Each year hundreds of oil or electric customers call Boston Gas to ask about fuel-switching. What do they look for?A gas utility can boost sales only one way-by gaining new customers. And in today's slowly growing economy, conservation trends limit growth opportunities. The average household today uses two-thirds the energy of 15 years ago. Commercial and industrial (C/I) customers also conserve. If a gas utility is to grow, a majority of its new customers will likely come from competing fuels, such as oil or electricity.

But which customers will make the switch? Why, exactly, does a business or individual replace its old oil-fired or electric furnace with a gas-fired unit?

Gas utilities can adopt several different approaches to study the reasons for fuel-switching. Among these approaches, a statistical analysis is probably the most efficient. It doesn't cost as much as a customer survey, which requires the utility to design, print, and mail questionnaires, and then follow up with long telephone conversations, focus groups, and so forth. A successful statistical analysis can also create a mathematical model suitable for forecasting gas demand and evaluating the costs and benefits of advertising and marketing strategies. That's exactly what we've done at Boston Gas. We have developed a forecasting model that estimates the annual inflow of new, fuel-switching C/I customers.How Many Will Ask?

Every year hundreds of managers and owners of C/I establishments contact Boston Gas about converting existing oil-fired or electric equipment to gas. We wanted to know whether we could predict customer behavior-both for intial inquiries and for actual equipment conversions.

To find the answer, we modeled a number of these customer contacts or applications using discrete choice analysis. We developed models based on 1986-1992 historical data, using 1993 as the test year. All potential customers were divided into three classes: "small" would use less than 500 million cubic feet per year (Mcf/yr); "medium," between 500-5000 Mcf/yr; "large," more than 5000 Mcf/yr. We studied each establishment class separately to account for differences in economic behavior between small and large organizations.

Our goal was to understand how fuel prices, the state of the local economy, and the company's marketing activities would affect the inflow of applications for switching to gas. We also tested the effect of environmental regulation on fuel-switching activity by introducing a special categorical variable. This variable reflects major changes in environmental policy from 1986 to 1993. Starting in 1988, customers began to show significant concern about the condition of underground oil tanks. Many businesses and organizations began to consider installation of gas-fired equipment as a way to avoid potential liability from leaky oil tanks. The Clean Air Act (1992) heightened the pressure to replace oil burning equipment with natural gas.

All in all, we found our models statistically reliable. Table 1, covering an eight-year period, shows that the model estimates come very close to the actual numbers of customer contacts. The statistical analysis revealed that small- and medium-sized customers make more inquiries about fuel-switching when the company raises its level of spending for advertising. Among those two classes, environmental concerns affected medium-sized customers more than small firms. That result agreed with observations by Boston Gas's marketing staff that environmental concern was the primary reason driving medium-sized commercial firms and organizations to replace old oil-fired equipment with gas-burning units.

The oil/gas price ratio affects inflow of applications in every customer class. But the ratio plays the primary role in predicting how often large-volume customers will seek out fuel-switching. The environmental factor was statistically insignificant for large customers. Does that mean that large establishments ignore environmental issues? Certainly not. They just behave differently. Large firms tend to be well educated; they are aware of trends in regulations and plan ahead. They can also afford installation of air filters and other pollution abatement technologies. The cost of these technologies is (in some sense) an addition to the oil cost. Thus, the oil/gas price ratio indirectly takes that additional cost into account, eliminating the need in a special (artificial) environmental factor.How Many Will Switch?

Of course, after a prospective customer inquires about fuel-switching, the key question remains: Will the customer say "yes" and actually install gas-fired equipment?

Discret choice analysis showed that the probability of actual fuel conversion by small- or medium-sized prospective customers is largely a function of the economy, reflected by the unemployment rate. For large-volume customers, the most important variable is the anticipated change in the unemployment rate (unemployment rate for the next year minus the rate for the current year). It seems as though large-volume customers arrive at their fuel-switching decisions based upon the future state of the economy. These customers will make more conversions if they believe the unemployment rate will fall during the next year.

In many cases, especially for large-volume customers, the time period between the inquiry date and the gas-on date (project completion time) exceeds one year. For forecasting purposes we will want to know in what year a prospective new customer will start buying gas. Therefore, we analyzed distribution of the completion time.

All told, the combination of formulas we derived from our statistical analysis of customer inquiries, final conversion decisions, and timing of actual equipment installation provides Boston Gas with a reliable model for forecasting the inflow of new C/I customers switching to natural gas from competing fuels. The comparison of the model estimates with the actual equipment installations (Table 2) demonstrates that the model simulates reality quite well. The errors, with one exception, do not exceed 10 percent. And in 50 percent of the cases they fall inside the 2-percent interval. The 1993 (test year) projection estimates are close to actual results. Even the 9-percent deviation for large-vloume establishments does not look high, taking into account that the estimate of 36 is not too far from the actual count of 33.Alexander Lonshteyn, PhD, is supervisor, business forecast and analysis, Boston Gas Co.

Possible Callouts:

Environmental worries prompted more inquiries from medium-sized customers than from small firms; large firms reacted only to the oil/gas price ratio.Large-volume customers actually convert to gas when they see unemployment rates dropping.Table 1

Annual Fuel-Switching Inquiries to Boston Gas Co.

(Actual vs. Model)Establishment* Data 1986 1987 1988 1989 1990 1991 1992 1993

TypeSmall Actual 753 915 796 807 844 765 865 601

18,000 Model 766 901 793 825 849 746 854 605

Diff. +13 -14 -3 +18 +5 -20 -1 +4Medium Actual 386 473 545 491 613 538 703 486

13,000 Model 384 473 554 498 612 527 701 514

Diff. -2 0 +9 +7 -1 -11 -2 +28Large Actual 37 63 70 65 158 128 92 75

3,000 Model 45 57 60 71 156 130 94 63

Diff. +12 -6 -10 +6 -2 +2 +2 -12Total Actual 1176 1451 1411 1363 1614 1431 1660 1162

Model 1199 1431 1407 1394 1616 1402 1659 1182

Diff. +23 -20 -4 +31 +2 -29 -1 +20*"Small" denotes potential gas use < 500 Mcf/yr; "Medium" denotes 500-5000 Mcf/yr.; "Large" denotes > 5000 Mcf/yr.Table 2

Installations of Gas-fired Equipment

(Actual vs. Model)Year Small Market Medium Market Large Market

Actual Model Diff. % Actual Model Diff. % Actual Model Diff. %1988 650 643 -1 371 372 0 26 26 0

1989 594 653 10 339 344 2 24 27 11

1990 679 656 -3 370 375 1 49 47 -4

1991 602 589 -2 365 338 -7 57 57 0

1992 651 637 -2 418 423 1 49 52 6

1993 469 475 1 379 358 -6 33 36 9

Articles found on this page are available to Internet subscribers only. For more information about obtaining a username and password, please call our Customer Service Department at 1-800-368-5001.