nbn cba analytical framework (final 16 may … · web viewconsider a government spending proposal...
TRANSCRIPT
![Page 1: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/1.jpg)
Cost-Benefit Analysis and Review of Regulatory Arrangements for the National Broadband Network
Cost-Benefit Analysis of Public Investment in High
Speed Broadband Network Infrastructure:
An Analytical Framework
Dr Alex RobsonDirector, Economic Policy Analysis Program
Department of Accounting, Finance and Economics
Griffith University
May 2014
![Page 2: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/2.jpg)
Contents
Executive Summary 5
Background and Context...........................................................................................................................5The Relevant Baseline or Reference Scenario(s)........................................................................................8
Benefits...................................................................................................................................................13
Costs........................................................................................................................................................16
Non-Quantifiable Benefits and Costs.......................................................................................................18
Overall CBA Strategy................................................................................................................................19
1. Introduction 21
1.1. Background and Context........................................................................................................21
1.2. What is Cost-Benefit Analysis? And Why Conduct One?.......................................................22
1.3. Statement of the Economic Problem.....................................................................................24
1.4. Structure of the Paper...........................................................................................................25
1.5. Key Analytical Issues..............................................................................................................26
2. What Should Project Outcomes be Measured Against?
Baseline/Reference Scenario(s) 29
3. Benefits 35
3.1. The Indirect Demand for Faster Broadband Speeds..............................................................36
3.2. Pure Time Savings..................................................................................................................39
3.3. Reduced Transaction Costs....................................................................................................41
3.4. Travel Time Savings................................................................................................................42
3.5. Productivity Improvements...................................................................................................42
3.6. Estimation of Private Willingness to Pay................................................................................43
3.6.1. Hedonic Regression Using Market Data.................................................................................43
3.6.2. Discrete Choice Estimation Using Survey Data.......................................................................44
3.7. Benefits Over Time................................................................................................................45
3.7.1. Product/Innovation Diffusion Models....................................................................................46
3.8. Network Effects.....................................................................................................................49
3.8.1. Direct Network Externalities..................................................................................................49
3.8.2. Indirect Network Externalities...............................................................................................52
3.9. Fiscal Effects...........................................................................................................................52
3.10. Wider Economic Benefits (WEBs)..........................................................................................56
3.10.1. Economies of Agglomeration.................................................................................................56
3.10.2. Labour Supply and Taxation...................................................................................................58
CBA Analytical Framework Page 2 of 95
![Page 3: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/3.jpg)
3.10.3. Competition/Market Power Effects.......................................................................................59
4. Costs 60
4.1. Direct Costs............................................................................................................................60
4.2. The Social Discount Rate........................................................................................................61
4.3. The Timing of Investment and Option Values........................................................................64
4.3.1. Dynamically Efficient Investment...........................................................................................64
4.3.2. Option Values........................................................................................................................64
4.4. Taxation.................................................................................................................................66
4.4.1. General Considerations..........................................................................................................66
4.4.2. The Marginal Cost of Public Funds.........................................................................................67
4.4.3. Taxation and Uniform National Wholesale Access Pricing.....................................................69
4.5. Non-Quantifiable Benefits and Costs.....................................................................................71
5. Conclusion and Recommendation for Overall CBA Strategy 72
6. References 74
Appendix 1. The Indirect Demand for Faster Broadband Speeds 78
Appendix 2. The Simple Welfare Economics of the Gains from Productivity Improvements
80
Appendix 3. Welfare Analysis in Discrete Choice Models 81
Appendix 4. The Dynamic Efficiency of S-Shaped Product Diffusion Curves 83
Appendix 5. A Simple Model of Network Effects 86
Appendix 6. The Dynamic Efficiency of Investment 90
Appendix 7. The Costs of Taxation 92
CBA Analytical Framework Page 3 of 95
![Page 4: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/4.jpg)
Figures
Figure 1.1: Network Access Versus Take-up and Usage........................................................................27
Figure 3.1: Illustration of the Effect of a Marginal Increase in Broadband Speeds on Welfare and Demand for Other Products................................................................................................37
Figure 3.2: Illustration of the Effect of a Marginal Increase in Broadband Speeds on Welfare and Demand for Other Products: The Case of a New Good.......................................................38
Figure 3.3: Faster Broadband Speeds in a Simple Time-Use Model......................................................40
Figure 3.4: The Impact of Faster Broadband Speeds on Consumer Welfare.........................................41
Figure 3.5: Example of a Logistic Product Diffusion Curve (=0.3, =2)................................................46
Figure 3.6: Change in Take-up Rates Implied by the Logistic Product Diffusion Curve (=0.3, =2)........................................................................................................................47
Figure 3.7: The Effects of Lower Unit Costs on Optimal Spending and Provision of a Publicly Provided Good...................................................................................................................................54
Figure 3.8: The Effects of Lower Unit Costs on Optimal Spending and Provision of a Publicly Provided Good – Taxes are Distortionary...........................................................................................55
Figure 4.1: The Present Value of $1 Received at Different Points in the Future, for Different Discount Rates...................................................................................................................................62
Figure 4.2: Welfare Effects of a Change in Demand When an Existing Good is Taxed..........................67
Figure 4.3: The Marginal Cost of Funds and the Optimal Provision of a Publicly Provided Good..........68
Figure 4.4: Cross-Subsidies as an Implicit Tax-Subsidy Scheme............................................................71
TablesTable 2.1: Two-Way Classification of Policy Alternatives and Scenarios......................................................34
CBA Analytical Framework Page 4 of 95
![Page 5: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/5.jpg)
Executive SummaryThis paper provides an analytical framework for the conduct of the panel’s cost-benefit
analysis of the national broadband network. Its aim is to set out the issues,
considerations and calculations that economic and cost-benefit analysis theory tells us
should be thought about in a cost-benefit analysis of this type. As such, the paper is
primarily theoretical and delves into the full range of items about which a cost-benefit
analysis practitioner will want to think, including, in some cases, options for estimating
or assessing costs and benefits.
The panel has done this to ensure that its cost-benefit analysis has a sound theoretical
basis and that it is transparent about the analytical framework it is using.
In doing this, the panel acknowledges that theory is just part of the process. There is
significant work in the next steps of the cost-benefit analysis applying the analytical
framework to produce an actual cost benefit analysis. This work involves taking the
theory and applying it to the practical world of high speed broadband in Australia and
the tools, techniques and data available to make estimates. Part of this work will
determine practical ways of measuring costs and benefits and the significance or
materiality of various theoretical issues (which may determine how they are treated in
the estimates or even if they are excluded). This work is underway and will be in the
final cost-benefit analysis.
Background and Context
In December 2013 the Minister for Communications announced a panel to conduct an
independent cost-benefit analysis (CBA) of broadband policy and review the regulatory
arrangements for the National Broadband Network.
The panel’s terms of reference and further background information are available on the
Department of Communications’ website.1
1 http://www.communications.gov.au/broadband/national_broadband_network/cost-benefit_analysis_and_review_of_regulation
CBA Analytical Framework Page 5 of 95
![Page 6: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/6.jpg)
The aim of this paper, commissioned by the panel (through the Department of
Communications), is to develop an analytical framework which will assist the panel in its
development of this CBA.
The basic purpose of any CBA is to inform government decision making by systematically
assessing the extent to which the social benefits of pursuing a particular policy or
project exceed the social opportunity costs, given economic and other constraints. CBAs
focus on economic efficiency, which is broadly defined as the dollar value of the gains,
less the dollar value of losses for those individuals and businesses directly and indirectly
affected by the policy change. In a standard CBA, benefits and costs are summed across
all individuals in the community. Questions regarding the overall dollar value of net
gains are separated from issues regarding the equity of end-results and the distribution
of those net gains.
The use of dollar values means that policies can be compared with one another in a
consistent fashion, which facilitates a ranking of policy options against each other. A
CBA also considers non-financial effects or broader social and environmental effects and
the costs and benefits of goods and services not exchanged on markets. A thorough
CBA should, to the greatest practicable extent, take into account and estimate the dollar
value of all of the effects of a particular policy change on the community (economic,
social and environmental) and should include both market and non-market values
where relevant. All non-quantifiable benefits and costs – either because of their
inherent nature or because of a lack of suitable data – must also be fully explored and
the findings presented alongside the valued effects.
The steps involved in a typical CBA are:
1. Collect existing information and data.
2. Establish the policy objectives. What is the economic problem that policy is
directed at addressing?
3. Determine a baseline or reference scenario.
4. Establish policy options.
CBA Analytical Framework Page 6 of 95
![Page 7: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/7.jpg)
5. Identify costs and benefits and estimate their value or qualitatively assess those
that are non-quantifiable.
6. Apply an appropriate discount rate to future cash flows to calculate net present
value, rank policy options (taking into account both the quantitative and
qualitative work) and suggest the most beneficial options.
7. Conduct a sensitivity analysis.
The ultimate purpose of a CBA is to inform policy decisions. By identifying and
measuring all costs and benefits, a CBA can help policy makers make decisions that
maximise net benefits to the community.
One of the most important first steps in a CBA is to identify the economic problem that
policy is meant to address. In the present context, the first step is to ask: what objective
is to be achieved by undertaking public investment in high speed broadband (HSBB)
network infrastructure? The main features of HSBB networks are their coverage (i.e.
the set of consumers that are able to access the network) and the speed available to
those who are connected (this should be taken to be the download and upload speeds
as actually experienced, rather than a hypothetical maximum speed). Given these key
features, the basic policy question underpinning any CBA of public investment in HSBB
network infrastructure could be described as follows:
What are the characteristics – in terms of the network’s key features – of the socially
optimal (i.e. efficient) network rollout, given current community preferences,
technology, costs, fiscal constraints and regulatory settings, and the likely evolution of
these ‘parameters’ over time?
Although the primary purpose of a CBA is not to solve this complex, constrained,
optimisation problem, in practice assessing various policy options may involve either
implicitly or explicitly benchmarking the consequences of various policy options
(including the alternative of “no policy change”) against some socially optimal set of
arrangements (i.e. arrangements that maximise net benefits). Hence, insofar as a CBA is
intended to inform the discussion of broadband policy more generally, identifying the
broad characteristics (in terms of coverage and speed) of the efficient rollout path
would be a valuable exercise.
CBA Analytical Framework Page 7 of 95
![Page 8: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/8.jpg)
The stated purpose of the Government’s independent CBA and review of regulation is to
“analyse the economic and social costs and benefits (including both direct and indirect
effects) arising from the availability of broadband of differing properties via various
technologies, and to make recommendations on the role of Government support and a
number of other longer-term industry matters”. Given the range of new and highly
complex issues that are likely to be involved in a CBA of this type, there are significant
risks in proceeding without a sound analytical framework. This paper develops such a
framework from a theoretical perspective. The purpose is to provide conceptual
guidance for the CBA, identifying and exploring the main sources of the potential
economic impacts of policies that are designed to achieve greater network coverage and
access to faster broadband speeds, as well as providing broad guidance on how to
estimate the magnitude of these impacts. Using such a framework can help to avoid
double counting of benefits and costs in a CBA. It can also, where appropriate, provide
guidance as to the likely economic importance and sign and size of particular impacts.
This paper focuses on three key areas:
The Relevant Baseline or Reference Scenario(s);
Benefits; and
Costs.
The Relevant Baseline or Reference Scenario(s)
As discussed above, the central question in undertaking a CBA of various broadband
policy options is whether the incremental social benefits of an increase in network
coverage and speeds associated with a particular policy are likely to outweigh the
incremental social costs of that particular option. The question is: incremental
compared to what? Using the language of the basic policy question set out above, what
is required is a set of inputs or parameter assumptions that can be fed into the
optimisation problem. In practice this means constructing a reasonable, credible
baseline or reference scenario in which some or all of the social benefits from a
particular policy may be realised over time (but not necessarily at the same time) even
without major policy changes. In this context, it is important to note that, over the last
CBA Analytical Framework Page 8 of 95
![Page 9: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/9.jpg)
decade, neither the availability of broadband nor adoption rates have remained
stagnant in Australia; nor would they be expected to do so in the future.
It is crucial for any CBA to ascertain, as a first step, the extent to which existing
applications, services, products and devices – as well as those likely to be developed,
made available and taken up over the medium term – are able to achieve an acceptable
level of functionality at speeds that are either already available or likely to become
available using existing infrastructure and technology, or using infrastructure that is
likely to be constructed and technological changes that are likely to occur over the
medium term, in the absence of major policy changes.
On the demand side these are essentially technological questions, which require an
examination of peak transmission requirements of various services and combinations of
services (such as high quality video conferencing, Internet Protocol television (IPTV),
streamed video, video and software downloads and general web usage), the extent to
which these transmission requirements are likely to change over the medium term due
to technological innovation (e.g. in video compression) or moves towards “content-rich”
environments, and the extent to which increases in download and upload speeds are
likely to yield economically noticeable changes in existing overall service quality.2
Similarly, on the supply side answering these questions will require an examination and
assessment of existing infrastructure, including the download and upload speeds that
can be experienced by customers using copper lines connected to ADSL2+ enabled
exchanges, and the coverage and capability of existing hybrid fibre coaxial (HFC)
networks, fixed wireless networks, and 3G and 4G mobile networks. It will also require
an assessment of likely future supply side developments (e.g. developments in wireless
and satellite technology, increased use of VDSL) that may occur in the absence of major
policy changes.3
2 Relevant here is Robert Kenny and Tom Broughton’s Domestic Demand for Bandwidth, an approach to forecasting requirements for period 2013 to 2023, see <http://www.commcham.com/storage/publications/BSG-Domestic-demand-for-bandwidth.pdf>.
3 Much of the work for such an assessment has recently been carried out by the Australian Government; see the Department of Communications’ 2013 report, Broadband Availability and Quality.
CBA Analytical Framework Page 9 of 95
![Page 10: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/10.jpg)
It is impossible to precisely determine the dynamically efficient path4 of high speed
broadband infrastructure investment in the Australian context (taking into account
demand and supply side uncertainty, irreversibility of investment, and the potentially
large number of option values that are embedded in investment decisions). However, it
would be highly desirable for a CBA to determine the most important economic features
of such a dynamically efficient path – including the economic trade-offs involved, and
the incentives on the demand and supply side of the relevant markets. Given the results
of such an analysis, it would also be desirable to examine and determine the extent to
which market outcomes – supported by appropriate regulation – will deliver the
features that characterise the dynamically efficient investment path. Such features
include the timing of aggregate investment, the extent of population coverage over
time, the extent of competition, pricing arrangements, capacity choice and congestion,
download and upload speeds, as well as other characteristics.
Addressing these issues necessitates the construction of a credible reference or baseline
scenario, which in turn involves careful consideration. In the language of the basic
policy question set out earlier, varying the reference scenario involves varying the
parameters of the optimisation problem. Given that there is a significant amount of
uncertainty regarding the precise evolution of technology and the development of
consumer applications and devices over the medium term, it would be preferable to use
a range of parameter values. In other words, the CBA should develop a suite of different
reference scenarios, each of which can ultimately be used as an input into sensitivity
analysis.
4 The dynamically efficient path is the one that maximises the expected net present value of benefits minus costs.
CBA Analytical Framework Page 10 of 95
![Page 11: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/11.jpg)
Factors that should be addressed in the construction of these scenarios include:
The current state of Australia’s broadband market, and the economic drivers of
the evolution of the market to date.
The evolution of private sector telecommunications infrastructure investment in
Australia, and the incentives that competition and market forces have provided
for such investment.
The role that regulation has played in influencing the evolution of market
outcomes, including its influence on incentives for investment, consumer prices,
capacity, network coverage and speed.
Given supply and demand side uncertainties, this paper is deliberately non-prescriptive
regarding the specific choice of baseline scenarios. However, at the minimum, the CBA
could examine three broad classes of parameter assumptions or reference scenarios:
Pessimistic scenarios in which private investment in – and take-up of – faster
speeds are inefficiently delayed to a significant extent, or where the breadth of
coverage is significantly inefficiently narrow. In constructing this scenario it is
crucial to carefully identify the reasons that may drive such an outcome – that is,
the market or regulatory failures that are likely lead to inefficient delay.
Intermediate reference scenarios, in which market outcomes are, in theory,
likely to be dynamically efficient. In developing these scenarios it would be
important to explore the reasons why market failures are likely to be minimal,
and the policy and regulatory settings that are likely to lead to such an outcome.
Optimistic or over investment scenarios, in which private investment in network
coverage and speed – as well as take-up rates – develops more rapidly than the
dynamically efficient path, or where network coverage is extended to areas
where it is inefficient to do so (i.e. where costs exceed benefits). Again, in
constructing such as scenario it will be important to carefully identify the market
or regulatory failures that may lead to inefficient overinvestment.
CBA Analytical Framework Page 11 of 95
![Page 12: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/12.jpg)
Different features from a subset of these scenarios could be chosen and combined with
each other to form hybrid reference scenarios. The extent to which such scenario-
building is employed is ultimately a matter of judgement for the CBA analyst.
The quantitative results of the implicit optimisation exercise undertaken in a CBA will be
a function of the assumed parameters (i.e. the reference scenario). This is a strength of
a CBA rather than a weakness, as results can be mapped clearly and transparently back
to the assumptions made. On the other hand, it is possible that the qualitative results of
a CBA – for example, whether a particular project yields positive incremental social net
benefits – may not vary greatly with the assumed reference scenario. Importantly, if
there is a great deal of uncertainty about modelling parameters or assumptions and the
qualitative features of the CBA vary widely with these assumptions, this suggests there
is a great deal of risk associated with the project. This is likely to be useful information
for decision makers.
Once these reference or baseline scenarios have been constructed, they can then be
compared against the likely outcomes under various policy scenarios. In comparing
reference scenarios against policy scenarios, it is important that the CBA examines and
realistically assesses policy failures and implementation failures. That is, the extent to
which policies address the sources of market or regulatory failure that are assumed to
drive the evolution of outcomes in the reference scenario, and the extent to which
stated policies are likely to be implemented in such a way as to achieve their stated
goals. Policy changes that are subject to these kinds of failures are less likely to yield
incremental net benefits relative to the reference scenario, and under some
circumstances may actually yield negative incremental benefits even assuming a
relatively pessimistic baseline scenario.
CBA Analytical Framework Page 12 of 95
![Page 13: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/13.jpg)
Benefits
Sources of Benefits
Conceptually, faster broadband speeds may provide a range of benefits to consumers
and producers. The paper identifies these as:
Increase in Benefits from Using Existing Products, and the Benefits from the
Ability to Use New Products.
Pure Time Savings.
Reduced Transaction Costs.
Travel Time Savings.
Productivity Improvements.
Whilst conceptually these are separate effects, since the benefits accrue directly to
individuals or firms, in many (but not all) cases they will be captured in estimates of
consumers’ and producers’ direct willingness to pay for faster broadband speeds.
Because there may be a range of benefits, one of the risks in undertaking any CBA is that
it may either fail to rigorously specify the conceptual foundation of benefits; double
count benefits; or count costs as benefits. Another risk is adopting narrow definitions of
particular effects because they make valuation easier.
Although levels of willingness to pay for faster speeds are important, a key consideration
for ascertaining the characteristics of the dynamically efficient network investment path
is the likely overall shape of willingness-to-pay curves (as a function of download speed),
which provide an indication of how willingness to pay may change over time. Even if the
quantum of future benefits is difficult to precisely determine, it would be useful to
obtain qualitative information regarding the likely evolution of future incremental
benefits.
There are a number of good economic reasons to expect that willingness-to-pay curves
for faster broadband speeds are concave – that is, other things being equal, consumer
marginal willingness to pay for faster speeds is likely to be positive, but decreasing.
CBA Analytical Framework Page 13 of 95
![Page 14: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/14.jpg)
First, existing economic studies suggest that there are diminishing marginal benefits to
faster speeds.5 Second, since overall service quality will depend on the data
requirements for each broadband activity, and whether they are undertaken one at a
time or together, moving from 25 Mbit/s to 50Mbit/s for example may make little
difference to overall service quality if the customer’s demand for existing applications,
services, products and devices can be satisfied by a peak requirement of less than 25
Mbit/s. On other hand, applications that have yet to be invented or widely adopted
could require faster individual speeds to achieve a minimal level of functionality, and so
a move from 25 Mbit/s to 50Mbit/s would mean a great deal for service quality if the
introduction of these products into the broadband activity mix required a peak
requirement greater than 25 Mbit/s. For these kinds of combinations of services, it will
be important for a CBA to determine the range of points at which further increases in
speed will be unlikely to significantly improve service quality, and to subject this to
sensitivity analysis. Third, once a domestic broadband network is operating at
reasonably high speeds, the links between the first point of domestic traffic aggregation
and the global Internet may act as one of the main constraints on further improvements
in speed. Fourth, as discussed further in the paper, for some kinds of benefits, basic
economic principles suggest that there will be diminishing marginal returns.
On the demand side, it is important to note that the demand for faster broadband
speeds is a derived demand. The goods that consumers value directly are services,
applications, products and devices. Access to faster broadband speeds is one of a
number of product characteristics or attributes that can enhance the value that
consumers attach to these end-use products. Faster speeds can be a substitute or
complement for other product attributes. Willingness to pay for faster speeds stems
from the ability of those faster speeds to enhance the marginal value from consuming
existing goods, or to increase demand for new applications, products, services and
devices.
5 For example, Dutz et al (2009) estimate the net consumer benefits from home broadband in the United States are of the order of US$32 billion per year, whilst the benefits of an increase in broadband speed from 100 times the typical historical speed of dial-up service to 1,000 times dial-up, are of the order of US$6 billion per year for existing home broadband users. They also document an “experience effect” – consumers who currently enjoy fast speeds value faster speeds over 40 percent more than households who currently have dial-up connections.
CBA Analytical Framework Page 14 of 95
![Page 15: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/15.jpg)
This has a number of important implications that are explored in the paper. For
example, one simple implication is that even if faster speeds increase the demand for
yet-to-be-invented applications, products, services and devices, any increase in marginal
value must be compared against the prices at which those innovations are actually
made available, which in turn is related to their production cost.
Estimation of Private Benefits
The paper briefly discusses two major methods that have been used to estimate private
willingness to pay in other contexts, using market based data or consumer survey
responses. These are standard approaches in the empirical literature and there are
many examples which can be used as a guide for a CBA of faster broadband speeds.
Benefits Over Time
The intertemporal pattern of demand for applications, products and services will drive
the evolution and growth of aggregate benefits of faster speeds over time. Since any
public infrastructure project will take many years to construct and will likely have very
high upfront costs, it becomes crucial to specify exactly when future benefits are likely
to accrue in both the project and baseline scenarios. Therefore, credible projections of
likely take-up rates are required, and these should be subjected to detailed sensitivity
analysis. The paper briefly discusses some standard approaches that are used in the
product diffusion literature to forecast take-up/adoption rates. The paper notes that if
these approaches are used, a CBA should ideally ensure that take-up projections are
consistent with assumptions regarding the evolution of consumer benefits and other
market variables over time.
Other Possible Sources of Social Benefits
A distinguishing feature of a CBA is that it considers external costs and benefits, as well
as private costs and benefits. This paper discusses the possible sources of benefits that
may not be reflected in measures of private willingness to pay. These include:
Network Effects: These are positive demand side externalities that reflect the
notion that consumers may value accessing a network if it has more users. A key
CBA Analytical Framework Page 15 of 95
![Page 16: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/16.jpg)
question here is the extent to which network effects are impacted by faster
connection speeds.
Fiscal Effects: These are effects associated with changes in tax revenue and
spending, as well as the quality of provision of government services. The analysis
in this paper suggests that projected reductions in government spending are
likely a poor proxy for measuring the benefits that can be attributed to public
sector activity. A preferable approach would be to directly estimate the changes
in social benefits and costs as a result of public sector costs falling, or the
benefits from changes in the quality or characteristics of publicly provided goods
and services. In assessing the net benefits of new government services that may
be introduced, or changes in quality, it is important to net out the additional
costs that may be incurred in securing these benefits – particularly in e-health
and online education.
Wider Economic Benefits: These are positive external benefits associated with
agglomeration, changes in market structure, and labour supply. This paper
considers these effects, and concludes that some of these effects could be
ambiguous in the case of HSBB (that is, they involve costs and well as benefits).
These effects should therefore be treated with particular care.
Costs
For the purposes of a CBA of HSBB network infrastructure projects, estimating the
quantifiable social costs of any project will require three important components: direct
costs; the cost of capital (the social discount rate); and the opportunity costs of taxation
needed to fund the project. Pricing arrangements (such as uniform pricing, which may
involve a scheme of implicit taxes and subsidies) may also impose social costs that
should ideally be captured in a CBA.
Regarding direct costs, ideally a “bottom-up” cost modelling approach should be
employed for this purpose, using engineering and construction data and assumptions
regarding exchange areas, services, lines, contention ratios and so on. This task is likely
to require highly specialised technical knowledge regarding the physical engineering and
construction requirements of different project scenarios.
CBA Analytical Framework Page 16 of 95
![Page 17: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/17.jpg)
Different investment paths result in different present values of costs. For the purposes
of a CBA of HSBB networks, the investment paths of some infrastructure projects will
more closely resemble the dynamically efficient path than others. The costs of
inefficiently timed investment should ideally be accounted for in a CBA. Similar
considerations apply to option values. Typically, investment in HSBB networks is largely
irreversible, with uncertain costs and benefits. This means that delaying investment can
be privately and socially valuable. That is, construction of a HSBB network involves
purchasing and exercising a series of embedded real options. These options have value,
and this value should ideally be accounted for in any CBA.
Since a national HSBB network will take many years to construct and will have high
upfront costs, the choice of discount rate – which is used to compare streams of costs
and benefits over time – is crucial. Social risk is a key issue – there is considerable
uncertainty around both benefits and costs, and along multiple dimensions
(geographical/spatial, intertemporal etc.). This paper advocates the use of a market
based approach to choosing a discount rate, whilst acknowledging that there is little
consensus among economists as to what the “correct” choice of the discount rate
should be. A range of rates should therefore be used and subjected to sensitivity
analysis. The discount rate can be estimated using standard techniques, by calculating a
project’s weighted average cost of capital (WACC) using, for example, the capital asset
pricing model (CAPM) to determine the cost of equity.
In the present case, given uncertainty around take-up and revenue growth, government
investment is likely to have a high degree of systematic or non-diversifiable risk. It is
therefore appropriate for a significant risk premium (determined by an appropriately
chosen asset beta) to be added to the risk free rate.
On the costs of taxation, a basic principle of welfare economics is that the direct
resource costs of government spending should be grossed up by the marginal cost of
public funds (MCF) – that is, the costs of taxation – to determine the full social cost of
the project, which is the reduction in private surplus that occurs when the project goes
ahead. The Department of Finance’s Handbook of Cost-Benefit Analysis6 states that:
6 http://www.finance.gov.au/publications/finance-circulars/2006/docs/Handbook_of_CB_analysis.pdf
CBA Analytical Framework Page 17 of 95
![Page 18: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/18.jpg)
“The excess burden of taxation means the supply cost of public investment or services
are greater than the actual amount of funds used. Consequently, it is appropriate to
make an upward adjustment to financial costs in cost benefit analysis to ensure the
calculated net present value is unbiased. This is, however, only where there is a
significant net cost to the budget. It excludes cases where costs are fully recovered (such
as where there is a user charge) or the resources are already committed (which is
effectively so for cost-effectiveness analysis and lease-purchase analysis).”
Ignoring the MCF can have deleterious welfare consequences, and should be included in
any CBA where relevant.
A final important aspect of taxation that should be taken into account in a CBA is the
implicit tax and subsidy arrangement that uniform national wholesale access pricing
may entail. Imposing a uniform price means that downstream providers (and final retail
consumers) in low cost areas will effectively pay a tax, the revenue from which is used to
subsidise consumers in high cost areas. The social costs of such an arrangement should
be estimated where relevant.
Non-Quantifiable Benefits and Costs
As discussed above, ideally a CBA should place dollar values on all sources of costs and
benefits. However, there may be benefits and costs which are difficult to estimate
accurately, or which are simply non-quantifiable. For example, consumers and
producers may (due to, for example, a lack of information) systematically underestimate
(or overestimate) the likely impact that faster broadband speeds may have on future
innovations and the value of new applications, services and products (including, for
example, e-health and e-education applications). If these systematic biases are present
across a sufficient number of consumers and producers, then estimates of private
willingness to pay, which rely on information provided by these groups, may not
correctly and/or fully reflect relevant benefits. Other benefits may be difficult or
impossible to quantify, such as the benefit from being able to access the “knowledge
commons” of the global internet or, more relevant for policy, the effects of faster
speeds on these benefits. Similarly, there may be significant non-quantifiable social
costs associated with broader network coverage and faster speeds, including issues
CBA Analytical Framework Page 18 of 95
![Page 19: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/19.jpg)
related to privacy, data security, national defence and security, intellectual property
protection, net neutrality, cyber-crime and freedom of speech. These are clearly
important policy issues in their own right, and the extent to which they are affected by
various investment options deserves to be discussed and explored qualitatively in a CBA.
Overall CBA Strategy
Undertaking a comprehensive CBA is a challenging exercise. In practice a range of
theoretical assumptions and empirical compromises will need to be made. The key to
undertaking a robust CBA is to test how sensitive these are to changes in these
assumptions and compromises, to clearly demonstrate to decision makers how they
affect the CBA’s final results.
Given that the ultimate purpose of a CBA is to inform policy decisions, the most
straightforward way to proceed would be to begin with a (relatively) simple partial
equilibrium analysis of the effects of various policy options on network coverage, speed,
and incremental net benefits. Under this approach, the analysis would begin with the
simplest adequate model: social benefits would be assumed to be fully reflected in
private willingness to pay, and estimates of consumer valuation could be compared with
direct project costs. Factors such as externalities and other effects (which are discussed
further below) could be temporarily put to one side.
Once a partial equilibrium analysis is completed, some of the more conceptually difficult
issues explored in this paper could be dealt with. In the first instance, these factors
could be examined qualitatively, by exploring the extent to which a more complicated,
general equilibrium analysis would increase the economic value of various policy
options, and whether such effects are likely to vary greatly across options. For example,
it may be the case that the presence of demand side network externalities would
increase the net benefits of all policies by a similar magnitude, leaving unchanged the
overall ranking of policies that emerged from a simple partial equilibrium analysis.
Alternatively, if a simple partial equilibrium analysis suggests that the net benefits of
certain policies are likely to be negative, a qualitative analysis could be used to show
how positive the benefits from other sources would need to be to reverse the results of
the simpler analysis.
CBA Analytical Framework Page 19 of 95
![Page 20: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/20.jpg)
CBA Analytical Framework Page 20 of 95
![Page 21: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/21.jpg)
1. Introduction
1.1. Background and Context
In April 2009 the Australian Government announced the establishment of a new
company (the majority government-owned NBN Co) to build and operate the National
Broadband Network (NBN). The policy objective was to invest up to $43 billion over 8
years to connect 90 percent of Australian premises to broadband services with speeds
up to 100 megabits per second and to connect all other premises in Australia with next
generation wireless and satellite technologies that will deliver broadband speeds of 12
megabits per second.7
In December 2013 the Minister for Communications announced a panel to conduct an
independent cost benefit analysis (CBA) of broadband policy and review the regulatory
arrangements for the National Broadband Network. This followed a change of
government in September 2013.
The Terms of Reference for the panel8 state that:
“The purpose of the independent cost-benefit analysis and review of regulation is to
analyse the economic and social costs and benefits (including both direct and indirect
effects) arising from the availability of broadband of differing properties via various
technologies, and to make recommendations on the role of Government support and a
number of other longer-term industry matters.”
In particular, the review is to examine, among other things, the following questions:
“1. What is the direct and indirect value, in economic and social terms, of increased
broadband speeds, and to what extent should broadband be supported by the
government?
7 Prime Minister’s Media Release, 7 April 2009. <http://pmtranscripts.dpmc.gov.au/browse.php?did=16491 >.
8 <http://www.communications.gov.au/__data/assets/pdf_file/0019/207046/NBN_Panel_of_Experts_-_Terms_of_Reference.pdf >.
CBA Analytical Framework Page 21 of 95
![Page 22: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/22.jpg)
a. This should consider the economic and social benefits of bringing forward
improvements in broadband speed and the respective benefits of alternative /
potential technologies.
b. It should also consider the extent to which market pricing mechanisms
can capture the value of benefits (including benefits to Australian governments).”
Given the range of new and highly complex issues likely to be involved in such a CBA,
the panel decided to first develop a sound analytical framework. The aim of this paper,
commissioned by the panel (through the Department of Communications), is to provide
that framework. It provides conceptual guidance for the CBA, identifying and exploring
the main sources of the potential economic impacts of policies that are designed to
achieve greater network coverage and access to faster broadband speeds, as well as
providing broad guidance on how to estimate these impacts and avoid double counting
of benefits and costs.
1.2. What is Cost-Benefit Analysis? And Why Conduct One?
The basic purpose of any CBA is to inform policy by systematically assessing the extent
to which the social benefits of pursuing a particular government policy or project exceed
the social opportunity costs, given economic and other constraints. A CBA focuses on
the dynamic efficiency effects of a policy: the present value, measured in dollars, of the
gains that accrue over time, less the present value of losses for those directly and
indirectly affected by the policy change. The use of dollar values is a convenient
analytical tool and measurement device that has a number of distinct advantages
relative to other approaches [such as multi-criteria analysis (MCA), for example].9 In
particular, the use of dollar values means that policies can be compared with one
another in a consistent fashion, which facilitates a ranking of policy options against one
another.
The use of dollar values does not mean that a CBA ignores non-financial or broader
social and environmental effects; nor does a CBA ignore the costs and benefits of goods
and services that are not exchanged on markets. It should, to the greatest practicable
extent, take into account and estimate the dollar value of all of the effects of a particular
9 The advantages of CBA relative to MCA are examined by Dobes and Bennett (2009) and Ergas (2009).
CBA Analytical Framework Page 22 of 95
![Page 23: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/23.jpg)
policy change on the community (economic, social and environmental), and should
include both market and non-market values where the latter are relevant. All non-
quantifiable benefits and costs – either because of their inherent nature or because of a
lack of suitable data – must also be fully explored and the findings presented alongside
the valued effects.
In a standard CBA, benefits and costs are summed across all individuals in the
community. Questions regarding the overall dollar value of net gains are separated
from issues regarding end-results equity and the distribution of those net gains. Ideally
the latter issues should be dealt with in a separate policy analysis, so that assumptions
about distributional effects – and the nature of any trade-offs between equity and
efficiency – are explicit and are made transparent to both policy makers and the wider
community.
Generally speaking, the steps involved in a typical CBA are:
1. Collect existing information and data.
2. Establish the policy objectives: What is the economic problem that policy is being
asked to address?
3. Determine a baseline or reference scenario.
4. Establish policy options.
5. Identify costs and benefits and estimate their value or qualitatively assess those
that are non-quantifiable.
6. Apply an appropriate discount rate to future cash flows to calculate net present
value, rank policy options (taking into account both the quantitative and
qualitative work) and suggest the most beneficial options.
7. Conduct sensitivity analysis.
By identifying and measuring all costs and benefits, a CBA can help decision makers
maximise net benefits to society. Robust assessments of the costs and benefits of
government projects can improve the probability of sound policy decisions (and reduce
the probability of costly policy errors), and increase the chances that taxpayer-funded
projects that go ahead will produce net benefits for the community.
CBA Analytical Framework Page 23 of 95
![Page 24: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/24.jpg)
In the context of government infrastructure projects, there are at least two possible
errors that can be made:
Type I error: an economically beneficial project is rejected because it is
incorrectly assessed as creating a net welfare loss.
Type II error: an economically wasteful project is approved because it is
incorrectly assessed as creating a net welfare gain.
Rigorous CBA can reduce the probability of both types of errors.10
1.3. Statement of the Economic Problem
As discussed above, one the most important steps in a CBA is to identify the economic
problem to be addressed and the objective that policy is attempting to achieve. The key
features of HSBB networks are coverage (i.e. the set of consumers able to access the
network) and the speed available to those who are connected (this should be taken to
be speeds as actually experienced, rather than a hypothetical maximum). Given these
key features, the basic policy question underpinning any CBA of public investment in
HSBB network infrastructure is: What are the characteristics – in terms of key network
features – of the dynamically efficient network rollout, given current community
preferences, technology, costs, fiscal constraints and regulatory settings, and the likely
evolution of these parameters over time?
Whilst the primary purpose of a CBA is not to solve this complex, constrained
optimisation problem, in practical terms assessing various policy options may involve
(either implicitly or explicitly) benchmarking the consequences of various policy options
(including the alternative of “no policy change”) against some socially optimal set of
arrangements (i.e. which maximise net benefits). Hence, insofar as a CBA is intended to
inform the discussion of broadband policy more generally, identifying the broad
characteristics (in terms of coverage and speed) of the efficient rollout path would be a
valuable exercise. Once the characteristics of an efficient path are identified, a CBA of
public investment in HSBB network infrastructure would also need to identify and take
into account the potential effects of possible market failures on coverage and speed, as
10 The analytical framework developed by Little and Mirrlees (1994) can be used to quantify the gains from cost-benefit analysis.
CBA Analytical Framework Page 24 of 95
![Page 25: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/25.jpg)
well as realistically assessing the behaviour of relevant government and regulatory
agencies and the likely effect (both positive and negative) they may have on the
evolution of the network’s key features.
1.4. Structure of the Paper
Although many of the basic principles of CBA are set out in detail elsewhere,11 a number
of specific issues are likely to arise in a CBA of investment in HSBB network
infrastructure. This paper therefore aims to provide “broadband-specific” conceptual
guidance for the panel’s CBA by exploring the main sources of the potential economic
impacts of access to faster broadband speeds, as well as providing guidance on how to
estimate these impacts. Where appropriate, commentary is also provided on the likely
economic importance and empirical size of particular impacts.
The paper examines three broad issues relating to public investment in high speed
broadband network infrastructure:
The relevant baseline or “reference scenario”.
Benefits.
Costs.
The paper is structured as follows. Section 2 outlines the considerations that should be
taken into account in designing the reference scenario against which policy changes can
be compared in a CBA. Section 3, which constitutes the bulk of the paper, examines the
conceptual foundations of the possible main benefits of faster broadband speeds.
Section 4 sets out some relevant considerations for direct and indirect costs that may be
associated with infrastructure investment in this area. Section 5 concludes. Advanced
analytical material is provided in the Appendices at the end of the paper.
1.5. Key Analytical Issues
Given the dynamic nature of technology on the supply side, and the rapid development
of applications and the associated benefits on the demand side, uncertainty will be a key
11 See, for example, the Department of Finance’s Handbook of Cost-Benefit Analysis. Other useful references include Boardman et al (2010), Pearce (1983) and Pearce and Nash (1981).
CBA Analytical Framework Page 25 of 95
![Page 26: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/26.jpg)
analytical and practical issue in any CBA. To ensure robustness of the CBA, sensitivity
analysis of benefits and costs will be vital. As a general rule, if the results of a CBA vary
wildly with minor changes in assumptions, this suggests that there is a great deal of
uncertainty about the net benefits of the project. This would be valuable information to
decision makers.
Extensive recent literature analyses the economics of the Internet in general, and the
benefits of fast broadband speeds in particular. This literature is not be surveyed in
detail here.12 What clearly emerges from the literature, however, is that broadband
network infrastructure possesses a number of specific economic features.
On the demand side, it is important to note that consumers do not directly value access
to faster speeds. Rather, the goods that are valued directly are services, applications
products and devices. Access to faster broadband speeds is one of many characteristics
that can enhance the value that consumers attach to these end-use products.
Consumers may be able to access a network, but if the access price exceeds their
willingness to pay, they will choose not to take up the services offered. However, if
willingness to pay is uncertain (i.e. subject to unanticipated random shocks –- for
example which may occur when new products come on the market), then consumers
may value the mere fact that they have an option to take up services at some point in
the future. In other words, as shown in Figure 1.1, it may be important to differentiate
between: the option value to consumers of being able to access a HSBB network
(without actually taking up the service); the take-up decision itself; and the demand for
usage of the network’s services (volume and speed of service). Demand for access to
the network (the extensive margin) depends on an individual’s consumer surplus at
existing prices for other goods, compared to the price of access. On the other hand, the
demand for usage depends on the marginal willingness to pay for higher volumes and
faster speeds (given that the decision has already been made to take-up the service),
compared to marginal usage prices. Policy interventions can affect each decision.
12 For a recent Australian example, see Deloitte Access Economics (2013), who estimate that in 2020 the average Australian household will receive $3800 in gross benefits from high speed broadband. However, the authors of this paper carefully note that because their analysis does not include the costs of constructing and operating broadband networks, it does not constitute a cost-benefit analysis; nor was their analysis a comparison of different broadband policies or modes of delivery. Hayes (2011) provides a selective overview of other recent work in this area.
CBA Analytical Framework Page 26 of 95
![Page 27: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/27.jpg)
Figure 1.1: Network Access Versus Take-up and Usage
Network Access?
No
TakeupService?
Usage Decision (Volume & Speed)
Yes
NoYes
On the supply side, faster speeds ultimately occur as the result or “output” of a
“production function” or supply chain that involves a number of intermediate stages.
Retail service providers and application/content service providers supply services
directly to end users. In contrast, any HSBB network will lie in the middle of an end-to-
end service, and provide retailers with wholesale access to the network. Retail access to
such a network is necessary, but not sufficient for final consumers to obtain higher
speeds. Indeed, as NBN Co has stated:
“The speeds actually experienced by consumers over fibre, wireless, satellite or copper
will depend on a number of factors including the retail broadband plan they choose, their
equipment and their in premises connection.”13
Access to faster speeds may be complementary to other inputs (such as encoding or
latency), or it may be a substitute. If network access is complementary to other inputs,
then providing access to the network will mean that those other inputs increase,
boosting the ultimate effect of faster network speeds. On the other hand, if access to
the services provided by a network is a substitute for other inputs (such as caching),
then there will be an offsetting reduction in those inputs, and this will mitigate the effect
of being able to access faster speeds on the network.
13 <http://www.nbnco.com.au/nbn-for-home/how-it-works/network-features.html >.
CBA Analytical Framework Page 27 of 95
![Page 28: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/28.jpg)
An important implication is that due to responses on the production or supply side,
provision of access to faster speeds via the customer access network may not translate
one-for-one into faster final speeds for consumers. In addition, it may not be
reasonable to make the usual ‘all other things being equal ‘assumption on the supply
side, i.e. it may not be reasonable to assume that building and providing access to a
network will not influence the incentives and behaviour of retail service providers.
Finally, there are a number of good economic reasons to expect that faster network
speeds may yield positive but declining gross marginal benefits. First existing economic
studies suggest that there are, in fact, diminishing marginal benefits to faster speeds.14
Second, since overall service quality will depend on the data requirements for each
broadband activity and whether they are undertaken one at a time or all together,
moving from 25 Mbit/s to 50Mbit/s for example, may make little difference to overall
service quality if the customer’s demand for existing applications, services, products and
devices can be satisfied by a peak requirement of less than 25 Mbit/s. On other hand,
applications that have yet to be invented or widely adopted could require faster
individual speeds to achieve a minimal level of functionality, and so a move from 25
Mbit/s to 50Mbit/s would mean a great deal for service quality if the introduction of
these products into the broadband activity mix required a peak requirement greater
than 25 Mbit/s. For these kinds of combinations of services, it will be important for a
CBA to determine the range of points at which further increases in speed will be unlikely
to significantly improve service quality, and to subject this to sensitivity analysis. Third,
once a domestic broadband network is operating at reasonably high speeds, the links
between the first point of domestic traffic aggregation and the global Internet may act
as one of the main constraints on further improvements in speed. Fourth, as discussed
further in the paper, for some kinds of benefits, basic economic principles suggest that
there will be diminishing marginal returns.
14 For example, Dutz et al (2009) estimate the net consumer benefits from home broadband in the US are on the order of $32 billion per year, whilst the benefits of an increase in broadband speed from 100 times the typical historical speed of dial-up service to 1,000 times dial-up are on the order of $6 billion per year for existing home broadband users. They also document an “experience effect” – consumers who currently enjoy fast speeds value faster speeds over 40 percent more than households who currently have dial-up connections.
CBA Analytical Framework Page 28 of 95
![Page 29: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/29.jpg)
2. What Should Project Outcomes be Measured Against? Baseline/Reference Scenario(s)
Cost-benefit analysis assesses the incremental net benefits of a policy, relative to a well-
specified baseline or reference scenario. Hence, as a general proposition, a reference
scenario in a CBA should be transparent and credible. There is little formal literature on
the optimal selection of reference scenarios in CBAs. Both the project scenario and the
reference scenario will involve the construction of forecasts of what is likely to happen a
number of years into the future, and are therefore subject to a great deal of uncertainty.
In public infrastructure projects where there is a relatively low likelihood of public
investment substituting for (or “crowding out”) private investment, the reference
scenario is often relatively obvious and straightforward, and is likely to be based heavily
on the status quo. However, in the case of broadband, the status quo is not a
reasonable baseline assumption. In particular, a baseline scenario in which all
consumers remain at current speeds is unlikely to be a plausible or sensible base case.
Data compiled by the Australian Bureau of Statistics (ABS) and the Organisation for
Economic Co-Operation and Development (OECD) show that access to broadband and
speeds have increased gradually over time in Australia, and suggest that an upward
trend would likely continue in any reference scenario.15
Although this data shows that the Australian situation has been changing over the last
decade, this should not be interpreted as indicating that observed outcomes are
necessarily socially optimal. Rather, the data simply illustrates the changing nature of
the market and highlights the additional complications that this may entail for a CBA,
particularly for specifying a reference scenario.
Given the dynamic and uncertain nature of the future evolution of broadband
characteristics and take-up rates, one useful way to think about the reference and policy
scenarios is that higher broadband speeds for a large proportion of the population may
be achieved in both scenarios but that outcomes are accelerated and benefits may be
realised sooner in some scenarios. Viewed in this way, the task of a CBA potentially
15 See, for example Australian Bureau of Statistics, Cat. No. 8146.0, Household Use of Information Technology, Australia, 2012-13; ABS Cat. 8153.0. Internet Activity, Australia, June 2013 and <http://www.oecd.org/sti/broadband/oecdbroadbandportal.htm > particularly Tables 1g(1) and 1g(2).
CBA Analytical Framework Page 29 of 95
![Page 30: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/30.jpg)
becomes less complicated: it is simply to compare similar community benefits that may
be realised over different timeframes, with different costs imposed on the community.
Two major factors are likely to complicate this approach. First, there is the question of
what should be assumed about the starting point for policy. Current policy
arrangements may be irreversible and therefore involve sunk costs – which should not
be counted in any CBA – whilst other policy settings may be reversed or renegotiated in
the short or medium term. Second, there is the question of what regulatory
arrangements will apply in the policy and reference scenarios. Regulation affects
incentives on both the supply side (by altering the timing and extent of investment) as
well as the demand side (by altering prices and take-up rates). Careful consideration will
need to be given to both factors, and ideally a range of scenarios should be examined
and subjected to sensitivity analysis.
Given recent broadband developments, it is crucial for any CBA to ascertain, as a first
step, the extent to which household and business broadband activities involving existing
applications, services, products and devices – as well as those that are likely to be
developed and made available and taken up over the medium term – are able to take
place with an acceptable level of functionality at speeds that are either already
available, or that are likely to become available, using existing infrastructure and
technology, or using infrastructure that is likely to be constructed and technological
changes that are likely to occur over the medium term in the absence of major policy
changes.
On the demand side these are essentially technological questions, which require an
examination of peak transmission requirements of various services and combinations of
services (such as high quality video conferencing, IPTV, streamed video, video and
software downloads and general web usage), the extent to which these transmission
requirements are likely to change in the medium term due to technological innovation
(e.g. video compression) or moves towards “content-rich” environments, and the extent
to which increases in download and upload speeds are likely to yield economically
noticeable changes in existing overall service quality.16
16 Relevant here is the work of Robert Kenny and Tom Broughton, Domestic demand for bandwidth, an approach to forecasting requirements for period 2013–2023, see
CBA Analytical Framework Page 30 of 95
![Page 31: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/31.jpg)
Similarly, on the supply side, answering these questions will require an examination and
assessment of existing infrastructure, including the download and upload speeds that
can be experienced by customers using copper lines connected to ADSL2+ enabled
exchanges, and the coverage and capability of existing hybrid fibre coaxial (HFC)
networks, fixed wireless networks, and 3G and 4G mobile networks. It will also require
an assessment of likely future supply side developments (e.g. developments in wireless
and satellite technology, increased use of VDSL) that may occur in the absence of major
policy changes.17
It is impossible to precisely determine the dynamically efficient path of high speed
broadband infrastructure investment in an Australian context (taking into account
demand and supply side uncertainty, irreversibility of investment, and the potentially
large number of option values that are embedded in investment decisions). However, it
would be highly desirable for a CBA to determine the most important economic features
of such a dynamically efficient path – including the economic trade-offs that are
involved, and the incentives on the demand and supply side of the relevant markets.
Given the results of such an analysis, it would also be desirable to examine and
determine the extent to which market outcomes – supported by appropriate regulation
– will deliver the features that characterise the dynamically efficient investment path.
Such features include the timing of aggregate investment, the extent of population
coverage over time, the extent of competition, pricing arrangements, capacity choice
and congestion, download and upload speeds, as well as other characteristics.
Addressing these issues necessitates the construction of a credible reference scenario,
which in turn involves a careful consideration. Given that there is a significant amount
of uncertainty regarding the precise evolution of technology, as well as the development
of consumer applications and devices over the medium term, it would be preferable to
develop a suite of different reference scenarios, each of which can ultimately be used as
an input into sensitivity analysis.
<http://www.commcham.com/storage/publications/BSG-Domestic-demand-for-bandwidth.pdf>.
17 Much of the work for such an assessment has recently been carried out by the Australian Government; see the Department of Communications’ 2013 report, Broadband Availability and Quality.
CBA Analytical Framework Page 31 of 95
![Page 32: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/32.jpg)
Factors that should be addressed in the construction of these scenarios include:
The current state of Australia’s broadband market, and the economic drivers of
the evolution of the market to date.
The evolution of private sector telecommunications infrastructure investment in
Australia, and the incentives that competition and market forces have provided
for such investment.
The role that regulation has played in influencing the evolution of market
outcomes, including its influence on incentives for investment, consumer prices,
capacity, network coverage and speed.
Given supply and demand side uncertainties, this paper is deliberately non-prescriptive
regarding the specific choice of reference scenarios. However, at the minimum, the CBA
could examine three broad sets of reference scenarios:
Pessimistic scenarios in which investment in – and take-up of – faster speeds are
inefficiently delayed to a significant extent, or where the breadth of coverage is
significantly inefficiently narrow. In constructing this scenario it is crucial to
carefully identify the reasons that may drive such an outcome – that is, the
market or regulatory failures that are likely lead to inefficient delay.
Intermediate baseline scenarios, in which market outcomes are, in theory, likely
to be dynamically efficient. In developing this scenario it would be important to
explore the reasons why market failures are likely to be minimal, and the policy
and regulatory settings that are likely to lead to such an outcome.
Optimistic or overinvestment scenarios, in which investment in – and take-up of
– faster broadband occurs more rapidly than the dynamically efficient path, or
where network coverage is extended to areas where it is inefficient to do so.
Again, in constructing such as scenario it will be important to carefully identify
the market or regulatory failures that may lead to inefficient overinvestment.
Different features from a subset of these scenarios could be chosen and combined with
each other to form hybrid baseline scenarios. The extent to which such baseline
scenario-building is employed is ultimately a matter of judgement for the CBA analyst.
CBA Analytical Framework Page 32 of 95
![Page 33: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/33.jpg)
The quantitative results of the implicit optimisation exercise undertaken in a CBA will be
a function of the assumed parameters (i.e. the reference scenario). This is a strength of
a CBA, rather than a weakness, as results can be mapped clearly and transparently back
to the assumptions that are made. On the other hand, it is possible that some of the
qualitative results of a CBA may not vary greatly with the assumed reference scenario.
For example, it may be the case that for all reasonably likely paths of willingness to pay
and costs, a particular policy never yields positive net benefits. Importantly, if there is a
great deal of uncertainty about the reference scenario (or uncertainty about other
modelling parameters or assumptions) and the qualitative features of the CBA vary
widely with these assumptions, then this suggests that there is a great deal of risk
associated with the project. This is likely to be useful information for a decision maker.
Once these reference scenarios have been constructed, they can then be compared
against the likely outcomes under various policy scenarios. The way in which this can be
done is illustrated in Table 2.1 below. The assumptions and parameters used to develop
such scenarios could also form part of the project scenarios. For example, one project
scenario could be that private suppliers target consumers with willingness to pay greater
than the incremental costs of higher broadband speeds, whilst the government also
mandates a community service obligation (CSO) to serve some or all other consumers.
CBA Analytical Framework Page 33 of 95
![Page 34: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/34.jpg)
Table 2.1: Two-Way Classification of Policy Alternatives and Scenarios
Private
Investment
Scenarios
Policy Alternatives
Policy A
(No policy
change)
Policy B Policy C Policy D
1: Low $0
2: Medium $0
3: High $0
Note: Entries in the cells in each row are the net social benefits of different policies relative to the relevant
baseline in that row. In some instances combinations of policies may be involved. Not every cell needs to
be filled with an estimate.
In comparing baseline scenarios against policy scenarios, it is important that the CBA
examine and realistically assess the extent to which policy changes are likely to achieve
their stated goals, and the extent to which they will actually address the sources of
market or regulatory failure that are assumed to drive the evolution of outcomes in the
baseline scenario. Policy changes that fail to adequately target these sources of failure
are less likely to yield incremental net benefits relative to the baseline, and under some
circumstances may actually yield negative incremental benefits even assuming a
relatively pessimistic baseline scenario.
CBA Analytical Framework Page 34 of 95
![Page 35: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/35.jpg)
3. BenefitsThere are two broad groups who may benefit from faster broadband speeds, and for
which private willingness to pay could therefore be estimated:
Consumers – individuals and households for whom faster speeds directly
enhances the value of existing and new entertainment, educational and other
applications.
Producers – small and large businesses, for whom faster upload and download
speeds may result in higher profits.
Conceptually, faster broadband speeds may provide a range of benefits to consumers
and producers. Consumers and producers may experience greater benefits from using
existing products, and are also likely to benefit from the ability to use new products.
Faster upload speeds for business may result in pure time savings uploading
information, whilst households may experience similar gains as a result of both faster
upload and download speeds. Other possible benefits include reduced transaction
costs, cost savings related to shorter travel times, and possible productivity
improvements. Finally, faster broadband speeds, by reducing effective “travel costs”
(where these costs encompass a broad range of factors that may inhibit worker
flexibility) may improve labour force participation, enhancing the ability of workers to
take advantage of employment opportunities that would otherwise not be present due
to the constraints imposed by slow broadband speeds.
This section examines these sources of benefits. Any CBA will have to acknowledge that
there is a considerable amount of uncertainty regarding:
The size or quantum of benefits.
The geographical/spatial pattern of benefits – e.g. urban versus regional.
The intertemporal pattern of benefits.
This section also explores some of the issues involved in estimating benefits at a point in
time as well as over time, and social benefits from faster broadband speeds that may
not be reflected in estimates of private willingness to pay.
CBA Analytical Framework Page 35 of 95
![Page 36: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/36.jpg)
3.1. The Indirect Demand for Faster Broadband Speeds
On the demand side, it is important to note that consumers do not directly value access
to faster speeds. Rather, the goods that are valued directly are services, applications
products and devices. Access to faster broadband speeds is one of many characteristics
that can enhance the value that consumers attach to these end-use products. Even in
the case of faster download speeds and direct travel cost savings (explored further
below), faster speeds not valued directly – they are only valued because they open up
opportunities for more downloads, greater leisure time or longer working hours (which
allow consumers to increase their labour income and consumption possibilities).
To see this more clearly, consider an extreme case in which there were no new or
existing applications that required very fast speeds. Then consumers would obviously
attach no value to accessing a network that promised further speed upgrades. Hence,
the value attached to access to faster speeds is an imputed or derived value, and the
marginal valuation curve for network access and the faster speeds it may provide (or the
“demand curve”) is a derived demand curve.
To understand this analytically, consider Willig’s (1978) approach to valuing changes in
the attributes of goods.18 Holding prices and income constant, the welfare gain from
being able to access a network with a greater speed of (compared to the current
network speed of ) can be estimated in one of two ways:
1. Willingness to pay can be estimated directly.
2. Willingness to pay can be indirectly estimated by adding the changes in
consumer surpluses in markets for various products and services when a change
in speeds induces a shift of those demand curves, assuming prices and income
are held fixed.
The basic idea is illustrated in Figure 3.2 below, where an increase in broadband speed
induces an increase in demand for good A (e.g. streamed video), but reduces demand
for good B (e.g. standard video). The change in consumer surplus in market A is the
change in the area underneath the demand curve for good A, whereas the change in
18 See also Johansson (1987), Chapter 6.
CBA Analytical Framework Page 36 of 95
![Page 37: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/37.jpg)
consumer surplus in market B is the change in the area underneath the demand curve
for good B. The overall effect on consumer welfare of this change in attributes or
characteristics is the sum of these two changes.
Figure 3.2: Illustration of the Effect of a Marginal Increase in Broadband Speeds on
Welfare and Demand for Other Products
0AP
1Ax
1Bx
11
BDAD
Market A Market B
0Bx
APBP
0BP
00
0Ax
'BD'AD
ACSBCS
Note that since the welfare effects of changes in Internet speeds are isolated, prices and
incomes are assumed not to change, and so the consumer’s overall spending is also
assumed not to change. Instead of overall spending changing, there is a shift in
spending towards goods whose marginal value has increased as a result of faster
speeds, away from goods whose marginal value has fallen. This analysis therefore
suggests that estimating changes in overall spending on final goods and services – i.e.
gross domestic product (GDP) – may not completely capture the full welfare effects of
changes in product quality and attributes in general, and the welfare effects of faster
broadband speeds in particular.19
For new goods, by definition the marginal cost exceeds the greatest marginal willingness
to pay. In this case, the above approach that directly links changes in demand to
willingness to pay for faster speeds needs to be modified. Suppose, for example, that
19 This does not mean that changes in broadband speeds will have no effect on GDP. The point is that changes in GDP are neither a necessary nor a sufficient condition for changes in broadband speeds to affect economic wellbeing.
CBA Analytical Framework Page 37 of 95
![Page 38: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/38.jpg)
the marginal cost of a new good is so high that even with an increase in demand
following an increase in speed, the market equilibrium quantity is still zero. Then, even
though demand has changed, no benefits are in fact realised. In Figure 3.3 below, the
initial price of good B (which would reflect its marginal production costs) is so high that
there is no demand for it at slow speeds. The increase in speed results in positive
demand, but the benefits need to be computed carefully. The main conclusion is that
the marginal willingness to pay for faster speeds will depend, among other things, upon
current speeds, prices, and income.
Figure 3.3: Illustration of the Effect of a Marginal Increase in Broadband Speeds on
Welfare and Demand for Other Products: The Case of a New Good
0AP
1Ax
1Bx
1
1
BDAD
Market A Market B
APBP
0BP
0
0
0Ax
'BD'AD
ACS
BCS
It may also be possible that faster broadband speeds may increase the variety of
products available to consumers. The above framework can be extended to cover this
case. Since such improvements in variety directly increase consumer and/or producer
welfare, it is reasonable to expect that a large share of these benefits is likely to be
captured in consumer and producer willingness to pay for faster broadband speeds.
CBA Analytical Framework Page 38 of 95
![Page 39: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/39.jpg)
3.2. Pure Time SavingsFaster broadband speeds may have a pure time saving effect with respect to
downloading information, which may result in both work and leisure (i.e. non-work)
time savings. Ergas and Robson (2009) develop a simple model which illustrates this
idea, using the simple time-allocation model originally developed by Becker (1965).20 In
this model, a consumer spends time either working or downloading. In that model the
consumer has a lump-sum income of Y and earns labour income at a nominal wage rate
of w. Let L be the number of hours of labour supplied. Assuming that the consumer
consumes both downloads and other consumption goods, their combined time and
budget constraint is:
The consumer maximises utility from consumption goods and the quantity of
downloads.21 The basic model is illustrated below.
20 Goolsbee and Klenow (2006) use Becker’s framework to compute the consumer benefits of access to the Internet, but they do not examine the welfare effects of greater download speeds.
21 The assumption that workers either work or download may be unrealistic. For example, individuals would do other work while waiting for downloads. The same argument applies for savings in leisure time from faster downloads. In these cases, the pure time savings from faster download speeds would clearly be smaller than those described in the model.
CBA Analytical Framework Page 39 of 95
![Page 40: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/40.jpg)
Figure 3.4: Faster Broadband Speeds in a Simple Time-Use Model
Consumption c
Tpw
pY
max ,0ipY sTp p
sTDownloads (D)
Constraint: iDY w T pc p Ds
Slope of Constraint is: ipdc wdD sp p
Source: Ergas and Robson (2009)
The consumer derives incremental benefits from an increase in download speeds, as this
frees up additional time for working, may boost the individual’s labour supply and widen
the individual’s consumption possibilities. The effect of an increase in speed in this
model is shown in Figure 3.4 below.
The increase in speed has both an income and substitution effect, altering the relative
returns to leisure. The volume of downloads increases, but time spent downloading
may either rise or fall, depending on the elasticity of the volume of downloads with
respect to changes in speed. Overall, faster speeds create marginal benefits that
depend inversely on the square of the speed.
CBA Analytical Framework Page 40 of 95
![Page 41: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/41.jpg)
Figure 3.5: The Impact of Faster Broadband Speeds on Consumer Welfare
Consumption c
Tpw
pY
0s T 1s T Downloads D
Source: Ergas and Robson (2009)
3.3. Reduced Transaction Costs
Faster broadband speeds may lower transaction costs in markets (e.g. the costs of
searching for goods and services). However, since reductions in transaction costs
increase consumer and/or producer surplus, a significant share of these benefits may be
captured in consumer and producer measures of willingness to pay for faster broadband
speeds. Hence, even if such effects can be identified and measured, a great deal of care
needs to be taken to ensure that there is no double counting of benefits.
In markets where transaction or time costs are significant, consumer demand depends
on the full price of the good (the money price, plus the transaction costs needed to
obtain the good). Similarly, the full price obtained by the seller is the money price
received from the buyer, less the transaction costs associated with finding a suitable
buyer, negotiating, and so on. Welfare in such markets depends on the sum of the
transaction costs on each side of the market. If transaction costs fall, then the incidence
is shared between buyers and sellers, depending on the relative size of demand and
supply elasticities. It is possible that money prices could rise following a fall in
CBA Analytical Framework Page 41 of 95
![Page 42: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/42.jpg)
transaction costs – although overall consumers would still be better off because the full
price of the good would have declined.22
3.4. Travel Time Savings
Faster broadband speeds may lower travel times for individuals travelling to and from
their place of work. The effects of shorter travel times on labour supply are discussed in
section below. Opportunity cost is the key driver of time costs. That is, conceptually,
the value of time saved depends on the utility (and disutility) experienced during travel,
compared with the value of the next best alternative activity. More generally, faster
broadband speeds may allow workers greater flexibility in arranging their work
schedules to fit in with family and other commitments, inducing individuals to explore
employment opportunities that were previously unavailable to them. Faster network
speeds may also allow individuals to work across state and international borders
effectively, and hence improve the ability of individuals to collaborate and compete
internationally. These improvements, which come about as a result of a reduction in
“travel costs” (where these costs are broadly defined and include vehicle operating
costs, parking costs, and the costs of inflexibility)23 would result in an expansion in
labour force participation. However, since such reductions in travel costs directly
increase consumer and/or producer wellbeing, a significant share of these benefits may
be captured in consumer and producer willingness to pay for faster broadband speeds.
Again, care needs to be taken to avoid double counting.
3.5. Productivity Improvements
Faster broadband speeds may increase the productivity of existing factors of production
such as labour and capital, leading to a direct increase in GDP. Increases in total factor
productivity (TFP) feed directly into reductions in unit costs of production (holding
22 This proposition is demonstrated in numerous economics textbooks. See, for example, Stockman (1996) pages 143–144. A more detailed analysis demonstrating these and other results on transaction costs can be found in Chapter 1 of Robson (2012).
23 If the negative effect on physical commuting is sufficiently large, there may also be reduced congestion costs.
CBA Analytical Framework Page 42 of 95
![Page 43: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/43.jpg)
wages and other input prices constant),24 and as such are likely to be captured in
producer willingness to pay for faster broadband speeds.
As explained in Appendix 2, the welfare effects of any productivity gains do not depend
on the effect of productivity changes on employment. Hence, in measuring any welfare
gains from improvements in total factor productivity, there is no need to measure the
number of jobs that are created or destroyed. Section explains the circumstances
under which changes in labour supply may have important implications for a CBA and
may need to be computed and included.
3.6. Estimation of Private Willingness to Pay
As outlined at the beginning of section 3, the benefits of faster broadband speeds will
likely accrue to both consumers and producers. The previous sections examined some
of the conceptual microeconomic foundations of faster broadband speeds. The analysis
demonstrated that these can be usefully studied using the “characteristics” approach as
well as other common economic approaches, and can be incorporated into standard
microeconomic models of consumer choice. Conceptually the direct drivers of
household willingness to pay for faster speeds are likely to derive from time savings
accessing existing products, services, applications, as well as making new products,
services and applications available, e.g. new health and educational products. Faster
upload and download speeds may also lower business costs and improve profits, lower
consumer prices, or some combination of both. The theoretical analysis suggests that
widely used econometric techniques may be used to estimate willingness to pay for
faster broadband speeds.
3.6.1. Hedonic Regression Using Market Data
One approach to estimating willingness to pay for faster speeds is to use aggregate time
series market data to estimate market demand curves, from which the marginal effect
of faster speeds can be ascertained. This approach can be summarised very briefly as
follows. Suppose that actual data on the (logarithm of) prices (pt), speeds (st), and a
vector of other package characteristics (zt) was available. A posited relationship
between price and these variables might take the log-linear form:
24 Harberger (1998) explores this interpretation of changes in total factor productivity.
CBA Analytical Framework Page 43 of 95
![Page 44: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/44.jpg)
The basic idea of hedonic regression methods is to use the coefficient on speed in to
provide some indication of the “implicit price” or marginal willingness to pay for faster
speeds, holding all other variables constant. This “hedonic” approach using market data
has been applied in a wide range of settings, including to the valuation of housing
characteristics, automobiles and other consumer goods.25
3.6.2. Discrete Choice Estimation Using Survey Data
An alternative approach to estimating willingness to pay for faster speeds is to use
discrete choice econometric methods. These may be applied to market data at a single
point in time (assuming the data set is sufficiently rich). Alternatively, data could be
collected using a carefully designed consumer survey.
There is a vast literature on these methods and their applications, but the basic
approach applied to broadband speeds can be briefly summarised as follows.26 In a
discrete choice or random utility estimation framework, consumers are confronted with
a set of J different broadband packages, labelled , and are asked to choose
one of them. Let be the direct utility gained by the individual when they choose
package j. This direct utility is assumed to be a function of observable and unobservable
factors:
where only one of the packages is actually chosen, so that . Here, is a vector
of characteristics of package j, and is a composite other good, whilst is a random
disturbance term reflecting unobserved characteristics. Let the price of package j be
25 A number of obvious difficulties arise in estimating an equation such as and isolating consumer willingness to pay. For example, this ignores the supply side and the marginal cost of producing different attributes. There is a vast literature exploring these techniques. For a summary of the theory behind the hedonic approach, see Rosen (2002). Triplett (2006) provides an overview of examples and information technology (IT) related applications, and outlines the practical difficulties and econometric issues that are typically encountered when using hedonic methods.
26 McFadden (1999) is a basic reference for the material in this section. See also Bateman, Bockstael and McConnell (2007).
CBA Analytical Framework Page 44 of 95
![Page 45: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/45.jpg)
. For convenience, assume that if alternative j is chosen, then ; otherwise .
If the individual chooses alternative j, indirect utility is then:
where it is assumed that the characteristics of packages that have not been chosen do
not affect utility.
As shown in the Appendix 3, under certain technical assumptions the expected welfare
effects of (willingness to pay for) faster broadband speeds can be directly recovered and
computed from the estimated parameters. However, if alternative assumptions are
employed, then it is not possible to obtain closed form expressions for willingness to
pay, and the use of numerical methods would be required.
3.7. Benefits Over Time
Depending on survey design, estimates from a discrete choice framework may reflect
consumer judgements regarding the present value of their willingness to pay for a
stream of benefits over time. Alternatively, such an analysis could be used to estimate
the willingness to pay for monthly or yearly access to faster speeds. The intertemporal
pattern of demand for applications, products and services will drive the evolution and
growth of aggregate benefits of faster speeds over time. Since any large public
infrastructure project will typically take many years to construct and will likely have high
upfront costs, it becomes crucial to specify exactly when future benefits are likely to
accrue in both the policy and baseline scenarios. Therefore, credible projections of
likely take-up rates are required, and these should be subjected to detailed sensitivity
analysis.
Forecasts and projections of aggregate take-up from NBN Co and from NBN Co.’s
December 2013 Strategic Review may assist with the construction of such projections.
However, in utilising these forecasts and projections in a CBA, it would be necessary to
clarify the assumptions on which they are based and the methods used to generate
them, so that they can be suitably modified under different project and reference
scenarios, and sensitivity analysis can be conducted.
3.7.1. Product/Innovation Diffusion Models
CBA Analytical Framework Page 45 of 95
![Page 46: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/46.jpg)
One approach to forecasting likely take-up rates is to study the patterns of similar
technologies, as well as international evidence on broadband take-up rates over time.
In the marketing and forecasting literature new product or innovation diffusion models
have been specifically developed and used for this purpose.27 Generally speaking, these
are pure time series forecasting approaches, but which sometimes incorporate other
variables such as price and advertising expenditure.
Product diffusion curves are typically assumed to follow an “S-shaped” path: there is
rapid take-up initially, but then take-up rates slow down as the market becomes fully
saturated. For example, a widely used model is the logistic diffusion model. Let c be the
take-up or adoption rate, and let be the rate of change in take-up over time. Then
according to the logistic model, the growth rate of take-up obeys:
where , and where is an upper bound on the take-up rate. As this
model results in the widely used Gompertz function:
Figure 3.6 below plots an example of the logistic production diffusion curve, whilst
Figure 3.6 plots the change in the take-up rate implied by such a curve. “S-shape” take-
up paths have been estimated in telecommunications markets (e.g. mobile phones) and
this approach may be usefully applied to broadband take-up rates. There are a wide
range of flexible approaches which could be used.
Figure 3.6: Example of a Logistic Product Diffusion Curve (=0.3, =2).
27 For a survey of the extensive forecasting literature in this area, see Meade and Islam (2006). Geroski (2000) reviews the literature on the microfoundations of diffusion models.
CBA Analytical Framework Page 46 of 95
![Page 47: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/47.jpg)
0
10
20
30
40
50
60
70
80
90
100
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41
Time
Take
up
Figure 3.7: Change in Take-up Rates Implied by the Logistic Product Diffusion Curve
(=0.3, =2).
0
1
2
3
4
5
6
7
8
9
10
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39
Time
Cha
nge
in T
akeu
p R
ate
One of the main concerns with using such curves in a CBA is that they may fail to specify
the “deep parameters” (i.e. underlying consumer preferences) that are supposed to
have generated the forecast take-up path. Appendix 4 shows that if benefits take a
special functional form, then these “S-shaped” take-up paths can be derived from a
simple intertemporal consumption optimisation problem. The main point of the
analysis in the appendix is to show that by making an assumption about the path of
adoption, a CBA is also implicitly making an assumption about the evolution of
CBA Analytical Framework Page 47 of 95
![Page 48: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/48.jpg)
consumption benefits over time. Assumptions about one cannot be made
independently of the other, and they should be consistent with each other.
CBA Analytical Framework Page 48 of 95
![Page 49: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/49.jpg)
3.8. Network Effects
A further possible source of net benefits that may be important for the analysis of faster
broadband speeds falls under the general heading of “network effects”. In the context
of a CBA of broadband network infrastructure, attention has focused on two
conceptually separate effects: direct demand side network externalities, and indirect
demand side network externalities. The next two subsections examine these effects
separately.
3.8.1. Direct Network Externalities
Direct network effects are said to be present when an individual’s marginal benefit from
using the network increases with the number of other users of the same network.
There is an extensive literature analysing the demand side externalities that may be
associated with Internet use [see, for example, Shapiro and Varian (1999) or Varian
Farell and Shapiro (2004)]. In other economic contexts (such as the analysis of social
norms or peer or group effects on economic behaviour), these effects are referred to as
“strong complementarities” [Becker and Murphy (2000)]. The basic idea is that for a
given speed, consumer marginal willingness to pay may depend in some way on the
number of other users on the network. In some cases, marginal social benefits of a
larger network may exceed private marginal willingness to pay for access to the
network. On the other hand, at some point as more users are added, congestion may
occur, reducing private marginal willingness to pay below marginal social benefits.
In theory, the presence of strong positive demand side network externalities may imply
that in the absence of policy interventions there will be an inefficiently low number of
broadband users, as each user imposes an uncompensated benefit on the others. On
the other hand if, at the efficient network size, the marginal externality is negative,
there will be too many network users.
It is important to remember that direct network externalities typically refer to the
demand side rather than the supply side. If there is an existing network and the
presence of a direct demand-side externality leads to an inefficiently small network of
adopters (so that there is excess network capacity) then it is unclear why building
CBA Analytical Framework Page 49 of 95
![Page 50: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/50.jpg)
additional network capacity would improve network utilisation. In addition, many users
of broadband networks only interact with a small group of other users (such as close
friends or family members), rather than the entire network. This increases the
likelihood that users will make private arrangements to internalise any direct network
externalities that may exist between them.
Furthermore, instances in which there a large number of interactions between network
users are likely to involve commercial transactions – in which case externalities are once
again more likely to be internalised via market arrangements. For example, any
government project is likely to be a wholesale network, not a retail network.
Intermediaries (retail service providers) will likely provide a range of plans with various
speeds that can exclude some users from accessing faster speeds. They may be able to
commit to providing faster speeds to consumers who exhibit network effects, and then
charge users a price so that users obtain their reservation utility [see, for example, Segal
(1999)]. In that scenario, any network externalities would effectively be internalised and
could be captured by measuring intermediaries’ willingness to pay.
One of the tasks of a CBA would be to examine, from an empirical perspective, the
extent to which current broadband adoption rates in Australia are inefficiently low due
to demand-side network externalities, and the extent to which faster speeds may
impact upon any network externalities. Whilst network effects typically refer purely to
increases in the number of users on a given network, the logic of network externalities
may also apply – with appropriate modifications – to changes in network functionality.
For example, following an increase in network speeds, some users may able to broaden
their interactions with other users (who may already be on the ‘network’ in a physically
connected sense) using a new technology (e.g. video) that was not previously available.
In this scenario, the new interactions would simply not have been possible but for the
increase in network speed.
A related concern regarding network effects is the extent to which they may cause
estimates of private willingness to pay for faster speeds to differ from the social benefits
of faster speeds.28 If there are strong network effects present and individuals are asked
to report their willingness to pay for faster speeds, then their stated willingness to pay is
28 See, for example, Quiggin (2013).
CBA Analytical Framework Page 50 of 95
![Page 51: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/51.jpg)
likely to be based on their expectation of the number of users of the network – even if
they are not explicitly asked about this in a survey question. This estimate could be
larger or smaller than the efficient number of users, and so the implications for the
results of willingness to pay estimates are ambiguous. In any case, it may be possible to
use discrete choice econometric methods to directly estimate the size of network
effects, and these could then be incorporated into a CBA. There is a growing empirical
literature that provides guidance on how these kinds of network interdependencies can
be accounted for and estimated using standard discrete choice econometric
techniques.29
Appendix 5 contains a theoretical analysis of network effects. One of the conclusions of
this analysis is that the incentive for the private sector to provide access to faster
maximum speeds depends on the marginal willingness to pay for higher maximum
speeds of the marginal consumer, whereas efficiency is concerned with the average
marginal willingness to pay, taken over the entire network. Even in the presence of
direct network externalities, it is not at all clear that the marginal consumer would be
willing to pay a lower amount for faster maximum speeds than the average consumer.
Hence, even if the presence of network externalities creates a discrepancy between
market outcomes and the socially efficient allocation of resources, it is not clear what
the sign or size of that discrepancy actually is.
If the network is subject to congestion effects (i.e. negative marginal network
externalities), matters are even less clear. If there are too many network users in a
market setting and the network becomes congested, then we are in a second best
economic environment. In this setting, commercial realities could mean that firms may
try to offset some of the congestion effects and compete for the marginal customer by
offering inefficiently fast maximum download speeds.
29 See, for example, Brock and Durlauf (2001).
CBA Analytical Framework Page 51 of 95
![Page 52: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/52.jpg)
3.8.2. Indirect Network Externalities
Indirect network externalities arise when a larger number of network users means that a
larger range of associated applications, products, services and devices can be supported,
or that as the market expands these associated products can be produced at lower unit
costs and therefore sold at lower prices to consumers. They differ from direct
productivity gains identified earlier, in that they do not result in a downward shift in a
firm’s cost curve – instead, they involve a movement along a downward sloping average
cost curve. As a number of authors have noted,30 it is more appropriate to treat these
effects as pecuniary externalities (i.e. effects that are mediated through the price
system) rather than technological externalities. For the purposes of welfare analysis,
these kinds of externalities can be ignored, as the gains to consumers from lower prices
are merely transfers from producers.
Consider, for example, the case in which a larger number of network users translates
into a large number of users of a particular “app”, and that this drives down the unit
costs of supplying the “app”. Analytically, this is simply an example of a competitive
industry which exhibits a declining long run supply curve. As the demand curve for the
good shifts out, input prices fall, and marginal product costs fall. The reduction in prices
is a transfer from the owners of the scarce factors of production to consumers, and
there is no overall change in welfare.
3.9. Fiscal Effects
Changes in broadband speeds may have implications for the pattern of tax revenue and
government spending in the economy as a result of changes in private demands. These
changes in revenues may or may not have welfare consequences. These issues are dealt
with further below.
There is an additional, conceptually separate class of welfare effects to consider in
relation to government-provided goods and services. Faster broadband speeds may
reduce the costs of supplying publicly provided private goods such as health and
education (some of which have public good characteristics, but which may be more
30 See, for example, Liebowitz and Margolis (1995).
CBA Analytical Framework Page 52 of 95
![Page 53: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/53.jpg)
appropriately regarded as merit goods), may improve the quality of existing services, or
may allow new services to be offered. In these cases, overall economic benefits depend,
among other things, on whether those services are provided efficiently (both in a cost
sense and in a wider economic sense).
Many existing studies have attempted to estimate the benefits that faster broadband
speeds may bring in the form of new or higher quality government services, but they
suffer from a major drawback: they do not provide estimates of the costs that are likely
to be involved in delivering those services. If faster broadband speeds allow
governments to provide new or higher quality services in health, for example, then new
medical equipment may be needed, doctors and nurses may need to be trained in new
techniques, and patients may need to purchase their own equipment. Failing to
subtract the costs associated with new or higher quality services will produce incorrect
estimates in any CBA.
Claims about reductions in government spending should also be treated with caution.
Consider Figure 3.8 below, which assumes that taxes impose no deadweight cost, so
that the optimal provision of a public good obeys the standard Samuelson rule (i.e.
supply up to the point where the social marginal benefits equal the marginal resource
costs). Suppose that faster broadband speeds reduce the marginal cost of providing the
relevant good (MC0 to MC1). Then in the Figure below, it is not only optimal to increase
the level of G that is provided (from G0* to G1
*); it is also optimal to increase spending on
the good. In this example, although the reduction in unit costs reduces spending
(reducing it be the shaded blue area), efficiency requires that more of this good be
supplied, which increases spending (by the shaded pink area). Overall, spending
increases in this example. But in general the sign and size of the effect depend on the
elasticity of the social marginal benefit curve. If this curve is relatively inelastic, then a
fall in costs will reduce spending. The welfare gain here is the gain in social benefits,
which is equal to the blue shaded area plus the green shaded area. The analysis
suggests that although changes in government spending and efficiency effects are
obviously related, they are conceptually very different and so it is important to treat
them separately.
CBA Analytical Framework Page 53 of 95
![Page 54: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/54.jpg)
Figure 3.8: The Effects of Lower Unit Costs on Optimal Spending and Provision of a
Publicly Provided Good
Quantity
Marginal Benefit, Marginal Cost
*0G
*1G
0MC
1MC
MB
This point is illustrated further in Figure 3.9 below, which modifies the previous case by
assuming that the taxes needed to fund public goods are distortionary (this point is
discussed further below). The welfare gain from lower costs is now the area bounded
by the points a, b, c, and d. Overall, even though unit costs fall, spending still rises – and
this is efficient.
CBA Analytical Framework Page 54 of 95
![Page 55: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/55.jpg)
Figure 3.9: The Effects of Lower Unit Costs on Optimal Spending and Provision of a
Publicly Provided Good – Taxes are Distortionary
Quantity
Marginal Benefit, Marginal Cost
*0G
*1G
0MC
1MC
MB
ab
cd
In these two examples, government was assumed to be acting efficiently and responding
in an efficient manner to changes in costs. If this is not the case, then the link between
reductions in costs, spending and efficiency becomes even weaker and more tenuous.
Hence, in the case of publicly provided goods, looking solely at changes in government
spending induced by public investment in an HSBB network is likely to be a poor proxy
for the community benefits of faster broadband speeds.
CBA Analytical Framework Page 55 of 95
![Page 56: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/56.jpg)
3.10. Wider Economic Benefits (WEBs)
Wider economic benefits are those not typically captured elsewhere in a standard CBA
(i.e. by estimating individual consumer and producer willingness to pay). This
subsection examines three wider economic benefits that are typically considered in
CBAs.
3.10.1. Economies of Agglomeration
In the context of urban economics, technological economies of agglomeration are said
to arise when a business or worker’s productivity is higher if they locate in a large city
rather than in a smaller city (or, alternatively, if they locate within a “cluster” of similar
workers or businesses).31 It is typically argued that these agglomeration effects act as
positive technological externalities that occur when a new road or railway is
constructed.
Similarly, “virtual” economies of agglomeration (i.e. the counterpart, in an online
environment, of these more traditional spatial agglomeration effects) refer to the
possibility that faster broadband speeds may lead to similar productivity benefits from
“clustering”, but without the need for firms or workers to be physically located in the
same geographic area.32
In urban economics the standard approach is that reducing transport costs increases the
extent to which activities are physically linked to each other, and hence individuals are
better able to take advantage of any agglomeration economies. However, whilst such
effects are indeed likely to be present, they need to be treated with a great deal of care
in a CBA: to generate the additional economies of agglomeration, workers or firms must
have located from one place to another, and a CBA should take account of the
agglomeration diseconomies that would presumably have occurred in the location from
which workers and firms have relocated. Only the net agglomeration effect – which
could be negative – should be counted in a CBA.
31 See, for example, Glaeser (2008), Small and Verhoef (2007) and Brueckner (2011)
32 Plum Consulting (2008) contains a brief discussion of this idea in the context of broadband policy.
CBA Analytical Framework Page 56 of 95
![Page 57: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/57.jpg)
Similar considerations are likely to apply to virtual economies of agglomeration. If
virtual “clustering” is a substitute for physical “clustering”, then it is only the net
agglomeration effect that should be counted in a CBA. For example, if workers decide to
no longer commute to a large city and instead work from a regional location and
communicate via teleconference over an HSBB connection, then, although this may
produce virtual agglomeration economies, it would have the opposite effect on physical
agglomeration economies. Whilst the workers themselves would clearly be better off
(having revealed a preference for not travelling to the city), the inclusion of economies
of agglomeration in a CBA is an attempt to capture external benefits which have not
been captured in private benefits. It is not obvious whether the net external effect
would be positive.
On the other hand, the ability to engage in “virtual clustering” may be complementary
to physical clustering. That is, the ability of workers or firms to access HSBB may make
cities even more attractive places to work. In this case, HSBB may produce further
physical economies of agglomeration (which, as before, would need to be compared
with physical diseconomies as workers or firms move from regional areas to cities).
Overall, it is not clear what the sign of “virtual” economies of agglomeration would
actually be, nor the size of these effects. Therefore, as is the case with more traditional
physical economies of agglomeration, such effects should be treated with a great deal of
care in any CBA. In any case, in a CBA the ultimate question is whether such effects,
even if they are positive, are likely to be greater in the project scenario compared to the
reference scenarios, and if they are, whether these gains justify the additional costs that
may be associated with the project.
Finally, even if there were significant “virtual” agglomeration economies, there are a
number of significant practical obstacles to be overcome before these could be
accurately estimated. For example, it would be necessary to separate productivity
differences due to virtual linkages from those due to more traditional transport linkages.
There could also be a risk of double counting with respect to conventional benefits and
benefits from network economies.
CBA Analytical Framework Page 57 of 95
![Page 58: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/58.jpg)
3.10.2. Labour Supply and Taxation
It is often argued that faster broadband speeds will increase the supply of labour.33 This
may not only have a beneficial economic effect in its own right, but may also increase
the revenue from taxes on labour income – which should be counted as an additional
source of benefits.
Both effects can be examined using basic principles of economics. The simple time use
model of the effects of broadband speeds, explored earlier in . Pure Time Savings,
indicates that faster broadband may increase labour supply. In this model, the
consumer’s entire non-work time is assumed to be downloading. Faster broadband
speeds make this form of leisure more valuable but because more time becomes
available as a result of faster speeds, this may also increase labour supply.
If faster broadband speeds reduce work travel times, then labour supply may also
increase. Using a standard model of time use and labour supply, it is straightforward to
show that a reduction in travel times has the same economic effect as an increase in an
individual’s lump sum income, and does not change the real wage. It follows that if
leisure is a normal ‘good’, then a reduction in travel times will lead to an increase in
leisure, but labour supply may also increase.
However, in these models the welfare effects of faster broadband speeds do not depend
on the response of labour supply. The labour supply effect of faster broadband speeds
becomes important when there is a wedge between the marginal benefit and marginal
cost of labour (e.g. an income tax). These taxes reduce the compensated supply of
labour, and anything that reverses these disincentive effects (i.e. increases the quantity
of labour supplied) will increase revenue collected from labour income taxation, without
requiring an increase in income tax rates. Hence, any revenue gains are welfare gains.
Similarly, any losses due to reductions in labour supply are pure welfare losses. The
same analytical considerations apply to pure productivity improvements which result in
increased tax revenues. The above analysis suggests that in the presence of distorting
income taxes it may be important to examine the “spending effect” of faster broadband
33 See, for example, Dettling (2013) who finds that access to high-speed broadband services at home has improved the labour force participation of married women.
CBA Analytical Framework Page 58 of 95
![Page 59: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/59.jpg)
speeds (see the analysis of the marginal cost of funds in the Appendix 7) since this may
alter the welfare effect of any government project.
3.10.3. Competition/Market Power Effects
A further argument that is often made in the context of faster broadband speeds is that
they may increase competitive forces and reduce market power in some markets.34 This
can happen if transaction costs fall by a sufficiently high amount so that the extent of
the market is effectively enlarged as consumers now find it attractive to purchase from
other firms.
If affected markets are characterised by imperfect competition, then reductions in
transaction costs may bring additional benefits beyond those usually ascribed to falls in
transaction costs in a CBA. For example, suppose are transaction costs in a retail market
served by a single profit maximising monopolist firm. If faster broadband speeds
eliminate these transaction costs, consumer prices will fall. This is a welfare gain.
However, if the reduction in transaction costs opens up the market to a second
competitor, which itself is a monopolist in a different market, then the market
effectively becomes a duopoly. Assuming that there is no collusion between these
duopolists, consumers in both markets will gain further as a result of competition
forcing prices lower.
Since many markets are workably competitive or are already subject to competition
from imported goods, it is unlikely that these additional, second round effects will be
widespread. However, there may be certain industries in which they are important. For
example, in television and video-on-demand markets, faster broadband speeds may
lower transaction costs and intensify competition in the domestic market, as
international firms compete more vigorously against domestic counterparts.
Any effects would depend heavily on assumptions regarding market structure. For
example, if oligopolistic markets are characterised by Bertrand competition rather than
Cournot competition35, then this would have welfare implications. Under Bertrand
34 See, for example, Plum Consulting (2008).
35 In the Cournot model of imperfect competition, firms choose quantities, whilst in the Bertrand model they choose prices. See, for example, Chapter 12 of Mas-Colell, Whinston and Green (1995).
CBA Analytical Framework Page 59 of 95
![Page 60: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/60.jpg)
competition, price may already be bid down to levels close to marginal costs, and the
consumer gains from further competition would be limited.
4. CostsFor the purposes of a CBA of various broadband network infrastructure projects,
estimating the quantifiable social costs of any project will require three important
components: direct costs; the cost of capital; and the opportunity costs of taxation
needed to fund the project. Pricing arrangements may also impose social costs that
should ideally be captured in a CBA. This section explores these issues.
4.1. Direct CostsDirect project costs (which are comprised of construction costs and maintenance costs)
could be estimated from existing engineering and construction data, taking into account
evolution of prices in relevant input markets and likely delays given past experience.
Estimating the future costs of different scenarios – which is vital for a CBA – is a more
complex task. A detailed cost model should be developed to investigate the costs of
different policy options (which may vary the speed of rollout, the extent of any network
coverage, the mix of technologies and so on). Ideally, a “bottom-up” cost modelling
approach should be employed for this purpose, using assumptions regarding exchange
areas, services, lines, contention ratios and so on. This task is likely to require highly
specialised technical knowledge regarding the physical engineering and construction
requirements of different project scenarios. Probabilistic costing – which could be used
to produce costs at P10, P50 and P90 levels (percentiles of the probability distribution) -
is becoming widely used in CBAs. If the probability distribution is reasonably
symmetrical, the P50 (i.e. median) cost could be used as the expected value in a CBA,
and the P10 and P90 costs could be used for sensitivity testing.
CBA Analytical Framework Page 60 of 95
![Page 61: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/61.jpg)
4.2. The Social Discount Rate
Since a HSBB network will typically take many years to construct and will have high
upfront costs, the choice of discount rate – which is used to compare streams of costs
and benefits over time – is crucial. As discussed earlier, risk is a key issue. There is
considerable uncertainty around both benefits and costs, and along multiple dimensions
(geographical/spatial, intertemporal etc.).
As a general rule, there are two ways of discounting streams of uncertain benefits and
costs:
Use expected values in the numerator, but use a higher risk-adjusted discount
rate in the denominator.
Use certainty equivalent values in the numerator, but risk-free discount rate in
the denominator.
For systematic risk and where the assumptions of the CAPM hold, both approaches are
formally equivalent.36 There are a number of ways of choosing a discount rate37, two of
which are:
Descriptive: Under this approach, the discount rate is the opportunity cost of
capital, so one should use financial market interest rates as a guide.
Prescriptive/normative approach: Under this approach, the discount rate is the
welfare-preserving rate of return on saving. This can be derived using ethical
assumptions regarding the weighting of welfare across time and across
generations.
It is important to note that there is little professional consensus among economists
regarding the appropriate choice of discount rate. Even within these two very different
36 For a demonstration of this result, see Trigeorgis (1996).
37 As noted by Harrison (2010, page 41), a common problem with cost-benefit studies is optimism bias, which is the tendency to underestimate costs and overestimate benefits. Altering the discount rate is a relatively poor way of correcting for such downside risks – they are most appropriately accounted for in the estimated of expected values.
CBA Analytical Framework Page 61 of 95
![Page 62: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/62.jpg)
approaches, there is much room for disagreement. The most appropriate way to
proceed in a CBA is to use a range of discount rates, and to conduct sensitivity analysis.
In the present case, given that any government investment may partially or completely
substitute for similar private sector infrastructure investment, the descriptive approach
would seem to be the most appropriate. The discount rate can be computed using
standard techniques, by calculating the project’s weighted average cost of capital
(WACC) using, for example, the capital asset pricing model (CAPM) to determine the
cost of equity. A range of rates should be used and subjected to sensitivity analysis. For
example, one straightforward approach would be to use the three rates recommended
by Infrastructure Australia (4, 7 and 10 percent).
Given the uncertainty around take-up rates and revenue growth, it is appropriate for a
significant risk premium (determined by an appropriately chosen asset beta) to be
added to the risk free rate. Arrow and Lind (1970) argue that if a government project is
small and its returns are uncorrelated with national income, then the social cost of risk
tends to zero as the risk is spread among a sufficiently high number of taxpayers. The
implication is that the correct rate to use for public investment projects is the risk-free
rate. In this case, however, it is highly unlikely that the Arrow-Lind assumptions hold. In
the context of HSBB, any government investment is likely to be relatively large, and its
returns are likely to be positively correlated with national income. In other words, any
public HSBB infrastructure project is likely to have a high level of systematic or non-
diversifiable risk.
below demonstrates the general sensitivity of present values to the choice of discount
rate. The figure plots the present value of $1 that is received in either 10, 15 or 20
years’ time under various discount rates. For example, using a discount rate of 15
percent, $1 worth of benefits that is received in 10 years’ time is worth only $0.25
today, and the same benefit received in 20 years’ time is worth only $0.06 today.
Figure 4.10: The Present Value of $1 Received at Different Points in the Future, for
Different Discount Rates
CBA Analytical Framework Page 62 of 95
![Page 63: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/63.jpg)
$0.00
$0.10
$0.20
$0.30
$0.40
$0.50
$0.60
$0.70
5 6 7 8 9 10 11 12 13 14 15 16
Pres
ent V
alue
of $
1
Discount Rate (percent)
10 years15 years20 years
The choice of discount rate may obviously have important implications for the
qualitative and quantitative results of a CBA, but may also affect the strategy employed
to estimate future benefits and costs. Choosing a higher discount rate means that long
term predictions (i.e. those extending more than 20 years) regarding future take-up
rates and the benefits from applications, services, products and devices that have not
yet been widely adopted (or have not yet been invented) receive a very low weight –
with a high discount rate, any future benefits received from these end-use products are
simply not worth very much in today’s dollars. On the other hand, if a low discount rate
is used, then benefits received in the future become more valuable in today’s dollars,
and greater accuracy in the estimation of future benefits becomes more important.
CBA Analytical Framework Page 63 of 95
![Page 64: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/64.jpg)
4.3. The Timing of Investment and Option Values
4.3.1. Dynamically Efficient Investment
Appendix 6 develops a simple model of the optimal construction of an infrastructure
project over time under conditions of certainty. The main point of that analysis is that
conceptually there is an optimal path of investment in which construction occurs
incrementally – it is almost never optimal to invest all resources immediately. Indeed, it
is straightforward to construct examples in which the optimal investment path is for
investment in capacity to steadily ramp upwards over time, and for most investment to
occur later rather than sooner. The factors that affect the optimal path of investment in
an environment of complete certainty include the discount rate, the depreciation rate,
and the incremental costs and benefits of investment at any point in time. In an
uncertain environment, risk factors will also affect the optimal path.
For the purposes of a CBA of construction of a HSBB network, the main implication is
that the investment paths of some infrastructure projects will more closely resemble the
dynamically efficient path than others, and that the costs of inefficiently timed
investment should ideally be accounted for in a CBA.
4.3.2. Option Values
High speed broadband networks possess features of experience goods: the
characteristics and quality (marginal benefit) of an experience good are uncertain, costly
to determine, and cannot be determined precisely prior to consumption. A common
concern with such goods is that producers will have an incentive to misrepresent the
quality of the good, with consumers (including governments) paying excessively high
prices for goods that, before consumption, are perceived to possess high marginal
benefits but which, after consumption, turn out to have a relatively low marginal value.
Markets have developed a number of mechanisms to deal with the potential
inefficiencies that may be associated with experience goods. For example,
intermediaries may provide specialised verification services, examining the quality of
various goods in exchange for a payment from consumers. Firms may establish
CBA Analytical Framework Page 64 of 95
![Page 65: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/65.jpg)
reputational capital as a commitment device, signaling to consumers that if they are
detected “cheating” (selling a low quality experience good for a high price), they will
incur a high cost as this capital is quickly eroded.
Consumers can also take steps to mitigate their exposure to the risks associated with
experience goods by deliberately delaying their consumption (or by limiting
consumption to smaller amounts) until more information about the incremental benefit
of the good becomes available and uncertainty is partially or fully resolved.
Investment in HSBB networks is largely irreversible, with uncertain costs and benefits.
Uncertainty is likely to be only gradually resolved over time, as technology evolves and
end-user applications develop. Construction of a HSBB network therefore involves
purchasing and exercising a series of embedded real options. For example, the decision
to roll out infrastructure in a particular geographic area when demand is unknown
involves exercising a real option. These options have value, and this value should ideally
be accounted for in any CBA.
If consumers are uncertain about future value of applications, they may value the option
of having faster speeds made available even if they do not immediately take up the
service. For a given level of uncertainty about future benefits, the gain to consumers
from possessing this option is higher, the higher their current network usage and
current benefits. However, given that marginal benefits are declining, and given the
likely distribution of preferences for faster speeds, the overall value of this option may
be low.
On the other hand, given the irreversible nature of infrastructure investment and the
uncertainty associated with the development of new and improved technologies, there
may be a benefit to delaying investment. There is an economic tradeoff involved in
directly adopting a “target” technology on the one hand, versus adopting a
technologically inferior but less costly intermediate technology on the other. All else
being equal, an infrastructure project which allows for construction flexibility and keeps
these supply side real options open is more likely to be superior to a project that does
not.
CBA Analytical Framework Page 65 of 95
![Page 66: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/66.jpg)
4.4. Taxation
4.4.1. General Considerations
In any CBA it may be important to consider the treatment of goods that are subject to
taxes and subsidies. In particular, if the goods whose demand is significantly affected by
faster broadband speeds are taxed or subsidised, then changes in tax revenue should
also be taken into account in any CBA. In particular, an increase in tax revenue (or
reduction in subsidy payments) induced by changes in demand represents a welfare
gain, whilst a fall in tax revenue (or an increase in subsidy payments) represents a
welfare loss.
For most products these effects will be relatively small and may be safely ignored. In
other cases these tax interaction effects may be important. For example, if an increase
in broadband speeds leads to a significant increase in demand for online retail goods
[some or all of which may not be subject to Australia’s goods and services tax (GST)] and
a shift away from purchases in more traditional “bricks and mortar” retail stores, then
tax revenue may be lower than it otherwise would be, and this would represent a
welfare loss. Similarly, if an increase in broadband speeds leads consumers to reduce
their labour supply (by inducing individuals to undertake more leisure as they spend
more time on the internet or using entertainment applications ), then this would
exacerbate the effect of existing labour market distortions, reduce income tax revenue
and should be counted as a welfare loss.
The basic idea is illustrated in Figure 4.11, where the increase in broadband speed again
increases demand for good A, but reduces demand for good B. In this example, good B
is now assumed to be subject to an existing tax, with revenue returned to consumers in
a lump sum fashion. The reduction in demand for B reduces the amount of revenue
collected by the tax on B (this is the shaded area), and since this reduction in revenue
was not offset anywhere by a reduction in tax rates, it represents a pure welfare loss
which must be added to the sum of the changes in consumer surpluses in each market.
Similar considerations apply to goods whose consumption may be subject to positive or
negative spill overs (i.e. externalities). For example, if an increase in speeds leads to an
CBA Analytical Framework Page 66 of 95
![Page 67: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/67.jpg)
increase in demand for goods which are associated with an unpriced (or under-priced)
negative external effect, then the change in demand multiplied by the size of the
marginal externality must be subtracted from the change in consumer surplus in the
relevant market to compute the welfare effect in that market.
An increase in demand for goods which are associated with unpriced positive
externalities (e.g. education) would need to be treated in a similar way. Note, however,
that if such goods are already appropriately taxed or subsidised then there are no
additional welfare effects to consider beyond changes in private consumer surplus.
Thus, for example, if health and education are already subsidised optimally (i.e. to the
point where marginal social benefits equal marginal social costs – an assumption which
would need to be verified), then the fact that more of these goods may be demanded as
a result of faster broadband speeds would have no overall welfare consequences,
beyond the changes in private consumer surpluses that come about as a result of those
changes.
Figure 4.11: Welfare Effects of a Change in Demand When an Existing Good is Taxed
0AP
1Ax
1Bx
1
1
BDAD
Market A Market B
0Bx
APBP
0BP
0
0
0Ax
'BD'AD
ACSBCS
0B BP t
4.4.2. The Marginal Cost of Public Funds
A basic principle of welfare economics is that the direct resource costs of government
spending should be grossed up by the marginal cost of public funds (MCF) – i.e. the costs
CBA Analytical Framework Page 67 of 95
![Page 68: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/68.jpg)
of taxation – to determine the full social cost of the project, which is the reduction in
private surplus that occurs when the project goes ahead. The Department of Finance’s
Handbook of Cost-Benefit Analysis states that:38
“The excess burden of taxation means the supply cost of public investment or services is
greater than the actual amount of funds used. Consequently, it is appropriate to make
an upward adjustment to financial costs in cost benefit analysis to ensure the calculated
net present value is unbiased. This is, however, only where there is a significant net cost
to the budget. It excludes cases where costs are fully recovered (such as where there is a
user charge) or the resources are already committed (which is effectively so for cost-
effectiveness analysis and lease-purchase analysis).”
Ignoring the MCF can have deleterious welfare consequences. Consider Figure 4.12
below, which plots the marginal benefits and marginal costs of government spending. In
the example, marginal costs are assumed to be constant.
Figure 4.12: The Marginal Cost of Funds and the Optimal Provision of a Publicly
Provided Good
MB
MC*MCF
MC
Quantity
Marginal Benefit, Marginal Cost
Welfare Loss if DWL of Taxation is Ignored
G* GOptimal Provision
38 <http://www.finance.gov.au/publications/finance-circulars/2006/docs/Handbook_of_CB_analysis.pdf >.
CBA Analytical Framework Page 68 of 95
![Page 69: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/69.jpg)
The curve is the marginal resource costs MC, grossed up by the costs of
taxation, and this represents the social opportunity cost of government provision. This
curve has a convex shape, as tax rates are raised higher to fund greater expenditure.
The optimal provision here is G*, where marginal benefits equate with marginal
opportunity costs. Ignoring the costs of taxation would result in overprovision of G,
which is a point at which marginal benefits are less than marginal opportunity costs.
Ignoring the costs of taxation leads to the welfare loss triangle indicated in the diagram.
Appendix 5 derives the rule of thumb that is applied in the simplest cases.
A range of estimates of the MCF may be applied in the present case. For example, the
Australia’s Future Tax System (AFTS) Review estimated an MCF of 1.24 for labour
income taxes.39 Ideally the MCF should be one of the items subjected to sensitivity
analysis to determine its effect on the final CBA results.
Although financial payments to third parties are transfers between government and the
private sector and net out in welfare terms, the distortionary costs of the taxes that are
levied to raise those funds should be counted as a welfare cost. Similar considerations
apply to project revenues – they are transfers between the private and public sector,
but in principle they could be used to reduce other taxes and so revenues should be
grossed up by the MCF. In summary, net profits (or losses) should be grossed up by the
MCF.
4.4.3. Taxation and Uniform National Wholesale Access Pricing
Another important aspect of taxation that should be taken into account is the implicit
tax and subsidy arrangement that any uniform national wholesale access pricing may
entail. Imposing a uniform price means that downstream providers (and final retail
consumers) in low cost areas will effectively pay a tax, the revenue from which is used to
subsidise consumers in high cost areas. This implicit tax has a number of important
implications for the welfare analysis of various scenarios.
The basic idea is shown in below, where for simplicity it is assumed that there are two geographically separated markets, A and B. The access provider has different marginal costs in each market, labeled cA and cB. There is a uniform price of P that is charged.
39 Jones (2010) and Robson (2005) survey estimates of the MCF.
CBA Analytical Framework Page 69 of 95
![Page 70: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/70.jpg)
Consumption in each market is . This arrangement leads to welfare losses in each market, with the welfare effects increasing non-linearly with the size of the implicit tax.40
Note that even though there may be a case for the firm setting price above marginal
cost in market A in order to recover its fixed costs, the prices as drawn in this example
cannot be second best efficient. Prices could be reduced in market A and increased in
market B so as to hold profits constant and reduce the aggregate welfare loss. Indeed,
the standard Ramsey pricing rule applies to such a situation, and states that prices
should be set above marginal costs in each market so that there is an equal percentage
reduction in each market below the first best consumption levels . In this case,
such a second best optimum would necessitate higher prices in market B, and lower
prices in market A (even though market A has a relatively low elasticity of demand at the
first best consumption point ).
40 It is important to note that uniform pricing is also likely to have dynamic deadweight losses as well, by influencing technology choices and distorting the mix of technology that is adopted in urban and regional areas.
CBA Analytical Framework Page 70 of 95
![Page 71: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/71.jpg)
Figure 4.13: Cross-Subsidies as an Implicit Tax-Subsidy Scheme
Ac
1Ax
1Bx
1 1
BDAMB
Market A Market B
0Bx
AP BP
0
0
0Ax
Bc
P P
4.5. Non-Quantifiable Benefits and Costs
As discussed earlier, ideally a CBA should place dollar values on all sources of costs and
benefits. However, there may be benefits and costs which are difficult to estimate
accurately, or which are simply non-quantifiable. For example, consumers and
producers may (due to, for example, a lack of information) systematically underestimate
(or overestimate) the likely impact that faster broadband speeds may have on future
innovations and the value of new applications, services and products (including, for
example, e-health and e-education applications). If these systematic biases are present
across a sufficient number of consumers and producers, then estimates of private
willingness to pay which rely on information provided by these groups may not correctly
and/or fully reflect relevant benefits. Other benefits – such as the benefit from being
able to access the “knowledge commons” of the global Internet (or, more relevantly for
policy, the effects of faster speeds on these benefits) – may be difficult or impossible to
quantify. Similarly, there may be significant non-quantifiable social costs associated
with broader network coverage and faster speeds, including issues related to privacy,
data security, national defence and security, intellectual property protection, net
CBA Analytical Framework Page 71 of 95
![Page 72: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/72.jpg)
neutrality, cyber-crime and freedom of speech. These are clearly important policy
issues in their own right, and the extent to which they are affected by various
investment options deserves to be discussed and explored qualitatively in a CBA
5. Conclusion and Recommendation for Overall CBA Strategy
This paper has developed an analytical framework to assist the panel of experts in its
development of a CBA of public infrastructure investment in HSBB networks. The paper
has provided some “broadband-specific” conceptual guidance for a CBA by exploring the
main sources of the potential economic impacts of access to faster broadband speeds,
as well as providing guidance on how to estimate these impacts.
The paper has demonstrated that there are a number of significant challenges that need
to be overcome when conducting a CBA of policy options regarding public investment in
HSBB networks. The main challenges are the specification of the baseline scenario and
the significant amount of uncertainty surrounding benefits, as well as the estimation of
benefits.
In practice a range of theoretical assumptions and empirical compromises will need to
be made. The key to developing a robust CBA will be to test the sensitivity of the results
to these assumptions and compromises, in order to clearly demonstrate to decision
makers how they affect the CBA’s final results.
Given that the ultimate purpose of a CBA is to inform policy decisions, the most
straightforward way to proceed would be begin with a (relatively) simple partial
equilibrium analysis of the effects of various policy options on network coverage, speed
and incremental net benefits. Under this approach, the analysis would begin with the
simplest model that is fit for purpose: social benefits would be assumed to be fully
captured by private willingness to pay, and estimates of consumer valuation could be
compared with direct project costs. Factors such as externalities and other effects
discussed in this paper could be temporarily put to one side.
Once a partial equilibrium analysis is completed, some of the more conceptually difficult
issues explored in this paper could be dealt with. In the first instance, these factors
CBA Analytical Framework Page 72 of 95
![Page 73: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/73.jpg)
could be dealt with qualitatively, by exploring the extent to which a more complicated,
general equilibrium analysis would increase the economic value of various policy
options, and whether such effects are likely to vary greatly across options. For example,
it may be the case that the presence of network externalities would increase the net
benefits of all policies by a similar magnitude, leaving unchanged the overall ranking of
policies that emerged from a simple partial equilibrium analysis. Alternatively, if a
simple partial equilibrium analysis suggests that the net benefits of certain policies are
likely to be negative, a qualitative analysis could be used to show how positive the
benefits from other sources would need to be in order to reverse the results of the
simpler modelling exercise.
CBA Analytical Framework Page 73 of 95
![Page 74: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/74.jpg)
6. ReferencesArrow, K. and Lind, R. (1970) “Uncertainty and the Evaluation of Public Investment
Decisions”, American Economic Review, 60(3), pp. 364–378.
Basu, A., Mazumdar, T. and Raj, S. (2003) “Indirect Network Externality Effects on
Product Attributes”, Marketing Science, 22(2): 209–221.
Bateman, I. et al (2007) Environmental Valuation: Revealed Preferences and Welfare
Economics: A Theoretical Guide to the Empirical Literature, Dordrecht: Springer.
Becker, G. S. (1965), “A theory of the allocation of time”, The Economic Journal, 75(299),
pp. 493–517.
Becker, G. and Murphy, K. (2000) Social Economics, Cambridge, MA: Bellknap Press.
Boardman, A. et al (2010) Cost Benefit Analysis, Fourth Edition, New York: Prentice Hall.
Brock, W. and Durlauf, S. (2001) “Discrete Choice with Social Interactions”, Review of
Economic Studies, 68(2): 235–260.
Brueckner, J. (2011) Lectures on Urban Economics, Cambridge, MA: MIT Press.
Deloitte Access Economics (2013) Benefits of High Speed Broadband for Australian
Households, Report for the Commonwealth of Australia as represented by the
Department of Broadband, Communications and the Digital Economy, Canberra.
Department of Communications (2013) Broadband Availability and Quality, Canberra,
December.
Dettling, L. (2013) “Broadband in the Labor Market: The Impact of Residential High
Speed Internet on Married Women’s Labor Force Participation,” Staff Working Paper,
Finance and Economics Discussion Series, Divisions of Research & Statistics and
Monetary Affairs, Federal Reserve Board, Washington, D.C.
http://www.federalreserve.gov/pubs/feds/2013/201365/201365pap.pdf
CBA Analytical Framework Page 74 of 95
![Page 75: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/75.jpg)
Dobes, L. and Bennett, J. (2009) “Multi-Criteria Analysis: ‘Good Enough’ for Government
Work?”, Agenda: A Journal of Policy Analysis and Reform, 16(3): 7–30.
Dutz, M., Orszag, J. and Willig, R. (2009) The Substantial Consumer Benefits of
Broadband Connectivity for US Households, Paper Commissioned by the Internet
Innovation Alliance, July.
Ergas, H. (2009) “In Defence of Cost-Benefit Analysis”, Agenda: A Journal of Policy
Analysis and Reform, 16(3): 31–40.
Ergas, H. and Robson, A. (2009) “The Social Losses from Inefficient Infrastructure
Projects: Recent Australian Experience”, Productivity Commission Roundtable on
Strengthening Evidence-Based Policy in the Australian Federation, Canberra, August.
Gandal, N. (2008) “Network Goods (Empirical Studies)”, in Durlauf, S. and Blume, L. (eds)
The New Palgrave Dictionary of Economics, Second Edition, London: Palgrave Macmillan.
Glaeser, E. (2008) Cities, Agglomeration and Spatial Equilibrium, Oxford: Oxford
University Press.
Goolsbee, A. and P. J. Klenow (2006) “Valuing Consumer Products by the Time Spent
Using Them: An Application to the Internet”, American Economic Review, 96(2), pp.
108–113.
Harberger, A. (1998) “A Vision of the Growth Process”, American Economic Review,
88(1): 1-32.
Harrison, M. (2010) Valuing the Future: The Social Discount Rate in Cost Benefit Analysis,
Visiting Researcher Paper, Productivity Commission, Canberra.
Hayes, R. (2011) Valuing Broadband Benefits: A Selective Report of Issues and Options,
Version 1.1, Institute for a Broadband-Enabled Society, University of Melbourne.
Jackson, M. (2010) Social and Economic Networks, Princeton: Princeton University Press.
Johansson, P. (1987) The Economic Theory and Measurement of Environmental Benefits,
New York: Cambridge University Press.
CBA Analytical Framework Page 75 of 95
![Page 76: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/76.jpg)
Kenny, R and T Broughton (2013), Domestic Demand for Bandwidth, An approach for
forecasting requirements for the period 2013-2023, UK Broadband Stakeholder Group.
Liebowitz, S. and Margolis, S. (1995) “Are Network Externalities a New Source of Market
Failure?”, Research in Law and Economics, 17: 1-22.
Little, I. M. D. and J. A. Mirrlees (1994) “The costs and benefits of analysis: project
appraisal and planning twenty years on”, in Layard, R. and S. Glaister (eds.), Cost Benefit
Analysis, 2nd ed., Cambridge University Press, Cambridge [England]; New York, NY.
Livingston, J. , Ortmeyer, D., Scholten, P. and Wong, W. (2013) “A Hedonic Approach to
Testing for Indirect Network Effects in the LCD Television Market”, Applied Economics
Letters, 20: 76–79.
Mas_Colell, A. Whinston, M. and Green, J. (1995) Microeconomic Theory, New York:
Oxford University Press.
McFadden, D. (1999) “Computing Willingness-to-Pay in Random Utility Models”, in
Melvin, J. et al (eds) Trade, Theory and Econometrics, New York: Routledge.
Meade, N. and Islam, T. (2006) “Modelling and Forecasting the Diffusion of Innovation: A
25 Year Review”, International Journal of Forecasting, 22: 519–545.
Pearce, D. (1983) Cost Benefit Analysis, Second Edition, London: Macmillan.
Pearce, D. and Nash. C. (1981) The Social Appraisal of Projects, New York: Halsted Press.
Plum Consulting (2008) A Framework for Evaluating the Value of Next Generation
Broadband: A Report for the Broadband Stakeholder Group, London.
Quiggin, J. (2013) “National Accounting and the Digital Economy: The Case of the
National Broadband Network”, Colin Clark Memorial Lecture, Brisbane.
Robson, A. (2012) Law and Markets, London: Palgrave Macmillan.
Rosen, S. (2002) “Markets and Diversity”, American Economic Review, 92 (1): 1–15.
Rosston, G., Savage, S, and Waldman, D. (2010) “Household Demand for Broadband
Internet Service”, Final report to the Broadband.gov Task Force, Federal
Communications Commission.
CBA Analytical Framework Page 76 of 95
![Page 77: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/77.jpg)
Samuelson, P. (1954) “The Pure Theory of Public Expenditure,” Review of Economics and
Statistics, 36(4): 387-389.
Segal, I. (1999) “Contracting with Externalities”, Quarterly Journal of Economics, 114(2):
337–388.
Shapiro, C. and Varian, H. (1999) Information Rules: A Strategic Guide to the Network
Economy, Cambridge, MA: Harvard Business School Press.
Shy, O. (2001) The Economics of Network Industries, Cambridge: Cambridge University
Press.
Small, K. and Verhoef, E. (2007) The Economics or Urban Transportation, London:
Routledge.
Small, K. and Rosen, H. (1981) “Applied Welfare Economics with Discrete Choice
Models”, Econometrica, 49(1): 105–130.
Spulber, D. and Yoo, C. (2009) Networks in Telecommunications: Economics and Law,
Cambridge: Cambridge University Press.
Stockman, A. (1996) Introduction to Economics, New York: Dryden Press.
Trigeorgis, L. (1996) Real Options: Managerial Flexibility and Strategy in Resource
Allocation, Cambridge: MIT Press.
Triplett, J. (2006) Handbook on Hedonic Indexes and Quality Adjustments in Price
Indexes: Special Application to Information Technology Products, Paris: OECD Publishing.
Varian, H., Farell, J. and Shapiro, C. (2004) The Economics of Information Technology: An
Introduction, London: Cambridge University Press.
Willig, R. (1978) “Incremental Consumer’s Surplus and Hedonic Price Adjustment”,
Journal of Economic Theory, 17: 227–253.
CBA Analytical Framework Page 77 of 95
![Page 78: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/78.jpg)
Appendix 1. The Indirect Demand for Faster Broadband Speeds
This appendix considers some of the analytical foundations of the value of faster
broadband speeds. Consider Willig’s (1978) approach to valuing changes in the
attributes of goods.41 Let be a consumer’s indirect utility when prices of goods
are , income is y and broadband speed is s. That is:
In this formulation, faster broadband speeds are assumed to affect consumer decisions
by altering the marginal utility from consuming different goods and services. The effect
of faster speeds is isolated by deliberately assuming that prices and incomes are held
fixed at current levels. The welfare gain from being able to access a network with a
greater speed of (compared with the current network speed of ) is then simply
.
At the consumer’s optimum, using the envelope theorem at the margin, a small or
incremental change in speeds gives a change in consumer welfare of:
and so:
It is possible to express marginal changes more conveniently by noting that:
where is the marginal utility of income, is the current price of good i, and is the
price for which (so that changes in speed do not affect demand if prices exceed
41 See also Johansson (1987), Chapter 6.
CBA Analytical Framework Page 78 of 95
![Page 79: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/79.jpg)
this level), and where the last equality uses Roy’s identity42 (which relates the indirect
utility function to the ordinary demand curve for the good).
As a technical aside, the expression in can be derived more precisely by replacing
ordinary demand curves with their compensated counterparts. If is the
compensated demand curve for good i at the initial level of utility, then the welfare
effect of a change in Internet speeds can be measured by the amount of money that,
given the new speeds, could be taken away from the consumer to restore them to their
initial utility level, which is .
Alternatively, the welfare effect could be measured by the amount of money that, given
current speeds, the consumer would be willing to pay for a level of wellbeing that they
would have received with faster speeds. This is . In
the absence of income effects, both of these expressions are equivalent to .
42 Roy’s identity links changes in consumer welfare (brought about by price changes) to the ordinary
demand for the good. Specifically, it states that - the change in a consumer’s utility with
respect to a small change in prices is equal to the negative of the marginal utility of income, multiplied by the level of ordinary demand for the good. If the marginal utility of income is constant, then the area to the left of an ordinary demand curve is an exact measure of welfare change. See, for example, Chapter 3 of Mas-Collell, Whinston and Green (1995)
CBA Analytical Framework Page 79 of 95
![Page 80: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/80.jpg)
Appendix 2. The Simple Welfare Economics of the Gains from Productivity
Improvements
This appendix develops a very simple high level model to illustrate the static economic
gains from an economy-wide total factor productivity (TFP) improvement and the effect
on labour supply and demand. Suppose that the aggregate production function is
, where L is labour input, A is a total factor productivity parameter, and F() is
a strictly increasing, strictly concave production function. Utility is a function of
consumption and leisure, so . All output is consumed. The economy faces a
time constraint of .
Substituting these constraints into the utility function gives:
By the envelope theorem:
or:
This states that, in a static setting, the marginal welfare gain from an improvement in
productivity is equal to the marginal utility of consumption, multiplied by the current
level of output, multiplied by the percentage increase in productivity. Note that this
expression, which quantifies the benefit from an improvement in productivity, does not
depend on the number of jobs that are created as a result of the productivity shock i.e.
whether the improvement in productivity increases or decreases the quantity of labour
employed. Indeed, depending on the relative size of substitution and income effects,
the equilibrium quantity of labour could fall. In this case, the expression in can still be
used to quantify the benefits from an increase in TFP.
CBA Analytical Framework Page 80 of 95
![Page 81: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/81.jpg)
Appendix 3. Welfare Analysis in Discrete Choice Models
The approach outlined in section 3.6.2 can be used to directly estimate willingness to
pay for faster broadband speeds. A standard approach in the literature is to assume
that the random component of indirect utility is additive, so that:
The probability that alternative j is chosen is then simply:
This probability depends on the assumed distribution of the random variable .
Since this is a random utility model, it is necessary to work with expected values, that is,
we need to find:
and then adjust these for changes in characteristics. As a general rule, it is not possible
to easily compute numerical solutions for this expression. To numerically estimate the
willingness to pay for different package characteristics (such as speed), it is necessary to
make choices about the distribution of the random error terms and the functional form
of the indirect utility function.
For example, a multinomial logit model assumes that the disturbance term follows a
Gumbel (or type 1 extreme value distribution):
Under this assumption, the differences follow a logistic distribution:
If a linear function for the observable characteristics of the utility function is assumed,
then:
CBA Analytical Framework Page 81 of 95
![Page 82: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/82.jpg)
Then the probability that alternative j is chosen is:
For the extreme value distribution, it is:
where C is an undetermined constant.
The compensating variation of the change in package characteristics from to is
defined implicitly by:
For the linear functional form, it is:
so that the actual CV is:
CBA Analytical Framework Page 82 of 95
![Page 83: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/83.jpg)
Appendix 4. The Dynamic Efficiency of S-Shaped Product Diffusion Curves
In the product diffusion literature there is often little connection between “S-shaped”
take-up paths and the standard economic theory of consumption and welfare. This
appendix examines the conditions under which such a take-up path is economically
efficient. The implication of the analysis is that by making an assumption about the path
of take-up or adoption, a CBA is also implicitly making an assumption about the
evolution of consumption benefits over time. Assumptions about one cannot be made
independently of the other, and they should be consistent with each other.
The approach is a simple consumption/wealth accumulation model that is standard in
the literature. Let c be the take-up rate, where 0<c<1. Let w be wealth. Let be the
social rate of time preference, and let r be the interest rate on wealth accumulation.
Consider the problem:
subject to:
and:
The utility function u(c) can be interpreted as benefits of take-up, or a social planner’s
objective function, i.e. a social welfare function over aggregate consumption (the take-
up rate). The purpose is to seek to find benefit or social welfare functions for which the
following patterns of consumption are optimal:
Where . Note that if then the Gompertz function applies:
CBA Analytical Framework Page 83 of 95
![Page 84: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/84.jpg)
It is possible to use standard techniques (i.e. Pontryagin’s maximum principle) to solve
this problem. The current value Hamiltonian is:
where is the current value costate variable. The necessary first order conditions are:
From it is:
whilst gives:
Hence:
or:
where:
is the elasticity of the marginal utility of consumption. This is a standard result in the
literature. To match with apply:
CBA Analytical Framework Page 84 of 95
![Page 85: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/85.jpg)
Hence if:
With then the generalised logistic function in , will be optimal.
For the specialised Gompertz case:
This gives:
where is an arbitrary constant and is a negative number. This has therefore
established that the Gompertz product diffusion curve is an optimal consumption path if
(and only if) the aggregate benefits of take-up obey .
CBA Analytical Framework Page 85 of 95
![Page 86: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/86.jpg)
Appendix 5. A Simple Model of Network Effects
Consider a simple model with a continuum of consumers. Their individual valuations
from accessing broadband are . The formula below allows this individual valuation
to depend on network speed.
Assume that there is a network externality present of where I is the number of
users on the network, and, in the case of positive direct marginal network
effects, for all I.
Total welfare is the sum of valuations, less cost:
where for simplicity it is assumed that marginal costs of network capacity are constant
and that the scaling effect of network externality is the same for all users, which means
that users with a higher valuation gain a higher absolute level of utility, the more users
there are on the network.
Efficiency requires that:
The first term is the effect on the marginal consumer’s own welfare when he joins the
network. The second term is the direct marginal network externality – it is the change
in the size of the overall network effect when the marginal consumer joins multiplied by
the effect of this on all existing inframarginal users.
Clearly, if network externalities in this environment are ignored and only the individual
WTP component is measured, then problems would arise for a CBA. Total welfare from
the network would be incorrectly measured as:
And the CBA would advocate building additional network capacity as long as:
CBA Analytical Framework Page 86 of 95
![Page 87: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/87.jpg)
Since the left hand side of is less than the left hand side of when , then a CBA
which ignored network externalities in such an environment could give incorrect results.
However, there are two mitigating factors. First, if there are network effects present
and individuals are asked to report their willingness to pay, then they will report
, where I is the size of the network that they expect, even if that is not explicitly
stated. This could be larger or smaller than the efficient network size. Hence, the
extent of the error may not be as large as would be implied by the earlier discussion.
Second, discrete choice econometric methods could be used to estimate the size of
network effects, and these could then be incorporated into the CBA.
Now consider a market environment. Suppose that there is a price of p that is charged
to access the network. To compute the equilibrium network size, first compute the
demand curve. Suppose that I users are on the network. Then the marginal user is
indifferent between joining and not joining:
This condition defines the demand curve for the good.43 Note that there may be
multiple equilibria depending on the strength of the network effect. In a perfectly
competitive market, users will join up until the point where marginal valuation equals
marginal cost:
Since it is assumed that , it is clear that, in a competitive market, the size of
the network will be inefficiently small.
Note, however, that this result depends on a number of assumptions. In particular, it
assumes that for all I. By way of contrast, suppose that at the optimum,
(i.e. there are negative marginal network effects – or marginal congestion
effects – at the optimum). Then the second term in is negative, and in a competitive
market the size of the network could be too large.
43 See, for example, Varian (2010), Chapter 35.
CBA Analytical Framework Page 87 of 95
![Page 88: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/88.jpg)
Maximum Network Speed
To examine the effect of a uniform increase in maximum network speeds in this model,
again assume that there is a direct network externality present, but that valuations
depend on the maximum available speed of access, labelled s. Assume that the
marginal cost of building network capacity also depends positively on s. Finally, assume
that the maximum is the same for all individuals on the network.
Total welfare is now:
Suppose that the network capacity has been optimally chosen. Then, by the envelope
theorem and Leibniz’s rule, it is socially beneficial to increase speed up until the point
where:
On the other hand, in a competitive market environment, increases in maximum speed
will be provided as long as the marginal consumer is willing to pay for them. Hence we
must have:
The question of whether the market will provide speeds that are too fast or too slow in
the presence of network externalities depends on a comparison of and . That is, market
maximum speeds will be inefficiently slow if:
If the marginal network externality is everywhere positive, then: . The
question is therefore whether:
CBA Analytical Framework Page 88 of 95
![Page 89: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/89.jpg)
This involves a comparison of the marginal willingness to pay for higher maximum
speeds of the marginal consumer in a competitive market, versus the average marginal
willingness to pay of the entire population assuming that an efficient network capacity.
It is not at all obvious that the right hand side of should be larger than the left hand
side.
Moreover, if the network is subject to congestion effects, matters are even less clear. If
there are too many network users in a market setting and the network becomes
congested, then it is a second best environment, and firms may try to offset some of
these congestion effects and compete for customers by offering inefficiently fast
maximum download speeds.
CBA Analytical Framework Page 89 of 95
![Page 90: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/90.jpg)
Appendix 6. The Dynamic Efficiency of Investment
Consider the following simple model of investment. There is an infrastructure project
which builds up a stock of capital of over time, and yields a flow of benefits of
at each instant, with . Capital is built up via investment,
but depreciates at a rate . Adding to the capital stock has a cost of , where I is
the amount of investment. Assume that and . Future net benefits
are discounted at the rate of r.
The purpose is to seek the path of investment that maximises the present value of the
flow of net benefits. In other words, to solve:
subject to:
and:
This is an optimal control problem which can be solved and analysed using standard
techniques.
The current value Hamiltonian is:
where is the current value of the co-state variable or the shadow price of capital. The
first order necessary conditions for this problem are:44
44 The Hessian of the Hamiltonian is jointly concave in I and K, and so the first order conditions are also sufficient.
CBA Analytical Framework Page 90 of 95
![Page 91: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/91.jpg)
Expression states that investment in each period should occur up to the point where
the marginal cost of investment equals the shadow price of capital. Expression states
that the shadow price of capital changes over time according to:
These expressions can be used to qualitatively examine the optimal path of investment
using phase diagram analysis. Using such an approach, it is straightforward to
demonstrate there are instances where it is optimal for investment to begin at a low
level and increase over the life of the project.
CBA Analytical Framework Page 91 of 95
![Page 92: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/92.jpg)
Appendix 7. The Costs of Taxation
This section derives a commonly used rule for accounting for the costs of taxation in the
supply of publicly provided goods. Let be the direct resource cost of supplying G,
which may be a private or public good supplied by the government. The government’s
budget is assumed to be balanced, so that:
where R is tax revenue and is the ordinary demand for good i, and ti is the tax rate on
good i.
The government chooses the level of G and the vector of tax rates t in order to
maximise indirect utility subject to . The Lagrangean is:
The first order necessary conditions are:
and:
for each j = 1,…,n.
Dividing by yields:
CBA Analytical Framework Page 92 of 95
![Page 93: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/93.jpg)
Finally, dividing both sides by the marginal utility of income and applying Roy’s identity:
The left hand side is the social marginal benefit of an additional unit of the government
supplied good, expressed in dollar terms. The right hand side can be decomposed as
follows. Recall that the marginal excess burden per dollar of revenue is:
Hence, the first term on the right hand side, , is one plus the absolute value
of the marginal excess burden, or the marginal cost of public funds of MCF for tax ti. The
expression in implies that taxes should be adjusted until the MCFs for each tax are
equalised.
The term inside the brackets on the right hand side of is therefore the marginal cost of
G, minus the effect of higher G on revenue from goods that are already taxed. This
needs to be taken into account because as G increases, consumption of goods that are
complementary to G will rise, and to the extent that these are taxed this will raise
revenue that the government would not have otherwise received.
On the other hand, consumption of goods that are substitutes for G will fall, and this will
reduce tax revenue. Hence the term (which is known as the “spending
effect”), may be positive or negative. If a project reduces labour supply, then the
spending effect is negative, which means that all else being equal, benefits must be
higher in order to justify the project. The opposite conclusion applies if the project
increases labour supply.
CBA Analytical Framework Page 93 of 95
![Page 94: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/94.jpg)
Ignoring this spending effect, the revised Samuelson (1954) rule is to provide G up to the
point where:
If taxes are not distortionary, then and the conventional Samuelson (1954)
rule applies:
For discrete projects or policies, a discrete version of can be used. Suppose that there is
no spending effect. Consider a government spending proposal which yields social
incremental benefits of B for each additional unit of G that is provided.
Suppose that each unit of the project has a price of P, which is equal to its direct
resource cost. Hence the budgetary cost or expenditure is P×G.
The project must be funded by distortionary taxation. Let the tax rate be t. As far as the
private sector is concerned, the cost of this taxation is the foregone gains from trade;
that is, the lost producer and consumer surplus. This in turn is equal to the sum of tax
revenue plus the deadweight loss of the tax. The change in these costs (i.e. the marginal
cost) is therefore R+DWL.
Now suppose that taxes must increase by a very small amount in order to fund an
additional unit of G. Government tax revenue increases by R. This allows additional
amount of the public good of G = R/P. This additional amount yields benefits of B
×G = B × R/P.
Hence the additional unit of G should be provided if the incremental benefits B ×G
exceed the incremental costs:
or:
CBA Analytical Framework Page 94 of 95
![Page 95: NBN CBA Analytical Framework (Final 16 May … · Web viewConsider a government spending proposal which yields social incremental benefits of (B for each additional unit of G that](https://reader036.vdocuments.site/reader036/viewer/2022070610/5aeb80567f8b9ae5318dd467/html5/thumbnails/95.jpg)
An additional dollar of government spending is justified on efficiency grounds if the
marginal benefit of that spending exceeds the price or direct cost, grossed up by the
marginal cost of public funds, which depends on the deadweight costs of taxation. The
term in is the marginal deadweight loss per dollar of revenue. The MCF is
equal to one plus the marginal deadweight loss per dollar of revenue.
CBA Analytical Framework Page 95 of 95