Consumer Welfare in the Deregulated Swedish Electricity Market. Jens Lundgren *

Transcription

1 Consumer Welfare in the Deregulated Swedish Electricity Market Jens Lundgren * Abstract The deregulation of the Swedish electricity market in 996 affected both the market design and the ricing of electricity. Since 996, the electricity rice faced by consumers has increased dramatically. Due to the high electricity rice and large comany rofits, a debate about the success of the deregulation has emerged. As such, the aim of this aer is to investigate whether or not the deregulation of the Swedish electricity market has imroved consumers welfare. The theoretical framework is an equivalent variation method and the analysis is erformed using monthly data for the eriod January 996 to January The results indicate that deregulation has ket the ower rice (excluding taxes) down and increased consumer welfare in Sweden. Keywords: Equivalent variation, Consumer welfare, Power market. JEL classification: D60, L3, L43. * Deartment of Economics, Umeå University SE Umeå, Sweden. jens.lundgren@econ.umu.se. The author would like to thank the editor of this journal, two anonymous referees, Niklas Rudholm, Jörgen Hellström and seminar articiants at Umeå University, University of Gävle, the Energy Markets Insectorate and at the Energy and Value Symosium in Amsterdam for helful comments and suggestions. Financial suort from Sarbankernas Forskningsstiftelse is gratefully acknowledged. Of course, the usual disclaimer alies. 0

2 Introduction High electricity rices in Sweden have sarked a debate among olicy-makers and oliticians. Consumers have seen their electricity bills increase while ower generators have earned significantly higher rofits. The ublic debate is often steered by the media, which ortrays the electricity market deregulation as the cause of high consumer rices and large roducer rofits. The media often emhasizes that deregulation has lowered consumers well-being, desite the fact that the objective was the oosite. The urose of this aer is to analyze the effects on consumer welfare caused by the deregulation of the Swedish electricity market. Sweden deregulated its electricity market in 996. Power roduction and retailing were oened u to cometition, while transmission remained regulated. The rationale for deregulation is to develo a more a cometitive market which should lead to a more efficient use of resources. Efficiency gains should imly a lower electricity rice (Bergman et al, 994). The fact that the oosite aears to have haened does not necessarily imly that deregulation has failed; instead, the increase in consumer rices may be misleading. The rice of ower together with transmission costs and taxes are the comonents that sum to the electricity rice consumers face. A review of the rice develoment shows that taxes have gone u by more then 200 ercent since 996 while transmission costs have remained almost constant in real terms. Furthermore, environmental fees have been introduced during this eriod. 2 In summary, factors other than the deregulation of the market may have caused the electricity rice increases for consumers. In this aer, I analyze how the deregulation of the Swedish electricity market has affected the ower rice and how this, in turn, has affected consumers welfare. The aroach adoted in revious studies has been to look at the consumer welfare in terms of consumer rices (e.g. Damsgaard and Green, 2005). This study uses the Swedish area rice at the Nord Pool sot market in order to isolate the effect of the deregulation on The transmission cost is the aggregated costs of transmission and distribution. 2 The electricity taxes discussed are taxes on households and the commercial sector. The electricity taxes on industry have had a less raid growth. Since May 2003 a system of green certificates is in oeration in Sweden to encourage investment in environmentally friendly ower roduction. The CO 2 allowance market introduced in 2005 was a Euroean Union initiative and therefore the otential imact on electricity rices is not a result of Swedish deregulation. 02

3 electricity rices er se, aart from exogenous increases in taxes or the introduction of green certificates. 3 To answer the question of how deregulation has affected consumers welfare through shifts in the ower rice, the actual rice after the deregulation is comared to a hyothetical rice ath in absence of the deregulation. This alternative and hyothetical rice ath is based on the regulation guiding the Swedish ower rices before 996. The difference between the alternative rice and the revailing Swedish rice is then used to calculate the effects on consumer welfare, making it ossible to quantify the effects on consumer welfare from changes in the ower rice while excluding the imact of tax increases and other governmental interventions such as the effects of introducing green certificates. Worldwide deregulation in the electricity sectors began in the late seventies. In 978, Chile started to liberalize its electricity market introducing a wholesale market ool in which generators would sell their ower to retailers. In 982, the liberalization was exanded when large end-users were allowed to choose their retailer and negotiate rices. Deregulation of Euroean electricity markets started in England and Wales (E&W) in 990, when the industry was rivatized, and cometition between generators was introduced. The E&W deregulation was followed by the deregulation of the Norwegian market in 99. Throughout the 990 s deregulation continued and today most of the OECD-countries have to some extent liberalized their electricity markets (Al-Sunaidy and Green, 2006). Sweden deregulated its electricity market in 996 and joined Norway in the first international electricity market, Nord Pool, the same year. In recent years Nord Pool has exanded to also include Finland (998) and Denmark (2000). 4 Newbery and Pollitt (997), Green and McDaniel (998) and Domah and Pollitt (200) discuss the welfare effects of the E&W deregulation. Newbery and Pollitt (997) analyze the rivatization of the Central Electricity Generation Board (CEGB), the owner of the generation facilities and the highvoltage net, and conclude that the welfare effects of the rivatization will deend on the cororation tax level and the time san until rices have reverted to the re-deregulated level. Domah and Pollitt (200) analyze the 3 It may be that deregulation has generated some sill over effects to the regulated arts of the market. Damsgaard and Green (2005) assumed that transmission costs fell one ercent er year after 996 due to the deregulation. According to Damsgaard and Green, low rofits in generation in the late 990s forced the vertically integrated comanies to cut costs in distribution. Excluding these otential sill over effects will, if they exist, result in an underestimation of otential consumer welfare gains. 4 In 2005 art of Germany was included to Nord Pool when the KONTEK bidding area oened. 03

4 effects of the rivatization of twelve regional electricity comanies and find that the rivatization yielded significant net social benefits. However, they conclude that the benefits of the rivatization were unevenly distributed, both across time and between grous in society. The deregulation in 990 did not allow all electricity consumers to choose their electricity sulier, so it was not until 998 that the E&W market was oened u to full cometition. Green and McDaniel (998) use Cost-Benefit analysis (CBA) to analyze the social welfare effects of the 998 E&W restructuring. Green and McDaniel (998) argue that consumer rices will fall, and that electricity suliers will face high additional transactions costs in the first five years after the restructuring. Based on lower rices and additional transactions costs their main scenario shows that consumers will gain while roducers will lose. Producers will lose more than consumers gain so their results indicate a social welfare loss due to the restructuring. However, the welfare loss could be converted into a surlus if cometition forces the comanies to make further cost savings which would be larger than the transactions costs involved. Bowitz et al (2000) and Damsgaard and Green (2005) discuss the welfare effects of the Swedish deregulation. Damsgaard and Green (2005) study the effect of the Swedish deregulation using CBA. They build counterfactual scenarios based on how they believe the electricity rice and international trade with electricity would have evolved without deregulation. They then comare the counterfactual scenarios to the actual result. Their analysis focuses on the electricity rice end customers face by slitting customers into three categories: households, commercial and industry. The result in their main scenario shows that Sweden has benefitted from deregulation by a total of 0.9 billion SEK 5 u to Finally, Bowitz et al (2000) analyze the combined effect of the deregulations in Norway, Sweden and Finland. Bowitz et al (2000) conclude that until 999 deregulation had a negative welfare effect. They found that lower average rices in the revailing situation led to a ositive consumer surlus but a negative roducer surlus, which resulted in an annual net surlus of.5 billion NOK 6. However, the reason for the negative overall welfare effect was an environmental dimension that assumed that the ollution costs were higher than the ositive net value of consumer and roducer surlus. The rest of this aer is organized as follows. Section 2 discusses the theoretical method, section 3 describes the data and discusses the emirical 5 SEK equals aroximately 0.25 USD. 6 NOK equals aroximately 0.5 USD. 04

5 model. Section 4 contains results and section 5 includes discussion and conclusions. 2 Method This aer studies the change in consumer welfare due to the deregulation of the Swedish electricity market using an equivalent variation (EV) aroach. The difficulty with the EV aroach is to find the direct or indirect utility function in order to calculate the EV. Two alternative solutions to this roblem have been used in the literature. The first aroach starts with secifying the direct or the indirect utility function and then derives the observable demand functions by maximizing the direct utility function subject to a budget constraint or by integrating the indirect utility function using Roy s identity. The second aroach uses the observable Marshallian demand to secify the required demand systems. The second aroach is econometrically referable, as it will generate demand functions that fit the data well (Hausman, 98). In this aer, the latter aroach is used. We secify a linear Marshallian demand function for electricity based on model fit and earlier literature (e.g. Johnsen (998), Hjalmarsson (2000) and Steen (2003)). The demand function is reresented by x y i k, i,..., N, () where is the ower rice, y income and k a vector containing variables that affect electricity demand. Since the interest is on the demand for electricity, an incomlete demand model for electricity is secified. A comosite good is secified using the consumer rice index (CPI). The ower rice and income are deflated using the CPI, meaning that the comosite good disaears from the equation. This rocedure results in a demand function that is zero degree homogeneous in rice and income, and the corresonding quasi-indirect utility function and quasi-exenditure function will be exact welfare measures if they fulfil the integrability conditions (Hausman, 98). Using exenditure functions it is ossible to secify the EV measure and, if we consider a rice change from 0 to, the EV measure can be written 0, u e, u, EV e (2) 05

6 which can be used to quantify the effects on consumer welfare. Hausman (98) showed that under these conditions all that is needed to establish the exact measure of welfare changes is knowledge of the Marshallian demand function. Following Hausman (98) starting with equation () and using Roy s identity gives v, y x y ik, (3) v, y y which is a linear artial differential equation. In order to make welfare comarisons we need to stay on a given indifference curve, indeendent of the rice level. This is secured by using the equation vt, yt u where u is the utility level in the equivalent variation case. Differentiating v t, yt and setting it equal to zero will assure that we stay on the same indifference curve when the rice changes, that is v t, yt dt vt, yt dyt * dt y * dt 0. Rearranging equation (4), substituting into equation (3), and using the imlicit function theorem gives an exression which exresses y as a function of in an ordinary differential equation dy d (4) y k. (5) i Solving equation (5) gives the solution y Ae k, i (6) where A is the constant of the integration that deends on the initial level of utility. If we choose A u as the cardinal utility index we can solve equation (6) for the quasi-indirect utility function associated with the incomlete demand system 06

7 07., 0 0 k y e A y v i (7) Inverting equation (7) gives the quasi-exenditure function., k u e u e i (8) Before measuring the equivalent variation the integrability conditions should be checked to verify that the derived Marshallian demand function has its origin in a utility function. LaFrance and Hanemann (989) derived the necessary restrictions to obtain exact welfare measures from an incomlete demand model. They showed that by augmenting the incomlete demand system with a comosite commodity reresenting total exenditure on all other goods, it is ossible to roceed as if the augmented system was comlete. That is, for the demand function in equation () to be weakly integrable, equation (7) has to be continuous and homogeneous of degree zero in rices and income. Equation (7) also has to be decreasing in rice and increasing in income, 0 and 0. Finally, for equation () to be weakly integrable, the Slutsky substitution terms j i j i ij x y x x s * ) ( have to be symmetric and negative semidefinite (LaFrance and Hanemann, 989). When the integrability conditions are fulfilled it is straight forward to calculate the equivalent variation. Using equation (2) and (8) we find that in the case of a rice change from 0 to, the EV-measure can be written, y k k y e EV i i (9) where 0 indicates a ositive income effect.

8 3 - Data and emirical model 3. Data The data used in this aer covers the eriod January 996 to January The variable descrition and descritive statistics are dislayed in Table. Table Variables, descritive statistics, definition and data sources. Variable Mean* Descrition Price (06.9) Swedish area rice. SEK/MWh Source: Nord Pool ASA Consumtion.34E+ 07 (2.0E+06) Electricity consumtion in Sweden in MWh (including exort and imort). Source: Swedish Energy Agency 0.99 (0.3) Swedish industrial roduction index, SNI C-E. Source: Statistics Sweden Industrial roduction Temerature 7.04 (7.32) Average temerature in Sweden. Weighted average using average temerature (in C ) from Malmö, Stockholm, Östersund, Umeå and Luleå. Source: SMHI Length of day 2.34 (3.85) Number of hours the sun is above the horizon in Gothenburg, Sweden. Source: Inflow ( ) Inflow to the water reservoirs, recalculated into MWh. Source: Nord Pool ASA. Consumer Price Index.02 (0.003) Index of the consumer rices in Sweden. Source: Statistics Sweden. Income.9E+ (.78E+ 0) Gross Domestic Product in SEK. Source: Statistics Sweden. *Standard error in arenthesis All variables excet income are monthly observations and in real terms. Income is measured with real GDP which is not available on a monthly basis. To get monthly observations on income, GDP is interolated from yearly observations. 08

9 3.2 The demand model The variables in the demand model are the ower rice, income, temerature, length of day, and industrial roduction. The rice that is used is the Swedish area rice at the Nord Pool sot market, which is the rice that the Swedish retailers face when buying ower at Nord Pool. 7 The Nord Pool sot rice is the foundation for the rices the Swedish consumers face. 8 Due to mark us, green certificates and taxes, the correlation between the sot rice and the retail rices is not erfect. However, calculating the correlation between the sot rice and contract tyes for small and medium large consumers shows that the correlation between the sot rice and the retail rices is high. For the flexible rice contracts that retailers offer to costumers, a tye of contract that has become increasingly more common each year, the correlation is as high as 99 ercent. The average correlation between the sot rice and the observable tyes of contracts is Correlation for large consumers (large industries) are not ublically available but are generally believed to be at least as high. 0 In order to deal with the aarent endogenity roblem when estimating a demand model including both rices and quantities, we rely on an instrumental variable aroach. In this study, the ower rice is instrumented using inflow to the water reservoirs, temerature and the lagged ower rice as instruments. We estimate the demand model with OLS, using standard errors corrected for autocorrelation and heteroscedaticity. Income measures the urchasing ower of the consumers and higher income is exected to increase the demand for electricity. Temerature is measured in degrees 7 The sot rice is imortant not just as the rice at Nord Pool but also as a reference rice for bilateral and financial contracts. 8 Comaring the retail market in Sweden to the ower market, the Nord Pool market, these markets are quite similar in terms of market structure, with the four largest firms covering over 50 ercent of the market and over 00 other, small firms active in the market. 9 The observable tyes of contracts are one, two, and three year contracts, flexible rice contracts and standard rice contracts. The correlation between Nord Pool sot rice and the observable contracts are: one year 0.75, two year 0.69, three year 0.65, flexible rice 0.99 and standard rice contract Large consumers often negotiate their contracts bilaterally and therefore rices are not ublic. Large customers often have contracts with variable rices, but the contracts are secured by hedging. An alternative is fixed rice contracts with an otion to re-sell electricity if the electricity is not consumed. Both of these contract tyes imly that the Nord Pool sot rice is the relevant reference rice and that that the correlation between the large customer contracts and the Nord Pool sot rice should be at least as high as the correlation for the small and medium large costumers contracts. 09

10 Celsius and is used as a roxy for electricity demand for heating uroses. The climate in Sweden leads to a reverse electricity demand which, eaks in the winter (for heating use) rather then in the summer (for air conditioning use). Length of day is a roxy used to measure the demand for electricity for lighting. The winter days are shorter resulting in higher demand for electricity for lighting than in summer. The industrial roduction variable is used to measure the industry demand for electricity; a ositive correlation between industry roduction and the demand for electricity is exected. Finally, monthly dummy variables are incororated to account for any additional seasonality in electricity demand. 3.3 Alternative rice To construct the alternative rice ath it is imortant to understand how the re-deregulated rice was determined. Before deregulation, the electricity market was divided in two arts based on customer characteristics: a high and low voltage market. The high voltage market was a market for generators, large industries and retail distributors, while the low voltage market was a market for retail distributors and their consumers (Bergman et al, 994). In addition to these markets there was a market for temorary energy exchange for generators. This market worked like an "electric ower ool" and was characterized by extensive cooeration and information exchange between the articiants. Generators with a shortage (surlus) of ower could buy (sell) ower from the "ower ool". The exchange rice was based on the so called slit-savings rincile, a method where the rice was determined by the average of the buying firms marginal cost and the selling firms marginal cost (Hjalmarsson, 993). Before deregulation, the electricity market was characterized by local and regional monoolies in both generation and distribution. There was no cometition between generators or between distributors (SPK 989:8). The high voltage market, where the state owned comany Vattenfall was the dominating actor with 50 ercent of the generation caacity, was not directly regulated. Pricing was, however, indirectly regulated through state-ownershi of Vattenfall and the formal objective of Vattenfall to break even subject to a required return. This established Vattenfall as the rice leader and also worked to set a rice ceiling. The generators collaboration in the "ower ool" resulted in a rice floor which also revented further entry to the market The required return was equal to the dereciation on relacement values and a rate of interest on loans from the government at the bond rate level. 0

11 (Bergman et al, 994). The ricing in the high voltage market was conducted through high voltage tariffs consisting of four elements: a fixed fee, a contractual demand charge, a eak load charge and an energy charge (SPK, 989:8). The energy and the eak load charge were riced by marginal cost while the contractual charge and the fixed charge were determined by the rate of return constraint. The four elements of the high voltage tariff can be aggregated into three variables: transmission, energy and a fixed fee. In turn, the fixed fee and transmission are variables describing the network costs. In reality, network costs have been constant in real terms (see Figure ). This leaves the energy art of the market as the remaining variable to exlain changes in rices in the high voltage market. SEK/ MWh Figure Price decomosition, Taxes Energy Network The slit-saving rincile used in the ower ool before deregulation imly that the rice on ower could be described as an average cost of roduction. Damsgaard and Green (2005) also argue that the required rate of return restriction on state- and municiality-owned generators makes average cost ricing a reasonable aroximation of the ricing rincile on the high voltage market before deregulation. In this aer, average cost of electricity roduction is measured by the roducer rice index (PPI) for electricity roduction, and since the ricing rincile in the regulated market was aroximately equal to average cost ricing, the PPI for electricity roduction will be used to derive the alternative rice ath if the market had not been deregulated. To convert the PPI from index form to a rice ath, the

12 January 996 ower rice will be used as the base. That is, the January 996 ower rice is multilied with the PPI for each month to get a rice ath that shows what the ower rice would have been without the deregulation. However, a requirement for the revailing PPI to work as an alternative rice ath is that the roduction structure has not changed significantly during the years under study. Table 2 shows the roduction structure before and after the deregulation, and it shows that the roduction structure has remained relatively constant during these years. Table 2 Electricity roduction in Sweden , roortions. Year Hydro Nuclear Wind Other* Net imort** Source: Swedish Energy Agency. *Include Industrial back-ressure ower, Combined heat and ower, Cold condensing ower and Gas turbines. **Positive sign equals net imort, negative sign net exort. Before 996, hydro and nuclear ower reresented on average ercent of the generation. This has not changed after the deregulation. Note, however, that the year 996 was a dry year with relatively little reciitation resulting in limited hydro ower roduction, while the year 2000 was a wet 2

13 year with relatively high reciitation and hydro ower roduction. In addition, the year 2003 was a very dry year with all time low levels in the water reservoirs. 2 As such, hydro ower roduction has varied deending on the reciitation, but this does not imly a change in the roduction structure. In other words, reduced reciitation leads to increased nuclear ower roduction both before and after deregulation Emirical EV-measure To comute the effects of the deregulation on consumer welfare, the alternative rice is set against the revailing Swedish area rice from the deregulated eriod. The estimated arameters from the demand function are used together with the alternative rice in equation (0) to calculate the effects on consumer welfare. 0 alt 0 2 EV e y alt tem j dummy j 2 (0) tem j dummy y. j 2 The EV measure is calculated for each month in the samle and then aggregated over the whole eriod. 4 Results 4. The demand model The estimation is erformed using a general to secific model aroach, where variables whose arameters are not statistically significant 2 The reciitation attern is reflected in the ower rice. Low reservoir levels in 996 and 2003, resulted in higher ower rices since more exensive roduction sources were used to meet demand. The higher rices in 2003 comared to 996 may have been a result of the more exensive energy sources in Denmark. The even higher rices during 2005 and 2006 may have been caused by the introduction of the Euroean market for CO 2 allowances. 3 Net imort of electricity (see Table 2) has also increased since deregulation. In this study the net imort of electricity, which is four ercent of market suly on average, will not be taken into account. 3

14 are removed from the model. The estimation result for the demand model is resented in Table 3. Table 3 Estimation results of the demand function. Variables Coefficient Standard Error Constant 37500*.72E+ 06 Price * Temerature * Income 2.97E- 05* 9.40E- 06 dfebruary * dmarch * daril * dmay * djune * djuly * daugust * dsetember * doctober * dnovember * ddecember * R 2 = 0.97 Adjusted R 2 = 0.96 N= 32 SBIC= Log-Likelihood= Ljung-Box Q-statistics Statistic Critical value Q() Q(2) Q(4) Q(8) Q(2) * Significant at the % level 4

15 All coefficients have the exected signs. The rice has a negative sign meaning that a higher rice reduces electricity demand. Increased income increases the demand for electricity. Warmer temeratures lead to reduced demand for electricity for heating. Length of day should intuitively have the same sign as temerature: the longer the day the less electricity required for lightning. However, we remove this variable from the demand model due to a high correlation with temerature (83 ercent), causing multicollinearity roblems. In addition, industrial roduction is not significant and is therefore excluded from the final demand model. To account for seasonal effects, monthly dummy variables for all months excet January are used. The resulting negative signs tell us that the electricity demand is highest in January and lowest in the summer. The long run elasticity of demand for ower is calculated and equals , which is in line with revious studies. 4 Finally, the results from the estimation of the demand model show that all integrability conditions are fulfilled. The Slutsky substitution terms are negative definite, the income arameter is ositive and the rice arameter is negative. 4.2 Welfare effects To measure the effects on consumer welfare, the revailing Swedish area rice is comared to the alternative rice ath described in section 3.3. The resulting monthly EV-measures are discounted back to 996 using a discount rate of 5 ercent and then added together. The aggregated EV measure tells us that the deregulation has been advantageous for electricity consumers in Sweden. In total, consumer welfare has increased by 4 billion SEK in 996 rices over the eriod studied. 5 The reason for this result is that while the rice ath under deregulation has been volatile with both dis and eaks, the alternative rice has been more stable but at a higher rice level (see Figure 2 6 ). 4 For an extensive discussion about the elasticity of demand for electricity, see Nilsson and Pettersson, (2008). 5 Discounting the EV-measure to 2007 does not qualitatively alter the results. The results dro from 4 billion SEK to 3.8 billion SEK. 6 The underlying data used in the calculations are available from the author on request. 5

16 Figure 2 Price comarison, Nord Pool Price Alternative Price 500 SEK/MWh jan-96 jul-96 jan-97 jul-97 jan-98 jul-98 jan-99 jul-99 jan-00 jul-00 jan-0 jul-0 jan-02 jul-02 jan-03 jul-03 jan-04 jul-04 jan-05 jul-05 jan-06 jul-06 jan-07 The less volatile alternative rice ath is due to differences in the ricing methods. The deregulated rice is more volatile because the rice is determined by suly and demand in a cometitive environment, and set equal to the cost of the marginal unit of energy necessary to meet demand, while the re-deregulated rice was based on the average cost of roduction in the different generation facilities. 7 The average cost based rice does not fluctuate as much as the marginal unit based rice because a sudden demand increase (decrease) and the subsequent rice jum (di) is averaged out by the other facilities. In the deregulated case, a sudden demand increase will allow those generation facilities with a marginal cost lower than the marginal facility, to increase their rofits. To consider the ossibility that the alternative rice ath calculated using the PPI starts from an incorrect level, for examle due to 996 being a dry year, a sensitivity analysis is erformed. The PPI is reconstructed to start both from a rice level 5 ercent below and 5 ercent above the base scenario. The sensitivity analysis shows that a 5 ercent lower initial rice will lower the increase in consumer welfare from 4 billion SEK to 7 The cometitiveness of the Nord Pool market has been studied by Hjalmarsson (2000) and Bask et al. (2009). Both studies show that the Nord Pool sot rice has been very close to the marginal cost of roduction after the deregulation. 6

17 653 million SEK, and that a 5 ercent increase in the start level will imrove the welfare effect to 7.3 billion SEK. However, note that the welfare effects are ositive in the sensitivity analysis as well, indicating that the ositive welfare effect calculated in the base scenario is quite robust to changes in the initial rice level. Comaring the results from this study to earlier studies of the deregulation of the Swedish electricity market (e.g. Damsgaard and Green, 2005 and Bowitz et al., 2000) reveals that these studies also found that the deregulation increased consumer welfare. The studies are made using different aroaches and different datasets, although the results are qualitatively the same. This gives further suort to the results resented in this aer. It might be interesting to set the increase in consumer welfare in ersective. An increase in consumer welfare of 4 billion SEK is a lot of money, but sread over the years of deregulation covered in this aer, and the aroximately 9 million consumers (households, commercials and industry), the consumer welfare gains equal less then 00 SEK er consumer er year. The average electricity use er caita in Sweden in 2003 was aroximately kwh (Statistics Sweden). Given an average Swedish ower rice of 0.22 SEK/kWh, the cost for ower er caita would be aroximately SEK er year, indicating a welfare gain just below three ercent. In this context the magnitude of consumer welfare gain from deregulation may aear small, but the results are consistent with revious studies: deregulation has benefitted the tyical Swedish electricity consumer. 5 - Conclusion The urose of this aer is to calculate the effects on consumer welfare due to the deregulation of the Swedish electricity market in 996. The analysis is erformed using an equivalent variation method and an alternative rice ath based on how the ower rice was set before the deregulation. When this alternative rice ath is comared to the revailing Swedish area rice it is clear that the deregulated rice a rice based on marginal cost ricing has been lower, on average, during the studied eriod, leading to an increase in consumer welfare. The results from the base scenario in this aer show that the deregulation has increased consumer welfare by 4 billion SEK. Sensitivity analysis using alternative scenarios reinforces the conclusion that the deregulation has increased consumer welfare. Set into ersective, the welfare gains from deregulation is not that imressive, less than 00 SEK er 7

18 consumer er year. However the deregulation effects are ositive and the consumer welfare has imroved. The results from this study can be comared to revious studies of the welfare effects of the deregulation of the Swedish electricity market. Bowitz et al (2000) found that the Nordic deregulation had increased consumer welfare by 5.6 billion NOK 8 er year between the eriod 996 to 999. Disaggregation indicates that the Swedish consumers gained.3 billion NOK er year after the deregulation. If this trend is extraolated until 2006, this equals a welfare gain for consumers of 4.3 billion NOK. It should, however, be noted that extraolating robably overestimates the welfare effects. This is because the Nord Pool rice has been very high during 2003 and 2006, and high rices will decrease the welfare gain from the deregulation. Damsgaard and Green (2005) calculate several different scenarios, all showing ositive effects of the deregulation. Their main scenario estimates welfare gains for all electricity consumers to be 4.2 billion SEK in total for the eriod 996 to It should be noted that Damsgaard and Green (2005) use the electricity rice faced by consumers, including taxes and environmental fees, while this aer uses the Swedish ower rice on Nord Pool. In this aer, the market effects of the deregulation have thus been searated from the olitical decisions that affect the rice of electricity. Using this aroach, it is clear that the increased rice of electricity for consumers after 996 is not due to the deregulation as such, but rather due to olitical decisions to increase energy taxes and to introduce green certificates. References Al-Sunaidy, A. and R.J. Green, Electricity Deregulation in OECD Countries, Energy, 3, Bask, M., J. Lundgren and N. Rudholm, Market Power in the Exanding Nordic Power Market. Alied Economics (forthcoming). Bergman, L., T. Hartman, L. Hjalmarsson and S. Lundgren, 994. Den nya elmarknaden. (SNS Förlag, Stockholm). Bowitz, E., T. Bye, O. Rosnes and H. Vennemo, (2000). The Nordic Electricity Reform: Economic and Environmental Consequences. Working Paer 3/00, ECON Centre for Economic Analysis. 8 NOK equals aroximately.2 SEK 8

19 Damsgaard, N. and R.J. Green, Den nya elmarknaden- Framgång eller misslyckande? (SNS Förlag, Stockholm). Domah, P. and M.G Pollitt, 200. The Restructuring and Privatisation of Electricity Distribution and Suly Business in England and Wales: A Social Cost-Benefit Analysis. Fiscal Studies, 22 (), Green, R.J. and T. McDaniel, 998. Cometition in Electricity Suly: Will 998 be Worth it? Fiscal Studies, 9 (3), Hausman, J., 98. Exact Consumer s Surlus and Deadweight Loss. American Economic Review, 7, Hjalmarsson, E., Nord Pool: A Power Market without Market Power. Discussion Paer No. 28, Gothenburg University. Hjalmarsson, L., 993. From Club-Regulation to Market Cometition in the Scandinavian Electricity Suly Industry. Memorandum no 93, Deartment of Economics, Gothenburg University. Johnsen, T.A., 998. Modelling the Norwegian and Nordic Electricity Markets. Dissertation, University of Oslo. LaFrance, J.T. and W.M. Hanemann, 989. The Dual Structure of Incomlete Demand Systems. American Journal of Agricultural Economics, 7, Newbery, D.M. and M.G. Pollitt, 997. The Privatisation and Restructuring of the CEGB-Was it Worth it? Journal of Industrial Economics, 45 (3), Nilsson, M., and F. Pettersson, Surrise, surrise The Customer Acts Like One. A Short and Painless Killing of a Long-lived Myth, Working Paer, Luleå University of Technology. SPKs raortserie, 989. Elmarknaden, ris- och konkurrensförhållanden R 989:8. (Statens ris- och konkurrensverk, Stockholm). Steen, F., Do Bottlenecks Generate Market Power? An Emirical Study of the Norwegian Electricity Market. Discussion Paer No. 26, Norwegian School of Economics and Business Administration. 9