How Has the Global Financial Crisis Affected Syndicated Loan Terms in Emerging Markets? Evidence from China

This paper examines the impact of the recent global financial crisis on the cost of debt capital (syndicated loans) in a leading emerging market, namely China, using difference-in-differences and GARCH approaches. Before the crisis China adopted banking reforms allowing entry of foreign banks and more domestic participation in the syndicated loan market. As a result, during the crisis the volume of syndicated loans grew steadily, in contrast to other countries. In addition, the amount of foreign syndicated loans decreased and average maturity increased compared to the precrisis period. Our findings provide useful information to policy makers to devise effective responses to financial crises.


Introduction
During the last two decades loans have dominated the corporate debt market in the developed economies (Drucker and Puri, 2007); in particular, the volume of syndicated loans has increased at a very rapid rate (Ferreira and Matos, 2012). A similar trend has been observed in emerging markets (Godlewski and Weill, 2008).
However, the recent financial crisis has led to a sharp decline (by 67 percent) in gross syndicated lending. Since in most cases the lead arrangers are international banks and financial institutions (Chui et al., 2010), the financial crisis that originated in the developed economies has also affected emerging markets (Dovern and Roye, 2014).
Given the borrower-lender and arranger-participant relationship in syndicated loans (Esty, 2001), financial shocks can be transmitted across countries through crossborder syndicated lending (Cetorelli, and Goldberg, 2011;De Haas and Van Horen, 2012;Ding et al., 2013). The increase in international infrastructure financing has resulted in foreign banks participating more in syndicated loans to reduce the risk of default from a single bad project (Brealey, Cooper and Habib, 1996;Ramamurti and Doh, 2004). Factors such as institutional weakness (Young et al., 2014), bank-level governance, country-level governance and previous profitability position (Beltratti and Stulz, 2009;Ivashina and Scharfstein, 2010;Berger and Bouwman, 2013) have all had a negative impact on syndicated loans. The performance of bank-dependent borrowers has also been affected (e.g. Chava and Purnanandam, 2011). A 'flight home effect' (Giannetti and Laeven, 2012) is another possible explanation for the decline of the syndicated loan market.
China being one of the biggest economies in the world (Berger et al., 2010), it is interesting to examine the impact of the crisis on its banking system (for some of its features see Hasan et al., 2009, andJia, 2009). In China, state controlled banks make most loan decisions expecting corporate borrowers to perform poorly (Bailey et al. 2011), and therefore syndicated loans are the most popular source of corporate finance (Pessarossi and Weill, 2013). The syndicates with lead arrangers from China have increased the loan amount even during the global financial crisis (Chui et al., 2010).
However, it remains to be seen how the crisis has affected the cost of such loans, and in particular how the syndicated loan terms with foreign arrangers compare to those with domestic arrangers.
Banks usually diversify their portfolio (Berger et al., 2010), avoid single name exposure, diversify their income sources by incorporating fee income as lead arranger and participate in syndicated loans to address the problems associated with origination capabilities (Godlewski and Weill, 2008). Borrowers also benefit from syndicated loans as larger amounts (Godlewski and Weill, 2008) can be arranged very quickly (Altunbas and Gadanecz, 2004); therefore, other debt markets have almost disappeared in China (Pessarossi and Weill, 2013).
The existing literature on syndicated loans documents agency conflicts arising from the lead arrangers having an information advantage over other participants (Strahan, 1999;Godlewski and Weill, 2008). In addition, there is a moral hazard problem as a higher number of participants leads to less monitoring by banks (Pennacchi, 1988). The agency problem persisted in China during the crisis owing to information asymmetries and poor accounting disclosure systems. Moreover, foreign participation forced the Chinese firms to disclose more information and consequently the extent of monitoring increased (Berger et al., 2010). In fact monitoring and opacity of firms are highly correlated with the loan contract terms. Therefore, unlike existing studies on China only focusing on the volume of syndicated loans during the crisis (Chui et al., 2010), in this paper we examine the impact of the financial crisis on both price and non-price contract terms of syndicated loans in China. To our knowledge, ours is the first study of this type.
Our empirical approach is twofold. Firstly, we apply a difference-indifferences method to data on 644 non-financial Chinese firms during the period 2000-2012. Secondly, we estimate a dynamic conditional correlation-GARCH model aggregating the data at the monthly syndicated loan level. We find that foreign lead arrangers tend to attract more lead arrangers in a single syndicated loan to overcome the financial difficulties in their home country, and offer a lower spread than the domestic lead arrangers to be competitive in the Chinese market. However, the amount of foreign syndicated loans has decreased during the crisis and their maturity has shortened. Our analysis highlights how the impact of the crisis was mitigated in China by agreeing appropriate syndicated loan contract terms with domestic arrangers, and more generally the strategy that should be followed in emerging markets during global financial crises.
The layout of the paper is the following. Section 2 reviews the relevant literature and develops the hypotheses to be tested. Section 3 gives details of the data and the methodology. Section 4 presents the empirical results. Section 5 concludes.

Changes in syndicated loans before and during the crisis
Demandable debt liabilities of banks give them an incentive advantage over other intermediaries. In the last two decades the debt market has witnessed an acceleration in the growth of syndicated loans (Focarelli et al., 2008;Dennis and Mullineaux, 2000) in both developed (Sufi, 2007) and emerging markets (Godlewski and Weill, 2008). A difference in bank capital before and during the crisis creates difficulties for bank-dependent borrowers (Demirguc-Kunt et al., 2013;Chava and Purnanandam, 2011). This is evident from the sharp fall in the volume of syndicated loans (Ivashina and Scharfstein, 2010). However, the spread remains quite high and the amount borrowed from banks associated to Lehman Brothers and other failed banks is quite low (Santos, 2011). Syndicated loans have generally declined, but China has been an exception (Okazaki, 2007).
Organising syndication, monitoring and due diligence are the responsibility of one or more lead arrangers (Dennis and Mullineaux 2000). For emerging markets, these are generally international banks (Chui et al., 2010) and the contraction in banks' foreign claims affects the syndicated loan market (De Haas and Van Horen, 2012). Ramamurti and Doh (2004) argue that lenders of a syndicated loan can earn attractive fees and interest rates. Moreover, the pro-market reforms in China allows them to 'securitise' their loans and pass the financial risk onto other investors. This type of securitisation activity has changed the role of lenders, and also stabilises the financial system as risk is allocated economy-wide. Another way of reducing risk is to involve local banks and investors to maintain a business network with political parties. These strategies possibly have enabled China to keep a constant growth of issuance of syndicated loans during the financial crisis. Chui et al. (2010) find ample supply of credit through local banks in China during that period. This evidence could partially explain the changes in the loan contract terms in China as the country underwent a series of banking sector reforms from 2002 to become a leading market-based economy (see Okazaki, 200;Ahlstrom et al., 2003;Young et al., 2011), and also joined the World Trade Organization in 2001. Murali and Banalieva (2015) showed that the relationship between market reforms and firms' performance is U-shaped because in the initial stages profitability decreases due to a monitoring vacuum but then, when the reforms are implemented, it rises, which attracts foreign banks. Foreign participants started their local currency business in China in December 2006. Their participation, in addition to other initiatives from the government (such as tax exemptions, strict disclosure rules, acceptance of international accounting rules, enhancing corporate governance norms etc.), has expanded the syndicated loan market in China. Moreover, in China most of the loans originate from state-owned and joint stock commercial banks (Okazaki, 2007). Resource endowment and organising capabilities together help Chinese firms aiming for outward internationalisation (Liang et. al., 2012). Globalisation and faster economic growth also create a greater need for domestic capital, with the presence of foreign banks increasing competition and improving the overall culture of the banking industry (Hasan et al., 2009). Domestic banks expand their activities through their networks (Bartoli et al., 2013), with borrowers preferring them because the government acts as a guarantor in most cases (Jia, 2009). Therefore, since 2008, whilst foreign banks started withdrawing from the Chinese market, the volume of syndicated loans arranged by domestic banks has stayed quite high, and on the whole the syndicate loan market has grown.
The volume of non-performing loans in China has been increasing, despite banking sector reforms and more monitoring of borrowers; this has led to poor profitability, an inadequate level of capital and contraction of the credit supply of banks (Albertazzi and Marchetti 2010). There is a high percentage of non-performing loans and no adequate risk pricing (Okazaki, 2007). Increased participation intensifies competition and the lack of proper accounting information disclosure by firms restricts the monitoring capabilities of banks. In the developed markets information asymmetry shapes syndicated loans (Preece and Mullineaux, 1996;Jones et al. 2005) because of less transparent borrowers (Dennis and Mullineaux, 2000), their reputation (Lee and Mullineaux, 2004) and their relation with lead arrangers. In addition, there is a correlation between borrower opaqueness and concentration in credit syndicates, which leads to the problem of moral hazard in communication (Sufi, 2007).
According to Albertazzi and Marchetti (2010), there is a possible 'flight to quality' effect. In other words, larger foreign lead arrangers might have reallocated loans away from risky and opaque Chinese borrowers during the crisis. In such a situation the alternative is 'zombie lending' (Bruche and Llobet, 2014), with some insolvent Chinese banks exploiting credit demand in the domestic market by trying to avoid credit losses and arranging syndicated loans for the risky borrowers with flexible loan contract terms.
The literature provides evidence of the impact of syndication on loan spread, maturity and loan amounts in other countries (Focarelli et al., 2008), and also of changes during the crisis (e.g. Strahan, 1999;Ivashina and Scharfstein, 2010;Santos, 2011). Chui et al. (2010) show that the volume of syndicated loans increased during that period, but do not examine the possible effects on loan amounts, spread, maturity and the number of lead arrangers of syndicated loans.

Hypothesis development
The cost of bank credit remained quite high in the US during the crisis compared to the pre-crisis period (Santos, 2011). Shocks were transmitted to emerging markets through different channels such as cross-border lending, direct foreign bank participation etc. Information asymmetry has kept rising in countries such as China, and therefore foreign arrangers have been finding it difficult to assess the credit worthiness of borrowers.
Before the crisis, domestic banks were the lead arrangers for most syndicated loans in China. Usually, firms prefer to establish relations with well-capitalised banks (Berger et al., 2008); consequently, banks with a higher capital ratio tend to have more information about borrowers and charge a spread premium. In a hierarchical banking structure, it becomes difficult to produce and transmit soft information (Stein, 2002). As a result, information asymmetry between lead domestic arrangers and domestic participants increases. Chinese small and medium industries, in particular, may suffer from a credit crunch (see . Following Rajan (1992), one can argue that, because of information asymmetry and less transparency, the interest rate on syndicated loans is higher during a crisis period. Recent studies also show that firms have paid more to obtain guaranteed access to liquidity during the global financial crisis (Santos, 2011;Bord and Santos, 2014). Jiangli et al. (2008) concluded that lending relationships mattered during the Asian crisis. There is evidence of a strong relationship between domestic banks and firms before the crisis that also continued during the crisis (Bartoli et al., 2013;Chodorow-Reich, 2014). Therefore, domestic arrangers are always in a more advantageous position than foreign arrangers because of their past relationships with firms. On the other hand, during the crisis foreign arrangers, mainly from the developed countries, have viewed China as a more financially stable market than their own economically imbalanced domestic ones.
Thus, in order to offset the losses arising from non-performing loans in their home countries, they have offered lower interest to credit-worthy Chinese borrowers. The above discussion suggests the following hypotheses to test:

H1: During the financial crisis in China the interest rate increased less for foreign than for domestic syndicated loans.
Syndicated loans contribute towards financial development and stability in emerging markets (Godlewski and Weill, 2008). During the crisis they fell in developed markets where lead arrangers were severely hit (Santos, 2011). However, in China, where the financial sector had been growing steadily and had been strengthened by various reforms (Okazaki, 2007), the supply of credit remained steady during the crisis.
Because of the sovereign debt crisis and the collapse of several financial institutions in 2008-2009, with the consequent crisis in confidence for the syndicated loan arrangers (mainly from developed markets), lending fell during the financial crisis (Popov and Van Horen, 2013). Foreign arrangers withdrew from the Chinese syndicated loan market, and also made more use of securitisation and reduced their loan supply (Bonaccorsi di Patti and Sette 2012). Since the liquidity position of the domestic lead arrangers did not change during the crisis, the total amount of syndicated loans was affected.

H2: During the financial crisis in China foreign syndicated loans decreased relative to domestic syndicated loans.
Usually banks prefer to lend for longer maturities to reduce moral hazard (Coleman et al., 2006). In China, the banking system is almost 100 percent government-owned (Dobson and Kashyap, 2006), and owing to lack of monitoring by state-owned banks and a weak institutional framework, the country is far behind in terms of capital allocation efficiency. Moreover, if there are many lenders the necessary monitoring decreases because the lead arrangers may exploit their informational advantage to obtain an information rent (Bruche and Llobet, 2014). Then, participants cannot understand the willingness of lead arrangers to monitor borrowers, which leads to non-performing loans. In the case of syndicated loans, when there is a need for diligence and monitoring by borrowing firms, the lead arrangers usually prefer concentrated loans, i.e. holding a large fraction of the loan (Sufi, 2007); however, if there is less information asymmetry, they tend to have a preference to hold smaller amounts (Focarelli et al., 2008). Consequently, the syndicate requires more arrangers and participants. Many banks take this opportunity to diversify their risks by becoming a member of the syndicate.
Loan maturity reflects the borrower risk (Nandy and Lodh, 2012), which is also associated with the loan spread. According to the credit quality hypothesis, lenders prefer a short maturity period for any loan as it gives them the opportunity to assess regularly the credit position of firms (Diamond, 2004). On the other hand, the trade-off hypothesis states that the loan spread increases with the maturity period (Gottesman and Roberts, 2004). A recent study (Alexandre et al., 2014) provides evidence that firms managed to obtain longer maturities during the crisis when they had a stronger lending relationship before the crisis. Therefore, we hypothesise that in China more domestic banks' participation and poor accounting disclosure allow the arrangers to assess the credit worthiness of firms, and as a result information asymmetry between the syndicated lenders and the borrowers is significant. In the crisis period the borrowers go through a tight screening process by foreign banks when these enter the Chinese market. This reduces the firms' opacity to some extent.
But due to the contraction in the operation of foreign banks in the Chinese credit market during the crisis, information asymmetry widened. Therefore, we test the following hypothesis: H3: During the financial crisis in China loan maturity remained longer for foreign syndicated loans compared to domestic syndicated loans.
In a syndicated loan the lead arrangers take the responsibility of originating it and share it with other financial institutions (Ivashina and Scharfstein, 2010a). They usually keep one third of the syndicate loan and sell the rest to other syndicate investors. This may create information asymmetry between the lead arranger and the other participants, with the former possessing more information. But if the participants are not satisfied with the information about the borrowers, then the lead arranger(s) might want to share the risk with other lead arrangers both in the domestic and foreign markets. In such a situation, they may hold less than one third of the syndicated loan. On the other hand, lenders are always more inclined to giving loans to firms with high profitability (Berger and Udell, 1990;Saidenberg and Strahan, 1999): the lead arrangers may charge less interest and may arrange loans with a longer maturity to attract more borrowers for the syndicated loans.
Banks' lending portfolios carry a considerable amount of country-specific risk (Fang and Lelyveld, 2014). During the financial crisis, the capital position of the foreign arrangers in their home country remained quite weak. They were attracted to China because of its financial stability. A single lead arranger cannot provide the required syndicated loan amount due to capital inadequacy and therefore might involve other lead arrangers from the domestic and foreign markets. This leads to formulating the following hypothesis:

H4: During the financial crisis in China the number of lead arrangers increased for
foreign syndicated loans compared to domestic syndicated loans.

Sample and Variable Description
To test the effects of the global financial crisis on both price and non-price terms of syndicated loans we use loan information for China from the ThomsonOne Deal database. We also match a few companies with the Worldscope and Bloomberg database to increase the number of observations. We start with all borrowers in the database and then identify the non-financial firms. In China in our sample period, Any loan contract consists of both price and non-price terms (Melnik & Plaut, 1986); Strahan (1999) argues that firms pay a higher interest rate when non-price terms become more restrictive. Therefore, we consider both price and non-price terms of syndicated loans as follows. Loan Spread is an all-in-drawn spread: it displays all spreads at multiple levels based on the margin in basis points and includes the base rate spread and facility, upfront, utilisation or fronting fee in the database. Loan Amount in the ThomsonOne database is the full loan package amount for the target market for all tranches and is reported in millions. Loan Maturity is another important loan contract term and is measured in years in our study. It is calculated as the difference between the maturity date and the issue date of the loans, where the former is the latest possible maturity date and, if the loan is extendable, the extra years are added to obtain the final maturity, and the issue date for syndicated loans is the announcement date of the transaction. The last loan term considered in the model is the Lead Arranger. The mandated arrangers are the lead agent banks named in a mandate letter for a particular loan. The mandated arranger title has been in use since January 2000. In Asia, mandated arrangers are the named lead agents in a mandate letter for a particular syndicate and may not be restricted to the Administration, Syndication or Documentation Agents.
Banks assess the creditworthiness of firms before deciding on loan contract terms and focus on several firm-level factors. Therefore, following the literature (e.g. Santos, 2011;Strahan, 1999) we control for firm characteristics. Big firms are assumed to have a lower default probability; therefore we include Firm Size, which is defined as the log of total assets. These may need more loans with long maturities for their activities but the spread could be lower than for smaller firms because of the lower default probability. Profitability is measured by the return on assets (ROA).
Higher returns for firms implies less risk from the bank's perspective. More profitable firms may require more loans but may pay less interest as they are considered to be less risky. Older firms are more established and are also viewed as less risky. We capture this by including Firm Age, which is defined as the log of age. Such firms may obtain more loans with long-term maturity and also pay less interest. Financial Leverage is long-term debt over total equity. There is a higher default probability if the firm borrowing is highly debt-dependent, especially during a crisis period. These borrowers may get more loans with a shorter maturity. However, the spread may be higher.
We also include the PE Ratio, which is defined as the current price divided by earnings. High growth firms may get more loans with a shorter maturity and a bigger spread. Another variable is EBIT, i.e. earnings before interest and tax. Higher earnings suggest a lower default probability. The lead arrangers of a syndicated loan can influence the loan terms with their contribution to the loan (Jones et al., 2005), therefore we control for the percentage of loans (principle amount) of lead arrangers.
The variable Share of lead arranger is also included in the model. Most banks check credit ratings. We use Moody's Credit Rating. According to their generic rating, firms have minimal default risk if they belong to Aaa and the risk is higher for category B and C. Moody's appends numerical modifiers 1, 2, and 3 to each generic rating classification from Aa through Caa. The modifier 1 indicates that the obligation ranks at the higher end of its generic rating category; 2 indicates a mid-range ranking, and 3 a ranking at the lower end of that generic rating category. Different industries may be associated with different levels of risk, therefore we also include Industry. Table 1 shows the summary statistics for three categories: domestic syndicated loans (all lead arrangers in a syndicated loan are from China), foreign syndicated loans (at least one lead arranger of a syndicated loan is from a foreign country) and the full sample. The maximum loan amount is 39,000 (US$, mil), which is for a domestic syndicated loan group. The maximum foreign syndicated loan is 6,000 (US$, mil); it is arranged by a maximum of 23 lead arrangers, whereas a maximum 8 lead arrangers are involved in a domestic syndicated loan. The average all-in-drawn spread for domestic syndicated loans is higher than for foreign syndicated loans. However, the average maturity period (years) is higher for the former than for the latter.

Data Summary
Insert Table 1

Empirical results
We proceed in two steps to test the hypotheses of interest: first, we analyse an unbalanced panel; second, we aggregate the data to create time series at the loanmonth level and estimate a dynamic conditional correlation-GARCH model to investigate the co-movement of loan contract terms in China during the financial crisis.

Model
To examine the impact of the global financial crisis on the price and non-price terms of syndicated loan terms in China we estimate the following model: We estimate the model with clustering at the industry level using the difference-in-differences (diff-in-diff) method. The underlying assumption is that the time trend on the treatment and control groups is the same as in the absence of treatment, which is difficult to verify. Therefore, as a robustness check we use pretreatment data to see if the trends are indeed the same.

Results for the difference-in-differences model
Column 1 of Table 2-5 shows the estimation results for the model with diff-in-diff without the control variables. The 2 is small for all four models (see Table 2 and 5 in particular). The results when including all the relevant firm-level control variables and the syndicated loan terms variables step-wise are reported in the other columns in Tables 2-5.

Insert Table 2 about here
In Table 2  Insert Table 3 about here Table 3 shows that the coefficient of the interaction term is statistically significant and negative in all cases. In particular, there was a 19 percent decrease (significant at the 5 percent level) of foreign syndicated loans relative to the pre-crisis period (see Column 4 of Table 3). This very strongly supports our Hypothesis 2 that during the crisis foreign syndicated loans decreased. In the follow-up period we do not find any significant changes (except for the year 2011 in Column 3). We interpret this result as suggesting that the total amount of loans remained unchanged owing to the more active involvement of domestic banks in China as the focus of foreign banks shifted to their domestic markets. This is consistent with previous findings (Chui et al., 2010).

Insert Table 4 about here
In Table 4 the coefficient of diff-in-diff is positive and statistically different from zero (at the 1 percent and 5 percent level), which supports Hypothesis 3. This implies that during the crisis the foreign syndicate loan providers imposed longer maturities and that this continued in the follow-up period. The reason is that the Chinese market was stable during the crisis and foreign banks experienced less risk compared to other countries including their own. Moreover, due to financial market and banking reforms in China, the improved credit scoring reduced uncertainty about borrowers (see .
Insert Table 5

about here
The statistically significant and positive coefficient in Table 5 supports Hypothesis 4, i.e. that during the financial crisis the number of lead arrangers increases for the foreign syndicated loans to diversify risk and to compensate capital inadequacy in their home country. It also appears that in the follow-up period (in 2012) the number of lead arrangers continued to increase.
Overall, we find empirical support for the hypotheses formulated above.
During the financial crisis foreign syndicated loans decreased despite a higher number of lead arrangers and longer maturities. However, to cope with the imbalances in the global economy and the credit market crunch, the foreign syndicated loan providers kept lower spreads by diversifying their risk through a number of lead arrangers.

Time-series approach (DCC-GARCH model)
In this section, we explore the effects of the financial crisis on the aggregate loan spread-maturity, loan spread-amount and loan amount-maturity relationships. A simple correlation analysis would not be sufficient for this purpose; we use instead Engle's (2002) time-varying dynamic conditional correlation (DCC)-GARCH model.
To begin with, we carry out some diagnostic tests. The results are presented in Table 6. The Ljung-Box Q statistic indicates serial correlation in all the variables. The Insert Table 6 about here In order to investigate the impact of financial shocks on the co-movement between loan amount, spread and maturity, we follow the following DCC-GARCH modelling approach (Engle 2002  The vector has the following conditional variance-covariance matrix: where = (�ℎ ) is a 2 × 2 diagonal matrix of time-varying standard deviations from univariate GARCH models, and ≡ [ ] for i, j=1 and 2, is a correlation matrix containing conditional coefficients. Note that varies over time.
The standard deviations in D t follow the univariate GARCH (P,Q) process as follows: The DCC-GARCH model is estimated in two steps (see Engle, 2002); the likelihood function of the DCC estimator is as follows: It has two components: volatility (D t ) and correlation (R t ).
The DCC (M, N) component can be written as: � is the time-invariant variance-covariance obtained from estimating equation (3), and * is a 2 × 2 diagonal matrix containing the square root of the diagonal elements of . In particular, the key element of interest is = , � , , , where j=1, 2. R t is a 2 × 2 matrix comprising the conditional correlations. Table 7 presents the results for the DCC (1,1) model. All the parameters of the univariate GARCH model ( , 1 , 2 , ) are more appropriate for syndicated loan amount, spread and maturity when one (or more) of the mandated arrangers is a foreign bank than when the syndicated loan is purely from domestic banks. The significance of the GARCH parameters indicates the presence of time variation and dependence of the variance. A sum of a m and b n close to 1 indicates high persistence in the conditional variance (see Equation (5)).

Estimation results for the DCC-GARCH model
Insert Table 7

Robustness Tests
Next we check the sensitivity of our main results on the existence of flight to quality (or banks' response to heterogeneity of borrowers) during the financial crisis, in particular after the Lehman Brothers' collapse. The results of the robustness tests are reported in Table 8. Insert Table 8

about here
We then select firms on the basis of their credit rating. We exclude firms with rating Aa2, A2 and A3 (top rating grades with A's in our sample). These are the firms with the lowest default rate, i.e. the lowest risk. We re-estimate the model in this case (with foreign syndicated loans only offered to the less risky borrowers in the Chinese market to reduce their risk of default) and find again that the main results are robust (see Panel B of Table 8). Therefore, we conclude that during the financial crisis the foreign syndicated loan arrangers targeted the entire Chinese market, irrespective of the borrowers' risk. This is also evident from our finding that foreign syndicated loans have flexible loan contract terms, such as lower spread and longer maturity.

Conclusions
The importance of syndicated loans in the corporate debt market has been highlighted both in the theoretical and empirical literature. During the global financial crisis their volume was squeezed in most countries (Ivashina and Scharfstein, 2010) and banks from the developed countries quit the emerging markets (Chava and Purnanandam, 2011). In China, however, the reforms implemented before the crisis enabled domestic banks and financial institutions to play a bigger role in the syndicated loan market. As a result the volume of syndicated loans in China grew steadily during the crisis (Chui et al., 2010). The present paper examines not only lending volumes, but also the cost of debt, and more generally both the price and non-price terms of syndicated loans. It emerges that foreign syndicated loans offered lower interest rates to attract more Chinese borrowers. Moreover, the loan amount was lower for longer maturities in the case of foreign syndicated loans in China. Domestic syndicated loan arrangers tended to offer better non-price than the price-terms. This has resulted in a constant credit supply in China during the global financial crisis.
Our findings contribute to the literature on cross-border syndicated loans and on syndicated loans in emerging economies during financial crises. Information on banks' lending volumes in emerging markets is not sufficient to design policy responses to financial crises, the amount and cost of debt should also be examined.
Our study of the Chinese case suggests that the impact of the financial crisis in emerging markets was mitigated by appropriate syndicated loan contract terms. It provides important information to policy makers of other emerging countries aiming to design an effective debt market strategy to tackle future global crises, since bank credit has a significant impact not only on firms' activities but also on the macroeconomy (Campello et al., 2010). A follow-up study will investigate such effects in the post-crisis period. Notes: domestic refers to the domestic syndicated loan defined as those loans when all lead arrangers of a syndicated loan are from China. Foreign refers to the foreign syndicated loan defined as those loans when at least one lead arranger of a syndicated loan is from foreign country. 0.08 0.20 0.67 0.63 Notes: Robust t-statistics are in parentheses. The standard errors are obtained using clustering on industry as explained in the methodology. *, ** and *** represent coefficients significant at the 5%, 1% and 0.1% respectively. Models are estimated with firm fixed effect. In all models, firm level controls, such as firm size, financial leverage, profitability and price-earnings ratio (in one year lag) and firm age, Moody's credit rating are included. Robust t-statistics are in parentheses. The standard errors are obtained using clustering on industry as explained in the methodology. *, ** and *** represent coefficients significant at the 5%, 1% and 0.1% respectively. Models are estimated with firm fixed effect. In all models, firm level controls, such as firm size, financial leverage, profitability and price-earnings ratio (in one year lag) and firm age, Moody's credit rating are included.    (15) and for domestic syndicated loans ARCH (10) LM test are done, because of differences in sample size. * Denotes significance at 5% level. a The estimated structural breaks (Month, Year) for variables are as follows: foreign syndicated loan amount (February, 2011), spread (September, 2006, maturity (August, 2006)