1、 AN ECONOMIC ANALYSIS OF CONSUMER RESPONSE TO NATURAL GAS PRICES FREDERICK JOUTZ AND ROBERT P. TROST PREPARED FOR THE AMERICAN GAS ASSOCIATION MARCH, 2007 An Economic Analysis of Consumer Response to Natural Gas Prices Frederick Joutz and Robert P. Trost1Prepared for the American Gas Association Mar
2、ch, 2007 Published by The American Gas Association 400 North Capitol Street, NW, Suite 450 Washington, DC 20001 www.aga.org 1Professors of Economics, George Washington University. Contact information: and trostgwu.edu. We are grateful for the support from the AGA, especially the helpful comments fr
3、om Bruce McDowell, David Shin, and Paul Wilkinson. We are responsible for any remaining errors. Copyright 2007 American Gas Association. All rights reserved. The enclosed materials were developed by the authors under an agreement with the American Gas Association. The statements, proposals, informat
4、ion or concepts expressed in these materials do not necessarily represent those of the American Gas Association or its members. For permission to reprint contact the American Gas Association. The American Gas Association (AGA) disclaims liability for any personal injury, property or other damages of
5、 any nature whatsoever, whether special, indirect, consequential or compensatory, directly or indirectly resulting from the publication, use of, or reliance on these materials. All warranties, express or implied, are disclaimed, including, without limitation, any and all warranties concerning the ac
6、curacy of the information, its fitness or appropriateness for a particular purpose or use, its merchantability and its non-infringement of any third partys intellectual property rights. Anyone using these materials should rely on his or her own independent judgment or, as appropriate, seek the advic
7、e of a competent professional in determining the exercise of reasonable care in any given circumstances and consult applicable federal, state, and local laws and regulations. Table of Contents Executive Summary.1 Introduction and Key Findings 1 Background 2 Decline in Use per Customer . 3 Price Elas
8、ticity and “Natural” Conservation Estimates. 4 Future Research .7 Introduction 8 Section 1: Background. 10 Section 2: Data. 17 Section 3: Approaches to Estimating Short- and Long-run Price Elasticity of Demand. 19 Section 4: Empirical Results Using the AGA Sample of LDCs 21 Shrinkage Estimators . 22
9、 National Results. 23 Regional Results 26 Regional OLS Estimates 27 Shrinkage Estimates. 37 Section 5: Summary of Results and Policy Implications. 43 Suggestions for Future Research . 45 References 46 Appendix A: Construction of Weather-Normalized Series for Use per Customer 50 Appendix B: U.S. Cens
10、us Regions 51 Appendix C: Literature Review. 53 Appendix D: Statistical Hypothesis Testing 55 Executive Summary Introduction and Key Findings The consumption of natural gas per household has been declining, on a weather-normalized basis, since about 1980. Over time, natural gas consumers have been t
11、ightening their homes, purchasing more efficient appliances and turning down their thermostats. Given the significant increase in natural gas prices since 2000, the American Gas Association (AGA) decided to examine whether or not the trend in declining use has changed in this higher-priced environme
12、nt. The results of this study are based on monthly data submitted by 46 local natural gas distribution companies that serve nearly 30 percent of all residential natural gas customers throughout the U.S. Some companies submitted data as far back as the early 1980s. The key findings of the study are a
13、s follows. A trend in declining use per residential natural gas customer of 1 percent annually has been documented2back to 1980. This decline rate has accelerated since the year 2000. null Weather-adjusted use per residential customer fell by 13.1 percent from 2000 through 2006. null The annual rate
14、 of decline in this 2000 to 2006 timeframe more than doubled relative to the pre-2000 period, increasing to 2.2 percent annually. null Further acceleration was witnessed in the 2004 to 2006 period, as evidenced by a 4.9 percent annual rate of decline. null The decline in use per customer has acceler
15、ated since 2000 in all 9 geographic regions analyzed. No appreciable changes in the price elasticity of demand were observed post-2000. Price elasticity of demand refers to the percentage change in demand for a good relative to a percentage change in price. Although the elasticity has not changed ov
16、er time, it should be noted that natural gas is an essential product that provides heat, hot water and cooking. Despite the essential nature of natural gas, consumers have continued to reduce their consumption at a relatively constant rate with respect to changing prices. Therefore, the large price
17、increases post-2000 have resulted in the large consumption declines noted above. 22004 AGA Energy Analysis: Patterns in Residential Natural Gas Consumption, 1980-2001. 2 null This study found a short-run price elasticity of 0.09 and a long-run price elasticity of -0.18. (Long-run elasticity refers t
18、o a period of time long enough for consumers to change the capital stock of their energy consuming equipment and the shell efficiency of their homes.) null These price elasticity estimates are relatively consistent with previous works on this subject. null The econometric analysis presented in this
19、study predicts a decline of 13.9 percent between 2000 and 2006; the actual decline was 13.1 percent. The decline is attributable to a price effect and the longer-run trend towards tighter homes and more efficient appliances. The price elasticity effect is 7.9 percent - equal to the elasticity estima
20、te of -0.18 times the 44 percent real price increase. The remaining 6.0 percent is explained by the longer-run trend towards tighter homes and more efficient appliances. null As a general rule of thumb, at the national level we would expect a 10 percent increase in the price of natural gas to result
21、 in nearly a 3 percent decline in the average residential use per customer 12 months later 1 percent attributable to more conservation with existing appliances, 1 percent attributable to the price-induced purchase of more efficient appliances, and 1 percent attributable to the natural turnover of eq
22、uipment that occurs annually. Background Residential natural gas consumption is strongly influenced by three factors: seasonal heating needs; response to price change; and the efficiency changes in appliances and home shells caused by a natural turnover rate to more efficient homes and gas appliance
23、s. On a weather-adjusted basis, the price and the long run conservation effects are key determinants of changes in residential natural gas consumption. The price effects can be further decomposed into short-term and long-term effects. Short term effects are decisions made by consumers with the curre
24、nt capital stock. Residential customers “turning down the thermostat” would be considered a short term effect. Long term effects are distinguished from short term effects by the inclusion of the decision to purchase more efficient energy consuming appliances and prematurely retiring less efficient o
25、nes. The price elasticity in the long-run is the sum of (1) the short-run demand and (2) the additional changes that occur to quantity demanded one year later because of natural gas price effects on the efficiency of the appliance capital stock and on the shell efficiency of homes3. While the separa
26、te efficiency and conservation effects due to 3It should be noted that if natural gas prices decrease, consumers will not replace recently purchased efficient equipment with less efficient equipment. So there maybe asymmetry with respect to the impact of natural gas prices on appliance and shell eff
27、iciency. The efficiency gains in appliance equipment that have occurred in the last several years will not disappear if natural gas prices go down. However, declining prices may lead consumers turning up thermostats to increase comfort levels (in the short-run). In the very long-run, a decline in pr
28、ices could lead to an increase in burner tips per customer. 3 appliance and housing shell turnover are difficult to disentangle in the current sample, they do appear to be discernable from the long term price effects. To address these issues, AGA commissioned a study to document changes in use per r
29、esidential customer on a weather normalized basis, particularly since the year 2000, and to identify the reasons for these changes. Other objectives of this study were: to obtain updated elasticity estimates for all nine US Census Regions and for the US; to test for an increase in the price elastici
30、ty of demand for natural gas since the year 2000; and to estimate a natural rate of decline in use per customer due to technology-induced gains in appliance and shell efficiency and a change in conservation attitudes that would occur even in an environment of constant real natural gas prices. Declin
31、e in Use per Customer Demand for natural gas per residential customer has been declining since the 1980s, and in recent years this decline has accelerated. Between 1980 and 2001, weather adjusted natural gas use per consumer in the US declined almost 1 percent on an annual basis. Since 2000, however
32、, the decline for winter only use has accelerated, decreasing 13.1 percent nationally between 2000 and 2006 for the sample of companies analyzed in this report. Figure ES1 below shows the winter season use per customer in actual and weather normal dekatherms from 1996-2006 using the data collected b
33、y AGA.4It is clear that actual and weather normalized use per customer has been declining since 1997 and this decline has accelerated since 2004. Figure ES1 US Annual Winter Use per Customer 4045505560657096 97 98 99 00 01 02 03 04 05 06Actual Weather NormalDekatherms4The data was collected from 46
34、Local Distribution Companies (LDCs) in 29 states, representing 28 percent of all residential customers. An LDC is a gas utility that serves a specific rate jurisdiction. Some of the companies in this sample have multiple jurisdictions in their corporate structure. The winter season for this report i
35、s defined as the sum of the monthly consumption between October and March. 4 Table ES1 disaggregates the national winter season weather normal use per residential customer across the nine US Census Regions and for the US. The decline in weather normal use per customer has occurred across all US Cens
36、us regions. The decline ranges from 5.7 dekatherms per customer for the West South Central region to 10.9 dekatherms for the East North Central region. The percentage decline in use per customer ranged from 9.2 percent for the Middle Atlantic Region to 14.8 percent for the Pacific Region. Table ES1
37、Annual Winter Season Weather Normal Natural Gas Use per Residential Customer, By Region and for the U.S. (Dekatherms per Customer) Census Region 2000 2001 2002 2003 2004 2005 2006 Percent ChangeNational 64.3 62.8 60.6 62.0 61.9 58.9 55.9 -13.1%East North Central 81.1 79.2 80.1 77.8 76.1 73.1 70.2 -1
38、3.4%East South Central 64.9 64.2 61.3 62.2 60.8 58.7 55.9 -13.9%Middle Atlantic 93.7 95.0 91.2 93.5 92.8 88.3 85.1 -9.2%Mountain 80.6 77.9 75.8 76.4 71.8 72.0 70.5 -12.5%New England 80.7 79.8 75.3 82.3 80.3 75.9 72.4 -10.3%Pacific 43.8 40.9 40.0 41.8 40.6 40.4 37.3 -14.8%South Atlantic 71.7 69.4 63.
39、8 69.1 62.0 62.5 62.5 -12.8%West North Central 80.1 79.5 79.8 80.4 78.3 75.9 70.2 -12.4%West South Central 46.3 46.4 40.2 44.1 54.1 41.7 40.6 -12.3%Price Elasticity and “Natural” Conservation Estimates This study found that neither a practical nor statistically significant change in the price elasti
40、city of residential natural gas consumption occurred in the post year 2000 period. The price elasticity of residential natural gas demand appears to have remained relatively constant since the 1990s. This implies the large percentage price increase since 2000 accounted for the decline in natural gas
41、 use, rather than an increased sensitivity or greater response by households to a given price change. The study also found that independent of natural gas price increases, the naturally occurring decline due to the technology driven gain in appliance and home thermal shell efficiency, as well as cha
42、nges in conservation attitudes was 1 percent per year. Table ES2 illustrates that for the sample of companies in the study, the short run price elasticity of demand averaged -0.09, while the long run estimated averaged -0.18. Therefore, given a 10 percent increase in the price of natural gas, consum
43、ption would decline 2.8 percent; 1.8 percent for price response, added to 1.0 percent decline due to the normal turnover of appliances and other “natural” conservation measures. There is very little regional variation in the total impact of a 10 percent increase in real prices on use per 5 customer.
44、 The impact in all regions was close to the national estimate of 2.8 percent, with the Mountain region being the lowest at 1.9 percent and the South Atlantic region being the highest at 3.7 percent. The study also found that the elasticity estimates calculated using the sample data were generally co
45、nsistent with the elasticity estimates found in the energy economics literature.5Table ES2 Summary of National and Regional Natural Gas Price Elasticity Estimates* Region Short-run elasticity Long-run elasticity*Annual Time Trend Total Response to a 10% Price Increase*National -0.09 -0.18 -1.0% -2.8
46、% East North Central -0.08 -0.22 -1.0% -3.2% East South Central -0.01 -0.01 -2.0% -2.1% Middle Atlantic -0.10 -0.20 -1.3% -3.3% Mountain -0.07 -0.10 -0.9% -1.9% New England -0.08 -0.25 -0.4% -2.9% Pacific -0.07 -0.12 -0.8% -2.0% South Atlantic -0.12 -0.29 -0.8% -3.7% West North Central -0.09 -0.15 -
47、1.1 % -2.6% West South Central -0.13 -0.16 -1.6% -3.2% * Estimates obtained from the “fixed effects” pooled regression * Cumulative: includes impacts of short-run elasticities * The total response to a 10% price increase is the sum of the long-run elasticity and the annual time trend effect. Implica
48、tions These price elasticity estimates and the natural conservation trends are able to explain the post 2000 winter consumption per household per customer actual experience. Between 2000 and 2006, real natural gas prices for the sample companies in this study rose 44 percent, which according to our
49、analysis would lead to approximately a 7.9 percent (0.18 x 44 percent) decline in use per customer by the year 2006. In addition to this 7.9 percent price induced decline in weather normal use per household, there would be an additional 6.0 percent (6 x 1.0 percent) decline because of the natural annual rate of turnover of old gas appliances to newer more efficient appliances. Hence, our analysis predicts a decline of 13.9 percent over the six-year period, which is very close to the actual decline of 13.1 percent. 5See Appendix C of the main report
