Abstract
In the U.S., natural gas pipeline transport has undergone a wave of deregulatory actions over the past several decades. The underlying motive has been the presumption that removing regulatory frictions would facilitate spot price arbitrage, helping to integrate prices across geographic locations and improve efficiency. Yet certain frictions, specifically the effect of congestion on transportation costs, inhibit positive deregulatory impacts on efficiency. With the increase in domestic production and consumption of natural gas over the coming decades, upward pressure on the demand for transport will likely result in an increased occurrence of persistently congested pipeline routes. In this paper we explore the relationship between congestion and spot prices using a simple network model, paying particular attention to the influence of storage. We find that as congestion between two hubs increases, the scarcity value of transmission capacity rises, driving a wedge between spot prices. We empirically quantify this effect over a specific pipeline route in the Rocky Mountain region that closely resembles our structural design. Although our results paint a stark picture of the impact that congestion can have on efficiency, we also find evidence that the availability of storage mitigates the price effects of congestion through the intertemporal substitution of transmission services.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Several researchers found evidence of convergence in regional gas spot prices prior to 2000 (Vany and Walls 1993, 1994a, b; Walls 1995; Serletis 1997; Dahl and Matson 1998), arguing that local, regional, and national gas markets evolved in response to increased arbitrage opportunities. Finnoff et al. (2004) find further evidence that FERC Order No. 636 spurred changes in pipelines’ operational and financial behavior that reduced ‘balkanization’, increased competition, and reduced expense preference behavior.
Where delivery constraints between major trading hubs exist, prices at the trading hubs can be impacted. This effect occurs irrespective of distance. De Vany and Walls (1995, p. 46) state, “…if there is no link [between markets] or if there are limits on the flow of the commodity over the link, then the prices of the commodity can move farther away from each other, especially in short time periods.”
Vickrey (1969) defines a bottleneck as “a situation in which a network segment has a fixed capacity substantially smaller relative to flow demand than that of preceding and succeeding segments.”
Importantly, the effect of congested transport infrastructure on price differentials is not limited to the natural gas pipeline network. A similar and widely publicized example had been occurring between two key oil price indices in the U.S. Starting in early 2011, increased oil production in Canada and the central U.S. overwhelmed the pipeline infrastructure transporting oil from the Cushing, OK hub to the Gulf Coast. This bottleneck resulted in a price differential between the West Texas Intermediate (Cushing) and Brent Crude (Gulf Coast) indices that averaged roughly $20 per barrel, and persisted through mid-2013. However, once new pipeline and rail links between Cushing and the Gulf Coast came online, the differential shrunk to a manageable $6 per barrel on average (DiColo 2013).
Two early empirical studies (Hollas 1994, 1999) examined the impacts of FERC’s push toward open access pipeline transport and restructuring of the natural gas market on public utility pricing. Following implementation of FERC Order No. 636 (as well as its pre-cursor, Order No. 436), industrial customers enjoyed significant reductions in retail gas rates relative to residential and commercial users.
Prior to the passage of Order No. 636, Alger and Toman (1990) presented experimental evidence that a market-based approach for this class of transaction could “outperform traditional rate-setting regulation” used in interstate pipeline transmission, with the caveat that short-term resale rates could greatly exceed the regulated primary market rates during peak demand periods.
See Tussing and Tippee (1995, p. 231) for a complete discussion. Certain types of buy-sell transactions are prohibited by FERC. For example, a capacity holder cannot buy from a seller with intent to resell to a pre-specified buyer after transport. This is considered a violation of open access policy. FERC’s ‘shipper-must-have-title’ rule requires a shipper to own any gas transported on the pipeline (FERC 2012). Firm capacity owners wishing to exploit a constraint must either release unused capacity directly to shippers, or buy the gas commodity from suppliers, ship, and then resell to any willing buyers at the destination market price.
Personal communication (April 20, 2012) with Gregory Lander, President of Skipping Stone, LLC energy consulting group.
Shippers do have direct access to the pipeline via ‘interruptible transport’, the rate for which is also regulated by FERC. However, this service is by nature less reliable, as transmission may be interrupted at any moment by a firm claim on capacity (McGrew 2009).
For large industrial users, for whom transmission contract costs can be passed on directly to final consumers, a higher degree of risk-aversion might lead to a greater willingness to contract firm capacity for reasons of reliability of supply. But for unregulated gas traders, who by nature face greater risk from fluctuations in both demand conditions and uncertain contract cost recovery, there are significant primary market risks.
Additionally, we assume the regulated primary market rates to be fixed over the time period under consideration, and have verified this to be the case in our empirical application.
We augment the Cremer et al. (2003) network model by allowing for storage, as resource firms have incentive to hold inventories to smooth production over time when prices are stochastic and sufficiently volatile (Mason 2010). In practice, storage plays a vital role in facilitating the use of natural gas through hedging and network balancing (INGAA 2009). That point noted, our central focus is on the manner in which pipeline transportation costs impact the markets at Hubs 1 and 2; the simplified model we discuss below is able to produce the testable hypotheses of interest.
This mediation by storage was demonstrated in the peak-load literature (Nguyen 1976; Gravelle 1976; Crew and Kleindorfer 1979). The ability to store reduces the price differential between peak and off-peak demand periods. Hollas (1990) found empirical support for this effect in the natural gas pipeline transmission industry, using firm and interruptible LDC transmission rates as proxies for peak and off-peak prices.
We assume that production in each basin is exogenous, reflecting the conventional wisdom that natural gas production is generally price inelastic in the short-run (IEA 1998, p. 36; Krichene 2002). The idea is that wells that are actively producing are operated at production capacity, and that the costs associated with shutting in production from a natural gas well, and then reopening the well later, are typically too great to justify doing so in response to small variations in price.
In practice, a pipeline’s ‘capacity’ is defined as the maximum throughput per unit time (typically expressed in daily increments) that can be maintained over an extended interval, and is subject to various technological, safety, and regulatory constraints. Capacity does not refer to the maximum physical throughput capability of the system or segment, which can greatly exceed the pipeline’s certificated capacity (www.eia.gov).
While spot traded volumes are presumably decreasing in the spot price at the source hub and increasing in the spot price at the destination hub or end-user market, overall traded volumes need not respond in kind, particularly when spot traded volumes are small in proportion to overall traded volumes.
We should note that two major regional export pipelines begun operations in 2011. Opening of these lines lies outside our sample period, and they are not shown in Fig. 2.
Capacities and scheduled flows are defined in units of volume per period. For natural gas, the generally applicable time period is one day, and volume is either posted in 1000’s of cubic feet (Mcf), or is converted into millions of British thermal units (MMBtu). The conversion ratio is roughly 1.02/1 for Mcf/MMBtu. Although rare, on occasion scheduled volume can exceed maximum certificated capacity.
The daily index midpoint is simply the midpoint between the high and low recorded spot prices on a given day. This is the value that is typically reported by industry newsletters (for example, Platts Gas Daily).
Due to copyright law, the terms and conditions of the purchase agreements prohibit publication and/or sharing of purchased data. Please contact the corresponding author for authentication.
Intuitively, this separation is related to the likelihood that differing technological and institutional factors govern upstream versus downstream sales. Our simple model relies on the assumption that the upstream price must exceed the downstream price for the transportation charge over a given segment to be positive. We thus consider days when the Cheyenne price exceeds the Opal price to be representative of our model’s design. In reality, this is not always the case, although the Opal price exceeding the Cheyenne price certainly seems to be the exception rather than the rule.
The 22 days on which these prices are equal are considered uninformative and are ignored in our formal empirical analysis. This is due to the fact that they do not fit into either of the other two cohorts, and because the small sample size makes testing the subset on its own difficult with respect to formal empirical inference.
Calculated as the Cheyenne spot price minus the Opal spot price (presented using absolute values in Fig. 3).
Measured as the difference between daily operating capacity and scheduled flow volume, which we consider to be a reasonable measure of congestion over the bottleneck route.
Using natural logs can be thought of as allowing for non-linear effects in demand curves. While there are other published price indices in the region, including Northwest Wyoming Pool, Northwest South of Green River, and White River, they are all are considered to fluctuate closely with the Opal Hub price.
Calculated as the net of spot prices \(p_{t}^{c} -p_{t}^{o} \), in levels, implying that the Cohort 2 basis differential is recorded as negative in our data.
There is no storage facility near Cheyenne. CIG has a significant amount of system storage, however it all lies to the east and southeast of Denver. For this reason, we do not consider this system storage to be connected to the Cheyenne Hub.
We consider statewide demand for Colorado as a geographically eastern variable only. Roughly 85 % of Colorado’s population lives in the eastern half of the state (www.colorado.gov), and all gas coming from the west is routed through the Cheyenne Hub.
To assess the appropriate lag structure, we estimated models containing zero, one, two, three, four and five lags. For each of these structures, we obtained the Akaike Information Criterion and Bayesian Information Criterion from the underlying 2SLS estimation. Based on these statistics, we inferred that the optimal lag structure contained four lags for each cohort. In an evaluation of 2SLS and 3SLS estimators with structural dynamic models of non-stationary and ‘possibly’ cointegrated variables, with unknown unit roots or rank of cointegration, Hsaio and Wang (2007) show that the 2SLS \(t\)-statistics and 3SLS \(z\)-scores of individual coefficients are asymptotically distributed as standard normal random variables.
The equation for flows on the bottleneck pipeline segment, \(y_t^b\), is used as the central equation for our 2SLS instrumental variables procedure, as it contains all other endogenous variables as regressors.
While the first- and second-stage weak identification tests are inconclusive, each of the three weak-instrument robust inference tests pass at 99 % confidence; we therefore conclude that weak identification is not a major cause for concern in our estimation.
Due to the simultaneous nature of our system of equations, 3SLS is the natural estimation procedure for obtaining coefficient estimates. One caveat with using 3SLS is that coefficient estimates may be inconsistent if there is any serial correlation in the error structure. Although the errors may display some contemporaneous correlation across equations, we have no reason to suspect any other correlations or heteroskedasticities in the error structure, and thus maintain confidence in the validity of our 3SLS procedure.
That this reduction is not statistically significant may be an artifact of the feature that the bottleneck constraint does not bind every day. On days when there is no congestion, other motivations for storage will trump any incentives to intertemporally arbitrage, weakening the significance of the impacts from days with tight constraints.
The remaining determinants of storage at Clay Basin (LHS: \(stor)\) conform to logic. Storage increases as production in the western basins increases (RHS: \(q^{w})\). It is drawn down (i) in the winter (RHS: \(w)\); (ii) when outflows west and through the bottleneck increase (RHS: \(y^{w},y^{b})\); and (iii) as consumption in Utah increases (RHS: \(ut)\).
In other words, one would arrange for additional purchases from other sources at Cheyenne, for example the Powder River Basin, and reduce injections into the pipeline at Opal, thereby freeing up the extra gas to be sold at Opal.
It is also important to note that prices were at their peak in 2008, adding to the propensity for high basis differentials relative to other intervals in our sample.
Applying our Cohort 1 coefficient estimates in Table 3 to the system of equations (1), and setting all variables equal to their Cohort 1 means following the second major capacity expansion, we obtain a basis differential of 7 cents.
Average scheduled volume for the sub-sample is 2,837,330 MMBtu/day. Using the FERC estimate, 646,213 MMBtu/per day of that volume would be transacted using spot prices. Multiplying this value by 1.6 cents, the increase in the basis differential we estimated in the text, yields an estimated increase in total transport costs of $10,399 per day. Multiplying that by the average number of days in a month (30.4) equals $316,129.
References
Alger, D., & Toman, M. (1990). Market-based regulation of natural gas pipelines. Journal of Regulatory Economics, 2(3), 263–280.
Atkinson, S., & Kerkvliet, J. (1986). Measuring the multilateral allocation of rents: Wyoming low-sulfur coal. The Rand Journal of Economics, 17(3), 416–430.
Baum, C., Schaffer, M., & Stillman, S. (2007). Enhanced routines for instrumental variables/generalized method of moments estimation and testing. The Stata Journal, 7(4), 465–506.
Black, & Veatch, LLC. (2012). Natural Gas Infrastructure and Electric Generation: A Review of Issues Facing New England. Prepared for: The New England States Committee on Electricity, December 14, 2012.
Bowring, J. (2013). Capacity markets in PJM. Economics of Energy & Environmental Policy, 2(2), 47–64.
Brown, S., & Yücel, M. (2008). Deliverability and regional pricing in U.S. natural gas markets. Energy Economics, 30(5), 2441–2453.
Cramton, P., Ockenfels, A., & Stoft, S. (2013). Capacity market fundamentals. Economics of Energy & Environmental Policy, 2(2), 27–46.
Cremer, H., Gasmi, F., & Laffont, J. J. (2003). Access to pipelines in competitive gas markets. Journal of Regulatory Economics, 24(1), 5–33.
Cremer, H., & Laffont, J. J. (2002). Competition in gas markets. European Economic Review, 46(4–5), 928–935.
Crew, M., & Kleindorfer, P. (1979). Public utility economics. New York: St. Martin’s Press.
Dahl, C., & Matson, T. (1998). Evolution of the U.S. natural gas industry in response to changes in transaction costs. Land Economics, 74(3), 390–408.
De Vany, A., & Walls, W. D. (1993). Pipeline access and market integration in the natural gas industry: Evidence from cointegration tests. The Energy Journal, 14(4), 1–19.
De Vany, A., & Walls, W. D. (1994a). Open access and the emergence of a competitive natural gas market. Contemporary Economic Policy, 12(2), 77–79.
De Vany, A., & Walls, W. D. (1994b). Natural gas industry transformation, competitive institutions, and the role of regulation. Energy Policy, 22(9), 763–775.
De Vany, A., & Walls, W. D. (1995). The emerging new order in natural gas: Markets versus regulation. Westport, CT: Quorum Books.
De Vany, A., & Walls, W. D. (1999). Cointegration analysis of spot electricity prices: Insights on transmission efficiency in the western US. Energy Economics, 21(5), 435–448.
DiColo, J. (2013). End of the line for Brent-WTI Trade. The Wall Street Journal. Retrieved June 24, 2013, from URL: http://blogs.wsj.com/moneybeat/2013/06/24/end-of-the-line-for-brent-wti-trade/.
Energy Information Administration (EIA). (2010a). Annual Energy Review 2009. Washington, DC: Office of Energy Markets and Use, U.S. Dept. of Energy.
Energy Information Administration (EIA). (2010b). Summary: U.S. Crude Oil, Natural Gas, and Natural Gas Liquids Proved Reserves 2009. Washington, DC: Office of Oil, Gas, and Coal Supply Statistics, U.S. Dept. of Energy.
Energy Information Administration (EIA). (2011). Annual Energy Outlook 2011 with Projections to 2035. Washington, DC: Office of Integrated and International Energy Analysis, U.S. Dept. of Energy.
Federal Energy Regulatory Commission (FERC). (1999). Cost-of-Service Rates Manual [for Natural Gas Pipelines]. Washington, DC.
Federal Energy Regulatory Commission (FERC). (2008). Order No. 712. Promotion of a More Efficient Capacity Release Market. Docket No. RM08-1-000. Issued: June 19, 2008.
Federal Energy Regulatory Commission (FERC). (2009). Order No. 712-B. Promotion of a More Efficient Capacity Release Market. Docket No. RM08-1-003. Issued: April 16, 2009.
Federal Energy Regulatory Commission (FERC). (2010). 2009 Analysis of Physical Gas Market Transactions Using FERC Form 552 Submissions. Item No. A-3, December 16, 2010.
Federal Energy Regulatory Commission (FERC). (2012). Order Approving Stipulation and Consent Agreement. Docket No. IN12-5-000. Issued: January 4, 2012.
Federal Energy Regulatory Commission (FERC). (2013). Jurisdictional Storage Fields in the United States by Location. Retrieved May 23, 2013, from Available on http://www.ferc.gov at URL: http://www.ferc.gov/industries/gas/indus-act/storage/fields-by-location.pdf.
Finnoff, D., Cramer, C., & Shaffer, S. (2004). The financial and operational impacts of FERC order 636 on the interstate natural gas pipeline industry. Journal of Regulatory Economics, 25(3), 243–270.
Gilbert, R., & Harris, R. (1984). Competition with lumpy investment. Rand Journal of Economics, 15(2), 197–212.
Gravelle, H. (1976). The peak load problem with feasible storage. The Economic Journal, 86(342), 256–277.
Hollas, D. (1990). Firm and interruptible pricing patterns: Public versus private gas distribution utilities. Southern Economic Journal, 57(2), 371–393.
Hollas, D. (1994). Downstream gas pricing in an era of upstream deregulation. Journal of Regulatory Economics, 6(3), 227–245.
Hollas, D. (1999). Gas utility prices in a restructured industry. Journal of Regulatory Economics, 16(2), 167–185.
Hsaio, C., & Wang, S. (2007). Lag-augmented two- and three-stage least squares estimators for integrated structural dynamic models. Econometrics Journal, 10(1), 49–81.
International Energy Agency (IEA). (1998). Natural Gas Pricing in Competitive Markets. Paris: OECD/IEA.
Interstate Natural Gas Association of America (INGAA). (2009). Interstate Natural Gas Pipeline Desk Reference: Summer 2009 Edition.
Krichene, N. (2002). World crude oil and natural gas: A demand and supply model. Energy Economics, 24(6), 557–576.
MacAvoy, P. (2007). Quantitative study number three: The basis differentials on partially deregulated pipeline transportation. In P. MacAvoy, V. Marmer, N. Moshkin, & D. Shapiro (Eds.), Natural gas network performance after partial deregulation: Five quantitative studies (pp. 97–129). Singapore: World Scientific Publishing.
Marmer, V., Shapiro, D., & MacAvoy, P. (2007). Bottlenecks in regional markets for natural gas transmission services. Energy Economics, 29(1), 37–45.
Mason, C. (2010). On stockpiling natural resources. Resource and Energy Economics, 33(2), 398–409.
McGrew, J. (2009). FERC: Federal energy regulatory commission (2nd ed.). Chicago, IL: American Bar Association, ABA Publishing.
Nguyen, D. (1976). The problems of peak loads and inventories. The Bell Journal of Economics, 7(1), 242–248.
Serletis, A. (1997). Is there an East-West split in the North American natural gas market? The Energy Journal, 18(1), 47–62.
Tussing, A., & Tippee, B. (1995). The natural gas industry: Evolution, structure, and economics. Tulsa, OK: Penwell Books.
Walls, W. D. (1995). Competition, prices, and efficiency in the deregulated gas pipeline network: A multivariate cointegration analysis. The Journal of Energy and Development, 19(1), 1–15.
Vickrey, W. (1969). Theory and transport investment. The American Economic Review, 59(2), 251–260.
Acknowledgments
This article has greatly benefited from the help and insights of David Aadland, Brian Jeffries, Erik Johnson, Gregory Lander, Jason Shogren, Alexandre Skiba, Brian Towler, and Aaron Wood. Two anonymous referees provided useful input, and pressed us to clarify our arguments and contribution. Earlier versions were presented at the 13th Annual CU Environmental and Resource Economics Workshop, Vail, CO (Oct. 7–8, 2011), the 2012 Occasional Workshop in Environmental and Resource Economics, Santa Barbara, CA (Feb. 24–25, 2012), and the 2012 Association of Environmental and Resource Economists (AERE) Summer Conference, Asheville, NC (June 4–5, 2012). We thank the participants of those events for helpful comments and observations. The School of Energy Resources at the University of Wyoming provided financial support for this research.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Oliver, M.E., Mason, C.F. & Finnoff, D. Pipeline congestion and basis differentials. J Regul Econ 46, 261–291 (2014). https://doi.org/10.1007/s11149-014-9256-9
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11149-014-9256-9