Nat­u­ral Monopoly and Mixed Own­er­ship Re­form - Based on Nat­u­ral Ex­per­i­ment and Cost Func­tion Anal­y­sis Method

- Based on Nat­u­ral Ex­per­i­ment and Cost Func­tion Anal­y­sis Method

China Economist - - Articles - ChenLin(陈林)

Ab­stract: De­spite a mul­ti­tude of the­o­ret­i­cal dis­cus­sions on China’s mixed own­er­ship re­form, very few stud­ies have ad­dressed re­al­is­tic ques­tions con­cern­ing the im­ple­men­ta­tion of the re­form. The Res­o­lu­tions of the Third Plenum of the 18th CPC Cen­tral Com­mit­tee and other re­form strate­gies have out­lined the re­form of sec­tors with nat­u­ral monopoly, in­clud­ing ur­ban pub­lic util­ity sec­tors. The ques­tion is how mixed own­er­ship re­form should be car­ried out in sec­tors of nat­u­ral monopoly, or which pub­lic util­i­ties sec­tors should en­joy pri­or­ity of mixed own­er­ship re­form. To an­swer this ques­tion, this pa­per em­ploys data of large pub­lic util­ity en­ter­prises in China from 1998 to 2008, and es­ti­mates the nat­u­ral monopoly at­tribute at the in­dus­try level and cor­po­rate to­tal fac­tor pro­duc­tiv­ity (TFP) us­ing cost func­tion anal­y­sis method ex­clud­ing the im­pact of prod­uct price fac­tor. Based on the dif­fer­ence-in­dif­fer­ences-in-dif­fer­ences (DDD) method of nat­u­ral ex­per­i­ment, an em­pir­i­cal test is car­ried out for the re­la­tion­ship among nat­u­ral monopoly, mixed own­er­ship re­form and cor­po­rate pro­duc­tiv­ity. Our re­sults sug­gest that: (1) Sta­tis­ti­cally, mixed own­er­ship re­form can­not sig­nif­i­cantly in­crease cor­po­rate TFP in sec­tors with nat­u­ral monopoly; (2) mixed own­er­ship re­form should not be car­ried out in­dis­crim­i­nately on a na­tion­wide ba­sis and for all pub­lic util­i­ties sec­tors. Such an at­tempt of re­form with­out dis­tin­guish­ing nat­u­ral monopoly and the level of com­pet­i­tive­ness is fraught with pol­icy un­cer­tain­ties; (3) rel­a­tive to sec­tors with nat­u­ral monopoly, cor­po­rate pro­duc­tiv­ity in com­pet­i­tive sec­tors af­ter mixed own­er­ship re­form will im­prove more sig­nif­i­cantly and en­joy greater “pol­icy div­i­dends” of in­sti­tu­tional re­form. There­fore, mixed own­er­ship re­form should be car­ried out first in com­pet­i­tive sec­tors.

Key­words: mixed own­er­ship re­form, nat­u­ral monopoly, cost func­tion, nat­u­ral ex­per­i­ment, dif­fer­ence-in-dif­fer­ences-in-dif­fer­ences (DDD) method

JEL Clas­si­fi­ca­tion Codes: H54; L98

DOI: 1 0.19602/j .chi­nae­conomist.2018.09.05

1. In­tro­duc­tion

Monopoly and com­pe­ti­tion, which com­pose a main­stream frame­work of eco­nomic the­o­ries, rep­re­sent ma­jor is­sues fac­ing so­cial­ist mar­ket econ­omy. For in­stance, re­form path­way for nat­u­ral monopoly is dis­cussed sep­a­rately in two parts at length in the Res­o­lu­tions of the Third Plenum of the

18th CPC Cen­tral Com­mit­tee1. The Res­o­lu­tions marks an at­tempt of top-level de­sign2 for the re­form of China’s mixed own­er­ship sec­tors and mo­nop­o­lis­tic sec­tors: sec­tors of nat­u­ral monopoly must be con­trolled by state cap­i­tal, while pri­vate cap­i­tal is per­mit­ted to en­ter “com­pet­i­tive sec­tors” with­out nat­u­ral monopoly; re­form path­way for sec­tors of nat­u­ral monopoly is an im­prove­ment of pre­vi­ous reg­u­la­tory ap­proach, i.e. state-owned en­ter­prises are sep­a­rated from the gov­ern­ment and op­er­ate un­der gov­ern­ment sanc­tion and over­sight, but does not in­clude mixed own­er­ship re­form.

Put sim­ply, in sec­tors of nat­u­ral monopoly, mixed own­er­ship re­form should not be car­ried out. In­stead, ex­ist­ing gov­ern­ment reg­u­la­tory sys­tems should be im­proved.

How­ever, top- level de­sign was not im­ple­mented in the re­form process, re­sult­ing in the lack of co­or­di­na­tion in the in­sti­tu­tional re­forms. First of all, the Min­istry of Hous­ing and Ur­ban-Ru­ral De­vel­op­ment (MOHURD) as the reg­u­la­tory au­thor­ity for ur­ban pub­lic util­i­ties en­acted the Im­ple­ment­ing Opin­ions on Fur­ther En­cour­ag­ing and Guid­ing the En­try of Pri­vate Cap­i­tal in Pub­lic Util­i­ties Sec­tors. This doc­u­ment calls for break­ing monopoly with­out mak­ing any dis­tinc­tion of nat­u­ral monopoly. It also states that “pub­lic util­i­ties in­vest­ment, con­struc­tion and op­er­a­tion mar­kets should be opened up” - such open­ing up nat­u­rally in­cludes all sec­tors. Pri­vate cap­i­tal is en­cour­aged to par­tic­i­pate in the mixed own­er­ship re­form of pub­lic util­ity sec­tors un­der MOHURD’s ad­min­is­tra­tion. In ad­di­tion to re­mov­ing monopoly, the new sys­tem also re­quires “stan­dard­iza­tion of mar­ket ac­cess in ac­cor­dance with gov­ern­ment li­censed op­er­a­tions.” How­ever, it does not spec­ify which sec­tors need dereg­u­la­tion and which oth­ers re­quire li­censed op­er­a­tion due to nat­u­ral monopoly. In fact, such a lack of clar­ity stems from the Opin­ions on En­cour­ag­ing and Guid­ing the Healthy De­vel­op­ment of Pri­vate In­vest­ment is­sued by the State Coun­cil in 2010, which is the up­per pol­icy reg­u­la­tion for MOHURD rules. This pol­icy reg­u­la­tion also makes no dis­tinc­tion of nat­u­ral monopoly in its pol­icy state­ment that “pri­vate cap­i­tal should be en­cour­aged to proac­tively par­tic­i­pate in the re­or­ga­ni­za­tion and re­form of mu­nic­i­pal pub­lic util­ity en­ter­prises.” This lack of clar­ity in­creased when lo­cal gov­ern­ments for­mu­lated lower-level reg­u­la­tions.

The root cause be­hind such vague­ness is the “Mar­shall Con­flict” that has baf­fled econ­o­mists for decades - should pol­i­cy­mak­ers pro­mote the economies of scale from nat­u­ral monopoly or in­tro­duce other types of cap­i­tal to in­spire mar­ket com­pe­ti­tion. In the the­o­ries of modern nat­u­ral monopoly, Wil­liam Bau­mol em­ployed “cost func­tion anal­y­sis method” to math­e­mat­i­cally prove the ex­is­tence of nat­u­ral monopoly (Bau­mol, 1977), and thus de­vel­oped the con­testable mar­kets the­ory - economies of scale may lead to ex­ces­sive en­try (com­pe­ti­tion) and thus harm mar­ket dy­namism and so­cial wel­fare. Co­in­ci­den­tally, the Res­o­lu­tions of the Third Plenum of the 18th CPC Cen­tral Com­mit­tee iden­ti­fies nat­u­ral monopoly and economies of scale as crit­i­cal ques­tions in rel­e­vant in­sti­tu­tional re­forms. Hence, the Res­o­lu­tions rep­re­sents a top-level de­sign of great the­o­ret­i­cal im­por­tance.

The ques­tions are: Whether mixed own­er­ship re­form should be car­ried out for sec­tors of nat­u­ral monopoly? Should mixed own­er­ship re­form be car­ried out in­dis­crim­i­nately for all ur­ban pub­lic util­ity sec­tors or for com­pet­i­tive sec­tors first? The­o­ret­i­cally, monopoly and com­pe­ti­tion rep­re­sent a main­stream frame­work for eco­nomic re­search. The ques­tion is how to bring the dis­cus­sions on mixed own­er­ship

re­form un­der this frame­work and seek a more re­li­able the­o­ret­i­cal ba­sis for mixed own­er­ship re­form (such as nat­u­ral monopoly the­ory and cost func­tion the­ory). The above prac­ti­cal and the­o­ret­i­cal ques­tions all con­tain great aca­demic value. In ad­di­tion, monopoly and state-owned econ­omy are both crit­i­cal ques­tions of eco­nomic re­search and aca­demic dis­cus­sions in China. Un­der the re­search frame­work of cost func­tion anal­y­sis cre­ated by Wil­liam Bau­mol (Bau­mol et al., 1982), this pa­per in­ves­ti­gates both ques­tions in com­bi­na­tion.

2. The­ory and Hy­poth­e­sis

In sec­tors of nat­u­ral monopoly, mixed own­er­ship re­form could be fraught with more pol­icy un­cer­tain­ties. The rea­son lies in the dif­fer­ence of cost func­tion for sec­tors of nat­u­ral monopoly.

Af­ter the first In­dus­trial Revo­lu­tion, econ­o­mists like John Stu­art Mill and Al­fred Mar­shall dis­cov­ered that ur­ban light­ing, wa­ter sup­ply and drainage and other mu­nic­i­pal in­fra­struc­ture net­works were char­ac­ter­ized by the ded­i­cated use of as­sets, sunk cost and economies of scale (later economies of range), and de­vel­oped the aca­demic con­cept of “nat­u­ral monopoly” - the fewer ex­ist­ing en­ter­prises there are, the lower to­tal cost of so­cial pro­duc­tion will be. Such sec­tors with spe­cial cost func­tions are en­tirely dif­fer­ent from com­pet­i­tive sec­tors, i.e. only mo­nop­o­lis­tic or oli­garchic mar­ket struc­ture will max­i­mize so­cial wel­fare, be­cause ex­ces­sive com­pe­ti­tion may harm the level of so­cial wel­fare for sec­tors of nat­u­ral monopoly. This the­ory was com­pre­hen­sively sup­ple­mented and im­proved by Bau­mol (1977), Pan­zar and Wil­lig (1981) et al. us­ing math­e­mat­i­cal mod­els. Af­ter­wards, Will­ner (1994) at­tempted to break the cost func­tion as­sump­tion of in­creas­ing mar­ginal cost and di­min­ish­ing re­turn to scale in the mixed own­er­ship re­form the­o­ret­i­cal model of De Fraja and Del­bono (1989), and in­cluded the qua­dratic term cost func­tion with vari­able av­er­age cost to ex­am­ine the ef­fect of economies of scale on mixed own­er­ship re­form. Since the economies of scale is a ba­sic char­ac­ter­is­tic of nat­u­ral monopoly (suf­fi­cient but not nec­es­sary con­di­tion), Will­ner’s study can be re­garded as the first pa­per of the­o­ret­i­cal ex­plo­ration on the mixed own­er­ship re­form in sec­tors of nat­u­ral monopoly.

Sunk cost and net­work ex­ter­nal­i­ties are also com­mon char­ac­ter­is­tics of sec­tors of nat­u­ral monopoly. Estrin and de Meza (1995) uses the for­mer and Will­ner (2006a) si­mul­ta­ne­ously uses both as the model as­sump­tion for sec­tors of nat­u­ral monopoly to in­ves­ti­gate the re­la­tion­ship be­tween nat­u­ral monopoly and mixed own­er­ship re­form. Re­sults in­di­cate that un­less a highly sig­nif­i­cant cost re­duc­tion oc­curs af­ter SOE re­form or the sunk cost is min­i­mal ( weak at­tribute of nat­u­ral monopoly), a higher pro­por­tion of state cap­i­tal and gov­ern­ment reg­u­la­tion should be re­tained in sec­tors of nat­u­ral monopoly. Will­ner (2006b) fur­ther demon­strates that for sec­tors of nat­u­ral monopoly with economies of scale, even if there is no dif­fer­ence of mar­ginal cost be­tween SOEs and pri­vate en­ter­prises ( pri­vate en­ter­prises do not have any com­par­a­tive ad­van­tage in terms of ef­fi­ciency), mixed own­er­ship re­form will still lead to greater to­tal so­cial wel­fare com­pared with a purely pri­vate or purely state-owned eco­nomic struc­ture.

Ex­ist­ing the­o­ret­i­cal stud­ies sug­gest that the form of cost func­tion in­flu­ences ev­ery as­pect of mixed own­er­ship re­form’s per­for­mance.

On the other hand, the most im­por­tant cri­te­rion for the in­sti­tu­tional per­for­mance of SOE re­form is ef­fi­ciency. There has been a pro­tracted de­bate over the ques­tion of ef­fi­ciency. Some schol­ars be­lieve that SOE re­form can bring about ef­fi­ciency im­prove­ment (Yang, 1997; Zhang, 1999; Zhang, 2004; Xu and Zhang, 2015; Wu and Zhang, 2015). Oth­ers con­sider that SOE re­form will not nec­es­sar­ily lead to an ef­fi­ciency im­prove­ment, and still less over­come the prob­lem of pol­icy bur­den on the state sec­tor of econ­omy (Lin, et al., 1987; Lin and Liu, 2001). Em­pir­i­cal ev­i­dences pro­vided by both sides also fo­cus on ef­fi­ciency. Some stud­ies find that SOE re­form is in­deed con­ducive to cor­po­rate pro­duc­tiv­ity (Sh­leifer et al., 1997, 1998; Liu, 2004; Liu and Li, 2005; Song and Yao, 2005; Jef­fer­son and Su, 2006; Dong, et al., 2006; Li and Qiao, 2010; Liu and Shi, 2010; Yu et al., 2013; Li and Yu, 2015). Some stud­ies hold

neg­a­tive views on the ef­fi­ciency ef­fect of mixed own­er­ship re­form (Lin and Li, 2004; Bai et al., 2006; Zhang and Zhang, 2011; Liu and Sun, 2013).

Among var­i­ous pol­icy ef­fects of mixed own­er­ship re­form, this pa­per will fo­cus on ef­fi­ciency dis­cus­sions to in­ves­ti­gate the ef­fi­ciency im­prove­ment ef­fect of re­form.

In real­ity, prod­ucts in sec­tors of nat­u­ral monopoly have a strong na­ture of pub­lic goods. Their pro­duc­tion, price and qual­ity are of great im­por­tance to pub­lic wel­fare. Sec­tors of nat­u­ral monopoly are of vi­tal im­por­tance to the econ­omy. In over­seas in­sti­tu­tional re­form prac­tices over re­cent years, own­er­ship does not ap­pear to be a ma­jor is­sue for re­form­ing sec­tors of nat­u­ral monopoly with the na­ture of pub­lic goods. For in­stance, in the United King­dom where the mixed own­er­ship re­form for sec­tors of nat­u­ral monopoly started in the 1970s, the re­form was crit­i­cized by aca­demi­cians for “turn­ing state monopoly into pri­vate monopoly” and fail­ing to achieve sig­nif­i­cant re­sults (Xiao, 2001). Sub­se­quently, the Bri­tish gov­ern­ment re­vamped in­sti­tu­tional re­form path­way: “The orig­i­nal mas­sive and unified sys­tem was dis­in­te­grated to achieve ef­fec­tive com­pe­ti­tion in sec­tors with­out nat­u­ral monopoly and ef­fec­tive reg­u­la­tion of sec­tors of nat­u­ral monopoly” (Xiao, 2011). For com­pet­i­tive busi­nesses in pub­lic util­ity sec­tors free from nat­u­ral monopoly, own­er­ship re­form re­mains a fea­si­ble ap­proach. Thus, mixed own­er­ship re­form in sec­tors of nat­u­ral monopoly may have some un­cer­tain­ties, and may not nec­es­sar­ily in­crease pro­duc­tiv­ity for mo­nop­o­lis­tic firms.

Based on the above anal­y­sis, this pa­per puts for­ward Hy­poth­e­sis 1 to be tested.

Hy­poth­e­sis 1: In ur­ban pub­lic util­ity sec­tors of nat­u­ral monopoly, mixed own­er­ship re­form can­not sub­stan­tially in­crease cor­po­rate pro­duc­tiv­ity.

Chi­nese schol­ars ex­ten­sively dis­cussed the mo­nop­o­lis­tic and com­pet­i­tive seg­ments of sec­tors of nat­u­ral monopoly (Qie, 2002; Lin and He, 2004; Li, 2004; Chen and Jiang, 2008). For this rea­son, the Third Plenum of the 18th CPC Cen­tral Com­mit­tee made the pol­icy state­ment that “while state cap­i­tal con­tin­ues to con­trol sec­tors of nat­u­ral monopoly, re­form should be car­ried out to sep­a­rate gov­ern­ment ad­min­is­tra­tion from en­ter­prise man­age­ment with gov­ern­ment sanc­tions and over­sight. In­fra­struc­ture net­work should be sep­a­rated from op­er­a­tion ac­cord­ing to the char­ac­ter­is­tics of dif­fer­ent in­dus­tries. Com­pet­i­tive sec­tors should be dereg­u­lated to pro­mote mar­ket-based al­lo­ca­tion of pub­lic re­sources,” and that “the scope of gov­ern­ment pric­ing should be lim­ited to crit­i­cal pub­lic util­i­ties, pub­lic-in­ter­est ser­vices and in­fra­struc­ture net­works of nat­u­ral monopoly.” Re­cently, Zhang and Zhang ( 2011) finds that SOE ef­fi­ciency is rather dif­fer­ent in mo­nop­o­lis­tic and com­pet­i­tive sec­tors. Mak­ing no dis­tinc­tion be­tween sec­tors with and with­out nat­u­ral monopoly is likely a chief rea­son for the in­ef­fec­tive mixed own­er­ship re­form of ur­ban pub­lic util­ity sec­tors and pol­icy un­cer­tain­ties.

Based on the above anal­y­sis, this pa­per puts for­ward Hy­poth­e­sis 2 on the ba­sis of Hy­poth­e­sis 1.

Hy­poth­e­sis 2: With­out mak­ing dis­tinc­tion be­tween sec­tors of nat­u­ral monopoly and com­pet­i­tive sec­tors, mixed own­er­ship re­form can­not sub­stan­tially in­crease cor­po­rate pro­duc­tiv­ity.

Ac­cord­ing to ex­ten­sive em­pir­i­cal ev­i­dences pro­vided by the academia, mixed own­er­ship re­form in the gen­eral sense can in­crease cor­po­rate pro­duc­tiv­ity (Liu, 2004; Song and Yao, 2005; Jef­fer­son and Su, 2006; Dong, et al., 2006; Hu Yi­fan, et al., 2006; Li and Qiao, 2010, Yu, et al., 2013). Hence, if the pub­lic util­ity sec­tors of the city are free from nat­u­ral monopoly and are com­pet­i­tive sec­tors in a so­cial­ist mar­ket econ­omy, re­lax­ing ac­cess thresh­old for pri­vate cap­i­tal, en­cour­ag­ing em­ployee stock own­er­ship and im­ple­ment­ing mixed own­er­ship re­form will nat­u­rally give rise to an ef­fi­ciency im­prove­ment ef­fect. Hence, we have Hy­poth­e­sis 3.

Hy­poth­e­sis 3: In com­pet­i­tive sec­tors free from nat­u­ral monopoly, mixed own­er­ship re­form has a sig­nif­i­cantly pos­i­tive ef­fect on the pro­duc­tiv­ity of mu­nic­i­pal pub­lic util­ity en­ter­prises.

Once the above hy­pothe­ses are proven, mixed own­er­ship re­form will not be the only op­tion for in­sti­tu­tional re­form of ur­ban pub­lic util­ity sec­tors of nat­u­ral monopoly, and should in­stead be car­ried out first in com­pet­i­tive sec­tors free from nat­u­ral monopoly.

3. De­sign of Study 3.1 Sam­ple Se­lec­tion

In China, ur­ban pub­lic util­ity sec­tors that may have nat­u­ral monopoly mainly in­clude the sec­tors of power sup­ply, heat sup­ply, fuel gas sup­ply, wa­ter sup­ply, sew­er­age treat­ment, waste treat­ment, mu­nic­i­pal land­scap­ing, etc. Given data avail­abil­ity, this pa­per se­lects five four-digit-code sec­tors in­clud­ing “4420 Power Sup­ply,” “4430 Heat Pro­duc­tion and Sup­ply,” “4500 Fuel Gas Pro­duc­tion and Sup­ply,” “4610 Tap­wa­ter Pro­duc­tion and Sup­ply” and “4620 Sewage Treat­ment” as re­search sub­jects. Data source is large Chi­nese en­ter­prises dur­ing 1998-2008 from the “Data­base of China’s In­dus­trial En­ter­prises.” In 2003, there was a change in the sta­tis­ti­cal scope of four-digit-code sec­tors. Us­ing the sec­tor codes af­ter 2003, this pa­per has ad­justed the sec­tor codes of 1998-2002.

This pa­per iden­ti­fies pri­vate en­ter­prises by the fi­nal con­trollers of listed com­pa­nies, and de­ter­mines the own­er­ship na­ture of en­ter­prises ac­cord­ing to the share­hold­ing sta­tus. With­out con­sid­er­ing the im­pact of le­gal per­son cap­i­tal, this pa­per iden­ti­fies en­ter­prises as state-owned en­ter­prises, col­lec­tive en­ter­prises, pri­vate en­ter­prises, en­ter­prises with in­vest­ment from Hong Kong, Ma­cao and Tai­wan, and for­eign-in­vested en­ter­prises if any of state cap­i­tal, col­lec­tive cap­i­tal, per­sonal cap­i­tal, cap­i­tal from Hong Kong, Ma­cao and Tai­wan, or for­eign cap­i­tal ac­counts for more than 50% in their “to­tal paidin cap­i­tal,” or if any of state cap­i­tal, per­sonal cap­i­tal, cap­i­tal from Hong Kong, Ma­cao and Tai­wan, or for­eign cap­i­tal is the max­i­mum cap­i­tal in their “to­tal paid-in cap­i­tal”; their “sta­tus of state con­trol­ling share” are ac­cord­ingly reg­is­tered as “ab­so­lute (rel­a­tive) con­trol­ling share held by the State,” “pri­vately held con­trol­ling share,” “con­trol­ling share held by in­vestor from Hong Kong, Ma­cao or Tai­wan,” and “con­trol­ling share held by for­eign in­vestor” re­spec­tively. When a state-owned or col­lec­tive en­ter­prise in­tro­duces pri­vate cap­i­tal and thus be­comes con­trolled by pri­vate cap­i­tal, this sam­ple is de­fined as an en­ter­prise hav­ing com­pleted mixed own­er­ship re­form. Fi­nally, we ob­tained 55,101 ob­ser­va­tions in five sec­tors, in­clud­ing 7,111 ob­ser­va­tions hav­ing com­pleted mixed own­er­ship re­form. On av­er­age, there are 711 en­ter­prises which com­pleted mixed own­er­ship re­form each year, as well as 6,135 ob­ser­va­tions whose fi­nal con­trollers are “le­gal per­son’s cap­i­tal” and whose own­er­ship na­ture thus can­not be de­ter­mined.

3.2 Nat­u­ral Ex­per­i­ment, “Ob­ser­va­tion Pe­riod” Method and Econo­met­ric Model

As an ear­lier study, Bai et al. (2006) em­ploys cor­po­rate level data from “Data­base of Chi­nese In­dus­trial En­ter­prises,” and in­tro­duces the time-dif­fer­ence vari­able of “whether re­form has been car­ried out” into its econo­met­ric model for the first time to de­pict the dy­namic change in cor­po­rate per­for­mance be­fore and af­ter re­form. Li and Qiao (2010), Yu, et al. (2013), Chen and Tang (2014), Yu, et al. (2016) and Sheng and Liu (2016) fur­ther com­bine the group­ing dif­fer­ence vari­able of “par­tic­i­pa­tion in re­form” with time-dif­fer­ence vari­able to carry out a dif­fer­ence-in-dif­fer­ences study based on nat­u­ral ex­per­i­ment. How­ever, the dif­fer­ence-in-dif­fer­ences method does not com­pletely ap­ply to the topic of re­search in this pa­per, since such a frame­work can­not ex­am­ine the in­ter­ac­tive ef­fect be­tween nat­u­ral monopoly and SOE re­form per­for­mance. In or­der to test Hy­pothe­ses 1 and Hy­poth­e­sis 3, this pa­per will fur­ther em­ploy the dif­fer­ence- in- dif­fer­ence- in- dif­fer­ences ( DDD) method. DDD method is ex­ten­sively ap­plied in in­ter­na­tional stud­ies (Gru­ber, 1994; Meyer, 1995; Yelowitz, 1995; Hut­tunen, et al., 2013; Garth­waite, et al., 2014; Chen, 2017), and Chi­nese stud­ies us­ing this method in­clude Deng, et al. (2014), Fu, et al. (2015), Wang (2016), etc.

Or­di­nary DDD stud­ies are all based on the frame­work of nat­u­ral ex­per­i­ment with “unique ex­per­i­men­tal pe­riod (si­mul­ta­ne­ous pol­icy im­pact).” For in­stance, in Garth­waite et al. (2014), the pol­icy im­pact oc­curred dur­ing the US pub­lic med­i­cal in­sur­ance re­form of 2005. In Hut­tunen, et al. (2013), the pol­icy im­pact oc­curred dur­ing the EU re­form of salary tax sub­sidy im­ple­mented on the New Year’s

Day of 2006. In Yelowitz (1995), the pol­icy im­pact oc­curred dur­ing the US women’s med­i­cal in­sur­ance ex­pan­sion pro­gram in 1991. In Gru­ber ( 1994), the pol­icy im­pact oc­curred dur­ing the US Fed­eral gov­ern­ment’s la­bor in­sur­ance re­form of 1978. The above stud­ies have the fol­low­ing things in com­mon: The sub­jects of ex­per­i­ment suf­fered uni­form pol­icy im­pacts that were ex­pe­ri­enced through­out the United States and the Euro­pean Union. Most rel­e­vant Chi­nese stud­ies are also based on nat­u­ral ex­per­i­ments of unique ex­per­i­men­tal pe­riod. How­ever, mixed own­er­ship re­form was car­ried out in dif­fer­ent pe­ri­ods of time across var­i­ous lo­cal­i­ties in China. Dif­fer­ent en­ter­prises im­ple­mented mixed own­er­ship re­forms in dif­fer­ent pe­ri­ods of time as well. Hence, ref­er­enc­ing the “ob­ser­va­tion pe­riod” method of Yu, et al. (2013), this study or­ga­nizes ex­per­i­ment group sam­ples with in­con­sis­tent ex­per­i­men­tal pe­ri­ods into nat­u­ral ex­per­i­ment sam­ples with ap­prox­i­mately con­sis­tent ex­per­i­men­tal pe­ri­ods.

First, the en­tire time span is roughly di­vided into three seg­ments: 2002- 2005 ( four years) is ob­ser­va­tion pe­riod of pub­lic pol­icy ex­per­i­ment; 1998-2001 (three years) is pre-re­form pe­riod; 20062008 (three years) is post-re­form pe­riod. Such dif­fer­en­ti­a­tion of time pe­ri­ods aims to test the dif­fer­ence of per­for­mance be­tween ex­per­i­ment group and con­trol group and be­fore and af­ter the mixed own­er­ship re­form.

In or­der to test Hy­poth­e­sis 2, this pa­per con­ducts re­gres­sion of nat­u­ral monopoly sam­ples and non­nat­u­ral monopoly sam­ples in a mix­ture to in­ves­ti­gate whether the re­sults of mixed own­er­ship re­form are af­fected, and de­signs a dif­fer­ence-in-dif­fer­ences econo­met­ric model. is ex­plained vari­able; is the group­ing vari­able of whether sam­ples par­tic­i­pated in mixed own­er­ship re­form; time-dif­fer­ence vari­able is a dummy vari­able which de­notes the time pe­ri­ods be­fore and af­ter mixed own­er­ship re­form; is un­ob­serv­able in­di­vid­ual fixed ef­fect; is the fixed ef­fect of year (dummy vari­able of year); is the fixed ef­fect of sec­tor (sec­tor dummy vari­able), which con­trols the un­ob­serv­able fac­tors be­tween var­i­ous sec­tors; is sto­chas­tic dis­tur­bance term. Con­sid­er­ing the im­pact of en­tity and time fixed ef­fects, this pa­per in­tro­duces ref­er­enc­ing Chen (2017), Sun, et al. (2017), Shi and Wang (2017), Shi and Yue (2016), Fu, et al. (2015), Garth­waite, et al. (2014), Jian (2013) and Lu, et al. (2013), and in­tro­duces sec­tor fixed ef­fect ref­er­enc­ing Wang (2016).

If a firm car­ried out mixed own­er­ship re­form dur­ing 2002-2005, it is then de­fined as ex­per­i­ment group, and the value of group­ing dummy vari­able is 1; sam­ples which did not carry out mixed own­er­ship re­form dur­ing 1998-2008 are de­fined as con­trol group, 03. The value of timed­if­fer­ence vari­able be­fore 2002 is 0, and the value af­ter 2005 is 1. is a dif­fer­ence-in-dif­fer­ences es­ti­ma­tor, and if mixed own­er­ship re­form can in­crease cor­po­rate pro­duc­tiv­ity, its re­gres­sion ef­fect should be sig­nif­i­cantly pos­i­tive. The spe­cific values of group­ing and time-dif­fer­ence vari­ables are as fol­lows:

DDD econo­met­ric model is spec­i­fied as:

is group­ing vari­able for nat­u­ral monopoly, and is 0 when the ur­ban pub­lic util­ity sec­tor in the city where sam­ples are lo­cated is of nat­u­ral monopoly. Oth­er­wise, it is 1. This vari­able re­flects the level of nat­u­ral monopoly and com­pet­i­tive­ness of the sec­tor of a firm. Spe­cific es­ti­ma­tion is shown be­low. DDD es­ti­ma­tor is the ba­sis for the assess­ment of whether the ex­per­i­ment treat­ment (pol­icy im­ple­men­ta­tion) has any sig­nif­i­cant im­pact on the ex­per­i­ment de­pen­dent vari­able ( ex­plained vari­able). Its re­gres­sion co­ef­fi­cient is the ex­per­i­men­tal ef­fect of ex­per­i­men­tal vari­able (DDD es­ti­ma­tor) on the de­pen­dant vari­able (ex­plained vari­able) of ex­per­i­ment group, i.e. the in­te­grated pol­icy ef­fects of nat­u­ral monopoly and mixed own­er­ship re­form. If re­gres­sion re­sult is sig­nif­i­cant, it sug­gests that as long as sam­ples are in a non-nat­u­ral monopoly state, mixed own­er­ship re­form will have a sig­nif­i­cant ef­fect on the ex­plained vari­able.

The above is a nat­u­ral ex­per­i­ment de­sign whose time of ex­per­i­ment is not unique. This pa­per marks such a DID/DDD econo­met­ric model based on “ob­ser­va­tion pe­riod” method as DID( 1) ( DDD( 1)). This pa­per as­sumes that the gov­ern­ment is the ex­per­i­ment’s oper­a­tor, and that the ex­per­i­ment started in 1998 and fin­ished in 2008 (i.e. the time span of data sam­ples). The gov­ern­ment’s pur­pose in con­duct­ing such an ex­per­i­ment was to ex­am­ine whether the ex­per­i­men­tal vari­able (mixed own­er­ship re­form) was able to af­fect the ex­per­i­men­tal de­pen­dant vari­able (TFP) 4.

In the sta­tis­ti­cal sense, if the time of mixed own­er­ship re­form is in­con­sis­tent, the nat­u­ral ex­per­i­ment’s 0 may not be guar­an­teed. Hence, nat­u­ral ex­per­i­ments whose ex­per­i­men­tal pe­ri­ods are not unique usu­ally com­pare an ex­per­i­men­tal group with rel­a­tively con­sis­tent pol­icy im­pact pe­ri­ods with a con­trol group. Based on the DID/DDD econo­met­ric model of “ob­ser­va­tion pe­riod” method, this pa­per ex­cludes the en­ter­prise sam­ples whose re­form pe­riod is too dis­tant (dur­ing 19982001) and those whose re­form pe­riod is too re­cent (2006-2008). In this man­ner, this pa­per cre­ates a nat­u­ral ex­per­i­ment with ap­prox­i­mately con­sis­tent ex­per­i­men­tal pe­riod (the re­form took place dur­ing 2002-2005), so as to in­ves­ti­gate whether the ex­per­i­men­tal vari­able (mixed own­er­ship re­form and nat­u­ral monopoly) can in­flu­ence the ex­per­i­men­tal de­pen­dent vari­able (cor­po­rate pro­duc­tiv­ity).

3.3 Panel Data DID/DDD Econo­met­ric Model

In or­der to en­sure the ro­bust­ness of re­sults, this pa­per em­ploys panel data DID/DDD econo­met­ric model to con­duct a ro­bust­ness test. Mixed own­er­ship re­form is a grad­ual process. Each year, it is car­ried out by a dif­fer­ent num­ber of en­ter­prises in the pub­lic util­ity sec­tors of var­i­ous cities. There­fore, this pa­per em­ploys panel data DID econo­met­ric model to ex­am­ine dif­fer­ent pe­ri­ods of cor­po­rate re­form

ref­er­enc­ing the method of Lu, et al. (2013), Yu, et al. (2013), Lu and Yu (2015), Sheng and Liu (2016): is DID es­ti­ma­tor, and ex­am­ines the fixed ef­fects of en­tity, time and sec­tor. The ex­per­i­ment group of equa­tion (5) is the to­tal sam­ples which car­ried out mixed own­er­ship re­form dur­ing 1999-2008 (1998 as the ini­tial year), and the value of its group­ing dummy vari­able

is 1. Sam­ples that did not carry out mixed own­er­ship re­form dur­ing this pe­riod of time are de­fined as con­trol group, 0. If ex­per­i­ment group sam­ples i con­ducted mixed own­er­ship re­form in year t0 , then = 1, = 0 and of re­main­ing sam­ples is spec­i­fied as 0. We have: DDD model is then cre­ated: For DDD es­ti­ma­tor , its re­gres­sion co­ef­fi­cient is ex­pected to be sig­nif­i­cantly pos­i­tive. The above DID/DDD model is marked as DID( 2) ( DDD( 2)).

Com­pared with “ob­ser­va­tion pe­riod” method DID( 1) and ( DDD( 1)), panel data DID/DDD model DID( 2) / DDD( 2) has cer­tain prob­lems of het­eroscedas­tic­ity and time-se­ries au­to­cor­re­la­tion. In­tro­duc­ing may con­trol for the above prob­lems and re­duce the risk of spu­ri­ous re­gres­sion to some ex­tent. re­spec­tively de­notes three dif­fer­en­tial vari­ables , and . Due to multi- collinear­ity prob­lem, fixed ef­fect and dif­fer­en­tial vari­able ( , , ) nor­mally will not si­mul­ta­ne­ously en­ter into re­gres­sion equa­tion.

The pair­wise in­ter­ac­tion ef­fects of cor­re­spond to ,

and , and are si­mul­ta­ne­ously in­tro­duced into re­gres­sion equa­tion5 as equa­tion (4).

The pair­wise com­bi­na­tion ef­fect of fixed ef­fect can­not be ne­glected, since the re­gres­sion co­ef­fi­cient of pair­wise cross-mul­ti­ply­ing term of dou­ble dif­fer­ence in DDD(1) has spe­cific ex­per­i­men­tal sig­nif­i­cance.

Main­stream DDD stud­ies have all con­sid­ered the DID vari­able or the joint ef­fect of pair­wise com­bi­na­tion, and the for­mer is more com­mon. Among Chi­nese stud­ies, Wang (2016) at­tempted to con­trol the fixed ef­fect of pair­wise com­bi­na­tion, and ex­cluded the dif­fer­en­tial vari­able of ex­per­i­men­tal group­ing (time). In ad­di­tion to en­tity and time fixed ef­fects, the pa­per also in­cludes the fixed ef­fects of prov­ince, sec­tor, etc. Time of ex­per­i­ment is also not unique for the nat­u­ral ex­per­i­ment cre­ated by Fan and Peng (2017). The study also ex­cludes ex­per­i­ment group­ing vari­able, re­plac­ing it with sec­tor fixed ef­fect and adding en­tity fixed ef­fect to par­tially avoid the multi-collinear­ity prob­lem. Garth­waite et al. (2014) only re­tains the third dif­fer­en­tial vari­able while ex­clud­ing ex­per­i­men­tal group­ing and time dif­fer­en­tial vari­ables and adding en­tity and time fixed ef­fects, but only con­trols for the fixed ef­fect of

pair­wise com­bi­na­tion. Take the vari­able names of this pa­per for in­stance, the in­de­pen­dent vari­able of the DID model of Garth­waite et al. (2014) is , and its is the dummy vari­able of prov­ince (pre­fec­ture). The cross-mul­ti­ply­ing term con­tain­ing in this pa­per’s en­ter­prise panel data model does not have very strong eco­nomic sig­nif­i­cance, and its ma­trix cal­cu­la­tion is hard to process. Thus, it is re­placed with sec­tor fixed ef­fect .

In sum­mary, this pa­per has cre­ated a panel data DDD eco­nomic model, which is ex­pressed as DDD( 3):

3.4 Core Vari­able Treat­ment

(1) Mea­sure­ment of nat­u­ral monopoly

is the out­put of firm i in year t ; is the price of type j pro­duc­tion fac­tor ob­tained by firm i in year tk ; is the num­ber of types of in­put fac­tor, and is the to­tal cost of firm pro­duc­tion. The size of re­gres­sion co­ef­fi­cient de­ter­mines the char­ac­ter­is­tics of pro­duc­tion func­tion and its dual pro­duc­tion func­tion. In or­der to en­sure that the cost func­tion is sec­ond or­der dif­fer­en­tiable,

. In or­der to sat­isfy the cost func­tion’s ho­mo­ge­neous­ness of de­gree one with re­spect to fac­tor price vec­tor (i.e. to­tal fac­tor price in­creases with to­tal cost by the same pro­por­tion), we must en­sure that

0, and use Shep­hard’s Lemma, as the con­di­tion of re­gres­sion con­straint. is the share of the con­sump­tion of type pro­duc­tion fac­tor in to­tal cost.

Based on data avail­abil­ity and pre­vi­ous re­search ex­pe­ri­ence, this pa­per as­sumes that firm pro­duc­tion con­forms to DID pro­duc­tion cost, and that cap­i­tal K (mu­nic­i­pal in­fra­struc­ture net­works and rel­e­vant equip­ment, etc.) and la­bor L are the two ba­sic in­put fac­tors for the pro­duc­tion of ur­ban pub­lic util­ity firms, i.e. the value of k is 2. The trans-log cost func­tion of the two fac­tors will em­ploy seem­ingly un­re­lated re­gres­sion mod­els (SUR) method (Zell­ner, 1965). Since the pro­duc­tion func­tions of the pub­lic util­ity sec­tors of five cities should be dif­fer­ent, the re­gres­sion of cost func­tion should be car­ried out by sec­tor group­ing.

Ac­cord­ing to Bau­mol et al. (1982, page 17), when a sec­tor reaches a cer­tain ag­gre­gate out­put Q, once cor­po­rate pro­duc­tion cost func­tion sat­is­fies = 0 , the sec­tor will have a strict “cost sub-ad­di­tiv­ity,” i.e. nat­u­ral monopoly. Cost weak ad­di­tiv­ity and nat­u­ral monopoly are mu­tu­ally suf­fi­cient and nec­es­sary con­di­tions6. is the ar­ti­fi­cially em­bed­ded share of the to­tal out­put of sep­a­rated mo­nop­o­lis­tic sec­tors. For the sim­plic­ity of cal­cu­la­tion, ’s value is 0.5.

Nat­u­ral monopoly is a sec­tor’s in­di­ca­tor, and ur­ban pub­lic util­ity en­ter­prises gen­er­ally will not

en­gage in cross-re­gional com­pe­ti­tion. Thus, in es­ti­mat­ing , Q is the an­nual to­tal out­put of a pub­lic util­ity sec­tor of a city. In fit­ting C(Q), co­ef­fi­cient ob­tained from re­gres­sion of equa­tion (10) will be em­ployed, and cap­i­tal and la­bor prices are the av­er­age lev­els of a cer­tain pub­lic util­ity sec­tor of a cer­tain city.

When a cer­tain pub­lic util­ity sec­tor of a cer­tain city 0, the im­pli­ca­tion is that its pro­duc­tion meets cost sub-ad­di­tiv­ity, i.e. sec­tor of nat­u­ral monopoly; 1 sug­gests that the sec­tor is a com­pet­i­tive sec­tor of non-nat­u­ral monopoly. More de­tailed stud­ies on the mea­sure­ment of nat­u­ral monopoly of trans-log cost func­tion in­clude: Chris­tensen et al. (1975), Evans and Heck­man (1984), Gils­dorf (1995), Wilson and Zhou (2001), Fraque­lli et al. (2004), Chen and Liu (2014), Wang and Liu (2016).

From the fol­low­ing cal­cu­la­tions, we may ob­tain the nat­u­ral monopoly at­tribute of the power sup­ply sec­tor (4420) of 3,214 cities above pre­fec­ture level, heat sup­ply sec­tor (4430) of 1,517 cities, gas sup­ply sec­tor (4500) of 1,680 cities, wa­ter sup­ply sec­tor (4610) of 3,680 cities, and sewage treat­ment sec­tor of 364 cities (4620). On av­er­age, we are able to es­ti­mate the nat­u­ral monopoly at­tribute of 948 cities for each year. Thus, we cre­ate the group­ing vari­able of nat­u­ral monopoly:

(2) Mea­sure­ment of cor­po­rate pro­duc­tiv­ity

In or­der to com­pre­hen­sively eval­u­ate over­all cor­po­rate pro­duc­tiv­ity, the ex­plained vari­able in this pa­per fol­lows to­tal fac­tor pro­duc­tiv­ity (TFP). Ac­cord­ing to the def­i­ni­tion of TFP and eco­nomic growth ac­count­ing equa­tion, when cost func­tion is con­stant re­turn to scale, ,where

is the growth rate of cap­i­tal and la­bor fac­tor in­puts; are the shares of cap­i­tal and la­bor fac­tor con­sump­tion in to­tal cost; is TFP and out­put growth rates. Ac­cord­ing to the cost func­tion deriva­tion of vari­able re­turn to scale by Wu et al. (2015) and Zhang et al. (2016), we ob­tain TFP growth rate on the ba­sis of vari­able re­turn to scale in equa­tion (10):

By ob­tain­ing from equa­tion ( 10), we may ar­rive at in equa­tion ( 12). is the re­turn to scale pa­ram­e­ter of­ten em­ployed in aca­demic re­search. When is small than 1, firm pro­duc­tion will be in a stage of in­creas­ing re­turn to scale. Oth­er­wise, it is in a stage of di­min­ish­ing re­turn to scale.

How­ever, the re­gres­sion of equa­tion (10) has a cer­tain prob­lem of er­ror term con­tem­po­ra­ne­ous cor­re­la­tion. Even if seem­ingly un­re­lated re­gres­sion method is em­ployed for con­trol, the pos­si­bil­ity of ro­bust­ness prob­lem in es­ti­mat­ing can­not be ex­cluded. In or­der to bet­ter con­trol for si­mul­tane­ity and se­lec­tion bias prob­lems, this pa­per at­tempts to in­tro­duce the semi-para­met­ric es­ti­ma­tion method pro­posed by Ol­ley and Pakes (1996) (“OP method” for short) to re-es­ti­mate the sec­ond ex­plained vari­able and con­duct ro­bust­ness test.

Since the pro­duc­tion func­tion un­der OP method has a con­stant re­turn to scale, the cost func­tion in equa­tion (10) is re­gressed into con­stant re­turn to scale, and ref­er­enc­ing Chen and Zhu (2017), we have:

is the cur­rent in­vest­ment of firms, and de­notes un­ob­serv­able pro­duc­tiv­ity. con­forms to first-or­der Markov process , i. e. , and is sto­chas­tic pro­duc­tiv­ity im­pact that con­forms to first-or­der Markov process. Ex­pected pro­duc­tiv­ity of a fu­ture phase is a func­tion of cur­rent-phase pro­duc­tiv­ity and cap­i­tal price . Thus, we may as­sume that there ex­ists a pro­duc­tiv­ity thresh­old : if firm pro­duc­tiv­ity is above the thresh­old, the firm will opt to stay in the mar­ket. Oth­er­wise, it will opt to exit the mar­ket.

The spec­i­fi­ca­tions of in­vest­ment de­ci­sion equa­tion and sur­vival prob­a­bil­ity equa­tion are con­sis­tent with Ol­ley and Pakes (1996). Specif­i­cally, firm in­vest­ment de­ci­sion-mak­ing is sub­ject to pro­duc­tiv­ity level and cap­i­tal price in the cur­rent phase . Sur­vival prob­a­bil­ity is fit­ted us­ing Pro­bit model , i. e. to ob­tain fit­ted value of prob­a­bil­ity . Given the lack of fixed as­set in­vest­ment in­di­ca­tor in Chi­nese In­dus­trial En­ter­prises Data­base, our es­ti­ma­tion is con­ducted based on . is the net value of fixed as­sets, and is cur­rent-phase de­pre­ci­a­tion.

is the pro­duc­tiv­ity which can be ob­served by firms but can­not be ob­served by re­searchers, and is the pro­duc­tiv­ity volatil­ity and mea­sure­ment er­ror which nei­ther firms nor re­searchers are able to ob­serve. Hence, will not af­fect firm de­ci­sions, while will in­flu­ence the cur­rent phase de­ci­sions of firms. is a sec­ond-or­der poly­no­mial which con­tains for ap­prox­i­ma­tion. Equa­tion (13) em­ploys OLS es­ti­ma­tion. Since semi-para­met­ric poly­no­mial con­trols for ob­serv­able pro­duc­tiv­ity volatil­ity, the es­ti­ma­tion of its re­gres­sion co­ef­fi­cient has con­sis­tency.

In the next step, we will es­ti­mate the elas­tic­ity co­ef­fi­cient of cap­i­tal price. Af­ter the es­ti­ma­tion re­sults of and sur­vival prob­a­bil­ity are known, equa­tion (14) is cre­ated. Specif­i­cally, is ap­prox­i­mated us­ing the sec­ond-or­der poly­no­mial which con­tains and . Based on non-lin­ear OLS es­ti­ma­tion, we may ob­tain the es­ti­mated value of and TFP es­ti­ma­tion equa­tion (15). Ac­cord­ing to equa­tion (15), the es­ti­mated value of is neg­a­tive, i.e. 1. may also be con­strued as the level of firm in­ef­fi­ciency. The higher it is (greater ab­so­lute value), the less ef­fi­cient the firm is, i.e. smaller .

4. Em­pir­i­cal Test 4.1 Treat­ment of Other Vari­ables and Data Ex­pla­na­tions

is the la­bor price fac­ing firms. La­bor in­put em­ploys cur­rent-year to­tal amount of payable com­pen­sa­tion. La­bor price equals di­vided by the to­tal num­ber of em­ploy­ees , i.e. av­er­age em­ployee wage. is cap­i­tal price fac­ing firms. This pa­per adopts a vari­a­tion of fixed cap­i­tal sim­i­lar to a vari­a­tion of per­pet­ual in­ven­tory method (Oum and Zhang, 1995), and takes into ac­count de­pre­ci­a­tion fac­tor (Jara-Dı́az et al., 2004). De­ter­mi­na­tion of the above-men­tioned two meth­ods and ref­er­enc­ing Chen and Liu (2014) and Luo and Ni (2015), we have:

Where, is the quan­tity of cap­i­tal fac­tor in­put (con­sump­tion). is to­tal cap­i­tal stock (in­clud­ing fixed cap­i­tal and ac­tive cap­i­tal), mea­sured by the as­set ag­gre­gate in­di­ca­tor in the data­base of in­dus­trial en­ter­prises. is the net value of fixed as­sets, and de­notes the fixed cap­i­tal stock of firms. r is the bench­mark in­ter­est rate for one-year time de­posits at the be­gin­ning of the year re­leased by the Peo­ple’s Bank of China, and de­notes the op­por­tu­nity cost of cap­i­tal. is cur­rent-year de­pre­ci­a­tion8. is the num­ber of years for the de­pre­ci­a­tion of fixed as­sets. Ref­er­enc­ing China’s rel­e­vant ac­count­ing stan­dards,

is the min­i­mum num­ber of years for the de­pre­ci­a­tion of houses and struc­tures, i.e. 20 years. is cur­rent as­sets, and is in­ven­tory. Bench­mark de­posit in­ter­est rate data for es­ti­mat­ing cap­i­tal price is from the of­fi­cial web­site of the Peo­ple’s Bank of China (China’s cen­tral bank).

Firm pro­duc­tion cost . Ac­cord­ing to the cost cat­e­gory of in­dus­trial firms in the NBS In­dus­trial Sta­tis­ti­cal State­ment Sys­tem ( man­u­fac­tur­ing cost, in­ven­tory at the be­gin­ning of year, sales cost, man­age­ment ex­penses, etc., with­out con­sid­er­ing fi­nan­cial cost, etc.), the value is the sum of cap­i­tal in­put , la­bor in­put , cost of pri­mary busi­ness, in­ven­tory, sales cost and man­age­ment cost.

Firm out­put is the phys­i­cal out­put of mu­nic­i­pal firms ex­clud­ing the im­pact of prod­uct price, i.e. prod­uct sales rev­enue of prod­ucts di­vided by the price of pub­lic util­ity prod­ucts in the city. House­hold elec­tric power con­sump­tion of 35 cities in­clud­ing pro­vin­cial cap­i­tals in China Price Year­book of 19992008 (yuan/kWh), nat­u­ral gas (yuan/m³), pipe­line gas (yuan/m³), house­hold do­mes­tic wa­ter (yuan/m³, ex­clud­ing sewage treat­ment cost) and sewage treat­ment cost (yuan/m³) are the prices of power sup­ply, heat sup­ply (nat­u­ral gas price is used due to miss­ing data), gas price, wa­ter sup­ply and sewage treat­ment prices . Since China Price Year­book of 1998 is not yet pub­lished, es­ti­ma­tion is con­ducted us­ing the price of 1999 for the same year de­duct­ing in­fla­tion­ary fac­tor. Due to miss­ing data, ex­cept for big cities such as Shen­zhen, Xi­a­men, Dalian and Qing­dao whose data is pub­lished in China Price Year­book, the price data of other pre­fec­ture-level cities is the prices of pro­vin­cial cap­i­tals. With­out dif­fer­en­ti­a­tion be­tween in­dus­trial and civil prices of mu­nic­i­pal prod­ucts, prod­uct prices cited are civil prices. In 1999, sewage treat­ment cost and mu­nic­i­pal wa­ter sup­ply cost were not yet sep­a­rated, and data of var­i­ous lo­cal­i­ties was miss­ing. Thus, es­ti­ma­tion is car­ried out us­ing the pro­por­tion of do­mes­tic wa­ter and sewage treat­ment cost for the same city in 2000. House­hold wa­ter tar­iff of Shenyang City for 2001-2003 is miss­ing, and re­placed with com­pos­ite wa­ter tar­iff. Data not pub­lished in China Price Year­book of 2007 and 2008 is re­placed with year-end data pro­vided by the Mon­i­tor­ing and Anal­y­sis Di­vi­sion of the Price De­part­ment of the Na­tional De­vel­op­ment and Re­form Com­mis­sion (i.e. data source of China Price Year­book).

Var­i­ous cap­i­tal stocks and de­pre­ci­a­tions are de­flated by the fixed as­set in­vest­ment price in­dex in var­i­ous sta­tis­ti­cal year­books with 1998 as base pe­riod. In ad­di­tion, mone­tary in­dex vari­able is de­flated by the ex-fac­tory price in­dex of in­dus­trial goods in the sta­tis­ti­cal year­books of var­i­ous years. Re­fer to the work­ing pa­per edi­tion for vari­able sta­tis­ti­cal char­ac­ter­is­tics and rel­e­vant anal­y­sis.

This pa­per has 15,000 valid sam­ples. Af­ter PSM, there are still around 2,000 sam­ples in the ex­per­i­ment group and con­trol group, which guar­an­tees sam­ple ca­pac­ity. In 1998, only 171 out of 4,696 firms par­tic­i­pated in mixed own­er­ship re­form, ac­count­ing for only 3.6%. With the ad­vance­ment of mixed own­er­ship re­form, this pro­por­tion kept in­creas­ing year by year. By 2007, the num­ber of

ex­per­i­ment group sam­ples reached 864, ac­count­ing for 18.1% of to­tal sam­ples in the year, which more than dou­bled com­pared with 1998. Ob­vi­ously, China’s eco­nomic re­form was car­ried out with sig­nif­i­cant in­ten­sity dur­ing this pe­riod of time. Mixed own­er­ship re­form was car­ried out with the great­est in­ten­sity for gas sup­ply sec­tor, with the share of ex­per­i­ment group sam­ples in­creas­ing from 13.1% in 1998 to 46.7% in 2007. About half of en­ter­prises be­came mixed own­er­ship en­ter­prises. The ex­per­i­ment group sam­ples of heat sup­ply sec­tor in­creased from 8.2% in 1998 to 35.4% in 2007, rank­ing the sec­ond. Then, the ex­per­i­ment group of sewage treat­ment sec­tor in­creased from 7.7% in 2003 (all mixed own­er­ship en­ter­prises were es­tab­lished in 2003) to 29.6% in 2007. The ex­per­i­ment group of wa­ter sup­ply sec­tor in­creased from 3.4% in 1998 to 13.4% in 2007. Re­form was car­ried out with the least in­ten­sity for power sup­ply sec­tor, whose ex­per­i­ment group sam­ples in­creased from 1.7% in 1998 to only 3.9% in 2007.

Mixed own­er­ship re­form was car­ried out with the high­est in­ten­sity in 2005, in­volv­ing 1,025 en­ter­prises. Af­ter 2005, how­ever, re­form en­coun­tered some re­sis­tance. Dur­ing 2006-2008, sam­ples that car­ried out re­form re­duced to 721, 694 and 764 re­spec­tively. Prior to 2005, the in­ten­sity of re­form in­creased with volatil­ity, up from 260 en­ter­prises in 1999 and 190 en­ter­prises in 2000 to 353 in 2002, 550 in 2003 and 544 in 2004.

In terms of cap­i­tal struc­ture, the an­nual av­er­age share of state cap­i­tal in power sup­ply and wa­ter sup­ply sec­tors both ac­counted for over 80% in 2004 (state cap­i­tal of all sec­tors di­vided by to­tal cap­i­tal); the shares dropped to 91.0% and 94.3% re­spec­tively in 1998, and started to de­crease at a rapid pace af­ter 2004, down to 75.7% and 67.8% re­spec­tively in 2008. Cap­i­tal struc­ture also shows that mixed own­er­ship re­form was car­ried out with greater in­ten­sity for heat and gas sup­ply sec­tors: In 1998, their sec­tor-wide share of state cap­i­tal was 87.6% and 94.3% re­spec­tively. Dur­ing 1998-2003, the shares de­creased by 19.7 and 33.7 per­cent­age points re­spec­tively in the five years from 1998 to 2003. Af­ter 2004, the av­er­age share of state cap­i­tal fur­ther re­duced to 44.6% and 29.9% re­spec­tively in 2008, and con­tin­ued to re­duce by 23.4 and 30.7 per­cent­age points in the fol­low­ing five years.

4.2 Cost Func­tion Re­gres­sion and Core Vari­able Re­port

Re­fer to Ta­ble 1 for the cost func­tion re­gres­sion of equa­tion (10) and equa­tion (14). Econo­met­ric equa­tion has a high goodness-of-fit, and re­gres­sion co­ef­fi­cient is gen­er­ally sig­nif­i­cant. Firm cost can be well fit­ted.

of pub­lic util­ity sec­tors in five cities is all greater than , which sug­gests that its sen­si­tiv­ity to pro­duc­tion cost and la­bor price volatil­ity is not strong, and that cap­i­tal price hike has a ma­jor im­pact on cor­po­rate cost. This shows that mu­nic­i­pal pipe­line con­struc­tion re­quires tremen­dous sunk cost, and in­volves sig­nif­i­cant economies of scope. Thus, the pub­lic util­ity sec­tors of five cities are all typ­i­cal cap­i­tal-in­ten­sive sec­tors.

Us­ing re­gres­sion co­ef­fi­cient of Ta­ble 1, we ob­tain the spe­cific form of the cost func­tion of ur­ban pub­lic util­ity sec­tors in five cities, so as to cal­cu­late dif­fer­en­tial vari­able us­ing equa­tion (11), equa­tion (12) and equa­tion (15) re­spec­tively. Re­sults of cal­cu­la­tion are shown in the work­ing pa­per edi­tion.

As can be seen from the tem­po­ral trend of , the num­ber of ur­ban pub­lic util­ity sec­tors with nat­u­ral monopoly shows a down­ward trend, down from 569 in 1998 to 462 in 2008. This is con­sis­tent with the main­stream views of academia: With the im­prove­ment of mar­ket-based eco­nomic sys­tem, pro­duc­tion sec­tors with nat­u­ral monopoly will re­duce, and mar­ket com­pet­i­tive­ness will in­crease grad­u­ally (Qie, 2002; Yu and Yu, 2004). In terms of sec­tor dis­tri­bu­tion, power sup­ply sec­tor had the high­est ra­tio of nat­u­ral monopoly, i. e. 100%, and sewage treat­ment sec­tor had the low­est ra­tio of monopoly, i.e. 13%. The ra­tio for wa­ter sup­ply sec­tor is the sec­ond-low­est, i.e. 27%, and heat and gas sup­ply sec­tors had medium ra­tios of nat­u­ral monopoly, i.e. 60% and 51% re­spec­tively. The pro­por­tions of nat­u­ral monopoly sam­ples for the lat­ter four sec­tors re­duced over time.

This pa­per ob­tains the nat­u­ral monopoly group­ing vari­able based on the cri­te­rion of

whether or not firm sam­ples are in sec­tors of nat­u­ral monopoly. Re­sult shows that the nat­u­ral monopoly sam­ples have 0 sim­i­lar dis­tri­bu­tion ra­tios at the level of ur­ban sec­tors. The level of nat­u­ral monopoly di­min­ishes in the or­der of the sec­tors of power sup­ply, heat sup­ply, gas sup­ply, wa­ter sup­ply and sewage treat­ment.

Among the 10,425 ur­ban pub­lic util­ity sec­tor sam­ples dur­ing 1998-2008, 6,003 are sec­tors of nat­u­ral monopoly, but power sup­ply sec­tor ac­counts for the big­gest share of nat­u­ral monopoly. The nat­u­ral monopoly at­tribute is in­signif­i­cant for ur­ban pub­lic util­ity sec­tors in­clud­ing heat sup­ply, gas sup­ply, wa­ter sup­ply and sewage sup­ply sec­tors. For these sec­tors, more cities are of non-nat­u­ral monopoly, but cities of nat­u­ral monopoly also ex­ist.

Mu­nic­i­pal pub­lic ser­vices of dif­fer­ent cities and sec­tors have dif­fer­ent nat­u­ral monopoly at­tributes. There­fore, sec­tor man­age­ment must be car­ried out ac­cord­ing to the spe­cific con­di­tions of var­i­ous lo­cal­i­ties and sec­tors. Over­look­ing the nat­u­ral monopoly at­tributes of spe­cific pub­lic util­i­ties sec­tors

will harm the ef­fi­ciency of their reg­u­la­tion and man­age­ment and jeop­ar­dize rel­e­vant eco­nomic re­forms.

Af­ter es­ti­mat­ing the re­gres­sion co­ef­fi­cient of cost func­tion , vari­ables of firms such as are sub­sti­tuted us­ing equa­tions ( 12) and ( 15) to cal­cu­late at the firm level, which is shown in Fig­ure 1. The “semi-para­met­ric method TFP” se­quence in Ta­ble 1 is the an­nual mean value of

. The “semi-para­met­ric method TFP growth” is the mean value of growth rate, and “TFP growth rate” is the mean value of .

TFP growth rate of firms in ur­ban pub­lic util­i­ties sec­tors re­duced at first and then in­creased. This find­ing is con­sis­tent with the man­u­fac­turer es­ti­ma­tion re­sult of Yang (2015). In gen­eral, the growth rates of es­ti­mated us­ing cost func­tion with vari­able re­turn to scale and es­ti­mated us­ing cost func­tion with con­stant re­turn to scale share a con­sis­tent time trend. From the per­spec­tive of TFP es­ti­ma­tion, this study is an in­no­va­tive at­tempt to c om­pletely ex­clude prod­uct price ef­fect.

4.3 DID Re­gres­sion Re­sult

Ac­cord­ing to DID econo­met­ric mod­els of equa­tions (1) and (2), Ta­ble 2 pro­vides the test re­sult of the cor­re­la­tion be­tween mixed own­er­ship re­form and TFP. Line 3 of Ta­ble 2 shows that the re­gres­sion re­sult es­ti­mated us­ing DID method is not ro­bust. The model re­gres­sion re­sult with 2002- 2005 as ob­ser­va­tion pe­riod shows that mixed own­er­ship re­form has a cer­tain pos­i­tive ef­fect on ex­per­i­ment group sam­ples, but such an ef­fect is in­signif­i­cant (such as Col­umns 2-3). In Model DID( 2) with to­tal sam­ple par­tic­i­pa­tion, the pol­icy ef­fect of mixed own­er­ship re­form is not ro­bust as well, and can be neg­a­tive.

Causal­ity be­tween mixed own­er­ship re­form and TFP is sta­tis­ti­cally in­signif­i­cant. There­fore, en­ter­prises hav­ing com­pleted mixed own­er­ship re­form may not nec­es­sar­ily im­prove their pro­duc­tion ef­fi­ciency. This find­ing is con­sis­tent with Hy­poth­e­sis 2 that with­out dif­fer­en­ti­a­tion be­tween sec­tors of nat­u­ral monopoly and com­pet­i­tive sec­tors, mixed own­er­ship re­form will not be able to sig­nif­i­cantly im­prove cor­po­rate pro­duc­tion ef­fi­ciency. In the real sense, in­dis­crim­i­nate mixed own­er­ship re­form for ur­ban pub­lic util­ity sec­tors ir­re­spec­tive of nat­u­ral monopoly at­tribute is sub­ject to un­cer­tain in­sti­tu­tional per­for­mance. At least, it is not good for so­cial pro­duc­tion ef­fi­ciency.

Hence, Hy­poth­e­sis 2 is ver­i­fied. The fun­da­men­tal rea­son for Hy­poth­e­sis 2 to hold true is that the group­ing vari­able of nat­u­ral monopoly has a crit­i­cal ef­fect on re­gres­sion as a whole. Only by sep­a­rat­ing the nat­u­ral monopoly at­tribute and com­pet­i­tive­ness of sec­tors will this pa­per be able to ac­cu­rately cap­ture the pol­icy ef­fect of mixed own­er­ship re­form for ur­ban pub­lic util­ity sec­tors.

4.4 DDD Re­gres­sion Re­sult

Ac­cord­ing to the DDD re­gres­sion re­sult in Ta­ble 3, DDD re­gres­sion has im­proved the sig­nif­i­cance

of some ex­plana­tory vari­ables, and the re­gres­sion goodness- of- fit has also im­proved to some ex­tent. This study is most con­cerned with DDD es­ti­ma­tor . 1 is the ur­ban pub­lic util­ity en­ter­prise of non-nat­u­ral monopoly which par­tic­i­pates in mixed own­er­ship re­form, and its re­gres­sion co­ef­fi­cient is the pol­icy ef­fect of mixed own­er­ship re­form in ur­ban pub­lic util­ity sec­tors of non­nat­u­ral monopoly.

Na­tional monopoly has a sig­nif­i­cant in­flu­ence on cor­po­rate TFP. First, in Model DDD( 1), once the nat­u­ral monopoly at­tribute is not con­sid­ered, the ag­gre­gate pro­duc­tiv­ity im­prove­ment ef­fect for the ex­per­i­ment group sam­ples is rel­a­tive to ur­ban pub­lic util­ity en­ter­prises which have not par­tic­i­pated in the mixed own­er­ship re­form, i.e. re­form will lead to ap­prox­i­mately an ad­di­tional -0.01 unit of TFP growth rate and about 0.24 units of TFP growth for ex­per­i­ment group sam­ples. Sec­ond,

’s re­gres­sion co­ef­fi­cient is sig­nif­i­cantly pos­i­tive. Namely, a city’s pub­lic util­i­ties sec­tor be­ing a com­pet­i­tive sec­tor of non­nat­u­ral monopoly will sig­nif­i­cantly in­crease the city’s pro­duc­tiv­ity. Third, with the im­prove­ment of mar­ket eco­nomic sys­tem, the ef­fi­ciency losses caused by nat­u­ral monopoly will di­min­ish, as ’s re­gres­sion co­ef­fi­cient is sig­nif­i­cantly neg­a­tive. It can be learned based on that nat­u­ral monopoly dur­ing 1998-2001 would cause 0.07 units of TFP growth loss of en­ter­prises and about 0.27 units of TFP loss. Dur­ing 2006-2008, such losses re­duced to about 0.05 and 0.15 units re­spec­tively.

Based on the above anal­y­sis and the re­sult that DID es­ti­ma­tor is rel­a­tively in­signif­i­cant and not ro­bust, this pa­per con­sid­ers that Hy­poth­e­sis 1 is ver­i­fied: In ur­ban pub­lic util­ity sec­tors of nat­u­ral monopoly, mixed own­er­ship re­form can­not sig­nif­i­cantly in­crease cor­po­rate pro­duc­tiv­ity.

Con­sis­tent with Hy­poth­e­sis 3, Ta­ble 3 shows that is all pos­i­tive value and gen­er­ally sig­nif­i­cant. The im­pli­ca­tion is that China’s mixed own­er­ship re­form dur­ing 1999-2008 has ex­erted a sig­nif­i­cant pos­i­tive ef­fect on com­pet­i­tive ur­ban pub­lic util­i­ties sec­tors of non­nat­u­ral monopoly and in­creased the TFP of mu­nic­i­pal ser­vice en­ter­prises.

’s ag­gre­gate ef­fect on TFP is . Nat­u­ral monopoly will cause ad­di­tional 0.12 units of TFP growth rate of ex­per­i­ment group sam­ples dur­ing 2006-2008 and about 0.21 units of TFP losses. In com­pet­i­tive sec­tors free from nat­u­ral monopoly, mixed own­er­ship re­form has a sig­nif­i­cantly pos­i­tive ef­fect on the pro­duc­tiv­ity of mu­nic­i­pal pub­lic util­ity en­ter­prises.

ob­tained based on Model DDD( 2) is also sig­nif­i­cantly pos­i­tive. Thus, Hy­poth­e­sis 3 is proven. Rel­a­tive to sec­tors of nat­u­ral monopoly, mixed own­er­ship re­form has a greater im­prove­ment ef­fect on TFP and its growth rate in com­pet­i­tive sec­tors ( 0). In sec­tors of nat­u­ral monopoly, mixed own­er­ship re­form led to about - 0.07 units of TFP growth rate and about 0.04 units of TFP growth ( ). In com­pet­i­tive sec­tors, mixed own­er­ship re­form brought about ap­prox­i­mately 0.05 units of TFP growth rate and about 0.25 units of TFP growth (

).

There­fore, the pos­i­tive ef­fect of mixed own­er­ship re­form on cor­po­rate pro­duc­tiv­ity mainly oc­curred un­der a com­pet­i­tive mar­ket struc­ture rather than in sec­tors of nat­u­ral monopoly. In other words, if com­pet­i­tive sec­tors carry out mixed own­er­ship re­form, the pro­duc­tiv­ity of ur­ban pub­lic util­ity en­ter­prises will be greatly in­creased, un­leash­ing more “pol­icy div­i­dends” of in­sti­tu­tional re­form.

Re­fer to Ta­ble 4 for the re­gres­sion re­sult of DDD model of panel data. Ir­re­spec­tive of whether or not the fixed ef­fect of pair­wise com­bi­na­tion is con­trolled for, the re­gres­sion co­ef­fi­cient of is all pos­i­tive, i.e. mixed own­er­ship re­form will in­crease pro­duc­tiv­ity in com­pet­i­tive sec­tors. Rel­a­tive to the im­prove­ment ef­fect of TFP , the im­prove­ment ef­fect of its growth rate is not very sig­nif­i­cant. Re­fer to Col­umn 2-3 of Row 3, Ta­ble 4.

5. Ro­bust­ness Test 5.1 Fur­ther Causal­ity Iden­ti­fi­ca­tion and Propen­sity In­dex Match­ing Method

The gov­ern­ment will take into ac­count the size and tax bur­den of en­ter­prises in se­lect­ing the pri­or­ity targets of re­form. While the for­mer in­volves re­form dif­fi­culty and so­cial sta­bil­ity is­sues, the lat­ter is a mat­ter of re­form in­tent. There­fore, this pa­per iden­ti­fies phys­i­cal out­put, in­dus­trial val­ueadded, cost of pri­mary busi­ness, payable in­come tax, payable VAT, gov­ern­ment sub­sidy and amount of losses as char­ac­ter­is­tic vari­ables9 to con­duct core match­ing for the Gaus­sian ker­nel func­tions of ex­per­i­ment group and con­trol group, with re­sults shown in the work­ing pa­per edi­tion. Af­ter PSM treat­ment, all char­ac­ter­is­tic vari­ables and stan­dard­iza­tion er­rors are smaller than 5% for the new con­trol group and ex­per­i­ment group sam­ples, and t test re­sult ba­si­cally does not re­ject the null hy­poth­e­sis that the ex­per­i­ment group and con­trol group have no sys­tem­atic dif­fer­ences. This im­plies that af­ter PSM treat­ment, sam­ple het­ero­gene­ity of dif­fer­ent groups has sub­stan­tially re­duced, which is of sig­nif­i­cant help to es­tab­lish­ing the “coun­ter­fac­tual frame­work.”

Af­ter con­duct­ing a more strin­gent causal­ity iden­ti­fi­ca­tion through PSM, Ta­ble 3 and Ta­ble 4 list rel­e­vant re­sults. PSM- DDD( 1)’s re­gres­sion re­sult is gen­er­ally con­sis­tent with pre­vi­ous re­sult. In ad­di­tion, the ag­gre­gate ef­fect of on TFP growth rate be­comes more ev­i­dent (

), i.e. ap­prox­i­mately 0.17 units of TFP growth loss. Com­pared with DDD( 2), de­spite a re­duc­tion in the ob­ser­va­tions of PSM-DDD( 2), the t statis­tic of its DDD es­ti­ma­tor re­gres­sion co­ef­fi­cient fur­ther in­creased, and the pol­icy ef­fect of mixed own­er­ship re­form is more sig­nif­i­cant.

Based on causal­ity deriva­tion of equa­tion (18) and ro­bust­ness test, this pa­per be­lieves there is sta­tis­ti­cal causal­ity among mixed own­er­ship re­form, nat­u­ral monopoly and the TFP of ur­ban pub­lic util­ity en­ter­prises con­sis­tent with Hy­pothe­ses 1-3.

5.2 Change of Ob­ser­va­tion Pe­riod

In or­der to fur­ther test ro­bust­ness, this pa­per short­ens (pro­longs) the ob­ser­va­tion pe­riod of re­form into 2003-2004 (2001-2006), and con­ducts a test us­ing DDD( 2) model. The test re­sult is still gen­er­ally con­sis­tent with the above, with the re­sult shown in work­ing pa­per edi­tion.

5.3 Ex­am­i­na­tion of Re­form Per­for­mance on Other Di­men­sions

Di­men­sions for mea­sur­ing pol­icy gains are var­ied, since the same pol­icy may have dif­fer­ent di­men­sions of “pol­icy div­i­dends.” There­fore, this pa­per se­lects dif­fer­ent op­er­a­tional per­for­mance in­di­ca­tors as cri­te­ria for eval­u­at­ing the re­sults of mixed own­er­ship re­form. Ref­er­enc­ing the def­i­ni­tion of Sheng and Liu (2016) on SOE per­for­mance, this pa­per se­lects sales profit mar­gin (prof­its di­vided by in­come from pri­mary busi­ness), share of man­age­ment ex­penses (man­age­ment ex­penses di­vided by in­come from pri­mary busi­ness), la­bor pro­duc­tiv­ity ( phys­i­cal out­put di­vided by the num­ber of em­ploy­ees), cap­i­tal pro­duc­tiv­ity ( phys­i­cal out­put di­vided by to­tal cap­i­tal stock), as well as other cor­po­rate per­for­mance in­di­ca­tors as ex­plained vari­ables to ex­am­ine the cor­po­rate busi­ness im­prove­ment ef­fect of mixed own­er­ship re­form.

Re­gres­sion re­sults show that un­der the ef­fect of nat­u­ral monopoly, mixed own­er­ship re­form’s cor­po­rate per­for­mance im­prove­ment ef­fect is sig­nif­i­cantly dif­fer­en­ti­ated for var­i­ous ur­ban pub­lic util­ity sec­tors. Rel­a­tive to sec­tors of nat­u­ral monopoly, mixed own­er­ship re­form in a com­pet­i­tive mar­ket struc­ture plays a big­ger role in im­prove­ment of cor­po­rate prof­itabil­ity, share of man­age­ment ex­penses, la­bor pro­duc­tiv­ity and cap­i­tal pro­duc­tiv­ity. Thus, it can be learned that the pos­i­tive ef­fect of mixed

own­er­ship re­form on cor­po­rate per­for­mance oc­curs pri­mar­ily un­der a com­pet­i­tive mar­ket struc­ture, rather than in sec­tors of nat­u­ral monopoly.

Ex­cept for the share of man­age­ment ex­penses, the per­for­mance of DDD( 1) model with sales prof­itabil­ity, la­bor pro­duc­tiv­ity and cap­i­tal pro­duc­tiv­ity as ex­plained vari­ables is not ro­bust. This im­plies that re­form may not di­rectly im­prove fi­nan­cial in­di­ca­tors such as cor­po­rate la­bor pro­duc­tiv­ity, cap­i­tal pro­duc­tiv­ity and sales prof­itabil­ity. In­stead, re­form should re­duce re­dun­dant man­age­ment ex­penses and im­prove cor­po­rate gov­er­nance struc­ture to in­di­rectly im­prove pro­duc­tiv­ity. Re­gres­sion re­sult 0 shows that mixed own­er­ship re­form will in­crease the share of man­age­ment ex­penses for sam­ples of nat­u­ral monopoly sec­tors. It can be de­duced that in less com­pet­i­tive sec­tors of nat­u­ral monopoly, mixed own­er­ship re­form will not lead to any im­prove­ment in cor­po­rate gov­er­nance struc­ture.

6. Con­clud­ing Re­marks

“Cross­ing the river by feel­ing your feet on the stones” is a con­cise sum­mary of China’s in­sti­tu­tional trans­for­ma­tion since re­form and open­ing-up. From the na­tion­al­iza­tion and planned eco­nomic sys­tem at the be­gin­ning of the found­ing of the Peo­ple’s Repub­lic of China in 1949 to the house­hold con­tract re­spon­si­bil­ity sys­tem at the in­cep­tion of re­form and open­ing-up and the cre­ation of mar­ket eco­nomic sys­tem in the 1980s, China’s in­sti­tu­tional trans­for­ma­tion has al­ways been in­ter­mit­tent. From the per­spec­tive of in­sti­tu­tional change, such an in­ter­mit­tent in­sti­tu­tional change is fraught with more sig­nif­i­cant un­cer­tain­ties of eco­nomic per­for­mance com­pared with a grad­u­al­ist in­sti­tu­tional change. Per­for­mance change aris­ing from abrupt in­sti­tu­tional change is of­ten more sur­pris­ing, but this seems to have be­come key to the suc­cess of China’s rapid growth over three decades. In its re­form en­deav­ors never ex­per­i­mented be­fore, China has ex­plored a unique path of its own. While the old sys­tem in­creased in size, each in­di­vid­ual in­sti­tu­tional re­form ad­dress­ing spe­cific is­sues nat­u­rally had di­min­ish­ing mar­ginal re­turn.

For ur­ban pub­lic util­ity sec­tors and sec­tors of nat­u­ral monopoly, the Third Plenum of the 18th CPC Cen­tral Com­mit­tee at­tempted to in­tro­duce a fun­da­men­tal change and de­signed a com­plete new sys­tem. State cap­i­tal must main­tain con­trol­ling shares in sec­tors of nat­u­ral monopoly. Their mixed own­er­ship re­form must be car­ried out with cau­tion. The pri­or­ity targets of mixed own­er­ship re­form should be com­pet­i­tive sec­tors of non­nat­u­ral monopoly. How­ever, since the propo­si­tion of mixed own­er­ship re­form in the 1980s, aca­demic dis­cus­sions have long fo­cused on the ne­ces­sity of re­form with­out ad­dress­ing the ques­tion as how re­form should be car­ried out. This pa­per aims to ad­dress the ques­tion as how mixed own­er­ship re­form should be car­ried out in sec­tors of nat­u­ral monopoly - an im­por­tant the­o­ret­i­cal and re­al­is­tic ques­tion fac­ing China’s cur­rent in­sti­tu­tional re­form, with a new ap­proach of dis­course that own­er­ship re­form can­not be car­ried out for own­er­ship re­form’s sake.

On the other hand, monopoly and com­pe­ti­tion have al­ways been clas­si­cal top­ics of dis­cus­sion in eco­nom­ics. For nat­u­ral monopoly as a spe­cial form of monopoly, mixed own­er­ship re­form should be well in­cluded into the frame­work of nat­u­ral monopoly the­ory. This pa­per marks such an at­tempt.

Our key find­ings are as fol­lows: 1) In ur­ban pub­lic util­i­ties sec­tors of nat­u­ral monopoly, mixed own­er­ship re­form can­not sig­nif­i­cantly in­crease cor­po­rate pro­duc­tiv­ity; 2) with­out dif­fer­en­ti­a­tion be­tween nat­u­ral monopoly and com­pet­i­tive sec­tors, tra­di­tional DID econo­met­ric model will not be able to test the sig­nif­i­cant ef­fects of mixed own­er­ship re­form, which means that in­dis­crim­i­nate im­ple­men­ta­tion of mixed own­er­ship re­form is likely to be fraught with pol­icy un­cer­tain­ties; 3) in com­pet­i­tive sec­tors of non­nat­u­ral monopoly, mixed own­er­ship re­form has a sig­nif­i­cant pos­i­tive ef­fect on the pro­duc­tiv­ity of mu­nic­i­pal pub­lic util­ity en­ter­prises. Rel­a­tive to sec­tors of nat­u­ral monopoly, the pro­duc­tiv­ity of com­pet­i­tive sec­tors of the mixed own­er­ship re­form will be in­creased much more sig­nif­i­cantly with greater “pol­icy div­i­dends” of in­sti­tu­tional re­form. There­fore, the mixed own­er­ship re­form of ur­ban pub­lic util­ity sec­tors

should be car­ried out first in com­pet­i­tive sec­tors of non­nat­u­ral monopoly.

De­ter­min­ing the nat­u­ral monopoly at­tribute of spe­cific pub­lic util­i­ties sec­tors in spe­cific cities is key to the im­ple­men­ta­tion of mixed own­er­ship re­form. The ac­cu­racy and open­ness of cost data, which is the ba­sis for such de­ter­mi­na­tion, are vi­tally im­por­tant. Hence, this pa­per puts for­ward the fol­low­ing pol­icy rec­om­men­da­tions: The cost data of ur­ban pub­lic util­ity en­ter­prises must be made open to gov­ern­ment depart­ments. Such open­ness should be fa­cil­i­tated by re­cent leg­isla­tive work. The de­part­men­tal rule Ad­min­is­tra­tive Mea­sures for In­fra­struc­ture and Pub­lic Util­ity Li­censed Op­er­a­tions en­acted in 2015 stip­u­lates that rel­e­vant in­for­ma­tion such as audited fi­nan­cial state­ments of the pre­vi­ous year should be made pub­lic in ac­cor­dance with rel­e­vant pro­vi­sions. This leg­isla­tive ap­proach is sim­i­lar to the manda­tory in­for­ma­tion dis­clo­sure clauses of price reg­u­la­tion agree­ment in the Price Reg­u­la­tion Agree­ment ex­e­cuted by the Gov­ern­ment of Hong Kong Spe­cial Ad­min­is­tra­tive Re­gion with fuel gas sup­ply en­ter­prises over the years (“In­for­ma­tion and Con­sul­ta­tion Agree­ment”), which is a fairly ad­vanced gov­ern­ment reg­u­la­tory rule. This pa­per con­sid­ers that the Reg­u­la­tions on the Li­censed Op­er­a­tions of Mu­nic­i­pal Pub­lic Util­i­ties cur­rently un­der the State Coun­cil’s leg­isla­tive process should put greater premium on the open­ness of cost data.

In the fi­nal anal­y­sis, mixed own­er­ship re­form can­not be car­ried out in­dis­crim­i­nately for pub­lic util­ity sec­tors which are of great rel­e­vance to pub­lic in­ter­est and wel­fare. Rel­e­vant in­sti­tu­tional re­forms should fol­low the “top-level de­sign” ap­proach of the Third Plenum of the 18th CPC Cen­tral Com­mit­tee. Sim­i­larly, mixed own­er­ship re­form in other sec­tors should also be car­ried out in other sec­tors with spe­cific pri­or­i­ties.

Nev­er­the­less, this pa­per still has many de­fi­cien­cies. For in­stance, such pub­lic util­ity sec­tors as waste treat­ment and land­scap­ing are not brought un­der anal­y­sis due to lack of data. Data of 2009-2013 is not brought un­der anal­y­sis in this study due to poor qual­ity of data. This pa­per is ex­pected to serve as an ini­tial study to in­spire sup­ple­men­tary stud­ies in the fu­ture.

Newspapers in English

Newspapers from China

© PressReader. All rights reserved.