Insights into Capital Flows in Business Groups

Revisiting the Bright and Dark Sides of Capital Flows in Business Groups Joseph P. H. Fan The Chinese University of Hong Kong Li Jin Oxford University & Peiking University Guojian Zheng Sun Yat-sen University

Basic Business Group Structures Parent Co Parent Co V=50% Y V=C=50% Y V=C=10% Z V=20% Z V = 20%, weakest link in the chain, C = 10%. Pyramid structure allows leveraging up in control Joseph P.H. Fan Organization and Value 2

Wang Qunbin Fan Wei Liang Xinjun GUO Guangchang 22% 10% 10% 58% Shanghai Guangxin Technology Development Co. Ltd. Shanghai Fusun High Technology Co. Ltd. 10 49 (2) China Medical Holding Co. ACCORD PHARM (000028) 43.33 (2) 95% Shanghai Fusun Business Investment Co. Ltd. 10 5% Shanghai Fusun High Technology (Group) Co. Ltd. 90 53.92 24.53 Estate 30 Iron&Steel 20 1.94 Nanjing Iron &Steel United Co.,Ltd. 3.77 20 20 Medicine Fusun Pharm (600196) FORTE (HK2337) 13.53 Zhaojin Mining Co. Ltd. YYTM (600655) 21 70.95 NISC (600282) 25 (2) 90.3% 11.36 Commerce 8.81 Shanghai Fusun I.T. Development Co. Ltd. (Subsidiary) Shanghai Fusun Pharmaceutical Development Co. Ltd. 30 SFGIC (600827) LRGF (600285) 15.04 (2) Tangshan Jianlong Steel Co. Ltd. 36.03 (1) 26.04 (1) 11.95%(2) Shanghai Friendship- Fusun (holding) Co. Ltd. Lianhua Supermarket (HK0980) JianMin Pharm (600976) Tianjin Pharmaceutical Holdings, Ltd. TJPC (600488) 67.12 48 (2) Fusun Group (Shanghai, China) Ownership Joseph Fan 3

Business Group Structure Business Group Structure v.s v.s. Conglomerate Structure . Conglomerate Structure Group-like organizations are commonplace in the world. Comparing with the conglomerate organization Similarity: complex internal resource flows Difference: legal boundaries, top down control as oppose to horizontal control, complex ownership structure Because divisions are legally separate entities in a business group, intra-group capital flow becomes an issue to outside minority owners

Costs of the Group Structure Costs of the Group Structure Expropriation of minority shareholders by the controlling parent, e.g., tunneling(Johnson et al.,2000) From the perspective of the whole group, such tunneling might not be a zero-sum game. additional resources to cover up tunneling, potential legal penalties, ex ante distortion of incentives on investments Investors not systematically fooled, the insider of the business group ultimately bears the welfare loss from tunneling. Cost of the tunneling is reflected in lower security prices (Claessens et al., 2002; La Porta et al., 2002).

Motivation Motivation Given the costly group structure, what explains the persistent existence and prevalence of business groups? For the whole group in under-developed financial markets: tunneling may be a constrained optimum even if not the first-best outcome , if it alleviates severe financing constraints of member firms, and enables the undertaking of positive NPV projects Many papers in the conglomerate literature (Stein, 1997 ) Business group literature: Morck, Wolfenzon, Yeung, 2005; Almeida and Wolfenzon, 2006, 2010; Khanna and Yafeh, 2007; Gopalan et al., 2007; Masulis et al., 2012

Empirical Challenges Key challenge to empirical research demonstrating this tradeoff: to disentangle the resources diverted to facilitate group efficiency from those diverted to satisfy the private benefits of the controlling shareholder For minority shareholders: both are tunneling. But from the business group perspective, sacrificing a division may benefit the whole group Parent firms typically not observable because of non- listed status Measurement problem (Kaplan and Zingales, 1997, 2000)

Our attempt In this paper, we make a modest attempt to bypass the above difficulties by focusing on transfers of financial resources within business groups, and testing the hypothesis that intra-group capital flow may be motivated by both group capital allocation efficiency and pure expropriation of minority shareholders.

Business Group and Pyramidal Control Structure in China SAMB Private Owner Parent SOE (Parent) Parent Co. (Parent) Listed Firm (Listed sub) Listed Firm (Listed sub)

Ningbo Group (China) Local Government 100% 100% NingBo (Group) Co. LTD. Textile Holding Ningbo Industry Investment Co. LTD. 93.19% 2.33% 17.5% NINGBO GROUP CO., LTD VEKEN 6.13% 29.7% NINGBO VEKEN ELITE GROUP CO., LTD(600152) Joseph P.H. Fan Organization and Value 11

Predictions Predictions Cash flowing in the group is almost one- directional : from the listed sub to the parent. We should observe more intra-group cash flow activity (tunneling) if the parent and the listed sub are more severely misaligned in incentives, or if parent faces more financing constraints.

Predictions on magnitude and efficiency of intra- group capital flow activity (tunneling) D1: High parent incentive alignment with sub, less severe fin constraint D2: High parent incentive alignment with sub, severe fin constraint D3: Low parent incentive alignment with sub, less severe fin constraint D4: Low parent incentive alignment with sub, severe fin constraint Magnitude of capital flow: D4 > D2 & D3 > D1 Efficiency of tunneled capital: D2 > D1 & D4 > D3

Sample and Data Sample and Data 624 firm-year observations from 1999-2005 in China Each obs. includes a pair of firms (Listed sub and Parent). Financial information from both the listed subs and the non-listed parents Parent sample comes originally from National Bureau of Statistics (NBS) Annual Industrial Survey Database . Exclude : Shell or holding companies With missing data Can t be indentified in NBS Less than 20% shares of the listed sub Parent and the listed sub has the same 3-digit industry code Negative cash flow 16

Regression Model Regression Model Measure of ICF activities: investment of group member firm A out of cash flow of member firm B, controlling for cash flow of firm A (Shin and Stulz ,QJE,1998). + low OtherCashF lativeQ * Re 4 = + + Capex OwnCashFlo 1 w OtherCashF Re low 0 2 + + + lativeQ YearDummie s 3 Adding firm fixed effects and year dummies. Run regression for both Listed sub and Parent, compare the results: ICF exists when 2is positive and significant. For parent, financial tunneling is efficient when 4is positive and significant. 17

Relative investment opportunity Relative Q = difference in industry Tobin s q between the parent and the sub Use industry average q because parent is not listed and to mitigate measurement bias if firm level q is otherwise used

Cash flow measures in the literature Cash flow measures in the literature Cash flow=income after tax+ depreciation dividend payments Hoshi, Kashyap and Scharfstein (QJE,1991 ) Cash flow=earnings before extraordinary items+ depreciation Kaplan and Zingales (QJE,1997) Cash flow=earnings before interest and tax +depreciation + amortization (EBITDA) Kaplan and Zingales (QJE,1997) Cash flow=operating profit+ depreciation Shin and Stulz (QJE,1998); Shin and Park(JCF,1999)

Cash flow measures in our paper Cash flow measures in our paper Traditional cash flow measure: EBIT + depreciation Three adjusted cash flow measures Adjusted Cash Flow Measure 1: (EBIT)+ depreciation - net change in trade credits net change in trade credits =increase in accounts receivables - increase in payables We do not have amortization data. Rationale: (EBIT+ depreciation) is the accounting profit. Reasonable in conglomerates (EBIT + depreciation - net change in trade credits ) is the amount of cash that is available for use by either own firm or the other firm. A large fraction of EBIT take the form of trade credits.

Cash flow measures in our paper Cash flow measures in our paper Example: Suppose a listed sub has a total EBIT of $100,of which $30 is the increase of trade credits, then available CF is $70. Two possibilities of this $30 trade credits : naturally arise due to normal transactions only $70 available for tunneling, adjusted measures is appropriate implicit loans from one firm to the other $100 available for tunneling, traditional measures is appropriate 21

Cash flow measures in our paper Cash flow measures in our paper Which one more closely resembles the reality is an empirical question. If adjusted CF measure underestimates tunneling relative to the traditional CF measure, investment should be less sensitive to the adjusted CF measure than to the traditional CF measure. we found stronger sensitivity between investment of the parent and the adjusted CF of the listed sub, suggesting that the adjusted CF measure does not underestimate tunneling. Traditional CF measure may be noisy. It appears that tunneling in China takes less obvious forms than through the extension of trade credit 22

Cash flow measures in our paper Cash flow measures in our paper Adjusted Cash Flow Measure 2: EBIT + depreciation -net change in trade credits- income tax Adjusted Cash Flow Measure 3: EBIT + depreciation -net change in trade credits - income tax + net increase of bank debts and equities. For adjusted cash flow measure1, we have data from both Parent and Listed sub For adjusted cash flow measure 2 and 3, we only have data from Listed sub. 23

Summary Statistics Summary Statistics Variable Obs. Mean Listed Sub 0.1191 -0.0392 0.1016 0.0624 0.0455 0.0849 1.6380 0.0800 0.1150 52.0038 0.2447 229261 Median Mean Median Parent Capital Expenditure Net Trade Credits Traditional Cash Flow Measure Adjusted Cash Flow Measure 1 Adjusted Cash Flow Measure 2 Adjusted Cash Flow Measure 3 Industry Q Relative Q Industry Growth Cash Flow Right of Parent Bank Ownership Dummy Size (thousand Yuan) 604 604 604 604 604 604 604 604 604 604 604 604 0.0761 -0.0362 0.0966 0.0664 0.0474 0.0847 1.6914 0.0952 0.0925 54.4250 0.0000 143129 0.0864 0.0284 0.0699 0.0983 1.5580 -0.0800 0.1070 542179 0.0405 0.0197 0.0595 0.0789 1.5422 -0.0952 0.0869 235155 24

Panel A Listed Sub Regression Cash Flow Measure Traditional CF Measure 0.4143 (0.1154)*** 0.0441 (0.0301) 0.0214 (0.0126)* Adjusted CF Measure 1 0.3601 (0.0547)*** 0.0223 (0.0178) 0.0201 (0.0112)* 0.4235 (0.1261)*** 0.0476 (0.0357) 0.0243 (0.0138)* 0.0212 (0.0176) 604 0.3637 (0.0603)*** 0.0254 (0.0189) 0.0214 (0.0119)* 0.0168 (0.0210) 604 Own Cash Flow Other Cash Flow Relative Q Other Cash Flow * Relative Q Adj_R2 604 604 Panel B Parent Regression Cash Flow Measure Traditional CF Measure Adjusted CF Measure 1 Adjusted CF Measure2 Adjusted CF Measure 3 0.5041 0.5088 0.5114 0.5132 0.5262 0.5251 0.5278 0.5299 Own Cash Flow (0.0565)*** (0.0579)*** (0.0654)*** (0.0689)*** (0.0502)*** (0.0511)*** (0.0534)*** (0.0541)*** 0.0938 0.0802 0.2243 0.2067 0.2012 0.1874 0.1387 0.1031 Other Cash Flow (0.0445) ** (0.0466) * (0.0564)*** (0.0576)*** (0.0413)*** (0.0389)*** (0.0367)*** (0.0278)*** 0.0162 (0.0081)** 0.0159 (0.0102) 0.0842 (0.0551) 0.0177 (0.0068)** 0.0167 (0.0087)* 0.1504 (0.0750)** 0.0178 (0.0101)* 0.0166 (0.0113) 0.1231 (0.0682)* 0.0161 (0.0122) 0.0153 (0.0112) 0.0962 (0.0449)** Relative Q Other Cash Flow * Relative Q Adj_R2 604 604 604 604 604 604 604 604 25

Impact of corporate governance and Impact of corporate governance and financing constraint constraint on ICF financing on ICF Ownership is fundamental to and cash flow rights theoretically at the core of corporate governance. Empirically strongly related to the incentives of large shareholders to tunnel the listed firms that they control (Bertrand et. al, 2002; Claessens et.al, 2002). Especially when legal protection for outside investors is weak (La Porta,et.al, 1997, 1998, 1999) Bank ownership has been argued to be important to firms for raising external finance(e.g, Hoshi, et.al,1991) Bank ownership in this paper: A dummy variable whether listed sub(under Parent s control ) owns shares of local financial banks. 26

Cash Flow Measure Adjusted CF Measure 1 0.5042 (0.0434)*** 0.1578 (0.0972) 0.0151 (0.0093) 0.1343 (0.0622)** -0.0347 (0.0322) 0.5085 (0.0483)*** 0.1689 (0.0884)* 0.0156 (0.0089)* 0.1172 (0.0618)* Own Cash Flow Other Cash Flow Relative Q Other Cash Flow * Relative Q Low Cash Flow Right -0.0296 (0.0297) No Bank Ownership 0.1486 (0.0676)** -0.0612 (0.0303)** Low Cash Flow Right * Other Cash Flow Low Cash Flow Right * Other Cash Flow* Relative Q 0.1068 (0.0483)** 0.0831 (0.0401)** 604 0.24 No Bank Ownership * Other Cash Flow No Bank Ownership * Other Cash Flow* Relative Q Obs. Adj_R2 604 0.24

Putting the two effects together Putting the two effects together We next examine the joint effects of corporate governance and financing constraints. Consider four types of interactions of corporate governance and financing constraints: D1: high cash flow right and with bank ownership; D2: high cash flow right and without bank ownership; D3: low cash flow right and with bank ownership; D4: low cash flow right and without bank ownership D 1 as the benchmark 28

Cash Flow Measure Adjusted CF Measure 1 (2) (1) 0.5110 (0.0412)*** 0.1145 (0.0698) 0.0167 (0.0084)** 0.5145 (0.0475)*** 0.1062 (0.0657) 0.0150 (0.0091) 0.1142 (0.0693) 0.0325 (0.0384) -0.0278 (0.0754) -0.0411 (0.0332) 0.1124 (0.0642)* 0.1004 (0.0527)* 0.1924 (0.0825)** 0.1465 (0.0728)** -0.1012 (0.0358)*** 0.0334 (0.0452) 604 0.26 Own Cash Flow Other Cash Flow Relative Q Other Cash Flow* Relative Q 0.0312 (0.0347) -0.0269 (0.0691) -0.0398 (0.0287) 0.1247 (0.0621)** 0.1126 (0.0558)** 0.2043 (0.0726)*** D2 D3 D4 D2* Other Cash Flow D3* Other Cash Flow D4* Other Cash Flow D2* Other Cash Flow* Relative Q D3* Other Cash Flow* Relative Q D4* Other Cash Flow* Relative Q Obs. Adj_R2 604 0.24 29

Robust Tests Robust Tests Using Alternative Using Alternative Measure of Intra Measure of Intra- -group Cash Flows group Cash Flows Conventional measure of investment-cash flow sensitivity still comes as an estimate of the true capital flows Two more direct measures for intra-group capital flow : ORECTA : Other Receivables deflated by total assets Jiang et al (2010) ORECTA_Parent Other Receivables provided to controlling shareholder deflated by total assets

Dep. Variable ORECTA ORECTA_Parent (3) 0.0096 (0.0034)*** (1) (2) (4) 0.0075 (0.0040)* Low Cash Flow Right 0.0036 (0.0019)* 0.0049 (0.0021)** No Bank Ownership -0.0191 (0.0094)** -0.0078 (0.0024)*** Low Cash Flow Right * Relative Q 0.0252 (0.0120)** 0.0112 (0.0111) -0.6147 (0.0341)*** -0.0019 (0.0012) -0.0185 (0.0066)*** -0.0023 (0.0016) -0.0074 (0.0036)** 604 0.2356 0.0094 (0.0042)** 0.0102 (0.0066) -0.3117 (0.0231)*** -0.0036 (0.0009)*** -0.0055 (0.0045) -0.0037 (0.0011)*** -0.0053 (0.0024)** 604 0.1910 No Bank Ownership * Relative Q 0.0028 (0.0056) -0.6070 (0.0344)*** -0.0022 (0.0012)* -0.0173 (0.0066)*** -0.0021 (0.0014)* -0.0077 (0.0035)** 604 0.2365 0.0036 (0.0032) -0.3169 (0.0234)*** -0.0036 (0.0009)*** -0.0064 (0.0045) -0.0037 (0.0011)*** -0.0055 (0.0024)** 604 0.1918 Relative Q ROA Size State Marketization Layer Obs. Adj_R2

Dep. Variable ORECTA ORECTA_Parent (3) 0.0064 (0.0028)** 0.0072 (0.0031)** 0.0102 (0.0036)*** (1) (2) (4) 0.0036 (0.0021) 0.0095 (0.0051)* 0.0111 (0.0040)*** 0.0016 (0.0022) 0.0080 (0.0052) 0.0081 (0.0042)* 0.0459 (0.0200)** -0.0599 (0.0193)*** -0.0242 (0.0161) 0.0413 (0.0284) -0.6009 (0.0345)*** -0.0019 (0.0012) -0.0161 (0.0067)** -0.0022 (0.0016) -0.0079 (0.0036)** 604 0.2389 0.0059 (0.0027)** 0.0065 (0.0033)** 0.0091 (0.0049)** 0.0208 (0.0105)** -0.0367 (0.0126)*** -0.0188 (0.0130) 0.0209 (0.0124)* -0.3167 (0.0235)*** -0.0038 (0.0009)*** -0.0066 (0.0046) -0.0035 (0.0011)*** -0.0055 (0.0024)** 604 0.1914 D2 D3 D4 D2* Relative Q D3* Relative Q D4*Relative Q -0.0073 (0.0047) -0.6082 (0.0345)*** -0.0019 (0.0012) -0.0150 (0.0067)** -0.0021 (0.0016) -0.0078 (0.0036)** 604 0.2342 0.0028 (0.0032) -0.3174 (0.0234)*** -0.0037 (0.0009)*** -0.0065 (0.0045) -0.0036 (0.0011)*** -0.0055 (0.0024)** 604 0.1910 Relative Q ROA Size State Marketization Layer Obs. Adj_R2

Other Robust Tests Other Robust Tests Alternative proxy for investment opportunities: Industry Q: industry average Q matched from listed firms in China s stock market. Industry Growth: Industry average sale growth calculated from NBS Event of corporate governance change: Capital market regulation against expropriation by controlling shareholder from 2003. Alternative proxy for financing constraints: Firm size (Almeida and Campello, 2007; Erickson and Whited, 2000) 33

Industry Q and Industry Growth 0.1046 (0.0512)** 0.0941 (0.0486)* 0.1525 (0.0684)** 0.1628 (0.0714)** -0.2046 (0.0701)*** -0.0512 (0.0357) 604 0.25 0.1040 (0.0516)** 0.0747 (0.0486) 0.1721 (0.0791)** 0.1542 (0.0472)*** -0.0576 (0.0273)** 0.0847 (0.0619) 604 0.26 D2* Other Cash Flow D2* Other Cash Flow D3* Other Cash Flow D3* Other Cash Flow D4* Other Cash Flow D4* Other Cash Flow D2* Other Cash Flow* Industry Q D2* Other Cash Flow* Industry Growth D3* Other Cash Flow* Industry Q D3* Other Cash Flow* Industry Growth D4* Other Cash Flow* Industry Q D4* Other Cash Flow* Industry Growth Obs. Obs. Adj_R2 Adj_R2

Capital Market Regulation Panel A: Investment-Cash Flow Sensitivity Model Panel B: Capital Flow Determinant Model Cash Flow Measure Adjusted CF Measure 1 Dep. Variable ORECTA ORECTA_Parent (1) (2) (1) (2) 0.5012 (0.0446)*** 0.2042 (0.0655)*** 0.0161 (0.0084)* 0.4915 (0.0425)*** 0.1733 (0.0754)** 0.0156 (0.0099) 0.1285 (0.0665)* 0.0685 (0.0266)** -0.0542 (0.0288)* 0.1156 (0.0658)* 604 0.2598 -0.0415 (0.0078)*** 0.0103 (0.0053)* 0.0128 (0.0076) -0.6180 (0.0341)*** -0.0019 (0.0012) -0.0176 (0.0066)*** -0.0021 (0.0016) -0.0079 (0.0036)** 604 0.2341 -0.0245 (0.0054)*** 0.0053 (0.0020)** 0.0061 (0.0048) -0.3140 (0.0231)*** -0.0036 (0.0009)*** -0.0053 (0.0044) -0.0037 (0.0011)*** -0.0055 (0.0024)** 604 0.1914 Own Cash Flow Regulation Regulation * Relative Q Other Cash Flow Relative Q Relative Q ROA Other Cash Flow * Relative Q Size 0.0723 (0.0226)*** -0.0624 (0.0356)* Regulation State Regulation * Other Cash Flow Marketization Regulation * Other Cash Flow* Relative Q Layer Obs. Adj_R2 604 0.2438 Obs. Adj_R2 Regulation =1 if sample year is 2004-2005

Firm Size Panel A: Investment-Cash Flow Sensitivity Model Panel B: Capital Flow Determinant Model ORECTA_Paren t (2) 0.0042 (0.0021)** 0.0097 (0.0050)* 0.0084 (0.0048)* -0.3142 (0.0233)*** -0.0037 (0.0009)*** -0.0055 (0.0044) -0.0036 (0.0011)*** -0.0054 (0.0024)** 604 0.1918 Cash Flow Measure Adjusted CF Measure 1 Dep. Variable ORECTA (1) (2) (1) 0.4952 (0.0528)*** 0.1945 (0.0755)** 0.0171 (0.0084)** 0.4978 (0.0513)*** 0.1661 (0.0822)** 0.0141 (0.0086) 0.1278 (0.0725)* -0.1335 (0.0301)*** 0.0918 (0.0496)* 0.0809 0.0133 (0.0044)*** 0.0041 (0.0020)** 0.0060 (0.0041) -0.6037 (0.0343)*** -0.0016 (0.0012) -0.0159 (0.0066)** -0.0022 (0.0016) -0.0079 (0.0036)** 604 0.2388 Own Cash Flow Small Size Small Size * Relative Q Other Cash Flow Relative Q Relative Q ROA Other Cash Flow * Relative Q Size -0.1402 (0.0286)*** 0.1047 (0.0486)** Small Size State Small Size * Other Cash Flow Marketization Small Size * Other Cash Flow* Relative Q Layer (0.0391)** 604 0.2525 Obs. Adj_R2 604 0.2447 Obs. Adj_R2 Small Size=1 when parent firm size is below the median of sample.

Conclusion Conclusion We document the existence of two aspects of intra- group financing using 604 pair-years of Chinese listed firms and their non-listed parents : cross-financing to mitigate severe financing constraints, and the exploitation of minority shareholders due to weak corporate governance. Both can account for the rise of intra-group financing, but they have opposite impacts on group capital allocation efficiency: highest when the motivation is purely the mitigation of financial constraints, and lowest when it is purely expropriation of outside investors. 37

Thank You!