Below is the draft of a lecture given on Tuesday 31st October as part of the 'The Beesley Lectures: Lectures On Regulation Series X 2000' organised by Professor David Currie of the London Business School and Professor Colin Robinson of the IEA. This is reproduced with the kind permission of Professor Yarrow
The new gas trading arrangements (NGTA) is the name given to a package of reforms in the gas industry, the first elements of which were introduced in October 1999. The subsequent process tends now to be referred to as the reform of gas trading arrangements (RGTA) programme.
The NGTA were and are part of an evolutionary process of development in the industry, with origins stretching several years back. The reforms have been more incremental than in electricity, where the abolition of the Pool and the introduction of the new trading arrangements (NETA), analysed by David Currie in an earlier Beesley lecture, marks a larger step change. Nevertheless, the pace of change in gas has sometimes been quite rapid, as many here will know all too well, and the introduction of NETA, coupled with the strengthening interactions between gas and electricity markets, now increases the priority that needs to be given to outstanding issues.
An occasion such as this provides an opportunity to take a stand-back look at the general direction of market and policy developments, and that is what I propose to do. The emphasis is therefore on general issues, rather than on the detail of latest proposals. I hope that this will make the material of more interest to the general listener, whilst not disappointing those at the coal-face (or is it the well-head?) of the RGTA programme.
The nature of the issues
Gas is a network industry, by virtue of the economic significance of the dedicated facilities required to transport the commodity from point of production to point of use.
In the most general terms, a network is simply a set of interconnected nodes. Visually, the transportation structure can be thought of as a set of points (nodes), with lines joining some of those points to one other (connectors).
Network industries differ according to the characteristics of the facilities required at nodes and to connect nodes, and gas has its own particular set of characteristics. In air transport, for example, nodes are populated by airports and associated facilities, such as parking and maintenance areas. Airplanes provide the physical means of connecting nodes, but there are no direct ‘physical’ connection, only a set of ‘routes’. In gas, the nodes are physically connected by means of pipes, but there is no equivalent of airplanes.
A general feature of network industries is that what goes on in one section of the network can have significant economic implications for what can go on in another section of the network. When a network is run as a fully integrated operation, this economic interdependence is not transparent to outside parties — the trade-offs are internalised within the integrated utility. There has, however, been a trend toward separation/unbundling of network activities, which brings the interdependence issues to the fore.
I will not discuss the various drivers of this trend tonight, but will simply make four points about it:
It has clearly been given a substantial boost by developments in information and transactions technologies.
The fact that the UK is a world leader in these developments in the energy sector means that, in pushing forward, there is frequently (though by no means always) a lack of directly applicable international experience to draw upon.
With the visual picture of a network in mind, there is obviously considerable scope for unbundling of the various elements of the network structure, right down to the individual facility level (whether the facility be a particular pipeline connection or assets at a particular node, such as a compressor or storage facility). There are, however, trade-offs in the unbundling process, which are influenced by factors such as transactions costs, information conditions and market power. Unbundling issues therefore raise one of the most fundamental questions in industrial organisation: what is the most effective division of activities between markets and firms (the Coase question)?
Given at least some unbundling, such that network users can choose which particular combinations of services they wish to purchase, a ‘textbook’ solution might say that prices of network components should be set to reflect the costs imposed by particular patterns of use. Thus, if one activity affects what can go on in other parts of the network, these effects should be reflected in the charges for the first activity.
Such an outcome can be achieved to at least some degree of approximation via market processes governed by appropriate rules, but not exactly (at least with current transactions technologies). If it could be achieved, the necessary co-ordination required for system balancing in gas could be handled by the prices determined in markets. As in electricity, however, a difficulty emerges in that there can be high costs of system imbalance, so that the degree of approximation required for efficient outcomes is beyond current capabilities. In such circumstances, the role of system operator emerges as a distinct, unbundled activity (in a fully integrated network, system operation is bundled with all other activities).
The role of the system operator is to deal with those co-ordination problems not dealt with via the market mechanism, or, put another way, to improve the market approximation to a level required to prevent inefficient imbalance. Yet another way of putting this is to say that the system operator corrects for economic externalities in the networks (i.e. effects on other parties not reflected in prices/charges).
It can be seen, therefore, that the scope of the role of the system operator depends upon the effectiveness of markets: the better are network effects reflected in market prices, the smaller the system operator role will be.
Given these points, developments in the gas industry can be seen as part of a general shift from internal organisation to markets as mechanisms for allocating economic resources and co-ordinating behaviour. Unbundling/network disaggregation is one aspect of this process; developing trading arrangements that more effectively promote value discovery and co-ordinate the activities of network users is another. The first directly increases the scope of markets; the second is aimed at making markets function more effectively.
I need hardly add that the second of these tasks is not easy. Integrated network operation has, historically, been a favoured option precisely because the complex economic interactions that take place within networks can be difficult to capture in trading arrangements. It can also be noted that inappropriate trading arrangements can sometimes cause more problems than they solve — they can increase, rather than reduce, network externalities, so some care is required in evaluating new sets of rules.1 Nevertheless, whilst challenging problems lie ahead, significant progress has been made, and there is no indication that the (current) limit of the general trend from internal organisation to markets has yet been reached.
Before leaving general issues, let me just add a few words about market power. As noted above, this is one of the potential limiting factors on the extent to which resource allocation and co-ordination can be effectively handled by markets. Network disaggregation/unbundling may lead to a situation in which significant market power can be exercised by parties controlling services provided in parts of the network, or controlling rights to use parts of the network. Exploitation of such market power is itself a source of economic externalities, and it is therefore potentially dysfunctional in terms of overall network performance (as well as being more directly damaging to consumers). In relation to traded markets, market power issues also surface in relation to questions of liquidity and depth, about which more will be said later.
The general point that I would make is an obvious one: the extent to which such power limits market developments depends in part on the availability of mechanisms to mitigate its exploitation. The greater the mitigation, the greater will be the scope for markets. There is, however, an obvious tension here: network disaggregation can enhance the role played by the price mechanism, but it may also create new sources of market power. Dealing with market power issues in network industries is one of the major challenges facing regulatory policy today, and policy performance in this area will be a major determinant of the effectiveness of developing trading arrangements.
Capacity, energy and spatial differentiation
In Nirvana Economics, strong competition could be imagined for all network components. The result would be a set of spatially differentiated prices for gas (energy), varying from moment to moment as inputs, offtakes and the state of the system changed. Every possible location would have its gas price, although it is easier to think only of prices either side of the network nodes (entry and exit points, compressors, storage facilities, pipeline junctions, etc.). The difference between gas prices at two locations would be the price for transporting gas from one location to another. As should be obvious, in these circumstances the price for transportation from A to B is equal and opposite to the price for transportation from B to A.
I have stressed the relevance of the state of the system in price determination because of its practical importance for what follows. It includes factors such as temporarily reduced capacity along connectors and at nodal facilities (due to failures and planned maintenance, say) and also pressures within pipelines. Much more than in electricity2, the storage capacity of network components provides a buffer between inputs and offtakes, so it is not necessary to maintain an instantaneous (flow) balance between the two. Prices through the network can, therefore, vary substantially for any given input and offtake flows depending upon whether the system is relatively full (pressures are near upper limits) or relatively empty (pressures are near lower limits).
The GB gas system is not, and likely never will be, close to Nirvana. The issues before us are to do with lower order incarnations. By the mid-1990s the position reached in relation to the National Transmission System (NTS) can be summarised as follows:
Administered transportation charges were based upon capacity and commodity elements.
Spatial differentiation occurred via locationally differentiated entry and exit capacity charges, in that different charges were set for different entry points and exit zones.
Energy could be traded at the National Balancing Point (NBP), as well as at particular locations (beach terminals).
Transco, in its role as system operator (SO), could buy and sell gas through the flexibility mechanism, a trading arrangement in which the SO was always one of the two contracting parties.
These arrangements have an interesting structure. The NBP can be viewed as a ‘virtual hub’ in that its physical location is not specified. One interpretation is that it is anywhere inside the Transco system. Another is that it is everywhere inside the Transco system, in that the values of gas is, in effect, equalised within the system for trading purposes. Such equalisation has the effect of increasing liquidity in trading, at the cost of losing some spatial signals (it is one of the ‘approximations’ judged to be acceptable).
It can be noted in passing that this approach depends upon the fact that the NTS is not itself disaggregated into separate pipeline companies. In jurisdictions where there are several pipeline companies (e.g. the US), the natural tendency is for energy trading to emerge at geographically defined hubs, such as where major pipeline systems connect with one another. That is, disaggregation of transmission forces greater spatial segmentation in energy markets.
Given these points, the NTS entry and exit charges can be interpreted respectively as charges to transport capacity from beach terminals to the NBP and from the NBP to NTS offtake points. This yields a conceptual hub and spoke pattern: the movement of gas from, say, the St Fergus terminal to inner London can be decomposed into a move from St Fergus to the NBP plus a move from the NBP to inner London. And, implicitly, spatial differentiation is confined to the defined entry points and exit zones.
If there are n entry points and m exit zones, the arrangements potentially lead to n + m + 1 energy prices (the +1 is the price at the NBP). Liquidity is concentrated at the NBP, since, on the hub and spoke analogy, all gas has to pass through the NBP. At some entry points, however, there may be a much smaller number of players on the selling side of the energy market, and at exit points the number of buyers of gas may be restricted.
Two major sets of issues can be identified in relation to the above arrangements:
A general lack of liquidity in energy trading, notwithstanding the pro-liquidity effects of the NBP concept.
Inefficient pricing signals in the transportation charging regime.
I will focus on the second of these in this section, which is most readily discussed by restricting attention to entry charges (an area where reforms have subsequently been implemented).
The release of primary entry capacity at prices based upon estimates of long-run costs clearly does not necessarily provide efficient signals of system costs in the shorter term (a problem exacerbated by the lack of temporal differentiation in the charges — there were no summer/winter variations for example). The spatial pricing structure is thereby distorted, and this will potentially lead to inappropriate decisions concerning use of the network.
Secondary markets can potentially improve the (relatively poor) first approximation. If, say, there is excess entry capacity at a particular terminal in a particular period, prices in the secondary market may be much lower, providing incentives for greater use of the network. Such incentives can be reinforced by use-it-or-lose it provisions to prevent capacity hoarding, and by with short-term sales of available capacity by Transco (e.g. on a day-ahead basis).
This does not, however, resolve the problem of more persistent excess capacity, since primary acquisition of capacity has to occur at the regulated price. There is also a more difficult problem that arises when there is excess demand for entry capacity at a particular location in a particular period. Under the old arrangements, Transco was required to offer as much capacity as network users wished to buy at the regulated price (an arrangement motivated by a desire to facilitate new entry by preventing capacity hoarding). Secondary markets can not deal with this issue since no-one would want to purchase at a price higher than the equivalent regulated price. That is, the secondary market price is effectively constrained by the regulated price.3
In the face of excess demand, therefore, network users have an incentive to purchase extra capacity from Transco, and Transco is then faced with the problem of rationing available capacity among holders of entry rights. This was the issue that emerged at the St Fergus terminal, in the period before the introduction of the first tranche of the NGTA, as a result of lack of capacity availability for a period.
One of the major issues addressed by the NGTA was a set of problems surrounding the flexibility mechanism, and which was seen as leading to high costs of maintaining system balance. A new on-the-day commodity market (OCM) was introduced in October 1999, providing opportunities for shipper-to-shipper as well as shipper-to-Transco trading. This encompassed both NBP and locational trades, and provided enhanced capabilities (supplementing OTC trading) for shippers to more closely balance their own input and offtake positions. The OCM is a screen-based, cleared market, operating on a near twenty-four hour basis.
A second major reform was the introduction of auctions for primary entry capacity, in monthly blocks and initially for six months ahead. These are sold as firm rights so that, if, in the event, Transco is unable to meet its contractual obligations, it is required to buy back capacity from the holders of rights to capacity. On the other hand, if Transco can, in any period, make more capacity available than has been sold in the primary auctions, it makes that extra capacity available to the market on a short-term basis, via incremental sales. As with energy balancing (see below), Transco is incentivised to some extent in relation to these day-ahead and on-the-day buy and sell actions.
These arrangements not only increase the scope for value discovery at the stage of primary release of capacity (although elements of the administered arrangements continue to exist in that levels of reserve prices in the auctions are high, at least in relation to short-run marginal costs at unconstrained terminals) but, perhaps more importantly, eliminate restrictions on secondary trading.
Whilst there has been much discussion of the design of the Transco auctions, the most substantive issues raised by the new entry capacity arrangements concern the duration of the rights offered at any one time and the amount of capacity that is made available. As the gas storage auctions indicated, there is clearly scope for making some capacity available on a longer-term basis, although the implications of market clearing prices for Transco’s controlled revenue is an added complicating factor that is not present in storage. No major conceptual issues are involved in such a development, and the prime regulatory requirement is only that, whatever the form of the contracts, there is no discrimination among different categories of network users.
In relation to the quantity of capacity offered for sale, difficulties arise because the capacity available varies over time: as noted in section 2, what can be done in one section of the network depends on what is happening in other parts of the network. Thus, at any time, the capacity available at a particular terminal will depend not only upon the physical dimensions of relevant network assets, but also upon the pattern of inputs and offtakes throughout the system, and on the state of the system itself (e.g. in the very short term, the pressure or linepack position). The solution adopted in the NGTA was to profile capacity offered to patterns of seasonal normal demand (SND).
In some circumstances, the amount of capacity offered in primary auctions does not matter a whole lot: the more important consideration is that, in any relevant period, all the capacity that can be made available is offered to the market. Thus, if, in the primary auctions, Transco systematically made available less (more) capacity than was actually available, the difference could be made good by short-term sales (buy-backs).
The issue acquires significance largely because of perceived failures in these short-term capacity markets and/or inappropriate Transco incentives. Entry capacity markets are locationally specific, and are therefore vulnerable to problems of liquidity and market power. This means that outcomes will not be neutral as to the division of capacity sales between the longer-term and shorter-term mechanisms. In addition, there can be concerns that Transco incentives are insufficiently strong to counteract the general tendency, not completely eliminated by the overall price control, to favour higher capacity prices (and therefore to be conservative in releasing capacity to the market).
For the future, it will be necessary to consider more closely the interactions between Transco SO incentive schemes and the incentives provided by the overall price control, which are of great significance for longer-term investment decisions. One possible resolution is to incorporate, into the price control process, forward commitments by Transco to offer certain levels of capacity by way of primary auctions. Such capacity could be offered on contracts for a variety of durations, and could potentially be sold forward in advance of physical capabilities. Adjustments in capacity availability, again of varying durations, around these forward commitments could then be handled via an enhanced SO incentive scheme, with Transco relying on a variety of contractual instruments, including capacity options as well as spot purchases and sales.
As explained in section 2, increased reliance on markets implies a diminished role for the system operator. One aspect of developments in gas has been to encourage greater self-balancing by shippers, leaving fewer problems to be resolved by the system operator. However, while the general intent of policy in this area is clear, balancing arrangements in gas have proved particularly problematic.
The intent is that shippers should be confronted with the costs that their imbalances (differences between inputs and offtakes) impose. To date, such costs have been approximated by a cash-out regime that is intended to pass back costs incurred by Transco to shippers in imbalance.
The targeting of (SO) imbalance costs has, however, been poor. Shippers have been granted tolerance bands for imbalances, within which the imbalances are cashed-out at the system or market average price. Only when imbalances fall outside the tolerance band do cash-out prices reflect the incremental actions taken by Transco when the system is short or long.
Cash out also only occurs at the end of the gas day (i.e. every twenty four hours). It is therefore incapable of reflecting the costs imposed by within day imbalances, which raises mounting problems in relation to interactions between the gas and electricity markets (where imbalances are cashed-out every half hour). To date, within day problems have been addressed only via rules concerning within-day profiling of flows, which leave something to be desired in terms of effectiveness.
The first phase of the NGTA reforms has made some contribution to the improvement of the balancing regime by:
Establishing the OCM, which allows shippers more easily to adjust their balance positions through the day.
Enhancing incentives for shippers to balance their own positions through a reduction in tolerances.
The second of these measures does, however, have limitations. Like end-of-day balancing, questions can be raised about whether such reforms are even working within the right conceptual framework. For example, as indicated earlier, it is not in generally efficient for inputs and offtakes to be exactly balanced at every moment through the day, and even less is it necessarily efficient for the cumulative inputs and offtakes to be matched at the end of every twenty four hour period. Within certain bounds, which themselves vary with inputs and offtakes, the inventory of gas within the Transco system can be run up or down. A more fully cost reflective set of arrangements would signal this through to shippers, and would incentivise them to take account of the state of the system when adjusting their own positions.
A set of trading arrangements that might better fit with the general objectives of moving to more market based arrangements is as follows:
Shippers acquire linepack or inventory rights from Transco.
Inputs and offtakes are monitored on a hourly basis.
Differences between inputs and offtakes in each hour either add to or subtract from inventory holdings.
If inventory holdings move outside the acquired limits, shippers are exposed to charges reflecting balancing costs imposed on the system operator.
Shippers can manage their positions by some combination of sales/purchases of gas and of inventory rights.
In addition to a requirement for this ‘missing market’ (in linepack or very short-term storage) to be developed, a further set of issues arises in connection with interactions between gas and electricity markets/networks. We already have two inter-related networks, such that what is done in one part of one affects what can be done in some parts of the other, but the interactions will likely grow stronger over time. For example, changing within-day patterns of electricity generation from CCGT plant could have very major implications for the operation of the gas network.
There is not time to discuss these further issues today, but it is clear that there will be a strong requirement for co-ordination and harmonisation of balancing arrangements for gas and electricity. The advent of NETA will likely mean that any initial burden of adjustment will fall mostly on the gas sector, but it will not always be a one-way street. The initial NETA framework is by no means the last word, and adjustments can be expected over time.
I am conscious of having touched upon only a few of the outstanding issues now being addressed in the gas sector, and then only at a broad brush level. I think, however, that it is useful to frame the issues in terms of a shift from internal organisation to markets, because this broad perspective draws immediate attention to major factors that can be expected to influence where, in any given state of the art, the most appropriate boundary between the two mechanisms is likely, very roughly, to lie. Such factors include transactions costs in general, information conditions in particular, and market power. Each of these can be expected to play a major role in determining the detail of the next incarnation of gas trading arrangements.
The other general point that may be worth emphasising is the degree of linkage between different aspects of the trading arrangements. Commercial conduct can be expected to be determined by many interacting factors, and progress in one direction may be impeded by lack of progress in others.
The development of the OCM illustrates both points. In its July 2000 review of the NGTA, Ofgem concluded that the OCM had experienced periods of low liquidity since its launch in the previous October. One of the factors influencing this outcome was identified to be the charging structure (transactions costs). This could be potentially addressed by, among other things, increasing the scope of the screen-based trading offered to encompass trading in such things as entry capacity and storage, thereby spreading fixed costs, and reducing transactions costs.
Other relevant factors that were identified in the review, and that are more directly related to other aspects of the trading arrangements and therefore more amenable to influence through the RGTA process, included shippers’ use of imbalance tolerances and Transco’s approach to trading in the market. In effect, the arguments here are that an individual shipper’s imbalances are being cross-subsidised under the existing cash-out regime – reducing incentives to trade so as to balance the position – and that Transco has insufficient incentives (or, for some reason, is failing to respond to existing incentives). Reform of other, relevant aspects of the gas regime could, therefore, have beneficial effects on the OCM.
Whether such developments (increasing the coverage of the trading platform, reforming cash-out, changing Transco conduct) would themselves be sufficient to address all major issues of liquidity is a more open question. Lurking in the background is the issue of market power. On this I am relatively optimistic. The NBP concept helps concentrate liquidity in energy trading and, despite the high headline level of concentration in the wholesale market, competition in gas supply has developed reasonably well, not least because of interactions, at the retail level, with electricity. There are also other policy instruments at hand that are targeted on market power problems.
I am less optimistic about liquidity (and market power effects more generally) in capacity markets. Although there is some scope for substitution among entry and exit flows, locational specificity is a much more significant issue. Market trading is characterised by smaller numbers of players, and recent evidence on prices paid by Transco in buying back capacity at St Fergus (during a period in which capacity was reduced by levels of flow in other parts of the system) provide no reassurance that markets are functioning in a fully competitive way. More generally, liquidity in secondary trading of capacity has been low.
Here, I think, we can begin to see some of the existing limits on how far market trading can be taken in certain areas. There are steps that can be taken to improve the position. The exploitation of market power can be monitored for compliance with competition law and licence obligations, Transco could be more active in its use of forward contracting, use-it-or-lose it provisions can help prevent capacity hoarding, and so on. It is also relevant that markets do not have to be fully competitive to make a useful contribution to improving economic performance. Even where limited, they can provide extra options and flexibility for network users. It might, however, be too much to hope that secondary markets in entry capacity will become deep and liquid at every entry terminal (to say nothing of trading at exit points, which the RGTA process has not yet touched).
The limitations of secondary capacity markets attaches a greater degree of importance to the way in which primary rights are sold by Transco. With effectively functioning secondary markets, the primary allocation process would not matter a whole lot from the perspective of economic efficiency – although there would still be issues to do with discrimination, including possible impacts on retail supply competition – since shippers’ positions could readily be adjusted through secondary trading. Illiquid secondary markets, however, mean that it is more important for network users that they secure a better approximation to their final requirements at the primary allocation stage.
The design of the existing Transco auctions is a relatively simple one, and it leaves scope for development. In my view, however, the bigger issues are to do with the questions of what it is that is being sold (should rights of longer duration be offered?) and with shippers’ skills in purchasing their requirements. The auction outcomes to date indicate that, while there was considerable price dispersion at the outset, significant learning has since taken place. This can be expected to continue as participants gain experience and as successive auctions increase the information that is available to bidders.
There are obviously some distortions to auction outcomes arising from the existence of the overall Transco price control, since over-recovery (or under-recovery) of allowable revenues leads to adjustments in commodity charges for gas transportation, the effects of which can, to some extent, be anticipated by bidders. Close analysis indicates that the effects are not likely to be large, however: a bidder considering whether or not to raise an offer price will only benefit from revenue recycling on the amount by which the (raised) price exceeds the price offered for any quantity that is displaced by the higher bid; and whilst bigger players have an incentive to bid a bit more, they have counteracting incentives to bid less in order to exploit monopsony power. In the round, therefore, the primary auctions would not appear to be fatally flawed on account of price control effects.
The big issues for the future appear to lie in those areas that have not yet been fully addressed by RGTA – long-term investment incentives and the exit capacity regime (including interruption) – and in the balancing regime. The latter is not well founded conceptually, and its weaknesses are being increasingly exposed by developments such as the emergence of substantial interconnector flows and interactions with the electricity market. Its reform will inevitably, therefore, be a high priority in the next stage of the RGTA process.
Key to such reform will be a more appropriate treatment of gas inventories within the Transco system (linepack), since only if all storage options are available, through markets, to network users is it likely that efficient self-balancing incentives can be established. Other desirable developments would be a shift to a shorter-term balancing period, together with other measures to ensure effective co-ordination between gas and electricity trading arrangements, and with improved incentive arrangements for Transco in its role as system operator. Among other things, Transco should be encouraged to rely on a wider mix of contractual instruments, subject always to the overarching requirements that it trades in a non-discriminatory manner and in a way that, where appropriate, complies with the Financial Services regime.
Finally, there is the issue of market power. Short-term demand and supply inelasticities are a less acute problem than in electricity, and the incidence of market power problems can be expected to be more weighted towards locational issues. Nevertheless, pricing anomalies in the wholesale gas market this summer indicate that not all is currently well. How the combination of the Competition Act, the existing licensing regime (including obligations in regard to system balancing) and reformed balancing arrangements will cope with the potential problems remains to be seen. All I would say, by way of conclusion is: get this one wrong and the likely outcome will be a more administered, less market-based set of arrangements than is currently hoped for.
1 The sensitivities here can be high — a poorly constructed rule can have substantial adverse effects, and deficiencies in rules may be hard to spot.
2 Inertia in electricity also provides some buffer between inputs and offtakes, but the timescales involved are much shorter.
3 Note that the restriction of secondary markets is not necessarily the same as a price cap at the regulated price, on a daily equivalent basis. If primary capacity is only available on an annual basis, a market participant seeking additional capacity for a period of a month may be willing to pay much more than the daily equivalent (regulated) price.