銀行往來關係及議價能力如何影響聯合貸款條約中使用依據績效調整利率的條款 - 政大學術集成

全文

(1)國立政治大學財務管理研究所. 碩士論文. 政治 "銀行往來關係及議價能力如何影響聯合貸款條約中使用. 大. 立. 依據績效調整利率的條款". ‧ 國. 學. "Banking Relationship, Bargaining Power and the Use of. ‧. Nat. n. al. er. io. sit. Syndicated Loan Contracts”. y. Performance Pricing Covenants in. Ch. engchi. 指導教授：張元晨. i n U. v. 博士. 研究生：李光耀中華民國一百零三年十二月 1.

(2) Acknowledgement 歷經了半年的辛勤耕耘，論文終於順利完成。對於當時再過幾天就要至挪威當交換學生的我而言，在口試委員宣布論文通過的那一刻，身上的大石頭總算是順利放下，除了喜悅之外，心中更是充滿了感謝之心。首先要感謝的當然是指導教授張元晨老師。自從暑假開始，每星期便有一次的 meeting，我自己也看了 2~4 篇的國際期刊論文，在老師細心地指導並討論之後，真的讓我的論文進度突飛猛進，能夠順利地在八月底就把論文的主軸訂好。雖然在寫作論文上總是會遇到一些瓶頸，老師也是很迅速地給予我們. 政治大並完成口試。真的很幸運能夠成為老師的指導學生，謝謝老師!!! 立. 正確的方向，多虧了老師的鼓勵及引導，才能讓我在有限的時間把論文寫完. ‧ 國. 學. 其次，我也要感謝柏凱老師及湘萍老師在口試過程中給予許多寶貴的建議，使最後的論文內容更為完善，也更能符合學術上的要求。在論文寫作的過程. ‧. 中，我也需要另外感謝兩位貴人，之寧及依婷兩位學姊。每次想破頭都想不. sit. y. Nat. 通 SAS 以及 STATA 程式碼的時候，兩位都可以很精準的解決我的問題，使. al. er. io. 我可以順利進展到下一步驟。如果沒有兩位學姊的幫忙，我可能真的要在奧. v. n. 斯陸邊上課邊寫程式碼了，祝福妳們之後在學術界發展更為順利囉。. Ch. engchi. i n U. 我在這邊也要感謝財管所上的同學們：同一個指導老師的聖雯、柯南以及皓雯，彼此交流了很多論文的寫作方式，恭喜你們論文也順利完成口試啦。另外，也感謝常駐研究室的其他同學們(以及研究室管理員 XD)，在我暫時沒心情寫論文的時候，能夠陪我打世紀帝國以及玩桌游，甚至於去河堤打球，紓解了我很多壓力，使我有更多動力完成論文的下一階段。我回國之後大家可能就各奔東西了，希望之後還有機會多多相聚。最後，我要感謝在這漫長的半年中所有關心我和幫助我的人，願將完成論文的這份喜悅也分享與你們。國立政治大學財務管理研究所光耀 2015.07.28. 2.

(3) Abstract It is shown in this thesis that long-term banking relationships and bargaining power are important determinants of performance-pricing covenants (PPC) inclusions in syndicated loan contracts. Using a large sample of syndicated loans data (1993-2010), I find that syndicated loans tend to include more interest-increasing PPC when a long-term banking relationship exists and when borrowers have fewer financing alternatives. The presence of banking relationship with lead arrangers reduces the odds of using interest-increasing PPC, because lead arrangers might not be able to. 政治大 table ranking are likely to立 use more interest-decreasing PPC to attract borrowers, capture all rents from holding-up borrowers. Finally, I find lenders with lower league. ‧. ‧ 國. 學. which again is consistent with the hold-up hypothesis.. n. er. io. sit. y. Nat. al. Ch. engchi. 3. i n U. v.

(4) Index Acknowledgement ........................................................................................................ 2 Abstract ......................................................................................................................... 3 Table of Content ........................................................................................................... 4 1. Introduction .......................................................................................................... 6 2. Review of the Literature .................................................................................... 10 3. Hypothesis Development ................................................................................... 12 4. Data Descriptions ............................................................................................... 15 4.1 Data Sources ................................................................................................. 15 4.2 Performance Pricing Types ......................................................................... 16. 政治大 4.4 Lead Arranger Ranking .............................................................................. 18 立 4.3 Measuring Relationship Strength ............................................................... 17. 4.5 Descriptive Statistics .................................................................................... 19. ‧ 國. 學. 5.. Empirical Analysis ............................................................................................. 21 5.1. Banking Relationships and the Use of Performance Pricing. ‧. Covenants. ................................................................................................ 21. y. Nat. 5.3. Borrower Opacity and the Use of Performance Pricing Covenants. . 24. 5.4. Bargaining Power and the Use of Performance Pricing Covenants. .. 27. n. al. er. sit. Syndication and the Use of Performance Pricing Covenants. ............ 22. io. 5.2. Ch. engchi. i n U. v. 6. Conclusion .......................................................................................................... 29 References ................................................................................................................... 30 Appendix ..................................................................................................................... 43. 4.

(5) Table of Content Table 1: PPC Use Percentage by Financial Institution Types……………..................32 Table 2: Total Lending Amount from 1990 to 2010 in USA……………………........33 Table 3: PPC Use Percentage between Domestic and Foreign Bank………...............34 Table 4: Five largest arrangers of USA from 1993 to 2010…………………….........35 Table 5: Summary Statistics………………………………………………………….36 Table 6: PPC Contract Types…………………………………………………………37. 政治大 Table 8: Syndication and the 立Use of Performance-Pricing Covenant…………......... 39. Table 7: Banking Relationships and the Use of Performance-Pricing Covenant…… 38. ‧ 國. 學. Table 9: Borrower Opacity and the Use of Performance-Pricing Covenant…………40 Table10: Market share and the Use of Performance-Pricing Covenant……………...41. ‧. Table11: Bank Ranking and the Use of Performance-Pricing Covenant…………….41. n. er. io. sit. y. Nat. al. Ch. engchi. 5. i n U. v.

(6) 1.. Introduction. Syndicated loan, which is provided by a group of lenders and arranged by one or several commercial or investment banks, is one of the major funding channels for corporations worldwide. It often occurs in situations where a borrower requires a large sum of capital that may be too much for a single lender to provide, or outside the scope of a lender's risk exposure levels. The syndicated loan market has become dominant for issuers to tap banks and other institutional capital providers for loans. 政治大. because syndicated loan is less expensive and more efficient to administer than traditional bilateral loans.. 立. ‧ 國. 學. It is common for syndicated loan agreements to include financial covenants and performance pricing covenants. Financial covenants in a syndicated loan contract. ‧. enforce either maximum or minimum financial performance measures against. sit. y. Nat. borrower’s performance. For example, with a minimum current ratio specified in the. n. al. er. io. loan contract, the borrower must maintain its current ratio above this level to avoid. i n U. v. technical defaults. If not maintained, the covenant will offer lenders the opportunities. Ch. engchi. to negotiate a loan contract and give them the right to terminate the agreement or force the borrower into default. In most cases, the lender would choose to demand immediate repayment, to increase spread, or to include additional collaterals if borrowers are in violation of certain covenants. Performance pricing covenant (PPC) was first introduced in the 1970s and has been widespread in many syndicated loan contracts since the early 1990s. Contracts with PPC include a pricing grid that defines performance levels based on certain criteria and their corresponding interest spreads. PPC may be tied either to the firm's credit rating or to an accounting measure of risk, such as the leverage or the. 6.

(7) debt-to-EBITDA ratios. Performance pricing provisions can be divided into two categories: an interest-increasing covenant when firms experience bad performance and creditors can charge higher interest rates and an interest-decreasing covenant when firms have good performance and creditors can lower interest spreads specified in the loan contract. Loans with performance pricing will punish borrowers with higher spreads if borrowers’ performance deteriorates while will reward borrowers with lower spreads if they outperform their initial state.. Loan contract may include both. 政治大 covenants, and the same financial variables can be set on all these provisions. 立. interest-increasing and interest-decreasing performance pricing as well as financial. The theoretical literature showed that PPC can reduce transaction costs, alleviate. ‧ 國. 學. asset substitution and signal firm’s financial conditions (Asquith, 2005, Bhanot, 2006,. ‧. Manso et al, 2010). While these theories explain some reasons of PPC inclusion, the. sit. y. Nat. inclusion still remains a puzzle in some situations. Adam and Streitz (2014) identified. io. er. that there is only a few public bond issues include performance pricing covenants. In contrast, the use of performance pricing is common in loan contracts1 (47% of their. al. n. v i n C h covenants). ByUanalyzing our sample of USA sample contain performance pricing engchi loan market data, we observe that the use of PPC differs between types of financial. institutions. Table 1 shows the percentage of usage of PPC by four types of financial institutions. It is clear that traditional commercial banks tend to use more performance pricing covenants in loan contracts than the other three types of financial institutions (Finance Company, Investment Bank, and Insurance Company).. [Insert Table 1 here]. 1. Adam and Streitz (2014) identify only 115 public bond issues with performance pricing provisions from 74 distinct companies between 1989 and 2012. 7.

(8) Furthermore, we examine the difference of using PPC by separating financial institutions into “Domestic” and/or “Foreign”. Table 2 outlines the total lending amounts from 1990 to 2010 of both “Domestic” and “Foreign” institution in the USA loan market. As showed in Table 2, we know that syndicate loans seem to clustered in local financial institutions. Table 3 shows that the probability of using PPC by domestic financial institution is higher than by foreign financial institutions.. [Insert Table 2 and 3 here]. 政治大 There should be no concerns about hold-up problem in public bond markets but 立. there is serious hold-up problem in loan contracts, where the four types of financial. ‧ 國. 學. institutions may have different market strategies toward different types of loan. ‧. customers. Finally, a fundamental difference between foreign- and domestic banks is. y. Nat. that domestic banks emphasis more on long-term relationship. The difference of in. er. io. sit. loan market share between “Domestic” and “Foreign” financial institutions may also affect their bargaining power. Foreign banks may price products bellow domestic. n. al. competitors and offer better. v i n C terms in loan agreements h e n g c h i U to. capture more market. share .The situations mentioned above address three questions. Why do firms issue loan with PPC? What is the main reason for creditors to include PPC in loan contracts? Does the bargaining power or lending relationship affect the decision to include PPC in loan contracts? Adam and Streitz (2014) argue that PPC is used to reduce hold-up problem in long-term banking relationships. The use of PPC is more common if the borrower is “locked in” by lenders. For example, banks can exploit borrowers by charging higher interest rates or refusing interest rate reductions when the borrower`s performance improves. However, they do not explain how hold-up problem affects the use of 8.

(9) different types of PPC (mixed, interest-increasing, interest-decreasing)2. To explore the difference of using PPC, we examine whether a long-term lending relationship or the bargaining power may lead to the discrepancy in using different types of PPC. We posit that there is a relation between the use of different types of PPC and use several proxies for lending relationship and lender-borrower bargaining power. Using a large loan sample from the USA syndicate loan market data between 1993 and 2010, we find that the use of PPC is more common if there is a steady long-term lending relationship.. 政治大 types of borrowers. We find that interest-increasing PPC is more common in 立 We further analyze whether the use of PPC varies systematically across different. relationship lending arrangements with smaller firms, firms with a lower long-term. ‧ 國. 學. issuer credit rating at the time of the loan origination, and firms with lower analyst. ‧. coverage. The presence of a lending relationship between borrowers and lead. sit. y. Nat. arrangers reduces the probability of using interest-increasing PPC. This is consistent. io. er. with the argument that lead arrangers cannot capture all rents from holding-up borrowers. Thus for lead arrangers hold-up is a less attractive strategy than preserving. n. al. the relationship with the client.. Ch. engchi. i n U. v. Finally, we find that the use of interest-decreasing PPC is more likely to happen when lead arrangers rank lower in loan market league table. The results support our hypothesis that lenders with less loan market share may try to use more interest-decreasing PPC to attract borrowers in relationship loans. The rest of the paper proceeds as follows. Section 2 reviews the related literature about the use of performance pricing covenant. Section 3 develops hypotheses based on different types of PPC. Section 4 describes methodology. Section 5 discusses our empirical results, and Section 6 concludes. 2. The authors only show the three types of PPC are positively correlated with relationship strength. 9.

(10) 2.. Review of the Literature. Performance pricing covenants are widely used clauses in loan contracts to adjust loan spread when certain pre-determined criteria are met. Prior literature has provided theoretical models of PPC from an optimal dynamic contracting perspective, which linked PPC to reducing debt renegotiation costs, signaling, and alleviating asset substitution. Asquith et al. (2005) argue that interest-decreasing performance pricing covenants reduce loan renegotiation costs which are associated with prepayment and. 政治大 interest-increasing performance 立 pricing covenants reduce moral hazard costs without the adverse selection costs which are associated with information asymmetry while. ‧ 國. 學. imposing liquidity risk on borrowers. Manso et al. (2010) develop a screening model to demonstrate that PPC can be used as a signaling device in a setting with. ‧. information asymmetry. Finally, Koziol and Lawrenz (2010) show that it is optimal to. sit. y. Nat. solve the agency conflict by the inclusion of a performance pricing provision.. n. al. er. io. However, Bhanot and Mello. (2006) argue that PPC is an inefficient method to reduce. i n U. v. the equityholder–debtholder conflict due to asset substitution. Adam and Streitz (2014). Ch. engchi. have linked PPC to mitigating hold-up problem in long-term banking relationships after they identify the phenomena that PPC is rare in public but common in private debt markets. Other studies document a link between PPC and earnings management (Beatty and Weber, 2003), manager equity incentives (Tchistyi et al. 2011), and managerial optimism (Adam et al. 2014). Adam and Streitz (2014) are closely related to our work. While they link PPC to banking relationships by distinguishing between accounting-based and rating-based PPC, our study emphasize more on the determinants to include different types of PPC. Our paper is also related to the literature of lending relationships. Bharath et al.. 10.

(11) (2011) find that repeated borrowing from the same lender translates into lower loan spreads and likelihood that collateral will be pledged. Sharpe (1990) and Rajan (1992) show that one cost of relationship lending is the hold-up problem imposed by the lender. Lenders may take advantage of its monopoly power over information about borrowers and thus exploit borrowers by charging higher interest rates or by refusing interest rate reductions when the borrower’s performance improves .This is especially the case for opaque borrowers which have fewer financing alternatives. Schmidt(2006) argues that the use of covenants, which is common in loan contracts, further. 政治大 borrowers to lenders. Von Thadden (1995) argues that one solution to the hold-up 立 strengthen the hold-up problem because covenants shift bargaining power from. problem is to specify contract terms ex ante, thereby limiting the direction of the. ‧ 國. 學. lender. Bharath et al. (2006) argue that a relationship lender’s informational advantage. ‧. over a non-relationship lender may generate a higher probability of being chosen to. sit. y. Nat. provide future loans. In other words, borrowers that suffer from greater information. io. al. n. relationship lenders.. er. asymmetry (e.g., firms with no rating) are more likely to get future loans from their. Ch. engchi. 11. i n U. v.

(12) 3. Hypothesis Development. Sharpe (1990) and Rajan (1992) show that banks which previously lend to firms learn more about borrower`s characteristics than do other banks. The difference of information asymmetry between the relationship lender and other potential lenders can be costly for borrowers. Adverse selection makes it difficult for one bank to attract another bank`s good customers without drawing off the less desirable ones because good firms in relationship lending have difficulties in conveying information. 政治大 makes borrowers difficult to 立switch to another lender. Relationship lenders, which are about their superior performance to other banks. On the other hand, adverse selection. ‧ 國. 學. also informed banks, can take advantage of their information monopoly and capture some rents from borrowers.. ‧. This is especially the case in the event of covenant violation, because financial. sit. y. Nat. covenants shift bargaining power from borrowers to lenders in this situation. Von. n. al. er. io. Thadden (1995) argues that one solution of mitigating hold-up problem is to. i n U. v. pre-specify contract terms ex ante, thereby limiting the discretion of the lender. PPC. Ch. engchi. can be regarded as such contract terms. PPC specify higher (lower) interest spreads if firms have bad (good) performance in the future. Relationship lenders could not improperly raise the interest rate in the event of covenant violation by using PPC. Similarly, Relationship lenders could not refuse interest spread reductions once a borrower`s performance improve. Since there is a higher probability that hold-up problem exists in repeated relationship lending, we formulate our first hypothesis as follows:. H1: Relationship loans are more likely to include performance-pricing covenants. 12.

(13) than none-relationship loans. When renegotiating a loan contract, a relationship lender must strike a balance between the short-term benefits of exploiting borrowers and the long-term benefits of maintaining the relationship. Lead arrangers must share the benefits of hold-up with the rest of arrangers or participants. Therefore, a relationship lead arranger is less likely to hold-up a borrower while the short-term benefits of relationship accrue mostly to the relationship lender. Accordingly, the use of interest-increasing performance-pricing covenants should be less likely compared to non-relationship. 政治大. deal. We therefore expect the hypothesis that. Relationship. loans. are. less. likely. to. include. interest-increasing. 學. ‧ 國. H2:. 立. performance-pricing covenants than non-relationship loans in syndicated loan. ‧. contracts.. sit. y. Nat. io. er. Our next research question is whether the severity of the hold-up problem can vary systematically for different types of borrowers. The degree to which a borrower. al. n. v i n C h relationship depends is “locked in” by a lender in a lending on whether a firm can get engchi U. sufficient fund through other financing sources, such as initial public offering or public bond market access, and the opaqueness of the borrower. In addition, if these borrower characteristics strengthen the severity of hold-up problem, their lender should be more likely to use interest-increasing PPC but less likely to use interest-decreasing PPC. Our second analysis will place an emphasis on testing whether borrower characteristics can affect hold-up problem and therefore the use of performance-pricing covenants.. H3: Firms with less outside financing alternatives or worse characteristics are 13.

(14) more likely to use interest-increasing performance pricing covenants, rather than interest-decreasing performance pricing covenants.. A lead arranger with a higher loan market league table rank must have more bargaining power when negotiating a syndicated loan contract. A relationship bank with more bargaining power may strengthen the severity of hold-up problem and thus more likely to require the inclusion of performance-pricing covenants in loan contracts. Loans with performance-pricing provision will punish borrowers with. 政治大 they outperform their initial state. 立. higher spreads if borrower performance deteriorates while will reward borrowers with lower spreads if. An interest-decreasing. performance pricing provision can be valuable for a lender who is trying to attract. ‧ 國. 學. high quality borrowers. On the other hand, high quality borrowers may also demand. ‧. interest-decreasing PPC in loan contracts. We expect that (1) lenders with lower. sit. y. Nat. ranking may try to use more interest-decreasing PPC to attract borrowers, thus. io. er. maintaining their relationship with borrowers (2) high quality borrowers require including interest-decreasing PPC in loan contracts. We therefore test the following. n. al. hypothesis:. H4:. High. quality. Ch. borrowers. engchi are. more. i n U. likely. v. to. require. including. interest-decreasing performance-pricing covenants in relationship loans. On the other hand, Lead arrangers with higher ranking are more likely to include interest-increasing. and. less. likely. to. include. performance-pricing covenants in relationship loans.. 14. interest-decreasing.

(15) 4. Data Descriptions. In this section, we briefly describe primary data sources of our research and how we construct our data set, e.g., how performance pricing types are defined. Then we show the explanatory variables we will use in our empirical analysis. Lastly we provide descriptive statistics regarding loan facilities, borrower characteristics for our full sample as well as subsamples of loans that include interest-decreasing and interest-increasing performance pricing provisions.. 立. 4.1 Data Sources. 政治大. ‧ 國. 學. We obtain loan sample from the Thomson Reuters Loan Pricing Corporation Dealscan. ‧. (LPC`s Dealscan) database, which contains detailed information on syndicated loan. y. Nat. sit. contracts since 1982. We extract our sample of loans issued by U.S non-financial. n. al. er. io. borrowers from 1993 to 20103. We conduct our analysis on the facility (tranche) level. i n U. v. because performance pricing features are determined for each facility. We obtain. Ch. engchi. information on loan characteristics such as facility amount, facility maturity, loan type, as well as loan facility purpose. In addition, we record whether a loan is secured or not. We exclude loans whose size (loan amount), and maturity are missing from Dealscan. To obtain borrower-specific information, We merge our loan data with Standard and Poor`s Compustat North American database4, such as firm size, S&P credit rating of firms, and market-to-book ratio , from the last available fiscal quarter before the 3. Prior to 1993, virtually no contracts include a performance-pricing provision according to LPC. We conclude that Dealscan’s data quality with respect to PPC in insufficient prior to 1993 as PPC was first introduced in the 1970s. 4 We use Michael Robert`s Dealscan-Compustat Linking Database to merge Dealscan with Compustat. 15.

(16) loan issue. We exclude those borrowers whose financial data (firm size, leverage, market value ... etc.) are missing in Compustat. Using the Institutional Brokers' Estimate System (I/B/E/S) data base, we obtain the number of analysts following the firm, which is a proxy for borrower opacity. Our final dataset consists of a sample of 26,877 closed or completed loan facilities in the U.S. market. Appendix I contains the definitions of all variables in our analysis.. 4.2 Performance Pricing Types. 政治大 We follow Cai et al. (2012)立 to distinguish between different performance pricing types.. ‧ 國. 學. Because Dealscan database cannot indicate whether a performance pricing provision is interest-increasing or interest-decreasing directly, we use the pricing grid of. ‧. Dealscan to identify the specific type. The pricing grid provides us detailed pricing. sit. y. Nat. information and the requirement based on either financial ratio or credit rating at each. n. al. er. io. degree. Financial ratios can be divided into two groups: the first group with smaller. i n U. v. ratios indicating better performance (e.g., leverage ratio), while the second group with. Ch. engchi. higher ratios indicating better performance (e.g., EBITDA). Based on the pricing grid, we determine the highest and lowest levels of pricing. According to Dealscan, all-in-spread-drawn, which expressed as a spread over LIBOR, is a measure of the overall cost of loans at starting point of loan facilities. A performance-pricing covenant is defined as interest-increasing if the starting all-in-spread-drawn is at the lowest level in the pricing grid and the performance measure (financial ratio or credit rating) is at the lowest or highest level depending on which group it is in. On the contrary, a performance pricing covenant is defined as interest-decreasing if the all-in-spread-drawn is the highest at starting point of the loan. 16.

(17) facility and the performance measure is the highest or lowest depending on its group. If the all-in-spread-drawn at the time of origination is neither at the highest nor lowest level of the pricing grid, we label the loan facility as one with both interest-increasing and interest-decreasing performance pricing (mixed). We further identify whether interest-increasing and interest -decreasing performance pricing covenants are based on financial ratios or credit ratings. Accounting-based PPC uses measures of financial ratios and Rating-based PPC relies on firm`s credit rating.. 4.3 Measuring Relationship 治政Strength. 立. 大. ‧ 國. 學. We follow Bharath et al. (2011) and construct three alternative proxies of relationship strength by searching the past borrowing record of borrowers. These proxies are. ‧. designed to pick up the existence of prior lending by the same lender in the past. To. sit. y. Nat. construct these proxies, we first need to identify the lead lender(s) for each loan. al. er. io. contract. We classify a lender as the lead lender if the variable "Lead Arranger Credit". v. n. (provided by LPC’s Dealscan) takes on the value "Yes", or if the lender is the only. Ch. lender specified in the loan contract.. engchi. i n U. Next, we search the borrowing record of borrowers over the past five years. The first lending relationship strength proxy, Rel(Dummy), is a dummy variable which equals one if the firm borrowed from the same lead lender in the previous five years and zero otherwise. If there are multiple lead lenders retained, we calculate the proxy separately for each lender and assign the highest value to the loan. The second proxy, Rel(Number), measures the relative number of loans obtained from the relationship lender. For bank m lending to borrower i, it is calculated as follows.. 17.

(18) Re l ( Number )m . Number of loans by bank m to borrower i in the last 5 years Total number of loans by borrower i in the last 5 years. Again, the highest value is assigned to a loan if there are multiple lead lenders. The third proxy, Rel(Amount), measures the relative loan amounts obtained from the relationship lender. For bank m lending to borrower i, it is calculated as follows.. Re l ( Amount ) m . Amount of loans by bank m to borrower i in the last 5 years ($) Total amount of loans by borrower i in the last 5 years ($). Again, the highest value is assigned to a loan if there are multiple lead lenders.. 政治大 4.4 Lead Arranger 立 Ranking. ‧ 國. 學. To assign the lead arranger ranking score, we construct the league table for lead. ‧. arrangers in USA syndicated loans market between 1993 and 2010. The lending. sit. y. Nat. amount for each bank is estimated from each deal’s lending amount times its bank. n. al. er. io. allocation. We use bank-allocation and lending amount in LPC`s Dealscan database. If. i n U. v. bank-allocation is missing or the sum value of each facility is less than 100%, we. Ch. engchi. assume leader arrangers take responsibility for the whole lending amount. Table 4 shows five largest arrangers of USA syndicated loans from 1993 to 2010.. [Insert Table 4 here]. Because our analysis is on the facility (tranche) level, we construct a variable “Bank Ranking” which refers to the lead arranger ranking score of each tranche according to the league table from 1993 to 2010, the lead arranger ranking score ranges from ten to zero. For example, a score of ten means the lead arranger in one. 18.

(19) facility is the top lead arranger in syndicated loan market and a score of zero means it is out of the top ten lead arrangers .If there are multiple lead arrangers in one facility, we assign the highest lead arranger ranking score to the facility.. 4.5 Descriptive Statistics. Table 5 presents descriptive statistics for our sample consisting of 26,877 loan tranches issued by 4,930 non-financial borrowers between 1993 and 2010. The data. 政治大 characteristics, which are立 consistent with prior studies (e.g. Adam (2014)).For. are winsorized at the 1% and 99% levels to remove outliers. Panel A reports loan. ‧ 國. 學. example, the mean/median facility amounts in our sample are $317/$106 million, which is large given the mean/median book value of assets of $4,559/$814 million. ‧. and an average leverage ratio of 30%. The average maturity is 47 months. 27% of. sit. y. Nat. loan tranches are term loans. We can also observe that roughly 42% of loans include a. n. al. er. io. performance-pricing provision. Panel B reports borrower characteristics variables. In. i n U. v. 59% of cases, borrowers do not have a credit rating. But if a rating exists it tends to be. Ch. engchi. around the investment grade threshold. Panel C shows the mean/median lead arranger ranking score we assign in our samples. Because we assign lead arranger ranking score from 10 to 0 , we observe that at least 50% facility is conducted by the top 4 lead arranger in our sample( median arranger ranking score of 7). Panel D reports descriptive statistics on the three relationship lending proxies. A lending relationship exists in 39% of all loan contracts. On average, 17% of the total capital raised over the course of 5 years was raised from the same lead lender. Table 6 shows the various performance measures used in PPC contracts. The most common performance measure is the Debt-to-EBITDA ratio (48%), followed by. 19.

(20) the senior debt rating (28%). The remaining performance measures are mostly other types of leverage ratios. In at least 4% of cases, leverage performance measures are used. We define PPC as accounting-based PPC whenever a financial ratio is used as a measure of firm performance. Rating-based PPC comprise all PPC contracts, which use borrower’s credit rating as a performance measure.. [Insert Table 5 here]. 政治大. [Insert Table 6 here]. 立. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 20. i n U. v.

(21) 5. Empirical Analysis 5.1. Banking Relationships and the Use of Performance Pricing Covenants. We begin by analyzing the interaction between lending relationship and the choice between PPC and straight debt. As noted in Section 4.2, we distinguish between accounting-based and rating-based PPC. We therefore estimate a multinomial logistic regression.. it. it. it. (1). 學. ‧ 國. PPCit . 政治大  +  Rel 立  M +   X + . The dependent variable, PPCit, is a discrete variable, which equals one if the. ‧. loan contract contains a performance pricing provision based on an accounting. Nat. sit. y. measure, two if the loan contract includes a performance pricing provision based on. n. al. er. io. the borrower’s credit rating, or three in the case of straight debt (control group). Rel. i n U. v. (M) represents one of our three measures of relationship strength, and X it are control. Ch. engchi. variables to control for heterogeneity. We use firm size, measured by the log of total assets, the market-to-book ratio of assets, leverage, tangibility, profitability, the current ratio, the loan amount, the deal maturity, and a dummy variable for secured loans as control variables. We also include loan purpose and loan type indicators, time fixed effects, industry fixed effects (we use Fama French industry classification), and dummy variables for each rating level. We cluster the standard errors at the firm level to account for non-independent observations within firms. Table 7 reports the regression results.. 21.

(22) [Insert Table 7 here]. Adam and Streitz (2014) argue that accounting-based PPC is used to address hold-up problems in repeated lending relationships, while rating-based PPC is more likely used to signal credit quality. In their samples, relationship strength appears to be uncorrelated or marginally negatively correlated with the use of rating-based PSD 5 . Thus, rating-based PSD are unlikely to be motivated by hold-up considerations due to lending relationships.. 政治大 with Hypothesis 1, we find that relationship strength is positively and significantly 立. However, our empirical result is different from Adam and Streitz (2014).Consistent. correlated with the use of both accounting-based PPC and rating-based PPC.. ‧ 國. 學. Therefore, we still use both accounting-based and rating-based PPC in the following. sit. y. Nat. Syndication and the Use of Performance Pricing. io. al. n. Covenants.. er. 5.2. ‧. analysis.. Ch. engchi. i n U. v. A significant portion of our sample consists of syndicated loan contracts. According to Hypothesis 2, a relationship lead arranger should find it less beneficial to hold-up a borrower compared to a single lender because the gains from hold-up would have to be shared with the rest of the syndicate. As a result, the use of performance-pricing provisions should be less likely if lead arrangers are relationship banks. To test Hypothesis 2, we estimate the following multinomial logistic regression.. 5. In their samples, Rel(Dummy) is uncorrelated with rating-based PPC; Rel(Number) and Rel(Amount) are marginally negatively correlated with the use of rating-based PSD. 22.

(23) PPCTYPE it =  + 1  Rel  M it + 2  Syndicationit. + 3  Rel  Mit  Syndicationit +   Xit +  it. (3). The dependent variable, PPCTYPE, is a discrete variable, which equals one if the loan contract contains a mixed PPC, two if the loan contract includes an interest-increasing PPC, three if the loan contract includes an interest-decreasing PPC, or four in the case of straight debt (control group). Syndication is a dummy. 政治大. variable, which equals to one if our sample is from syndicated loan contracts and. 立. zero otherwise.. ‧ 國. 學. As reported in Table 8, we find the use of performance pricing provisions is indeed less likely in syndicated loan contracts when a banking relation exists. The. ‧. coefficient on Rel(Dummy)*Syndication in the interest-increasing section is. sit. y. Nat. negative and statistically highly significant6, which supports our hypothesis that the. n. al. er. io. use of interest-increasing performance-pricing covenants should be less likely. i n U. v. compared to non-relationship deal in syndicated loan contracts... Ch. engchi. [Insert Table 8 here]. 6. The result is not significant to using our other measure of relationship strength.(See Appendix Table A.IV. ) 23.

(24) 5.3. Borrower Opacity and the Use of Performance Pricing Covenants.. Our empirical result so far shows that relationship lending is positively correlated with the use of PPC (both accounting-based and rating-based included). To establish whether this positive correlation is due to hold-up, we make use of the fact that the severity of the hold-up problem is likely to vary systematically across different types of borrowers. For example, more opaque borrowers have fewer financing alternatives, so that these borrowers are more subject to hold-up. We use firm size,. 政治大. which is measured by the log of total assets, as our first proxy for opacity. We expect. 立. larger borrowers are less likely to include interest-increasing performance-pricing. ‧ 國. 學. provisions because larger borrowers have more financing alternatives. Next, we use a variable “Rating”, which is a debt-rating score we assign according to the. ‧. borrower`s long-term S&P rating in each facility start year. For example, a borrower. Nat. sit. y. with “AAA+” rating gets the highest score and a borrower with no long-term S&P. n. al. er. io. rating gets the lowest rating score of zero. Another proxy for opacity is the number. i n U. v. of analysts following the firm. Finally, we use a dummy variable, IPO7, which. Ch. engchi. equals to one if borrowers have initial public offering at the time of the loan origination and zero otherwise. Larger firms, firms with higher rating, firms with larger analyst coverage, and public traded firms are more likely to have multiple financing alternatives, and are thus less "locked in" by lenders in bank lending relationships. To test for the cross-sectional variation in the severity of the hold-up problem induced by lending relationships, we estimate the following model of multinomial logistic regression.. 7. Company with IPO usually has financial benefit in the form of raising capital 24.

(25) PPCTYPE it =  + 1  Rel  M it + 2  BorrowerOpacityit. + 3  Rel  M it  BorrowerOpacityit +   Xit +  it. (2). The dependent variable, PPCTYPE, is a discrete variable, which equals one if the loan contract contains a mixed PPC, two if the loan contract includes an interest-increasing PPC, three if the loan contract includes an interest-decreasing PPC, or four in the case of straight debt (control group). BorrowerOpacity stands for the above-mentioned proxies for borrower opacity. We include interaction variables. 政治大 variables. Due to the high correlation of interaction variables, we include one 立. of relationship strength and BorrowerOpacity to test for the joint effect of these two. variable at a time in the regressions. The results are reported in Table 9.. Nat. sit. y. ‧. ‧ 國. 學 [Insert Table 9 here]. io. er. The empirical results of all interaction variables of borrower opacity are as follows: First, the coefficient of interaction variables of borrower opacity is negative and. al. n. v i n Ch statistically significant in the interest-increasing section when we use firm size as the engchi U. proxy. Second, the coefficients of the two interaction variables of borrower opacity are negative and statistically significant in the interest-increasing section when we use firm`s rating 8 and the number of analysts as the proxy. However, the. coefficients of the two interaction variables are not statistically significant in the interest-decreasing part. Finally, the coefficient of interaction variables of borrower opacity is insignificant in the interest-increasing section when we use IPO as the proxy for borrower opacity but it became positive and statistically significant in the. 8. We also use the other dummy variable ” Rated” which equals one if company has a S&P rating (or zero otherwise ). However, the result is not statistically significant (see Appendix Table X.) 25.

(26) interest-decreasing part. In summary, the first three coefficients of interaction variables of borrower opacity (firm size , firm rating ,and the number of analysts following the firm ) are negative and statistically significant in the interest-increasing section, which supports our hypothesis that opacity in the presence of a lending relationship increases the severity of hold-up, and hence the likelihood of using interest-increasing PPC 9. The result of the last proxy shows that public traded firms are more likely to alleviate the severity of hold-up between repeated banking. 政治大. relationships, thus increase the likelihood of using interest-decreasing PPC10.. 立. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 9. i n U. v. However, The results are not significant to using our other measure of relationship strength.(See Appendix Table A.II. and A.III.) 10 The result is robust to using our other measure of relationship strength. 26.

(27) 5.4. Bargaining Power and the Use of Performance Pricing Covenants.. In this section, we investigate whether the use of performance pricing is influenced by the bargaining power between lenders and borrowers. Firstly, we use market share of firms in their own industry to measure the bargaining power of borrowers. To calculate the market share of one borrower in each year, we use the financial data of sales from compustat and the first two number of SIC code to distinguish between different industries. The market share of each borrower then is matched according to. 政治大. the facility start year. In other words, one borrower has different market share because. 立. of its facilities in different years. Next, as mentioned in section 4.4, we use lead. ‧ 國. 學. arranger ranking score of lenders as a proxy for the bargaining power of lenders. To. Nat. sit. PPCTYPE it =  + 1  Rel  M it + 2  MarketShareit. y. ‧. test Hypothesis 4 and 5, we estimate the following multinomial logistic regression.. al. n PPCTYPE it. (4). er. io. + 3  Rel  Mit  MarketShareit +   Xit +  it. i n U. v. Ch e hi =  + 1  Rel  M it +ng2 cBankRanking it + 3  Rel  Mit  BankRankingit +   Xit +  it. (5). The dependent variable, PPCTYPE, is a discrete variable, which equals one if the loan contract contains a mixed PPC, two if the loan contract includes an interest-increasing PPC, three if the loan contract includes an interest-decreasing PPC, or four in the case of straight debt (control group). MarketShare and BankRanking stand for the above-mentioned proxies for borrower and lender bargaining power. We include interaction variables of relationship strength and above-mentioned proxies to 27.

(28) test for the joint effect of these two variables.. [Insert Table 10 here]. As shown in Table 10, the coefficient of interaction variables are not significant in both of interest-increasing and interest-decreasing section when we use firm`s market share to evaluate the bargaining power of firms11. This appears not to be supportive of Hypothesis 4.. 立. 政治大 [Insert Table 11 here]. ‧ 國. 學. However, As shown in Table 11, the result is different when we use lead arranger. ‧. ranking score as a proxy of bargaining power of lenders. Although the coefficient of. y. Nat. interaction variable is still not significant in the interest-increasing section, the result. er. io. sit. of interest-decreasing section is negative and statistically significant 12 , which is consistent with our Hypothesis 4 that lead arrangers with higher bank ranking in. al. n. v i n C toh include interest-decreasing relationship loans are less likely performance-pricing engchi U. covenants in syndicated loan contracts.. 11. 12. The result is still not significant to using our other measure of relationship strength.(See Appendix Table A.V. ) The result s is robust to using our other measure of relationship strength. (See Appendix Table A.VI. ) 28.

(29) 6. Conclusion This thesis explores whether long-term banking relationships and bargaining power are important determinants of performance-pricing covenants (PPC) inclusions in syndicated loan contracts. Sharpe (1990) and Rajan (1992) argue that hold-up problem imposed by the lender is one cost of relationship lending. Von Thadden (1995) shows that pre-specifying loan contract terms can be an efficient way to mitigate hold-up problems in long-term banking relationships. An example is performance-pricing. 政治大 would otherwise trigger debt 立renegotiations. In this paper, we test the hypothesis that. covenants (PPC), which pre-specifies loan contract terms in prevention of events that. ‧ 國. 學. performance-pricing covenants are used because of potential hold-up problems in long-term banking relationships.. ‧. Consistent with this hypothesis, we find that interest-increasing PPC is more. sit. y. Nat. common when hold-up problem exists in relationship lending. This is especially the. n. al. er. io. case if borrowers are opaque and have fewer financing alternatives, both of which. i n U. v. imply a greater potential for hold-up. However, we do not have enough evidence that. Ch. engchi. the use of interest-decreasing PPC is associated with the opaqueness of borrower. In syndicated deals, relationship lenders as lead arrangers should find it less beneficial to hold-up borrowers as the gains from hold-up would have to be shared with the other syndicate members. This reduces the need for interest-increasing PPC. Indeed, we find that the presence of a lending relationship between borrowers and lead arrangers reduces the use of interest-increasing PPC. In addition, we find that the use of interest-decreasing PPC is more likely when lead arrangers rank low in loan market league table. The result is consistent with our expectation that lenders with less market share may try to use more. 29.

(30) interest-decreasing PPC to attract borrowers in relationship loans. In conclusion, we find that hold-up and bargaining power are important determinants to include different types of performance-pricing covenants.. References. Adam, T., D. Streitz,. 2014. Hold-Up and the Use of Performance-Sensitive Debt,. 政治大 Cai, J., Mattes, J. A., & Steffen, S. 2012. Performance Pricing versus Financial 立. Working Paper.. Covenants: Agency Costs and Incentive Alignment, Working Paper.. ‧ 國. 學. Manso, G., B. Strulovici, and A. Tchistyi. 2010. Performance-sensitive debt. Review. ‧. of Financial Studies 23 (5), 1819–1854.. sit. y. Nat. Asquith, P., A. Beatty, and J. Weber .2005. Performance pricing in bank debt contracts.. io. er. Journal of Accounting and Economics 40, 101–128.. Roberts, M. R., & Sufi, A. 2009. Control rights and capital structure: An empirical. n. al. i n C h 64(4), 1657-1695. investigation. The Journal of Finance, engchi U. v. Demiroglu, C., & James, C. M. 2010. The information content of bank loan covenants. Review of financial Studies, 23(10), 3700-3737. Murfin, J. 2012. The Supply‐Side Determinants of Loan Contract Strictness. The Journal of Finance, 67(5), 1565-1601.. Renzis, De T., Francis, B., Hasan, I., 2010. Performance Pricing Provisions in Bank Loans: A Cross-Country Analysis, Working Paper. Adam, T., V. Burg, T. Scheinert, and D. Streitz. 2014.Managerial Optimism and Debt Contract Design, Working Paper. Panyagometh, K., Roberts, G. S., Gottesman, A. A., & Beyhaghi ,M. 2013. 30.

(31) Performance Pricing Covenants and Corporate Loan Spreads. Working Paper. Bhanot, K. and A. S. Mello .2006. Should corporate debt include a rating trigger? Journal of Financial Economics 79, 69–98. Bharath, S., S. Dahiyab, A. Saunders, and A. Srinivasan .2007. So what do i get? the bank’s view of lending relationships. Journal of Financial Economics 85, 368–419. Bharath, S. T., S. Dahiya, A.Saunders, and A.Srinivasan .2011. Lending relationships and loan contract terms. The Review of Financial Studies 24, 1142–1203. Dass, N. and M. Massa .2011. The impact of a strong bank-firm relationship on the. 政治大 Degryse, H.and S.Ongena .2005. Distance, lending relationships, and competition. 立 borrowing firm. Review of Financial Studies 24, 1204–1260.. The Journal of Finance 60, 231–266.. ‧ 國. 學. Agarwal, S. and R. Hauswald .2010. Distance and private information in lending.. ‧. Review of Financial Studies 23, 2757–2788.. sit. y. Nat. Koziol, C.and J. Lawrenz .2010. Optimal design of rating-trigger step-up bonds:. io. 182–204.. er. Agency conflicts versus asymmetric information. Journal of Corporate Finance 16,. al. n. v i n Coutsiders: Rajan, R. G. 1992. Insiders and choice between informed and arm’s U h e n gThe i h c length debt. The Journal of Finance 47, 1367–1400.. Ross, D. G.2010. The “dominant bank effect:” how high lender reputation affects the information content and terms of bank loans. Review of Financial Studies 23, 2730–2756. Tchistyi, A., D. Yermack, and H.Yun .2011. Negative hedging: Performance-sensitive debt and ceos’ equity incentives. Journal of Financial and Quantitative Analysis 46, 657–686. Von Thadden, E.-L. 1995.Long-term contracts, short-term investment and monitoring. Review of Economic Studies 62, 557–575. 31.

(32) Table 1：PPC Use Percentage by Financial Institution Types This table reports the financial institution types and frequencies of using PPC in our sample of PPC from 1993 to 2010 Observations. Frequency. 10,822. 12.83%. Interest-Increasing. 1,738. 2.06%. Interest-Decreasing. 8,218. 9.74%. Straight debt. 63,588. 75.37%. Total. 84,366. 100%. Panel A：Traditional Bank Mixed. Panel B： Finance Company Mixed Interest-Increasing. 立. 6.56% 1.77% 83.98%. Total. 12,147. 100% 7.1%. Interest-Increasing. 100. 1.82%. Interest-Decreasing. 561. 10.22%. 學. 10,201. y. Straight debt. sit. 7.69%. ‧. 934. ‧ 國. Interest-Decreasing. 政治大797 215. Panel C：Investment Bank Mixed. 390. Nat. al. n. Total. 4,440. er. io. Straight debt. i n U. v. 80.86%. 5,491. 100%. 5. 5.1%. Interest-Increasing. 0. 0.00%. Interest-Decreasing. 0. 0.00%. Straight debt. 93. 94.9%. Total. 98. 100%. Ch Panel D：Insurance Company Mixed. engchi. 32.

(33) Table 2：Total Lending Amount from 1990 to 2010 in USA This table reports the leader arranger lending Amount from 1990 to 2010 in USA loan market from LPC database. Domestic Year. Foreign. Lending Amount (US$). %. Lending Amount (US$). %. 1990. 184,924,162,080 83.29%. 37,107,015,213 16.71%. 1991. 204,811,925,778 85.24%. 35,456,873,005 14.76%. 1992. 272,939,025,109 80.89%. 64,494,937,467 19.11%. 1993. 408,856,954,618 83.86%. 78,705,938,032 16.14%. 1994. 636,063,781,563 86.48%. 2001. 140,819,539,872 11.22%. 781,772,895,703 87.50%. 111,719,401,238 12.50% 110,535,616,900. 9.79%. 1,187,757,315,379 90.35%. 126,866,081,210. 9.65%. 1,168,635,591,166 89.76%. 133,308,219,513 10.24%. y. 1,018,946,314,565 90.21%. Nat. 2000. 16.19%. 1,114,368,788,896 88.78%. ‧. 1999. 18.09%. io. sit. 1998. 13.52%. 學. 1997. er. 1996. ‧ 國. 1995. 政治大 99,422,063,169 691,299,386,499 81.91% 152,688,680,677 立 817,601,909,415 83.81% 157,915,329,096. 2002. 1,026,780,937,960 90.05%. 2003. 960,754,356,825 85.63%. 2004. 1,353,641,214,438 86.22%. 2005. 1,603,230,336,393 83.66%. 313,068,891,223 16.34%. 2006. 1,808,901,674,294 79.23%. 474,274,635,750 20.77%. 2007. 2,045,298,824,651 79.08%. 541,067,529,882 20.92%. 2008. 882,358,253,169 76.40%. 272,518,610,702 23.60%. 2009. 592,574,665,318 70.13%. 252,342,793,694 29.87%. 2010. 684,779,624,379 66.96%. 337,917,742,584 33.04%. Total. 19,446,297,938,199 83.18%. 3,931,243,800,217 16.82%. n. al. Ch. engchi. 33. 113,399,107,802. v i161,266,211,125 n U. 9.95% 14.37%. 216,348,582,061 13.78%.

(34) Table 3：PPC Use Percentage between Domestic and Foreign Bank This table reports the use percentage of PPC between Domestic and Foreign Bank in our sample of PPC from 1993 to 2010 Observations. Frequency. Mixed. 9,652. 11.5%. Interest-Increasing. 1,707. 2.03%. Interest-Decreasing. 8,176. 9.74%. Straight debt. 64,407. 76.73%. Total. 政治大 Panel B：Foreign financial institution 立. 83,942. 100%. Mixed. 1,228. 8.26%. Panel A：Domestic financial institution. ‧ 國. 學. Interest-Increasing. 1,152. 7.76%. 12,233. 82.36%. io. y. sit. 14,853. n. al. er. Total. Nat. Straight debt. 1.62%. ‧. Interest-Decreasing. 240. Ch. engchi. 34. i n U. v. 100%.

(35) Table 4: Five largest arrangers of USA from 1993 to 201013 Year. Top 1. Top 2. Top 3. Top 4. Top 5. 1993. Chemical Bank. Chase Manhattan. Morgan Guaranty Trust. Citigroup. Bank of America. 1994. Chemical Bank. Citigroup. Chase Manhattan. Morgan Guaranty Trust. NationsBank. 1995. Chemical Bank. Citigroup. Morgan Guaranty Trust. NationsBank. Bank of America. 1996. Chase Manhattan. Citigroup. Bank of America. Chemical Bank. NationsBank. 1997. Chase Manhattan. Citigroup. Bank of America. NationsBank. 1998. Chase Manhattan. Bank of America. Citigroup. Morgan Guaranty Trust. 1999. Bank of America. Chase Manhattan. Citigroup. BANK ONE. 2000. Chase Manhattan. Bank of America. Citigroup. BANK ONE. 2001. JP Morgan Chase. Bank of America. 政治大 Citigroup. BANK ONE. 2002. JP Morgan Chase. Bank of America. Citigroup. BANK ONE. 2003. JP Morgan Chase. 2004. JP Morgan Chase. 2005. JP Morgan Chase. 2006. JP Morgan Chase. 2007. JP Morgan Chase. 2008. JP Morgan Chase. Bank of America. 2009. JP Morgan Chase. Bank of America. 2010. Bank of America. JP Morgan Chase. ‧ 國. 學. Credit Suisse Boston Credit Suisse Boston. Citigroup. Bank of America. Citigroup. Bank of America. Citigroup. Deutsche Bank. Wachovia Bank. Bank of America. Citigroup. Deutsche Bank. Wachovia Bank. Bank of America. Citigroup. Deutsche Bank. Goldman Sachs Bank NA. Citigroup. Credit Suisse Boston. Wachovia Bank. Wells Fargo. Credit Suisse. Citigroup. Deutsche Bank. n. Ch. y. sit. er. io. al. ‧. Bank of America. Nat. 13. 立. Morgan Guaranty Trust BT Alex Brown Inc Trust Morgan Guaranty Trust Credit Suisse Boston Boston Credit Suisse Boston Credit Suisse Boston. n U e nCredit hi g cSuisse Citigroup. iv. BANK ONE Wachovia Bank. We consider the M&A activities of the top 25 biggest banks from 1990 to 2010 in the USA. 35.

(36) Table 5: Summary Statistics This table reports summary statistics for a sample of 26,877 loan facilities by 4,930 non-financial firms between 1993 and 2010.14 Mean Median Std Min Max Panel A: Loan Characteristics PPC(Accounting) 0.3 0 0.46 0 1 PPC(Rating) 0.12 0 0.32 0 1 Facility Amount (M. USD) 316.88 106.25 771.17 0.04 30,000 Facility Maturity(months) 47 48 29 1 420 Term Loan 0.27 0 0.44 0 1 Sole Lender 0.09 0 0.29 0 1 Secured 0.52 1 0.5 0 1 Loan Purpose: General 0.31 0 0.46 0 1 Loan Purpose: Refinance 0.19 0 0.39 0 1 Loan Purpose: Takeover 0.12 0 0.32 0 1 Loan Purpose: Working Capital 0.18 0 0.38 0 1 Panel B: Borrower Characteristics Total Asset (M. USD) 4559.5 813.71 12694.857 0.406 284508 Leverage 0.3 0.27 0.26 0 6.88 Market-to-Book 1.03 0.72 1.23 0 60 Tangibility 0.34 0.29 0.24 0 0.99 Profitability 0.03 0.03 0.04 -1.42 0.36 Current Ratio 1.89 1.54 2.96 0 411.8 # Analysts 4.35 0 7.92 0 57 Market Share 0.02 0.003 0.05 0 1 Rating AAA 0 0 0.06 0 1 Rating AA 0.01 0 0.12 0 1 Rating A 0.08 0 0.27 0 1 Rating BBB 0.14 0 0.34 0 1 Rating BB 0.16 0 0.36 0 1 Rating B 0.11 0 0.32 0 1 Rating C (or below) 0.01 0 0.12 0 1 Rated 0.41 1 0.5 0 1 Panel C:Lender Characteristric Bank Ranking(score) 5.31 7 4.41 0 10 Panel D:Relationship Lending Proxies Rel (Dummy) 0.39 0 0.49 0 1 Rel (Number) 0.13 0 0.24 0 1 Rel (Amount) 0.17 0 0.27 0 1. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. 14. Ch. engchi. i n U. v. Both loan contracts with performance-pricing covenants and straight bond (loan contracts without performance-pricing covenants)are included in the sample of 26,877 loan facilities. 36. N 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877 26,877.

(37) Table 6: PPC Contract Types This table reports the types and frequencies of performance-pricing provisions in our sample of PPC. Frequency. Observations. Panel A: Accounting-Based PPC Debt-to-EBITDA User Condition15 Multiple16 Leverage Senior Debt to Cash Flow. 0.48 0.06 0.01 0.04 0.03. 5462 715 75 506 338. Fixed Charge Coverage Other Accounting Measures Outstanding17. 0.03 0.01 0.02. 316 115 234 151 283. Panel B:Rating-Based PPC Senior Debt Rating Other Credit Rating. 0.01 0.02 0.00 0.28 0.00. Total. 1.00. 立. 政治大. ‧ 國. y. sit. n. al. er. io. 16 17. ‧. Nat. 15. 學. Debt-to-Tangible Net Worth Interest Coverage. Ch. engchi. i n U. User Condition is classified according to company`s financial status The other financial ratios(e.g. current ratio…EBITDA..etc.) Shares outstanding 37. v. 3192 27 11414.

(38) Table 7: Lending Relationships and the Use of Performance-Pricing Covenant This table reports the marginal effects of multinomial logit regressions to evaluate the likelihoods of using rating-based or accounting-based PPC. The dependent variable equals one if the loan contract contains a performance pricing provision on an accounting measure, two if the loan contract includes a performance pricing provision on the borrower’s credit rating, and three for non performance-pricing loan contracts (straight bond). Marginal effect for each covariate are constructed as the difference in predicted probabilities for a particular outcome computed at their mean values holding all other covariates constant. All item are defined in the Appendix Table A.I. Standard errors are heteroskedasticity robust.*, **,*** indicate statistical significance at the 10%,5%,1% level. (1) (2) (3) PPC(Accounting) PPC(Rating) PPC(Accounting) PPC(Rating) PPC(Accounting) PPC(Rating) Rel(Dummy) 0.0282*** 0.00689* (0.00560) (0.00387) Rel(Number) 0.0497*** 0.0214*** (0.0108) (0.00800) Rel(Amount) 0.0507*** 0.00984 (0.00947) (0.00709) Ln(Total Assets) -0.0253*** 0.00445** -0.0243*** 0.00486** -0.0243*** 0.00470** (0.00305) (0.00213) (0.00304) (0.00213) (0.00304) (0.00213) Leverage 0.0668*** -0.0998*** 0.0704*** -0.0976*** 0.0695*** -0.0987*** (0.0144) (0.0126) (0.0144) (0.0125) (0.0143) (0.0125) Market-to-Book -0.000820 0.00451 -0.000962 0.00465 -0.000896 0.00467 (0.00354) (0.00289) (0.00354) (0.00289) (0.00354) (0.00289) Tangibility -0.0243* 0.0189* -0.0259* 0.0182* -0.0257* 0.0187* (0.0142) (0.0108) (0.0142) (0.0108) (0.0142) (0.0108) Profitability 1.273*** 0.208** 1.284*** 0.206** 1.277*** 0.208** (0.107) (0.0988) (0.107) (0.0987) (0.107) (0.0988) Current Ratio 0.0138*** 0.00270 0.0135*** 0.00263 0.0136*** 0.00266 (0.00271) (0.00245) (0.00271) (0.00245) (0.00271) (0.00245) Ln(Facility Maturity) 0.0835*** -0.0281*** 0.0825*** -0.0281*** 0.0830*** -0.0283*** (0.00478) (0.00303) (0.00476) (0.00301) (0.00477) (0.00301) Ln(Facility Amount) 0.0528*** 0.0431*** 0.0532*** 0.0431*** 0.0528*** 0.0432*** (0.00304) (0.00231) (0.00303) (0.00231) (0.00303) (0.00231) Secured 0.187*** -0.0938*** 0.188*** -0.0935*** 0.188*** -0.0938*** (0.00577) (0.00489) (0.00577) (0.00489) (0.00577) (0.00489) Obs. 26,877 26,877 26,877 Adj R2 0.2728 0.2729 0.2729 Industry Fixed Effect Yes Yes Yes Time Fixed effect Yes Yes Yes Credit Rating Fixed effect Yes Yes Yes Loan Purpose Fixed effect Yes Yes Yes Loan Type Fixed effect Yes Yes Yes. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 38. i n U. v.

(39) Table 8: Syndication and the Use of Performance-Pricing Covenant This table reports the marginal effects of multinomial logit regressions to evaluate the likelihoods of using mixed , interest- increasing ,or interest-decreasing PPC.*, **,*** indicate statistical significance at the 10%,5%,1% level. Mixed 0.154*** (0.0373) 0.269*** (0.0224) -0.146*** (0.0377). Rel(Dummy) Syndication Rel(Dummy)* Syndication Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. 立. PPC Increasing 0.0313*** (0.0110) 0.00239 (0.00699) -0.0323*** (0.0113) 26,877 0.1653 Yes Yes Yes Yes Yes Yes Yes. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 39. i n U. v. Decreasing -0.00596 (0.0248) 0.0887*** (0.0127) 0.0181 (0.0253).

(40) Table 9: Borrow Opacity and the Use of Performance-Pricing Covenant This table reports the marginal effects of multinomial logit regressions to evaluate the likelihoods of using mixed, interest- increasing, or interest-decreasing PPC. Marginal effect for each covariate are constructed as the difference in predicted probabilities for a particular outcome computed at their mean values holding all other covariates constant. All item are defined in the Appendix Table A.I. Standard errors are heteroskedasticity robust.*, **,*** indicate statistical significance at the 10%,5%,1% level. (A) Firm Size Mixed Rel(Dummy) Ln(Total Assets) Rel(Dummy)* Ln(Total Assets). 立. 0.0556** (0.0227) -0.00534* (0.00318) -0.00479 (0.00304). 0.0923*** (0.0181) -0.00597** (0.00279) -0.0113*** (0.00263). ‧ 國. ‧. y. sit. n. er. io. Ch. engchi Mixed. Rel(Dummy)* Rating. Decreasing. 學. Nat. al. (B) Firm S&P long-term debt rating. Rating. 0.0288*** (0.00959) -0.00167 (0.00159) -0.00401*** (0.00136) 26,877 0.1515 Yes Yes Yes Yes Yes Yes Yes. 政治大. Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. Rel(Dummy). PPC Increasing. 0.0307*** (0.00833) 0.0153*** (0.00137) -0.00158* (0.000819). Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. 40. i n U. v. PPC Increasing 0.00567* (0.00332) -0.00281*** (0.000611) -0.000821* (0.000426) 26,877 0.1541 Yes Yes Yes Yes Yes Yes Yes. Decreasing 0.0188*** (0.00660) -0.00768*** (0.00110) -0.000116 (0.000782).

(41) (C) Numbers of analyst Mixed Rel(Dummy). 0.0253*** (0.00633) 0.00101** (0.000504) -0.00122* (0.000642). #Analysts Rel(Dummy)* #Analysts Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. (D) Initial Public Offering. 立. 0.0156** (0.00745) 0.000651 (0.00726) 0.00772 (0.0105). sit. n. al. er. io. 0.00180 (0.00401) 0.00724** (0.00346) -0.000563 (0.00512) 26,877 0.1511 Yes Yes Yes Yes Yes Yes Yes. y. Nat. Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. PPC Increasing. ‧. Rel(Dummy)* IPO. Mixed. ‧ 國. IPO. 政治大. 0.00402 (0.00305) -0.000114 (0.000234) -0.000553* (0.000318) 26,877 0.1512 Yes Yes Yes Yes Yes Yes Yes. 學. Rel(Dummy). PPC Increasing. Ch. engchi. 41. i n U. v. Decreasing 0.0209*** (0.00544) -0.000134 (0.000514) -0.000954 (0.000669). Decreasing 0.00724 (0.00703) 0.00108 (0.00606) 0.0189** (0.00931).

(42) Table 10: Market share and the Use of Performance-Pricing Covenant This table reports the marginal effects of multinomial logit regressions to evaluate the likelihoods of using mixed , interest- increasing ,or interest-decreasing PPC.*, **,*** indicate statistical significance at the 10%,5%,1% level. Mixed 0.0199*** (0.00593) -0.0504 (0.0838) -0.0631 (0.106). Rel(Dummy) Marketshare Rel(Dummy)* Marketshare Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. 立. PPC Increasing 0.000104 (0.00292) -0.0783 (0.0516) 0.0920 (0.0632) 26,877 0.1509 Yes Yes Yes Yes Yes Yes Yes. Decreasing 0.0179*** (0.00521) -0.0146 (0.0785) -0.0202 (0.119). 政治大. ‧ 國. 學 ‧. Table 11: Bank Ranking and the Use of Performance-Pricing Covenant This table reports the marginal effects of multinomial logit regressions to evaluate the likelihoods of using mixed , interest- increasing ,or interest-decreasing PPC.*, **,*** indicate statistical significance at the 10%,5%,1% level.. n. al. Rel(Dummy)* Bank Ranking. Ch. engchi. Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. 42. PPC Increasing 0.00494 (0.00484) 0.000540 (0.000423) -0.000627 (0.000616) 26,877 0.1595 Yes Yes Yes Yes Yes Yes Yes. er. io. Bank Ranking. sit. y. Nat. Rel(Dummy). Mixed 0.0151 (0.0117) 0.0141*** (0.000929) -0.00109 (0.00143). i n U. v. Decreasing 0.0414*** (0.00815) 0.00453*** (0.000790) -0.00474*** (0.00111).

(43) Appendix Table A.I Variable Definitions Variables. Definition. Loan Characteristics PPC(Accounting). Performance pricing provision based on an accounting measure.. PPC(Rating). Performance pricing provision based on the firm`s credit rating.. Mixed PPC. The pricing grid of PCC allows for both interest rate increases and interest rate decreases.. Increasing PPC. The pricing grid of PCC allows for interest rate increases only.. Decreasing PPC. The pricing grid of PCC allows for interest rate decreases only.. Facility Amount. 立. ‧ 國. Facility amount in million USD. 學. Facility Maturity. 政治大. Time to maturity in Months.. Initial all in drawn spread above LIBOR.. Term Loan. A dummy variable which equals one if the loan tranche is specified as. y. sit. io. Syndication. “ Term Loan”. A dummy variable which equals one if the loan is secured.. er. Nat. Secured. ‧. All-in-drawn Spread. A dummy variable which equals one if the distribution method of the. n. al. Ch. i n U. v. loan tranche is defined as “Syndication” according to LPC.. engchi. Sole Lender. A dummy variable which equals one if the. Purpose: General. A dummy variable which equals one if the loan purpose is specified as “ corporate purpose”. Purpose: Refinance. A dummy variable which equals one if the loan purpose is specified as “ debt repayment”. Purpose: Takeover. A dummy variable which equals one if the loan purpose is specified as “ takeover” or“ acquisition”. Purpose: Working Capital. A dummy variable which equals one if the loan purpose is specified as “ working capital”. Borrower Characteristics. 43.

(44) #Analysts. The number of analysts covering the borrower at the time of the loan origination.. IPO. A dummy variable which equals one if the borrower has initial public offering at the time of the loan origination.. MarketShare. Firm sales over the total sales in its own industry (we classify the industry by the first two number of SIC code).. Total Assets. Firm`s total assets in million USD.. Leverage. Long-term debt divided by total assets.. Market-to-Book. Market value of the firm divided by the book value of assets.. Tangibility. Net Property plant and equipment divided by total assets.. Profitability. EBITDA divided by total assets.. 立. Current Ratio. Current assets divided by current liabilities.. ‧ 國. 學. A rating score of the borrower which ranges from 0 to 22 according to the S&P long-term debt rating.. ‧. of AAA…C (or below) at the time of the loan issue.. y. Nat. Rated. A dummy variable which equals one if the borrower has an S&P rating. io. sit. Rating AAA…C(or below). A dummy variable which equals one if the borrower has an S&P rating. al. n. at the time of the loan issue.. Lender Characteristics Bank Ranking. Ch. engchi. er. Rating. 政治大. i n U. v. A ranking score of the arranger leader bank which ranges from 0 to 10 according to the league table which match the facility start year.. Relation Lending Proxies Rel(Dummy). A dummy variable which equals one if the firm borrowed from at least one of the lead lenders in the five years before the present loan.. Rel(Number). The number of loans from the same lead bank(s) over the total number of loans issued in the five years before the present loan.. Rel(Amount). The dollar value of loans from the same lead bank(s) over the total dollar value of loans issued in the five years before the present loan.. 44.

(45) Table A.II. Borrow Opacity and the Use of Performance-Pricing Covenant (using Rel(Number) as proxy for relationship strength) This table reports the marginal effects of multinomial logit regressions to evaluate the likelihoods of using mixed, interest- increasing, or interest-decreasing PPC. We use the other measures of relationship strength, Rel(Number) to evaluate the regressions .Marginal effect for each covariate are constructed as the difference in predicted probabilities for a particular outcome computed at their mean values holding all other covariates constant. All item are defined in the Appendix Table A.I. Standard errors are heteroskedasticity robust.*, **,*** indicate statistical significance at the 10%,5%,1% level. Ln(Total Assets) Rel(Number). 立. Ln(Total Assets). ‧ 國. y. sit. n. al. er. io. Rel(Number) Rating Rel(Number)* Rating. Decreasing 0.106*** (0.0337) -0.00763*** (0.00270) -0.0125** (0.00518). ‧. Nat. Rating. 學. Rel(Number)* Ln(Total Assets) Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. PPC Increasing 0.0210 (0.0160) -0.00281* (0.00155) -0.00194 (0.00244) 26,877 0.1508 Yes Yes Yes Yes Yes Yes Yes. 政治大 Mixed -0.0639 (0.0426) -0.00839*** (0.00304) 0.0143** (0.00604). Ch. engchi Mixed 0.0261* (0.0144) 0.003*** (0.001) 0.001 (0.002). Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. 45. i n U. v. PPC Increasing 0.00782 (0.00530) 0.0142*** (0.00130) 0.00104 (0.00154) 26,877 0.1538 Yes Yes Yes Yes Yes Yes Yes. Decreasing 0.0311*** (0.0116) -0.00756*** (0.00103) -0.000710 (0.00150).

(46) #Analysts Mixed 0.0466*** (0.0122) 0.000786* (0.000423) -0.00267* (0.00138). Rel(Number) #Analysts Rel(Number)* #Analysts. 立. Rel(Number)* IPO. n. al. er. io. sit. Nat. Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. PPC Increasing 0.0137** (0.00673) 0.00857*** (0.00317) -0.00977 (0.00922) 26,877 0.1511 Yes Yes Yes Yes Yes Yes Yes. ‧. IPO. Mixed 0.0518*** (0.0144) 0.00960 (0.00647) -0.0361* (0.0210). 學. Rel(Number). ‧ 國. IPO. Decreasing 0.0209** (0.0106) -0.000828* (0.000433) 0.00145 (0.00132). 政治大. y. Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. PPC Increasing 0.00682 (0.00523) -0.000460** (0.000217) 0.000309 (0.000585) 26,877 0.1509 Yes Yes Yes Yes Yes Yes Yes. Ch. engchi. 46. i n U. v. Decreasing -0.00357 (0.0136) 0.000824 (0.00544) 0.0569*** (0.0185).

(47) Table A.III. Borrow Opacity and the Use of Performance-Pricing Covenant (using Rel(Amount) as proxy for relationship strength) This table reports the marginal effects of multinomial logit regressions to evaluate the likelihoods of using mixed, interest- increasing ,or interest-decreasing PPC. We use the other measure of relationship strength, Rel(Amount) to evaluate the regressions . Marginal effect for each covariate are constructed as the difference in predicted probabilities for a particular outcome computed at their mean values holding all other covariates constant. All item are defined in the Appendix Table A.I. Standard errors are heteroskedasticity robust.*, **,*** indicate statistical significance at the 10%,5%,1% level. Ln(Total Assets) Rel(Amount). 立. Ln(Total Assets). ‧ 國. y. sit. n. al. er. io. Rel(Amount) Rating Rel(Amount)* Rating. Decreasing 0.0924*** (0.0300) -0.00754*** (0.00271) -0.0110** (0.00458). ‧. Nat. Rating. 學. Rel(Amount)* Ln(Total Assets) Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. PPC Increasing 0.0209 (0.0144) -0.00275* (0.00156) -0.00210 (0.00219) 26,877 0.1508 Yes Yes Yes Yes Yes Yes Yes. 政治大 Mixed -0.0180 (0.0376) -0.00782** (0.00306) 0.00728 (0.00528). Ch. engchi Mixed 0.0318** (0.0129) 0.0144*** (0.00131) -9.41e-05 (0.00137). Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. 47. i n U. v. PPC Increasing 0.00835* (0.00482) -0.00321*** (0.000598) -0.000373 (0.000734) 26,877 0.1538 Yes Yes Yes Yes Yes Yes Yes. Decreasing 0.0246** (0.0102) -0.00765*** (0.00104) -0.000222 (0.00131).

(48) #Analysts Mixed 0.0444*** (0.0108) 0.000925** (0.000440) -0.00272** (0.00121). Rel(Amount) #Analysts Rel(Amount)* #Analysts. 立. Rel(Amount)* IPO. n. al. er. io. sit. Nat. Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. PPC Increasing 0.0121* (0.00634) 0.00872*** (0.00327) -0.00875 (0.00842) 26,877 0.1511 Yes Yes Yes Yes Yes Yes Yes. ‧. IPO. Mixed 0.0462*** (0.0129) 0.00961 (0.00665) -0.0291 (0.0184). 學. Rel(Amount). ‧ 國. IPO. Decreasing 0.0203** (0.00920) -0.000704 (0.000456) 0.000512 (0.00122). 政治大. y. Obs. Adj R2 Industry Fixed Effect Time Fixed effect Credit Rating Fixed effect Loan Purpose Fixed effect Loan Type Fixed effect Loan characteristics Borrower characteristics. PPC Increasing 0.00575 (0.00473) -0.000479** (0.000231) 0.000321 (0.000562) 26,877 0.1509 Yes Yes Yes Yes Yes Yes Yes. Ch. engchi. 48. i n U. v. Decreasing -0.00446 (0.0119) -0.000221 (0.00557) 0.0493*** (0.0161).