NSP inputs for use in economic benchmarking

NSP Inputs for Use in
Economic Benchmarking
AER Economic Benchmarking Workshop #3
20 March 2013
Denis Lawrence and John Kain
Data requirements
Require data on the price and quantity (and hence value) of
all outputs and inputs and on the quantities of operating
environment variables (noting that output prices may be
shadow prices where the output is not explicitly charged for).
This then allows any of the key economic benchmarking
methods to be implemented.
Given its tops-down nature, it is important to concentrate on
a relatively small number of key outputs and inputs in
economic benchmarking
Desirable to have robust and consistent data
Data backcasting may be possible and facilitate early use
Different types of inputs
Non-durable versus durable inputs
Opex (labour, materials and services) versus capital (lines,
cables and transformers)
Different treatment of these inputs required in economic
benchmarking studies
Obtaining price and quantity of non-durable inputs is relatively
But obtaining accurate annual cost and quantity of capital
inputs is more challenging
Capital service flow is the quantity of capital inputs’
contribution to production each year
Annual user cost is cost of using capital inputs each year
Capital input quantities
An asset’s capital service flow will depend on its physical
depreciation profile
If there is no physical depreciation over asset life then the
profile is said to be ‘one hoss shay’ or ‘lightbulb’ type
If there is more physical depreciation over time then profile
could be of the geometric, straightline or hyperbolic type
Distinction between physical depreciation profile and
regulatory depreciation profile
Capital service flow is unobservable so need proxy
measures – usually assume it to be proportional to the
capital stock
Best aggregate capital service flow proxy will depend on
physical depreciation profile of components, robustness of
data used, whether aggregate profile mirrors components
Annual user cost of capital
Annual user cost of capital consists of:
plus opportunity cost
minus capital gains
Equivalent building block concepts are return of capital
(depreciation less inflation allowance) and the return on
capital (opportunity cost)
Building blocks also implicitly include concept of ex-ante
financial capital maintenance based on RAB and WACC
Preferable to have (approximate) consistency between the
ways annual capital costs are calculated in building blocks
and economic benchmarking
Items excluded
Relevant capital input measures for economic benchmarking
are capital service flow and annual user cost of capital
Capex only measures additions to capital stock each year
It is therefore not appropriate to use capex (nor ‘totex’) as an
input in economic benchmarking studies
Focus of economic benchmarking will be on the efficiency of
operating and maintaining the network
Therefore NSP inputs used for the construction of capital
assets should be excluded to avoid double counting
Some grey areas such as treatment of pole replacement
Input selection criteria
1) input coverage is comprehensive and non–overlapping
2) measures of capital input quantities are to accurately reflect
the quantity of annual capital service flow of assets
employed by the NSP
3) capital user costs are to be based on the service provider’s
regulatory asset base (RAB) and should approximate the
sum of the return of and return on capital components used
in building blocks, and
4) be consistent with the NEL and NER
Opex coverage
Are separate data on labour, materials and services
Given extent of contracting out, generally use aggregate
Diverse composition requires deflating opex by a price
index to derive the quantity of opex inputs indirectly
Sometimes need to remove accounting adjustments
that do not reflect opex input usage that year
Need uniform treatment of asset refurbishment and
allocation of corporate overheads
Need some consistent disaggregation of opex to allow
consistency checking and to provide more information
Do we need to allow for differing system structures and
opex coverage across NSPs?
Opex price index
• Around 60% of opex is typically made up of labour costs
• Remainder covers operational, office and professional
services and consumables
• NSP-specific price indexes generally not available
• Use ABS broad EGWW sectoral and economy-wide price
indexes instead
• Ongoing debate over whether labour price index should
• Use combination of Producer price indexes for rest of
opex – reclassification by the ABS often necessitates
index splicing
• Do opex input price levels vary across NSPs?
• How representative are the high level price indexes of the
price changes NSPs actually face?
Opex specification
Nominal opex / Weighted
average price index
Opex (adjusted to
remove accounting
items not reflecting input
use that year)
Weighted average of
an ABS labour price
index and several ABS
producer price indexes
- Uses readily available data
- Unlikely to be any practical alternatives
- Dependent on changes in ABS sectoral and economy-wide price indexes
accurately reflecting changes in opex prices faced by all NSPs
- Assumes all NSPs face the same levels of opex component prices
- Assumes the same range of opex coverage and functions for all NSPs
Capital input quantities
Capital service flow is not directly observable – have to
use proxy measures instead
No proxy measure will be perfect – all have pros and
Best proxy measure will depend on:
 Physical depreciation profile of constituent assets
 Robustness of the data used to form the proxy – the
more accurate the better and the fewer assumptions
that have to be made, the better
 Whether depreciation profile of the aggregate mirrors
that of the components
Physical depreciation profiles
• Important to distinguish between physical and financial
regulatory depreciation
• Asset physical ‘carrying capacities’ can have one of four
different physical depreciation profiles:
 one hoss shay (no reduction in carrying capacity
throughout the asset’s life)
 hyperbolic (small decline in early years, larger decline
in later years of the asset’s life)
 straight–line (ie equal absolute reduction each year of
the asset’s life), or
 geometric (equal percentage decline over lifetime, ie
bigger absolute falls in early years and smaller
absolute falls in later years)
NSP asset physical depreciation
• NSP assets subject to little physical deterioration over
their lifetime provided they are properly maintained
• Failure rates may be higher towards the end of asset life
leading to Weibull distribution of asset ages
• But service flow relatively constant over the asset’s life
meaning one hoss shay depreciation is closest
• Can approximate service flow under one hoss shay by
using physical measures (eg MVA-kms) or gross capital
stock in constant prices
• Using constant price depreciated asset value proxies
assumes depreciation profile other than one hoss shay
• If the decay of the asset over time is overestimated by the
proxy then efficiency will also be overestimated
Data robustness
• Critical to have accurate input quantity measures
• Best to minimise reliance on indirect measures where
quantities derived by deflating value by a price index
• Capital goods price indexes are particularly problematic
• EGWW sector prices may not be close to prices actually
paid by NSPs
• Also small sample and lumpiness problems
• Starting asset value data may not be consistent and often
adjusted – particularly a problem for long lived assets
• Asset value proxies assume constant asset lengths of life
but regulatory depreciation rates have varied widely through
time for some NSPs, across NSPs and sources
• Asset register data likely to be the most accurate data
Asset aggregation
• US BEA advanced ‘portfolio effect’ example in support of its
use of geometric depreciation and a constant price
depreciated asset value for macro productivity data
• But the US BLS is the main US productivity agency and it
has been at the forefront of using a ‘productive capital stock’
proxy based on the hyperbolic physical depreciation profile
• ABS and SNZ also use the BLS hyperbolic model which is
closer to one hoss shay for structures
• Aggregation issues also not likely an issue for NSPs given
more bunched nature of network roll-outs and few NSPs
• Include 4 asset categories: overhead lines, underground
cables, transformers and other capital
• Use MVA-kms to aggregate up lines and aggregate up
cables, KVA to aggregate up transformers
RAB depreciation as a proxy
• AER Issues Paper raised possibility of using real RAB
depreciation as a proxy for service flow
• This has some similarities to a physical quantity based
one hoss shay proxy
• But it would need to be converted to capital goods
constant prices rather than the CPI constant prices used
in RAB calculations
• It would have the same limitations as other indirect
proxies, namely differences in the basis of initial
valuations, differing assumed asset lives across NSPs,
over time and across sources, and the need to rely on
sectoral capital goods price indexes that may not well
approximate the prices paid by NSPs
Conclusion on capital quantities
• While no capital service flow proxy will be perfect, the use
of physical quantity based proxies appears best on a
number of grounds:
 They reflect the one hoss shay physical depreciation
profile of individual NSP assets
 Asset register data is likely to be the most accurate
data available and will more accurately reflect actual
asset lives
 They are the available proxy that comes closest to
approximating the hyperbolic profile used by leading
statistical agencies for aggregate structures
• The constant price RAB depreciation proxy is likely to be
the second best proxy – this option is subject to further
Capital annual user costs
• Annual user cost can be measured either directly or
• Direct approach applies formula taking in depreciation
rate, opportunity cost rate and rate of capital gains
• This produces an exogenous measure of the ex ante or
shadow annual user cost
• Most exogenous formulas used to date differ from building
block return of and return on capital calculations
• Indirect approach simply takes residual of revenue minus
opex as the annual user cost
• It is an ex post measure of the annual user cost and will
not be FCM-consistent except by chance
• The annual user cost used in economic benchmarking
should be similar to the building blocks return of and
return on capital components
Alternative capital specifications (1)
Physical proxies
Annual user cost
O/H MVA-kms
Return of & on O/H capital O/H AUC/MVA-kms
U/G MVA-kms
Return of & on U/G capital U/G AUC/MVA-kms
Transformers & other KVA Return of & on Trf & other Trf & other AUC/KVA
- Reflects individual component carrying capacities
- Uses the most robust data available and captures actual asset lives
- Approximates productive capital stock used by leading statistical agencies
for aggregate structures
- Ensures consistency with building blocks calculations
- Small amount of extra data required
- May not fully reflect aggregate depreciation profile
Alternative capital specifications (2)
RAB depreciation proxy
Annual user cost
Nominal RAB
depreciation/ABS CGPI
Return of & on capital
AUC/Const price RAB
- Potentially reflects individual component carrying capacities
- Potentially approximates productive capital stock used by leading statistical
agencies for aggregate structures
- Uses existing regulatory data
- Assumes consistency of treatment over time and across NSPs
- May not capture actual asset lives
- Dependent on accuracy and consistency of initial capital base values
- Dependent on ABS CGPI accurately capturing prices paid by NSPs
Alternative capital specifications (3)
Depreciated RAB proxy
Annual user cost
Nominal depreciated
Revenue minus opex
AUC/Const price
depreciated RAB
- Easy to implement
- Uses existing regulatory data
- Unlikely to reflect carrying capacities of assets
- Dependent on accuracy and consistency of initial capital base values
- Assumes consistency of RAB treatment over time and across NSPs
- May not capture actual asset lives
- Dependent on ABS CGPI accurately capturing prices paid by NSPs
- Unlikely to be consistent with financial capital maintenance

similar documents