Effects of Privatization on Price and Labor Efficiency: The Swedish Electricity Distribution Sector.

• Author: Lundin, Erik

1. INTRODUCTION

   When a network industry is privatized, a fundamental controversy lies in the potential increase in productive efficiency due to increased incentives to minimize costs, vs. distortions in allocative efficiency and a subsequent redistribution from consumers to producers due to market power. The electricity distribution sector is particularly well suited for firm performance comparisons since electricity distribution is a homogenous good, and since data on the technical characteristics of the networks as well as accounting data are standardized and reported to a central regulator. At the same time, since the market is regulated, the possibility to extrapolate results to other markets should be done with care.

   In this study, I examine the effects of private acquisitions of publicly owned networks in the Swedish electricity distribution sector. Specifically, I study the performance of 34 municipally owned networks that were acquired by private firms around the turn of the century. I focus on two outcome variables: price and labor efficiency. I find evidence of an increase in labor efficiency in the acquired networks by 8-18 percent compared to the control group depending on model specification, while no acquisition effect is found on the price. Thus, the evidence suggests economically meaningful efficiency gains but that these are not fed through to consumer prices. All acquisitions examined in the study were conducted by two firms, E.ON and Fortum. The qualitative effects on both outcome variables are similar across firms, although the increase in labor efficiency is both statistically and economically more significant in the networks acquired by Fortum.

   To the best of my knowledge, this is the first study to use a synthetic control method to evaluate firm performance in the electricity distribution sector. For each acquired network, I create a synthetic control network from a weighted average of the control networks. The synthetic control network is constructed to have the same technical characteristics and pre-acquisition trend of the outcome variable as the acquired network. The effect of the acquisition is then estimated by comparing the factual post-acquisition trend of the outcome variable to that of its synthetic analogue. Conceptually, it is a generalization of the more commonly used difference-in-differences (DiD) estimator, and is particularly well suited to estimate the effect of an intervention when the number of potential control firms are large. Robustness tests using a conventional DiD estimator largely confirms the results from the synthetic control method, although the estimated efficiency gains are either less pronounced, or comparable to, the results using the synthetic control method. Further, the precision of the estimates are comparatively low under both methods for several of the specifications. Therefore, results should be interpreted with care.

   Although it is beyond the scope of this paper to identify the specific mechanisms driving the efficiency gains, one plausible explanation is returns to scale, since each of the acquisitions involved bordering networks that were previously operated by each respective municipality. For example, the same administrative staff may be used for several networks, and the use of equipment may be optimized by transporting it across regions. It is also likely that some publicly owned firms prefer to offer higher wages relative to private firms ceteris paribus, in which case variations in labor efficiency also reflects how revenues are distributed between workers and owners. Still, it should be noted that since prices are regulated, also publicly owned firms are allowed to make profits that can be transferred to the general municipal budget.

   The choice of the electricity distribution sector as a testing ground for studying privatization is not only motivated by the availability of data and the diversity of the ownership structure. Another important factor is the increased skepticism against private ownership in network industries that has developed during the last decade, which can be exemplified by the experiences in Sweden and Germany. Prior to electricity reform, distribution networks in Sweden and Germany were usually owned by the municipalities. In Sweden, the state-owned firm Vattenfall also owned a fair share of the networks, and continues to do so (NordReg, 2011; OECD, 2004). However, the last two decades have seen a transition towards consolidation and privatization of the network ownership structure. In Sweden, the new entrants Fortum and E.ON now have a market share of around 20 percent each. Consumer groups claim that the sharp price increases during the last decade are mainly driven by the largest firms (SABO, 2011), and some municipalities have expressed an interest in buying back their networks (Dalarna's newspaper, 2014). Germany has seen a similar development, where the four largest firms acquired shares in numerous local distribution networks during the privatization wave initiated by the municipalities around the turn of the century. However, in contrast to Sweden, there has been a reverse trend in network ownership during recent years. Municipalities, often guided by the will of the local people, have started to repurchase the networks (the so-called Rekommunalisierung, or remunicipalization). The two most notable events were the referendums in the two largest German municipalities, Berlin and Hamburg, which in 2013 both voted for the remunicipalization of the electricity, gas, and district heating networks (although the referendum in Berlin failed due to insufficient voter turnout). Similar skepticism towards private ownership of network industries has also been raised elsewhere. For example, the website www.remunicipalisation.org collects data on remunicipalization projects of water provision services throughout the world, recording 180 cases during the last 15 years, impacting 100 million people. The recent years' remunicipalization trend further highlights the importance of examining the effects of privatization in network industries.

   Previous studies on the relationship between ownership and efficiency in Swedish electricity distribution are not conclusive. Using a stochastic frontier model analyzing panel data during 2000-2007, Soderberg (2011) finds that private ownership is associated with relatively lower costs than public ownership. However, the economic significance of the effect is rather modest and the sample is different from the present study since networks that changed ownership status during the sample period are excluded from the analysis. Using data from 1970-1990, Kumbhakar and Hjalmarsson (1998) find that private ownership is associated with relatively higher labor efficiency in terms of cross-sectional variation, but there is no conclusive evidence that labor efficiency increased more within the privately owned networks during the sample period.

   International experience also does not provide any clear cut predictions, although most evidence suggests that private ownership is associated with increased efficiency. In the UK, Domah and Pollitt (2001) find that privatization improved efficiency, but costs and prices did not fall until about a decade after privatization, partly as a result of regulatory interventions. In line with this finding, Karahan and Toptas (2013) find that the privatization of electricity distribution companies in Turkey did not yield lower prices within the first 4 years of the program. Within the Ukrainian distribution sector, Berg et al. (2005) find that privatization is conducive to efficiency given that regulation is adequate. Borghi et al. (2016) study the interaction effects between quality of government and private ownership in determining the total factor productivity of electricity distribution firms using data from 16 EU countries. When the quality of government is poor, private ownership is associated with relatively higher productivity levels, while the opposite is true when the quality of government is good. In a study using data from 14 Latin American countries, Estache and Rossi (2005) find that the relative labor efficiency of private firms depend on the regulatory regime: private firms perform better under price cap regulation, but not under rate-of-return regulation. A common argument against privatization is that it may be accompanied by a deterioration in quality. For example, Kwoka (2005) finds that public ownership is associated with relatively higher quality of service, measured by the occurrence of service disruptions. Although quality of service is not the focus of the present study, I include a measure of service interruptions as a control variable to reduce the probability that this factor is driving the estimated efficiency gains.

2. INSTITUTIONAL BACKGROUND AND DATA

   2.1 Regulatory framework

   Traditionally, a majority of the Swedish distribution networks have been owned by the municipalities, or by private firms organized as economic associations owned by the electricity consumers in the area where they operated. Some distribution networks were also owned by the state-owned firm Vattenfall, which is also the largest electricity producer in the Nordic region. Before the liberalization of the wholesale market in 1996, a proper incentive regulation for the distribution sector was not considered necessary, although there was a general legal principle stating that publicly owned firms were not allowed to make profits (sjalvkostnadsprincipen). The non-profit rule did not apply to private firms, that in theory were free to set their own prices. Shortly after the liberalization of the wholesale market, a new regulatory framework was introduced. It was a type of rate-of-return regulation, albeit without being any more precise than that prices should be fair (skaliga). Heden (2012) characterizes the evaluation of the tariffs as "pretty ad-hoc", noting...