Monitoring intensity, investment inefficiency and institutional shareholders: Evidence from JSE listed companies in South Africa

This study investigates how variation in monitoring intensity affects the efficiency of firms’ investment decisions in an emerging market in South Africa. The study hypothesis argues that the distraction of institutional shareholders has a statistically significant positive effect on corporate investment inefficiency. Using a more robust Generalized Method of Moments (Sys GMM) estimation approach to analyze data collected for firms listed at the Johannesburg Stock Exchange (JSE) for the period 2004–2019, the results showed that the distraction of institutional shareholders has a positive and statistically significant impact on investment inefficiency. That is, when the attention of institutional shareholders is shifted, the intensity of their monitoring drops, and the executive is involved in investment decisions that are not profitable. This insight has an implication for stakeholders and the value-creating corporate governance mechanism. The study concludes that institutional shareholders must always sustain their monitoring intensity to ensure that corporate decisions are consistent with the firm’s value.


INTRODUCTION
The involvement of institutional shareholders in corporate finance has become a crucial issue that is open to comprehensive research in corporate finance literature. These are large investors who invest on behalf of their members. They include the followings: superannuation and pension funds, life and non-life insurance companies, investment and unit trusts, financial institutions such as banks and finance companies, credit co-operatives, building societies, and investment companies (Koh, 2003;Survé, 2009). They are essential in the capital market due to their large holding of shares. The market value of their investment in the United States, British, and South African stocks is vast and significant (Bhikha, 2014;Blume & Keim, 2012;Thomas, 2017;Ward et al., 2017). 80% of the US stock market's equity market is held by institutional investors (Intractive, 2017); institutional investors in FTSE 100 in the UK account for 62% of total ownership (Segerstrom, 2020). Institutional shareholders constitute the majority of investors on the Johannesburg Stock Exchange (JSE) (Zhang, 2016), and they mostly include provident and pension funds, insurance companies, and collective investment schemes(CIS) (Nhlapo & Gumata, 2011;Sibanda & Holden, 2014). The size and significance of institutional shareholders have grown over time. This category of investors accounts for about

Institutional shareholders' monitoring and limited attention
Standard economic models conjecture that shareholders use every information available to make prudent decisions. Nevertheless, psychology and behavioral finance literature assert that shareholders are contingent upon intellective restraints and emotional tendencies. The central intellectual-processing capacity of the human brain is limited. The volume of information appropriate for the firm's valuation is immense, and the time and intellectual resources needed to handle such information are substantial. Therefore, shareholders usually fail to integrate every appropriate information because of limited attention (Wang, 2017). This limited attention does not apply to individual shareholders alone but also to institutional shareholders. It was documented by Abarbanell and Bushee (1998) that analysts could not efficiently use available information in the financial ratios. Likewise, the analyst failed to discount discretional accruals of the firm's new issue sufficiently (Teoh & Wong, 2015). Furthermore, Hirst and Hopkins (1998) provided empirical evidence that a professional analyst usually fails to recollect and react appropriately to information in detailed financial disclosures.
The survey conducted by the Investor Responsibility Research Center Institute IRRC (2011) revealed that institutional shareholders have limited attention. They document a direct connection between restriction to institutional shareholders' attention and monitoring of corporate decisions. They stated that "three-fourths of the institutional shareholders submit that time is the most usual barrier to engagement with the corporations, while staff plan ranks second." So, institutional shareholders do not equally monitor firms they invested in with the same enthusiasm. The implication of this is that institutional shareholders may become 'distracted' at a particular time. Although 'distracted', they provide below the optimal control level (Kempf et al., 2017).
Since the determinants of investor attention are not known, measuring shareholders' attention becomes problematic. To sidestep this challenge, various empirical proxies have been suggested to secure shareholders' attention. These proxies have produced many exciting and keen findings regarding stock price movement around important corporate information events such as earnings announcements, analyst recommendations, prominent attention-grabbing events, etc. (Wang, 2017). A typical empirical proxy relating to shareholders' attention is firm size. More prominent firms are generally focused upon by shareholders for several reasons. For example, they have more analyst coverage; more significant firms have more news media coverage than smaller firms. However, because firm size is also being used as a proxy for other variables like information asymmetry, using it as a proxy for shareholders' attention makes it a noisy measure. Additionally, even though the firm size and analyst coverage are proxies for the volume of the available information in the public domain, the measure becomes indirect because it is difficult to determine the extent to which the shareholder will utilise the information (Wang, 2017).
In this regard, researchers have suggested several optional proxies for institutional shareholder's attention or inattention. Dellavigna  They discovered that the response to the day's announcement is smaller than the reaction after the announcement, which is better about several competitive announcements. However, the same day's earnings announcement from a different industry is a lot more distracting than industry-related announcements. Engelberg and Gao (2011) suggested that the Google search volume index offers a better direct measure for shareholders' attention. They discussed that the large number of requests for stocks on Google is an indication that many shareholders pay attention to and find information about the particular stock. They recorded a positive relationship between changes in the volume of search and shareholders' trading. Besides, they documented that an increase in shareholder's attention is related to stock returns of the first-day IPO.
Considering the optional empirical surrogates, trading volume is much more notable and broadly used. The perception is unambiguous. Shareholders hardly trade in stocks they pay less attention to. Whereas the likelihood of stock trade they pay greater attention to is high. That is to say, attention is well correlated with the volume of trade. Evidence from the literature has lent credence to the linkage between shareholder's attention and trade volume. Boone and White (2015) demonstrated that trade volume, especially in large stocks, attracts shareholders' attention. Chordia (2000) showed that, despite controlling for firm size, high volume stocks respond quickly to market returns information compared to low volume stocks. This indicates that trading volume reflects information regarding shareholder attention over and above firm size. Gervais et al. (2001) proposed that stock visibility is induced by the high trad-ing volume, which attracts shareholders. Barber  In the King IV draft report, critical attention was paid to institutional shareholder's responsibilities.
While the earlier King reports on the board of directors' positions as a contact point of corporate governance, the King IV report extends the implementation to cover institutional shareholder responsibilities referred to as institutional shareholder fiduciary duty. From the 17 principles of King's report beginning from 75 in King III, one relates explicitly to institutional shareholders. This shows their investee companies' gains when they are alive to their corporate monitoring activities (Harber, 2017). As stated in King IV principle 17, an institutional shareholder must ensure that profitable investment is initiated and practised by their investee companies to strengthen good governance and value creation (Harber, 2017;IoDSA, 2016). They should pursue and enforce high-yielding investments that guarantee long-term and lasting returns. Their actions and inaction will either strengthen or weaken good governance (IoDSA, 2016). Also, Zhang (2016) reiterated the responsibility of institutional shareholders to comply with their fiduciary duties by incorporating ESG contemplations and ensuring investee firms' continuing development (Zhang, 2016

Investment efficiency
A firm's investment decision is driven by investment opportunities that always lead to its growth due to the positive net present value (NPV) estimated. This investment decision is expected to be maintained until the minimum benefit is reached (Benlemlih & Bitar, 2018;Hayashi, 1982;Modigliani & Miller, 1958). In reality, firms face financial difficulties that prevent managers from executing all projects with positive NPV (Hubbard, 1997 The thinking behind Dt is that a given shareholder i in a given f firm is most probably distracted in the event of an attention-grabbing occurrence in another industry significant in the investor i-portfolio. Therefore, this study will first calculate a shareholder-level distraction score and, after that, sum it across all investors in the firm.
Precisely, Dt for each f firm at period t is modelled as: 11 1 , where

The effect of shareholder distraction on inefficient investment
Limited attention due to shareholder distraction has been proven from extant literature (

GMM model specifications
This study explored the dynamic panel data approach called the generalized method of moments (GMM) proposed by Arellano and Bond (1991). It generates a model that enhances the efficiency of the estimator. This equation modifies the fixed effect equation with the incorporation of instrumental variables. GMM can either be estimated using Difference GMM or System GMM.
Difference GMM: It was proposed by Arellano and Bond (1991). It corrects endogeneity by transforming independent variables through differencing. It also removes the fixed effect. However, the first differencing transformation has a weakness as it subtracts the previous observation from the contemporary one, thereby magnifying gaps in an unbalanced panel.
When the regressors are transformed through first differencing, the fixed effect is removed, since it did not vary over time, but endogeneity remains. From equation (8) The unobserved fixed effects no longer enter the equation as they are, by assumption, constant between periods. The first-differenced lagged dependent variable is instrumented with its past levels and now changes in the dependent variable as assumed to be represented by equation (8). So, equation (8) still shows that there is endogeneity in the model due to the lagged dependent variable  Arellano and Bover (1995) and Blundell and Bond (1998). It corrects endogeneity by introducing more instruments to improve efficiency significantly. Also, it transforms the instruments to make them uncorrelated (exogenous) with fixed effects. Moreover, it builds a system of two equations: the original equation and the transformed one. It used orthogonal deviations. That means rather than deducting the previous observation from the contemporaneous one, it deducts the average of all future available observation of a variable. Regardless of the data gaps, they can be estimated for all observations apart from the last for each individual, thereby reducing data loss.
For the system GMM model, the initial model under difference GMM equation (6) refers. The equation is assumed to be a random walk model, and the dependent variable ( ) y is persistent. In this case, applying the difference GMM estimator will yield both bias and an inefficient estimate of ϕ in limited samples, and this is especially keen when T is short. According to Blundell and Bond (1998), the underperformance of difference GMM in these circumstances is due to poor instruments. Therefore, the system GMM is applicable for the reasons stated below.
One equation is stated in the form of levels with the first difference as instruments, and the second equation is expressed in the first difference with levels as instruments. This approach includes more significant numbers of moment conditions. Still, in the study by Arellano and Bond (1991), Monte Carlo evidence indicates that when T is short, and the dependent variable is persistent, there are gains in accuracy, and small sample bias is minimized. Besides, when there are heteroscedasticity and serial correlation, a two-system GMM estimator should be used utilising a weighting matrix using residual from the first step. However, where there are limited samples, standard errors tend to be downward biased. In such a case, practitioners' usual approach is to use Windmeijer adjustment to correct for such small sample bias (Arellano & Bond, 1991;Arellano & Bover, 1995;Blundell & Bond, 1998;Windmeijer, 2005).
To estimate the dynamic model of this study in nature and control for endogeneity, the GMM estimation method was adopted. Extant literature established that a dynamic panel model improves estimators' efficiency in a panel model (Arellano & Bond, 1991). Oyedokun et al. (2009) stated that the combination of static specification of fixed effect model and autoregressive coefficients, the lagged value of the dependent variable, provides responses of both past and present shocks to the current value of the dependent variable. This specification method is referred to as GMM. The GMM eliminates temporal autocorrelation in the residual and averts running a spurious regression that may cause inconsistent estimators. The orthogonal conditions in the variance-covariance application control for correlation of errors, heteroscedasticity, simultaneity, and endogeneity issues (Antoniou et al., 2008;Vengesai, 2019 Subsequently, taking the first difference of equation (11), the following equation (12) To ensure that likely correlation between 1 it Inefinvest − and it ϕ is avoided, an instrumental variable N′ , which will not be correlated with the two, is achieved by matrix transcription of regressors. Equation (15)

Correlation matrix analysis
In the correlation analysis (see Table 2), the distraction measure (Dt), the variable of interest, is positively correlated with inefficient investment and is statistically significant. This means that an increase in Dt will lead to a rise in investment inefficiency. All other independent variables displayed a positive association with the dependent variable, showing positive and negative relationships. Generally, the analysis indicates that there is no multicollinearity between the variables. Consequently, prior-year investment behavior affects current trends. Incorporating the lagged dependent variable helps measure the previous investment impact on the current investment levels and minimize autocorrelation from misspecification (Arellano & Bond, 1991). A dynamic panel model vis-a-vis the estimating techniques mitigate against likely heterogeneity and endogeneity problems in the data sample. Over a while, investment dynamics are captured in the dynamic model, and partial adjustment instrument modeling is allowed (Baum et al., 2001;Vengesai, 2019). So, to determine the estimation's robustness, the study used dynamics panel data model techniques, system GMM. System GMM had been proved in the literature to produce an efficient estimate (Arellano & Bover, 1995;Blundell & Bond, 1998). It corrects endogeneity by introducing more instruments to improve efficiency significantly. According to Antoniou et al. (2008), the conventional estimation techniques that are the OLS, fixed, and random effects cannot control the dynamic biasness. Therefore, it is necessary to introduce stochastic variation into the model. System GMM had been confirmed as an appropriate estimation technique when there is serial correlation from idiosyncratic disturbances, heteroscedasticity, and endogenous regressors (Roodman, 2009).   Note: *** p < 0.01, ** p < 0.05, and * p < 0.1.

Two-step system GMM results
The finding shows a positive relationship between a shareholder's distraction and inefficient investment and is statistically significant. The result indicates a positive coefficient of 0.1674 and is statistically significant at 10%. This implies that a unit of a 0.1674 increase in a shareholder's distraction will lead to a 0.1674 increase in the level of inefficient investment. This confirms the assumption that institutional shareholders are subject to distraction due to extreme events in an unrelated industry in their portfolio, thereby weakening control over corporate monitoring. The executive takes advantage of that and engages in unprofitable investments that jeopardize the firm's value to shareholders' detriment. The results are similar to the study by Ward et al. (2017), where they link a less motivated institutional investor to investment inefficiency and obtained positive coefficients. The authors found out that institutional shareholders' higher motivated monitoring reduces over-investment and under-investment. Furthermore, the firm's tangible assets have a positive effect on investment inefficiency. The result can be the indiscriminate asset acquisition for personal interest. Other control variables, such as leverage, cash holding ratio, size, and age, reported negative effects on investment inefficiencies.

DISCUSSION
An important factor for considering a corporate investment ought to be the optimization of the firm's value. Corporate decisions related to investment should be effective and efficient to guarantee future returns to stakeholders, including the institutional shareholders. However, according to agency theory, the potential conflicts of interest between the shareholders and managers influences inefficient managerial decisions that lead to inefficient investments (Imegi and Nwokoye, 2015) The unproductive managerial decision prompted by a manager's opportunistic behaviour can be checked by adequate monitoring of the manager's corporate activities by the institutional shareholders. In other words, loosening the monitoring intensity of institutional shareholders exacerbate manager's free riders' tendencies where investment decisions seek to satisfy the personal interest. The study findings corroborate this fact. Distraction measure, which is a proxy for loosening monitoring intensity, show a positive and significant impact on inefficient investment. Furthermore, the results reflect one of the factors (uncontrolled acquisition spree) that contributed to the accounting scandal rocking Steinhoff International in South Africa.

CONCLUSION
The paper investigated the relationship between institutional shareholders' distraction and corporate investment inefficiency. The study's hypothesis shows that institutional shareholders' distraction is statistically significant and positively affects investment inefficiency. The more robust dynamic panel data estimation of system GMM is employed in the analysis to achieve the study's objectives. The study provides evidence from the findings: (1) Institutional shareholders' distraction significantly affects corporate investment inefficiency; (2) The firm asset tangibility significantly affects investment inefficiency. The paper's findings explain the recent accounting scandals in South Africa, where ineffective corporate governance was blamed for its cause. The ineffective corporate governance brought about by insufficient institutional shareholders' monitoring gives rise to opportunistic executive behavior, leading to financial scandals. Besides, although the effects of limited attention on institutional monitoring intensity had been explored in literature, this study, however, contributes to the extant literature by providing an understanding of how executives in an emerging economy respond to shifts in institutional shareholders' monitoring intensity, which, to the best of authors' knowledge is the first-hand evidence in South Africa. This understanding will spur action that will strengthen corporate governance in the face of future accounting scandals. Overall, the findings indicate that having insight into how corporate executives react to temporally loosening monitoring intensity can considerably enhance corporate governance perception of value creation in companies.