Debt on Credit Enhancement: Further Evidence on Internal Capital Market Efficiency

Debt on Credit Enhancement: Further Evidence on Internal Capital Market Efficiency Fang Chen Job Market Paper This version: December 2011 Fang Chen is a Ph.D. candidate in Finance at the University of Rhode Island. Email: fchen@mail.uri.edu. I would like to thank my thesis committee, Dr. John Burkett, Dr. Shaw Chen, Dr. Gordon Dash, Dr. Orlando Merino, Dr. Yan Xu (co-chair) and Dr. Tong Yu (chair) for their direction, dedication, and invaluable advice along this project.

Debt on Credit Enhancement: Further Evidence on Internal Capital Market Efficiency Abstract I examine the effect of internal capital allocation through credit enhancement (CEs) on investment and firm value, and provide a new evidence of internal capital market s inefficiency. An obvious benefit of CEs is the reduction of financial constraints induced underinvestment through liquidity pooling, on the other hand, a cost of CEs is the amplified overinvestment through the weakened monitoring role of debt. Using a sample of public corporate debts with CEs from 1990 to 2009, I find that CE is negatively related to firm value and the firms issuing debt with CEs underperform those without, suggesting the cost dominates the benefit. I also find that the extent of negative impact of CE is lower in the firms with higher growth opportunities and higher debt levels. Our finding remains strong after controlling a set of firm characteristics and correcting the self-selection bias. Further test reveals a negative effect of credit enhancement on the stock return upon the bond offering announcements. Key words: Internal Capital Markets; Credit Enhancements; Agency Problems; Overinvestments; Underinvestments

1 Introduction A key feature of internal capital markets is cross subsidization. That is, conglomerates have the potential to put together resources within the corporate border. The outlet of such resources is however under heated debate. Studies supporting the internal capital market efficiency contend that a key advantage of an internal capital market is that it shields investment projects from the information and incentive problems that plague external finance (e.g., Alchian, 1969, Williamson, 1975, Gertner, Scharfstein, and Stein, 1994, and Stein, 1997). Supportive to this argument, Khanna and Tice (2001) show that internal capital allocation in diversified firms functions efficiently by tying up capital to investment opportunities. A competing line of research, in contrast, argues that internal capital market can hinder investment efficiency. Scharfstein and Stein (2000) and Rajan, Servaes, and Zingales (2000), among others, highlight the effect of agency problems and power grabbing to generate inefficient cross subsidization across projects. The vast literature on the internal capital market efficiency mainly focuses on the investment efficiency of internal capital markets. Little attention is paid to the role of financing of internal capital markets. In this study, I explore the financing side of the internal capital markets with a sample of credit enhancements (CEs thereafter) firms. With an CE arrangement, bond issuers acquire a guaranty from a third party to secure their payments when bond issuers are at default. A salient fact about CE is that up to this day, the lion s share of CEs is conducted within the corporate border, either through the guaranty of parents for subsidiaries or the guaranty of subsidiaries for their parents. Therefore CE within conglomerate is an unambiguous operation conducted through internal capital markets: one division s assets are used as collateral to raise financing that is diverted to other divisions. Therefore if CE affects firm valuation, then studying CE can further our understanding about the efficiency of internal capital markets. It is highly likely that use of CE will affect firm valuation. The cross subsidization 1

of the internal capital markets is clearly reflected by use of CE. By combining the divisional cash flows into a smooth aggregate cash flow, firms can raise their debt capacity and enjoy tax benefits (Hege and Ambrus-Lakatos, 2005). In the case of conglomerates, the parent pools the liquidity and channel the funds to subsidiaries with worthy projects through an efficient internal capital market. There is also a dark side for use of CEs. Measured in terms of par value, only a small fraction of corporate bonds were issued with CEs. It may seem puzzling that no more firms consider CEs. After all, the direct cost is out of proportion relative to the potential benefit. With the internal guaranty arrangement, firms can considerably reduce the expensiveness of external capital. In contrast, the direct cost associated with the use of CE includes only the small cost of filing to SEC. I conjecture that there must exist some indirect costs related to procurement of CEs, and it must potentially be associated with the existence of internal capital market. Our research questions then become, first, as an internal capital market arrangement, would CE impact firm value? If so in which way? Second, firms are heterogeneous, and the differences in their firm characteristics may influence the impact of internal capital allocation through CE on firm value. Does the impact of CE on firm value differ across firms with different level of growth opportunities? Would the impact of CE on firm value differ between firms with a high debt level and those with a low debt level? To better understand the nature of CE, I first investigate the determinants of the use of credit enhancement with a probit analysis for the period from 1990 to 2009. Our unique sample includes all CEs used by the public firms and the guarantors are all subsidiaries. The results show that firms with lower rating and shorter history are more likely to use CE for debt. Next I examine the effect of CE on firm value and I find that CE firms have a Tobin s Q that is -0.17 (11.2%) lower than that of non-ce firms. This finding is robust after controlling a set of firm-specific characteristics and correcting the self-selection bias. By documenting a negative valuation on internal capital allocation, I provide 2

a direct evidence on the inefficiency of an internal capital market operation. Our evidence also implies a direct agency cost of managers, which may be exacerbated by the unique feature of CE. Specifically, CE use shifts the incentive of monitoring managers from many debt holders to one or two credit enhancers. However, the monitoring of subsidiaries on their parents perhaps is just weaker and inefficient. Next, in the subsample analysis, I specifically test whether the effect of CE on firm value differs across firms with different growth opportunities. The rationale is that for firms with greater growth opportunities, the benefit of expanded debt capacity from CE use is more likely to manifest, while resource misallocation is more likely the concern for CE users with lower growth opportunities. Our findings indeed support the conjecture. The results show that for firms with relatively low growth opportunities, CE has a significant negative effect on firm value. Specifically, in the group of firms with lowest growth opportunities, the coefficient of the CE dummy is -2.47 and statistically significant. In comparison, for firms with high growth opportunities, the coefficient of the CE dummy is not significant. Our finding supports the notion that the sensitivity of internal allocation of resource to growth opportunities has impact on firm value (Peyer and Shivdasani, 2001). It is also consistent with Billett and Mauer (2003) that the firms with more efficient internal capital markets are more highly valued. Moreover, in firms with high debt level, firm debt issuance capacity is more likely to be constrained, justifying the adoption of CE. I find that, for firms with a low debt level, the coefficient of the CE dummy is -1.28 and statistically significant. While for firms with a high debt level, CE has no significant effect on firm value. The extent of negative impact of CE appears to be lower in firms in the high leverage group. This finding is interesting given CE arrangements that I analyze are within the corporate border. Firms become more sensible in allocating internal resources when they face higher leverage in this particular setting. Our findings therefore are not consistent with the prediction in Hege and Ambrus-Lakatos (2002) such that the pooling of financial resources in an internal capital market may magnify financial 3

distress situations. Our evidence is also in contrast with those documented in Lins and Servaes (2000) which show that the conglomerate discount is actually steeper in poorly developed emerging markets, and those in Claessens et al. (1999b) which find that during the 1998 Asian financial crisis, the conglomerate discount in the Asian markets rose. Further test reveals the negative effect of credit enhancement on the stock return upon the bond offering announcements. The average 21 days buy-and-hold abnormal return (CAR) of bond issuers with CE is 0.6% (0.5%) lowered than that of bond issuers without CE. Our paper clearly points out a link between the internal resource allocation and conglomerate value reduction, and contributes to the large literature on internal capital market allocation efficiency. So far, the literature has largely focused on the direct allocation of internal capital among firms while not on the source of these internal funds. Our study fills this void. Our paper also contributes to the literature on CE. By investigating the relationship between CE and firm valuation, this study explores the implications of corporate financing in an important yet largely overlooked area. In all, our evidence documents the significantly negative effect of CE on firm value and its sensitivity to growth opportunities and debt level. Credit risk and various instruments purposed to mitigate credit risk indeed ignited the fire during the recent financial crisis. As a result, a careful study on the relationship among CE and firm value would not only benefit us by unveiling potential risks from using CE, but also shed light on the drivers for the recent financial crisis. The paper proceeds as follows. In Section 2, I introduce the background of CE. In Section 3, I review the literature on internal capital market and the relation between financial constraints, investment and firm value. Section 4 presents the data and sample, followed by the analysis on the impact of CE on firm value on Section 5. 4

Section 6 tests whether the relation between CE and firm value differs by growth opportunities and debt levels. Section 7 concludes the paper. 2 Backgrounds on Credit Enhancements A recent global survey shows that the majority of more than 1100 risk managers consider credit risk as one of the most important risks (Bodnar et al, 2011). Seeing CE as an effective way to reduce credit risk, a significant proportion of corporate bonds are issued with credit enhancement (CE). The percentage of corporate bonds with CE increased from 2% in 1990 to 26% in 2010 in terms of dollar value. Overall, corporate bonds have been sold for $16,711 billion (par value) during the period of 1990 to 2010. Of bonds issued in the same period, 17% ($2,853 Billion), by dollar value, were issued with CE. After 2000, SEC greatly reduces the filing cost for the parent/subsidiary guarantee transaction. CE is used to protect bond holders from defaults by providing the guarantee to the payment of principal, premium (if any) and interest of the underlying bonds in case of default of issuers. The focus of this study is the external credit enhancement for corporate bonds. Traditionally corporate bonds use three major types of external enhancements: guarantee, insurance and letter of credit (LOC) which takes 96%, 3% and 1% of the total credit enhancement respectively from 1990 to 2010. In forms of guarantee, guarantors are diversified. They can be parent firms/subsidiaries. For example, MGM Mirage used all its domestic subsidiaries as guarantors for its $225,000,000 bonds issued in 2001. They can also be independent firms. For example, Nestle SA provided guarantee on $150,000,000 bonds issued by EMC in 2005. Among 1432 CE bonds by public-listed firms from 1990 to 2009, 1092 CE bonds were guaranteed by subsidiaries. The major form of CE is that subsidiaries provide guarantee on bonds issued by parents. It accounts for about 70% of all CE used by public firms. 5

Credit ratings is an important factor in the requirement by several regulations on financial institutions and other intermediaries investments in bonds. For example, regulations restrict banks from investment in speculative-grade bonds since 1936 (Partnoy, 1999; West, 1973). In 1989, savings and loans were required to completely liquidate their speculative-grade bonds by 1994 (Kisgen, 2006). Finally, pension fund guidelines often prevent bond investments from speculative-grade bonds (Boot, Milbourn, and Schmeits, 2003). In order to attract financial institutional investors and other intermediaries, debt borrowers use CE to increase credit rating of debt before issuing. Should debt borrowers default, debt holders have recourse to guarantors. Once a bond obtains credit enhancement, the rating agency will assign two ratings: one is the rating for the underlining bond without the consideration of CE, another one is the rating of the guarantor/insurer. The rating agency then gives the higher of either one to the bond with CE. If the guarantor or insurer is downgraded, the rating agency will reevaluate the bond and adjust the rating if needed. Since the guarantor or insurer has the possibility of failing to fulfill its obligation of paying debt in case of the default of the bond issuer, the credit risk for bonds with CE is reduced to the least extent but not completely eliminated. 3 Literature Review and Hypotheses Development 3.1 Debate on Internal Capital Market Efficiency Is the internal capital market efficient than external capital market? The debate on the issue has been fierce. In the camp supporting internal capital market efficiency, Williamson (1975) argues that the internal capital market of diversified firms might allocate capital more efficiently than the external capital market because top management of a diversified firm knows more about the firm s investment opportunities than external investors. Further, Gertner, Scharfstein and Stein (1994) and Stein 6

(1997) propose models to identify the conditions for more efficient investment decisions. These modes are mainly related to investment comparison between two forms of organization: diversified firms and stand-alone firms. Specifically, Stein (1997) argues that managers will be unwilling to cut investment when they have poor investment opportunities. However, an internal capital market of diversified firms gives managers a way to redeploy capital from divisions with poor investment opportunities to those with good investment opportunities without compromising the overall capital budget. Khanna and Tice (2001) examine capital expenditure decision of diversified firms in response to WalMart s entry to their market. They find diversified firms make quicker decision of exit or stay and their capital expenditures are more sensitive to the their productivity than focus firms. Hence, internal capital allocation in diversified firms functions efficiently by tying up capital to investment opportunities. Converse, a competing line of studies argue that internal capital market is inefficient. One important measure of capital allocation inefficiency is defined as divisions investment being independent on their investment opportunities. Lamont (1997) shows that when oil prices are high, the non-oil divisions of diversified oil producers increase their investment more than their industry counterparts and these investments are inconsistent with their own investment opportunities. Shin and Stulz (1998) find evidence that investment of small divisions of conglomerates depends on cash flows of other divisions, not their own Q. Rajan, Servaes, and Zingales (2000) show that when divisions have similar level of resources and opportunities, internal fund allocation has a positive relation with their opportunities; when divisions have diversified level of resources and opportunities, the relation between internal fund allocation and opportunities turns into a negative one in which divisions with good opportunities are poached by other divisions. The inefficient capital allocation has also been investigated by comparing the investment efficiency before and after spinoff. Gertner, Powers, and Scharfstein (2002) examine spin-off divisions of diversified conglomerates and find their investment becomes more sensitive to industry Q. Ahn and Denis (2004) further find that for diversified firms after spin-off there is a signifi- 7

cant increase in measures of investment efficiency and firm value. They thus conclude that internal capital allocation in diversified firms is inefficient. Most of these studies compare conglomerate firms to their standalone counterparts and hence is subject to the criticisms of measurement errors (Whited, 2001; Villalonga, 2000) and sample selection issues (Campa and Kedia, 2002; Chevalier, 2004). Another limitation also exists in the use of an industry Tobin s Q as proxy for growth opportunities which implicitly assume all firms, conglomerates and singlesegment firms have similar investment opportunities within an industry (Marksimovic and Philips, 2002). The conditions leading to an inefficient internal market have been investigated by many researchers. Agency problem of division managers is regarded as one of the main reasons for inefficient internal capital market efficiency. Scharfstein and Stein (2000) argue that rent-seeking behavior of division managers will raise their bargaining power and influence CEO to overinvest in the divisions with bad investment opportunities and underinvestment in the divisions with good investment opportunities. Datta et al (2009) suggest that CEO compensation makes difference in internal capital allocation. Specifically, stock option is more effective than stock grant in reducing agency problem and being incentive mechanism for efficient allocation. Ozbas and Scharfstein (2010) find that in unrelated segments of diversified firms investment is less sensitivity than stand-alone firms. Moreover, the ownership stakes of top management has a positive relation with the extent of Q-sensitivity differences, suggesting that agency problem leads to the inefficient capital market. A key assumption in many studies is that division managers, especially in lowgrowth divisions, are empire builder and rent-seekers while parents or CEOs act in the maximum interest of firms. Mathew and Robinson (2008) base their model on the assumption of the rational and profit-maximizing behavior on the part of parents. Kolasinski (2009) builds his hypothesis on the assumption that the CEO would prefer to commit ex ante to invest more in the high-growth division. Stein (1997) points 8

out that the control right of parents is the key for parents to move funds from less desirable investments to more desirable ones. It implies that parent firms have to act in the best interest of firms to make internal capital market efficient. However, the fact that parents themselves are agents of investors raises the concern if they can act in the maximum interest of firms. Bolton and Scrafstein (1998) point out because allocating capital to divisions with opportunities aligns with parents empire-building preference, it is not obvious that agency problem of parents leads to inefficient internal capital allocation among divisions. Scharfstein and Stein (2000) argue that a two-tier agency problem, stemming from misaligned incentives at parents and at divisions, is necessary for corporate socialism in internal capital allocation. Parents use their authorities to have their bonds guaranteed by subsidiaries and such allocation may be a consequence of empire-building preference of parents rather than following the investment opportunities. The CE in the sample of this study is through subsidiaryguarantee. I therefore postulate the following hypothesis: H1. Credit enhancement has a negative effect on firm value. 3.2 Financial Constraints, Investments and Firm Value Whited (1992) includes financial constraints into firms investment analysis and finds firms ability to borrow money can explain the different investment patterns very well. The implication of this finding is that borrowing constraints can lead to underinvestment, subsequently reduce firm value. Billett and Mauer (1994) document that internal subsidies to small segments with financial constraints and relatively poor investment opportunities can increase the excess value of the segments. They conclude that financial constraints affect the relationship between internal capital market and firm value. A global survey of 1,050 Chief Financial Officers (CFOs) finds that the majority of firms had to forgo attractive investment opportunities because of financial constraints in the financial crisis of 2008 (Campello et al, 2010). CE can increase the ratings of bonds and lessen firms financial constraints, thus leading to increased firm 9

value. Growth opportunity is an important factor in influencing debt level and firm value. For example, Vogt (1994) finds overinvestment in US manufacturing firms is the strongest in the large, low-dividend firms with low Tobin s Q. Lang, Ofek and Stulz (1994) divide the firms into the high and low growth opportunities group and found that only in the group with low growth opportunities leverage is negatively related to subsequent growth of employees number and capital expenditure. McConnell and Servaes (1995) empirically investigate the relation between leverage and firm value, and find that this relation is impacted by growth opportunities. Specifically, firm value is negatively correlated to leverage for the high-growth firms and is positively correlated to the leverage for the low-growth firms. CE is in nature an internal capital market allocation and thus it is better if parents have more favorable investment opportunities. The fewer the growth opportunities, the more likely investment will go to unprofitable projects. Moreover, the fewer the growth opportunities for a firm, the less the benefit the firm would receive from reducing financial constraints. This reasoning leads to the following empirical hypothesis: H2. The higher growth opportunities of a bond issuer, the less negative effect of credit enhancement on its firm value is. Firm value is positively related to debt level since debt has a positive effect in preventing overinvestment. Jensen (1986) argues that debt can reduce managers interest in overinvestment due to the obligation of paying debt. Because of the separation of corporate equity ownership and management, managers intend to reward themselves by increasing the size of the firm beyond the optimal level for the shareholders. One explanation is that managers consider the firm a source of self-esteem and a means to increase their own human capital (Zingales, 1998). Once firms borrow debt, managers will have to limit their investment on unprofitable projects to avoid bankruptcy. Therefore, debt plays a monitoring role in reducing agency problem. D Mello and Mi- 10

randa (2010) investigate the monitoring role of debt and find the new debt offering by unlevered firms leads to a reduction in abnormal capital expenditure in firms with overinvestment in real assets. As discussed before, CE weakens the monitoring of debts. When parents have high level of non-ce bonds, the strengthened monitoring from non-ce bondholders can mitigate weakening effect from CE to some extent. Thus, I state the following hypothesis: H3. The higher level of non-ce bonds of a bond issuer, the stronger monitoring it has and the less negative effect CE on its firm value. 4 Data I start with Mergent Fixed Income Securities Database (FISD) data for this study. From 1990 to 2009, there are 101,428 corporate bonds issued by 7,110 firms which include 13,455 bond issued by 2,428 public-listed firms. Among all these bond issues, there are 1,273 firms that used external credit enhancement on their 11,765 bond issues, including 1,658 bonds with credit enhancement issued by 690 public-listed firms. The stock information for this sample is obtained from CRSP and the financial data and the ratings are from COMPUSTAT. Following the convention on bond research, I exclude the financial firms (SIC codes 6000-6999) and regulated utilities firms (SIC codes 4900-4999) from the sample. I also exclude the firms missing sufficient data to compute valid Tobin s, P/E and debt. Further, for each credit enhancement, I identify the relationship of issuer and guarantor from the guarantor description in FISD. If there is still unclear relationship, I manually check it from Mergent online. All credit enhancements which are not guaranteed by subsidiaries are excluded. The final sample consists of 3,956 bonds issued by 1,460 public-listed firms, including 448 bonds with credit enhancement issued by 303 public-listed firms. The US corporate bonds include US corporate debt, US corporate MTN, asset-backed security and other US corporate bonds but exclude the US corporate convertible, preferred stock and US asset-backed security. 11

The firms are evaluated by their Tobin s Q. Tobin s Q is computed as (total asset - book value of equity + market value of equity)/total asset. Because all the bonds with credit enhancement are long-term bonds, I choose the S&P Domestic Long Term Issuer Credit Rating one month before the offering date from COMPUSTAT. To reduce the effect of possible spurious outliers, I winsorize the top and bottom 1 percentile of observations. To investigate the underinvestment problem of CE, I analyze the valuation difference between parents with the highest growth opportunities and those with the lowest growth opportunities; similarly, to investigate overinvestment problem of CE, I examine the valuation difference between parents with a low debt level and those with a high debt level. For the proxy for the growth opportunity, I follow the method of McConnell and Servaes (1995) by using a firm s price-to-operating-earnings (P/E) ratio. The ratio is computed as stock price divided by the operating earnings per share at the end of the corresponding year. Sales growth is used as a second proxy for growth opportunities and is calculated as the average sales growth percentage in the last two years. Descriptive statistics for firms with credit enhancement and firms without credit enhancement are presented in Table 3. The data period is in the year before the debt issuance. The differences in the firm characteristics between the CE firms sample and the non-ce firms sample are dramatic. For example, Tobin s Q for two groups is significantly different. Mean (median) Tobin s Q of the CE firms is 1.36 (1.25) and this value increases to 1.53 (1.31) for the non-ce firms. The results show that the firms with CE have a lower market valuation than those without CE. Two proxies of growth opportunities, P/E and sales growth, have a different result in two samples. P/E for the CE firms is significantly lower compared to the non-ce firms. In contrast, sales growth for CE firms is significantly higher for the CE firms that for the non-ce firms. The implication of this difference is that P/E and sales growth may capture a different aspect of growth opportunities. Considering this possibility, I estimate the regression for the high growth and the low growth samples using P/E and sales 12

growth respectively. The S&P long-term domestic debt rating for CE firms is lower than that for non-ce firms. As shown by Table 3, firms with credit enhancement usually have a higher debt level, smaller size, lower ROA, lower market-to-book, lower P/E, less free cash flow, lower long-term credit rating by S&P, higher sale and sale growth than those without credit enhancement. Except sales, all the differences are statistically significant. The lower value of financial data of the firms with credit enhancement suggests that these parents on average have less growth opportunities, lower return, and weaker pay-back ability than those firms without credit enhancement. 5 Analysis of the Use and the Valuation Effect of Credit Enhancements 5.1 Probit Analysis on the Use of Credit Enhancement The first step is to use probit regression to examine the determinants of CE use (Y). A bond issuer either takes CE (Y=1) or doesn t (Y=0) in the sample. A set of factors in a vector x explains the decision. I model the probability that an insurer uses CE as a probit function: Pr(Y ) = Φ(β X) (1) where Y* is not observable while I can observe y, Φ(.)denotes the standard normal distribution, X is a set of variables explaining bond issuers propensity to use CE (discussed below). The set of parameters β reflects the impact of changes in on the probability. In the setting of probit, I have: Y = 1 when Y > 0 (2) Y = 0 when Y <= 0 (3) 13

The benefit of new debt is the motivation of debt issuers. Two variables control for the benefit of new debt issuance. The lower the yield of bonds in the market, the more likely the firms issue new debt to retire the old debt. I use the average yield of Moody s AAA and Baa bonds. When a firm has high growth opportunities, the benefit from raising capital from debt is more likely to manifest. P/E ratio is used as the proxy for growth opportunities. The firms with financial constraints are in greater need of CEs than firms without financial constraints. Rating controls for a firm s default risk. The lower the credit rating, the higher the default risk a firm has.firms with lower credit rating are expected to be more likely to use CE. I use a firm s S&P long-term domestic credit rating. Size controls for a firm s ability to issue debt. Small firms lack collateral to back up their borrowing and have more information asymmetry than larger firms. Consequently, small firms have limit access to debt market. The implication is size affects the firms ability to issue debt (Whited, 1992).Dividend payout by a firm shows its debt capacity. A firm that doesn t pay dividends is regarded as a firm with difficulty for external capital (Fazzari, 1988). Bernanke and Campbell (1988) advocate that three variables can be used to measure a firm s collateral for external capital. These three variables are cash, inventory, PPE(property, plant and equipment). Debt has impact on the issuance of new debt. On one hand, high debt level has a high level of default risk. On the other hand, the over-hang problem from high level debt prevents the firms from issuing new debt. Table 4 reports the result of the probit regression. Overall, the firms with higher benefit from new debt and with higher financial constraints are more likely to use CEs. Specifically, the bond yield in the market, credit rating, size, dividend payout ratio, cash and PPE are significantly negatively related the choice of CE use, while the firms growth opportunities and debt level are significantly positively related to the choice of CE use. For example, the coefficient on rating is -0.03 and significant at 1% level, implying that lower credit rating increases the likelihood of CE use. The coefficient on growth opportunities is 0.02 and significant at 1% level, meaning firm s 14

growth opportunities increases the likelihood of CE use. However, the coefficient on inventory is not statistically significant. Hence, I fail to find evidence that firms with high inventory are easier to access capital market and less likely to use CE. 5.2 Valuation Effect of Credit Enhancements The next step of the analysis is to investigate the effect of CE on firm value. I estimate a regression of the Tobin s Q of the firms on CE dummy variable and a set of control variables. A number of firm characteristics serve as control variables in the regression. CE is a dummy variable for the use of credit enhancement. It takes a value of 1 for a CE firm and 0 for a non-ce firm. P/E ratio is the proxy for the growth opportunities. It has been shown to have impact on the relationship between firm value and debt level (McConnell and Servaes, 1995). Sales Growth is another proxy for growth opportunities (McConnell and Servaes,1995; Doidge et al 2004). Free Cash Flow is a proxy for overinvestment probability. In Jenson (1986), firms with free cash flow and low growth opportunities are more likely to have overinvestment problem when managers have empire-building preference. Debt has impact on investment and default risk. On one hand, high debt level induces underinvestment and reduces overinvestment. On the other hand, high debt level indicates a financial distress risk and may lower the firm value. ROA is related to the firm value directly. Size is total asset. Small firms are well known to outperform large firms. 15

Rating is used to measure the benefit of CE use. The lower the firm rating is, the more benefit it receives from CE. Therefore rating has an effect on the firm value. In addition, fixed Year Effect is used to capture the impact of time-varying macroeconomic factors. It is worth noting that firms with lower Q are more likely to use CE than those with higher Q. With this self-selection, the error in the regression is therefore likely to be correlated to a firm s decision to whether use CE or not. This correlation will create a selection bias in the estimate of the coefficient of the CE dummy variable in the valuation regression. The econometric problem I face is similar to the treatment effects in Heckman model, therefore I also use Heckman s (1979) two-step estimator. To examine the effect of CE, I use a valuation regression of Q as: Q i = α + β X i + δce i + ε i (valuation equation) (4) where X i is a set of exogenous variables, CE i is a dummy variable that equals one for a firms that use CE for bond issuing. α, β, δ is a vector of parameters to be estimated and ε i is an error term. δ measure the relation between CE i and Q i. If a firm s decision to use CE is related to Q, CE i and ε i are correlated and the estimate of δ will be biased. To correct the bias, according to Heckman (1979), a reduced form CE decision equation is given by: CE i = γ Z i + ν (CE decision equation) (5) where CE i is an unobserved latent variable. I observe only an indicator variable for the CE decision, defined as CE i = 1 if CE i > 0 and CE i = 0 if CE i <= 0. Z i is a set of variables that affects the decision to use CE and µ is an error term. In addition to 16

the basic structure, the Heckman model requires the following assumption: ( ) ( ) µ ε σε 2 ρσ ε σ nu (1)µ =, σ =, bivariate normal distribution; µ ν ρσ ε σ nu (2)(ε, ν) is independent of X and Z; (3)var(ν) σν 2 1; σ 2 ν Then the expected Q of the CE firm can be expressed as: E((Q i = CE i ) = 1) = α + β X i + δ + ρσ ε λ i,1 (γ Z i ) (6) Where λ i,1 (γz i is the inverse Mills ratio (IMR) and is estimated as φ(γ /Φ(γ Z i ), where φ(.) and Φ(.) are the density function and cumulative standard normal distribution functions respectively. Similarly, I have the expected Q of the non-ce firm as: E((Q i = CE i ) = 0) = α + β X i + δ + ρσ ε λ i,2 (γ Z i ) (7) Where λ i,2 is estimated as φ(γ Z i )/Φ(γ Z i )(1 Φ(γ Z i )). The difference of Q for CE firm and non-ce firms is computed as: E((Q i = CE i ) = 1) E((Q i = CE i ) = 0) = δ + ρσ ε φ(γ Z i )/[Φ(γ Z i )(1 Φ(γ Z i )] (8) From equation (8), if low Q firms tend to choose CE, the low Q after use of CE is positively related to the low Q before use of CE. Such the correlation of the error terms, ρ, between two equations are positive and the bias of valuation on CE firms is upwards. The first step in Heckman (1979) model is to run a probit model of using CE for all the firms. The estimates of the γ from this probit model are then used to computed the inverse Mills ratio (IMR, denoted as lambda in the below). In the second step, the valuation equation now becomes: Q i = α + β X i + δce i + θlambda i + e i (9) 17

Where the θ i captures the sign of correlation between the error terms in both equation. The result of Heckman model is reported in Table 5. Specifications (1) to (4) are the results for regressions of Q on the CE choice and a set of control variables after controlling for the self-selection bias. Lambda, inverse Millers ratio (IMR) in Heckman s model, is significantly positive at all regressions and it shows the error term in the selection regression and the valuation regression are positively related. The result confirms the previous finding that low Q firms are more likely to choose CE. After introducing the selection correction variable Lambda in the regression, the explanatory power of CE dummy variable becomes economically stronger and stays statistically significant. In Specification (1), the dummy variable has a negative coefficient of 0.25 with a t-statistic of -9.42. One may argue this negative effect of CE on bonds is due to the fact that firms with CE has less growth opportunities and thus lower firm value. In specification (2), I use two proxies for growth opportunities: P/E ratio and sales growth. The results show the coefficient of the dummy variable CE is still significantly negative even after controlling for the growth opportunities. Controlling for growth opportunities reduces the magnitude of negative effect from CE only slightly. In specifications (3) and (4), I add other control variables. The coefficient of the dummy variable CE is significant in both regressions. These overall results are consistent with the conjecture and suggest that CE has a significant negative impact on firm value. In the last specification (4), the regression includes the control variables CE dummy, P/E, free cash flow, sale growth, debt, ROA, size, and rating. The coefficient of CE dummy is -1.24 with a t-statistic -3.73 and R 2 = 0.27. Except for the sale growth, all other variables are significant. In particular, bond issuing firms with lower free cash flow, higher debt level and higher ROA have higher Q. Bond issuing firms with low rating (high rating number), large size, short firm history tend to have lower Q. In sum, the negative effect of CE on firm value is robust after controlling for a set of a firm s financial characteristics and self-selection bias. 18

6 Valuation Effect of Credit Enhancement for Different Subsamples This section investigates whether the impact of CE on firm value differs between firms with high growth opportunities and those with low growth opportunities, and differs between firms with a high debt level and a low debt level. 6.1 Valuation Effect of Credit Enhancement for Firms with Different Growth Opportunities For growth opportunities, I first divide the sample into five groups based on their P/E ratio in the year of the bond issuing. To quantify the impact of growth opportunities, I estimate two regressions with the same set of control variables on the highest P/E (growth opportunities) group and the lowest P/E (growth opportunities) group respectively. The regressions include a dummy variable CE that takes the value of one if it uses CE and zero otherwise. Comparing the coefficients CE dummy variable in the regression of the high P/E dummy variable and the low P/E dummy variable reveals the effect of P/E (growth opportunities) on the relationship between CE use and firm value. Table 6 reports the results of the regressions. In the lowest P/E (growth opportunities) group, coefficient for CE is -2.47 (t=-2.83) while that for the highest P/E (growth opportunities) group is -0.15(t=-0.19). It can be seen that the reduction in firm value from CE in the lowest grow opportunities group is 2.03 higher than that in the highest grow opportunities group. The magnitude of the coefficient of CE in the lowest growth opportunities group indicates that the negative effect of CE on firm value is economically significant when firms lack growth opportunities. Comparing the coefficient of CE in the lowest growth opportunities quintile with that in the highest growth opportunities quintile, one can see that the impact of CE on firm value is affected negatively by growth opportunities. 19

The empirical results may depend on the proxy of growth opportunities. As an alternative measure of growth opportunities, I used sales growth. Again, I divide the sample into five groups based on the sales growth in the year of bond issuing and perform one additional sensitivity test on the sales growth. The regression result is reported in Table 7. Similar to Table 6, there is a significant negative effect of CE on firm value in the lowest sales growth (growth opportunities) group and no significant effect of CE on firm value is found in the highest sales growth (growth opportunities) group. The result further confirms the conjecture that the impact of CE on firm value has a negative relation with growth opportunities. The finding supports the view that the internal capital allocation is inefficient (Scharfstein and Stein, 2000; Rajan, Servaes, and Zingales, 2000). The inefficiency stems from the fact that capital flows to the divisions or subsidiaries with few growth opportunities. In firms with few growth opportunities, the benefit of expanded debt capacity from CE use is small and overinvestment is more likely. The cost of CE is more likely to dominate the benefit of CE. Therefore the negative effect of CE on firm value is more pronounced in firms with few opportunities. 6.2 Valuation Effect of Credit Enhancement for Firms with Different Leverage Next, I break down firms into different leverage groups. To check the impact of debt level, I divide the sample into five groups based on their debt ratio in the year of the bond issuing. I then estimate two regressions with the same set of control variables on the highest debt group and the lowest debt group respectively. The regressions also include a dummy variable CE. Table 8 reports the results of the regressions. In the lowest debt group, coefficient for CE is -1.28 (t=-1.83). In contrast, in the highest debt group coefficient of CE is -0.66(t=-0.70), which only equals to half of the value in the other group. The 20

result supports the conjecture that firms with a high debt level is less sensitive to the negative effect of CE. The benefit of CE on firms with high debt is two-fold. First, there is reduction on high constraints of debt capacity. Secondly, the strong monitoring of non-ce debt holders achieves in weakening agency problem, which is the culprit for the inefficient internal capital market allocation. In sum, the growth opportunities and debt level have impact on the effect of CE on firm value and the internal capital allocation efficiency. 7 Stock Return at the Bond Offering While up to now the analysis focus on the long-term valuation effect of firms with credit enhancement, it is also important to understand the stock market reaction at the initial announcement of an CE bond offering. There is an extensive literature on the market reaction on bond offering announcements. Dann and Mikkelson (1984) identified a marginal negative valuation effect at the announcement of debt announcements. For an announcement of convertible debt offering, the negative valuation effect is significant. The valuation is measured using the abnormal return in the (-10,+10) days of the bonds offering announcement. For each day, abnormal return is computed as the difference between the stock return and the value-weighted market return. The total abnormal return is calculated as buy-and-hold abnormal return (B&H) and cumulative abnormal returns (CAR). The result in Panel A of Table 9 shows the Non-CE firms has experienced a higher abnormal return than the CE firms. Specifically, Non-CE firms have 0.6% B&H return and 0.5% CAR than CE firms in the 21 days around the announcement of the bond offering. A regression of the abnormal return on CE reveals the driving factors of the lower stock performance of the CE firms. The result is reported in Panel B. I control the fixed year and fixed industry effect in the regressions. In all models of both regression, credit enhancement use lowers the abnormal return 0.01% and the effect is statistically sig- 21

nificant. The negative effect of credit enhancement use is robust after controlling the firm characteristics. 8 Conclusions In an increasing trend, firms allocate their internal capital by internal sponsored guarantee on their debt. This study explores the impact of CE on firm value and whether the relationship between CE and firm value differs by growth opportunities and debt levels. CE may affect firm value both positively and negatively. The positive effect of CE on firm value stems from lessening financial constraints and reducing the risk of debt. The negative effect of CE is caused by weakened monitoring on agency problem, a well identified reason for inefficient internal capital market allocation (Scraftstein and Stein, 2000). The results of this empirical test show that bond issuing firms with CE have a significantly lower firm value than those without CE. The finding remains strong after controlling a set of firm characteristics and correcting the self-selection bias. The finding indicates the negative effect of CE on firm value dominates its positive effect on firm value. A further investigation reveals that the impact of CE on firm value differs by growth opportunities and debt levels. The extent of negative impact of CE is lower on firms with higher growth opportunities and higher debt levels. Moreover, the paper finds a negative valuation effect of credit enhancement on the stocks return upon the bond offerings. This study provides a new evidence of inefficient internal capital allocation and casts doubt on the assumption that CEOs act in the maximum interest of firms. A possible interpretation is that CEOs use the subsidiary with good rating to borrow debt and then subsidize the subsidiary with poor rating, a problem called corporate socialism (Scraftstein and Stein, 2000). The market perceives the internal capital allocation inefficiency and gives a low valuation to the firm using CE. This study also raises the question whether risk hedging between different entities 22

of a firm through CE is an efficient way. The results in this study confirm that the risk reduction by internal guarantee comes with a value reduction of firm value. This study suggests that the use of CE is subject to agency problems. Therefore, for firms with better corporate government, the decision of using CE is expected to less depend on the rating of bond issuers. Instead, the decision may more rely on the overall effect of CE on firm value. The effect of issuers corporate governance on the choice and valuation of CE merit further investigation. 23

Table 1: Corporate Bonds by All Firms This table reports the number of corporate bond issue, the number of corporate bond issues using credit enhancements, the percentage of corporate bond issues using credit enhancements in terms of issue number, par value of issued corporate bonds, par value of corporate bonds using credit enhancements, and the percentage of corporate bonds using credit enhancements in terms of par value. The par value is in millions. Year Issue # of Issue # of Percentage of Par Value of Par Value of Percentage of Corp. Bond Corp. Bond Issue # of CE Corp. Bond($) Corp. Bond $ value of CE with CE (%) with CE ($) (%) 1990 531 12 2.26 77,869 1,795 2.31 1991 1,224 37 3.02 138,141 1,991 1.44 1992 2,224 89 4 217,848 689 0.32 1993 3,349 239 7.14 307,044 8,678 2.83 1994 2,892 81 2.8 173,876 3,174 1.83 1995 4,677 138 2.95 273,075 7,644 2.8 1996 3,961 118 2.98 339,944 20,876 6.14 1997 4,920 447 9.09 509,463 70,612 13.86 1998 5,303 629 11.86 802,194 116,944 14.58 1999 4,527 462 10.21 807,173 120,489 14.93 2000 3,674 268 7.29 869,506 125,039 14.38 2001 4,173 456 10.93 1,029,108 133,929 13.01 2002 4,576 617 13.48 882,008 130,268 14.77 2003 6,700 568 8.48 985,258 125,094 12.7 2004 7,317 903 12.34 1,035,495 156,274 15.09 2005 7,309 1,173 16.05 1,040,805 166,188 15.97 2006 7,869 1,274 16.19 1,523,111 180,708 11.86 2007 10,432 1,637 15.69 1,504,498 207,292 13.78 2008 9,344 1,450 15.52 1,258,112 298,152 23.7 2009 6,426 1,167 18.16 1,672,110 643,852 37.33 2010 9,549 1,100 11.52 1,265,278 333,979 26.4 Total 110,977 12,865 11.59 16,711,916 2,853,667 17.07 24

Table 2: Corporate Bonds by Public Firms This table reports the number of public corporate bond issue, the number of public corporate bond issues using credit enhancements, the percentage of public corporate bond issues using credit enhancements in terms of issue number, par value of issued public corporate bonds, par value of public corporate bonds using credit enhancements, and the percentage of public corporate bonds using credit enhancements in terms of par value. The par value is in millions. Year Issue # of Issue # of Percentage of Par Value of Par Value of Percentage of Corp. Bond Corp. Bond Issue # of CE Corp. Bond($) Corp. Bond $ value of CE with CE (%) with CE ($) (%) 1990 225 2 0.89 39,350 100 0.25 1991 399 2 0.5 71,725 0 0 1992 628 29 4.62 103,524 15 0.01 1993 745 77 10.34 116,164 2,930 2.52 1994 289 14 4.84 50,625 1,095 2.16 1995 505 22 4.36 85,872 2,443 2.85 1996 573 35 6.11 119,931 8,267 6.89 1997 845 98 11.6 178,907 15,180 8.48 1998 1,100 154 14 274,812 28,554 10.39 1999 784 119 15.18 264,797 29,971 11.32 2000 528 45 8.52 244,918 17,563 7.17 2001 763 108 14.15 361,283 34,827 9.64 2002 675 104 15.41 271,914 32,945 12.12 2003 849 104 12.25 303,506 35,041 11.55 2004 751 135 17.98 285,568 37,277 13.05 2005 716 114 15.92 279,768 39,081 13.97 2006 742 112 15.09 330,453 51,110 15.47 2007 1,046 96 9.18 408,602 45,805 11.21 2008 588 63 10.71 291,779 44,533 15.26 2009 704 143 20.31 468,658 106,360 22.69 2010 747 186 24.9 425,644 89,496 21.02 Total 14,202 1,762 12.41 4,977,800 622,593 12.51 25