Investment coordination in network industries: the case of electricity grid and electricity generation

Abstract

Liberalization of network industries frequently separates the network from the other parts of the industry. This is important in particular for the electricity industry where private firms invest into generation facilities, while network investments usually are controlled by regulators. We discuss two regulatory regimes. First, the regulator can only decide on the network extension. Second, she can additionally use a “capacity market” with payments contingent on private generation investment. For the first case, we find that even absent asymmetric information, a lack of regulatory commitment can cause inefficiently high or inefficiently low investments. For the second case, we develop a standard handicap auction which implements the first best under asymmetric information if there are no shadow costs of public funds. With shadow costs, no simple mechanism can implement the second best outcome.

Appendix

1.1 Analysis of the full commitment case

We take up the framework of Sect. 3.2 in which the regulator decides only on building the network and where there is asymmetric information with respect to $c_{N}.$ We call “full commitment” a situation in which the regulator moves second, but can ex ante commit how to react to the firms’ decisions. Obviously, such “full commitment” can never be worse than any of the regulatory strategies discussed so far, since “full commitment” can at least mimic any of those. However, the additional gain from “full commitment” is surprisingly small. The only strategic options the regulator gains in addition to the ones already discussed $\left( always,never,nocom\right) ,$ are the following: (i) to commit to building the link if and only if at the first stage nothing was built by the firms, or (ii) commit to building the link if the firms had built in the South or if nothing was built by the firms.¹⁷ Both strategies avoid the investment forcing effect (since if the firms build in the North, the regulator commits not to build the link). The latter strategy additionally avoids the investment preempting effect (since the regulator commits to building the link if investment in the South occurred). Under both rules firms never build in the North, hence no information is ever revealed to the regulator. Therefore, as uncertainty becomes large and therefore if the probability mass on regions 1 and 2 of Fig. 1 vanishes and is shifted to the left and right of these regions, there is no additional advantage from full commitment. Thus, the best full commitment can achieve in this case is to mimic $no$ $commitment$.

1.2 Alternative institutional arrangements

In Sect. 3, we focused on vertically separated networks where the regulator decides on the network extension. Alternative institutional arrangements are used in practice or are discussed, in particular vertically integrated firms and private “merchant” transmission investors. We briefly discuss the effects of these alternatives by comparing them to the outcome with a fully informed regulator without commitment (the framework of Sect. 3.1; “regulator decides on the network extension” in what follows). We also maintain from Sect. 3.1 the price formation for electricity, the assumption that generation investors move before the decision on the network is taken, and the assumption of “free entry”.

1.2.1 Vertically integrated incumbent, with reimbursement of network cost

Consider a situation with an incumbent who owns all generation capacity which is initially installed. In addition, this incumbent decides on the network extension and he enjoys a first mover advantage with respect to generation investments (can invest in generation before entrants can do so). The network is “regulated” in the sense of a cost-reimbursement rule, i.e., the integrated incumbent gets all network investments reimbursed (e.g., via network fees paid by final customers).

Lemma 1

An arrangement with a vertically integrated incumbent, with reimbursement of network cost, is welfare inferior to an arrangement in which the regulator decides on the network extension.

In Proposition 1, we describe the inefficiencies the regulator suffers from when she is unable to commit: the investment preempting effect, and the investment forcing effect. The same inefficiencies are present also with the vertically integrated incumbent since he has not to bear the cost of the network extension. However, there is an additional inefficiency added, a foreclosure effect. Consider a situation in which building the link only is optimal, and where the regulator can indeed implement this outcome. The former requires that $q_{S}q_{N}B<\min \left( c_{S}-L,c_{n}\right) ,$ the latter requires $q_{S}q_{N}m-c_{N}<0,$ i.e., private generation investments are not profitable with the link. A vertically integrated incumbent who does not build the link earns $\pi ^{o}=q_{S}M+\left( 1-q_{S}\right) m+q_{N}m.$ If he adds the link, profits are $\pi ^{L}=q_{S}q_{N}2M+q_{S}\left( 1-q_{N}\right) 2m+\left( 1-q_{S}\right) q_{N}2m.$ Therefore, $\pi ^{o}>\pi ^{L}$ if:

$$\begin{aligned}&\displaystyle q_{S}M+\left( 1-q_{S}\right) m+q_{N}m-\left( q_{S}q_{N}2M+q_{S}\left( 1-q_{N}\right) 2m+\left( 1-q_{S}\right) q_{N}2m\right) >0,\\&\displaystyle m+Mq_{S}-mq_{N}-3mq_{S}-2Mq_{N}q_{S}+4mq_{N}q_{S}>0,\\&\displaystyle m\left( 1-q_{N}\right) -q_{S}\left( M\left( 2q_{N}-1\right) +m\left( 1+4q_{N}\right) \right) >0, \end{aligned}$$

which holds for $q_{S}$ sufficiently small. For the integrated incumbent, the downside of building the link is that supply in the North can then negatively affect the price in the South, in particular if demand in the North is low. The only benefit to the integrated incumbent from building the link is that in cases of high demand in the South, with a connected network, this high demand can drive up the price also in the North. If this event is sufficiently unlikely, i.e., $q_{S}$ is sufficiently small, then the integrated incumbent abstains from building the link.

A similar effect arises if the regulator wants to implement $\mathcal{NL }$, and is actually able to do so, which requires that building in the North is privately profitable, and more profitable than investing in the South, $q_{S}q_{N}m-c_{N}>q_{S}m-c_{S}.$ If the link is built, we will never have excess demand (if the incumbent would not built it, an entrant would do so). As a consequence, if the integrated incumbent builds the link, he also adds generation in the North and earns $\pi ^{NL}=q_{S}q_{N}m.$ This is obviously smaller than $\pi ^{o}$ for $M$ sufficiently large. Adding link and generation implies that the incumbent forgoes the chance to earn the high price $M;$ for $M$ sufficiently high, the integrated incumbent implements $\mathcal{S }$ (if $q_{S}m-c_{S}\ge 0,$ or $\varnothing$ otherwise), instead of the welfare superior outcome $\mathcal{NL }$ which the regulator would implement.

In both of the two examples, the integrated incumbent abstains from building the link as a means to restrict the addition of generation in order to support higher prices; he does not invest to foreclose additional supply.

1.2.2 Vertically integrated incumbent, with private network investment

Consider the previous setup, but additionally assume that the integrated incumbent has to bear the cost of the network himself. Any outlays for the network need to be covered by the revenues generated on the electricity market. This, in a way, is closest to the “old world” prior to electricity market liberalization.

Lemma 2

(i) If there is no investment forcing effect, then an arrangement with a vertically integrated incumbent who bears the networks costs privately, is welfare inferior to an arrangement where the regulator decides on the network extension. (ii) Only if there is a investment forcing effect, the vertically integrated incumbent who bears the network cost privately can lead to higher welfare compared to an arrangement where the regulator decides on the network extension.

Welfare inferiority is obvious from the above discussion of foreclosure effects. These become stronger, since building the link becomes even less attractive due to the network costs. However, different to the situation with a cost reimbursement for the integrated incumbent, we now can identify conditions under which integration performs strictly better than having the regulator without commitment power decide on the network. The reason is that the investment forcing effect can no longer occur. The investment forcing effect requires that $\mathcal{S }$ is preferred to $\mathcal{NL }$, i.e., $c_{S}<c_{N}+L,$ but investment in the North (with network extension) is privately more profitable, $q_{S}m-c_{S}<q_{S}q_{N}m-c_{N}$, as long as the investor has to pay only for the generation, but not for the network. Now the integrated incumbent has to pay for the network, too. He would then invest in the North only if $q_{S}m\left( 1-q_{N}\right) <c_{s}-c_{N}-L,$ which can never happen if $\mathcal{NL }$ is optimal (since the right hand side is then negative).

1.2.3 Merchant transmission investor

Imagine that a third private party undertakes the network investment. Such a “merchant transmission investor” bears all cost of building the network. His revenues depend on the usage of his network, i.e., the cases where the prices at the two nodes differ. Generally, the merchant investor will be able to capture some fraction of the gains from trade between the two nodes. From the perspective of the generators, this means that they have to pay some amount of money $F$ for using the network for trading electricity. In our simple model, network usage occurs if and only if the demand in the South is high and no generation was added in the South (i.e., only with $\mathcal{NL }$ or $\mathcal L)$. In these cases, without usage of the network the price in the South would be $M,$ while with the use of the network it would only be $m<M.$ ¹⁸ We therefore assume that the merchant investor’s revenues are increasing in the difference $\left( M-m\right)$.

Lemma 3

(i) If there is no investment forcing effect, then a merchant transmission investor leads to lower social welfare compared to an arrangement where a regulator without commitment power decides on the network. (ii) Only if there is an investment forcing effect, the merchant transmission investor can lead to higher social welfare compared to an arrangement where the regulator decides on the network extension.

Without the investment forcing effect, the regulator might face the investment preempting effect. Investment preemption can not be solved by a merchant transmission investor. Whenever it is more profitable for the generators to add capacity in the South than to add them in the North (although this might be efficient) they will do so - independent of who decides on the network. Neither the regulator nor the merchant will add a network once generation addition in the South ensures there will never be any shortage at either node. However, the merchant introduces an additional inefficiency. Consider the case where $\mathcal{NL }$ or $\mathcal{L }$ is optimal. The merchant can make money only if the demand in the South is high. Thus, his profits increase in $q_{S}$. Thus, if $q_{S}$ is sufficiently small, the merchant will not invest (since he always has to bear the cost $L$), while absent the investment preempting effect (for which there is no difference between regulator or merchant deciding on the investment), the regulator can always implement the desired outcome. ¹⁹

In addition, whenever $q_{S}$ is large, implying good chances for the network to be built by the merchant, it will not be needed, since for high $q_{S}$ it is increasingly privately profitable to invest in the South (instead of investing in the North), and it also tends to become socially more desirable to implement $\mathcal{S }$ instead of $\mathcal{NL }$ or $\mathcal{L }$.

However, a merchant transmission investor might be beneficial as an instrument to avoid the investment forcing effect. With the merchant transmission investor, the generators lose some revenues $F$ from trading between North and South. Thus, inefficient investment forcing only happens if $q_{S}m\left( 1-q_{N}\right) +F<c_{N}-c_{S}<L$ while at the same time a merchant would indeed build the network if there had been generation addition in the North, which requires $q_{S}$ to be large and $L$ to be small. Thus, the conditions for the investment forcing effect to occur are more restrictive compared to a situation where the regulator decides on the network extension.

To summarize, it is not surprising that a merchant transmission investor tends to worsen the situation. Investment forcing effect and investment preempting effect are a result of the separation of the decision of generation and network. This separation is not solved by the merchant transmission investment; even more, the merchant does not account for the social benefit of the network, and therefore adds additional inefficiencies. The only circumstances in which they can be beneficial are—ironically—situations where no network should be build, i.e., where delegating the network decision to the merchant is a credible mechanism not to build the network.