, is the coefficients of (1 − α) t and Q 4 × (1 − α) t summing to less than zero, indicating that in the fourth quartile of government share ownership, there is a negative association between value and government share ownership.

Control Variables
We include dummy flagging modes of government influence, other than via government shareholding. For some companies, substantial shareholders, members of the executive and/or directors have a link (via personal association) to a current or former politician or political party (Faccio 2006). These political connections guide the companies to develop business models that are consistent with public policy objectives. During periods of economic downturn, Malaysian companies with informal political connections experience higher audit quality (Gul 2006). The investigation period of the current paper falls during the global financial crisis (GFC). Hence, companies with informal political connections would produce higher-quality financial statements. Monitoring mechanisms using these disclosures would be more effective. Similar to companies subject to government share ownership, companies with an informal political connection have access to more sources of debt finance than non-politically connect companies (Fraser et al. 2006;Johnson and Mitton 2003). Prior evidence supports these arguments, suggesting that Malaysian companies with informal political connections enjoy higher market-to-book ratios (Fraser et al. 2006), higher share market returns (Faccio 2006) and lower required returns on equity (Boubakri et al. 2012). 4 Hence, we anticipate a positive coefficient attaching to LSInfPC t .
To collect data for LSInfPC t , we identified companies with an informal political connection in 1997/1998 from the Appendix A of Fung et al. (2015). For each of these companies represented in our sample, we perused the directors' and managers' biographies in the relevant annual reports. If the directors, owners and/or managers still had a connection, as at the relevant balance date, we classified the company as having a longstanding informal political connection.
We also control for Government-Linked Corporation (GLC) status. Naturally, these companies would also be flagged by the independent variables of interest, identifying observations subject to government share ownership. However, inclusion of GLC is still appropriate; companies subject to government shareholding by non-GLICs would not have GLC status. Naturally, we anticipate a positive coefficient. GLCs were identified from Table  2.7 of Gomez et al. (2018). Since a company's GLC status is unlikely to change, we classify companies from this list as being GLCs for every year of representation in our sample.
Equations (8)-(10) control for family company status. Family companies are subject to the agency relationship between controlling shareholder-managers and non-controlling shareholders in addition to the relationship between managers and shareholders. These relationships are, respectively, referred to as "Type 2" and "Type 1" agency relationships of equity (Ali et al. 2007). Evidence from the United States (Wang 2006;Ali et al. 2007) and Malaysia (Amran and Che Ahmad 2009;Amran 2011;Abdullah et al. 2015) indicates that governance outputs differ between family and non-family companies. This suggests that the magnitude of Type 1 versus Type 2 agency costs and the efficacy of agency mechanisms differs between family and non-family companies.
The relative magnitudes of these two agency costs of equity may also differ among types of family companies. Findings from the United States (Pérez-González 2006;Chen et al. 2013) and Thailand (Bertrand et al. 2008) highlight two facets of family companies that may account for this cross-sectional variation: founding Chief Executive Officer (CEO) status and the extent to which professional managers are employed. Hence, we control for both management strength and founding CEO status. We anticipate a positive coefficient attaching to the proxy for management strength. The aforementioned evidence suggests that the association between shareholder wealth-generating success and founding CEO status depends on other variables, generating unclear expectations regarding the sign of the coefficient. Data for founding CEO status were sourced from the company's websites.
Equations (8)-(10) control for Nanyang company status. Nanyang companies, a type of family company prevalent throughout the Association of Southeast Asian Nations (ASEAN) region, are Chinese family companies, managed according to Confucian tenets (Sinnadurai 2018). Nanyang companies have been extremely successful in maximising shareholder wealth. A positive coefficient is anticipated. The first criterion for being a Nanyang company is having an ethnic Chinese CEO. The second criterion is that the CEO's family were substantial shareholders. Family ownership details were obtained from Dynaquest Sendirian Berhad (2011Berhad ( -2015 and annual reports sourced from Bursa Malaysia Berhad (2021). 5 If a company met these criteria in any year of the investigation period, it was classified as a Nanyang company for all the years when it was in the sample to acknowledge evidence that family company status is temporally stable (Claessens and Yurtoglu 2013).
We control for leveraged free cash flow, to capture potential managerial overinvestment (Richardson 2006). The impact of leveraged free cash flow is jointly captured by Free Cash Flow to Company t and DE t . The former is unleveraged free cash flow divided by the sum of end-of-year total equity and total debt. Observations were collected from Dynaquest Sendirian Berhad (2011Berhad ( -2014. Expectations are for a negative coefficient of Free Cash Flow to Company t . Data for DE t , the firm's debt-to-equity ratio, were obtained from Dynaquest Sendirian Berhad (2011Berhad ( -2014. Sign expectations regarding the coefficient of DE t are unclear. This paper employs two sets of controls for industry membership. The first set comprises the variable H-H j,t , a measure of product market concentration. Product markets may become concentrated because only a few players are able to survive the competition (Ali et al. 2014). Furthermore, in Malaysia, the incidence of government shareholding and hence the degree of government financial support both affects and is affected by product market competition (Gomez and Micheaux 2017). H-H j,t is an Herfindahl-Hirschmann index calculated on a product market-year basis. The narrow Dynaquest industry groupings are used, to increase intra-industry homogeneity of business models (Ali et al. 2014). Data were sourced from Stockbase. 6 The final sample comprises 69 product markets. The second set of industry controls constitutes a series of dummy variables flagging the Malaysian stock exchange's industry classifications. In Malaysia, the degree of government investment clusters by broad economic sector, due to variations in the mix between laissez-faire economics and the developmental state, are as ideologies underpinning economic development policies (Gomez 2009). Sign expectations regarding the coefficients of both sets of industry controls are unclear.
Equations (8)-(10) include a series of yearly dummies. Evidence indicates that in Malaysia, corporate performance varies temporally, according to macroeconomic conditions and fiscal policy. Evidence also suggests that these trends differ according to the type of government-related investor (Johnson and Mitton 2003;Lau and Tong 2008;Mitchell and Joseph 2010). Sign expectations regarding the coefficients of the yearly dummies are unclear.
Models using DQ-Strength and Growth t+1 as the dependent variable control for return of equity, measured in the same year as the dependent variable. Observations of Return on Equity t+1 were sourced from Dynaquest Sendirian Berhad (2012)(2013)(2014)(2015). This control acknowledges evidence that earning quality models and the Dynaquest assessment of earnings stability may confound extreme performance with earnings quality. A positive coefficient is expected (Dechow et al. 2010). Models using DQ-Strength and Growth t+1 as the dependent variable also control for earnings stability, measured in the same year as the dependent variable. A positive coefficient is anticipated.
Equations (8)-(10) control for foreign influences on Malaysian companies. It is not uncommon for listed Malaysian companies to have foreign executives and shareholders (Gomez 2009). Evidence indicates that foreign investors implement sound governance practices from their home countries (Shayan-Nia et al. 2017). Foreign influences are captured as Foreign CEO t , a dummy variable that flags observations with a non-Malaysian CEO. Data were collected from the annual reports, sourced from Bursa Malaysia Berhad (2021). A positive coefficient is anticipated.
Equations (8)-(10) control for company age. Older companies may be more diversified, have more sophisticated operations and have some products at the maturity life cycle phase. However, the anticipated direction of association between wealth-generation capacity and company age is unclear. Companies just beginning to diversify may be at early operational stages in their new segments, generating losses in these segments. A company's life cycle phase is the sum of the life cycle phases of its products. Hence, a company may revert from the maturity life cycle phase to the growth phase, when it diversifies its product mix to include newer products (Dickinson 2011). Furthermore, there is a larger volume of publicly available information and hence lower manager-shareholder information asymmetry for older companies. Dates of listing were obtained from the companies' websites, accessed via Bursa Malaysia Berhad (2021). Company size is measured via a natural logarithm of operating sales. Data were collected from Dynaquest Sendirian Berhad (2011-2014). Sign expectations regarding the coefficient are unclear.
The models using ROE t+1 as the dependent variable control for operating risk via DQ-Earnings Stability t. . The latter variable is the Dynaquest sub-rating for earnings and dividend stability. A stable earnings stream indicates lower business risk (Dechow et al. 2010). Furthermore, subsidisation may provide similar benefits as upwards accruals management without incurring the costs. A company would avoid the large accruals reversal and hence increased volatility of reported earnings (He 2016). Both of these mechanisms generate the expectation of a positive association between ROE t+1 and DQ-Earnings Stability t. . Data for DQ-Earnings Stability t. were obtained directly from Dynaquest.
Equation (10) includes GovSHSoc t as a control variable. Naturally, we anticipate a positive coefficient.

Sample Selection
The investigation period is 2011-2014. The initial sample comprises the population of Malaysian listed company years, followed by Dynaquest. The initial sample was sourced from Dynaquest Sendirian Berhad (2011)(2012)(2013)(2014)(2015). Company years from the finance industry were deleted. We remove IPOs via a filter requiring an observation to have been listed for at least three years as of the September of the year of interest.
Other reasons for deleting observations are as follows. Some of the observations were not followed by Dynaquest in year t + 1. Others were deleted due to a change in balance date, either in year t or year t + 1. Other observations were deleted because their financial leverage and/or return on equity ratios were not meaningful, owing to a negative denominator. The final sample comprises 1732 company years, representing 544 companies. The entire sample of 1732 observations was used in tests of H1. Tests of H2 use the reduced sample of 203 observations subject to shareholding by a government investor with a mission related to economic policy. Table 1 presents descriptive statistics related to the continuous variables. The statistics in Panel (a) were calculated using the entire sample. The statistics in Panel (b) were calculated using the reduced sample of observations with a shareholder political connection related to economic policy. We winsorised ROE t+1 at the ninety-ninth and first percentiles. The upper quartiles of H-H t in Panels (a) and (b) are 0.29 and 0.40, respectively, indicating that most products markets in Malaysia have competitive structures. The lower quartiles of Free Cash Flow to Company t are −0.06 and −0.05, respectively, indicating that approximately three-quarters of the sample observations were generating sufficient cash flows from operations to cover capital investment requirements.  where all variables are defined in the Appendix A. The statistics in Panel (a) were calculated using the entire sample of 1732 company years. The statistics in Panel (b) were calculated using the sub-sample of 203 company years subject to shareholding by a government investor with a mission related to economic policy.

DQ-Strength and
Both panels also report the standard deviations of the continuous variables. The standard deviations of the variables constructed using Dynaquest data are substantially lower than the standard deviations of the other variables. This reflects that the Dynaquest variables have lower ranges. (e.g., our intrinsic value surrogate can only adopt values between 0.00 and 9.00; ROE t+1 can take the value of any real number at all.) The large standard deviation of (1 − α) t , in Panel (b) (24.15) indicates that a strength of our study is wide within-sample variation in this independent variable on interest. Table 2 presents univariate descriptive statistics related to the categorical variables. Panel (a) reveals that government shareholding by investors with a social objective is substantially more common than shareholding by investors with an economic policy mission. Panel (a) reveals that 41% of the sample observations were subject to any form of government shareholder. Conversely, Panel (b) reveals that only 7% of the sample observations had GLC status. This confirms that while all GLCs are subject to government shareholding, the converse is not true (i.e., there are many companies subject to government shareholding without GLC status). Panel (d) documents that approximately three-quarters of the observations were Nanyang companies, consistent with prior evidence (Sinnadurai 2016). Panel (e) shows that only one-quarter of the sample observations had founding CEO status, indicating that most of the Nanyang companies had progressed beyond the founding generation. Panel (g) reveals that the sample observations are distributed evenly across the investigation period.

Panel (a)-Subjugation to government shareholding
Number (percentage) of company years subject to ownership by a government-related investor with an economic policy mission 203 (12) Number (percentage) of company years subject to ownership by a government-related investor with a social policy mission 615 (36) Number (percentage) of company years without any ownership by a government-related investor 1019 (59) Total 1732 (100)

Panel (b)-Government-Linked Corporation Status
Number (percentage) of company years with Government-Linked Corporation Status 113 (7) Number (percentage) of company years without Government-Linked Corporation Status 1619 (93) Total 1732 (100)

Panel (c)-Status regarding having an informal longstanding political connection
Number (percentage) of company years with a longstanding information political connection 243 (14) Number (percentage) of company years without a longstanding information political connection 1489 (86) Total 1732 (100)

Panel (g)-Temporal distribution of final sample
Number ( The descriptive statistics in Table 2 were calculated using the entire final sample of 1732 observations. The percentages in Panel (a) sum to more than 100 because an observation can be subject to shareholding by both categories of government investors. Table 3 reports correlations between pairs of continuous variables. Panel (a) uses the full sample; Panel (b) uses the sub-sample of observations subject to ownership by government investors with a mission related to economic policy. where all variables are defined in the Appendix A. The correlations in Panel (a) were calculated using the entire sample of 1732 company years. The correlations in Panel (b) were calculated using the sub-sample of 203 company years subject to shareholding by a government investor with a mission related to economic policy. ***, ** and *, respectively denote significance at the one, five and ten percent levels (two-tailed).

Bivariate Correlations
Some correlations are common to both panels. (1 − α) t is positively correlated with H-H t , the measure of product market concentration (p < 0.01, two-tailed). Industries with output amenable to achieving economic policy may be subject to government-directed horizontal integration, auguring for increased industry concentration (Gomez and Micheaux 2017). (1 − α) t is positively correlated with company size (p < 0.01, two-tailed). In industries with high government ownership, the politically connected players may become dominant due to their resource base and government-directed horizontal and vertical integration (Lai 2012;Gomez and Micheaux 2017). Consistent with this explanation, both panels document positive correlations between H-H t and DQ-Strength and Growth t+1 (p < 0.01, two-tailed). Management strength is positively correlated with intrinsic value, return on equity and earnings stability, (p < 0.01, two-tailed), suggesting that stronger management is more successful at generating shareholder wealth (Hu and Kumar 2004). The correlations between H-H t and DQMS t are positive (p < 0.01, two-tailed), possibly due to some industries becoming concentrated due to the business acumen of the successful players (Ali et al. 2014). Earnings stability is positively correlated with intrinsic value, return on equity and product market concentration (p < 0.01, two-tailed), suggesting that successful companies have earnings streams that penetrate product market competition (Ali et al. 2014). Free cash flow is positively correlated with company age (p < 0.10, two-tailed), possibly due to some older companies being established industry players, with penetrating earnings streams (Ali et al. 2014). Table 3 reports some variables that are significantly correlated with return on equity in only one panel. Return on equity is positively correlated with product market concentration (p < 0.01, two-tailed) in Panel (a). This may be due to some product markets being concentrated because of a small number of successful players (Ali et al. 2014). Table 3 reveals that return on equity is negatively correlated with financial leverage (p < 0.01, two-tailed) in Panel (b) but not Panel (a). A possible explanation is that companies subject to shareholding by a government investor with an economic policy mission are afforded more debt from government-related sources (Fraser et al. 2006). These companies may experience difficulty servicing this debt, causing deterioration in financial performance (Mohd-Saleh and Ahmad 2005).
Free cash flow is positively correlated with (1 − α) t (p < 0.05, two-tailed) in Panel (a), possibly due to public policy projects generating substantial cash flow from operations. The fact that the counterpart correlation is insignificant in Panel (b) may be artefactual, due to the smaller number of observations, use to produce Panel (b). The correlation in Panel (b) is actually twice the size of its counterpart in Panel (a). The correlations are 0.10 and 0.05, respectively.
Panel (b) reports a negative correlation between management strength and financial leverage (p < 0.10, two-tailed), for companies subject to shareholding by government investors with missions related to economic policy. The corresponding correlation in Panel (a) is insignificant. A plausible explanation is that politically connected companies with poor quality management may be afforded "bail out" packages from the government, to assist their recovery from financial distress (Mohd-Saleh and Ahmad 2005;Fraser et al. 2006). In Panel (b), earnings stability is negatively correlated with financial leverage (p < 0.10, two-tailed). The corresponding correlations in Panel (a) are not significant. A possible explanation is that of the companies subject to shareholding by government investors with a public policy mission, older companies have more involvement in implementing public policy projects than younger companies. Hence, older companies would carry more debt, from government sources, to fund these projects (Fraser et al. 2006). The older companies would also have a greater portion of their capital budgeting projects at the growth life cycle phase, yet to produce stable earnings streams (Dickinson 2011). Table 4 presents the results of the empirical tests of H1. where: the models presented in Table 4 are estimated using the full sample of 1732 company-year observations. The basic forms of the models estimated are Boycko are defined in the Appendix A. β i , β* i , δ j and δ* j are regression parameters. ε i,t and ε* i,t are stochastic disturbance terms. Asymptotic t-ratios are reported in parentheses underneath the coefficients. One-tailed tests are conducted in cases where there are sign expectations regarding the coefficient. In other cases, two-tailed tests are conducted. *, ** and ***, respectively, denote significance at the ten-five-and one-percent levels in the anticipated direction (one-tailed). ++ and +++ , respectively, denote significance at the five-and one-percent levels, in the opposite direction from anticipation (two-tailed).

Empirical Results
Overall, the results from using intrinsic value as the dependent variable, reported in Models (1) and (2), support H1. In Model (1), the coefficient of GovSHAny t is positive and significant (p < 0.05, one-tailed). This suggests that in general, companies subject to government shareholding generate more shareholder wealth than other companies. Model (2) uses separate independent variables, flagging each type of government-related investor. The coefficient of GovSHSoc t is positive and significant (p < 0.01, one-tailed), consistent with expectations. Conversely, the coefficient of GovSHEcon t is negative and significant (p < 0.01, two-tailed), contrary to expectations. This indicates that support for H1 is driven by companies subject to ownership by government investors with a social policy mission.
Possible reasons for these results follow. In companies subject to shareholding by a government investor with an economic policy mission, the government investor assumes a direct role in corporate decision-making, generating conflicts between the goals of shareholder wealth maximisation and implementing public policy (Boycko et al. 1996;Lai 2012;Gomez et al. 2018, p. 47). Conversely, government investors with social policy missions tend to be actively managing investment funds, with the goals of portfolio return maximisation (Gomez et al. 2018, pp. 28-31, 65).
The coefficients of the control variables, capturing other modes of government input into corporate decision-making, have mixed conformity with expectations. The coefficient of LSInfPC t is not significant in Model (2). This suggests that longstanding political connections do not add value for shareholders, contrary to prior evidence (Fung et al. 2015). The coefficient of GLC t is positive and marginally significant (p < 0.10, one-tailed). This evidence weakly accords with prior studies (Lau and Tong 2008). The sound financial performance of GLCs may be due to the investigation period of this study coinciding with the period of the GLC Transformation Program (Gomez et al. 2018, p. 199).
Our finding differs somewhat from the consensus in the empirical literature that in general, government ownership of listed companies is not shareholder wealth-enhancing (Wang and Shailer 2018). However, our evidence suggests that a reason for these prior findings is that they have had insufficient regard for the fact the relation may differ among types of government-related shareholders.
Models (3) and (4), using return on equity as the dependent variable, do not support H1. In Model (3), the coefficient of GovSHAny t is insignificant. In Model (4), the coefficients of both GovSHEcon t and GovSHSoc t are insignificant. A possible reason for these results is that earnings are produced to facilitate decision-making of shareholders seeking to maximise their wealth, and hence have diminished value-relevance in companies with public policy objectives (Gomez et al. 2018, pp. 150-64). This also suggests that the evidence in Models (1) and (2), providing strong support for H1, is more credible than the weaker evidence from Models (3) and (4). However, this interpretation is inconsistent with evidence from the People's Republic of China that subsidies are value relevant (Chen and Wang 2004;Lee et al. 2014).
The coefficients attaching to the variables capturing family company attributes display mixed conformity with expectations. The coefficients of Founder CEO t are uniformly insignificant. A possible explanation is that the Nanyang companies, dominating the sample, may continue being financially successful, after the founder generation, due to the sound Confucian tenets underpinning their business models (Sinnadurai 2018). The coefficients of Nanyang t are positive and significant (p < 0.05, one-tailed) in Models (1) and (2), using DQ-Strength and Growth t+1 as the dependent variable. However, the coefficients of Nanyang t are negative and significant (p < 0.05, two-tailed) in Models (3) and (4), using ROE t+1 as the dependent variable. A possible explanation is that management of Nanyang companies may not prioritise production of quality earnings. The principal shareholders are also managers and hence able to obtain more specific information, reducing the demand for earnings quality (Wang 2006). 7 Most of the other control variables have coefficients conforming with expectations. The coefficients of DQ-Earnings stability t+1 are uniformly significant (p < 0.01, one-tailed). The coefficients of ROE t+1 are uniformly positive and significant (p < 0.01, one-tailed).
The coefficients of Company age t are all significant (p < 0.01, two-tailed). However, these coefficients are both positive in Models (1) and (2) and both negative in Models (3) and (4). A possible explanation is that company age is a non-monotonic proxy for the life cycle phase (Dickinson 2011). This explanation may also account for the fact that the coefficients of company size are uniformly insignificant. The coefficients of DE t are negative and significant (p < 0.01, two-tailed) in Models (1) and (2), suggesting that Malaysian companies are over-levered. Similarly, in Models (1) and (2), using intrinsic value as the dependent variable, the coefficients of H-H t are both positive and significant (p < 0.05, two-tailed). The coefficients of HH t and DE t and are uniformly insignificant in the two models using intrinsic value as the dependent variable. The coefficients of free cash flow are uniformly insignificant. These results are like due to offsetting mechanisms affecting these variables' association with shareholder wealth. Table 5 presents the results of the empirical tests of H2. Table 5. OLS estimation of models testing H2-Using default classifications of government-related investors.

Model (2)-DQ-ROE i,t+1 as the Dependent Variable
( Yes F-statistic testing null hypothesis that coefficient that coefficient of (1 − α) t is zero, against the alternative that this coefficient is positive 3.07 * 0.61 F-statistic testing null hypothesis that coefficient that the sum of the coefficient of (1 − α) t and Q 3 × (1 − α) t is zero, against the alternative that this sum is negative 1.67 0.70 F-statistic testing null hypothesis that coefficient that the sum of the coefficient of (1 − α) t and Q 4 × (1 − α) t is zero, against the alternative that this sum is negative 0.74 0.15 R 2 0.65 *** 0.31 *** where: The models are estimated using the sub-sample of 203 company-year observations subject to shareholding by a government investor with a mission related to economic policy. The basic forms of the models estimated are represented in Equation (10). All variables are defined in the Appendix A. β' 0 , β' 1,j , β' 2,j and δ' j are regression parameters. ε' i,t is a stochastic disturbance term. The coefficients of the intercept terms, β' 0 and β' 2,j (j = 2, 3, 4) are not reported. Asymptotic t-ratios are reported in parentheses underneath the coefficients. *, ** and ***, respectively denote significance at the ten-, five-and one-percent levels, in the anticipated direction (one-tailed).
The results in Table 5 partially support H2. The first arm of H2, postulating that at lower levels of (1 − α) t , shareholder wealth is positively associated with (1 − α) t , is partially supported. The coefficient of (1 − α) t is positive and significant (p < 0.05, one-tailed) in Model 1, consistent with expectations. The corresponding coefficient is not significant is Model 2, contrary to H2. The results in Table 5 do not support the second arm of H2, postulating that for companies with higher levels of (1 − α) t , there is a negative association between shareholder wealth and shareholding by government investors with an economic policy mission. The F-statistics, testing the null hypothesis that the slope coefficients of (1 − α) t equal zero for observations in the third and fourth quartiles of (1 − α) t are uniformly insignificant.
However, the results in Model (1) of Table 5, using intrinsic value as the dependent variable, support a different conclusion. In Model (1), the coefficient of (1 − α) t is positive (p < 0.05, one-tailed). This indicates that for companies with low levels of shareholding by government investors with mission elated to economic policy, shareholding by the government investor is value-enhancing. The coefficient of Q 3 × (1 − α) t is not significant. The coefficient of Q 4 × (1 − α) t is negative and significant (p < 0.05, one-tailed), consistent with expectations. This suggests that for companies in the top quartile of (1 − α) t , the association between shareholder wealth and ownership by the government investors (with an economic policy mission) is positive but not as strong as for companies in the lower three quartiles. Hence, in aggregate, the evidence from Model (1) indicates that in Malaysia, ownership by a government-related investor with an economic policy mission enhances shareholder wealth, with diminishing marginal returns. This is a slightly different conclusion from international evidence that this relationship follows an "inverted U" pattern (Boubakri et al. 2020). The difference may be due to the failure of previous studies to distinguish types of institutional investors.
The results in Model (2) of Table 5, using return on equity as the dependent variable, do not support H2. The coefficient of (1 − α) t is not significant. In Model (2), the coefficient of Q 3 × (1 − α) t is negative and significant (p < 0.01, two-tailed). However, the F-statistic (testing the null hypothesis that the sum of the coefficients of (1 − α) t and Q 3 × (1 − α) t equals zero) is not significant. This suggests that for companies in the third quartile (of ownership by a government investor with an economic policy mission), there is no net impact on earnings of this type of government ownership. The results using return on equity as the dependent variable also indicate that this is also true for companies in the top quartile of shareholding by government investors with a mission related to economic policy. These results may be due to earnings being a sub-optimal valuation metric for companies with a public policy mission (Gomez et al. 2018, pp. 150-64).
The principal limitation of this study is that the boundary between economic and social policy is blurred (Gomez 2009). Some of the government investors with a social policy mission have similar fund management strategies as their counterparts with an economic policy mission (i.e., preference to "buy and hold", rather than engage in active portfolio management and having an active input into the governance of the investee, in order to implement public policy). Hence, we repeated the analyses after re-classifying some of the government-related investors as having a mandate related to economic policy, rather than social policy. 8 The results (unreported) are qualitatively identical to those in the body of the paper.
Another limitation is that Dynaquest does not follow the entire population of companies listed on Bursa Malaysia. Hence, to some extent, our study may be plagued by sample selection bias. However, there are also benefits of using Dynaquest data. The use of country-specific proxies for theoretical constructs would result in less measurement error and represents an adaptation to the unique Malaysian setting (Claessens and Yurtoglu 2013). In order to reap this benefit, it is necessary to restrict our sample to companies followed by Dynaquest.

Conclusions
The study investigates the shareholder wealth impact, on listed companies, of share ownership and (by extension) subsidisation by the Malaysian Government. We assume that for companies subject to shareholding by a government investor with a mission related to economic policy, the level of government shareholding is a proxy for the degree of government input into the corporate decision-making and hence the extent to which the entity is subsidised (Boycko et al. 1996).
Contrary to the general consensus of international evidence (Wang and Shailer 2018), our results suggest that in general, government share ownership is value-enhancing. However, this result is driven by one particular type of government-related shareholder (government investors with a social policy mission). Our results suggest that in general, companies subject to shareholding by an investor with an economic policy mission have lower shareholder wealth than companies without any government shareholding at all. Hence, the findings of prior studies may be limited, due to their failure to acknowledge that the relationship may differ among types of government-related investors.
The results suggest that for companies subject to shareholding by a government investor with an economic policy mission, shareholder wealth increases with the level of government shareholding. The strength of this association is subject to diminishing marginal returns. This conclusion differs from the finding of Boubakri et al. 2020) that this relationship follows an "inverted U" shape. Our finding corroborates the conclusion that the Boycko et al. (1996) model accurately characterises Malaysian companies subject to ownership by government investors with an economic policy mission.
Our conclusions are principally based on the findings using intrinsic value as the dependent variable. The results using return on equity as the dependent variable do not support our hypotheses. We regard the evidence from using intrinsic value as more reliable. Earnings may have diminished value relevance for companies with a public policy mission (Gomez et al. 2018). Hence, a suggestion for further research would be to investigate the value relevance of earnings for companies subject to government share ownership. The Easton and Harris (1991) and Ohlson (1995) methodologies would be appropriate, similar to studies using data from the People's Republic of China (Chen and Wang 2004;Lee et al. 2014).
Our conclusions have implications for at least two categories of business stakeholder in Malaysia. The results suggest that for private sector investors, investment in companies subject to ownership by a government investor with a social (economic) policy mission would contribute positively (negatively) to maximisation of portfolio returns. However, if the private sector investor is compelled to own some shares in companies subject to ownership by a government investor with an economic policy mission, the potential of the stock for return maximisation would be positively related to the level of government ownership. Our findings also have implications for the Malaysian Government. Our evidence suggests that the shareholder wealth impact of the economic and social development policies is not straightforward and depends upon the type and degree of government shareholding. Another suggestion for further research would be to investigate this issue and the explanatory power of the "grabbing hand" theory (Boubakri et al. 2020;Yu and Wang 2020) in the specific context of the 1Malaysia Development Berhad scandal.
Funding: This research received no external funding.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
Most of the data used for this study are publicly available, from Bursa Malaysia (2021). The data provided by Dynaquest Sendirian Berhad are only available via the following means: subscription to Stockbase, purchasing the relevant issue(s) of the Stock Performance Guide, published by Dynaquest Sendirian Berhad (2011Berhad ( -2015, purchasing the data for any particular subset of the variables directly from Dynaquest Sendirian Berhad.

Conflicts of Interest:
The authors declare no conflict of interest.

SH Wealth
Indicator j,i,t+1 Value of shareholder wealth indicator j for company i in year t + 1. The two constituents of SH Wealth Indicator j,i,t+1 are DQ-Strength and Growth i,t+1 and ROE i,t+1 .

DQ-Strength and Growth i,t+1
Sum of Dynaquest sub-ratings for management strength, financial strength and earnings and dividend growth for firm i as at September, year t + 1.

ROE i,t+1
Return on equity for company i during year t + 1.
GovSHAny i,t 1 if, during year t, company i had shareholding by either type of government-related investor and 0 otherwise.
GovSHEcon i,t 1 if, during year t, company i was subject to shareholding by a government-related investor with a related to economic policy and 0 otherwise.
GovSHSoc i,t 1 if, during year t, company i was subject to shareholding by a government investor with a mission related to social policy and 0 otherwise.
(1 − α) i,t The level of shareholdings by government-related investors with an economic policy mission, company i, during year t.

Control j,i,t
Value of control variable j, for company i, during year t. The individual constituents of Control j,i,t are as follows.
LSInfPC i,t 1 if as at the balance date in year t, company i had a longstanding informal political connection. Otherwise, LSInfPC i,t = 0.
GLC i,t 1 if, during year t, company i was a Government-Linked Company and 0 otherwise.
Nanyang i,t 1 if during year t, company i had an ethnic Chinese Chief Executive Officer and was subject to a degree of share ownership by the family of the Chief Executive Officer. Otherwise, Nanyang i,t = 0.
Founder CEO i,t 1 if the Chief Executive Officer of company i as at the balance date in year, was the founding Chief Executive Officer. Otherwise, Founder CEO = 0.

DQMS i,t
Dynaquest sub-rating for management strength of company i as at September, year t.
H-H i,t Σ i (Sales revenue i,t /Total sales of all companies in industry j during year t) 2 .
IND j,t,i 1 (0) if company i was (was not) a member of industry j in year t.
Foreign CEO i,t 1 for observations with a non-Malaysian Chief Executive Officer and 0 otherwise.
Yr i,t 1 for observations drawn from year t and 0 otherwise.

Company age i,t
Number of years between the date of listing of company i and the balance date in year t.
Free Cash Flow to Company i,t Unleveraged free cash flow to total capital for company i, during year t, scaled by contemporary sales revenue.

DE i,t
Debt-to-equity ratio of company i, as at the end of year t.
Company Size i,t Natural logarithm of sales revenue of company i, during year t.

DQ-Earnings Stability i,t+1
Sub-rating for earnings and dividend stability for firm i, as at September, year t + 1.

1
There are three principal ethnic groups in Malaysia: Malays and indigenous Malaysians (collectively referred to as "bumiputera", "sons of the soil"), the ethnic Chinese and the ethnic Indians. Historical and cultural factors have caused ethnicity to be associated with economic function in Malaysia. The bumiputera have traditionally been poor, rural residents. The ethnic Chinese have historically been urban dwellers, owning and managing most of the country's businesses (Eng 2004). Successive policies of the Malaysian Government, starting with the New Economic Policy in 1971, have sought to redress these imbalances. The principal objectives have been to eliminate poverty and the association between ethnicity and economic function (Gomez et al. 2018, p. 34). Hence, the State Economic Development Corporations have been working towards these objectives, within their particular Malaysian states.
2 Barisan Nasional was ousted from power in the fourteenth general election in 2018, replaced by Pakatan Harapan, a new coalition of then-opposition parties (Nadzri 2018). Pakatan Harapan was ousted in 2020, replaced by another coalition, Perikatan Nasional.

3
For example, Gomez et al. (2018, p. 46) note that Khazanah Nasional Berhad was never intended to engage in active stock market trading. 4 Unsurprisingly, the evidence is not unanimous. The findings of Johnson and Mitton (2003) suggest that Malaysian companies with an informal political connection performed worse (or better) during period of economic crisis (when regulators impose capital controls).

5
"Family" is defined as first cousin or closer (via marriage or blood) (Sinnadurai 2016). 6 Stockbase is a database maintained by Dynaquest, available from Dynaquest Sendirian Berhad (2021). Observations in Dynaquest industries such as "multi-industry" were re-allocated to the single industry corresponding to the business segment with the largest sales revenue. 7 Wang (2006) discusses this mechanism. The results, supporting the opposite hypothesis that family companies produce higher quality earnings, may not generalise to Malaysia. Wang, Kun, and Greg Shailer. 2018. Does ownership identity matter? A meta-analysis of research on firm financial performance in relation to Government versus private ownership. Abacus 54: 1-35. [CrossRef] Yu, Xinyu, and Ping Wang. 2020. Government control and the value of cash: Evidence from listed firms in China. Review of Quantitative Finance and Accounting 55: 1341-69. [CrossRef]