ML20003E977
| ML20003E977 | |
| Person / Time | |
|---|---|
| Site: | Comanche Peak  |
| Issue date: | 12/31/1980 |
| From: | Owen L TEXAS, STATE OF |
| To: | |
| Shared Package | |
| ML19240B984 | List:
|
| References | |
| NUDOCS 8104170620 | |
| Download: ML20003E977 (13) | |
Text
__
-y I
i DOCKET N0. 3460 i
i APPLICATION OF DALLAS POWER & LIGHT PUBLIC UTILITY COMMISSION COMPANY FOR AUTHORITY TO CHANGE RATES l
OF TEXAS 8
ffE DIRECT TESTIMONY OF
[,
LAURA J. 0 WEN 4
ECONOMIC RESEARCH DIVISION i
PUBLIC UTILITY COMMISSION OF TEXAS f',
i
! ~
DECEMBER 1980
+#
b 1*
w.-
810.4170(ch
- s 1
9 DOCKET N0. 3460 Page I of 11 i
L h
I Q.
Please state your name and business address.
2 A.
Laura J. Owen, 7800 Shoal Creek Boulevard, Suite 400N, Austin, Texas.
3 Q.
By whom are you employed and in what position?
4 A.
I am employed by the Public Utility Comission of Texas as an Economic
'=
5 Consultant.
6 Q.
What are your principal areas of responsibility in this capacity?
7 A.
The majority of my duties involve review and evaluation of the econometric 8
models employed in proceedings before the Comission.
9 Q.
Please state briefly your educational background and professional 10 qualifications.
I 11 A.
I received a Bachelors degree in economics from Vassar College, Poughkeepsie, 12 New York, in May of 1980. I was elected to Omicron Delta Epsilon, a national 13 economics honor society, in the fall of 1978, and to Phi Beta Kappa in the 14 spring of 1980.
15 Q.
Have you ever testified before this Comission?
16 A.
Yes, I testified in Docket No. 3370 (Comunity Public Service) regarding 17 weather, customer, and other O&M adjustments.
l I
18 Q.
Would you please state the scope of your analysis and the purpose of your 19 testimony in this case, Docket No. 34607 lL l
20 A.
Yes. There are two areas in which I will present testimony. First, I will I
21 consider the weather and customer adjustments to kwh sales as prepared by the 22 Company and propose Staff adjustments in these areas. Second, I will discuss
,t 23 the Company's adjustment to "other O&M expenses" and propose a Staff 24 adjustment in this area.
This prepared testimony has, therefore, been l'
25 organized into two sections:
.i
.EL
DOCKET NO.
3460 Page 2 of 11 h
h f
I I.
Weather and Customer Adjustments 2
II. Other Operation and Maintenance Expense Adjustment 3
Weather and Customer Adjustments 4
Q.
Have you reviewed the Company's proposed adjustment to sales for the effects.
5 of abnormal weather?
6 A.
Yes, I have.
7 Q.
Would you briefly explain the Company's procedure?
~
8 A.
Yes.
The Company's weather adjustment can be outlined as a three step 9
procedure. In the first step the relationship between daily system input and 10 daily mean temperature was estimated using ordinary least squares regression
!5 11 techniques and the functional form:
S 2
12 Y = a + bX + cX 13 where Y = daily system input 14 X = daily mean temperature 15 and a,b,c = estimated coefficients 16 A separate equation was estimated for each month of the test year.
17 Using these twelve equations the Company calculated experienced system 18 input and normal system input for each month of the test year.
This 19 calculation was accomplished by plugging in values for daily mean temperature 20 and daily normal temperature (for the calendar month) into each monthly 21 equation. The difference between the calculated normal system input and the 22 calculated experienced system input was the total weather adjustment to 23 system input for each month of the test year. These twelve adjustments were 24 then added to actual (not calculated) system input for the corresponding l.'
25 months to obtain weather normalized sytem input for each month of the test
.0 L
DOCKET NO. 3460 page 3 of 11 M
i,'
~
l year.
2 In the second step of the Company's weather adjustment the relationship h
3 between total general business sales and total system input was estimated for 4
each month using data from 1969-1978 and ordinary least squares regression 5
techniques. The functional form for these twelve equations was:
6 St " "It + bit-1 for t = 1... 12 7
where S = total general business sales for month t t
8 I = total system input for the same month t
9 It-1 = total system input for the previous month l'
10 and a,b = estimated coefficients y
11 By plugging the monthly values for actual system input and weather 12 normalized system input (from Step 1) into the corresponding equations, the 13 Company calculated actual general business sales and weather normalized 14 general business sales for each month.
The difference between the weather 15 normalized and the actual general business sales was the total weather
{
16 adjustment to kwh sales for each month.
~
17 Step 3 of the Company's weather adjustment was used to allocate the 18 monthly weather adjustments (from Step 2) to the thirteen weather-sensitive l
l 19 ratt -lasses.
The basic procedure used by the Company was to estimate the 20 relatk ' ship between kwh sales and weighted degree days (the sum of cooling l
j 21 degree days and heating degree days weighted by the billings in each cycle).
6l
22 These equations were estimated using ordinary least squares regression 1.
23 techniques and monthly 1978 data.
Based on these estimated equations the j,
24 Company calculated weather adjustments for each month for each of the
'~'
25 thirteen weather-sensitive rate classes.
The Company then determined the L5
DOCKET N0.
3460 Page 4 of 11
' q 1
percentage of. the monthly weather adjustments that each rate class was i
2 responsible for and used these percentage figures to allocate the total
{
3 monthly weather adjustments calculated in Step 2.
4 Q.
Do you have any comments regarding the procedure just outlined?
5 A.
Yes.
Although I believe tiere is some validity in the Company's procedure 6
for adjusting for the effects of abnormal weather, I have difficulty placing 7
a great deal of confidence in certain steps and results of this procedure.
I 8
With regard to the first step of the weather adjustment process, I have 9
three coments. First, I believe that the Company's process of calculating 10
" normal" weather and ranking it on a daily basis within each month leads to h
11 daily temperature values which are not normal.
p~
12 Q.
Could you explain how the Company calculated the normal daily temperatures 13 for each month?
14 A.
Yes, I will use January as an example.
The Company's first step was to 15 collect thirty years (1946-1975) of daily weather data for the month of 16 January. Then, the coldest mean temperature in each of the thirty Januarys S
17 was averaged to determine the " normal" mean temperature for the first of j
18 January.
Then, the second coldest mean temperature in each of the thirty e
19 Januarys was averaged to calculate the " normal" mean temperature for January 20 2nd. This process was continued until the hottest mean temperature in each 21 of the thirty Januarys was averaged to obtain the " normal" mean temperature 22 for January 31st. This process led to ranked " normal" mean temperatures for 23 the month of January from 23.8 F to 65.4 F.
Employment of this process 24 insures a wider variation between the lowest and highest " normal" mean 25 temperatures for any month than if a straight average of the mean temperature d
i L
DOCKET NO.
3460 Page 5 of 11 4
I
~
l on a given day over thirty years was used. By insuring a wider variation in i
2
" normal" temperatures, this process will certainly insure a larger (in
,f 3
absolute terms) weatner adjustment for each month.
4 I do not believe that this averaging of the extremes leads us to the 5
best approximation of normal daily mean temperatures.
In fact, I find it 6
very difficult to believe that 23.80F and 65.40F are the actual normal 7
temperatures for the first and thirty-first of January in the Dallas Power 8
and Light service area. The ranking of the " normal" mean temperatures from 9
low to high in each month also seems to miss the best approximation of normal 10 weather. This ranking may seem realistic in months such as February, March, 11 and April when temperatures are generally colder in the beginning of the 12 month, but it is probably just as much unrealistic in months such as October, 13 November, and December when temperatures tend to get progressively colder 14 throughout the month.
15 Q.
Do you have any comments on the data used as input in the regression analysis 16 that is part of Step 17 17 A.
Yes, I do.
In estimating the relationship between system input and daily 18 mean temperature for each month of the test year, the Company did not use 19 system input and temperature values from each day of the month.
- Instead, 20 they gathered data for the working days of the month in question and for the 21 last five and fitst five working days of the past and following months.
I 22 realize that the system inout on weekends and holidays may vary from the 23 system input on workdays, but I do not see this variation as sufficient 1
24 justification for excluding non-working days from the estimation procedure.
25 In fact, it seems this exclusion would possibly lead to greater inaccuracy in
'i 4
L
o DOCKET N0.
3460 Page 6 of 11
{
l the weather adjustment.
What the Company has done is estimated the 2
relationship between working day system input and temperature. However, they f
3 use this estimated relationship to determine the weather adju?,tment for each 4
day of the month (workdays, weekends, and holidays). This can be seen in the 5
calculations of experienced system input and normal system input previously
~
6 outlined.
7 Q.
Do you have any coments on the statistical results the Company obtained in 8
the first step of their weather normalization procedure?
9 A.
Yes. With a sample of approximately thirty observations a t-statistic of at
~
10 least 2.048 is needed to be ninety-five percent confident that the estimated
{l 11 coefficient is significantly different from zero.
If an independent
)F 12 variable's coefficient is not significantly different from zero, then that 13 variable has not been shown to have any explanatory power over the dependent 14 variable.
In the Company's results there are twelve t-statistics that fall 15 below this desirable level. The results for the models of April, June, July, u
{
16 and August are particularly problematic because none of the coefficients for 17 either the mean temperature or the squared mean temperature variables are 18 significantly different from zero at a ninety-five percent confidence level.
19 This fact is particularly disturbing when one considers that the adjustments 20 for June and August are the two largest (in magnitude) of all the weather 21 adjustments to system input.
22 Q.
Do you have any comments regarding the second step of the Company's weather 23 adjustment procedure?
24 A.
Yes, I would like to note two aspects of Step 2 that cast some doubt on the 25 significance of the results that the Company obtained therein.
The first i
5L i
DOCKET NO.
3460 Page 7 of 11 4
'i
{
problem I see with Step 2 is the small sample that is used in the estimation l
2 of the monthly equations.
Monthly data is used for the period 1969-1978, f
3 which leads to the estimation of each equation on only ten observations.
I 4
would also point out that the sample upon which the twelve equations were.
5 estimated does not include any of the test year data (July 1979 - June 1980).
6 '
The second aspect of Step 2 that I would like to comment on is the same 7
problem found in the results of Step 1 -- low t-statistics.
With a sample 8
size of ten a t-statistic of at least 2.306 is necessary to be ninety-five 9
percent confident that an estimated coefficient is significantly different 10 from zero. One quarter of the t-statistics for the coefficients estimated in 1
[;.,
11 Step ? do not meet this minimum requirement.
The equations for January, M
12 March, April, May, June, and December all have one coefficient with a t-13 statistic below the desireable level.
With six of the twelve estimated 14 equations showing insignificant coefficients I
would question the 15 appropriateness of the functional form: S = alt + bit-1 It is also t
16 interesting to note that the months of April, June, and December show low 17 t-statistics in both the Step 1 and Step 2 estimations.
i 18 Q.
Do you have any comments on the allocation process (Step 3) of the Company's 19 weather adjustment?
t 20 A.
Yes, I have several comments on the data used and results obtained in Step 3.
21 Similar to my comment on Step 2, I am concerned with the small sample sizes
~
22 and the lack of test year data used in 'the estimation of the relationship 23 between kwh sales and weighted degree days. Nine of the thirteen equations 24 were estimated on a seasonal (winter and summer) basis using 1978 data, which
\\
25 resulted in sample sizes of six observations. The remaining four equations
.L
DOCKET NO.
3460 Page 8 of 11 Il 1:
~
l were estimated for the whole year with one observation from each month of 2
1978.
Again the Company did not include test year data in its estimation f
3 procedure.
4 Q.
Do you have any comments on the variables used in the Step 3 regression 5
analysis?
6 A.
Yes, the Company used the variable weighted degree days to explain kwh sales.
7 By calculating this variable as the weighted sum of cooling degree days and 8
heating degree days, the Company made the assumption that kwh sales respond in the same manne-to the temperature variations measured by cooling degree 9
10 days and heating degree days.
If the Company had used cooling degree days r-11 and heating degree days as separate variables, they would have al! owed for A
12 the separate measurement of the effects of each on kwh sales.
13 Q.
Would you like to comment on the Company's use of the results obtained in the 14 Step 3 regression analysis?
15 A.
Yes. I am concerned with the exclusion of the Commercial and Industrial G --
16 No Riders regression equations for the winter months (November through April) 17 from the allocation procedure. The t-statistics for the coefficients in both 18 equations are certainly of a magnitude to indicate a significant difference 19 from zero, and the r-squareds for both equations (87.8 and 81.7, 20 respectively) are considerably high for such simple models. I agree with the 21 Company that the negative coefficients obtained for the weighted degree day 22 variable in both models is not what I would have theoretically expected.
23 However, I would not consider results different from what I expected as 24 grounds for ignoring those results.
25 Q.
In consideration of the problems you have found with the Company's weather i
i w
DOCKET NO. 3460 Page 9 of 11
~
l adjustment, do you recommend that the entire adjustment be disallowed?
2 A.
No, I do not.
Although I am not confident of how accurately the Company's f
3 procedure measures the need for a weather adjustment, I do believe that test 4
year weather did deviate from normal weather and that a weather adjustment is 5
warranted. I have calculated a weather adjustment according to the Company's a
6 model but using a ninety-five precent confidence interval of normal mean 7
temperature.
Based upon these calculations I am recommending a weather 8
adjustment to kwh sales of 59,065,095 kwh as shown on Schedule I.
In 9
calculating this adjustment I have not corrected for the various problems 10 outlined above because I feel that overall changes, rather than piecemeal U
11 corrections, are needed to rectify the inadequacies of the Company's weather
'1 12 adjustment model.
13 Q.
Did you review the Company's adjustment to kwh sales for the effects of 14 serving an increased number of customers?
15 A.
Yes, I did.
16
{
Q..
Do you have any comments on this adjustment to kwh sales?
17 A.
Yes. I do not have any problems with the Company's procedure for calculating 18 a customer adjustment using weather normalized kwh per customer. However, I 19 have proposed a weather adjustment different from that of the Company's and t
20 therefore cannot agree with the customer adjustment they have recommended.
21 When I realized that I would probably be making an alternative recommendation 22 on the weather adjustment, I asked the Company to rerun their customer 23 adjustment program using experienced kwh sales per customer. It is upon the 24 results of this computer run that I have made my recommendations as presented 25 in Schedule I.
i L
DOCKET NO.
3460 Pag 2 10 of 11 N
d, r
I Q.
Do you have any comments regarding the negative adjustments to lines 5, 10, 2
and 15 of Column (g) on the Company's Schedule N-3, page 1 of 37 f
3 A.
Yes.
I have disallowed the negative adjustments to kwh sales for the 4
General, Industrial Primary, and Municipal Pumping rate classes. These three 5
negative adjustments are not known and measurable adjustments to kwh sales 6
for a change in the number of customers served and therefore are incorrectly 7
included under Column (g), Schedule N-3, page 1 of 3.
In addition, these 8
adjustments are not properly explained by the Company in the testimony or 9
schedules of the rate filing package.
10 Other Operation and Maintenance Expense Adjustment 11 A.
Did you review the Company's methodology for adjusting "other 0&M expenses"?
I; Vi 12 A.
Yes, I did.
13 Q.
Do you have any comments regarding this adjustment?
14 A.
Yes.
I agree with the general manner in which the Company calculated this 15 adjustment, but have slightly refined the analysis and proposed a Staff 16 adjustment based upon this analysis.
17 Q.
Would you please explain your refinement and the resulting adjustment?
18 A.
Yes. My refinement consists of further statistical techniques which enhance l9 the accuracy of the model proposed by the Company.
The calculation of my i
20 recommendation can be found in Schedule II.
The major change is that the 21 Company adjusted other O&M expenses to the predicted value with the year end 22 level of customers while the Staff recommends adjusting to the closest 23 boundary of a ninety-five percent confidence interval around that predicted 24 value.
25 Q.
Why should the adjustment be made to the boundary of a confidence interval
?
I
^
DOCKET NO.
3460 Page 11 of 11 ii G.
P 1
rather than to the predicted value?
2 A.
The predicted value, by itself, is only an indicator of the neighborhood in f
3 which the true value will lie.
By adjusting to the predicted value, the 4
Company is claiming accuracy which is simply not real. As stated in several
.c 5
previous rate cases before this Commission an adjustment of this type should 6
be made only to the boundary of a statistically reliable confidence interval.
7 Q.
Does this conclude your direct testimony in this case?
8 A.
Yes, it does.
9
~,
10
!)
II
'I 12 13 14 15 16 l..
17 19 20 1
21 22 23 24 25 s
i L
DOCKET NO. 3460 Schedule I Page 1 of 1 PUBt!C UTILITY CODMISSTON OF TEIAS DALLAS P0mER & L15 P COMPAM SL* MARY OF STAFF K.H MsTAIETS
'l Rate Experienced kWh Adjustments Adjusted Description of Service Schedule kWh Sales
_ Weatner
_ customers kWh Sales W
Residential RS 2,433,569.984 68,249,550 13.230,201 2,515,049,735 Residential Water Heating RS-WH 45.423,974 485.890 2,246,865 48,156,729
)
Residential Space Heating RS-RH 244,590,634
-6,920,211 82,050,713 319,721,136 Residential Water & Space Heating RS-WHRH 286,576,201
-12,705,343 54,751,500 328,622,358 General G
5,478,313,040 29,404,147 100,130,788 5,607,847,975 General Space Heating. Option A G-GHA 15,821,629
-135,452 478,388 16.164,565 General Space Heating, Option B G-GHS 1,747.126.012
-19,429.203 77.534,715 1,805,231,524 Ceneral Short-Term G-T 32,846,378 0
-3,563,843 29,282.535 General Stand-By & Supplementary G-S 32,455,440 115,717 0
32,571.157 m
Industrial Primary IPS 510,588,640 0
0 510,588,640 q
Outdoor Lighting OL 469,717 0
1,005 470.722
!'i Street Lighting SL 80,814,832 0
0 80,814,832 Traffic Signal TS 11,406,793 0
0 11,406,793 Miscellaneous Municipal J
15,106.285 0
32.252 15,138,537 e
Municipal Pumping MP 222.335.548 0
928 222.336.476 General Business Sales 11,157,445,107 59,065,095 326,893,512 11.543,403,714 Sales to Other Electric Utilities 192,319.000 0
0 192.319.000 Total Energy Sales 11,349.764.107 59.065.095 326,893.512 11.735.722.714 i
I i,
l t
l l -
i I k l
L t
l-Schedule II DOCKET NO. 3460 Page 1 of 1 PUBLIC UTILITY C0411SSION OF TEXAS DALLAS POWER & LIGHT COMPANY OTHER OPERATION & MAINTENANCE EXPENSE ADJUSTMENT Yt=a (1 - $ ) +
8 (Xt-6Xt-1) +$Yt-1 where 8
-148,227,000
=
[
$=
672.174
~
$=
.834961 X and X t
t-1 = 299,733 (year end number of customers)
Yt-1 = 44,639,454 il G
Y
$46,059,825
=
t Forecast Interval: i (S )(t value) f 46,059,825 (545520.242)(2.048) r, 1
ii_
Lower bound of forecast inverval:
44,942,600 l4 Amount of Other 0&M on books:
44,778,324 1
Adjustment to Other O&M.
5 164,276 Staff recomendation for the adjustment to other 0&M expenses:
$164,276.
l O
~
i i
L