The effect of the financial reporting quality of the subsidiary on the credit ratings of the MNC

The effect of the financial reporting quality of the subsidiary

on the credit ratings of the MNC

Name: Bas Smit

Student number: 3797155

Master Accountancy and Controlling, Rijksuniversiteit Groningen Supervisor: Mrs. Rusanescu

Date: June, 2020 Word count: (11.054)

Abstract

This study examines the relationship between the average financial reporting quality of the foreign subsidiary and the long-term issuer credit rating of the parent. My expectation is that financial reporting quality of the foreign subsidiary is positively associated with the long-term issuer credit rating of the parent. Prior literature states that credit ratings are positively related to accrual quality in stand-alone firms because high financial reporting quality leads to less uncertainty about credit risk and therefore higher credit ratings. Furthermore, parents have incentives to and can manage their earnings in their foreign subsidiaries through the reporting guidelines that they impose on them. Besides this, the financial statements of the subsidiaries are consolidated in the financial statements of the parent, hence the financial reporting quality of the foreign subsidiaries impacts the overall financial reporting quality of the parent. My second expectation is that an independent board strengthens the relationship between the financial reporting quality of the foreign subsidiaries and the credit rating of the parent by indirectly monitoring the financial reporting quality of the subsidiaries through monitoring the financial reporting quality of the parent. This will increase the transparency of the financial statements of the parent and indirectly those of the foreign subsidiaries. This leads to less default risk and credit rating agencies who will be better able to take into account the financial reporting quality of the parents and indirectly the financial reporting quality of the subsidiaries. Furthermore, an independent board is associated with better credit ratings because they can reduce default risk by mitigating agency costs. This study uses a final sample of 1143 parent firm-year observations from 245 unique parent firms located in the US that have 3992 foreign subsidiaries located in Europe over a sample period of 2011-2017. I conducted two Ordered logistic regression and two logistic regressions and found no significant relation between the average financial reporting quality of the foreign subsidiaries and the parent’s long-term issuer credit ratings. This finding implies that credit rating

agencies do not specifically look at the financial reporting quality of the foreign subsidiaries when determining the long term issuer credit ratings of the parent. Besides this, I also found that independence of the board of the parent has no significant effect on the relationship between foreign subsidiaries financial reporting quality and the credit rating of the parent. This could imply that the parent’s board is not able to monitor the financial reporting quality of the foreign subsidiary directly or indirectly through the financial reporting quality of the parent.

Keywords: Financial reporting quality, foreign subsidiaries, multinational enterprises, board independence, credit rating, earnings management.

Table of content

1. Introduction

This study examines the relationship between the subsidiaries’ financial reporting quality and the multinationals’ credit rating and the influence of the independence of the multinationals’ board on this relationship. In recent years, multinational enterprises (MNEs) have become increasingly important for the overall economy of the world. For instance, the United Nations Conference on Trade and Development (UNCTAD, 2019) estimated that in 2018 the sales of MNEs represented around 30% of the global gross domestic product (GDP). This suggests that MNEs have a great economic importance. Furthermore, foreign subsidiaries are of up most importance, since the global top 100 MNEs had 60% of their total assets invested in foreign countries and 60% of their total sales were made in foreign countries in 2018. Moreover, 55% of their employees worked in foreign countries (UNCTAD, 2019). Finally, the sales of the global top 100 MNE’s foreign subsidiaries accounted for around 7% of global GDP. Due to the process of consolidation, the consolidated financials of the MNEs consist of the financials that are reported by their subsidiaries. Consequently, the financial reporting quality (FRQ) of the MNEs is expected to be influenced by the FRQ of their subsidiaries. Besides the economic importance of the FRQ of the subsidiaries, there is also an increased attention in earnings management and the function of the board in constraining this. As a result, the Blue Ribbon Panel was established by the Public Oversight Board. This is an independent private sector group that oversees the programs of the SEC. The Blue Ribbon Panel stated that the board may have a role in constraining earnings management.

Specifically, the Blue Ribbon Panel stated that when the board consists of a majority of independent outside directors, the level of earnings management will be lower and the transparency of the financial statements will be greater.

Due to the economic importance of the MNEs, it is also interesting to look at the credit ratings of those MNEs. Credit ratings are issued by credit rating agencies and represent their opinion about the credit risk of the company. It showsif the company is able and willing to meet their financial obligations to their debtors (Cha et al., 2016). Standard & Poor’s (2003), one of the big credit rating agencies, claim that they look at the quality of the financial information in determining the credit ratings of the firms. Several researchers investigated why credit ratings are important for firms. First, they found credit ratings to be important for the cost of debt since higher credit ratings lower the cost of debt. This is due to the

information that credit ratings provide about default risk. Default risk determines the firms cost of debt (Demirtas & Cornaggia, 2013; Graham & Harvey, 2001). Second, credit ratings are important for the financing structure of the firm and the ability of the firm to continue trading. For instance, many firms have credit policies that prevent them from doing business with firms that do not have at least an investment grade rating. Third, a downgrading of a firm’s credit rating could result in penalty clauses when this is set forth in contracts with other parties (Gray et al., 2006). Fourth, credit ratings are an important factor in the decision to issue more debt (Graham & Harvey, 2001).

Another construct what is also important for firms is the FRQ. FRQ is the extend in which the financial statements of the firm adequately represents information about the firms operations that helps potential users of the financial statements in making well founded decisions (IASB, 2008). FRQ is important, because the financial statements bear information about the

performance of the firm (Bushman and Smith, 2001). FRQ is mostly measured through earnings management. Healy and Wahlen (1999) describe earnings management as follows: “Earnings management occurs when managers use judgement in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers” (p.368). This implicates that when managers engage in earnings management, the FRQ reduces since the earnings of the firm do not adequately represent the firms operations (Healy and Wahlen, 1999).

FRQ can influence credit ratings because high FRQ leads to less uncertainty about credit risk, and therefore to higher credit ratings (Akins, 2018). In line with this, Ashbaugh-Skaife et al. (2006) found that credit ratings are positively related to accrual quality, because high accrual quality leads to less opportunism of managers and better transparency of the financial

statements. Furthermore, a high FRQ may influence credit ratings because high accrual quality leads to higher bond liquidity and therefore to higher credit ratings (QI et al., 2010). The consolidated financials of the MNEs contain the financials that are reported by their subsidiaries. Consequently, the FRQ of the MNE is expected to be influenced by the FRQ of its subsidiaries. Prior research shows that MNE-parents are inclined to manage its earnings abroad by making use of their foreign subsidiaries (Beuselinck et al., 2019). Dyreng et al. (2012) stated that MNEs do this in order to avoid getting caught, since it is more difficult for local tax authorities and auditors of the MNE-parents to monitor the financial statements of the foreign subsidiaries (Durnev et al., 2017). Also, the reputational and legal penalties that are associated with misreporting are likely lower when the misreporting takes place at the subsidiary level than when it takes place at the parent level. A reason for this could be that when it takes place at the subsidiary level the plaintiffs must show that both the subsidiary and the parent have the same interest and control and that some form of injustice was resulted, which is difficult to proof in court (Dearborn, 2009).

MNE-parents have a great amount of influence over the business decisions that are being made by their subsidiaries (Robinson and Stocken, 2013). In addition, parents may insert influence on reporting choices (Principe, 2012), which means they are able to manage earnings through their subsidiaries (Beuselinck et al., 2019). One way how parents can manage earnings is through the reporting guidelines that they impose on their subsidiaries (Principe, 2012). Due to the consolidation process, the consolidated financial statements of the parents, which are used to determine the credit rating of the parent (Standard & Poor’s, 2003), contain the financial statements of their subsidiaries (Beuselinck et al., 2019; Sutton, 2004). So, the FRQ of the subsidiaries is likely to influence the FRQ of the MNE as a whole. Therefore, I expect that a higher subsidiary FRQ will lead to a higher parent FRQ. This reduces the uncertainty about the credit risk of the parent and leads to higher transparency of the financial statements of the parent, which will result in a higher parent’s credit rating. Additionally, previous research found board independence to have a positive relationship with credit ratings (Ashbaugh-Skaife et al., 2006; Bhojraj & Sengupta, 2003). Specifically, more independent boards can reduce default risk through mitigating agency costs and are

considered to be better monitors of the management and thereby reducing information asymmetry between the firm itself and lenders (Bhojraj & Sengupta, 2003). Also, prior studies found board independence to be positively related with FRQ (Kao & Chen, 2004; Alves, 2014; Klein, 2002; Peasnell et al., 2000; Davidson et al., 2005). The main reason for this is that independent boards are better monitors of the management, because they have less ties with the management. Also, independent boards are better able to protect the interests of shareholders because they are better able to mitigate the private motives of the managers (Dechow et al., 1996). This restricts the opportunities of managers to engage in earnings management and thus enhances the FRQ.

A parent’s board is likely to be concerned about the FRQ of the subsidiaries, since the managers of the parent use the individual financial statements of the subsidiaries to draw up the consolidated reports (Beuselinck et al., 2019). Beuselinck et al. (2010) documented that governance characteristics of the MNE could have an effect on the FRQ of their subsidiaries. More specifically, ownership structure and analyst coverage of the MNE-parent affect the extent of earnings management on the subsidiary level (Beuselinck et al., 2010). An

independent board of the parent could result in more transparent financial statements of the subsidiaries, because through the consolidation process there is a relation between the FRQ of the parent and the FRQ of the subsidiary. So when a parent board monitors the FRQ of the parent they indirectly also monitor the FRQ of the subsidiaries. Therefore, I predict that board independence of the parent will have a positive moderating effect on the relationship between the FRQ at the subsidiary level and the credit rating of the parent. Since the board indirectly monitors the FRQ of the subsidiary when they monitor the FRQ of the parent, the

transparency of the financial statements of both the parents and the subsidiaries increases. This will lead to less default risk and credit rating agencies who will be better able to take into account the FRQ of the parents and indirectly the FRQ of the subsidiaries and therefore to higher credit ratings of the parent.

The empirical analysis is conducted on a sample of 245 U.S. MNEs and covers the period 2011-2017. The material subsidiaries will be identified and hand-collected from exhibit 21.1 of the 10-K (Securities Exchange Commission’s (SEC) Edgar database). The financial data of the subsidiaries is obtained from the Orbis database while the necessary information about the MNEs is obtained from Compustat. The data about the composition of the parents’ boards are gathered from the MSCI (former KLD and GMI) database. I will use earnings management as a measure of subsidiaries’ FRQ. I determine the level of earnings management through

making use of discretionary accruals, which I estimate by using the modified Jones model (Dechow et al., 1995). The measure of the MNEs credit ratings will be long-term issuer credit ratings from Standard & Poor’s. The ratings range from AAA to D. This measure is

commonly used in the literature (Ashbaugh-Skaife et al., 2006; Cha et al., 2016; Cantor & Packer, 1996). Finally, in accordance with Klein (2002), I consider a board of directors to be independent when 51% (i.e. the majority) of directors is independent from the management. My findings suggest that the FRQ of the foreign subsidiaries do not have an impact on the credit rating of the parents. This could be due to the fact that credit rating agencies do not specifically look at the statements of the foreign subsidiaries and only at the statements of the

parent company when determining the rating of the parent. Furthermore, the results show that board independence of the parent does not strengthen the relationship between the foreign subsidiarie’s FRQ and the parent’s credit rating. This could be because of the geographical distance and operational complexity of foreign operations which are likely to impair effective monitoring and clear consolidation of the statements of the foreign subsidiaries (Beuselinck et al., 2011;Chin et al., 2009).

This study contributes to the current literature on the determinants of credit ratings. To the best of my knowledge, the relation between subsidiary FRQ and MNE credit ratings has not been studied before. Prior literature does documents a positive relation between FRQ and credit ratings in stand-alone firms (Ashbaugh-Skaife et al., 2006; QI et al., 2010). However, they did not study whether the FRQ of the subsidiaries has an effect on the credit ratings of the parent. This is important because the FRQ of the subsidiary is related to the FRQ of the parent through the consolidation process. Credit ratings are calculated based on the

consolidated reports of MNEs (Standard & Poor’s, 2003) and since those reports consist of the financial statements of the subsidiaries, there could be a relationship between the FRQ of the subsidiary and the credit rating of the parent. However my results suggest that there is not a significant relationship between the FRQ of the subsidiary and the credit rating of the parent.

Furthermore, this study contributes to the existing literature on the consequences of board independence. Board independence is a well-studied subject. Prior literature documented a positive relationship between board independence and FRQ, because independent boards are better monitors of the management (Kao and Chen, 2004; Alves, 2014; Klein 2002; Davidson et al., 2005). However, those papers looked at it in stand-alone firms and not in a MNE context. It is important to investigate the influence of board independence on the relationship between the FRQ of the subsidiaries and the credit rating of the parents. First, since MNEs have incentives to manage their earnings through their subsidiaries (Beuselinck et al., 2019). Second, because there could be a relationship since, due to the consolidation process, the parent’s board of directors indirectly monitors the FRQ of the subsidiaries when they monitor the FRQ of the parent. This could result in a moderating effect of board independence on the relationship between the FRQ of the subsidiaries and the credit rating of the parents .

Therefore this research also contributes to the research area of board independence with the MNE approach. My results however suggest that board independence of the parent does not have a significant moderating effect on the relationship between the FRQ of the subsidiaries and the parent’s credit rating. This could mean that the indirect monitoring of the subsidiary FRQ is limited.

The rest of the paper is organized as follows: In section 2, the theoretical framework and the hypotheses are developed. In section 3, the research methodology will be discussed. In section 4, the results of the empirical tests are reported. Lastly, in section 5, I will conclude and discuss this research.

2. Theoretical Framework

One of the most important concepts in this research is FRQ. This is a complex concept and therefore different definitions of FRQ have been used by previous researchers. Biddle et al. (2009) define FRQ as the precision with which the financial reports convey information about the firms operations, to inform the investors. Jonas and Blanchet (2000) define FRQ as ‘full and transparent financial information that is not designed to obfuscate or mislead user s’ (p.357). The above mentioned definitions are consistent with the objective stated by the International Accounting Standards Board (IASB), namely ‘to provide financial information about the reporting entity that is useful to present and helps potential e quity investors, lenders and other creditors in making decisions in their capacity as capital providers ’ (IASB, 2008). FRQ is reduced when earnings management occurs because the use of earnings management leads to more errors in accrual estimation that need to be corrected or reversed in the future, thus reducing earnings persistence and, ultimately, financial reporting qualit y (Dechow and Dichev, 2002). Besides this, earnings management could be used in order to misinform investors. This will reduces FRQ because through the use of earnings management the accounting numbers become less useful for decision makers (Kouki et al., 2011).

There are several incentives for firms to manage earnings. Firms closer to violation of their debt covenants make more use of earnings management than firms that are not, because earnings management enables firms to avoid violation (Franz et al., 2014). In line with this, Jha (2013) found that managers manage earnings upwards in the period before a

debt-covenant violation and downwards in the period that a violation occurs and they proceed with this while the firm remains in violation. This is because in the pre-violation period managers want to avoid the violation and in the period of the violation and while the firm remains in violation the managers have incentives to manage earnings downward because this might improve the bargaining position of their firm during renegotiations. So the literature suggests that debt covenant violation could be an incentive for a firm to manage their earnings.

Another incentive to manage earnings is stated by Burgstahler & Dichev (1997). They concluded that firms use earnings management to smooth their earnings and particular to get from a small loss to a small profit. An important reason why firms smooth their earnings is that earnings volatility is an important factor for rating agencies and for investors. Rating agencies prefer firms with smooth earnings, because than they have more faith that the firm is able to pay of their debtors in the long run (Jung et al., 2013). DeAngelo et al. (1996)

documented that firms that break a consistent earnings growth suffer from a high negative abnormal stock return in the year that the pattern is broken. Zong et al. (2007) studied another incentive to manage earnings, namely pressure from block holders. They concluded that when managers fear negative repercussions for declining performance from stakeholders, they make more use of aggressive earnings management.

Prior literature have addressed several ways in which good FRQ is important for firms. Biddle and Hilary (2006) found that FRQ is positively associated with higher investment efficiency because it mitigates information asymmetries. Garcia-Sanchez & Martinez-Ferrero

(2016) stated that good FRQ has a positive effect on how shareholders, market, investors and other stakeholders assess the firm. Furthermore, FRQ has a positive effect on firm

performance. Firms that try to reduce information asymmetries through high FRQ have better firm performance than firms with low FRQ, because high FRQ is among other things

associated with higher investment efficiency (Bushman and Smith, 2001; Ahmed and Duellman, 2011).

Another concept that is important for firms is their credit risk. This is the willingness and ability of a firm to pay of their debt (Standard & Poor’s, 2003). A form in which investors determine the credit risk of borrowers is through their credit rating (Stumpp, 2001). Cha et al. (2016) defined credit ratings as that they ‘represent the opinions of rating agencies about credit risk, which is the ability and willingness of a borrower or an issuer of a debt security, such as a corporation, state, or city government, to meet its financial obligations.’(p.621) The creditworthiness of the firm is determined by assessing if the future cash flows of the firm will be sufficient to pay the debt service costs. When the future cash flows of the firm become lower, the probability of default will increase and this will result in lower credit ratings for the firm (Ashbaugh-Skaife et al., 2006). Credit ratings are important because the higher the credit rating the lower the cost of debt will be (Demirtas & Cornaggia, 2013; Graham & Harvey, 2001). This is because other firms determine how much and under which terms they want to do business, based on the credit ratings of the firm. This is because a credit rating is a reflection of the ability and willingness of the firm to pay of their debt (Boot et al., 2006). Furthermore, Boot et al. (2006) stated that credit ratings get their value from two features, namely: (i) the monitoring role of credit rating agencies and (ii) the role that credit ratings play in the decisions of institutional investors. The monitoring role constitutes that the rating agency could get into an implicit contract with a firm, where the firm promises to take actions so that their credit rating will stay at the same level or improves. The role that credit ratings play in the decisions of institutional investors is mostly that these investors base their investment decisions on the rating that is given by the rating agency. Some investors for instance require a minimum rating, otherwise they won’t invest in the firm.

A small amount of papers looked at the relationship between credit ratings and earnings management in stand-alone firms and documented a negative relationship. For instance Ashbaugh-Skaife et al. (2006) found that credit ratings are positively related to accrual quality, because high accrual quality leads to less opportunism of managers and better transparency of the financial statements. In line with this, credit rating agencies claim that they take into account the quality of accounting information in determining the credit ratings of the firms (Standard & Poor’s, 2003). Building on this, Demirtas & Cornaggia (2012) found that current accruals are abnormally high around initial credit ratings which means that firms have an incentive to use earnings management to influence their credit ratings. Furthermore, Qi et al. (2010) documented that accrual quality has a positive effect on credit ratings, because a high accrual quality leads to a higher bond liquidity.

However, the relationship between FRQ and credit ratings could be different in MNEs, because they have some other opportunities to manage earnings than stand-alone companies.

One opportunity they have is that it could be more difficult for investors and other users of the financial statements to discover earnings management in the consolidated financial statements, because of the complexity of their statements in comparison with stand-alone companies (Prencipe, 2012). Furthermore, due to multinationals global presence they have more opportunities to manage earnings in other countries. For instance, in countries were the legal/judicial system is less developed it is more easy to get away with earnings management (Bushman & Piotroski, 2006). Besides, the reputational and legal penalties that are associated with misreporting are likely lower when the misreporting takes place at the subsidiary level than when it takes place at the parent level (Dearborn, 2009). In line with those opportunities, Durnev et al. (2017) stated that MNEs prefer to manage earnings abroad in order to avoid the pressure from the companies’ local monitors. Therefore, it is interesting to look at the FRQ of the subsidiaries instead of the FRQ of the parent in relation with the credit ratings of the parent.

Besides that MNEs have incentives and prefer to manage their earnings throu gh their subsidiaries, they also have the power to exercise it. For instance, Robinson and Stocken (2013) found that MNE-parent firms have a great amount of influence over the business decisions that are being made by their subsidiaries. In addition, parents may insert influence on reporting choices through the reporting guidelines that they impose on their subsidiaries (Principe, 2012). This means that they are able to manage earnings through their subsidiaries (Beuselinck et al., 2019). Due to the consolidation process, the consolidated financial

statements of the parents, which are used to determine the credit rating of the parent

(Standard & Poor’s, 2003), contain the financial statements of their subsidiaries (Beuselinck et al., 2019; Sutton, 2004). So, the FRQ of the subsidiaries is likely to influence the FRQ of the MNE as a whole. Therefore, I expect that a higher subsidiary FRQ will lead to a higher parent FRQ. This reduces the uncertainty about the credit risk of the parent and leads to higher transparency of the financial statements of the parent, which will result in a higher parent’s credit rating. Based on the above arguments I formulate the following hypothesis: H1: There is a positive relationship between the FRQ at the subsidiary level and the credit

rating of the MNE.

Earnings management is considered as a type of agency cost because managers will look after their own interests when using earnings management (Healy and Wahlen, 1999). This will result in financial reports that are not an accurate picture of the firm and therefore investors can make detrimental decisions, which will lead to agency costs (Prior et al., 2008). The agency view states that a board performs a monitoring role and can help to educate self-interested managers, through monitoring and advice giving (Fama & Jensen, 1983). Besides this, agency theory states that organizations with independent boards are better able to control agent self-interested behaviour and therefore reduce agency costs (Eisenhardt, 1989).

Therefore, in this study I use the agency view to come to the effect of board independence on the relationship between subsidiary FRQ and MNE credit ratings.

As already mentioned, an important activity of the board is that they monitor the managers of the firm. They monitor the management to prevent managers to pursue their private interests instead of those of the shareholders (Fama & Jensen, 1983; Eisenhardt, 1989). Another important function is their resource and support role in the decision making process within the firm because of the experience that board members have. They have a good amount of experience, because most of the time they are also former high positioned managers in other firms (Fama & Jensen, 1983). Boards consist of dependent and independent directors. Prior research mostly used the following classification in determining when a director is

independent or dependent: A director is a dependent director when he or she is employed by the firm and or has shares or family ties with for instance a manager in th e firm. He or she is independent when their only relationship with the firm is through the board. (Bhagat & Black, 2001; Xie et al., 2003; Byrd and Hickman, 1992). Independence of boards is found to be an important characteristic for the quality of the corporate governance (Ning et al., 2014). The board is more likely to ensure that decisions are made in the best interest of the shareholders when they are mostly composed of independent directors (Fama & Jensen, 1983). The main reason of this is that independent directors are better monitors of the management because they are not dependent on the managers of the firm and therefore more critically monitor management (Eisenhardt, 1989). This more critically monitoring of the management will result in less opportunities for the management to use earnings management in their financial reporting (Xie et al., 2003; Alves, 2014; Black & Kim, 2012; Jensen & Meckling, 1976). Prior research shows that board independences increases FRQ. For instance Kao & Chen (2004) found that outside directors are more independent from the management of the firms and are therefore better monitors of the management. They also found that firms with boards that have a higher proportion of independent directors are less likely to engage in earnings management than firms with boards with a lower proportion of independent directors. Alves (2014) investigated the relationship between an independent board and earnings quality within a sample of Portuguese listed firms. She found that independent board members reduce earnings management and therefore increase earnings quality. In line with this, Klein (2002) stated that earnings management is negatively associated with board independence. Peasnell et al. (2000) found empirical support that board independence is associated with a decrease of earnings management in UK firms. In line with this, Davidson et al. (2005) also found that a board that contains of a majority of independent directors reduces earnings management.

Other researchers focussed on the relationship between board independence and credit ratings. They found a positive relationship between board independence and credit ratings , because an independent board can reduce default risk by reducing agency costs and

monitoring management (Alali et al., 2012;Ashbaugh-Skaife et al., 2006; Bhojraj &

Sengupta, 2003). So prior research suggest that board independence is positively associated to both FRQ and credit ratings.

Besides this, a parent’s board is likely to be concerned about the FRQ of the subsidiaries, since the managers of the parent use the individual financial statements of the subsidiaries to

draw up the consolidated reports (Beuselinck et al., 2019). Beuselinck et al. (2010)

documented that governance characteristics of the MNE could have an effect on the FRQ of their subsidiaries. More specifically, ownership structure and analyst coverage of the MNE -parent affect the extent of earnings management on the subsidiary level (Beuselinck et al., 2010). An independent board of the parent could result in more transparent financial

statements of the subsidiaries because the parent uses the individual financial statements of the subsidiaries to draw up the consolidated reports (Sutton, 2004). This results in a relation between the FRQ of the parent and the FRQ of the subsidiary. So when a parent ’s board monitors the FRQ of the parent they indirectly also monitor the FRQ of the subsidiaries. Th is will result in more transparent financial statements. This will lead to less default risk and credit rating agencies who will be better able to take into account the FRQ of the parents and indirectly the FRQ of the subsidiaries. Subsequently this will lead to higher credit ratings of the parent. Based on the above arguments I formulate the following hypothesis:

H2: The positive relationship between the FRQ at the subsidiary level and the credit rating of

the MNE is stronger when the board of the MNE consist of a majority of independent directors.

3. Research Methodology

The sample selection started by identifying all listed firms from the US using Compustat. Next, I discard the financial firms because the high leverage that is normal in cases of financial firms do not have the same meaning for non-financial firms and it is common practice to exclude financial firms (Klein, 2002; Beuselinck et al., 2019; Dedman & Kausar, 2012). I also discard firms that do not have the necessary financial information to carry out the research. Furthermore, I exclude firms with missing credit ratings and board related information. Financial information and credit ratings of the parents are gathered from

Compustat, while the information about the parent’s board is gathered from the MSCI (former KLD and GMI) database. Then, for these listed non-financial firms I hand collect the names and jurisdictions of all material subsidiaries included in exhibit 21.1 or 21 of the 10-k fillings from Edgar. In the following step I match the names and countries of the material subsidiaries with those of the firms covered by Orbis. I select only subsidiaries located in 30 European countries (EU, Norway and Switzerland). Then, I delete subsidiaries operating in the financial industry because it is difficult to define abnormal accruals for financial services firms (Klein, 2002). Finally, after discarding firms with unavailable financial information, the final sample consists of 1143 parent firm-year observations of 245 non-financial public U.S. MNEs and 15351 corresponding material subsidiary firm-year observations during the period 2011-2017. The financial information of the subsidiaries is collected from the Orbis database. Table 1 and 2 provide further details about the distribution of my sample.

Table 1 shows the distribution of the sample by year. It shows that the years are somewhat equally distributed, but with a small overrepresentation of the years 2015 -2017 and a small underrepresentation for the years 2011-2013.

Table 1: Sample distribution by year

Year Freq. Percent

2011 133 11.64 2012 140 12.25 2013 145 12.69 2014 160 14.00 2015 183 16.01 2016 201 17.59 2017 181 15.84 Total 1,143 100.00

Table 2 presents the distributions of the parent firm-year observations by industry. The firms in the sample are located in 7 different industries. The most represented industries are

Table 2: Sample distribution by industry in which MNEs operate

SIC industry title Freq. Percent

A: Mining 37 3.24

B: Manufacturing 809 70.78

C: Transportation, Communications, Electric, Gas and Sanitary Services 10 0.87

D: Wholesale Trade 45 3.94 E: Retail Trade 43 3.76 F: Services 192 16.80 G: Public administration 7 0.61 Total 1,143 100.00 3.2 Variables Parents credit ratings

The dependent variable in this research is PAR_RATING. I specify my measure of the parent’s credit ratings, by converting the long-term issuer credit ratings from Standard & Poor into an ordinal scale. Standard & Poor calculate their long-term credit rating based on their opinion about the credit risk of the company, this means if the company is able and willing to meet their financial obligations to their debtors in the long term (Cha et al., 2016). The highest long-term credit rating is AAA and means that the firm’s capacity to meet its financial commitments on their loans is extremely strong. On the other hand, the lowest long-term credit rating is D and means that the obligation is in default or in breach of a promise, for instance when payments for the obligation are not made on the date due (Standard & Poor’s, 2003). In accordance with prior literature, I categorize the ratings from 0 to 9. I assign a value of 9 if the firm has an rating of AAA (the highest) and 0 if the firm has an rating of D (the lowest) (Ashbaugh-Skaife et al., 2006; Bhojraj and Sengupta, 2003; Attig et al., 2013). I categorize the credit ratings based on letters and not based on the plus or minus in the rating, because the sign is only to show the status of the rating within the rating category. Beside this, Standard & Poor also give their rating definitions based on only the letters and not also based on the sign behind the letter. (Ashbaugh-Skaife et al., 2006; Standard & Poor’s, 2003). To give some more insights, in line with Ashbaugh-Saife et al. (2006) I made a dummy variable PAR_RATINGinvest, which is given a 1 if the credit rating is BBB or above

(investment grade) and 0 otherwise. The difference between investment grade and junk grade ratings is that for companies with junk grade ratings it is speculative if they will be able to repay their debt (Standard & Poor’s, 2003).

FRQ of the subsidiary

The main independent variable in this study is the FRQ of the foreign subsidiaries of the MNEs. Prior research often measures FRQ by the level of earnings management (Beuselinck et al., 2019; Biddle et al., 2009; Durnev et al., 2017) Discretionary accruals are often used to measure the level of earnings management in a firm (Dechow et al., 1995; Biddle et al., 2009; Xie et al., 2003; Kao & Chen, 2004). Different models are used to measure the discretionary accruals. To estimate the discretionary accruals prior literature often use the original Jones

model (Jones, 1991) or the modified Jones model (Dechow et al., 1995). The modified Jones model decomposes accruals into discretionary and non-discretionary accruals (Dechow et al., 1995). The original Jones model (Jones, 1991) only looks at earnings management which is exercised through non-discretionary accruals. To be able to also capture earnings management through discretionary accruals, the modified Jones model of Dechow et al. (1995) adjusts the change in revenues for the change in receivables. The modified Jones model implicitly

assumes that all changes in the credit sales result from earnings management. This is based on the reasoning that “it is easier to manage earnings by exercising discretion over the

recognition of revenue on credit sales than it is to manage earnings by exercising discretion over the recognition of revenue on cash sales” (Dechow et al., 1995 p199). I use the modified Jones model from Dechow et al. 1995 for two reasons namely: (i) the model is more powerful in detecting earnings management than the original Jones model (Dechow et al., 1995) and (ii) the model is commonly used in prior literature (Klein, 2002; Kao & Chen, 2004). To be able to estimate the discretionary accruals, I calculate the total accruals with the following equation:

𝑇𝐴𝐶𝐶_𝑡 = 𝛥𝐶𝐴_𝑡− 𝛥𝐶𝑎𝑠ℎ_𝑡 − 𝛥𝐶𝐿_𝑡+ 𝛥𝐷𝐶𝐿_𝑡− 𝐷𝐸𝑃_𝑡 (1) Where:

TACCt = Total accruals in year t,

ΔCAt = Change in current assets in year t,

ΔCasht = Change in cash and cash equivalents in year t.

ΔCLt = Change in current liabilities in year t,

ΔDCLt = Change in short term debt included in current liabilities in year t,

DEPt = Depreciation and amortization expense in year t.

To estimate discretionary accruals of the subsidiaries, I use the modified Jones model (Dechow et al., 1995): TACCt TAt−1 = α1 1 TAt−1+ α2 (ΔREV−ΔRECt) TAt−1 + α3 ΔPPEt TAt−1+ ϵt (2) Where:

TACCt = Total accruals in year t,

ΔRECt = The change in receivables in year t,

PPEt = the property, plant and equipment in year t

TAt-1 = the total assets in year t-1

ϵt = the estimated discretionary accrual in year t

Equation (2) is estimated for each industry and year combination. I use the absolute value of discretionary accruals because I am interested in the extent of earnings management, rather than its direction. This is because both income increasing and income decreasing earnings management reduces the FRQ of a firm (Klein, 2002). Furthermore, since I am interested in how subsidiary FRQ influences parent’s long term credit ratings, I compute a compound measure of subsidiary FRQ named SUB_ DISCACC. This measure is the average

discretionary accruals in absolute value of all subsidiaries of a given parent in a given year. Parent board independence

The moderating variable in this research is PAR_BOARDIND which proxies for the

independence of the parent’s board. According to Klein (2002) a board is independent when a majority (i.e. at least 50.01%) of the members is independent of management. Similarly Dechow et al. (1996) argue that a board is inside-dominated if at least 50.01% of the

members are insiders. They stated this on the basis of that the majority rule dominates board actions. Therefore the 51% threshold is commonly used to define board independence. I define PAR_BOARDIND as a dummy variable equal to 1 when the board consists of a minimum of 50.01% independent directors, and 0 when not. Besides this, I also measure the parent’s board independence as the percentage of independent directors from the total number of board members (Klein, 2002). This variable is named PAR_BOARDINDperc.

Control variables

Prior research documents several firm characteristics that could have an effect on the long-term credit ratings of the firms. For instance, larger firms have lower market risk and are therefore expected to have higher ratings (Attig at al., 2013; Bhojraj & Sengupta, 2003; Ashbaugh-Saife et al., 2006; Oikonomou et al., 2014). Therefore, I include PAR_Size as a control variable. This variable is measured through the natural logarithm of total assets of the parent at the end of year t.

Also, firms with higher profitability are expected to have higher ratings because they are seen as better able to make a profit on their assets and therefore better able to pay their debtors (Attig at al., 2013; Bhojraj & Sengupta, 2003; Ashbaugh-Saife et al., 2006; Oikonomou et al., 2014). Therefore, I include PAR_Margin as a control variable. This variable is calculated through dividing the operating income by total assets of year t from the parent.

Besides this, firms that have higher (lower) debt to equity ratios are expected to have lower (higher) credit ratings because they have more (less) debt that they have to pay (Attig at al.,

2013; Bhojraj & Sengupta, 2003; Ashbaugh-Saife et al., 2006; Oikonomou et al., 2014). Because of this, I include PAR_Leverage as a control variable. This variable is measured by the ratio of long term debt to total assets.

Furthermore, firms who make a loss are expected to have lower ratings because they have less earnings to use to pay of their debt and therefore their market risk is higher than firms who make a profit (Attig at al., 2013; Bhojraj & Sengupta, 2003; Ashbaugh-Saife et al., 2006). For this reason I include PAR_Loss as a control dummy variable which is set to 1 if net income of the parent is negative and 0 otherwise.

Also, firms who have a high sales growth are expected to have higher ratings because firms with high sales growth are seen as better being able to pay their debt and therefore have a smaller credit risk (Dedman & Kausar, 2012). This is mainly because high sales growth could be seen as that the firm is performing well and that they have a better continuity. Thus, I control for sales growth (PAR_Salesgrowth), calculated by dividing parent’s sales of year t-1 by parent’s sales of year t.

In addition, firms with higher ratios of fixed assets to total assets are expected to have higher ratings because fixed assets may be used as collateral (Attig at al., 2013; Bhojraj & Sengupta, 2003; Ashbaugh-Saife et al., 2006). Therefore I include PAR_Capint. This variable is

calculated by the ratio of property, plant and equipment to total assets.

Being audited by a Big 4 audit firm (PWC, KPMG, Deloitte and EY) is expected to positively influence the credit rating of the firm. This is because credit rating agencies link being

audited by a Big 4 to higher reliability of the financial statements and can therefore base their opinion about the credit risk of the company on more reliable information (Dedman &

Kausar, 2012). Hence, I add a dummy variable to control for firms which are audited by a Big 4 accountant (PAR_Big4), with a value of one if the parent’s auditor is Big 4 and 0 otherwise. To conclude, I control for industry (Industry dummies) and year (Year dummies) effects by including dummy variables, because default risk may vary between industries (Oikonomou et al., 2014), I include the industry fixed effects by making use of the first two digits of the industry SIC codes (Klein,2002). Table 3 provides an overview of all the variables used and their definitions.

Table 3: Variables used

Variable Definition

PAR_RATING The level of which the company is able and willing to meet their financial obligations to their debtors in the long term. Furthermore categorized based on the letters and not on the signs, assigned a value of 9 if the rating is AAA (the highest) and a 0 if the rating is D (the lowest).

PAR_RATINGinvest Dummy variable set to 1 if the credit rating is BBB or above and 0 otherwise.

SUB_DISCACC The average discretionary accruals in absolute value of all subsidiaries of a given parent in a given year. Estimated with the modified Jones model.

PAR_BOARDIND Dummy variable equalling 1 if the board of the parent is

constituted of a majority of independent directors and 0 otherwise.

PAR_BOARDINDperc The number of independent directors divided by the total number

of parent board members.

PAR_ Size The natural logarithm of total assets of the parent at the end of year t.

PAR_Margin Parent operating income divided by parent total assets of year t .

PAR_Leverage Parents ratio of long-term debt to total assets.

PAR_Loss Dummy variable set to 1 if net income of the parent is negative and 0 otherwise.

PAR_Capint Parents ratio of property, plant and equipment to total assets

PAR_Big4 Dummy variable set to 1 if the parent’s auditor is Big 4 and 0 otherwise.

3.3 Empirical model

Because the dependent variable, PAR_RATING, is an ordinal variable I use an Ordered logit regression. I constructed two models to be able to investigate both the relation between subsidiary FRQ and the parent’s credit rating (model 3) and the effect of the independence of the parents board on this relation (model 4). Based on the first hypothesis, I expect for the main independent variable, SUB_DISCACC, a negative sign. Based on the second hypothesis, I also expect a negative sign for the interaction term, namely the SUB_DISACC multiplied with the PAR_BOARDIND.

PAR_RATING = β0 + β1SUB_DISACC + β2PAR_Size + β3PAR_Margin + β4PAR_Leverage + β5PAR_Loss + β6PAR_Capint+ β7PAR_Big4 + β8PAR_Salesgrowth + β9Year dummies +

β10Industry dummies + ɛ (3)

PAR_RATING = β0 + β1SUB_DISACC + β2PAR_BOARDIND + β3SUB_DISACC *

PAR_BOARDIND + β4PAR_Size + β5PAR_Margin + β6PAR_Leverage + β7PAR_Loss + β8PAR_Capint + β9PAR_Big4+ β10PAR_Salesgrowth + β11Year dummies + β12Industry

dummies + ɛ (4)

To give more insights, I also constructed the following models with PAR_RATINGinvest and

PAR_BOARDINDperc. For models 5 and 6 I use a logistic regression, because here the

dependent variable is a dummy variable. The expected signs for the dependent variable and the interaction term remain negative.

PAR_RATINGinvest = β0 + β1SUB_DISACC + β2PAR_Size + β3PAR_Margin +

β4PAR_Leverage + β5PAR_Loss + β6PAR_Capint+ β7PAR_Big4 + β8PAR_Salesgrowth +

β9Year dummies + β10Industry dummies + ɛ (5)

PAR_RATINGinvest = β0 + β1SUB_DISACC + β2PAR_BOARDINDperc + β3SUB_DISACC * PAR_BOARDINDperc + β4PAR_Size + β5PAR_Margin + β6PAR_Leverage +

β7PAR_Loss + β8PAR_Capint + β9PAR_Big4+ β10PAR_Salesgrowth + β11Year dummies +

4. Results

4.1 Descriptive statistics

Table 4 provides the descriptive statistics for all the variables used. I winsorized all the continuous variables at the 5% and 95% level to eliminate the effects of outliers. The

dependent variable (for model 3 and 4), parent’s credit ratings (PAR_RATING), has a mean of 5.794, a minimum of 3.000 and a maximum of 9.000. The mean of 5.794 means that the average credit rating in my sample is approximately BBB (which is an investment grade rating and means that the firm has adequate capacity to meet its financial commitments) . The minimum of 3.000 means that the lowest credit rating in my sample is CCC and the maximum of 9.000 implies that the highest credit rating in my sample is AAA. The main independent variable (SUB_DISCACC), computed as the average discretionary accruals in absolute value of all the parent’s subsidiaries, has a mean of 0.180 and a standard deviation of 0.115. The moderator (for model 3 and 4), the independence of the board (PAR_BOARDIND), has a mean of 0.918, which means that 91.8% of the parents in my sample have a board which is constituted of a majority of independent directors.

The dependent variable (for model 5 and 6), PAR_RATINGinvest, has a mean of 0.583, which means that 58.3% of the firm years in my sample got an investment grade long-term credit rating (BBB or higher). The moderator (for model 5 and 6), PAR_BOARDINDperc, has a mean of 0.783, which means that the average percentage of independent directors on the parents boards in my sample is 78.3%.

Furthermore, the mean of PAR_Size is 8.831, which means that the average natural logarithm of total assets in the sample is 8.831. Moreover, the parent firms in my sample have, on average, a profit margin of 14.6%. Besides this, the mean of the variable PAR_Leverage is 0.280, which means that the long-term debt of the firms in my sample represents for an

average of 28% total assets. Furthermore, the mean of 0.121 of the variable PAR_Loss, means that only 12.1% of the firm years in my sample reported a loss. The variable PAR_Capint has a mean of 0.429, which means that in my sample the average PPE, accounts for 42.9% of total assets. Furthermore, based on the PAR_Big4 mean of 0.974, I can state that 97.4% of the firms in my sample are audited by a Big 4 auditor (PWC, KPMG, Deloitte or EY). To

conclude, the 0.031 mean of PAR_Salesgrowth shows that the firms in my sample on average see their sales grow by 3.1% per year. However looking at the min of -0.178 and the

percentile 25 of -0.023, more than 25% of my observations correspond to parents that suffered a decrease in sales. The descriptives of my sample are similar to those reported in prior literature (Ashbaugh-Saife et al., 2006; Attig et al., 2013).

Table 4: Descriptive statistics

VARIABLES N Mean Median Std Dev Min 25% 75% Max

PAR_RATING 1,143 5.794 6.000 1.119 3.000 5.000 6.000 9.000 PAR_RATINGinvest 1,143 0.583 1.000 0.491 0.000 0.000 1.000 1.000 SUB_DISCACC 1,143 0.180 0.156 0.115 0.030 0.098 0.235 0.473 PAR_BOARDIND 1,143 0.918 1.000 0.275 0.000 1.000 1.000 1.000 PAR_BOARDINDperc 1,143 0.783 0.818 0.153 0.000 0.706 0.892 1.000 PAR_Size 1,143 8.831 8.669 1.242 6.796 7.949 9.628 11.430 PAR_Margin 1,143 0.146 0.136 0.060 0.058 0.101 0.178 0.286 PAR_Leverage 1,143 0.280 0.266 0.131 0.090 0.175 0.363 0.567 PAR_Loss 1,143 0.121 0.000 0.327 0.000 0.000 0.000 1.000 PAR_Capint 1,143 0.429 0.330 0.288 0.073 0.198 0.619 1.037 PAR_Big4 1,143 0.974 1.000 0.160 0.000 1.000 1.000 1.000 PAR_Salesgrowth 1,143 0.031 0.032 0.100 -0.178 -0.023 0.080 0.240 The variable definitions are provided in Table 3.

4.2 Correlations

Table 5 provides the Spearman’s rank correlations for all the variables used in this research. I used the Spearman correlations because my dependent variable PAR_RATING is an ordinal variable. To start with, the coefficient of the correlation between SUB_DISCACC and

PAR_RATING is not statistically significant, suggesting that the parent credit rating is not

associated with the average magnitude of subsidiary discretionary accruals. Next, the results show that PAR_RATING is positively and significantly correlated with both PAR_BOARDIND and PAR_BOARDINDperc at the 1% level, which means that as expected when the board of the parent consists of a majority (higher proportion) of independent directors the parent’s credit rating is better. Moreover, the correlation between PAR_BOARDIND and

SUB_DISCACC is not statistically significant, suggesting that the parent’s board

independence is not associated with the average magnitude of subsidiary discretionary accruals.

Looking at the dependent variable for model 5 and 6 (PAR_RATINGinvest), there is also no significant correlation with SUB_DISCACC. Furthermore, the results show that

PAR_BOARDINDperc is positively and significantly correlated with PAR_RATINGinvest at

the 1% level. Moreover the correlation between PAR_BOARDINDperc and SUB_DISCACC is not statistically significant, suggesting that the parent’s board independence percentage is not associated with the average magnitude of subsidiary discretionary accruals. So, these

correlations are in line with the correlations found with the PAR_RATING and the

PAR_BOARDIND, that are given above.

Furthermore, the correlations indicate that PAR_Size, PAR_Margin and PAR_Big4 are

significantly positive correlated with PAR_RATING at the 1% level. This suggest that the credit rating is higher for larger and more profitable firms and for those audited by a Big4 auditor.

PAR_Leverage and PAR_Loss are significantly negative correlated with PAR_RATING at the 1%

level and PAR_Salesgrowth is significantly negative correlated with PAR_RATING at the 5% level. This suggest that the credit rating is lower for firms with higher leverage and sales growth and for those who make a loss. Besides this, PAR_Capint is not significantly related with

PAR_RATING, indicating that capital intensity does not influence credit ratings. These

correlations are the same when I use the PAR_RATINGinvest as the dependent variable. Lastly, there are two variables which have a correlation coefficient higher than 0.7, but these variables (PAR_RATING and PAR_RATINGinvest) are substitutes of each other and are therefore not used in the same model. So multicollinearity is not a problem in my sample.

A2: PAR_RATINGinvest 0.889*** 1.000 B: SUB_DISCACC -0.029 -0.012 1.000 C: PAR_BOARDIND 0.087*** 0.048 0.005 1.000 C2: PAR_BOARDINDperc 0.224*** 0.179*** 0.002 0.453*** 1.000 D: PAR_Size 0.627*** 0.524*** 0.047 0.021 0.145*** 1.000 E: PAR_Margin 0.419*** 0.325*** -0.054* 0.049* 0.068** 0.044 1.000 F: PAR_Leverage -0.358*** -0.303*** 0.013 -0.035 -0.024 -0.204*** -0.052* 1.000 G: PAR_Loss -0.309*** -0.273*** -0.010 -0.025 -0.082*** -0.157*** -0.385*** 0.184*** 1.000 H: PAR_Capint -0.018 0.001 -0.126*** -0.021 -0.115*** -0.053* 0.264*** 0.043 0.028 1.000 I: PAR_Big4 0.192*** 0.192*** -0.051* -0.009 0.074** 0.178*** 0.144*** -0.149*** -0.240*** 0.127*** 1.000 J: PAR_Salesgrowth -0.074** -0.082*** 0.053* -0.006 -0.099*** -0.054* 0.078*** -0.014 -0.091*** -0.181*** -0.116*** 1.000 The variable definitions are provided in Table 3.

*= significant at the 10% level **= significant at the 5% level ***= significant at the 1% level

4.3 Regression analysis

Table 6 contains the results of the four regression estimations. For models 3 and 4, I used an Ordered logit regression for panel data, because the dependent variable of these models is measured on an ordinal scale and the observation are panel observations based on parent and year. For models 5 and 6, I used an logistic regression for panel data, because the dependent variable of these models is a dummy variable and the observation are panel observations based on parent and year.

First, I estimate the regression of model 3, which includes the dependent variable

(PAR_RATING), the independent variable and all the control variables used in this study. The results of this regression are showed in column 1. The coefficient of SUB_DISCACC is not statistically significant, which means that there is no relation between the average magnitude of subsidiary discretionary accruals and the parent credit rating. This result does not support hypothesis 1 which predicted a negative relation between the average level of earnings management in foreign subsidiaries and the parent´s credit rating. In terms of control variables, I observe that credit ratings are better for large and more profitable firms. Furthermore, credit ratings are lower for firms with higher leverage and sales gr owth. To be able to further investigate hypothesis 1, I also estimated model 5, which differentiates from model 3 in term of the dependent variable (PAR_RATINGinvest). In line with the results of the regression of model 3, the coefficient of SUB_DISCACC is not statistically significant, which means that there is no relation between the average magnitude of subsidiary

discretionary accruals and the parent’s credit rating being investment grade (BBB or above) or junk grade. This does also not support hypothesis 1. In terms of control variables the results are the same, with one exception: namely in this estimation also credit ratings are better for firms with a higher capital intensity.

Second, I estimate model 4, which includes the dependent variable (PAR_RATING), the independent variable, the moderator (PAR_BOARDIND), the interaction variable and all the control variables used in this study.As reported in column 2, the coefficient of the interaction term is not significant at conventional levels. This implies that the parent’s board

independence does not moderate the relation between the average magnitude of subsidiary discretionary accruals and the parent credit rating. This result does not support hypothesis 2, according to which parent board independence is expected to have a positive moderating effect on the positive relationship between the FRQ at the subsidiary level and the credit rating of the MNE. Lastly, the coefficients of the controls are very similar in comparison to the estimation of model 3. To be able to further investigate hypothesis 2, I also estimated model 6, which differentiates from model 3 in terms of the dependent variable

(PAR_RATINGinvest) and the moderator (PAR_BOARDINDperc). As reported in column 4, in line with the results of the regression of model 4, the coefficient of the interaction term (SUB_DISCACC*PAR_BOARDINDperc) is not significant at conventional levels. This implies that the parent’s level of board independence does not moderate the relation between the average magnitude of subsidiary discretionary accruals and the parent credit rating being investment grade (BBB or above) or junk grade. This also does not support hypothesis 2. In terms of control variables the coefficients are very similar in comparison to the estimation of model

Table 6: Regression analysis

Models 3 4 5 6

VARIABLES PAR_RATING PAR_RATING PAR_RATINGinvest PAR_RATINGinvest

SUB_DISCACC -0.695 2.193 0.816 8.742 (1.316) (4.183) (2.335) (11.979) PAR_BOARDIND 0.931 (0.971) SUB_DISCACC*PAR_BOARDIND -3.235 (4.421) PAR_BOARDINDperc 0.213 (4.925) SUB_DISCACC*PAR_BOARDINDperc -10.829 (15.593) PAR_Size 6.889*** 6.908*** 7.187*** 7.821*** (0.755) (0.758) (0.783) (0.720) PAR_Margin 25.123*** 25.011*** 33.445*** 33.004*** (5.604) (5.601) (10.799) (10.723) PAR_Leverage -16.899*** -16.933*** -17.149*** -19.104*** (2.736) (2.758) (4.684) (5.162) PAR_Loss 0.329 0.341 -0.423 -0.308 (0.537) (0.537) (1.103) (1.159) PAR_Capint 2.409 2.456 10.689*** 12.169*** (2.187) (2.195) (2.945) (3.811) PAR_Big4 2.852 2.863 0 0 (3.185) (3.249) (omitted) (omitted) PAR_Salesgrowth -6.201*** -6.206*** -6.157* -6.208* (1.737) (1.734) (3.423) (3.457)

Year effects YES YES YES YES

Industry effects YES YES YES YES

Pseudo R-square 0.438 0.439 0.513 0.517

Observations 1,143 1,143 1,143 1,143

Standard errors are reported in parenthesis The variable definitions are provided in Table 3. *= significant at the 10% level

**= significant at the 5% level ***= significant at the 1% level

5. Conclusion

This study investigates the relationship between the average FRQ of foreign subsidiaries and the parent’s long-term credit ratings. Based on previous literature, I expected a positive relationship between the FRQ of the subsidiaries and the parent’s long term credit ratings. Credit ratings are positively related to FRQ in stand-alone firms because high FRQ leads to less uncertainty about credit risk (Akins, 2018) and better transparency of the financial

statements (Ashbaugh-Skaife et al., 2006). Furthermore, through the reporting guidelines that parents impose on their subsidiaries, parents insert influence over the reporting choices of their subsidiaries (Principe, 2012). This means that parents are able to manage earnings through their subsidiaries (Beuselinck et al., 2019). Due to the consolidation process, the consolidated financial statements of the parents, which are used to determine the credit rating of the parent (Standard & Poor’s, 2003), contain the financial statements of their subsidiaries (Beuselinck et al., 2019; Sutton, 2004). So, the FRQ of the subsidiaries is likely to influence the FRQ of the MNE as a whole. Therefore, I expect that a higher subsidiary FRQ will lead to a higher parent FRQ. This reduces the uncertainty about the credit risk of the parent and leads to higher transparency of the financial statements of the parent, which will result in a higher parent’s credit rating. Based on this reasoning my H1 predicted that there is a positive

relationship between the FRQ at the subsidiary level and the credit rating of the MNE-parent. The results do not provide support for H1. This means that the FRQ of the subsidiaries does not influence the long-term credit ratings of the parent. This could be due to the fact that credit rating agencies do not look specifically to the statements of the foreign subsidiaries and only to the statements of the parent company. This is in line with that analysts don’t give much attention to subsidiaries in research that they publish because useful information about subsidiaries is sometimes difficult to obtain by analysts (Fieldman et al., 2014). Furthermore, parent firms could find it difficult to appropriately consolidate the financial statements of their foreign subsidiaries because the geographical distance and operational complexity of foreign subsidiaries (Chin et al., 2009). This leads to less transparent financial statements, which makes it harder for external users (like rating agencies) of the financial statements to take into account earnings management and the FRQ of the foreign subsidiaries (Chin et al., 2009). So, this study provides a new insight that subsidiary FRQ is not associated with higher parent’s long-term credit ratings.

Furthermore, I also tested if board independence of the parent’s board affect the relationship between FRQ of the subsidiaries and parent’s long-term credit ratings. Based on prior literature, I predicted that when a majority of the directors on the parent’s board are independent, this would strengthen the positive relationship between subsidiary FRQ and parent’s credit ratings. The managers of the MNE-parent are able to control the behavior of their subsidiaries (Doz & Prahalad, 1981). Extending this, Beuselinck et al. (2010)

documented that governance characteristics of the MNE could have an effect on the FRQ of their subsidiaries. More specifically, ownership structure and analyst coverage of the MNE-parent affect the extent of earnings management on the subsidiary level (Beuselinck et al., 2010). An independent board of the parent could result in more transparent financial

statements of the subsidiaries because the parent uses the individual financial statements of the subsidiaries to draw up the consolidated reports (Sutton, 2004). This results in a relation between the FRQ of the parent and the FRQ of the subsidiary. So when a parent board

monitors the FRQ of the parent they indirectly also monitor the FRQ of the subsidiaries. This increases the transparency of the financial statements of both the parents and the subsidiaries. This will lead to less default risk and credit rating agencies who will be better able to take into account the FRQ of the parents and indirectly the FRQ of the subsidiaries and therefore to higher credit ratings of the parent. Based on this reasoning my hypothesis 2 predicted that independence of the parent’s board strengthens the positive relationship between the FRQ of the foreign subsidiaries and the credit rating of the parents.

My results do not provide support to H2 either. This means that a board with a majority of independent directors (and the proportion of independent directors) does not influence the relationship between subsidiary FRQ and parent’s credit ratings. This could be because of the geographical distance and operational complexity of foreign operations which are likely to impair effective monitoring and clear consolidation of the statements of the foreign

subsidiaries (Beuselinck et al., 2011;Chin et al., 2009). Effective monitoring could become more difficult because the costs are high for the MNE parent’s board to gather information about the foreign subsidiaries (Beuselinck et al., 2011).

This paper contributes to the literature on the determinants of credit ratings. While all other studies look at this relation in stand-alone companies (Ashbaugh-Skaife et al.,2006;QI et al., 2010), I investigate the relation between subsidiary FRQ and parent’s credit ratings. This is important because the FRQ of the subsidiary is related to the FRQ of the parent through the consolidation process and therefore there could be a relationship between the FRQ of the subsidiary and the credit rating of the parent. Furthermore, parents prefer to manage their earnings through their subsidiaries in order to avoid the pressure from the companies’ local monitors (Durnev et al., 2017). My results suggest that credit rating agencies do not take into account the FRQ of the foreign subsidiaries.

Another strand of literature where this paper contributes to, is the literature on the consequences of board characteristics. Where most of the other papers look at board characteristics in stand-alone companies (Kao and Chen, 2004; Alves, 2014; Klein 2002; Davidson et al., 2005). I investigate whether independence of the board of the parent has a moderating effect on the relationship between the FRQ of the subsidiaries and the credit rating of the parent. There are several reasons why it is important to look at this effect. First, MNEs have incentives to manage their earnings through their subsidiaries in order to avoid the pressure from the companies’ local monitors (Durnev et al., 2017). Second, due to the consolidation process, the parents board of directors indirectly monitors the FRQ of their subsidiaries when they monitor the FRQ of the parent. Resulting in a possible effect of board independence on the relationship between subsidiary FRQ and parent credit ratings. My results however suggest that board independence does not moderate the relationship between subsidiary FRQ and parent long-term credit ratings.

A practical implication of this study is that the FRQ of the subsidiaries does not influence the credit rating of the parent. This is important for MNEs, because this could mean that they can manage their earnings in their foreign subsidiaries without having negative results on their credit ratings.

One of the limitations of this study is that the final sample consists for 91.8% of parent companies for which the board has a majority of independent directors. This could influence my results of hypothesis 2, because there are simply not enough parent companies in my sample in which the percentage of independent directors in the board is low , to be able to get a significant moderation with the independence of the board. Therefore I recommend future researchers to use a sample in which there is a higher percentage of boards with a small percentage of independent directors.

Another limitation of this study is the generalizability of the findings since this study only includes US parents. The default risk and therefore the long-term credit ratings of the parents could vary between countries (Oikonomou et al., 2014). Therefore, I recommend future researchers to use parents from other countries or even use an international sample consisting of MNEs headquartered in multiple countries to find generalizable results on the relation between subsidiary FRQ and parent long-term credit ratings.

A further limitation of this study could be that I use accrual based earnings manag ement as a measure for FRQ. Cohen et al. (2008) found that after the passage of the Sarbanes Oxley Act in 2002 there was a significant decline in accrual-based earnings management and a

significant increase in real earnings management. Because my sample co nsists of firm years from 2011-2017 it could be better to measure the FRQ of the subsidiaries based on real earnings management and not based on accrual-based earnings management.

A last limitation of my study could be that my measures of board independenc e only look at whether the director is independent or not and does not account for other personal attributes of independent directors. For instance, director tenure is expected to decrease the

independence of the director, which is expected to decrease their monitoring capabilities (Liu & Sun, 2010). Therefore, I recommend future researchers to also use other characteristics of directors to come to a better measure of board independence.