Investment and apital struture of partially private regulated rms Carlo Cambini y and Yossi Spiegel z Jauary 21, 2014 Abstrat We develop a model that examines the apital struture and investment deisions of regulated rms in a setting that inorporates two key institutional features of the publi utilities setor in many ountries: rms are partially owned by the state and regulators are not neessarily independent. Among other things, we show that regulated rms issue more debt, invest more, and enjoy higher regulated pries when they fae more independent regulators, are more privatized, and when regulators are more pro- rm. Moreover, regulatory independene, higher degree of privatization, and pro- rm regulatory limate are assoiated with higher soial welfare. JEL Classi ation: G32, L33, L51 We thank seminar partiipants at the 2009 EARIE Conferene in Ljubljana, the 2010 International Conferene on Infrastruture Eonomis and Development in Toulouse, the 2011 Industrial Organization: Theory, Empiris and Experiments workshop in Otranto, the 2011 CRESSE Conferene in Rhodes, the 2012 Conferene on the Eonomis of the Publi-Private Partnerships in Barelona, FEEM in Milano, and IAE in Paris. Yossi Spiegel thanks the Henry Crown Institute of Business Researh in Israel for nanial assistane and Carlo Cambini gratefully aknowledges nanial support from the Italian Ministry of Eduation (No. 20089PYFHY_004). y Politenio di Torino and EUI - Florene Shool of Regulation. Address: Politenio di Torino, DISPEA, Corso Dua degli Abruzzi, 24, 10129 Torino, Italy. email: arlo.ambinipolito.it. z Tel Aviv University, CEPR, and ZEW. Address: Reanati Graduate Shool of Business Administration, Tel Aviv University, Ramat Aviv, Tel Aviv, 69978, Israel. email: spiegelpost.tau.a.il, http://www.tau.a.il/~spiegel 1
Keywords: regulation, debt, investment, government ownership, regulatory independene, regulatory limate 2
1 Introdution Sine the early 1990 s, many ountries around the world have substantially reformed their publi utilities setor through large sale privatization and by establishing Independent Regulatory Agenies (IRAs) to regulate the newly privatized utilities. These reforms were intended to improve the e ieny and servie quality of utilities and boost their investments. The strutural reforms, however, were aompanied by a substantial inrease in the nanial leverage of regulated utilities. 1 This trend, oined the dash for debt, is widespread aross ountries and aross setors and has raised substantial onerns among poliy markers. For instane, a joint study of the UK Department of Trade and Industry (DTI) and the HM Treasury argues that the dash for debt within the UK utilities setor from the mid-late 1990 s ould imply greater risks of nanial distress, transferring risk to onsumers and taxpayers and threatening the future naneability of investment requirements (DTI and HM Treasury, 2004, p. 6). Likewise, the Italian energy regulatory ageny, AEEG, has reently expressed its onern that exessive nanial leverage ould lead to nanial distresses whih in turn ould ause servie interruptions (AEEG 2008, paragraph 22.13). The AEEG has also announed its intention to start monitoring the nanial leverage of Italian energy utilities in order to disourage speulative behavior that might jeopardize their nanial stability (see AEEG, 2007, paragraph 17.40 and AEEG, 2009, paragraph 11.8). To put the onerns about the dash for debt phenomenon in perspetive, it is worth noting that the investments of publi utilities in infrastruture aount for a signi ant fration of GDP. For example, Table 1 in the Appendix shows that in the EU14 states, the average rate of gross xed apital formation in the energy setor (eletriity and gas), teleommuniations, water supply, and transportation, was 15:24% of GDP in 2008. Given the sheer size of investments at stake and the overall importane of the publi utilities setor for the eonomy at large, it is learly important to understand the determinants of the investments and nanial deisions of regulated rms and study how these deisions a et soial welfare. 1 See Bortolotti et al. (2011) for evidene on the EU14 states and Da Silva et al. (2006) for evidene on Latin Ameria and Asia. 3
Earlier literature on this topi (e.g., Taggart 1981 and 1985; Dasgupta and Nanda, 1993; Spiegel and Spulber, 1994 and 1997; and Spiegel, 1994 and 1996) has shown that regulated rms may have an inentive to strategially issue debt in order to indue regulators to set a relatively high prie in order to minimize the risk that the rm will beome nanially distressed. 2 This literature however impliitly assumed that the regulated rm is privately owned and regulators are independent. 3 While these assumptions re et the institutional setting in the U.S. and more reently in the UK, in many other ountries around the world, inluding the EU, Latin Ameria, and Asia, entral or loal governments still hold signi ant ownership stakes (often ontrolling stakes) in many publi utilities (see e.g., Bortolotti and Faio, 2008; Boubakri and Cosset, 1998; and Boubakri et al., 2004), and IRAs do not exist in all setors. 4 Indeed, the large sale privatization proess that started in the 1990 s seems to have led to a new form of state apitalism, whereby governments hoose to remain partial owners of large rms, rather than try to ompletely sell their stakes to private investors (The Eonomist, 2012). 5 This phenomenon is widespread for example in Europe, where many large teleoms and energy utilities are partially held by the state, as well as in emerging markets. 6 The purpose of this paper is to develop a tratable model that will allow 2 Jaimison and Sappington (2013) examine how regulators an prevent exessive leverage by imposing penalties on the rm should it experiene nanial distress. 3 Moreover, with the exeption of Spiegel (1994), this literature has only onsidered the interation between apital struture and regulated pries, holding the rm s invetment level onstant. 4 For instane, fully or partially state-owned enterprises in the OECD area are valued at over 2 trillion USD and employ over 6 million people. Half of these rms by value operate in the network industries (teleoms, eletriity and gas, transportation and postal servies). See http://www.oed.org/daf/a/oeddatasetonthesizeandompositionofnationalstateownedenterprisesetors.htm As for regulation, IRAs were established and are fully operational in the EU only in the teleommuniations and energy setors, but in other setors, like transportation and water, most utilities are still regulated diretly by ministries, governmental ommittees, or loal governments (see Bortolotti et al. 2011). 5 Aording to this report, the ombined market value of state owned ompanies is over 2 trillion USD and total employment is around 6 million. 6 To illustrate, as of the end of 2013, Frane Teleom-Orange is 23:2% held by the Frenh Government, Deutshe Telekom is 31:9% held by the German Government, TeliaSonera is 37% held by the Swedish Government and 13:2% by the Finnish Government, and Telekom Austria is 28% held by the Austrian Government. Likewise, in the energy setor, the Frenh Government holds a 84:5% stake in EDF, while the 4
us to study how (partial) state ownership and regulatory independene a et the apital struture and investments of the regulated rm, regulated pries, and welfare. Our model onsiders the strategi interation between the managers of a regulated rm, who need to deide how muh to invest and how to nane this investment, and a regulator, who needs to set the regulated prie. A main assumption in our model is that the rm s ost is subjet to random shoks. Hene, when the rm is leveraged, a su iently negative ost shok may result in a ostly nanial distress. The regulator therefore faes a trade o between setting a low prie, whih bene ts onsumers, and a high prie, whih minimizes the probability of nanial distress. There is no general agreement in the literature on how to model the objetive of the management of a partially state owned rm. We follow two main strands in the literature and assume the rm s management hooses the rm s ations to maximize the expeted pro t of the rm, but in doing so, the management disounts to some extent the rm s osts. Aording to Sappington and Sidak (2003, 2004), the managers of partially state-owned rms are onerned not only with pro t, but also with revenue, and the weight assigned to revenue inreases with the state s stake in the rm. As a result, the rm s managers e etively disount the rm s ost, and more so when the state s stake in the rm is large. 7 Alternatively, aording to the soft budget onstraint approah, partially state owned rms are more likely to be bailed out by the state in ase of nanial distress, espeially when the state s stake in the rm is large. Hene, the ost of nanial distress from the perspetive of the managers is dereasing with the state s stake in the rm. Sine the regulator sets a regulated prie that takes the rm s objetive into aount, he sets a higher prie when the rm is more privatized and internalizes a larger fration of its ost. The higher regulated Italian Government holds a 32% stake in Enel and 30% stake in Eni, and the Austrian Government holds a 70% stake in Verbund. Aording to the Eonomist report on state apitalism (the Eonomist 2012), the share of national/state-ontrolled ompanies in the MSCI emerging-market index is over 65% in energy, around 55% in utilities, and around 35% in teleommuniation servies. 7 If managers maximize a weighted average of revenue and pro t, then they maximize the expression R + (1 ) (R C), where R is revenue, C is ost, and is the state s stake in the rm. This expression is equivalent to R (1 ) C, so the managers disount the rm s ost to a larger extent when the rm is less privatized (i.e., is high). 5
prie, in turn, allows the rm to issue more debt and indues it to inrease its investment. To model regulatory independene, we follow the literature on entral bank independene (see e.g., Cukierman, 1992) and assume that more independent regulators are more ommitted to the regulatory rule used to determine the regulated prie, while less independent regulators are more likely to behave opportunistially and deviate from their preannouned regulatory rule. Consequently, more regulatory independene leads to a higher regulated prie and hene indues the rm to issue more debt and raise its investment level. Altogether then, our model implies that regulated rms that fae more independent regulators and are more privatized will be more leveraged, will invest more, and will enjoy higher regulated pries. In addition, our results show that higher degrees of regulatory independene and privatization, as well as more pro- rm regulatory limate (the regulator assigns more weight the rm s payo in setting the regulated prie), are all assoiated with higher soial welfare. These results suggest that the dash for debt phenomenon mentioned above is a natural outome of the privatization proess and the establishment of IRAs, and moreover, these proesses are welfare improving. The rest of the paper is organized as follows. Setion 2 presents the model. Setion 3 haraterizes the equilibrium regulated prie for given ombinations of debt and investment. In Setions 4 and 5, we solve for the equilibrium hoie of apital struture and investment and study how these hoies are a eted by the main exogenous parameters of the model, namely the degree regulatory independene, the extent of privatization (i.e., the state s stake in the regulated rm), and the regulatory limate. In Setion 5, we onsider the rm s investment deision and study how it is a eted by the main exogenous parameters of the model. In Setion 6, we examine the impliations of our model for soial welfare. Conluding remarks are in Setion 7. All proofs are in the Appendix. 2 The model Consider a regulated rm, whih is partially owned by the state (at the national or the loal level). For simpliity (but without a serious loss of insights), we assume that the rm faes a unit demand funtion. The willingness of onsumers to pay depends on the rm s 6
investment, k, and is given by a twie di erentiable, inreasing, and onave funtion V (k). That is, k an be interpreted as investment in the quality of the rm s servies. Using p to denote the regulated prie, onsumers surplus is given by V (k) p. 2.1 The apital struture of the rm and its expeted ost The rm s ost of prodution is subjet to random ost shoks (e.g., utuating energy pries) and is given by a random variable,, distributed uniformly over the interval [0; ], where < V (0). Let D denote the fae value of the rm s debt, whih the rm needs to over from its operating inome p. If the rm annot pay D in full, it inurs a xed ost T due to nanial distress. 8 Using (p; D) to denote the probability of nanial distress, the total expeted ost of the rm is C + (p; D) T; 2 where 8 >< 0 D + p; (p; D) >: 1 p D D p < + D; 1 p < D: (1) Intuitively, when D + p, the rm an always pay D in full so (p; D) 0. On the other hand, when p < D, the rm annot pay D in full even when 0, so (p; D) 1. For intermediate ases, (p; D) 1 p D. Obviously, (p; D) is (weakly) inreasing with D and (weakly) dereasing with p: the rm is more likely to beome nanially distressed when its debt is high and the regulated prie is low. 2.2 The regulated rm s objetive Let denote the state s stake in the rm s equity. As mentioned in the Introdution, there is no generally agreed upon way to model the e et of on the objetive of the rm s manage- 8 Finanial distress does not neessarily mean formal bankrupty: it ould refer to any nanial problem that the rm may fae when it annot pay its debt in full and needs to reorganize it. For instane, nanial distress may make it harder for the rm to deal with ustomers and suppliers and raise apital for investment, and it also diverts managerial attention away from normal operations. 7
ment. Our modeling approah follows two main strands in the literature: the manageriallyoriented publi enterprise (MPE) approah, due to Sappington and Sidak (2003, 2004), and the soft-budget onstraint (SBC) approah introdued by Kornai (1986). Aording to the MPE approah, the managers of the (partially) state-owned rm are onerned not only with the rm s pro t, R C, where R is revenue and C is ost, but also with the rm s revenue, R, and their objetive funtion, after investment is already sunk, is given by 9 R + (1 ) (R C). This objetive funtion re ets the idea that the managers of state-owned enterprises often have onsiderable interest in expanding the sale or sope of their ativities and expand the rm s budget for politial reasons. Alternatively, the manager s of partially state-owned rms are less exposed to the disiplining fores of the apital market and to takeover threats, and hene nd it easier to expand the rm s budget in order to pursue their own private agenda. Noting that C + (p; D) T and realling that sine we have a unit demand 2 funtion, R p, the ex post payo of the rm s managers under the MPE approah an be written as R + (1 ) (R C) R (1 ) C p (1 ) 2 (1 ) (p; D) T: (2) As for the SBC approah, some authors (e.g., Shmidt, 1996, and Maskin and Xu, 2001) argue that publi ownership is a major ause of SBC. Aording to this view, stateowned rms are more likely to be bailed-out by the state in ase they beome nanially distressed. Using b to denote the probability of a bailout, and assuming for simpliity that the rm does not bear any ost of distress if it is bailed out (this assumption an be easily relaxed so long as the ost of distress is smaller under a bailout), the ex post payo of the rm s managers under the SBC approah is given by R 2 (1 b) (p; D) T p 2 (1 b) (p; D) T: (3) 9 For related papers whih model the e et of state ownership by modifying the rm s objetive funtion, see for example, Bös and Peters (1988), De Fraja and Delbono (1989), Fershtman (1990), Cremer, Marhand and Thisse (1989, 1991), and Lee and Hwang (2003). 8
There is evidene that suggests that the probability of a bailout, b, inreases with the state s stake in the rm. For instane, Glowika (2006) nds that distressed publi rms are more likely to reeive long-term government assistane ( restruturing aid ), while distressed private rms are more likely to reeive only short-term resue aid, whih is intended to keep them in operation until a restruturing plan is in plae. Borisova et al. (2011) examine stok purhases in publily traded ompanies by governments or state-owned investors and nd strong support for the notion that debtholders view government ownership as an impliit assurane of repayment and protetion against bankrupty. Similarly, Borisova and Megginson (2012) examine orporate bonds of fully and partially privatized rms and show that on average, a one-perentage-point inrease in government ownership is assoiated with a derease in the redit spread of roughly three-quarters of a basis point. If we take b to be linear in, the payo of the rm s managers under the SBC approah oinides with their payo under the MPE approah, exept for the oe ient of, whih is 2 equal to 1 under the MPE approah, and is equal to 1 under the SBC approah. Hene, we an apture both approahes with the following (ex post) payo funtion: p 2 (1 ) (p; D) T; (4) where 1 under the MPE approah and 1 under the SBC approah. Importantly, under both approahes, the managers of a partially state-owned regulated rm e etively behave as if they ignore a fration of the rm s expeted ost of nanial distress. Ex ante, before k is sunk, the objetive of the rm s managers is given by the same expression minus k. 2.3 The rate setting proess, regulatory independene, and regulatory limate Following Besanko and Spulber (1992), Dasgupta and Nanda (1993), and Spiegel and Spulber (1997), we assume that the regulator sets the regulated prie, p, in order to maximize a 9
welfare funtion de ned over onsumers surplus, V (k) p, and the rm s objetive funtion: 10 (V (k) p) (p 2 (1 ) (p; D) T k) 1 : (5) The parameter 2 (0; 1) aptures the regulatory limate: the higher, the more proonsumer the regulator is. The resulting regulated prie alloates the expeted surplus aording to the asymmetri Nash bargaining solution for the regulatory proess. Under this interpretation, the parameters and 1 re et the bargaining powers of onsumers and the rm. Our approah is therefore onsistent with models that view the regulatory proess as a bargaining problem between onsumers and investors (Spulber, 1988 and 1989). Alternatively, the welfare funtion (5) ould represent a redued form for the regulator s own payo from being involved in some politial eonomy game. It is often argued that a greater degree of regulatory independene improves the regulators ability to make long-term ommitments to regulatory poliies (see e.g., Levy and Spiller, 1994, Gilardi 2002 and 2005, and the disussion in Edwards and Waverman, 2006). 11 In line with this argument, we will assume that the regulator is ommitted to the regulatory rule given by (5) only with probability. With probability 1, the regulator happens to be opportunisti, and after k is sunk, he sets a lower regulated prie. The parameter 2 (0; 1) then re ets the regulator s ability to make long-term ommitments to the regulatory rule 10 Our approah is onsistent with the observation that in pratie, regulators set pries to balane the interests of onsumers and rms. For example, aording to the U.S. Supreme Court, prie regulation involves a balaning of the investor s and the onsumers interests that should result in rates within a range of reasonableness (see Federal Power Comm. v. Hope Natural Gas Co., 320 U.S. 591, 603 (1944)). Similiarly, Ofwat, the water and sewerage regulatory ageny in England and Wales states that...it is our role to protet the interests of onsumers while enabling e ient ompanies to arry out and nane their funtions. This is a deliate balaning at. On the one hand, we must be sure that ustomers ontinue to reeive the servies that they expet at a prie they are willing to pay now and over the long term. On the other, we must ensure that the ompanies have su ient resoures to deliver servies e iently and remain attrative to investors... (see Ofwat, 2010, p. 3). 11 Guash, La ont, and Straub (2008) provide empirial support for this argument by showing that the presene of an IRA lowered the probability of renegotiation of ontrats for the provision of utilities servies by 5% 7:3%. This e et is signi ant given that the average probability of renegotiation of any individual ontrat at any point in time is around 1%. The better ability of IRAs to make long-term ommitments suggests that IRAs are less opportunisti than non-independent regulators. 10
and therefore serves as our measure of regulatory independene, with higher values of indiating a greater degree of independene. 12 Spei ally, we will assume that while a ommitted regulator always sets a the prie p to to maximize (5), an opportunisti regulator takes advantage of the fat that p is set after k is already sunk, and hene sets p to maximize an ex post objetive funtion that ignores k: (V (k) p) (p (1 ) (p; D) T ) 1 : (6) 2 Again, the probability that the regulator is ommitted is while the probability that the regulator is opportunisti is 1. In a tehnial Appendix, we show that the main results of the paper remain virtually the same if we adopt an alternative approah and assume that an opportunisti regulator uses a more pro-onsumer rule when setting p (i.e., uses a higher for setting p) rather than ignore k when setting the regulated prie. 13 2.4 The sequene of events The strategi interation between the rm s managers and the regulator evolves in two stages. In stage 1, the rm s managers hoose k and issues debt with fae value D in a ompetitive apital market. 14 If the funds raised by issuing D exeed k, the rm pays the exess funds as a dividend. If the funds raised by issuing D fall short of k, the rm raises additional funds by issuing equity; to simplify matters, we assume that in this ase the state partiipates in 12 A similar appoah is used in the literature on entral banks independene, where a greater degree of independene is modeled by assuming that the publi assigns a larger probability to the event that the entral banker is ommitted to his preannouned level of in ation. By ontrast, an opportunisti entral banker hooses ex post an atual level of in ation whih may di er from the one that he announed. See for example Cukierman (1992). 13 For details, see http://www.tau.a.il/~spiegel/papers/cs-appendix-july-4-2012-gamma.pdf 14 Our approah di ers from De Fraja and Stones (2004) and Stones (2007) where the regulator, rather than the rm, hooses the apital struture of the rm. These papers also assume that the regulator must set p to ensure that the rm never goes bankrupt and shareholders earn their required rate of return. Our approah also di ers from Lewis and Sappington (1995) who examine the optimal design of apital struture in the ontext of an ageny model that involves a risk-averse regulator (a prinipal) and a risk-neutral regulated rm (an agent) under alternative assumptions regarding the prinipal s ability to ontrol the agent s apital struture. 11
the equity issue to maintain its original stake. 15 In stage 2, given k and D, the regulator sets the regulated prie p. As mentioned earlier, the regulator is ommitted to set p in order to maximize (5) with probability, but with probability 1, the regulator happens to be opportunisti and sets p in order to maximize (6). 16 Finally, the rm s ost is realized, output is produed, and payo s are realized. Our sequene of events (the rm makes its hoies before the regulated prie is set) is onsistent with the empirial nding in Bortolotti et al. (2011) and Cambini and Rondi (2012) that leverage Granger auses regulated pries, but not vie versa. 3 The regulated prie In stage 2 of the game, the regulator sets p to maximize the ex ante objetive funtion (5) with probability and the ex post objetive funtion (6) with probability 1. Sine the two funtions di er only with respet to k, we an rewrite the regulator s objetive funtion ompatly as (V (k) p) (p 2 (1 ) (p; D) T Ik) 1 ; (7) where I is an indiator funtion whih equals 1 with probability (the regulator keeps his ommitment to the regulatory rule) and equals 0 with probability 1 (the regulator behaves opportunistially and ignores k when setting p). It should be noted that at the extreme when 1, the regulator ares only about onsumers and hene sets a ost-based prie that simply overs the rm s expeted osts. At the opposite extreme when 0, the regulator ares only about the rm and sets p V (k); this prie is independent of the rm s ost. In general then, the lower is, the more responsive is the regulated prie to the rm s ost. Using (7), we an now solve the problems of both ommitted and opportunisti regu- 15 Without this assumption, there would be another link between the investment deision of the rm, its apital struture, and its ownership struture. However, taking this link into aount would require a theory of publi ownership (i.e., a theory that would endogenize the state s stake in the rm). Suh a theory is beyond the sope of the urrent paper. 16 More formally, one an think about the game as having three stages: Nature hooses the regulator s type (ommitted or opportuniti) in stage 1, the rm s managers hooses k and D in stage 2 wiithout observing nature s hoie, and in stage 3, the regulator sets p given his type. 12
lators by simply maximizing (7) with respet to p. Using the same steps as in Spiegel (1994), the solution to the maximization problem is given by 8 D 1 (k; I) + D D 1 (k; I) ; >< D + D p 1 (k; I) < D D 2 (k; I) ; (D; k; I) D 1 (k; I) + + M (D; I) D 2 (k; I) < D D 3 (k; I) ; >: D 1 (k; I) + + (1 ) T D > D 3 (k; I); where D 1 (k; I) (1 ) V (k) + + Ik; (9) 2 M(D; I) (1 ) T D + (2 ) Ik 2 ; (10) 1 + (1 ) T D 2 (k; I) (1 ) 1 + (1 ) T V (k) + + Ik 2 ; (11) 1 + (1 )(1 ) T and D 3 (k; I) is smaller than the value of D for whih D 1 (k; I) + + M (D; I) D. This solution is obtained under the assumption that < V (0) (the regulator is not too proonsumer). If this assumption is violated, then D 1 (k; 0) 0, though none of our results V (0) 2 is a eted. The regulated prie is illustrated in the following gure: (8) Figure 1: Illustrating the regulated prie as a funtion of D for I 0 (the solid red line) and I 1 (the dashed blue line), holding k xed 13
To interpret Figure 1, note that if we ignore nanial distress, i.e., assume that (p; D) 0, then the prie that maximizes (7) is given by D 1 (k; I) +. So long as D D 1 (k; I), this prie overs the rm s ost plus its debt obligation even in the worst state of nature. 17 Hene, (p; D) is indeed equal to 0 for all D D 1 (k; I). However, one D > D 1 (k; I), a prie of D 1 (k; I) + leaves the rm suseptible to nanial distress. So long as D does not exeed D 1 (k; I) by too muh, the regulator nds it optimal to set p D + to keep (p; D) just equal to 0. However, when D > D 2 (k; I), this strategy is no longer optimal for the regulator beause the resulting marginal loss in onsumers surplus beomes too large relative to the bene t of preventing nanial distress. The regulator now allows the rm to harge a prie premium, given by M(D; I), to lower the probability that the rm beomes nanially distressed. Although the prie premium M(D; I) is inreasing with D, its slope is less than 1; hene p is now smaller than D +, and as a result, (p; D) > 0. When D > D 3 (k; I), it is no longer optimal for the regulator to o set the e et of debt on the likelihood of nanial distress. Consequently, (p; D) 1, so p is now onstant and equals D 1 (k; I) + + (1 )T. It is easy to see from equations (9) and (11) that D 1 (k; 1) > D 1 (k; 0) and D 2 (k; 1) > D 2 (k; 0), and moreover, it is easy to hek from (8) that p (D; k; 1) p (D; k; 0): the regulated prie set by a ommitted regulator (who takes k into aount) is weakly higher than prie set by an opportunisti regulator (who ignores k). To limit the number of di erent ases that an arise, we make the following assumption: Assumption 1: D 1 (k; 1) < D 2 (k; 0). Assumption 1 ensures that the parameters of the model are suh that there exists an interval of D for whih p (D; k; 1) p (D; k; 0). 18 A su ient ondition for Assumption 1 to hold is that the soial surplus absent nanial distress is su iently large: V (k) 2 k > k (1 ) (1 ) T : 17 As mentioned above, if is relatively large, then D 1 (k; I) 0 and the regulator annot ignore the possibility of nanial distress, no matter how small D is. 18 Absent Assumption 1, p (D; k; 1) > p (D; k; 0) for all D, although none of our main results is a eted. 14
Assumption 1, together with the fat that D 2 (k; 0) < D 2 (k; 1), implies that, as Figure 1 shows, D 1 (k; 0) < D 1 (k; 1) < D 2 (k; 0) < D 2 (k; 1): 4 The hoie of apital struture Assuming that the apital market is perfetly ompetitive, the market value of new equity and debt is exatly equal in equilibrium to their expeted return. Hene, outside investors (debtholders and possibly new equityholders if the rm also issues new equity) must break even. This implies in turn that the entire expeted pro t of the rm, p ost of investment, k, must arue to the original equityholders. C, net of the sunk To write down the rm s objetive funtion, let (D; k; I) (p (D; k; I); D) be the probability of nanial distress, whih is obtained by substituting p (D; k; I) into equation (1). Now, reall that with probability, the regulator is ommitted to take k into aount, in whih ase the regulated prie is p (D; k; 1) and the probability of nanial distress is (D; k; 1). With probability 1, the regulator is opportunisti, so the regulated prie and probability of nanial distress are p (D; k; 0) and (D; k; 0). Using these expressions and equation (4), the expeted payo of the rm s managers is given by Y (D; k) p (D; k; 1) (1 ) (D; k; 1) T k 2 + (1 ) p (D; k; 0) (1 ) (D; k; 0) T k : (12) 2 The rm s managers hoose the rm s debt level, D, and investment, k, to maximize Y (D; k). The following proposition haraterizes the equilibrium hoie of debt. The proof, as well as all other proofs, is in the Appendix. Proposition 1: In equilibrium, the regulated rm will issue debt with fae value D 2 (k; 0) if <, and will issue debt with fae value D 2 (k; 1) if >, where (1 ) (1 ) T : (13) 1 + (1 ) (1 ) T Proposition 1 shows that the apital struture of the rm depends on, whih re ets the degree of regulatory independene. In what follows, we will say that the regulator is 15
independent if > (the regulator s ability to ommit to take k into aount is relatively high) and non independent if < (the regulator s ability to ommit is relatively low). Proposition 1 shows that the rm issues more debt when it faes an independent regulator. Note from (13) that the threshold above whih we onsider the regulator as independent is dereasing with both and : other things equal, a more pro-onsumer regulator (a higher ) who faes a less privatized rm (a higher ) is onsidered independent for a larger range of values of. We now establish two orollaries to Proposition 1. Corollary 1: When the regulator is non independent ( < ), the regulated prie is equal to D 2 (k; 0) + with probability 1. When the regulator is independent ( > ), the regulated prie is equal to D 2 (k; 1) + with probability and D 1 (k; 0) + + M (D 2 (k; 1); 0) with probability 1, where D 2 (k; 1) + > D 1 (k; 0) + + M (D 2 (k; 1); 0). The expeted regulated prie when > is therefore Ep (k) D 2 (k; 1) + (1 ) (D 1 (k; 0) + M (D 2 (k; 1) ; 0)) + : (14) Corollary 1 shows that the regulated prie is be fully antiipated when the regulator is non-independent ( < ), but not when the regulator is independent. This result may seem surprising beause an independent regulator has a greater ability to ommit to the regulatory rule and determine the regulated prie. However, preisely for this reason, the regulated rm is able to issue in this ase debt with a larger fae value. This debt level in turn indues an opportunisti regulator to set a lower prie than the prie set by a ommitted regulator. The next orollary deals with nanial distress. When the regulator is non independent ( < ), the rm issues debt with fae value D 2 (k; 0). Sine by Corollary 1, the resulting regulated prie is D 2 (k; 0) +, the rm is immune to nanial distress even when the highest ost shok is realized. When the regulator is independent ( > ), the rm s debt is D 2 (k; 1). By Corollary 1, the regulated prie in this ase is D 2 (k; 1) + ; with probability, this prie ensures one again that the rm never beomes nanially distressed. With probability 1, though, the regulated prie is D 1 (k; 0) + + M (D 2 (k; 1); 0); sine 16
this prie is below D 2 (k; 1) +, the rm now beomes nanially distressed when the ost shok is su iently large. Corollary 2: When the regulator is non independent ( < ), the rm is ompletely immune to nanial distress. When the regulator is independent ( > ), the rm is immune to nanial distress with probability (the regulator is ommitted); with probability 1 (the regulator is opportunisti), the rm beomes nanially distressed when is su iently high. Corollary 2 shows another impliation of Proposition 1: the regulated rm may beome nanially distressed only when the regulator is independent. As before, the reason is that in this ase, the rm allows itself to issue debt with a higher fae value. With probability 1, the regulator happens to be opportunisti, and sets a regulated prie that leaves the rm suseptible to nanial distress with a positive probability. With Proposition 1 in plae, we an now examine how the equilibrium debt level is a eted by the main exogenous parameters of the model, holding the rm s investment level, k, xed. Proposition 1 already shows that the rm will issue more debt when the regulator is independent ( > ) than when the regulator is non independent ( < ). In the next proposition, we examine how debt is a eted by the other two main exogenous parameters: the state s stake in the regulated rm,, and the measure of regulatory limate (i.e., how pro-onsumer the regulator is),. Proposition 2: Holding k xed, the debt level of the regulated rm is higher the lower and are. Combined, Propositions 1 and 2 imply that if we onsider a ross setion of regulated rms that di er in terms of the degree to whih they are privatized (the value of ) and in terms of the regulatory environment they operate in (the values of and ), then other things equal, rms should be more leveraged when they are more privatized ( is lower) and when they fae more independent and more pro- rm regulators ( is higher and is lower). These preditions are onsistent with Bortolotti et al. (2011) who study a omprehensive panel data of 92 publily traded EU utilities over the period 1994 2005 and nd that rms 17
tend to be more leveraged if they are privately ontrolled (i.e., the state s stake in the rm is below 50% or below 30%) and regulated by an IRA. 19 Although Bortolotti et al. establish their results without ontrolling for investments, we show in Proposition 7 below that the preditions of Propositions 1 and 2 generalize to the ase where k is determined endogenously. To see the intuition for Proposition 2, note that in equilibrium, the rm issues the largest D that still ensures that if the regulator is ommitted, the rm will be ompletely immune to nanial distress. Naturally then, the rm will issue a higher D if p is higher. When the state holds a smaller stake in the rm, the rm takes into aount a larger fration of its expeted ost of nanial distress, so the regulator, who sets p by taking into aount the rm s objetive funtion, will set a higher p. The reason why D is higher when is low is more subtle sine now there are two opposing e ets. On the one hand, other things being equal, p is higher when is low (the regulator is more pro- rm), so the rm has an inentive to issue more debt. But on the other hand, as noted above, a derease in makes pries less responsive to the rm s ost; onsequently, debt, has a weaker e et on the regulated prie. It turns out that the rst e et is always stronger, so a derease in indues the rm to raise D. Finally, sine other things being equal, p is higher when the regulator is independent, the rm will also issue a higher D when it faes an independent regulator. Next, we examine how the regulated prie is a eted by and. As in the ase of Proposition 2, for now we hold k xed. In Setion 5, we will show that our omparative statis results ontinue to hold even when k is determined endogenously. Proposition 3: Holding k xed, the expeted regulated prie is higher when the regulator is independent ( > ) than it is when the regulator is non independent ( < ). Moreover, the expeted regulated prie is dereasing with both the state s ownership stake, and with the measure of regulatory limate. The result that the regulated prie is dereasing with the state s ownership stake is 19 Bortolotti et al. (2011) do not have a diret measure of the regulatory limate and hene annot study the e et of the regulatory limate on leverage and on pries. Their analysis shows however that rms have a lower leverage when the government is more right-wing. Cambini and Rondi (2012) nd a similar result in a study that examines 15 EU Publi Telommuniation Operators (PTOs) over the period 1994-2005. To the extent that right-wing governments are more pro- rm, this nding is inonsistent with Proposition 2. 18
onsistent with Kwoka (2002) who shows that after ontrolling for ost di erenes, the pries of publily owned eletri utilities in the U.S. are 4:4% heaper, on average, than the pries of investor owned utilities. Moreover, together with Proposition 2, Proposition 3 implies that if we hold k xed, then any hange in the parameters, and shifts the rm s debt and the regulated prie in the same diretion. This implies in turn that in a sample of regulated rms that di er from eah other only in terms of, and, the rm s debt and regulated prie should be positively orrelated. This nding is onsistent with Bortolotti et al. (2011) and with Cambini and Rondi (2012). The latter paper shows that the leverage of PTOs has a positive e et not only on regulated retail rates, but also on the wholesale aess fees that PTOs harge alternative operators who wish to aess the PTOs networks. Finally, reall from Corollary 2 that the rm never beomes distressed if <. When >, the rm beomes distressed only when the regulator is opportunisti and sets a prie p (D 2 (k; 1); k; 0) D 1 (k; 0) + + M (D 2 (k; 1); 0). Sine the probability of this event is 1, the overall probability of nanial distress when > is (1 ) I (k), where, using equation (1), I (k) p (D 2 (k; 1); k; 0) D 2 (k; 1) 1 {z } (D 2 (k;1);k;0) D 2(k; 1) D 1 (k; 0) M (D 2 (k; 1); 0) k : 1 + (1 ) T (15) The following result is an immediate onsequene of equation (15): Proposition 4: Holding k xed, the probability of nanial distress when an independent regulator happens to be opportunisti, I (k), is inreasing with,, and k and is independent of. Under a non-independent regulator, the rm never beomes nanially distressed. At a rst glane, Proposition 4 seems ounterintuitive sine Proposition 2 implies that the rm issues less debt, D, when and are higher. Hene it might be thought that the rm would be less suseptible to nanial distress. Yet, Proposition 3 shows that when and are higher, the regulated prie, p, is also lower. It turns out that the derease in p has 19
a stronger e et on the probability of nanial distress than the derease in D, so overall, nanial distress beomes more likely. 5 The equilibrium level of investment Having haraterized the equilibrium hoie of debt, we next turn to the hoie of investment. Consider rst the ase where <, and reall from Corollaries 1 and 2 that in this ase, D D 2 (k; 0). The regulator in turn sets a prie D 2 (k; 0) +, whih ensures that the rm is ompletely immune to nanial distress. By equation (12) then, the expeted payo of the rm is Y NI (k) Y (D 2 (k; 0) ; k) D 2 (k; 0) + (2 ) k: (16) 2 When >, the rm issues debt with fae value D 2 (k; 1). Now, with probability, the regulator is ommitted and sets a regulated prie p (D 2 (k; 1); k; 1) D 2 (k; 1) +, whih ensures that the rm never beomes nanially distressed. With probability 1, the regulator is opportunisti and sets a prie p (D 2 (k; 1); k; 0) D 1 (k; 0) + + M (D 2 (k; 1); 0); with this prie, the rm beomes nanially distressed with probability I (k). Substituting these expressions in equation (12), using the de nition of M (D 2 (k; 1); 0), and rearranging terms (see the proof of Proposition 5 for details), the rm s expeted payo is Y I (k) Y (D 2 (k; 1) ; k) (1 (1 )) V (k) (1 ( )) k (17) (1 ) 1 + (1 ) T 2 : 1 + (1 ) (1 ) T Using Y NI (k) and Y I (k) we establish the following result: Proposition 5: The equilibrium level of investment, k, is independent of the degree of regulatory independene,, when <, but is inreasing with when >. Consequently, the rm invests more when the regulator is independent (i.e., > ) than when the regulator is non independent (i.e., < ). Sine regulatory independene in our model is assoiated with a smaller degree of regulatory opportunism, Proposition 5 is onsistent with Lyon and Mayo (2005) who show that a greater threat of regulatory opportunism leads rms to invest less. 20
Having fully haraterized k and showed how it is a eted by regulatory independene, we are now ready to examine how k is a eted by the state s stake in the rm,, and by the regulatory limate,, whih re ets the degree to whih the regulator is pro-onsumers. Proposition 6: The equilibrium level of investment, k, is dereasing with and. If in addition V 0 (k) V 00 (k) is nondereasing, then the negative e ets of and on k are larger when the regulator is independent, i.e., when >. To see the intuition for Proposition 6, reall from Proposition 2 that when and are higher, the regulator sets a lower regulated prie. Consequently, the marginal bene t of investment falls and the rm invests less. Proposition 6 shows that these e ets are stronger when the regulator is independent, i.e., when >. Propositions 5 and 6 imply that other things being equal, rms should invest more when they fae an independent regulator, when they are more privatized (i.e., is lower), and when they fae a more pro- rm regulator (i.e., is lower). These preditions are onsistent with a number of empirial ndings. Wallsten (2001) studies the investment of Teleoms in 30 Afrian and Latin Amerian ountries from 1984 to 1997. Among other things, he nds that privatization ombined with regulatory independene is positively orrelated with investment in apaity and phone penetration. Privatization alone, however, is assoiated with few bene ts, and is negatively orrelated with interonnetion apaity. Henisz and Zelner (2001) study data from 55 ountries over 20 years and nd that stronger onstraints on exeutive disretion, whih improves their ability to ommit not to expropriate the property of privately owned regulated rms, leads to a faster deployment of basi teleommuniations infrastruture. Gutiérrez (2003) examines how regulatory governane a eted the performane of teleoms in 22 Latin Amerian ountries during the period 1980 1997 and nds that regulatory independene has a positive impat on network expansion and e ieny. Alesina et al. (2005) examine the aggregate levels of investment in the transport, teleommuniations, and energy setors in 21 OECD ountries over the period 1975-1998. Among other things, they show that a larger ownership stake of the state is assoiated with lower levels of investment. Egert (2009) shows that inentive regulation implemented jointly with an 21
independent setor regulator has a strong positive impat on investment in various network industries (eletriity, gas, water supply, road, rail, air transportation, and teleommuniations) in OECD member ountries. Finally, Cambini and Rondi (2010) study a panel of 80 publily traded EU teleoms, energy, transportation, and water utilities over the 1994-2004 period and nd that utilities invest more when an IRA is in plae; moreover, they nd that onditional on the existene of an IRA, rms invest more when the IRA has a larger degree of formal independene. Next, reall that Propositions 1-4 examined the e ets of regulatory independene, privatization, and the regulatory limate on the rm s debt level, regulated prie, and the probability of distress, holding k xed. We now show that these results ontinue to hold even after the endogenous hoie of k is taken into aount. Proposition 7: Taking into aount the endogenous hoie of investment, the rm s debt and the regulated prie are higher when > (the regulator is independent) than they are when < (the regulator is non independent). Moreover, the rm s debt and the regulated prie are both dereasing with the state s ownership stake, and with the measure of regulatory limate. The probability of nanial distress when an independent regulator is opportunisti, I (k ), is inreasing with the degree of regulatory independene,. If in V addition is su iently small to ensure that 0 (k ) + (1 )(1+(1 )(1 ) T ) 0, then I (k ) V 00 (k )k is also inreasing with the state s ownership stake,, and with the measure of regulatory limate,. The result that I (k ) is inreasing with the degree of independene,, is surprising given that an inrease in means that the regulator is less likely to be opportunisti (reall that nanial distress ours only when the regulator is opportunisti). The reason for this surprising result is that when the regulator is independent, an inrease in indues the rm to invest more and to issue more debt to nane its investment. Indeed, Proposition 4 shows that I (k) is inreasing with k and Proposition 5 shows that k is inreasing with. As a result, an inrease in makes the rm more suseptible to nanial distress. Proposition 7 also shows that the result of Proposition 4 that the rm is more suseptible to nanial distress as and inrease ontinues to hold when k is endogenous, provided that is 22
su iently low. To get a better feel for the su ient ondition in the last part of Proposition 7, V suppose that V (k) log (a + k), where a < 1. Then, 0 (k 1 ) a+k V 00 (k )k k 1 + a (a+k ) 2 k. In the proof of Proposition 8 below we show that V 0 (k ) > 1. In the urrent example, this inequality implies that 1 a+k > 1, or k < 1 a: Hene. su ient ondition then is more likely to hold as a gets smaller. V 0 (k ) V 00 (k )k 1 + a k < 1 1 a. The 6 Soial welfare Having studied the rm s investment and naning deisions, we now turn to the impliations of our model for soial welfare. In partiular, we are interested in nding out how regulatory independene, privatization, and the regulatory limate a et soial welfare one the rm s and the regulator s deisions are taken into aount. In our model, the expeted value of soial welfare is given by the di erene between the willingness of onsumers to pay and the expeted ost of the rm, inluding its expeted ost of nanial distress and ost of investment: W (k) V (k) 2 (1 ) (D; k; I) T k. By Corollary 2, (D; k; I) 0 when the regulator is not independent. Hene, the expeted soial welfare, as a funtion of k, is given in this ase by W NI (k) V (k) 2 k. (18) When the regulator is independent, equation (15) shows that (D; k; I) Hene, expeted soial welfare, as a funtion of k, is given by W I (k) V (k) 2 (1 ) k T 1 + (1 ) T k (1+(1 ) T ). k. (19) In the next proposition we ompare the equilibrium level of investment, k, with the soially optimal level that maximizes W NI (k) and W I (k) and examine how soial welfare is a eted by regulatory independene, privatization, and the regulatory limate. Proposition 8: The equilibrium level of investment, k, is below the soially optimal level. Moreover, in equilibrium, 23
(i) soial welfare is independent of the degree of regulatory independene,, but is dereasing with the state s ownership stake, and with the measure of regulatory limate when the regulator is non-independent (i.e., when < ); (ii) assuming that 1 (1 ) T > 0, soial welfare is inreasing with the degree of regulatory independene,, and dereasing with the state s ownership stake, and with the measure of regulatory limate, when the regulator is independent (i.e., when > ). Proposition 8 shows that when we take into aount the endogenous determination of investment and apital struture, a higher degree of regulatory independene (a higher ), a larger extent of privatization (a derease in the value of ), and a more pro- rm regulatory limate (a lower value of ), are all welfare-enhaning. The reason for this is that as Propositions 5-6 show, regulatory independene, privatization and pro- rm regulatory limate strengthen the rm s inentive to invest and, while the regulated prie inreases too, the inrease in investment leads to an inrease in the total surplus generated by the rm. 7 Conlusion We studied the strategi interation between apital struture, regulation, and investment, in a setting that features partial ownership by the state in the regulated rm and regulation by agenies with various degrees of independene. Both features are ommon in many ountries around the world. Our model shows that regulated rms issue more debt and enjoy higher regulated pries when they fae more independent regulators, are more privatized, and when regulators are more pro- rm. At the same time, these fators also indue the rm to invest more and this inrease in investment is overall welfare improving. These results indiate that the dash for debt phenomenon observed in many ountries is a natural response of regulated utilities to the privatization proess and the establishment of IRAs. Moreover, our results suggest while privatization and regulatory independene lead to a dash for debt, these proesses also lead to higher soial surplus. 24
8 Appendix Investment rate of utilities relative to GDP in the EU14 states: The following table shows the rate of gross xed apital formation in the energy setor (eletriity and gas), water supply, transport, and teleommuniations, as a share of GDP in 2008, using the OECD s STAN (Strutural Analysis) Indiators database. This database provides annual setorial indiators on the prodution and employment strutures, labor produtivity and osts, investments, R&D expenditures, and international trade patterns in eah OECD ountry. Table 1: Investment rate as % of GDP in 2008 in the EU14 states State Investment rate as % of GDP Austria 13.94% Belgium 15.57% Denmark 18.80% Finland 15.79% Frane 9.84% Germany 11.70% Greee 14.59% Ireland 19.00% Italy 16.63% Netherlands 9.66% Portugal 20.24% Spain 14.58% Sweden 18.51% UK 14.47% Average EU14 15.24% 25