Firms_as_surrogate_intermediaries

Firms as Surrogate Intermediaries:

Evidence from Emerging Economies?

Hyun Song Shin?Laura Yi Zhao?

December2013

Abstract

A?rm can?nance investment either by borrowing or by drawing on cash balances, so that?nancial asset and liability changes tend to have opposite signs.In contrast,

?nancial intermediaries borrow in order to lend,so that?nancial asset and liability

changes have the same https://www.sodocs.net/doc/364569925.html,rge non-?nancial?rms in China and India behave like

intermediaries rather than textbook non-?nancial?rms.We explore the role of non-

?nancial?rms in the shadow banking system.The evidence from China and India is

in contrast to US non-?nancial?rms,which conform to the textbook predictions.

?Preliminary.This study forms part of the background research for the Asian Development Bank technical assistance program on“Financial Regulatory Reform in Asia”.

?Corresponding author:Bendheim Center for Finance,Princeton University,26Prospect Avenue,Prince-ton,NJ08540,USA;hsshin@https://www.sodocs.net/doc/364569925.html,

?Asian Development Bank;yzhao.consultant@https://www.sodocs.net/doc/364569925.html,

1Introduction

The market stresses faced by many emerging economies in the face of tighter global monetary conditions in2013have focused renewed attention on the transmission of?nancial conditions across borders.One conceptual challenge is to reconcile the small net external debt positions of many emerging economies with the apparently disproportionate impact of tighter global monetary conditions on their currencies and?nancial markets.Indeed,some commentators have wondered aloud why emerging economies with low net external debt positions are experiencing such severe stresses.1

The purpose of our paper is to o?er one missing piece in the puzzle,highlighting the role of non-?nancial corporations as surrogate?nancial intermediaries that operate across borders.When corporate activity straddles the border,measuring exposures at the border itself may not capture the strains on corporate balance sheets.For instance,if the London subsidiary of the company has taken on US dollar debt but the company is holding domestic currency?nancial assets at its headquarters,then the company as a whole faces a currency mismatch and will be a?ected by currency movements,even if no cross-border exposures are registered in the o?cial net external debt statistics.Nevertheless,the?rm’s fortunes (and hence its actions)will be sensitive to currency movements.In the case of?rms that straddle borders,it may be more illuminating to look at the consolidated balance sheet that motivates corporate treasurers,rather than the balance of payments statistics that are organized according to residence.

One aspect of?rms’access to international capital markets is the o?shore issuance of debt securities sold to international investors.If the debt securities issued o?shore are in foreign currency,o?shore issuance would mirror currency mismatches on the consolidated balance sheet.Hence,o?shore issuance goes beyond just a measurement issue on the size of the company’s debt and instead addresses the fundamental issue of how?rms will fare when 1For instance,Krugman(2013)“Asian Vulnerability,Then and Now”

https://www.sodocs.net/doc/364569925.html,/2013/08/29/asian-vulnerability-then-and-now/

1

B i l l i o n U S d o l l a r s

B i l l i o n U S d o l l a r s

Figure 1:China (left)and India (right):International debt securities outstanding for non-?nancial corporates by nationality and by residence (Source:BIS Debt Securities Statistics,Table 11D and 12D)

global ?nancial conditions and exchange rates change.

Figure 1shows BIS statistics on the amounts outstanding of international debt securities issued by non-?nancial corporate borrowers in China (left)and India (right)by residence of the borrower (blue)and the nationality of the borrower (red).The di?erence between the red and blue series re?ects the o?shore issuance of corporate debt securities.We see from Figure 1that o?shore issuance activity was small until the 2008crisis,but subsequently grew strongly.The period after 2010has seen a particularly steep increase so that by 2013,the o?shore amounts outstanding are equal in size to the onshore issuance outstanding.McCauley,Upper and Villar (2013)describe the recent trend of o?shore issuance of corporate debt securities.2

Our paper examines the role of the ?rm as a surrogate ?nancial intermediary that trans-mits ?nancial conditions across borders.The hallmark of banks and other ?nancial inter-mediaries is that they borrow in order to lend.As such,when their ?nancial assets increase through new lending or purchases of securities,their ?nancial liabilities,such as deposits,also increase.In this way,a distinctive feature of ?nancial intermediaries is that the change in their ?nancial assets has the same sign as the change in their ?nancial liabilities.In contrast,textbook non-?nancial ?rms behave in a very di?erent way.When a non-Agust′?n Villar “Emerging market debt securities issuance in o?shore centres”BIS Quarterly Review,September 2013,Box 2,pp 23-24.https://www.sodocs.net/doc/364569925.html,/publ/qtrpdf/r qt1309b.pdf

2

?nancial?rm undertakes an investment,it can?nance it either by drawing on its existing ?nancial resources or by external borrowing,or a combination of both.A prediction of the “pecking order”theory of corporate?nance(Myers(1984))is that the?rm will draw on internal funds?rst as the cheapest form of?nancing,and only tap outside funding when internal funds are inadequate.A prediction from such behavior would be that changes in ?nancial assets and changes in?nancial liabilities will have opposite signs,capturing those ?rms that raise outside funding while drawing down internal funds.

We show that non-?nancial?rms in emerging economies behave like?nancial intermedi-aries in that co-movements in?nancial assets and?nancial liabilities have a positive sign. This is true both in the cross-section,as well as in the time series.In other words,?rms that borrow more also hold more cash,and?rms that increase their borrowing also increase their cash holding.To the extent that?rms’cash holdings are claims on the domestic bank-ing sector,the?rms would be performing a?nancial intermediation role by making funding available indirectly to other domestic borrowers.

Our paper has a close parallel with Hattori,Shin and Takahashi(2009),who describe the role of non-?nancial corporates as surrogate intermediaries in Japan in the1980s.Hattori et al.(2009)show how the?nancial liberalization of the1980s enabled large manufacturing ?rms in Japan to gain access to funding by issuing securities,especially from international investors who sought yen exposure.As new funding sources opened up,?rms recycled yen funding through the banking system in the form of bank time deposits.Through this channel, the?nancial assets of non-?nancial?rms increased in step with their?nancial liabilities in the1980s.Banks in Japan su?ered a reversal of roles in which corporate borrowers became corporate depositors,and banks were pushed to seek borrowers in riskier sectors such as in commercial real estate.The parallel between Japan in the1980s and the emerging economies in2013lies in the role of non-?nancial corporates as surrogate intermediaries.

The evidence in our paper comes from a large panel of non-?nancial?rms emerging economies in the Compustat Global database in which we examine both the cross-section

3

patterns in corporate balance sheets,as well as the growth of individual?rms’?nancial assets and liabilities with?rm?xed e?ects.The evidence from the major emerging economies is especially noteworthy given its contrast to US non-?nancial?https://www.sodocs.net/doc/364569925.html,?rms are seen to conform to the textbook prescription for corporate?nancing choices in which?nancial assets and liabilities move in the opposite directions,consistent with the pecking order theory of ?nancing(Opler et al.(1999)).

Our paper has a point of contact with the many studies that have explored the trends and implications of corporate cash holdings.Traditional studies focus on?rm value,merges and acquisitions,and dividend issuance.Harford(1999)shows that cash-rich?rms are more likely to attempt acquisition and their mergers tend to be followed by a decline in operating performance.Lie(2000)?nds that a large increase of dividends mitigates the agency problem associated with excess cash-holdings.Denis and Sibilkov(2010)show that as due to costly external?nancing,greater cash holdings increase the value of constrained?rms.Our paper contributes to this literature by exploring the implications of the non-?nancial?rms’cash holdings for the liquidity in the banking system.

Given the importance of corporate liquidity,many works explore its determination.Opler et al.(1999)and Ferreira and Vilela(2004)?nd supportive evidence for a static trade-o?theory using data from the United States and European countries respectively.Bates et al. (2009)argues that precautionary motives explain the rise of US industrial?rms’cash-to-asset ratio.The role of corporate governance is also explored.As an example,Dittmar et al.(2003)?nds that the agency problem is an important determinant of corporate cash holdings,and that?rms in countries with poor shareholder rights hold twice as much cash as?rms in countries with good shareholder protection.Other factors are also found to a?ect corporate liquidity:tax cost for multinational companies to repatriate foreign income(Foley et al.(2007)),the predation risk(Haushalter et al.(2007)),the diversi?cation of investment opportunities(Duchin(2010)),and the incentive to hedge cash?ow shocks during bad times (Lins et al.(2010)).Adding to this line of literature,our paper explores a new perspective

4

to understand non-?nancial?rms’cash holdings through their role as surrogate?nancial intermediary.

Our paper is also related to the studies on the?nancing decisions of?rms,in particular the use of debt?nancing.This line of literature focuses on two competing theories:the trade-o?theory and the pecking order theory.The empirical evidence is mixed.Shyam-Sunder and Myers(1999)argues that the basic pecking order model has more explanatory power than the static trade-o?theory in explaining the?nancing patterns of public and mature?rms in the United States.On the other hand,Frank and Goyal(2003)and Fama and French(2005)?nd pervasive evidence contradicting the pecking order https://www.sodocs.net/doc/364569925.html,ter on, Leary and Roberts(2010)show that the pecking order theory performs better in explaining ?rms’?nancing decisions only when factors typically attributed to other theories are simul-taneously accounted for.These two theories focused mainly on the traditional explanations for corporate use of debt,for example taxes,bankruptcy cost,transaction costs,adverse selection and agency con?icts.Our paper,by investigating the surrogate?nancial interme-diary roles of the non-?nancial?rms,suggests that the non-?nancial?rms in China borrows in order to invest,especially in the form of deposit and other short-term investments.

Before documenting the key facts,we delve deeper into the institutions that underpin the empirical results.In particular,we explore how the availability of?nancing from inter-national capital markets induces large non-?nancial?rms to engage in?nancial transactions in the shadow banking system that have the tell-tale attributes of?nancial intermediation. The institutional backdrop of the shadow banking system in China is a tightly regulated formal banking sector,which sits alongside a highly open and trade-dependent economy. Even if capital account transactions through banks can be tightly regulated,the current account transactions of thousands of?rms generated in the course of international trade will be much harder to monitor and regulate.

By its nature,shedding light on the shadow banking system and?rms’roles in the system presents formidable challenges in measurement and for data availability.However,

5

Figure2:Non-?nancial?rms as intermediary.In this diagram,?rms with access to international capital markets act as an intermediary for outside funding when the banking sector has restricted access to international capital markets.

the advantage of our approach is that,however the?rms managed to change their?nancial claims and liabilities,the consequences of their actions will be captured in the snapshot of the consolidated balance sheet at the reporting period.As such,for the purpose of gauging the scale of intermediation performed by non-?nancial?rms,we can simply read o?the ?nancial assets and liabilities,without having to capture in detail all the speci?c practices that the?rms engage in reaching their?nal position.

When the availability of external?nancing from international capital markets varies with global liquidity conditions,a prediction of our approach is that the surrogate?nancial intermediation activity of non-?nancial?rms in emerging economies will re?ect(at least in part)the ebb and?ow of global liquidity conditions themselves.Consistent with this hypothesis,we?nd that the extent of intermediation activity of non-?nancial?rms co-moves strongly with indicators of credit availability at the global level.We contrast the evidence from emerging economy?rms and?rms from the United States.While US?rms conform closely to the textbook model,?rms from emerging economies exhibit the distinctive positive co-movement of?nancial assets and liabilities.We conclude with some broader lessons for the operation of the?nancial system in a tightly regulated economy.

6

Chinese corporate Hong Kong bank

Figure3:O?shore borrowing by a non-?nancial corporate in foreign currency

2Background

An economy with an open?nancial sector and convertible capital account will be sensitive to global?nancial conditions,but the sensitivity to external?nancial conditions also applies to economies that are tightly regulated and whose capital accounts are closed.Just as water will?nd cracks to trickle through a rock,so will international capital?nd ways into an open economy when it has a large volume of transactions associated with trade.This is so even when the?nancial sector is tightly regulated and external borrowing is restricted by regulations that govern capital in?ows.The role of non-?nancial?rms is crucial in this respect as the channel through which capital in?ows take place.

Figure2depicts an economy with a banking sector that has restricted access to wholesale funding in international capital markets,but where a subset of?rms have access both to the domestic?nancial system as well as international capital markets through trade?nancing or the operation of overseas o?ces.Although non-?nancial?rms are subject to regulations in their use of international capital markets,the sheer number of such?rms as well as the complexity of their transactions make them much harder to regulate than the banks.

As well as the corporate bond market,global banks provide another channel to the international capital markets.

Figure3is a schematic illustration of the activities of a non-?nancial?rm from China

7

-400

-300-200-1000100200300400Jan-2003

Oct-2003

Jul-2004

Apr-2005

Jan-2006

Oct-2006

Jul-2007

Apr-2008

Jan-2009

Oct-2009

Jul-2010

Apr-2011

Jan-2012

B i l l i o n H K d o l l a r s

Claims on non-bank customers in China (F.C.)

Liabilities to non-bank customers in China (F.C.)Figure 4:Hong Kong banks’claims and liabilities to non-bank customers in China in currencies other than Hong Kong dollars (Source:Hong Kong Monetary Authority)

with operations in Hong Kong,who borrows in US dollars from an international bank in Hong Kong and posts Renminbi deposits as collateral.The transaction would be akin to a currency swap,except that the settlement price is not chosen at the outset.The transactions instead resemble the operation of the old London Eurodollar market in the 1960s and 70s.For the Chinese corporate,the purpose of having US dollar liabilities and holding the proceeds in Renminbi may be to hedge their export receivables,or simply to speculate on Renminbi appreciation.

Figure 4provides some aggregate evidence for the transactions depicted in Figure 3.Figure 4plots the claims and liabilities of Hong Kong banks in foreign currency to customers in China.Foreign currency,in this case,would be US dollars mainly for the assets and Renminbi mainly for the liabilities.Both have risen dramatically in recent years,re?ecting the rapidly increasing US dollar funding of non-?nancial corporates from China.

As well as channeling capital ?ows into China,non-?nancial ?rms play a more direct role as a ?nancial intermediary through the institution of “entrusted loans”.

Entrusted

8

Figure5:Non-?nancial?rms as intermediary through“entrusted loans”.This diagram depicts the operation of“entrusted loans”where non-?nancial?rms lend to other non-?nancial?rms with limited access to bank lending.The bank acts as delegated manager of the loan contract.

loans are loans granted by one?rm to another?rm directly.However,a commercial bank administers the loan as a delegated manager.Figure5illustrates the operation of an entrusted loan,where a large?rm with access to bank loans recycles the loan by granting an entrusted loan to another?rm-typically a smaller?rm with restricted access to bank lending,or a property-related?rm.The commercial bank administers the entrusted loan, and the entrusted loan stays o?the bank’s balance sheet,and hence does not count against lending limits set for the commercial bank by the bank regulators.From Figure5,we see that the lending?rm in the entrusted loan relationship behaves like a?nancial intermediary, simultaneously borrowing and lending.Increased incidence of such intermediation activity will be captured in a snapshot of the lending?rm’s balance sheet as the simultaneous increase in both?nancial assets and?nancial liabilities.

Quantitatively,the intermediation conducted through entrusted loans is large relative to the lending through the formal banking sector.Figure6plots the quarterly?ow of entrusted loans and domestic currency bank loans in China,as published by the People’s Bank of China statistics on all systems?nancing(also called total social?nancing).We see that the?ow of entrusted loans have increased in recent quarters,reaching25%to30%of

9

0.00.51.01.52.02.53.0

T r i l l i o n R M B

0%

5%

10%

15%

20%

25%

30%

Figure 6:Quarterly ?ow of entrusted loans and Renminbi bank loans (Source:People’s Bank of China,https://www.sodocs.net/doc/364569925.html,/publish/diaochatongjisi/4032/index.html)

formal bank lending in China.

3Data

We now turn to our empirical analysis,starting with a description of the data used in our study.The activities illustrated in Figures 5and 2suggest that the transactions underlying the surrogate intermediation done by non-?nancial ?rms can be complex and not easy to disentangle.

Nor are these transactions easily measured or monitored.

Our strategy,

therefore,is to focus on the snapshot of the balance sheet at the end of the year,and investigate the co-movement in ?nancial assets and ?nancial liabilities of the ?rms,both in the cross-section and over time for each ?rm individually.

Our ?rm level data comes from Compustat Global.The advantage of this data is two-fold.

First,the database includes listed ?rms in China,which would include the large

non-?nancial ?rms that would be candidates for the intermediation activity described so far.Since we are interested in the ?rms engaged in the surrogate intermediation rather than the small and medium sized enterprises that are the ultimate borrowers,con?ning attention to

10

the large?rms will not miss the bulk of the surrogate intermediation.

Second,Compustat Global imposes accounting classi?cations that are designed to ensure that cross-country comparisons are possible.Cross-country comparability is important for our purpose,as one of the checks to our main investigation is to compare the empirical results for China with that for the United States.For such an exercise,cross-country comparability is crucial,and Compustat Global ensures broad comparability.

3.1Firm level data for China

Our sample of?rms from China in Compustat Global covers those?rms with Global Industry Classi?cation Standard(GICS)sector codes not equal to40.The sample period is from1990 to2012,with data cuto?date as November30,2013.For our benchmark regressions,we exclude the?rms which are outliers in terms of the ratio of cash and short-term investments to sales(above the99.5or below the0.5percentiles).After the sample selection,there are 1532?rms in our sample.

As our focus is on the surrogate intermediation activity of?rms,our focus is on the cash and short-term investment position of the?rms,as well as other?nancial assets.In what follows,“cash”is taken to mean cash and short-term investments.Financial liabilities are de?ned as the sum of the short—term debt and the long-term debt,which includes bank loans.Firm leverage is de?ned as?nancial liabilities divided by total assets.The summary statistics are presented in Table1.

We note the following features.First,cash-holdings of Chinese?rms grew rapidly over the sample period.The average cash-holding increased more than?ve-fold from RMB248.6 million in1990to RMB1,488.3million in2012.

Second,the growth of cash holdings was skewed to large?rms in the later periods,as suggested by the faster growth of the mean cash holding relative to the25and75percentiles. Therefore,the rapid growth of the average?nancial liabilities seems mainly to have been driven by large?rms.

11

Table1:Description of variables for the1990-2012Compustat sample for publicly traded Chinese non-?nancial companies.Cash includes short-term investments;?nancial liabilities are de?ned as the sum of the short—term debt and the long-term debt;?rm leverage is de?ned as?nancial liabilities devided by total

per-centile

per-

centile

?rms A.1990-2012

cash879.871.4198.4501.1179931532?nancial liabilities1,911.1119.0318.1893.0179931532 sales5,374.2351.1832.92,252.6179931532?rm leverage27.1%15.1%25.3%36.7%179931532 cash/sales31.5%10.4%19.9%37.9%179931532?nancial liabilities/sales59.3%19.0%39.2%72.4%179931532

B.1990-2001(Period1)

cash248.623.979.1205.84919775?nancial liabilities607.774.3170.0390.44919775 sales1,226.7192.2388.4838.34919775?rm leverage26.7%16.3%25.6%35.6%4919775 cash/sales31.1%7.3%17.1%38.4%4919775?nancial liabilities/sales63.9%24.0%45.6%78.3%4919775

C.2002-2007(Period2)

cash662.686.5206.1455.258731260?nancial liabilities1,533.5164.0382.8915.358731260 sales4,509.0395.5878.12,214.658731260?rm leverage28.6%17.1%27.0%38.0%58731260 cash/sales30.0%10.8%19.9%36.5%58731260?nancial liabilities/sales64.6%21.0%43.2%79.7%58731260

D.2008-2012(Period3)

cash1,488.3135.6354.0838.372011499?nancial liabilities3,109.5149.8447.21,450.072011499 sales8,912.8587.81,393.43,799.172011499?rm leverage26.1%12.8%23.7%36.0%72011499 cash/sales33.0%12.0%21.2%38.7%72011499?nancial liabilities/sales51.9%15.2%32.3%62.0%72011499

12

Figure7:China:Ratio of aggregate cash to aggregate sales of?rms in sample(positive bars)and ratio of aggregate?nancial liabilities to aggregate sales(negative bars)

Third,cash holdings grew faster than sales,which in turn grew more rapidly than?nancial liabilities.As a result,the ratio of cash to sales ratio has increased during the sample period, while the ratio of?nancial liabilities to sales has fallen in the sample period.

Figure7shows the cash to sales ratio and?nancial liabilities to sales ratio of the sample Chinese?rms.The chart indicates that the cash to sales ratio co-moved with the?nancial liability to sales ratio Figure8is the scatter plot of cash holdings versus sales,plotted in log scale.The slope of the scatter is close to1,suggesting that there is a roughly proportional relationship between cash holdings and sales,so that sales are a good normalizing variable for?rm size.A roughly proportion relationship between cash and sales would be consistent with a“bu?er stock”view of cash holdings,where?rms hold cash to serve as a bu?er against shocks to cash?ows.

13

Figure8:Scatter plot of cash vs sales in log scale for sample?rms in2000and2011.

3.2Bond Issuance

As well as?rm-level data,we will also employ aggregate corporate bond issuance series for non-?nancial?rms from China as an aggregate explanatory variable in the panel regressions. Aggregate corporate bond issuance serves as an indicator of the availability of credit through debt markets.When the corporate bond is issued in foreign currency,the issuance series also serves as an indicator of global capital market conditions and the availability of credit to?rms in China from international investors.

Figure11shows the total outstanding amounts of bonds for di?erent sectors in China. The chart uses total depository data from China Central Depository and Clearing Co.Cor-porate bonds grew from literally nothing to RMB5trillion between1997and2012.Even when the amounts are normalized relative to China’s GDP,we see from Figure10that the corporate bonds outstanding has increased very rapidly from only1%in2005to10%of China’s GDP by2012.

Figure11shows the breakdown of total corporate bonds by instrument.The corporate bonds category encompasses commercial paper(CP)and medium-term notes(MTN),both of which are shorter maturity instruments.Medium-term notes(MTNs)have grown most

14

Figure9:Total bond depository amount of China.Source:China Central Depository and Clearing Co..

Figure10:Total bond depository amount to GDP ratio in China.Source:China Central Depository and Clearing Co..

15

Figure11:Breakdown of total corporate bond depository amount.Source:China Central Depository and Clearing Co..

signi?cantly since their inception in2008,indicating the increasing need for medium-term ?nancing for Chinese?rms By2012,MTNs accounted for over half of the total corporate bonds.

Foreign currency bond issuance by Chinese?rms has also increased rapidly in recent years.The outstanding balance increased from USD4.7billion in2001to USD81.7billion in2012.Figure12plots the foreign currency bond outstanding relative to China’s GDP by sector.We see that private issuance by?rms was lower than the issuance by the government sector,but private issuance overtook government sector issuance in2008and has pulled away further since.As of2012,foreign currency corporate bonds outstanding is around1%of China GDP,while government bonds accounted for only0.25%.

4Panel Regressions for China

We proceed to examine panel regressions that ascertain how?nancial asset holdings vary with the?rm’s?nancial liabilities.

16

Figure12:Foreign currency bond outstanding to GDP ratio by government,?nancial institutions and other corporates.Source:Asian Development Bank.

4.1Panel regressions in log ratios

Our?rst set of panel regressions are for log of cash(including short-term investments)to sales ratio regressed on log of?nancial liabilities to sales ratio.Our interest is in the sign of the coe?cient on log of?nancial liabilities to sales ratio.We include log sales and?rm leverage,de?ned as?nancial liabilities to total assets,as control variables.We also include the full set of year?xed e?ects and?rm?xed e?ects.Table2presents the regression results. We present results below on the case where we have dropped observations for?rms that have zero?nancial liability at any date in the sample.Qualitatively,the results are unchanged when we include?rms with zero debt,although the coe?cient is smaller.

Column(1)is for the full sample.We see that the sign on ln(?n liab)is positive and signi?cant at the1%level.The coe?cient of0.209implies that a1%increase in?nancial liabilities to sales ratio in the cross-section translates into a0.21%increase in cash and short-term investment holdings to sales ratio.

In columns(2)to(5),we examine subgroups of?rms arranged into four size quartiles based on the average sales of the?rms over the period.As large?rms may have better

17

short-term investments;?nancial liabilities are de?ned as the sum of the short—term debt and the long-term debt;?rm leverage is de?ned as?nancial liabilities devided by total assets.Quartiles are for average sales Dependent variable:ln(cash to sales)(1)(2)(3)(4)(5)

Full sample Quartile1Quartile2Quartile3Quartile4

ln(?n liab to sales)0.209***0.218***0.08580.351***0.425***

(3.407)(2.672)(1.163)(3.713)(8.003)

ln(sales)-0.160***-0.131***-0.151***-0.163***-0.121***

(-5.544)(-2.688)(-3.271)(-3.388)(-3.200)

?rm leverage-1.353***-1.372***-0.735-2.373***-3.333***

(-2.656)(-2.636)(-1.444)(-3.780)(-12.22) manufacturing dummy×0.0185-0.02340.04870.06890.0249 ln(?n liab to sales)(0.611)(-0.377)(0.789)(1.016)(0.439) year?xed e?ects yes yes yes yes yes

?rm?xed e?ects yes yes yes yes yes constant-2.075***-0.201-2.739***-2.116***-0.755***

(-6.603)(-0.510)(-10.84)(-6.935)(-3.125)

Observations17,9933,9744,3734,7524,894

R-squared0.1880.2360.1650.2610.223 Number of?rms1,532383383383383

access to the capital market and are more likely to be engaging surrogate intermediation,

we would expect that large?rms to exhibit more of a positive association between their

?nancial liabilities and cash holdings.This is indeed what we?nd.Except for Quartile

2,the coe?cient on?nancial liabilities is signi?cantly positive at the1%level.We also see

that larger?rms have larger positive coe?cients,implying a higher cross-section elasticity

of cash holdings with respect to?nancial liabilities.For the highest quartile,the coe?cient

is0.425,implying an elasticity of42.5%for the increase in cash and short-term investments

relative to?nancial liabilities.

Table3is also in log ratios,but replaces the time?xed e?ects with the aggregate cor-

porate bond levels,given by the depository series of CCDC(China Central Depository and

Clearing Co.).We would expect a positive coe?cient on the corporate bond series,in-

dicating intermediary activity of the?rms leading to a positive association between bond

?nancing and?nancial asset holding.We restrict the sample in Table3to?rms with average

sales above the median(i.e.?rms in Quartile3and4),and divide the sample period into

three sub-periods:1990?2001,2002?2007,and2008?2012.

18

short-term investments;?nancial liabilities are de?ned as the sum of the short—term debt and the long-term debt;?rm leverage is de?ned as?nancial liabilities devided by total assets.Aggregate corporate bond is the total depository number from China Central Depository and Clearing Co.Robust t-statistics in parentheses. Dependent variable:ln(cash to sales)(1)(2)(3)(4)

1990-20121990-20012002-20072008-2012

ln(?n liab to sales)0.345***0.209*0.396***0.284***

(5.168)(1.803)(5.687)(3.456)

ln(sales)-0.137***-0.297***-0.135**-0.220***

(-4.075)(-3.130)(-2.505)(-4.673)

?rm leverage-2.420***-1.655*-3.663***-1.478**

(-4.710)(-1.747)(-10.57)(-2.474) manufacturing dummy×0.0211-0.01400.105-0.136** ln(?n liab to sales)(0.488)(-0.127)(1.540)(-2.310)

ln(aggregate corp bond)0.0446***0.523***-0.0522**0.196***

(3.453)(8.853)(-2.370)(6.908) year?xed e?ects no no no no

?rm?xed e?ects yes yes yes yes constant0.177-0.597 1.140***-0.781**

(0.998)(-1.238)(3.903)(-2.421)

Observations8,5191,6213,2663,632

R-squared0.1430.1840.2330.100 Number of?rms766438664751

Table3shows that the positive relationship between?nancial liabilities and cash holdings

is most pronounced in the period of2002?2007,consistent with the rapid growth of the

corporate bond market in this period.Corporate bonds have positive impacts on?rms’cash

holdings with the exception of the2002?2007period.

4.2Panel regressions in growth rates

So far,we have examined panel regressions in log ratios that are designed to capture cross-

section di?erences across?rms in their surrogate intermediation activity.

An alternative approach is to examine panel regressions where the dependent variable is

the log di?erence of cash holdings,and where explanatory variables are the log di?erence of

sales and the log di?erence of?nancial liabilities.The rationale for the panel regressions in

growth rates is to capture the direction of co-movement in the growth of cash holdings over

time for individual?rms with the growth in?nancial liabilities.The growth in sales plays

19