Mobile Edition
Print
Get the Mag
Weekly Updates1. INTRODUCTION
The natural gas markets in different regions are gradually becoming more integrated. This globalisation process is due to several reasons. First, the costs of transport (especially LNG--Liquefied Natural Gas) have fallen significantly over the last 1-2 decades (except for the last few years), and constitute a much lower share of the wholesale price of gas than 15-20 years ago (Brito and Hartley, 2007). Second, gas reserves in the main consuming areas are gradually reduced compared with annual consumption, (1) which implies an upward pressure on domestic prices and increased imports of gas. With a larger share of gas reserves located in a few geographical locations such as Russia and the Middle East, intercontinental trade becomes more profitable. Third, an increase in the share of spot trade means that short-term price differences between regions may be more easily exploited by re-routing the gas (especially LNG), cf. IEA (2006). (2)
In this paper we look into future scenarios for a globalised natural gas market. Using a detailed numerical partial equilibrium model of the international gas, oil and coal markets, we explore how regional gas prices and trade patterns may develop until 2030 under different scenarios about future market conditions. Not surprisingly, we find that intercontinental trade will grow considerably over the next decades, reducing the upward pressure on gas prices in import regions such as Europe and North America. This result depends crucially, however, on the absence of constraints in the expansion of the gas industry in the Middle East. If the growth in gas production from this region is suppressed, we may see quite higher prices from 2020 onwards, and less intercontinental trade.
Our numerical analysis builds on the assumption that the international gas markets are liberalised and integrated. This is in contrast with the results in Siliverstovs et al. (2005), who find no sign of price integration between the North American market and the European/Japanese markets in the period 1994-2003. Their findings may reflect that the gas markets in continental Europe and North-East Asia are not yet fully liberalised. On the other hand, the gas markets in North America and the UK have been liberalised for more than a decade, with prices linked to liquid spot markets, and Neumann (2007) finds increasing convergence of spot prices in the US and the UK based on data for 1999-2007. Moreover, the EU has adopted two directives on gas liberalisation over the last decade (EU, 1998, 2003), although the speed of implementation has been quite slow in several key countries (Haase, 2008). With rapid growth in international spot trade, gas suppliers may find it easier to sell their gas in new markets, especially in the short term. Jensen (2004) claims that a moderate level of spot trade may be sufficient to balance the regional markets. Thus, we believe that the international gas markets are heading towards a globalised market, although with region-specific prices.
Whereas gas consumption in OECD regions has grown slowly over the last decade, consumption outside OECD and the former Soviet Union has increased by more than 5 per cent annually since mid 1990's (BP, 2008). In China, gas consumption more than tripled from 1997 to 2007. A similar picture is seen on the supply side, where gas production in the OECD has been rather constant since the turn of the century. On the other hand, production both in the Middle East and in Africa has more or less doubled over the last decade. Although intercontinental trade has been modest so far, some arbitrage trading has occurred in the Atlantic Basin, and the Middle East has to some extent become a swing supplier to both South and East Asia and the Atlantic Basin. So far, most of the LNG from the Middle East has been shipped eastwards, to India, Japan and South Korea, and only about 15 per cent to the Atlantic Basin (BP, 2008).
Besides the studies presented in EMF (2007) (and in this special issue), there have been few previous numerical analyses of globalised natural gas markets. One exception is Rosendahl and Sagen (2009), who examine the effects of transport cost reductions on gas prices in different regions (using the same model as in this paper). They show that gas prices in some import regions may increase when transport costs decline, e.g., because of different choice of transport mode or because of different transport distances between trading regions. Numerical analyses of the European gas market are found in Golombek et al. (1998), Boots et al. (2004) and Egging and Gabriel (2006), whereas MacAvoy and Moshkin (2000) and Gabriel et al. (2005) simulate the North American gas market.
In the following section we briefly describe the numerical model FRISBEE. Then we go on to present the simulation results of future scenarios in Section 3. Finally we conclude.
2. MODEL DESCRIPTION
We use a numerical model of the international energy markets called FRISBEE. (3) It is a recursively dynamic partial equilibrium model with 13 global regions, cf. Table 1. Supply and demand of fossil fuels and electricity are modelled in each region. FRISBEE accounts for discoveries, reserves, field development and production of oil and natural gas, distributed on regions and field groups. Supply of coal and electricity are modelled in less detail. There are three demand sectors in the model: 'Manufacturing industries', 'Power generation', and 'Others' (including households). All markets clear each year, and annual, regional supply, demand, prices and trade flows are among the outputs of the model. Seasonal variations in demand and supply are not included in FRISBEE, which means that variations in e.g. trade directions over the year are not captured by the model. The base year of the model is 2000, and it is programmed in GAMS (Brooke et al., 2005).
Natural gas demand is a function of the end-user prices of all energy goods. The own price elasticities for 'Manufacturing industries' and 'Others' are on average around -0.3 in the long run (around -0.1 in the short run). Cross-price elasticities are in general small. In the long run, gas demand is particularly dependent on income growth--(per capita) income elasticities are on average around 0.6. The elasticity of population is set equal to one. Finally, a moderate, exogenous energy efficiency rate is assumed (0.25% p.a. within OECD and 0.5% p.a. outside OECD). Fuel demand in the 'Power generation' sector is driven by existing capacities and generation costs (including fuel prices) for different technologies, as well as the electricity price. Substitution possibilities are thus much higher in the power sector than in the two end-user sectors.
Traditionally, the natural gas markets in Europe and to some degree Asia-Pacific have been dominated by only a few large players, both upstream and downstream, and the markets have been highly regulated. As the gas markets become more integrated, the potential for upstream market power diminishes. (4) Moreover, liberalisation processes are taking place both in OECD and non-OECD regions (IEA, 2006), and this is gradually reducing the market power of large, downstream companies. The extent of spot trade is growing fast, and gas price indexation is partially replacing the oil price link in long-term contracts (Cornot-Gandolphe, 2005). (5) Consequently, although it might be seen as a simplification of the current market structure, in our future scenarios for the global natural gas markets we assume fully competitive and liberalised markets.
FRISBEE distinguishes between (oil and gas) fields in production, undeveloped fields and undiscovered fields. Data on field characteristics are based on an extensive database of global petroleum reserves in the year 2000, and data on production costs are based on the same source. Supply from developed fields in the model is set so that marginal operating costs equal producer prices net of gross taxes. Operating costs are increasing functions of production, but are generally low unless production is close to the fields' production capacity; then they increase rapidly. The cost functions are calibrated based on data on production costs in different locations.
Oil and gas companies may invest in new fields and in reserve extensions of developed fields. Investments decisions are driven by expected net present values (NPV), which are calculated for four field categories in each of the 13 regions. (6) Expected NPV depends on expected prices (adaptive), a pre-specified required rate of return (set to 10 per cent in real terms), unit operating and capital costs, and net and gross tax rates. Unit capital costs are convex in the short term, and increase when the pool of undeveloped reserves declines (for new fields), and when the recovery rate rises (for reserve extension).
New discoveries are modelled in a simpler way. The amount of discoveries depends on expected prices and expected undiscovered reserves in each region (USGS, 2000).
All arbitrage opportunities are assumed to be exploited in the model, so that price differences between two regions never exceed the corresponding transport costs. Unit costs of LNG and pipelines are assumed to be constant in this analysis. Both capital and operating costs are included in the cost figures, except for pipeline capacities existing before 2007 (where only operating costs matter). Total transportation costs are linear functions of the distance between the regions. No geopolitical or other constraints are restricting investments in new transportation capacity in the model. Each year the cheapest transport technology between a pair of regions is chosen (i.e., LNG or pipeline). Thus, a region may import both via LNG and pipeline transport, but not from the same region. Data on unit transport costs are mainly based on OME (2001).
FEED