1. 1 Some macro-economic estimates of Energy Security Externalities Alistair Hunt University of Bath NEEDS FORUM 2 Energy Supply Security – Present and Future Issues Krakow, July 5th, 2007.

2. 2 Structure of Presentation Taxonomy of energy security Definitions Welfare effects Measurement of macroeconomic impacts: method & results Outline of Method for measuring impacts on individuals e.g. risk aversion

3. 3 Definition of Energy Insecurity Lack of “availability of regular supply of energy at a reasonable price” (IEA) Physical availability dimension and price dimension Long term and short term dimensions Longer term: will there be enough energy available at an affordable price in the future? Shorter term: an unanticipated cut in supply and corresponding increase in price (sustained or not)  We focus here on short term effects of (supply shock) insecurity (i.e. 1-2 years after shock).

4. 4 Empirical Evidence: Dimensions of past shocks Typologies of oil disruptions Quantity shocks, related to physical constraints (political and military conflicts, strikes) Price shocks, related to producers decision (OPEC) or economic factors (Asian crisis) Technology shocks, related to new concepts and ideas, or to new constraints (i.e., an unanticipated technical advantage of nuclear over oil with the discovery of the climate change problem) (rare) Dimensions oil disruptions looking at the past Magnitude of supply shortfall: absolute value (4 mb/d limit value of main shocks, 3.2 mb/d IEA reference value for Emergency Response System) Magnitude of supply shortfall: relative value (7% reduction in oil supply, as IEA reference value for Emergency Response System). 2002 world consumption was 76 mb/day Variation of oil price (increase of 50%) Duration of shocks (maximum 9 months)

5. 5 Energy Security and Its Impacts

6. 6 Theoretical Justification for Public Policy on ES  Individual decisions on production, consumption and import of energy do not take account of full social costs (externality) Disruption of supply has macroeconomic impacts that individual do not take into account Producers and importers cannot accommodate the risks for competitive reasons (e.g. holding of stocks) Individuals may underestimate the risks of disruption

7. 7 Characterisation of Energy Insecurity Impacts Impacts on economy (derived from international macro models) Effects of supply shocks to energy markets (price changes) and indirect / induced effects on other markets That is … Pecuniary externality – when the actions of one economic agent affects the welfare of a 3rd party through changes in price(s). SR only: assume in LR adjustments take place in amount of energy related investment & rate of innovation

8. 8 Military Expenditures? – we exclude this because: 1. Military expenditures are cost of mitigating energy insecurity, not cost of insecurity itself. 2. Securing oil flows/stabilizing prices is one of a number of security interests targeted by expenditure. Problem with apportionment. 3. Benefits accrue to a large number of countries. Again, problem with apportionment. Military Expenditure Associated with Stabilizing Oil Exporting Areas

9. 9 Developing Unit Costs General Methodology Metric of welfare loss: GDP …. Assess importance of macroeconomic costs of energy price shocks and, if possible, draw inferences about potential empirical significance of externality component. These costs constitute upper bound (After Sanchez (1995)). Review of empirical literature relating to estimates of macro-economic costs of energy security (mainly oil import dependence, sustained price increases). Convert ‘externality’ to unit value (€ / kWh).

10. 10 Empirical Evidence: Dimensions of past shocks (I) Some estimates of main direct economic impacts of oil disruptions Reduction of GDP growth rate with 1-2 years lag Negative balance of payments with maximum 1 year lag Reduction of negative effects (especially for GDP growth rate) after 1973 oil shock due to: More appropriate policy responses Consistent reduction of oil consumption (demand restraint policies)

11. 11 Empirical Evidence: Dimensions of past shocks (II): EU impacts Some estimates of main indirect impacts of oil disruptions Increase of inflation rate with 1 year lag Increase of unemployment rate with 1-3 years lag Gradual reduction of negative effects for inflation rate Small gradual increase of unemployment rate due to structural conditions of European labour market

12. 12 Empirical Evidence: Dimensions of past shocks (III) Factors that affect the magnitude of economic costs Capacity to anticipate shocks Level and duration of the shortfall (and price increase) Response of the oil markets (increase of production elsewhere, price volatility) Internal oil production and dimensions of strategic stocks Specific characteristics at macroeconomic level for immediate impacts: Oil intensity of industrial sectors or transport sector (relative importance to economy) Degree of flexibility of the energy sector (fuel-switching capacity) Specific characteristics at macroeconomic level for indirect impacts: Monetary and fiscal policies (in order to reduce inflation) Level of petroleum products taxation Degree of flexibility of labour market

13. 13 Modelled Estimates of Macro-economic Losses

14. 14 Inter-EU country differences

15. 15 Inter-EU country differences The role of fiscal and monetary policy responses. 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 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 e.g. weight of energy costs in GDP, Flexibility of energy sector i.e. capacity to shift from one fuel to another

16. 16 Conversion to Unit Costs Distribution of % change in GDP Corresponding distribution of physical supply shock Mean barrels per day Mean % reduction in GDP Annual World / EU- 25 GDP averaged over corresponding period Loss in annual World / EU-25 GDP Conversion factors Mean kWh per annum Mean € per kWh Modelling studies

17. 17 Approximate ‘Externality’ Unit Value EU27 GDP loss over 1 year ( € ) 43,798,143,600 GDP loss over 4 years175,192,574,400 Original oil consumption (mb/day)82.5 Fall in oil consumption (mb/day)3 New oil consumption (mb/day)79.5 Change in GDP per barrel consumed (1 year loss) (€)1.5 Change in GDP per barrel consumed (4 year loss) (€)6.0 Each Barrel is equal to 1648.8 kWhs Thermal Efficiency40% Likelihood of shock0.2 Cost estimate per kWh - 1year loss (€)0.002289 Cost estimate per kWh - 4 year loss (€)0.009154 cost (€/kWh) 1 year loss0.000458 Cost (€/kWh) 4 year loss0.001831 Cost proportional to electricity generation (€/kWh)0.000004

18. 18 Possible model refinements Simplified model of economy that asks ‘what is cost of event, and what is likelihood of event?’.  estimates an upper bound to external pecuniary effects of fall in supply of oil to EU’s electricity generating sector. A dynamic model would look at effect of oil supply shock over time  how economy would adjust Precautionary expenditure e.g. on stockpiling and using the futures market. Currently exogenous to model, but models could be developed to capture the effect of these policies. Inclusion of risk aversion factors

19. 19 Measurement of value of actions that result in change in energy security Two elements of Willingness to Pay value economic losses using direct valuation methods value individual aversion to risk using Contingent Valuation (i.e. questionnaire). e.g. loss due to an act of terrorism value economic losses from earnings, employment etc. for each loss, an individual will be WTP something more than the value of the loss times the probability of the loss, because s/he is averse to the risk of the event. This WTP element needs use of CVM methods

20. 20 Energy Security and Its Impacts

21. 21 Methods of Estimating Benefits of Different Energy Security Impacts

22. 22 Sample of CVM Question

23. 23 CVM- Issues How to make risk change understandable? Use historic frequencies if possible/relevant? Increase to e.g. 1:100 risk change and assume linearity? Maybe realistic if consider more than one security event together. Issue of fear/dread distorting rationality -limit by questionnaire design Other CVM biases - e.g. free riding - limit through testing

24. 24 Future research directions Develop macroeconomic externality estimates from PDF of event sizes Exploration and derivation of individuals’ risk aversion to aspects of energy insecurity CBA of macro- and micro-policy responses