ExternE-Pol Externalities of Energy: Extension of ... · prices. Energy demand derives from transport (accounts for 67% of EU oil use), households ... reasons to energy insecurity

ExternE-Pol Externalities of Energy:

Extension of Accounting Framework and Policy Applications (Contract N° ENG1-CT-2002-00609)

Final Report on Work Package 3

The Externalities of Energy Insecurity

Anil Markandya and Alistair Hunt, University of Bath, Bath, UK

30 November 2004

Introduction The overall objective of this work package is: To estimate the external costs due to risk of quantity and price disruptions of energy supply (energy insecurity) for the European Union (EU); We define energy security as “a state in which consumers and their governments believe, and have reason to believe, that there are adequate reserves and production and distribution facilities available to meet their requirements in the foreseeable futures, from sources at home and abroad, at costs which do not put them at a competitive disadvantage or otherwise threaten their well-being. Insecurity arises as a result of physical failure of supplies or as a result of sudden and major price changes” (Belgrave, 1987 cited in Lockwood, 1997). This accords with the International Energy Agency definition as being the “availability of regular supply of energy at a reasonable price” (IEA: 2001). In the recent past there have been a number of disruptions in supply and in the price of energy. This has led to more focused attention to the concept of ‘energy security’. Most notably, the European Commission Green Paper (EC, 2002) concentrates on the need for reduced energy import dependence in order to reduce energy insecurity. This policy context is presented in broad terms in Box 1. This research responds to that policy need. Box 1: EU Energy Policy context The EC Green Paper was published in order to stimulate discussion on policy design, given the overall policy objective of ensuring guaranteed energy supply at predictable prices. Energy demand derives from transport (accounts for 67% of EU oil use), households (majority of demand is for gas) and industry (mainly electricity derived from gas, coal, oil).

Energy Supply of oil: Middle East is major supplier; EU supplies limited; gas: Russia is major supplier, also N. Africa; EU supplies limited; coal: EU uncompetitive; China is major supplier; renewables are seen as having limited competitiveness without more R & D The principal policy issues are: 1) Under current trends in demand and supply a greater dependence on energy supplies from outside the EU is likely. Why is this a problem? There is vulnerability to:

- world market instability – greater price fluctuations/variance

- Exporting countries political/social instability or supply manipulation - Transit countries’ political/social instability ⇒ price variance

The policy objective is therefore to reduce extra-EU dependence of energy sources. Policy responses may include diversification of sources of supply; promoting the build-up and co-ordination of strategic stocks of oil, including through increased integration of trans-European networks; encouraging use of EU Renewables; reducing energy demand, and; developing more attractive agreements to supplier and transit countries.

2) Environmental regulation of energy as a result of externalities and particularly climate change – Kyoto obligations to reduce carbon emissions. The policy objective is to reduce pollution emissions whilst policy responses include: increasing energy efficiency in use; reduction in demand e.g. through carbon levy etc., and; encouraging use of EU renewables.

3) Creation of competitive internal market to lower prices and enhance economic competitiveness. The policy objective in this case is economic efficiency whilst policy responses may include liberalisation of EU energy markets and establishment of trans-European energy networks. For European policy makers, energy security is an important issue because private decisions about energy use may not fully take into account the costs of energy insecurity. Disruptions in supply and dramatic price increases have macroeconomic impacts that individuals/firms do not take into account. Furthermore, agents tend to underestimate the risks of disruption and subsequent price adjustments, and there are other less tangible effects such as the psychological costs of people feeling insecure about their energy supplies. Therefore, it is important from a policy perspective, to estimate the size of the external costs of energy arising from energy insecurity. This policy-focussed research fits within an overall research programme for energy security that is suggested in Figure 1, below. Aspects of all phases of this research programme are addressed in the current research.

Methodology Previous work (e.g. European Commission (1999)) identified three potential kinds of externality associated with energy security: These are:

- Monopsony wedge externality e.g. when strategically increased imports of a fuel cause price to rise. In this case the cost of an additional unit of the fuel to the entire economy is higher than it is to the individual importer. Alternatively, imports may be restricted in order to lower price. Note that there is no market failure in terms of global inefficiency, but resulting financial transfers out of country may be of concern.

- Incomplete rent capture e.g. when decisions of energy suppliers changes the

value of economic rent (consumer surplus) that the consumer gains, due to the inability of the supplier to capture the full rent from consumers.

- Macroeconomic Externalities, where there are effects of decisions in the energy

market on prices and quantities in markets other than the energy market. This research found major limitations in techniques available for measuring the first two types in quantitative terms. For this reason, and because this original research suggested that macroeconomic externalities were likely to be dominant, we focus on this type in the current research. Below, we report on a literature review of the impacts of energy insecurity relating to non-nuclear energy sources including oil, gas, coal and electricity supply, focussing on the EU region. On the basis of the findings of this review we make first estimates of the ‘externality unit values’ for energy insecurities. The majority of empirical work has been in relation to oil and electricity supply, and our work reflects this. However, in order to compare alternative energy sources and technologies it would be important to have quantitative estimates for other energy sources. As far as the evidence allows, we report estimates for all energy sources. We also report the results of a policy modelling exercise that we undertook to identify appropriate policy instruments for the internalisation of energy insecurity into energy pricing, taking into account the additional constraint of climate change mitigation policy. Figure 1: Framework for Energy Security Research B. Identifying the Reasons for Concern about

Energy Security

A. Identifying the Types of Energy Insecurity Among the Population

As part of this project, some development work was also undertaken on the design of a survey that would estimate willingness to pay for economic agents to avoid energy insecurity risks. However, insufficient co-funding was available to carry out the survey. We therefore flag this exercise for future research efforts in this area. Macroeconomic costs of energy insecurity In order to understand the policy context in which energy security externalities are positioned, it is useful to be able to categorise the sources of energy insecurity. These categories may include random shocks e.g. terrorism; and strategic shocks e.g. OPEC manipulating the quantity and therefore price of crude oil. Both these shocks ultimately lead into significant changes in energy prices, and this is the focus of the next section. Since the 1950s, there have been a number of oil shocks. Figure 1 illustrates the series of supply shortfalls from the Suez crisis in 1957, all the way to the Venezuelan strike in 2002. As one can see, these are all exogenous shocks i.e. they have all been driven by political events in countries outside of the EU. Figure 2: Magnitude of oil supply shortfall (mb/d) 1957 to 2002

C. Measuring the Value of Actions to Reduce Energy Insecurity

D. Identifying Policies that Reduce Energy Insecurity

E. Linking these policy measures to the underlying reasons to energy insecurity and thus comparing the costs of policy measures with the values of changes in energy security.

Source: Harks 2003, in Costantini & Gracceva, 2004:2 In understanding the nature of externalities associated with energy security it is useful to distinguish between two types of externalities that generate external costs: technological externality – when the actions of an economic agent affect the welfare of another, other than by affecting prices; and a pecuniary externality – when the actions of one economic agent affects the welfare of another through price changes. Even though energy shocks can involve physical disruptions, the impact of these also comes through dramatic effects on price. Therefore, the external costs associated with insecurity of energy supply are thought to be primarily pecuniary externalities. The mechanisms through which pecuniary externalities are thought to arise are described in the paragraphs below. However, as is also reflected below, the classification of all macroeconomic costs as externalities should not be readily accepted and the identification of what is, or is not, an externality is important in determining appropriate policy responses. Outline of basic macro-economic mechanisms associated with energy market change Macro-economic consequences of energy market changes have to date largely been the consequences of activities in the world oil market. Thus, the following analysis has its empirical basis in that experience. In principle, however, the mechanisms would apply to other fuel sources were there to be similar dependence on other imported fuels in the energy mix. Below, we start by outlining the mechanism by which oil price increases bring about macro-economic changes. As shown below, the mechanism is not identical to that resulting from oil price volatility. However, it does allow us to illustrate the problem of defining what is an externality and what is internalised. Oil price increases and the macro-economy linkages We assume – in line with the majority of economic modelling of energy prices and their macro-economic linkages - that the following line of causation exists:

1) Payment for oil imports results in a worsening trade balance for an importing country, (since higher prices result in an increase in total payments - assuming price inelasticity); 2) The consequent current account and balance of payments deficits, and associated depreciation of the exchange rate, makes other imports more costly. 3) Higher import costs may lead to higher price levels and inflation; higher unemployment may result from the transfer of resources needed to pay for the oil imports; lower GDP may result. This is a simplification of the central mechanism. There is considerable discussion, (e.g. in Bohi and Toman (1996)), about a number of complicating factors in the operation of such a mechanism. For example, the terms of trade effects may be positive or negative. If a country is reliant and inflexible with regard to its use of oil there will be a greater decline in home output relative to other countries. Therefore whilst the shortage of home-produced goods in world markets will lead to a relative shortage and therefore improved terms of trade, the economy will still worsen. Bohi and Toman (1996) also argue that the relationship between a country’s exchange rate and price of oil is ambiguous. After an initial current account deficit from an oil price rise, the effect on the exchange rate in oil importing countries depends on the willingness of oil exporting countries to hold different foreign currencies. If they prefer to hold more dollars (or euro) than other currencies, the dollar exchange rate will rise. Additionally, whilst it is acknowledged that higher oil prices raise prices, unless these prices continue to rise there will not be on-going inflation. Thus, whilst energy price increases may aggravate existing inflationary processes, they should not necessarily be seen to be the cause of continuing inflation. Energy security externalities: Identification problems With these caveats in mind, the following sections provide a review of theoretical and empirical investigations into the macroeconomic effects of oil price shocks. These investigations do not distinguish between internalized and externalized costs although there are reasons to think that this distinction is important. For example, Bohi and Toman (1993) make the following argument. They state that in the long run, in response to an energy price change, adjustments take place in the amount of energy-related investment and rate of innovation. Thus, the long run effects on productivity are internalized. In the short run, they argue that the macroeconomic effects arise from slow adjustment in the factor and product markets. First, real wages may not adjust to maintain employment when energy prices rise. A rise in energy prices reduces demand for energy and when energy and labour are complementary, this will lower the marginal product of labour. Lower productivity means an increase in the unit cost of labour and employers will seek to reduce the level of employment in response if the wage rate does not adjust to reduce the unit cost. A reduction of employment will lead then to a reduction of overall output. In a similar way,

the resulting reduction in capital services makes some energy inefficient capital goods superfluous, either from competition, or if the demand for more expensive energy-intensive end products declines. Again, there is a reduction in productive capability and therefore output. These two influences – on the labour and capital markets - may be exacerbated if it is also difficult to reallocate factors in response to changes in the mix of final demand brought about by changes in product prices. From this analysis rises the question: are these effects on employment and output externalities? The initial price increase is simply representative of a change in resource cost resulting from the need to pay for imports demanded, and the re-alignment of domestic resources. So in this case it seems that costs are fully internalized in private decisions. It is clear, also, that output loss and unemployment may be interpreted either as resulting from imperfections in factor market adjustments or that these factor markets may be operating as well as possible given real world institutional constraints. Whichever interpretation we choose, externalities exist to the extent that parties to labour and capital transactions cannot fully avail themselves of the means to anticipate and respond to energy price shocks, whilst if they are anticipated and coped with, effects of energy price variability would be internalized. There also appears to be the threat of government – rather than market – failure that contributes to the macroeconomic impacts of energy price increases (Bohi and Toman, 1993). For example, because of regulation, an increase in oil prices will not lead to efficient adjustments in gas and electricity prices. These price rigidities may cause adjustment problems throughout economy where regulation cannot simultaneously constrain market power and allow regulated prices to adjust to market conditions. The preceding paragraphs have made the case that energy security externalities arising from macroeconomic consequences of energy price changes result to the extent that there are informational, and factor and product market, failures. We make estimates of these externalities in the following sections and highlight some issues relating to the policy implications in the concluding section. The externality estimates are based on the total macroeconomic costs associated with given changes in energy markets. The estimates are derived directly from a review of the historically observed macroeconomic effects of these changes, as well as changes that are simulated in macroeconomic models. In the following analysis we make a distinction between two types of price movement that have a bearing to energy security and macro-economic effects: sudden increases in price and volatility of oil prices. Oil Price Movements Increase in Price A review of the theoretical literature suggests that there are a number of channels that could contribute to an inverse relationship between oil prices and economic activity. For example, IMF (2000) proposes five channels through which higher oil prices might affect the global economy: transfer of income from oil consumers to producers; rise in costs of production of goods and services due to increases in price of inputs; inflation; direct and indirect impacts on the financial markets, with subsequent effects on interest rates; and

changes in relative prices. The consequences of such channels operating are expressed in macro-economic terms through e.g. losses of GDP (due to general increasing costs of supply); losses via negative balance of payments (due to increasing import prices); and through a rise in inflation (a sudden response to oil price increases due to direct market transmission mechanisms) and interest rates, reduced non-oil demand; lower investment in net oil importing countries. There are also a number of indirect impacts which are felt through the resulting effect of a reduction in tax revenues and increasing budget deficits. It is also argued that economic and energy-policy responses can exacerbate the negative impacts of an oil price shock (IMF, 2000). Here, we solely focus on the changes in GDP, which we interpret as expressing changes in welfare and hence can be interpreted as equivalent to other measures of externalities. Macroeconomic costs of energy price increases: Empirical evidence In order to substantiate the belief – constructed from the theoretical literature - that energy supply insecurity might give rise to welfare costs we reviewed the empirical evidence on linkages between oil price shocks and macro-economic consequences. The events, as illustrated in figure 2 above, have led not only to supply shortfalls, but also to increases in oil prices. For instance, in 1973-1974, when the Organization of Petroleum Exporting Countries (OPEC) first imposed an oil embargo and then greatly increased crude oil prices, the price of the barrel increased from $3.4 to $13.4; in 1978-1979, after disruption in oil supplies as a result of the Iranian revolution, the price rose from $20 to $30; and following Iraq’s invasion of Kuwait in 1990, oil prices went from $16 to $26; finally, and for the oil shock in 1999, prices went up from $12 to $24. A cursory glance at the statistical evidence – presented in Figure 2 below - illustrates how movements in oil prices correspond with GDP growth rates in the EU, between 1970 and 2001. It is apparent from this that there is some correlation between a higher oil price and lower GDP growth rates with a one to two year lag. Figure 3: Oil price and GDP growth rate in the European Union, 1970 - 2001

Source: Costantini & Gracceva, 2004: based on data from the World Bank (GDP growth rate) and EconStat (oil price). Other studies have confirmed this impression e.g. Hamilton (2001), which provides a literature review of the relationship between energy prices and aggregate output or expenditure, and Brown & Yucel (2000). Huntington (1983: cited in Brown and Yucel, 2000) showed that all (but one) post World War II recessions were preceded by substantial increases in oil price, and that other variables could not account for this.

Additionally, Muellbauer and Nunziata (2001); in Hamilton, 2001) successfully predicted the US recession of 2001 from a multivariate analysis which included oil prices. However, there are a number of factors that lead to a high variance in the economic costs of oil price increases, summarised in Costantini & Gracceva, 2004:10). These include:

• The role of fiscal and monetary policy responses. (IEA, (2001)), for example, argues that inappropriate policy responses to oil shocks have led to recessions, where, for instance, highly contractionary monetary policies and fiscal policies to contain inflation reduce national income and increase unemployment.

• Level and duration of the oil price increase - effects are greater the more sudden and pronounced the increase in price;

• Response of oil market. Whether – in the face of a price shock from one source – other suppliers can and do act to alleviate the impact;

• Amount of oil reserves available at the national level • Import dependence • Features of the individual national economy, including the weight of energy

costs in GDP, the share of energy intensive sector in industry and the prevailing macro-economic state;

• Flexibility of energy sector i.e. capacity to shift from one fuel to another Consequently, when analyzing the differential effects of oil prices on each of the countries within the European Union (Cunado et al (2000)) found significant differences between countries. They analysed the relationship between oil prices and economic activity for most of the European countries using quarterly data for the period 1960 to 1999. They found that the impact of increasing oil prices is higher when using national oil prices measured in national currency as opposed to the world oil price index, due to the role of the exchange rate. They found that the relationship between price and economic activity was only limited to the short-run, with no co-integrating relationship in the long-run. However, with regards to the impact on inflation there was evidence of co-integration for all countries except Germany, Luxembourg, Netherlands and Sweden. They also found that oil prices affect real activity not only through affecting inflation rates but by some other mechanism. In fact, there seems to be some evidence that the relationship between price and economic activity has weakened since the mid-1990s. Various explanations have been put forward for this, including reductions in the energy consumption to GDP ratio, and better experience in policy responses, particularly with regards to monetary policy. However, Cunado et al (2000) used new proxy variables of oil price shocks, and found asymmetric effects in the impact of oil prices on economic activity. In fact, while oil price increases have a negative and significant effect on IPI growth rates, the opposite result does not hold for oil price decreases. Furthermore, quantitative estimates of overall macro-economic damage vary substantially. Therefore, several, e.g. Hamilton (2001), have argued that the apparent relationship between oil price and economic activity is heavily reliant on the functional forms rather than a weakening relationship: a non-linear

specification of the oil price macro-economy relationship, may be most appropriate due to apparent asymmetries found between oil price increases and decreases (see e.g. Hamilton (2001)). Based on this empirical evidence, different asymmetric and non-linear transformations of oil prices have been proposed in order to evaluate the possible non-linearities of the oil price-macroeconomy relationship (Lee et al. (1995); Hamilton, (1996) etc). These immediately preceding paragraphs serve to indicate that the link between energy (oil) price increases and economic performance are not linear functions, if they exist at all. This conclusion therefore signals that the confidence in which we can have in estimates of externalities based on macroeconomic costs is diminished. However, it is hard to quantify the degree of uncertainty surrounding the macro-economic estimates. As shown below we therefore rely on interval analysis informed by our own subjective judgement to communicate this uncertainty. Price Volatility We define price volatility as being the standard deviation of return/price, which measures how widely actual values are dispersed from the average. The larger the difference between the actual value from the average value, the higher the standard deviation will be and the higher volatility. Oil prices have become more volatile over the medium and long-term since the mid-1990s, (IEA, 2001), and in part, this is explained by the fact that OPEC has changed from setting the price and letting production fluctuate to setting production quotas and letting the price fluctuate. Its possible importance as source of external costs stems from the fact that during periods of high volatility, the level of oil prices contains little information about future oil prices. This acts as a disincentive for rational agents to invest since the uncertainty associated with future returns is higher. This increasing uncertainty can push up risk premiums, and discourage oil companies investment. Thus, high volatility can be seen as a barrier to investment in the oil industry. Figure 4 below shows the evolution of the monthly average price of daily Brent for the last twelve years. Volatility is a measurement of change in price over a given period. The standard definition for volatility is the standard deviation of return/price, which measures how widely actual values are dispersed from the average. The larger the difference between the actual value from the average value, the higher the standard deviation will be and the higher volatility. It is apparent that prices have become more volatile over the medium and long-term since the mid-1990s. Since 1996, prices have tended to fluctuate much more on a monthly and annual basis. Between October 1997 and February 1999, oil prices fell from 23 $/b to as low as 10 $/b. Prices started to increase in March 1999, reaching 30$/b at the beginning of 2000. Analysis of short-term price movements shows that oil prices have become more volatile on a daily and weekly basis since the Gulf War (IEA, 2001).

Figure 4: Price volatility of dated Brent, 1987-2001

Source: IEA, 2001 The empirical work supports this hypothesis. For example, IEA (2001) tested the hypothesis empirically through a regression analysis of the annual change in the amount of investment in global oil exploration and production expressed as a function of both changes in the annual aggregates of daily oil price volatility and annual changes in price levels. Volatility and price changes were lagged by one year to take into account the full impact of price changes on investment decisions. The results suggest that the level of prices is the key determinant of exploration and production investments, although the degree of price volatility has a significant impact. It points to a strong inverse relationship between the investment changes and volatility, implying that an increase in volatility results in a decline in investments and vice-versa. These findings are supported also by Ferderer (1996; in Hamilton, 2001) whereby oil price volatility was found to depress spending and investment through its effects on uncertainty (Ferderer, 1996). In particular the analysis showed that the elasticity of investment with respect to changes in volatility is -0.11, while it is 0.44 against price change. In other words, a one percent increase in volatility is associated with 0.11 percent decline in investment, when other things such as interest rates are kept constant, and a one percent increase in price results in a 0.44 percent increase in investment. Results from Predictive models Several models have been developed to estimate the impact of energy price increases and/or supply disruptions on the macro-economy and these have largely focused on oil supplies. This section briefly reviews the outputs of a number of these models in the process of identifying appropriate estimates of macroeconomic costs of energy price changes.

The European Commission, in a communication to the European Parliament and the Council (EC (2002)), reported on an informal meta-analysis of economic studies undertaken that estimated that an increase of $10 in the price of a barrel of crude oil over a 3-month time period is likely to reduce economic growth in the industrialised countries by around 0.5% per annum. For developing countries, this would reduce economic growth by around 0.75% per annum. However, it notes that the impact of oil prices on economic growth is not linear (as discussed above): sudden unexpected and sharp increases in prices are likely to cause much greater impact to the economy than these estimates. IEA (2004) simulated the impact of a sustained increase in the price per barrel of crude oil by $10 – from $25 to $35 – over the whole of a five year period from 2004 to 2008, using the OECD’s in-house macro-economic model (Interlink). Crucially, nominal dollar exchange rates are held constant at late-2003 levels in both cases. In practice, any change in the value of the dollar would significantly affect the impact of higher nominal oil prices on the global economy. For countries in the Euro-zone, which are highly dependent on oil imports, the impacts would be worst in the short term. Estimates for the impact on GDP show a drop by 0.5% per annum, and inflation rising by 0.5% in 2004, with unemployment rising also. These GDP losses would exacerbate budget deficits, which are already large (close to 3% on average in the euro-zone). Interestingly, the United States would suffer the least, largely because oil dependency is relatively less - national production meets over 40% of its oil needs. However, unemployment would worsen significantly in the short term. Figure 6 shows how the impacts of a price increase vary between the US, Japan and the Euro-zone. Figure 6: OECD Macro-economic indicators in sustained higher oil price case, by region/country

Note: Oil prices are assumed to be $10/barrel higher than in base case. Source: IEA/OECD analysis.

IMF (2000) ran simulations of higher oil prices on macro-economic variables. Several simulations of a sustained $5 per barrel (20 percent) increase in the price of oil were run using the MULTIMOD model, focusing on the implications for real GDP and inflation. The results indicate that a permanent $5 per barrel increase in the price of oil would reduce the level of global output by around ¼ percentage point over the first 4 years, after which the output losses slowly fade away. The results for each region, including the Euro-zone, are reported in detail in table 1 below. These simulations are then compared to the following models, as listed in table 2 below: • OECD Interlink simulations, consider a shock in which the oil price averages around

13 percent above baseline in the first year (2000), and then 22.5% above the baseline in the second year, declining to 10 percent by the end of the third year;

• McKibbin-Sachs Global 2 (MSG2): The MSG2 simulation incorporate a permanent 20 percent price increase as the MULTIMOD scenario.

Table 1: Permanent $5 per Barrel Increase in the Price of Oil: Baseline Scenario

(Percent deviation from baseline unless otherwise specified) (IMF, 2000)

Table 2: Comparison of the Baseline Scenario with Simulations using different models

Notes: 1 Corresponds to OECD countries in the OECD and MSG2 simulations. 2 Baseline scenario. 3 The OECD simulations consider a shock in which the oil price averages about 13 percent above baseline in the first year (2000), is 22.5 percent above baseline throughout the second year, and declines to 10 percent above baseline by the end of the third year. A more recent analysis by the IMF (2004) applied the Global Economy Model (GEM). The GEM is a multi-country macroeconomic model built using recent economic research based on an explicit microeconomic framework in which consumers maximise utility and producers maximise profits. This model allows for the integration of domestic supply, demand, trade, and international asset markets into a single theoretical structure. Figure 7 shows how the Euro Area is affected by a 20% increase in oil prices after one year, in relation to other regions, using the GEM. This price increase is predicted to reduce real output by 0.4% for the Euro-zone. The differences between the regions of the world reflect the factors as discussed above that affect the impact of a price increase. The output effects are fairly similar to those of MULTIMOD, whilst the impact on inflation is estimated to be lower when using GEM. The impact on the current account is larger reflecting a high dependence on oil in the short-run i.e. firms find it hard to substitute away from oil in the short-run.

Figure 7: GEM: Impact of a Permanent 20 percent Increase in Oil Prices After One Year

Source: IMF (2004) based on staff calculations Other estimates that have been made through use of predictive models include: • Donald et al (2002), who estimate the magnitudes of effect of oil price shocks on

GDP through the use of impulse response functions of oil price shocks in the GDP equation of a vector autoregression (VAR). For the recent period, values between -

0.05 and -0.06 as an elasticity, spread out over one and a half to two years, are estimated for price changes exceeding a three-year high.

• Sauter and Awerbuch (2003) estimate that a 10 % rise in the oil price might cut GDP

growth by 1.5 % for a period of 3–6 months following the price increase. Extrapolating this to the EU 15 would result in GDP reductions of between EUR 35 billion and EUR 70 billion.

A summary of results from these models is presented in Table 1 below and focuses on the impacts on GDP growth for the EU. Table 3: Macroeconomic cost estimates Source Driver Estimate Units Country /

Region EC (2002) Sustained $10

increase in price of crude oil (per barrel)

- 0.5% GDP growth rate Industrialised countries

- 0.5%

GDP growth rate Euro Zone IEA (2004) Sustained increase from $25 to $35 i.e. by $10 per barrel of crude oil

0.5% Inflation Euro Zone

- 0.25% (over first four years, then fades away)

GDP growth rate World

- 0.4% (percentage deviation from baseline after one year)

GDP growth rate Euro Area

0.5% (percentage deviation from baseline after one year)

Inflation Euro Area

IMF (2000) Sustained increase of $5 per barrel of crude oil (20% increase)

- 7.8 ($ billion) Trade balance Euro Area IMF (2004) Sustained increase of

$5 per barrel of crude oil (20% increase)

- 0.4% (after one year) GDP growth rate Euro Zone

Donald et al (2002)

Price change exceeding a three year high

- 0.05 to – 0.06 Elasticity GDP to oil price shocks

USA

- 1.5% (for 3-6 months)

GDP growth rate Euro Zone Sauter and Awerbuch (2003)

Sustained 10% rise in oil price

- EURO 35 to 70 billion

GDP Euro Zone

- 0.25% (over first two years)

GDP growth rate Industrial countries

World Bank (2000) in IMF (2000)

50% increase in price in first year, then decline back to by the third year. 0.2% Inflation Industrial

countries Huntington Doubling of oil price -3.7% GDP USA and Euro

(2004) Zone Price level: 1% increase

0.44% % change in Investment

IEA member countries

IEA (2001)

Price volatility: 1% increase

- 0.11% % change in Investment

IEA member countries

The summary results presented in Table 3 belie the complexity of the models and the variety of their outputs for world regions. Nevertheless, it is worth noting that despite the model differences there is some consistency in the pattern and extent of GDP changes. The different models generally show that even when the oil price increase is assumed to be permanent, GDP impacts are likely to be greatest in the first four years of the price increase, and declining in size after that, suggesting that economic agents adapt their expectations to the new price level so that factor and product markets move to new equilibria over this time period. Perhaps the more striking conclusion is that there is a reasonable level of consistency in the size of annual GDP losses that are associated with given oil price increases. Thus for the industrialised countries a $10 price increase per barrel gives rise to a 0.5% loss of GDP (EC, 2002) or a proportionate linear scaling of that, on average (IMF, 2000). For the Euro zone countries, there is a similar consistency in the results though here it appears that the GDP is more sensitive to oil price increases than for the industrialised countries as a whole. For Euro zone countries, a $5 increase of a price per barrel of oil results in a 0.4% decline in annual GDP for the first year. Before using these results to make a first estimate of gross external cost per kilowatt hour we review other potential sources of external costs of energy security. Military expenditure Some have argued that military expenditure should be factored into the external cost of energy security given that without this type of expenditure (particularly in the Middle East) there would be a tangible threat to the secure supply of oil. Delucchi and Murphy (1996) go as far as arguing that if US motor vehicles (a major user of petroleum) did not use petroleum the U.S. would reduce its defence expenditures in the long run by between $1billion to 10 billion dollars per year. However, other authors such as Bohi and Toman, (1996) argue that there are good reasons why this expenditure should not be considered as part of the external cost calculations. These include the arguments that:

• Military expenditure is a cost of mitigating energy security rather than a cost of insecurity itself

• Other national security interests are being served, not just oil • Military presence is potentially on behalf of many other countries too

We are persuaded by these arguments to differing degrees; the first correctly identifies that the cost of doing something (preserving oil supplies) should not be interpreted as the benefit of that action. However, it may be interpreted as a minimum of what society is willing to pay for the benefit. The second and third arguments are correct and signal the practical difficulty of disentangling different types of security interests and associated expenditures for a wide range of countries when – as is most commonly the case –

national defence expenditures are not published according to geographical reason or purpose. These practical difficulties have meant that within the constraints of the project resources we have not been able to include a military expenditure component in the estimation of energy security externalities. Gas and Coal Gas The rising demand for energy and an increasing dependence on external sources of gas in the future are potential sources of energy insecurity in the EU. The insecurity derives from source dependence; transit dependence; and facility dependence (extent of spare capacity in the event of failure of a major component – almost all gas connections between European member countries are said to be fully used). The IEA forecasts that European Union (EU) natural gas imports will increase to 80% by 2030 from 44% now, while demand for natural gas will rise to 34% of the total energy mix from a current 23% unless the EU further legislates to boost the use of renewable energy. Projected import growth is largely caused by depletion of internal European reserves, such as the UK North Sea, the Netherlands Groningen field and the expansion of the Union by non-producing countries of Eastern Europe. By 2030 and without a new framework for renewables, Russian natural gas will account for 33% of imports while natural gas from Africa will account for 27% and Norway and the Middle East 17% each. The IEA therefore states that "from an energy security supply point of view to rely so much on imports for its gas would be a very risky business."1 A matrix of gas security risks is included in Figure 8, below.

Figure 8: from INQUIRY INTO EUROPEAN UNION ENERGY POLICY: GAS SUPPLY AND ACCESS http://www.oxfordenergy.org/pdfs/House%20of%20Lords3.04.pdf Perhaps the greatest risk of prolonged interruption comes from the destruction of a major production or processing facility or a deep water pipeline whose replacement might take many months to build. In 1998, this was most graphically demonstrated when an explosion at an onshore processing plant in the Australian state of Victoria caused the disruption of gas supplies to all customers throughout the entire state for a period of

1 ‘Energy security and liquefied natural gas’ Energy Security, prepared by the Institute for the Analysis of Global Security http://www.iags.org/n0929034.htm

nearly two weeks. Despite the fact that gas supplies originated from different domestic offshore fields through different pipelines, all gas production was dependent upon the availability of the plant. European gas supplies from particular sources are vulnerable to potential accidents at key transmission and import facilities, some of which are remote from European territory. The most important are the Yamal– Nenets pipeline corridor, which carries nearly 90% of Russian gas production; the Ukrainian pipeline corridor, which carries around 90% of Russian gas exports; the Trans-Mediterranean and GME pipelines from Algeria to Italy, Spain and Portugal; the Troll field and associated pipeline infrastructure, which account for more than half of Norwegian production and exports. Although it is highly unlikely that any of these facilities – particularly those involving multiple pipelines or LNG trains – would suffer a major failure for any significant period of time, such low-probability events could have a substantial impact on a particular source of transit route and therefore an entire European region.2 It is also the case that energy partnerships and dialogues created by the European Union with its major gas suppliers in Russia and North Africa also need to involve transit countries e.g. with Ukraine and (most recently) Belarus in the case of Russian gas. Problems in the commercial and political relationships between Russia and these transit countries present unresolved problems between states, which have the potential to create gas disruption risks for the European gas, market as a whole. However, the costs of such an event have not been estimated to date, making it impossible to make an estimate of associated external costs. Furthermore, no real threat analysis has been performed to find out possible vulnerabilities in transportation and storage procedures that could be utilized by terrorists. Indeed it has been commented that there is a lack of an overall framework within which the costs, risks and benefits of insurance investments can be judged.3 Coal Evidence relating to the energy security nature of coal to the European Union is disappointingly scarce, and there appears to be no evidence available that seeks to quantify the macro-economic costs of coal supply disruption in Europe. We suspect that this lack of evidence may reflect the fact that these costs are perceived to be low relative to those for oil. However, there are growing dependencies in the EU on imported coal, particularly from China, that suggests a potential source of insecurity. This suggests the need for new empirical modelling work in order to simulate these potential insecurities in a credible enough way to provide comparisons with the oil sector. Electricity The paragraphs above have reported evidence on energy insecurity relating to energy fuel sources, which manifest themselves principally in price changes. A separate component of energy insecurity is the non-supply of energy that occurs in the case of electricity

2 http://www.riia.org/pdf/research/sdp/Sec_of_Euro_Gas_Jul02.pdf 3 INQUIRY INTO EUROPEAN UNION ENERGY POLICY: GAS SUPPLY AND ACCESS http://www.oxfordenergy.org/pdfs/House%20of%20Lords3.04.pdf

blackouts, or power cuts. The following paragraphs summarise the findings of a literature review on this topic. Costs associated with electricity supply disruptions (termed as ‘social costs’ by Costantini & Gracceva, 2004) are:

• Expenditure for military, policy and emergency services • Expenditure on public transport e.g. costs of subway interruptions and increased

delays with flights etc • Health care expenditure e.g. costs relating to reduced refrigerating capacity • Sanitation and waste disposal e.g. interruption in sanitation services • Other public services e.g. interruption of schools • Human life values i.e. costs relating to mortality and ill health, as well as lost

leisure time and fear. Factors that influence the extent of social costs include: the area affected; existence of alternative energy sources; duration of disruption; time of day and season; availability of advance warning and information. The literature estimates the costs of supply disruptions by multiplying the energy not served by a factor called the Value of Lost Load (VOLL). VOLL can be estimated by different methods including econometric models and case studies of interruptions. However, customer surveys are the most prominent e.g. willingness-to-pay to avoid a supply disruption. The official value of VOLL used until recently by the Pool in England and Wales is a function of the duration of an outage, averaged across different kinds of customers. It ranged between €3.8/kWh for a one-hour outage to €1.8/kWh for an outage of longer than 24 hours (Egenhofer et al, 2004). A more recent survey by Kariuki and Allan (1996) found that a higher value of €4.6/kWh. This value is similar to that found in a study on the blackout in 2003 in New York City, where the estimated direct cost (e.g. lost production and wages) was €0.66/kWh, with indirect costs amounting to about €3.45/kWh. This figure then has to be adjusted by the frequency of such an event to derive an expected annual value. Ex post studies, not yet available, on the series of blackouts in Europe in 2003 – particularly in Italy, Denmark, Sweden, and London - are expected to up-date this estimate, which we adopt in the meantime in our subsequent analysis. It should be noted that there is some volatility in prices in the electricity market. However, there is little empirical work on the external costs of such volatility and so we do not consider this further here.

Estimation of Externality Unit Value The above sections provide estimates for the overall macroeconomic implications of increases in oil prices and of physical disruptions in electricity supply, but not for coal and gas. These estimates do not distinguish between internal and external costs and so they provide an upper bound to the external costs of energy insecurity. In order to

facilitate comparison with other energy external costs we wish to express these costs in terms of mEuro per kilowatt hour. Whilst the estimates for electricity given above are in this unit, the macroeconomic costs of oil price changes are not. We therefore convert these aggregate costs to mEuro per kilowatt hour. A number of strong assumptions are adopted in order to derive these values and these are explained in detail below. In addition, the macroeconomic cost estimates only allow us to include the effect of a sustained price increase. No separate estimates based on price volatility is made since there are - as yet - no good studies of sectoral or macro-economic costs resulting from price volatility (Sauter and Awerbuch (2003)). We therefore assume that the unit value estimates derived below are minima. We make the conversion from gross macro-economic costs to costs per kilowatt hour by making a series of steps. These are described in the following paragraphs. The first step is to identify a benchmark for the world oil price against which deviations in price – and therefore GDP - can be measured. as a reference value in the event of an oil price change for instance. We assume here that average global oil production of 8.5m barrels per day over the last three years (2001-2003) provides for a “normal” oil price, of €254 per barrel, excluding shocks. The second step is to apportion a price increase that can be seen as typical to a given oil supply reduction. This can then allow us - in subsequent steps - to relate a GDP change resulting from a price change to the energy production from a given volume of oil. In this step we therefore assume a given price increase of €10 per barrel, which results from a supply disruption of 3 million barrels per day5, lasting for one year, equivalent to 1,095 million barrels. The third step is to relate the oil price increase to GDP loss, where we use GDP to represent changes in welfare. The issue here is the geographical scope of the welfare loss that is appropriate for our purposes. We may think that since we are researching to inform EU policy we should only include welfare costs borne by the EU. However, this would be inconsistent with estimates of climate change costs whose estimates are made based on global impacts. In the first instance we estimate both. We estimate the resulting loss of GDP in the EU256 relating to a range of the estimates of macroeconomic costs presented in Table 1 above, specifically relating to Europe. Estimates were taken from EC (2002) and IMF (2004) since these studies present results from a meta-analysis of previous studies and the most recent macro-model run (presumably best reflecting current macro-economic linkages), respectively. The length of impact differed from one year to four years. Total GDP for the EU25 in 2002 equated to €9.7 trillion. Therefore, a 0.5% fall in GDP equates to €48 billion p.a., totalling 192 billion over the four year period over which we assume the effect to last. For the world as a whole the equivalent figure is €178 billion p.a. and 712 billion over the four year period.

4 We assume €1 = $1 5 IEA (2001a) 6 http://www.eurostat.eu.int

The fourth step is to divide total GDP loss by the number of oil barrels lost in the supply disruption. Since the supply disruption is assumed to last for one year, the GDP loses identified in the third step are divided by the 1,095 million barrels identified in step 2. This gives values of €178 and €1,300 per barrel for the EU25 and world, respectively.

In the fifth step, given that we have now derived a macroeconomic cost per barrel of oil not supplied to the market, i.e. its opportunity cost, we need to make the final conversion to the energy provided by one barrel of oil. We use the conversion factors: 1 barrel of oil = 5,800,000 British thermal units (Btu)7; 1 kilo-watt hour = 3.413 Btu8, together with a thermal efficiency factor of 50%, to generate unit cost estimates of 0.2 and 1.5 mEuro per kilowatt hour for the EU25 and world, respectively. Since we want an annual equivalent these figures need to be multiplied by the probability of the event occurring in any give year to give us expected values. As an indication the current probability of a price increase of this magnitude occurring can be approximated on the basis of historical data. Harks (2003) estimates that a 3 million barrel shortfall event may be expected to occur at present on a 1 in 5 frequency, based on historical events. Adopting this probability, the resulting expected values are 0.04 and 0.3 mEuro per kilowatt hour for the EU25 and world, respectively. Note that these are average, rather than marginal, cost estimates. These estimates are low compared to the marginal cost estimates for health externalities from coal (at maximum about 1/5th) but similar to health externalities from nuclear. We think, however, that estimates that included oil price volatility and risk aversion would be significantly higher. Estimates would also be higher if other, perhaps smaller-scale but higher frequency, price increase events were included in the total annual cost calculations. Policy Analysis Much of the cost associated with energy insecurity is the result of increased uncertainty about supplies. This causes lower profits to producers by increasing the variance in the returns on their activities, and causes costs to consumers in terms of potential unemployment, disruption of lifestyles etc. Although all this is generally acknowledged, the formal modelling of ES in an uncertainty framework is rare. Yet, the tools for such modelling are available and include risk aversion modelling within the Von-Neumann-Morgernstern expected utility framework. As part of this project, we constructed a simple expected utility model of the ES problem, to see what insights it offered. The Statement of the Problem

7 www.nol.org/home/NEO/statshtml/glossaryb.htm 8 http://www.eia.doe.gov/kids/energyfacts/science/unitsindex.html

The following model has been designed to capture the following elements of the energy security problem:

A. The cheapest source of energy is imported energy, when it is supplies under ‘normal’ conditions. However, if the imported energy supply fails, for one reason or another, the result is a shortage in the domestic markets and prices that are substantially higher.

B. Domestic production can and does meet a part of the national energy demand.

The higher the supply price of energy, the more will be met from domestic sources.

C. The risks of energy supply disruptions or failures are well understood and can

be characterized in probability terms, based on the historic experience. Few would argue with elements A and B but there could be some dispute as to whether risks of disruption can indeed be captured in probabilistic terms. To assume that is a simplification, but one that serves to provide some insight into the appropriate controls for responding to the security problem. It is also a simplification in that the probabilities are not stationary, but change as the geopolitical situation changes. Eventually we may wish to look at endogenous or Bayesian probabilities, but before we can do that we need to investigate the case of exogenously given probabilities. The social problem is also tremendously simplified to deal make the analysis tractable. Society’s well-being is a function of the utility it derives from the consumption of energy, and that utility function is a well behaved von-Neumann-Morgernstern utility function that exhibits risk aversion. The main argument of the utility function is the consumer surplus that a given level of energy provides. Recall that consumer surplus is the difference between the total willingness of consumers to pay for a given amount of a commodity and the amount that they actually pay. Society’s choice can be described in terms of setting the total consumption of energy to maximise the expected utility of consumer surplus. Of course, in a market economy the government does not directly determine levels of imports and domestic output. But, by setting the domestic price of energy, it can determine both these variables. This price will typically be higher than the ‘normal’ international price of imported energy, the ‘premium’ being added to encourage domestic production and reduce dependence on imports. The basic structure of the problem is illustrated in Figure 1 below. Figure 9: Optimal Response to Insecure Imported Energy Price/ Cost

Marginal Cost of Domestic Production

Demand for Energy as Function of Price

A Formal Representation of the Problem The problem can be represented mathematically as follows: The objective is to maximize with respect to x0 and x1:

(1) Where: U(.) = Von-Neumann Morgernstern concave utility function P(x) = Inverse demand function for energy (P’(x) < 0) C(x0) = Total cost function for domestic energy (C’(x0) > 0, C(0) < c1a) c1a = Normal cost per unit of imported energy c1b = Cost per unit of imported energy with disruption x0 = Quantity of domestic energy produced and consumed x1 = Quantity of imported energy produced and consumed 1-π = Probability of a disruption in supply x = Total energy consumed = x0 + x1

Paut

!!

"

#

$$

%

&

''

(

)

**

+

,---+

''

(

)

**

+

,-- ..

++ 1010

100

110

0

110,

))(')'(()1())(')'((...

xx

a

xx

axx

xcxCdxxPUxcxCdxxPUMax //

If the problem has an interior solution it is characterized with levels of xo*, x1* and x* by:

0

*0*

)()(

dx

xdCxP = (2)

))({)1())({ 1*/11*

/1 ba

cxPUcxPU !!+! "" (3) Equation (2) states that at the optimum the marginal cost of domestic production equals the consumer price. This has the important implication that at the optimum there are no specific subsidies or taxes on domestic production. Equation (3) states that the expected marginal utility from an additional unit of imports is equal to zero. It can also be shown that, the solution characterized above has the property that the optimal price P(x*) will have the following properties:

)().1(. *11 xPccba!"+ ## (4)

That is, the optimal domestic price is not less than the expected price of imports. The difference between the domestic price and the expected price of imports reflects the risk premium – the greater this risk or the greater the risk aversion, the greater will be this premium. Numerical Analysis Using Specific Functional Forms In order to gain some further insight into the problem we take a specific functional form and look at how the solution behaves for plausible parameter values. The forms of the equations are:

(a) The utility function is the familiar iso-elastic form:

(.))1(

(.)!

!"=U (5)

Where β < 1 and is the coefficient of relative risk aversion. (b) The inverse demand function is linear in prices

bxaxP +=)( (6) a > 0, b < 0 (c) The total cost function is a simple quadratic form

2000 )( xcxC = (7)

Note that the problem only exists if the price at which domestic production would satisfy the entire demand is greater than the ‘normal’ price in international markets. In terms of Figure 1 Paut > c1a. For the specific functional forms chosen this implies:

ac

bc

baa 1

0 )2(

.>

!+ (8)

Numerical Ranges for the Parameters The following ranges have been taken for the key parameters in making the calculations reported below. β: The range taken is 0.3 to 0.7, which is the typical range in the risk literature for

the coefficient of relative risk aversion. 1-π The probability of a disruption in supply in any year is set at between 0.1 and 0.3.

This guess is based on the frequency of disruptions observed in the last 50 years, when there have been 17 disruptions to oil deliveries, but only 4 of these were major.

a, b The values of the demand equation parameters have been chosen to that the demand elasticity ranges from -0.4 to –0.7.

c0 The coefficient on the cost equation is set so that at the world price of c1a, domestic production meets 25% of domestic demand. Note that the implies supply elasticity from domestic production is equal to one. Both these values are not regarded as untypical for European countries.

c1b/c1a The costs of foreign supply with a disruption relative to the costs without disruption are set to range from 1.5 to 3.0. Experience from oil disruptions indicates that the price can increase by over 100% during the period of the disruption and there are other costs to consider (e.g. costs in terms of direct loss of welfare).

Numerical Results The numerical results of the model are given below. In each case the sensitivity of the optimal solution to the parameters of the model are measured in terms of changes in the output of domestic energy, the level of imported energy, total consumption and the optimal tax rate as a percentage of the price of energy. Sensitivity to Costs of Disruption9 Figure 10 shows how the solution varies with the costs of disruption. At a cost of 1.5 times the normal price there is no restraint on total energy consumption and so the tax on 9 In each simulation the ‘default values of the parameters that are not changing are: costs of disruption (2.0), probability of disruption (0.20, elasticity of demand (0.5) and risk aversion (0.5).

energy is zero. As costs rise to 2 times the normal price of energy, however, total consumption falls by 24%, implying substantial demand restraint. Energy dependence also decline substantially with increasing costs of disruption. As the cost rises from 1.5 to 3.0 for example, the amount of imported energy declines by 80% and domestic production trebles. The taxes on energy go from zero with a cost of 1.5, to 70% with a cost of 3. Hence the solution is very sensitive to this parameter. Figure 10

Sensitivity to Probability of Disruption Figure 11 shows the sensitivity of the solution to the probability of disruption. With a probability of 0.1 (10%), demand is restrained by 22% (i.e. it is 22% less than it would be with no probability of disruption. When the probability increases to 0.3 (30%), however, the restraint increases only marginally – to 26%. Imports fall with the increased probability by 16% as we go from 0.1 to 0.3, and domestic production increases by 10%. The tax on energy ranges from 42% (0.1) to 53% (0.3). Hence, although the probability is important in determining the scope of the reduction in imports, the impact is not as much as one might expect. Sensitivity to the Price Elasticity of Demand Figure 12 shows the sensitivity to the price elasticity of demand. Lowering the elasticity from 0.7 to 0.5 has major implications for energy policy. The energy tax increase from 50% to 117%, imports fall by 60% and domestic production increases by 50%. Hence the model is very sensitive to this parameter. The more inelastic demand is, the greater the incentives to go for reduced energy dependence and to develop domestic resources10.

10 It is worth noting that the model does not work for elasticities lower than 0.45, as the expected utility is not defined. Further work on lower elasticity demand functions is being undertaken.

Varation By Cost of Disruption

0

10

20

30

40

50

60

70

80

1.5 2 2.5 3

Cost of Disruption as % of Normal Cost

0

20

40

60

80

100

Domestic

Imports

Tax Rate

Figure 11

Figure 12

Energy Variation by Probability Disruption

0

10

20

30

40

50

0.3 0.2 0.1

Probability of Disruption

En

erg

y

0

10

20

30

40

50

60

Domestic Imports Tax (%)

Variation with Elasticity of Demand

0

10

20

30

40

50

60

0.7 0.5 0.45

Price elasticity

En

erg

y

0

20

40

60

80

100

120

140

Domestic Imports Tax (%)

Sensitivity to Risk Aversion Parameter The results of varying the risk aversion parameter are shown in Figure 13. There is little impact within the range of this parameter: lowering the coefficient of relative risk aversion from 0.7 to 0.5 causes energy taxes to fall by about 6%, imports to rise by about 12% and domestic output to fall by 10%. Figure 13

Concluding thoughts on the policy modelling exercise This analysis has shown how an important tool in addressing energy security is the use of the energy tax, and how the level of that tax varied according to the key parameters of the system: risk aversion, probability of disruption, demand elasticity and cost of disruption. The model is of course, only a partial representation of reality. But it is an important one and captures the significant role that internal energy pricing can play in reducing the impacts of uncertainty of foreign supply. To make the model more ‘realistic’ we need to:

• Model risk and costs more realistically as joint probability distribution for the two • Take account of measures that reduce costs of disruption but have a cost

themselves (e.g. holding of stocks). Stock levels are not calculated in this way at present.

• Develop links between measures of dependence and vulnerability and parameters such as risk of disruption.

• Assess more carefully exactly how much ES is an externality – how much of the risk has been internalized.

We intend to continue this work under future projects on energy security

Varaiation By Risk Aversion

32

34

36

38

40

42

44

0.7 0.5 0.3

Coefficient of Risk Aversion

38

40

42

44

46

48

50

52

Imports

Domestic

Tax Rate %

Overall Conclusions Measurement of energy security externalities remains a complex and difficult exercise. Problems of definition as to what constitutes these externalities make agreement on what the policy issue is hazardous. Additionally, the range of assumptions that need to be made in order to calculate quantitative estimates of the size of these externalities means that these estimates should be viewed as indicative only. There are also a range of gaps relating to oil price volatility and the potential macroeconomic costs of gas and coal supply disruption that suggest that the values of 0.04 and 0.3 mEuro per kilowatt hour are much lower than the true costs, whether categorised as external or not. However, our policy analysis has shown that there are potentially important implications for the design of an optimal energy tax that incorporates energy security. In the case of the macroeconomic external costs, the appropriate policy responses may also include the following: capital services obsolescence may be coped with by hedging through the adoption of less energy intensive technologies or expanding energy storage capabilities. It is argued that hedging will not work perfectly without perfect foresight of energy changes or if capital is not perfectly malleable across sectors. In this case, the only way to improve hedging is through better data and information. Additionally, it is argued that energy policy may be able to reduce the consequences of output price rigidities in utility markets and spill-over effects in labour and capital markets, by helping to reduce input price variability or enhance adaptability. Further reductions may only be possible through factor market reform. Because of the wide range of policy implications indicated is hoped that future projects will allow us to make more robust estimates of energy security external costs, including other costs such as those associated with risk aversion, additional to the macro-economic costs, and those costs relating to other energy fuels including coal, gas and nuclear energy, so that these costs can therefore more fully be considered in the design of future energy policy and in related sectors. More detailed attention to the costs of electricity blackouts on the basis of new empirical evidence would also help inform the design of policy related to energy infrastructure.

Bibliography Bohi D.R., and M.A.Toman (1996) The Economics of Energy Security. Kluwer. Boston Brown & Yucel (2001) “Energy prices and aggregate economic activity: an interpretive study” (Forthcoming) Quarterly review of Economics and Finance. Federal Reserve Bank of Dallas. Costantini V. and F. Gracceva (2004) “Social Costs of Energy Disruptions” INDES Working Paper No. 6. Centre for European Policy Studies. Cunado J (2000) “Do oil shocks matter? Evidence for some European Countries”. University of Navarra. Delucchi M. A. and J. Murphy (1996) “U.S. military expenditures to protect the use of Persian-Gulf oil for motor vehicles”. Report #15 in the series: The Annualized Social Cost of Motor-Vehicle Use in the United States, based on 1990-1991 Data. Institute of Transport Studies. Davis, California. Egenhofer C., Gialoglou K., Luciani G., Boots M., Scheepers M., Costantini, V., Gracceva, F., Markandya A., and G. Vicini (2004) Market-based Options for Security of Energy Supply”. INDES Working Paper No. 1. Centre for European Policy Studies. Ferderer J.P., (1996) Oil Price Volatility and the Macroeconomy: A Solution to the Asymmetry Puzzle. Journal of Macroeconomics 18, 1-16. Hamilton (2001) “What is an Oil Shock?” Mimeo, Dept of Economics, University of California, San Diego, December Huntington H. G. (2002) “Energy disruptions, interfirm price effects and the aggregate economy” Energy Economics (Forthcoming) Huntington H. G. (2004) “Shares, Gaps and the Economy’s Response to Oil Disruptions”. Energy Modelling Forum, Stanford University, Stanford. ICF Consulting (2003) The Economic Cost of the Blackout – An Issue paper on the Northeastern Blackout August 13 2003. Washington. IEA (2004) “Analysis of the Impact of High Oil Prices on the Global Economy”. May 2004 Paris. IEA (2001) “Energy price volatility: trends and consequences”. IEA paper Paris IEA (2001a) Towards a Sustainable Energy Future, IEA paper, Paris.

IMF (2000) “The Impact of Higher Oil Prices on the Global Economy” December 8 2000. Research Department Paper. Washington. IMF (2004) GEM: A new International Macroeconomic Model. Research Department, January 2004. Washington. Jones, D.W., Leiby, P. N., and Inja K. Paik, 2002. "Oil Price Shocks and the Macroeconomy: What Has Been Learned since 1996," Proceedings of the 25th Annual IAEE International Conference, June 26-29, Aberdeen, Scotland. Lee, K., Ni, S. and Ratti, R., 1995. Oil shocks and the macroeconomy: The role of price variability. The Energy Journal 16, pp. 39–56 Lockwood B. (1997) Energy Security. Unpublished report from the ExternE Core Project (1996-1997) for European Commission DGXII, contract number JOS3-CT95-0002 EC (2002) Muellbauer J., and L. Nunziata (2001) "Credit, the Stock Market and Oil: Forecasting US GDP," CEPR Discussion Papers 2906. Sanchez B. (1995) Le cout externe de l’insecurite des approvisionnements petroliers en Europe, Centre de Recherche et Droite de l’Energie, Universite de Montpelier Sauter R., and S. Awerbuch (2003) Oil price volatility and economic activity: A survey and literature review, IEA Research paper, August 2003.