ML19343B837

Testimony on Behalf of Util Re Energy Alternatives.Tables of Actual & Projected Sales & Peak Loads for 1979,1990 & 2000 Encl

ML19343B837
Person / Time
Site:	Allens Creek File:Houston Lighting and Power Company icon.png
Issue date:	12/18/1980
From:	Anderson K NATIONAL ECONOMIC RESEARCH ASSOCIATES, INC.
To:
Shared Package
ML19343B832	List: ML19343B831 ML19343B833 ML19343B834 ML19343B835 ML19343B836 ML19343B837 ML19343B838 ML19343B839 ML19343B840 ML19343B841 ML19343B842 ML19343B843 ML19343B844 ML19343B845 ML19343B846 ML19343B847 ML19343B848 ML19343B849
References
NUDOCS 8012300677
Download: ML19343B837 (27)

Text

-

.

l

-

i. .

l 1

!

. t i  !

1 l

i,

>

i i-

!'

4 ,

4 i  :

.

J i I l

i 4

k

-

i i  :

!

'

!

l

!  !-

.

e i

!, ,.

j DIRECT TESTIMONY OF j

! KENT P. ANDERSON 1 i 1 i I

j ON BEHALF OF HOUSTON LIGHTING &. POWER COMPANY ,

-

i 1

4 RE ENERGY ALTERNATIVES l 6

'

i 1

'

.,

,

+

r l

!

• .

,

I l

8 012s o, g17:

.

.

.

I l

i l

1 TESTIMONY 2 OF KENT P. ANDERSON l 3 ON BEHALF OF 4 HOUSTON LIGHTING AND POWER CO.

,

5 6 Q. What is the purpose of your testimony?

7 A. The purpose of my testimony is to present an j

8 assessment of the reasonableness of the Company's latest 9 sales and load forecast based upon the results of an 10 econometric model that I have constructed for the Company.

11 Q. What are your qualifications?

.

12 A. I am a Senior Consultant at National Economic 13 Research Associates, Inc., 555 South Flower Street, Los 14 Angeles, California, an economic consulting firm speciali-15 zing in the economics of energy, the environment, antitrust 16 and labor. I received a B.A. degree in economics "with 4 17 High Distinction" from the University of Michigan in 1964 18 and a Ph.D. in economics in 1968'from the Massachusetts 19 Institute of Technology. From 1968 until late 1974 I 20 was a member of the Economics Department at the Rand 21 Corporation in Santa Monica, California. While at the 22 Rand Corporation I conducted research in a variety of 23 fields including the economics of national defense, 24 economic growth and development and the economics of energy.

25 My energy-related research included econometric analyses 26 of residential and industrial energy demands, and evaluation

• Ire /r/a

-

- - - _ -- -

. - _

1 of electric energy policy in California, studies of future 2 energy supply conditions and costs and a simulation analysis

,

l 3 of the nation's energy markets to 1995.

4 Since joining National Economic Research Associates, 5 Inc. in October 1974, I have assisted clients in solving 6 electric utility planning and management problems requiring 7 the application of economic theory and econometric modelling l

,

l 8 techniques. These include forecasting the growth of 9 electricity and fuels demands; projecting the effects 10 of time-of-use rates upon sales, loads, revenues and

,

11 earnings; evaluating the costs and benefits of time-of-

,

l 12 use rates, load management and conservation; determining 13 the desirability of new facilities and the optimal level l 14 of generating system reserves; estimating marginal and i 15 total production costs, system reliability and the costs l

16 of electricity shortages; and applying marginal cost l

'

17 principles to energy pricing. I have also aided clients 18 in such areas as evaluating the competitiveness of market l

19 behavior, the responsiveness of industrial location to l

20 regional economic conditions and the demand and supply l

21 of cable television services.

l 22 I have presented testimony on the subject of the

!

23 need for power before federal or state regulatory bodies 24 in Washington, Oregon, Montana, Arizona, California, New

,

25 Mexico, Colorado, Texas and Utah. I designed and constructed 26 an econometric model now used by Pacific Northwest Utilities then?ta' L

l 1 Conference Committee in its forecasting process. This 2 model projects electricity sales for each of the states 3 of Washington, Oregon, Montana and Idaho.

4 Some of my work has been published in The Journal 5 of Economic Theory, The Journal of Business and in EPRI 6 and Rand reports. I am a member of Phi Beta Kappa and 7 the American Economic Association.

8 Q. What are your findings?

9 A. Table 1 cummarizes the model's projections. Sales 10 range from 72.8 to 91.6 billion kilowatt-hours for 1990 11 and from 87.4 to 144.8 billion kilowatt-hours for 2000.

12 These figures correspond to annual growth rates of 3.0 13 to 5.2 percent from 1979 to 1990 and 1.9 to 4.7 percent 14 thereafter. Growth rates for the customer groups differ 15 somewhat from group to group depending upon the particular 16 model inputs and characteristics for each.

17 Q. Does the range of growth rates implied by the three 18 cases illustrated in Table 1 indicate absolute upper and 19 lower limits for growth?

20 A. No. The three cases correspo'..d only to different 21 assumptions about energy prices and economic and population

,

22 growth. The assumptions, even for this particular set 23 of inputs, do not cover the entire realm of possibilities.

24 The use of more extreme assumptions would widen the range 25 projected. It would also have been possible to generate 26 a wider range of outcomes by making different assumptions n/e/r/a'

.-- -..-- . . - - .- . .-. ,., . - - .

__.. - - . _ - . -- .-. . . . . _ . __.

1 about other factors, such as elasticity coefficients 2 or model structure. In my judgment, however, the range 3 of outcomes shown in Table 1 is suggestive of the more 4 probable growth paths for sales.

5 Q. What about peak loads?

6 A. The projections of peak-load growth derive from 7 application of the Company's projected system load factor 8 to projected saies (including resale sales as forecasted 9 by the Company). Projected peak load ranges from 13,300 10 to 16,700 megawatts for 1990 and from 16,000 to 26,500 11 megawatts for 2000. These levels imply growth at an 12 annual rate of 3.3 to 5.4 percent from 1979 to 1990 and 13 at a rate of 1.9 to 4.7 percent thereafter.

14 Q. Does use of an econometric model guarantee an accurate 15 pr'ojection of future growth?

16 A. A model is not an infallible oracka, and it cannot 17 replace judgment. It is, at best, an aid to judgment.

18 Its results are the product of an involved process resting 19 upon many assumptions. It can be a useful tool only if 20 its projections are subjected to some critical scrutiny.

21 I want to emphasize, however, that I believe the kilowatt-22 hour sales projections to be quite good, based, in part, 23 on the model's track record for the period 1974-1979.

24 (The base year for the model is 1973.) The errors of 25 prediction for total sales for each of the six years are l l

26 as follows (estimated vs. actual):

rhen?!aI l

. - - - - _ . - . . -. -. . - . .

. -

1 1974 2.1%

2 1975 3.1 3 1976 4.5 4 1977 0.5 5 1978 0.4 6 1979 2.3 to 3.7 7 Average 2.3 8 The model is calibrated for the prediction error observed 9 for 1979; that is, the projections .for 1990-2000 are 10 adjusted by 2.3 to 3.7 percent depending upon the case. ,

11 (There is a range of prediction errors for 1979 because 12 the model uses different, projected economic growth rates 13 for 1978-1979 depending upon the case studied. Actual 14 economic growth figures for this period are not yet 15 available.)

16 Q. How does the Company's forecast compare with your 17 projections?

18 A. Table 2 presents comparisons for 1990 and 2000. I 19 It shows that the Company's sales forecast falls inside 20 the range established by cases 1-3 of the NERA model.

21 Q. Does this mean that the Company's forecast is 22 correct?

.

23 A. Not necessarily. But since the NERA projections 24 result from a fully independent approach to the problem, 25 this agreement does increase the confidence that one can 26 place upon the Company's. forecast. In my opinion, the

• n/e/r/a q 1

_

~6-f 1- Company's forecast is reasonable, though possibly a bit 2 on the low side.

3 Q. You say thit your approach is fully independent.

4 What exactly do you mean?

5 A. I mean that in cons.tructing the sales forecasting 6 mode: I utilized data and methods fully of my own choosing 7 and without regard to the forecasting methods or assumptions 8 of the Company. As already noted, I rely upon the Company's 9 projection of sales for resale. I also make use of the 10 Company's projections of electricity and fuel prices, 11 but I do not rely upon them exclusively.

12 Q. Is your model econometric?

13 A. Yes. It contains separate energy usage equations 14 for the major classes listed in Table 1. Explanatory 15 factors include income, electricity prices, fuel prices, 16 commercial activity, industrial output, etc. A brief 17 sumicery of the modelling process is as follows:

18 The first si=p is to break electricity usage down 19 into basic cause-and-effect categories for the purpose 20 of statistically estimating demand relationships. In 21 practice, the extent to which one can decompose usage 22 into the theoretically ideal categories is limited, the 23 main reason being that sufficient data simply do not exist.

24 One could in principle collect the data but only at 25 significant cost and over a considerable period of time.

26 Consumption by ultimate customers was divided into

• n/e/r/a

_. . - - . _ -.

i

,

1 four user classes--residential, nonmanufacturing (compri-2 sing all business usage except manufacturing), manufac-3 turing and street lighting. The residential category was 4 further broken down into space heating, water heating, 5 cooking, clothes drying, air conditioning (single room, 6 multiple room and central) and all other uses, mainly 7 lighting and refrigeration. Separate equations were 8 estimated for each category. For nonresidential usage, 4

9 the to'tal was first broken into manufacturing and nonmanu-

,

10 facturing components. It was not possible to decompose 4 11 nonmanufacturing usage into product or. service groups or 12 into any process-related components; however, the equation 13 estimated for the model includes factors describing

, 14 the mix of businesses in the nonmanufacturing group.

!

15 Manufacturing usage was broken down into categories 16 corresponding to the federal government's 2-digit Standard 17 Industrial Classification (SIC) codes for manufacturing 18 industry groups. These include food and kindred products; 19 textile mill products; lumber and wood products; paper 20 and allied products; chemicals and allied products;  ;

21 petroleum refining and related industries; rubber and 22 miscellaneous plastic products; stone, clay and glass 23 products; primary metal industries; fabricated metal 24 products; nonelectrical machinery; electric machinery, 25 equipment and supplies; transportation equipment; and

-

26 all other manufacturing. Separate equations were

• n e/r/a

__ _ _ _ . . _ _ . . _ . . - __ . ._ _ _ _ _ _ _ . __

. - - - _. _ - _. - - . . ._

1 estimated for each group (except all other).

2 The next step in the process is to identify the 3 " causal" or " explanatory" variables to be included in

,

4 the equations to be estimated and to collect the data, 5 which generally cover the 48 contiguous states, each state 6 representing a data point. Depending upon availability, 7 data cover usage in one or more of the years in the period 8 1970-1976.

9 Once the data have been collected and organized, 10 the next step is to estimate a mathematical cause-and-11 effect relationship for each usage category. The 12 estimation process applied statistical techniques that 13 have come to be called "econometric"--hence the name, 14 econometric model.

15 To apply the econometric results to the Company's 16 service territory, I went through four steps. The first 17 was to obtain base-year (1973) data for the Company on

,

18 electricity usage by class, the number of residential 19 customers and economic activity. These data are necessary 20 to give the model the correct starting point. Values-21 are required for the following items:

22

• residential usage per residential customer, 23

• nonmanufacturing usage per dollar of nonmanufacturing 24 earnings,1 and 25 26 1 Deflated to 1967 dollars.

• n/e/r/a

.. .- . . . . - - . . . - -. .- - . . - . . - . - . ..

._ _ ._ . - - _ .- __

1

• manufacturing usage per dollar of manufacturing 2 earnings.2 3 The second step in applying the equacions to the model 4 was to use them to compute long-run elasticities of demand 5 with respect to the more important explanatory variables.3 6 Table 3 shows these elasticities by type and' user group.

7 The coefficients for the residential sector are composites 8 that reflect the relative importance of the various major 9 appliances in average household consumption. Those for 10 manufacturing similarly reflect the relative importance 11 of the various industry groups served by the company in 12 total manufacturing sales. In addition to these long-13 run coefficients, it is necessary to specify a lag coef-14 ficient for each user group. This coefficient determines 15 the ratio of short-run to long-run elasticity and governs 16 the rate at which pr.Je and other effects spread. (See 17 the Appendix for a concise description of the fore-18 casting equations.) Based upon econometric studies and 19 other data, I utilize the following values:

20

• residential 0.90 21

• nonmanufacturing 0.86 22

• manufacturing 0.85 23 24 2 Ibid.

25 3 An elasticity is a coefficient that measures the percent change in demand with respect to each 1 percent change 26 in a particular explanatory variable,

• nie/r/a

. _ _ _ _ . _ _ _ _ . _ . _ . _ . . ._. - . - _ _ .. .- _ _ _ . _ -

-- - _ _

1 The third step in applying the results is to develop 2 projections of the important explanatory variables: real 3 electricity and fuels prices, real wages, real income 4 per customer, number of residential customers, real 5 manufacturing earnings, and real nonmanufacturi.19 earnings.

6 The fourth step is to calibrate the model using 7 the data on actual sales for 1974-1979. This involves 8 setting the base-year value of the ratio of marginal usage 9 per customer (or per dollar of activity in the nonreri-10 dential sector) to average usage at a level th,at minimizes 11 the model's prediction error over the historical period.4 12 This ratio is not directly observable and can only with 13 difficulty and the risk of considerable error be estimated 14 independently. By estimating it implicitly through the 15 calibration procedure, the model's historical accuracy 16 is improved. This procedure also allows the model to 17 take into account any conservation activities that might 18 not have been predicted as part of the normal response 19 to price changes. A second calibration involves adjusting 20 the 1980-2000 projecticns for the prediction error remaining 21 for 1979 (after the first calibration).

22 Q. What assumptions did you make about the growth 23 of population and economic activity?

24 A. I relied heavily upon a set of projections made 25 26 4 This ratio is the parameter r g in Equation (1) of the l Appendix J i

n/e/r/a

. - . .

1 l

1 for the Houston-Galveston Standard Consolid.sted Statistical 2 Area (SCSA) by the Departments of Commerce and Agriculture.

3 These projections, made back in 1971, are known as the 4 "OBERS" projections--a name derived from the initials 5 of the agencies within the departments that did the work.

6 An updated study is now underway, but the results have 7 not yet been published. Despite its age, the OBERS report 8 is a valuable source because it presents projections that 9 are consistent across regions and various s6ctors of the 10 economy. This is because the OBERS method begins with 11 national population and output projections and then works 12 down to specific industries and regions.

13 Since actual data on population and economic activity 14 were available through 1978 for the SCSA, I was able to 15 check the OBERS projections against the actuals for that 16 year, as shown in Table 4. The table indicates that OBERS 17 substantially underpredicted growth. (This happened even 18 though OBERS overpredicted economic growth nationally.)

19 In using OBERS to make projections for 1978-2000, 20 I made three alternative assumptions: (1) After adjustment 21 for the error observed in 1978, the OBERS projections 22 will underpredict, but to a lesser extent than during 23 the 1971-1978 period and to a diminishing degree over 24 time. (2) After adjustment, the OBERS predictions will 25 be correct. (3) After adjustment, the OBERS predictions 26 will correctly describe the growth of the Houston-Galveston

'

IIIe/r/a

-. - _. . .

j 1 SCSA relative to the national rate of growth but will 2 actually overpredict to the extent that OBERS is too high 3 nationally. (OBERS projected growth of GNP of about 3.6 4 percent for the period 198r A00, but it overpredicted 5 growth in GNP per capita by about 0.65 percent per year 6 over the 1971-1980 period. Adjustment for tnis historic 7 bias gives projected growth of about 2.9 percent per year 8 nationally. This accords well with projections made 9 recently by other GNP forecasters.) Tables 5 through 10 7 display the projections obtained under each of these 11 three assumptions. Case (1) gives the highest overall

• 12 growth projection, and Case (3) the lowest. Case (3) 13 is extremely conservative, in that it totally discounts 14 the histcrical bias in the regional OBERS projection and 15 inrstead relies upon the historical bias in the national 16 OBERS projection, which was in exactly the opposite 17 direction.

18 Q. What assumptions did you make about future elec-19 tricity prices?

20 A. I relied in part upon the Company's projections 21 of electricity rates and in part upon recent Department 22 of Energy (DOE) forecasts of future energy prices for 23 its Region No. 6 comprising Texas, Oklahoma and New Mexico.$

24 J 25 5 U.S. Department of Energy,. Office of Conservational and Solar Energy, " Federal Energy Management and Planning 26 Programs; Methodology and Procedures for Life-Cycle Cost Analyses (Average Fuel Costs) ," Federal Register, October 7, 1980, pp. 66632-66635 and October 27, 1980, pp. 71344-71346 (Vol. 45, Nos. 196 and 209),

nietria' l

1 The Company projects growth in real electricity prices 2 ranging from 0.7 to 1.0 percent annually, depending upon 3 customer class, over the period 1980-1990. DOE, in what 4 it bills as a " pessimistic" forecast, projects nearly 5 zero growth from 1980 to 1985, annual growth of 1.5 to

,

6 2.0 percent from 1985 to 1990, depending upon class, and

,

7 slightly negative growth (about -0.3 percent per year) 8 from 1990-1995. The assumptions I have used are detailed 9 in Table 8. For Cases (1) and (2) real electricity prices 10 increase about as projected by the Company and DOE and 11 for Case (3) real electricity prices increase at 2.0 percent 12 per year, reflecting the possibility that future capacity 13 will be more expensive than currently planned, that older 14 gas- and oil-burning capacity will be displaced less 15 rapidly than presently anticipated or that future coal 16 costs will be higher than now forecasted.

17 Q. What assumptions did you make about fuel prices?

18 A. I relied in part upon the Company's projections 19 of natural gas prices and in part upon recent DOE forecasts 20 of future energy prices for its Region No. 6. The Company 21 projects growth in real gas prices ranging from nearly 22 zero for residential to 0.4 percent per year for indus-23 trial customers over the period 1980-1990. DOE projects 24 gas price increases of 1.7 to 1.8 percent per year, 25 depending upon class, from 1980 to 1985, annual growth 26 of 4.5 to 5.9 percent from 1985 to 1990, and growth ranging nie:r:a'

l 1

1 from 3.1 to 3.8 percent, again depending upon class, for 2 1990-1995. The price of distillate oil for industry usage J is projected to increase at about 3.5 percent annually 4 to 1995 and for residual oil at about 4.9 percent. The 5 assumptions I have used are indicated in Table 8. For 6 Case (1) real fuel prices increase about as projected 7 by DOE, which describes its assumptions as pessimistic.

8 For Cases (2) and (3) real fuel prices increase at roughly 9 the same rate as projected by the Company for natural 10 gas.

11 Q. How are these cases used in the model?

12 A. The Case (1) assumptions for economic growth and 13 prices are combined to yield an upper-end projection.

14 The Case (2) assumptions give a mid-range estimate, and 15 the Case (3) assumptions, again for both economic growth 16 and prices, yield a lower-end projection.

17 Q. How did you project the number of residential 18 customers?

19 A. I utilized the population projections noted above, 20 and assumed a constant ratio of customers to population, 21 since data on population per customer for the SCSA indi-22 cate no clear trend.

23 Q. What did you assume about self-generation?

24 A. I projected industrial sales net of self-generation, 25 because the model's price elasticity estimates, by virtue 26 of the way in which they were estimated, include an nie.T:a'

, -. .

- . . . -

l 1 allowance for the response of self-generation to price.

2 Q. Does your model take conservation into account?

3 A. Yes. It does so in connection with the projected 4 response to electricity price. That is, the estimated 5 effect of electricity price increases comprises such things 6 as adding insulation, stopping leaks, installing more 7 efficient equipment, etc. Table 9 gives some rough-cut 8 estimates of the size of this effect in percentage terms 9 for 1990 and 2000. On a kilowatt-hour per-customer basis 10 the residential-sector savings amount to 1300 kilowatt-

,

11 hours in 1990 and 2300 kilowatt-hours in 2000.

12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 I

l ruencial l i

- , . . _

I

. . , . . . , _ _ _ . _ _ _ , . - , . _ . . . _ _ _ . _..___ _ .-.

ACTUAL AND PROJECTED SALES AND PEAK LOADS 1979, 1990 and 2000 Year Actual 1990 2000 Item Units 1979 Case 1 Case 2 Case 3 Case 1 Case 2 Case 3 (1) (2) (3) (4) (5) (6) (7) (8)

Sales to Ultbaate Customers Residential gWh 11,079 17,714 15,524 14,764 25,430 19,712 16,736 Nonmanufacturing gWh 15,039 26,096 22,770 20,430 41,543 32,841 25,308 Manufacturing gWh 23,084 42,038 36,477 31,862 70,238 53,429 37,930 Other gWh 107 151 131 118 199 153 116 Subtotal gWh 49,309 85,999 74,901 67,,174 137,410 106,135 80,090 Resale Sales' gWh 3,052 5,585 5,585 5,585 7,346 7,346 7,346 Total Sales gWh 52,361 91,584 80,486 72,759 144,756 113,481 87,436 2 14,698 13,287 26,542 20,808 16,032 Gunner Peak Load led 9,336 16,725 Note: Totals may not add due to rounding.

I Company data.

• ExcluCes interruptible load.

Source Col. (2): Company data.

Cols. (3)-(8): NERA estimates, g b

i;

~

G

$

e

.

.

.. ..

.

.

. . .

.. .

.. .

COMPARISON OF THE COMPANY FORECAST WITH THE NERA PROJECTIONS 1990 and 2000 Year 1990 2000 NERA NERA Item Units Company case 1 Case 3 Company case 1 Case 3 (1) (2) (3) (4) (5) (6) (7) .

l 67,174 103,776 137,410 80,090 Total Sales

• gWh 76,826 85,999 2

15,050 16,725 13,287 20,375 26,542 16,032

Summer Peak MW 8 Excludes resale sales.

' Excludes interruptible load.

Source: NERA estimates.

l l

'

,

b 5;

l

.

~

l e.

.-.. o

-_ __ _ _ _ . -.

4 LONG-RUN ELASTICITY COEFFICIENTS USED FOR THE MODEL Explanatory User Group

Variables Residential Nonmanufacturing Manufacturing (1) (2) (3)

,

4 Price Electricity -0.75 -0.61 -0.90

, Fuels 0.40 0.29 0.33 Wages N/A N/A 0.19 Income Per Customer 0.41 N/A N/A Per Capita N/A 0.01 N/A Other Household. Size -0.61 N/A N/A Note: N/A indicates not applicable.

.

Source: NERA estimates based on Appendices A-C.

l e

b

N .

'

m a

d P.

_

-- - -

COMPARISON OF THE OBERS PROJECTION WITil AC'IUAL RESULTS FOR Tiik. HOUS' ION-GALVES'lON SCSA 1978 Percent Item Units Estimated Actual Error 3 (Units as Indicated) (Percent)

,

(1) (2) (3) (4)

Population Thousands of Persons 2,503 2,793 11.6%

Real Personal Income Millions of 1967 dollars 11,561 14,197 22.8 Real Income Per Capita 1967 dollars per capita 4,619 5,082 10.0.

Manufacturing Earnings Millions of 1967 dollars 2,201 2,510 14.1 Nonmanufacturing Earnings Millions of 1967 dollars 7,313 9,790 33.9 3

Actual vs. estimated.

Source: NERA estimates, i

d

>

! tn

,

M i to

• e

' 9 l

5, I

-. .

! .

.

AC'IUAL AND PROJECTED ECONOMIC ACTIVITY IN THE HOUSMN-GALVESMN SCSA CASE 1 c

Actual Units for Colhans Values Projected Values Projected Growth Item (2)-(4) 1978 1980 1990 2000 1978-1980 1980-1990 1990-2000

-(Units as Indicated)=---- -

(Percent)- ---

(1) (2) (3) (4) (5) (6) (7) (8)

~

Population (Thousands) 2,793 2,944 3,657 4,229 2.67% 2.224 1.44%

Income Per Capita (1967 $ per Person) 5,082 5,482 7,232 9,658 3.86 2.81 2.93 Psrsonal Income (Million 1967 $) 14,197 16,141 26,524 40,845 6.63 5.09 4.41 Mrnufacturing Earnings (Million 1967 $) 2,510 2,833 4,393 6,392 6.23 4.48 3.82 Nonmanufacturing Earnings (Million 1967 $) 9,790 11,239 18,H43 29,360 7.15 5.30 4.54 Source: NERA estimates, based on OBERS.

H b

R

=

D G

M P. .

_ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ . _ _ . _ _ _ _ _ _ _ _

ACTUAL AND PROJECTED ECONOMIC AC"flVITY IN THE HOUSEN-GALVES MN SCSA CASE 2 Actual Units for Columns Values Projected Values Projected Growth Item (2)-(4) 1978 1980 1990 2000 1978-1980 1980-1990 1990-2000

---

(Units as Indicated) = = - - -

(Percent) .

(1) (2) (3) (4) (5) (6) (7) (8)

Population (Thousands) 2,793 2,898 3,471 3,925 1.864 1.821 1.24%

Income Per Capita (1967 $ per Person) 5,082 5,408 6,894 9,051 3.15 2.46 2.75 Parsonal Income (Million 1967 $) 14,197 15,673 23,932 35,526 5.07 4.32 4.03 M nufacturing Earnings (Million 1967 $) 2,510 2,780 4,112 5,845 5.23 3.99 3.58 Nonmanufacturing Earnings (Million 1967 $) 9,790 10,780 16,284 24,084 4.94 4.21 3.99 Source: NERA estimates, based on OBERS.

8 i;

E5 m O

3 5

-- . . .- _. .- . .

ACRJAL AND PROJECTED ECONOMIC ACTIVITY IN THE HOUS1DN-GALVES10N SCSA CASE 3 Actual Units for Columns Values Projected Values Projected Growth Item (2)-(4) 1978 1980 1990 2000 1978-1980 1980-1990 1990-2000 (Units as Indicated) =- - = - - -

(Percent) -------

(1) (2) (3) (4) (5) (6) (7) (8)

Population (Thousands) 2,793 2,898 3,471 3,925 1.864 1.82% 1.24n

~

Income Per Capita (1967 $ per Person) 5,082 5,338 6,375 7,841 2.48 1.79 2.09 Parsonal Income (Million 1967 $) 14,197 15,470 22,131 30,778 4.39 3.65 3.35 Manufacturing Earnings (Million 1967 $) 2,510 2,744 3,803 5,064 4.55 3.32 2.91 Nonmanufacturing Earnings (Million 1967 $) 9,790 10,640 15,058 20,865 4.26 3.53 3.32 Source: NERA estimates, based on OBERS.

8 b

r,

<

e G

M

%

____

PROJECTED PERCENT GROWTH RATE I!4 ENERGY PRICES Item Case

- --(Percent per Year)----

(1) (2) (3)

Residential Prices Electricity 1980-1990 1.0% 1.0% 2.0%

1990-2000 0.0 0.0 2.0 Gas 1980-1990 3.0 0.5 0.5 1990-2000 2.0 0.0 0.0

.

Commercial Prices Electricity 1980-1990 1.0 1.0 2.0 1990-2000 0.0 0.0 2.0 Fuels 1980-1990 3.5 0.!; 0.5 1990-2000 3.0 0.0 0.0 Industrial Prices Electricity '

1980-1990 1.0 1.0 2.0 1990-2000 0.0 0.0 2.0 Fuels 1980-1990 4.0 0.5 0.5 1990-2000 3.0 0.0 0.0 e

Source: NERA estimates based on DOE and Company data. ,

M h m G

M We

ESTIMATED PERCENT CONSERVATION 1990 and 2000 Year / Case I User Group 1990 2000 I - - -- - - --- - - - - - - - - - ( P e r c e n t ) - - -- - -- - -- -- --- - - -

(1) or (2) (3) (1) or (2) (3) ,

l Residential 8.1% 8.4% 11.3% 14.6%

Nonmanufacturing 6.2 6.5 9.2 12.3 Manufacturing

• H24.6 25.0 28.7 33.0

• Includes usage reductions due to self-generation.

Source: NERA estimates.

a b

G

-

.

g

@

M Pn

i l

l APPENDIX MATHEMATICAL FORM OF THE FORECASTING EQUATIONS The mathematical form of the forecasting equations for the residential, nonmanufacturing and manufacturing sectors is the same for all three sectors. For each sector there are three equations:

C i

• m f X e U)

(1) t " 'o ' #o

• i=1 (X. 2/

eg =e *( }* (2) t -1 .

E t **t

• A t

II where e is electricity usage per unit of activity; r is the ratio of narginal to average usage per unit of activity; Xt is the ith explanatory variable (i=1,...,m); E is electricity usage; and A is the level of sectoral activity (number of customers in the residential sector and real employee earnings for the nonresidential sectors). The subscript t denotes future year t, and the subscript o denotes the base year.

The superscript

• denotes a long-run equilibrium value. The coefficient A is the lag coefficient. It determines the rapidity with which changes in the Xt's affect usage. The value (1-A) is the ratio of short- to long-run elasticity of demand. The coefficients c i are the long-run elasticities e

niett:a

. - - - - .. - - .- . ...

.

.

A-2 of demand with respect to the X i

's. The Eg's and 1 for each sector are computed from regression analyses. The values obtained appear in Table 3 of the testimony. Equation (1) indicates that the long-run target level of usage per unit of activity is a function of (a) the base-year level of i

usage per unit of activity, (b) the base-year ratio of marginal to average usage per unit of activity, and (c) changes in the explanatory variables from their base-year levels. Equation (2) states that this year's electricity usage per unit of activity is a weighted average of long-run target usage per unit of activity and last year's usage per unit of activity. Equation (3) states that this year's electricity usage is equal to the product of usage per unit of activity and the level of sectoral activity.

t i

D L

l l

l

!

<

@

!

n/e/r/a i

!