**************************************************************************
NOTICE
*********************************
***********************
This document was converted from WordPerfect to ASCII Text format.
Content from the original version of the document such as headers, footers, footnotes,
endnotes, graphics, and page numbers will not show up in this text version. All
text attributes such as bold, italic, underlining, etc. from the original document will not
show up in this text version. Features of the original document layout such as columns, tables,
line and letter spacing, pagination, and margins will not be preserved in the text version. If you
need the complete document, download the WordPerfect version or Adobe
Acrobat version, if available.
*********************************
********************************
APPENDIX B
METHODOLOGY FOR ESTIMATING OUTSIDE PLANT COSTS
I. Introduction
1. Section II in this appendix explains in specific detail the regression equations and the
adjustments to these equations for estimating the input values adopted in this Order for structure and cable
costs. These regression equations and these adjustments are set forth in this appendix on the following
tables: Table I., labeled "Regression Equations Derived From RUS Data For Estimating Cable And
Structure Costs;" Table II., labeled "Adjustments To Regression Equations Derived From RUS Data For
Estimating Cable And Structure Costs;" and Table III., labeled "Regression Equations Derived From Non-
Rural LEC Data For Estimating Cable Costs."
2. Section III illustrates use of the Huber methodology to derive reasonable estimates for 24-
gauge aerial copper cable costs. This illustration uses the diagram in this appendix labeled "Scatter
Diagram Of 24-Gauge Aerial Copper Cable Cost And Size With The Huber Regression Line." This
diagram shows RUS cable cost observations for 24-gauge aerial copper cable and the regression line fit to
these observations by using the Huber methodology. It also uses the frequency distribution in this appendix
set forth on Table IV., labeled "Frequency Distribution Of Huber Weights For 24-gauge Aerial Copper
Cable Cost." This frequency distribution shows the number of aerial copper cable observations to which
the Huber methodology assigns particular weights.
3. Section IV demonstrates that the Huber methodology generally does not have a statistically
significant impact on the level of the material costs reflected in the cable cost estimates adopted in this
Order. This finding provides support for the large LEC buying power adjustment reflected in these
estimates. This finding is supported by the statistical information set forth in this appendix on Table V.,
labeled "Analysis Of Coefficient For Cable Size Variable In The Huber Regression Equations."
II. Regression Equations For Estimating Outside Plant Structure Costs
A. Regression Equations Derived From RUS Data For Estimating Cable
And Structure Costs
4. Table I, labeled "Regression Equations Derived From RUS Data For Estimating Cable
And Structure Costs," sets forth the regression equations adopted in this Order for estimating the cost of:
(1) 24-gauge aerial copper cable; (2) 24-gauge underground copper cable; (3) 24-gauge buried copper
cable and structure; (4) aerial fiber cable; (5) underground fiber cable; (6) buried fiber cable and structure;
(7) poles; and (8) underground structure. These regression equations, other than the equations for poles
and underground structure, are developed by revising the regression equations for cable and structure costs
developed by Gabel and Kennedy in the NRRI Study. The regression equations adopted in this Order,
other than the equation for poles, are estimated by using the Huber methodology with RUS data. The
regression equations in the NRRI Study are developed by using ordinary least squares (OLS) with RUS
data. The regression equation for poles adopted in this Order is the regression equation for poles in the
NRRI Study. The regression equation adopted in this Order for poles is not estimated by using the Huber
methodology because the Huber regression for poles is not statistically significant at the five percent level.
5. Column A identifies, by type of cost, the regression equations adopted in this Order. Set
forth in columns B, D, F, H, J, L, and N are the intercepts and the slope coefficients reflected in these
regression equations. The coefficients set forth in these columns for these regression equations are for the
variables that indicate the size of a cable, density zone, soil surface texture, bedrock type, combined
bedrock and soil type, and the presence of a high water table. Columns C, E, G, I, K, M, and O display
the t-statistics used to measure the statistical significance of these intercepts and coefficients. Column P
displays the F-statistics used to measure the statistical significance of these regression equations. Column
O displays the number of observations in the data used to estimate these equations.
6. The coefficients for the variable that indicates the size of the cable in the regression
equations for 24-gauge copper cable cost and fiber cable cost do not reflect the adjustments adopted in this
Order for large LEC buying power. The intercepts and the coefficients in these equations do not reflect
splicing and LEC engineering costs because these costs are not reflected in the RUS data from which these
equations are derived. The intercepts and the coefficients for the water, soil, and bedrock indicator
variables in the regression equations for structure costs do not reflect LEC engineering costs because these
costs are not reflected in the RUS data from which these equations are derived. The intercept and the
coefficients for the water, soil, and bedrock indicator variables in the regression equation for pole costs do
not reflect costs for anchors, guys, and other pole-related items because these costs are not reflected in the
RUS data from which this equation is derived.
B. Adjustments To Regression Equations Derived From RUS Data For
Estimating Cable And Structure Costs
7. Table II, labeled "Adjustments To Regression Equations Derived From RUS Data For
Estimating Cable And Structure Costs," sets forth adjustments to the regression equations adopted in this
Order for estimating costs for 24-gauge copper cable, fiber cable, and structure. The equations that reflect
these adjustments, i.e., the adjusted equations, are used for estimating the cost of: (1) 24-gauge aerial
copper cable; (2) 24-gauge underground copper cable; (3) 24-gauge buried copper cable; (4) aerial fiber
cable; (5) underground fiber cable; (6) buried fiber cable; (7) aerial structure; (8) underground structure;
and (9) buried structure.
8. Column A identifies, by type of cost, the adjusted equations used to derive the cable and
structure costs adopted as input values in this Order.
9. Column B displays the intercepts in the adjusted equations. In the adjusted equations for
the cost of aerial and underground 24-gauge copper cable, fiber cable, and structure, the intercepts are
those in the regression equations for these costs. The intercepts in the adjusted equations for 24-gauge
buried copper cable and buried fiber cable represent the fixed cost of buried copper cable and the fixed cost
of buried fiber cable, respectively. The intercepts in the regression equations for 24-gauge buried copper
cable and structure and buried fiber cable and structure represent the fixed cost of buried copper cable and
structure and the fixed cost of buried fiber cable and structure, respectively, in density zone 1. The fixed
cost of 24-gauge buried copper cable used as the intercept in the adjusted equation for 24-gauge buried
copper cable, approximately $.46 per foot, is derived by subtracting from the intercept in the regression
equation for 24-gauge buried copper cable and structure, approximately $1.16 per foot, the value of the
fixed cost for buried structure in density zone 1 adopted in this Order, $.70 per foot. The fixed cost of
fiber cable used as the intercept in the adjusted equation for fiber cable, approximately $.47 per foot, is
derived by subtracting from the intercept in the regression equation for buried fiber cable and structure,
approximately $1.17 per foot, the $.70 per foot fixed cost adopted for buried structure in density zone 1.
The intercept in the adjusted equation for buried structure represents the fixed cost of buried structure in
density zone 1. The fixed cost of buried structure in density zone 1 used as the intercept in the adjusted
equation for buried structure is the $.70 per foot fixed cost adopted for buried structure in density zone 1.
10. Column C displays the coefficients for the cable size variable in the adjusted 24-gauge
copper and fiber cable equations. In the adjusted equations for the cost of aerial and underground 24-gauge
copper cable and fiber cable, the coefficients for the cable size variable are those for this variable in the
regression equations for these costs. In the adjusted 24-gauge copper cable equation, the coefficient for the
cable size variable is the coefficient for this variable in the 24-gauge buried cable and structure regression
equation. In the adjusted 24-gauge fiber cable equation, the coefficient for the cable size variable is the
coefficient for this variable in the buried fiber cable and structure regression equation.
11. Column D displays the large LEC buying power adjustment factors. These factors are
applied to the coefficients for the cable size variable in the adjusted copper and fiber cable equations.
Column E displays the values of the coefficients for these cable size variables in these equations, as
adjusted to reflect large LEC buying power.
12. Columns F, G, and H display the coefficients for the density zone, bedrock indicator, and
combined soil and bedrock indicator variables in the adjusted structure equations. In the adjusted equations
for the cost of aerial and underground structure, these coefficients are those for these variables in the
regression equations for these costs. In the adjusted buried structure equation, these coefficients are those
for these variables in the 24-gauge buried copper cable and structure regression equation. The coefficients
for the water and soil indicator variables in the structure regression equations are not reflected in the
adjusted equations because the value for these variables is set equal to zero to estimate structure costs used
as input values.
13. Column I displays the loading factors used to reflect splicing costs in the cable cost
estimates for 24-gauge copper cable and fiber cable.
14. Column J displays the loading factor used to reflect LEC engineering costs in the structure
cost estimates.
15. Column K displays the flat dollar loading used to reflect LEC engineering costs in the
cable cost estimates for 24-gauge copper cable and fiber cable.
16. Column L displays the adjusted equations used to estimate costs for aerial, underground,
and buried 24-gauge copper and fiber cable, buried and underground structure, and poles.
17. Columns M-O display adjustments to the adjusted pole equation. These adjustments add
to the cost of poles the costs for anchors, guys, and other pole-related items, including LEC engineering
costs associated with these additional items, and convert per pole costs, inclusive of costs for anchors,
guys, and other pole-related items, i.e., aerial structure costs, to per foot costs. Column M displays the
costs for anchors, guys, and other pole-related items for density zones 1 and 2 ($32.98 per pole), density
zones 3-7 ($49.96 per pole), and density zones 8 and 9 ($60.47 per pole). Column N displays the
loading factor used to reflect LEC engineering costs in the costs for anchors, guys, and other pole-related
items. Column O displays the distance between poles used to calculate aerial structure cost per foot for
density zones 1 and 2 (250 feet per pole), density zones 3 and 4 (200 feet per pole), density zones 5 and 6
(175 feet per pole), and density zones 7-9 (150 feet per pole).
18. Column P displays the adjusted equation used to estimate aerial structure cost per foot,
including poles, anchors, guys, and other pole-related items.
19. We illustrate how the adjusted equations are used to develop the input values adopted in
this Order by calculating the cost for a 100-pair 24-gauge aerial copper cable. Column L sets forth the
adjusted equation used to develop the input values adopted in this Order for 24-gauge aerial copper cable.
The adjusted equation set forth in column L for 24-gauge aerial copper cable is as follows:
A1 = (B1 + (E1)(# of Prs.))(1 + I1) + K1
where:
A1 = 24-gauge aerial copper cable cost per foot;
B1 = the intercept for 24-gauge aerial copper cable in dollars per foot;
E1 = the coefficient, adjusted for buying power, in dollars per pair per foot, for the variable
that represents the number of 24-gauge aerial copper cable pairs;
I1 = the splicing loading for 24-gauge aerial copper cable expressed as a percentage;
K1 = the LEC engineering loading for 24-gauge aerial copper cable in dollars per foot.
20. By substituting into the above equation for 24-gauge aerial copper cable the values from
Table II for the intercept, adjusted coefficient for the cable size variable, splicing loading, and LEC
engineering loading, and the number of cable pairs in this example, 100, we obtain the following estimate
for the cost of a 100-pair 24-gauge aerial copper cable:
A1 = (1.014907 + (.008329)(100))(1 + .094) + .19
= (1.014907 + .8329)(1.094) + .19
= (1.847807)(1.094) + .19
= 2.021501 + .19
= $2.21 per foot.
We adopt this estimate as the input in the model for the cost of a 100-pair 24-gauge aerial copper cable.
C. Regression Equations Derived From Non-Rural LEC Data For Estimating
Cable Costs
21. We adopt in this Order a methodology to derive estimates of 26-gauge copper cable costs
from 24-gauge copper cable costs. We first estimate by using the Huber methodology with RUS data the
cost for 24-gauge copper cable for each cable size. We then obtain by using the Huber methodology with
certain non-rural LEC data estimates of the cost for 24-gauge copper cable and 26-gauge copper cable for
each cable size. We next divide the 24-gauge copper cable cost estimate derived from the non-rural LEC
data into the estimate for 26-gauge copper cable cost derived from these data for each cable size. The
result is a ratio of 26-gauge copper cable cost to 24-gauge copper cable cost for each cable size. Finally,
we multiply this ratio by the estimate of the cost for 24-gauge copper cable derived from the RUS data to
obtain the cost for 26-gauge copper cable for each cable size. We adopt these estimates as inputs for 26-
gauge copper cable costs in the SM.
22. Table III, labeled "Regression Equations Derived From Non-Rural LEC Data For
Estimating Cable Costs," sets forth regression equations derived from the non-rural LEC data for: (1) 24-
gauge aerial copper cable; (2) 24-gauge underground copper cable; (3) 24-gauge buried copper cable; (4)
26-gauge aerial copper cable; (5) 26-gauge underground copper cable; and (6) 26-gauge buried copper
cable. We use these regression equations to develop the ratios of 26-gauge copper cable costs to 24-gauge
copper cable costs used to derive the cost for 26-gauge copper cable. Column A identifies these regression
equations by type of copper cable cost. Set forth in columns B and D are the intercepts and the slope
coefficients reflected in these regression equations. Columns C and E display the t-statistics used to
measure the statistical significance of these intercepts and coefficients. Column F displays the F-statistics
used to measure the statistical significance of these regression equations. Column G displays the number of
observations in the data used to estimate these equations. Column H shows the regression equations
derived from the non-rural LEC data for estimating costs for 24-gauge and 26-gauge copper cable.
23. These regression equations are derived from ex parte data on 24-gauge and 26-gauge
copper cable costs submitted by Sprint and Aliant, data on these cable costs submitted by BellSouth with
its comments, and the BCPM default values for these cable costs. These regression equations are
developed by using the Huber methodology. Using the Huber methodology with non-rural LEC data to
estimate cable costs for 24- and 26-gauge copper cable costs is consistent with use of this methodology to
estimate 24-gauge copper cable costs from the RUS data. The regression equations derived from non-rural
LEC data use the number of copper cable pairs as the sole independent variable. Using the number of
copper cable pairs as the sole independent variable in these regression equations is consistent with using
this variable as the sole independent variable in the regression equations for 24-gauge copper cable costs
estimated from the RUS data.
24. In this Order, we find it reasonable to rely on the non-rural LEC data for calculating the
ratio of the cost for 24-gauge copper cable to that for 26-gauge copper cable but not for calculating the
absolute cost for 24-gauge copper cable and 26-gauge copper cable. As discussed in this Order, we find
that the non-rural LEC data is not a reliable measure of absolute costs. Notwithstanding this finding, we
conclude that it is reasonable to use the non-rural LEC data to determine the relative value of the cost for
24-gauge copper cable to that for 26-gauge copper cable. We find that it is reasonable to conclude that
each LEC used the same methodology to develop both 24-gauge and 26-gauge copper cable costs.
Accordingly, any bias in the costs for 24-gauge and 26-gauge copper cable that results from using a given
methodology is likely to be in the same direction and of a similar magnitude. As a consequence, cost
estimates for 24-gauge and 26-gauge copper cable for each cable size developed from non-rural LEC data
by using the Huber methodology are likely to be biased by approximately the same factor. The ratios of
these estimates are not likely to be affected significantly because the bias in one estimate approximately
cancels the bias in the other estimate when the ratio is calculated.
25. We illustrate how we calculate the costs that we adopt in this Order for 26-gauge copper
cable by calculating the cost for a 100-pair 26-gauge aerial copper cable. As explained above, we derive a
ratio of 26-gauge copper cable cost to 24-gauge copper cable cost from non-rural LEC data to obtain costs
for 26-gauge copper cable. To calculate this ratio for a 100-pair aerial copper cable, we estimate
separately from non-rural LEC data the cost for a 100-pair 24-gauge aerial copper cable and a 100-pair
26-gauge aerial copper cable. We first estimate the numerator of this ratio, i.e., the cost for a 100-pair 24-
gauge aerial copper cable. Column H shows the regression equation derived from non-rural LEC data for
estimating the cost for 24-gauge aerial copper cable. The regression equation set forth in column H for 24-
gauge aerial copper cable is as follows:
A1 = B1 + (D1)(# of Pairs)
where:
A1 = 24-gauge aerial copper cable cost per foot;
B1 = the intercept for 24-gauge aerial copper cable in dollars per foot;
D1 = the coefficient in dollars per pair per foot for the variable that represents the number of
24-gauge aerial copper cable pairs.
26. By substituting into the above equation for 24-gauge aerial copper cable the values from
Table III for the intercept and the coefficient for the cable size variable, and the number of cable pairs in
this example, 100, we obtain the following result for the cost of a 100-pair 24-gauge aerial copper cable:
A1 = 2.1548 + (.012393)(100)
= 2.1548 + 1.2393
= $3.39 per foot.
27. We next estimate the denominator for the ratio of 26-gauge aerial copper cable cost to 24-
gauge aerial copper cable cost for a 100-pair aerial copper cable, i.e., the 26-gauge aerial copper cable cost
for a 100-pair cable. Column H shows the regression equation derived from non-rural LEC data for
estimating the cost for 26-gauge aerial copper cable. The regression equation set forth in column H for 26-
gauge aerial copper cable is as follows:
A4 = B4 + (D4)(# of Pairs)
where:
A4 = 26-gauge aerial copper cable cost per foot;
B4 = the intercept for 26-gauge aerial copper cable in dollars per foot;
D4 = the coefficient in dollars per pair per foot for the variable that represents the number of
26-gauge aerial copper cable pairs.
28. By substituting into the above equation for 26-gauge aerial copper cable the values from
Table III for the intercept and the coefficient for the cable size variable, and the number of cable pairs in
this example, 100, we obtain the following result for the cost of a 100-pair 26-gauge aerial copper cable:
A4 = 2.385108 + (.008721)(100)
= 2.385108 + .8721
= $3.26 per foot.
29. We next calculate the ratio of 26-gauge copper cable cost to 24-gauge copper cable cost
for a 100-pair cable. The ratio of 26-gauge copper cable cost to 24-gauge copper cable cost for a 100-pair
cable is .96 ($3.26 per foot divided by $3.39 per foot).
30. Finally, we multiply this ratio by the estimate of the 24-gauge copper cable cost for a 100-
pair cable derived from the RUS data, $2.21 per foot, to obtain the cost for a 100-pair 26-gauge copper
cable, $2.12 per foot. We adopt this estimate as the input in the SM for the cost of a 100-pair 26-gauge
aerial copper cable.
III. Huber Methodology
31. We find in this Order that it is reasonable to use the Huber methodology to develop input
values for cable and structure costs. The structure and cable cost inputs used in the SM should reflect
those that are typical for cable and structure for a number of different density and terrain conditions.
Otherwise, the model may substantially overestimate or underestimate the cost of building a network. The
Huber methodology produces estimates of costs that are typical for cable and structure by assigning zero or
less than full weight to cable and structure cost observations that have extremely high or extremely low
values. At the same time, it assigns full or nearly full weight to closely clustered cable and structure cost
observations.
32. Use of the Huber methodology to derive reasonable estimates from RUS data is illustrated
for aerial copper cable cost on the diagram labeled "Scatter Diagram Of 24-Gauge Aerial Copper Cable
Cost And Size With The Huber Regression Line" and on the frequency distribution set forth on Table IV,
labeled "Frequency Distribution Of Huber Weights For 24-Gauge Aerial Copper Cable Cost." The scatter
diagram shows RUS cable cost data points representing combinations of aerial copper cable costs
(measured on the vertical axis in dollars per foot) and cable size (measured on the horizontal axis by
number of pairs). It also shows the regression line that the Huber methodology fits to these data points.
The algebraic expression of this line explains or predicts the effects on aerial copper cable costs of changes
in cable size. The observations to which Huber assigns a weight that is less than .47 are identified with
an "o"; those to which it assigns a weight that is greater than .47 are identified with an "*". The frequency
distribution shows the number of aerial copper cable observations to which the Huber methodology assigns
particular weights.
33. The scatter diagram and the frequency distribution demonstrate that the aerial copper cable
estimates derived by using the Huber methodology with RUS data reflect most of the information contained
in nearly all of the observations. As depicted on the scatter diagram, the majority of the aerial copper cable
observations are clustered closely around the regression line. These are the observations to which Huber
assigns the greatest weight when fitting the regression line to the data. As the frequency distribution shows,
approximately 82 percent of the aerial copper cable observations is assigned a weight of at least .8. This
large majority of closely clustered observations clearly represents typical cable costs. The minority of the
aerial copper cable observations lies a considerable distance from the regression line. These are the
observations to which Huber assigns the least weight when fitting the regression line to the data. As the
frequency distribution shows, approximately 18 percent of the observations is assigned a weight of at less
than .8. This small minority of observations comprises extremely high and extremely low values that do
not represent typical cable costs. The scatter diagram also shows that some of the observations that receive
a relatively small weight lie a substantial distance above the regression line while others that receive such
weight lie a substantial distance below this line. This demonstrates that the Huber methodology excludes or
assigns less than full weight to data outliers without regard to whether these are high or low cost
observations.
IV. Analysis Of Coefficient For Cable Size Variable In The Huber Regression
Equations
34. In this Order, we derive equations to estimate the non-rural LECs' labor and material cost
for cable. We derive these equations by: (1) deriving regression equations by using the Huber
methodology with RUS cable cost data that reflect labor and material costs; and (2) adjusting downward
the coefficient for the variable that represents cable size in these regression equations to reflect the buying
power of large LECs in comparison to RUS companies. The coefficient for the variable that represents
cable size represents the additional cost for an additional pair of cable and therefore represents cable
material costs. The adjustment to this coefficient is based on the difference between the average cable
material prices that Bell Atlantic and the RUS companies pay for different cable sizes. The RUS
companies' average cable material prices are calculated by using unweighted RUS data. Conversely, the
Huber methodology used to estimate the regression equations assigns zero or less than full weight to data
points that have extremely high or extremely low values. Below we demonstrate that the Huber
methodology generally does not have a statistically significant impact on the level of material costs
reflected in the cable cost estimates. That is, in general, there is not a statistically significant difference
between the value of the coefficient for the cable size variable in the regression equations estimated by
using the Huber methodology and the value of this coefficient in the regression equations developed in the
NRRI Study by using OLS. Accordingly, the buying power adjustment for material is based on averages
of RUS companies' cable material prices calculated by using unweighted RUS data.
35. Table V, labeled "Analysis Of Coefficient For Cable Size Variable In The Huber
Regression Equations," displays the values of the coefficient for the cable size variable in the regression
equations estimated from RUS data by using the Huber methodology in this Order and the 95 percent
confidence interval surrounding the value of this coefficient in these equations in the NRRI Study estimated
from these data by using OLS. Except for 24-gauge buried copper cable, the value of the this coefficient
estimated by using the Huber methodology lies inside the 95 percent confidence interval surrounding the
value of this coefficient in these equations in the NRRI Study estimated from these data by using OLS.
That is, except for 24-gauge buried copper cable, the value of the cable size coefficient estimated by using
the Huber methodology lies within an interval that contains with 95 percent certainty the true value of the
OLS cable size coefficient. This statistical evidence supports a finding that the Huber methodology does
not have a statistically significant impact on the level of the material costs reflected in the cable cost
estimates derived by using this methodology. The cable size coefficient obtained by using the Huber
methodology for buried copper cable lies outside the 95 percent confidence interval associated with the
cable size coefficient obtained by using OLS for buried copper cable. This supports a finding that the
Huber methodology does have a statistically significant impact on the level of the material costs reflected in
the buried copper cable cost estimates.