Guest Lecture: Prof Peter Forsyth - Evaluating Investments: CBA or

advertisement
Evaluating Investments – CBA or CGE?
Peter Forsyth
Monash University
NZ Treasury
easu y Se
Seminar
a
September 6 2011
1
Problem
• CGE models are now used often to
assess whether investments are
worthwhile
• The traditional technique is to use CBA
• But which should be used?
• Which is correct?
2
• Both
3
Argument
• Properly specified, both should give the
same answer – i.e.
i e no inconsistency
• But there are limitations with both
techniques
• But these are complementary
• Possible to use both to improve
investment evaluation
• (If that is what decision makers want)
4
Theme
• Recognising the different roles of the
two
• Use the two to improve evaluation
• Reconciling the two obtain consistent
results
5
Agenda
• Use of CGE in project and policy
evaluation
• Characteristics of CBA and CGE
• Issues for
f analysis
• Integrating CGE and CBA- is there
convergence?
• Conclusions
6
Use of CGE in Project Evaluation
7
Literature Survey
• Growing recognition that CBA and CGE are on
about the same things
g
• Use of CGE to inform CBA (key shadow prices
etc)
• Critical discussion eg BTE 1999
• Texts eg
g Boardman et al 4 edn p124
p
• Recognition of the issue (eg Layman)
• But no integrated
g
or comprehensive
p
discussion
• We’re confused!
8
CGE as a Project Evaluation Technique
• CGE models are now used quite often as a way
of evaluating
g investments
• Esp for large infrastructure projects
• Melbourne CityLink, East West Transport Study
(Eddington)
( y
y
• Also, non infrastructure investments (Sydney
Olympic Games 2000)
9
The Tandem Approach- Pedalling in
Different Directions?
• Often done alongside a CBA
• CityLink,
CityLink East West study
study, Melbourne
Grand Prix
• Two different
ff
assessments off the same
project
• How comparable are they?
y may
y be close ((identical),
), but not
• They
couched in the same terms
10
What’s
What
s Missing?
•
•
•
•
A common welfare measure e.g.
Melbourne East West
CBA – Net Benefits
$1.0bn
CGE
(in 2031)
$852bn GSP
• A good or bad project - what does this mean?
• Results are not comparable
• But this can be resolved -we
we can get
comparable welfare measures for both
11
On the Same Track
•
•
•
•
CGE and CBA can address the same evaluation problems
How much better off is the economy as a result of this
project or policy?
Both have limitations and rely on approximations
“Once the additional used to increase production and
consumption are allowed for, CGE model results and
CBA are speaking the same language. CGE modelling
results
lt can th
then be
b used
d to
t enhance
h
the
th results
lt off CBAs,
CBA
rather than to be used as an alternative” Layman, 2004 (?)
12
Characteristics of the Techniques
13
Relative Characteristics of CBA and CGE
•
•
•
•
•
•
•
CBA can measure all costs and benefits of a project
CGE constrained by the model, usually at a high level of
aggregation
But details can be handled with sub models
Many effects are location specific
specific, eg noise
noise, congestion –
difficult to handle in CGE
Externalities difficult to handle with CGE (Not always)
CBA is usually used as a partial equilibrium approach
(Not always)
CBA tends to have difficulty with macro effects (eg
stimulation of a state econ)
14
Con t
Con’t
•
•
Difficult to handle distributional effects and measure
incidence of changes in CBA
Upshot- take advantage of complementarities
15
Caution
•
•
•
•
•
CBA and CGE are two approaches which are soundly
based in theory
Other alternatives are not
Eg using Input/Output Impact analysis
Guaranteed to produce very big “Benefits”
Increasingly used to “sell” projects (Dwyer et al 2004)
16
CGE and IO
•
•
CGE an outgrowth of IO
CGE to answer the limitations of IO
Should IO ever be used?
• In
I measuring
i
– eg carbon
b
footprint
f t i t (a
( snapshoth t nott
estimation of impact)
• In some regional
g
analysis
y
• When full CGE not available
17
Issues
18
Issues
•
•
•
•
•
•
•
•
•
•
Measuring welfare
Disaggregation and detail
CBA and PE
Externalities
U
Unemployment
l
t
Macro effects
Wider economic benefits (Transport)
Measuring key shadow prices
Misusing CGE
Data limitations and use of proxies
19
Welfare Measures
•
•
•
•
•
•
Theory of welfare measurement for CBA well established
Many CGE users are very cavalier (this does not have to
be so)
Simply identify benefits with change in GDP,
consumption
Are inputs to production free, as many implicitly argue?
Many CGE models have no explicit welfare criterion
B t some d
But
do – straightforward
t i htf
d to
t incorporate
i
t
20
Proxy Measures?
•
•
•
•
•
•
•
•
GDP is a welfare measure But under extremely restricted assumptions
Generally a bad measure
Can develop acceptable proxy measures, allowing for
costs of factor inputs
GNI (allowing for terms of trade effects), adjusted for cost
of additional capital and labour
Suitable for marginal changes
“Avoiding the Manna from Heaven assumption” (Dixon)
C retro-fit
Can
t fit a model
d l which
hi h does
d
nott have
h
an explicit
li it
welfare measure
21
Rigorous Measures
•
•
•
•
•
•
Theory of welfare measure measurement is well
developed
Consumers and producers surplus – properties well
known
Multi product measures can be used (eg King,
King equivalent
gain)
Require info available in a CGE model (eg demand
system)
t )
Can handle non marginal changes
Welfare measure can be as rigorous or casual as in CBA
22
Measuring Welfare
Welfare…
•
•
•
Essential if CGE and CBA results are to be compared and
differences can be reconciled
Fundamentally not a difficult exercise
Theoretical basis is the same
23
An Aside
Aside…
•
•
•
This is true for all policy evaluations using CGE
Do need to have the correct welfare measures
(Often not the case in Australia, even though Australia
relies heavily on CGE)
24
Disaggregation and Detail
•
•
•
•
•
•
CGE models are often at a high level of aggregation
Though not necessarily – detail can be brought in
Can expand the model to bring in the necessary detail
Eg, the model may have a “rail” sector- bet we can break
it down to “rail
rail and a “HST”
HST sector
And, indiv HSTs
Eg,
g develop a tourism specific model based on MMRF
Green (A state based model with a CO2 component)
25
CBA and Partial Equilibrium
•
•
•
•
•
•
•
•
Most CBA studies are primarily PE
Little detail except for big distortions
Need not be
Ideally, CBA is a GE technique, capable of shedding light
on indirect effects
GE theory for CBA- Dreze and Stern, Dinwiddy and Teal
But y
you need a way
y of measuring
g the GE effects
I.e . a CGE model
Some GE impacts are potentially large
26
Externalities
•
•
•
•
In markets are not included in a CGE model, impacts will
not be measured
But externalities will be left out – noise will not be
measured
Can adjust for this- measure and add in the cost of noise,
noise
just as for CBA
Don't need to include in the model-can retro fit
27
Incorporate Externalities?
•
•
•
•
•
•
•
•
Can include select externalities in a model
Eg, including CO2 in MMRF Green
Can then estimate the net effects on CO2 of a project,
allowing for indirect effects
Eg the effects on CO2 of a HST
Direct effects easy to measure
But not the indirect effects
Could even be negative
Easy to assess using a CGE model
28
Unemployment
•
•
•
•
•
•
•
Suppose there is less than full emp
What is the shadow wage, and impact on unemployment?
Very difficult to answer using a PE framework
Also difficult using some CGE models, which have simple
labour supply assumptions
The biggest practical problem in CBA?
Potentially
y large
g impact
p
on measurement of project
p j
benefits
Can at least put limits on the problem using a CGE model
29
Macro Effects
•
•
•
•
•
•
Often not well handled using CBA
Yet macro effects are sometimes big, esp when state
economies are considered (factors can move interstate)
The benefits from a road could include stimulating the
econ
What is the benefit to Victoria from an $1bn in tourism or
econ activity?
(Benefits of tourism are the sum of lots of small gains
and losses which add up to significant effects)
The gains and losses can be measured using a CGE
model
30
Wider Economic Benefits (of Transport)
•
•
•
•
•
Given prominence in EU, East West Study (Mel)
Two sources:
Agglomeration effects
Tax distortions
C th
Can
these b
be handled?
h dl d?
31
Can Effects be Measured With a CGE
Model?
•
•
•
•
•
Agglomeration effects- not really, unless you have a very
detailed model (not available) (econometric approaches
useful)
f l)
Tax distortions- easy, if you have a model which models
the tax structure accurately
y
Which some do
Conventional measures of the value of time do not
measure the
th impact
i
t on welfare
lf
off a saving
i
in
i time
ti
correctly (do consumers always use all of their time
savings in extra leisure? Forsyth)
The products of projects can be taxed or subsidised, and
this should be accounted for
32
Shadow Prices
•
•
•
•
•
A simple way of using a CGE model – show light on key
shadow prices
Marginal welfare cost of tax, exchange rate, shadow
wage, price of fuel
Eg estimates of MWC of tax (Stuart) – bigger than PE
measures
Canadian shadow exchange rate (Jenkins and Kuo)
Can’t be done using Monash/ MMRF – fixed labour supply
33
Data Availability
• Both approaches need a lot of data
• Esp CGE
• Need to have a complete model of the
economy with CGE
CG
• May use proxies and sensitivities
34
Large and Small Projects
•
•
•
•
•
•
CBA and/is used for large and small projects
What about CGE
Some examples of use in large projects (Melbourne
Eddington study)
Can in principle be used for small projects
If models are available, and model can pick up the
changes (not too aggregated)
Simulations are not necessarily expensive
35
Misusing CGE
• Easy to do
Eg loading the assumptions to get a result
• Eg,
• Such as assuming very large impacts on
employment even if the labour market is tight
• The result is that all projects look good and
g impact
p
on output
p and benefits
there is a big
• Using GDP/GNI as a welfare measure?
• Relevance for Australia at the moment?
36
Cost of CGE Models
•
•
•
•
Often claimed “CGE far too costly”
Usually existing models can be used
Model simulations are not expensive
Many models to choose from in Australia
37
Ways Forward
38
Ways Forward
•
•
•
•
•
We can seek to integrate CBA and CGE
They are complementary techniques
In theory, should give the same answer
In practice,
practice both have limitations
Can use both to get a better evaluation of a
project
39
Approaching the Answer from Two
Directions
• Often two reference points are better than one
• Express using a common welfare measure (a
challenge for Australia)
• Work out impact on welfare using both ways
• Work out why they are different- develop a
concordance
• Assess the accuracy of the approaches
40
Progress to Date/Priorities
1. Welfare measures- should be easy
2. Getting the theory of how to measure right (will a CGE
study pick this up?)
3. Do a study which brings CBA and CGE together
4 Develop a concordance4.
concordance why are they different and
why?
41
Conclusions
•
•
•
•
•
•
•
CGE now being used as an investment evaluation tool
Poses the question of what relationship this has with
CBA
Use so far has not been very illuminating
Many attempts,
attempts but there is no overall theory or guide
The two techniques are complementary
But they
y do need to be used in a comparable
p
way
y
Clear potential for improvements in evaluation
42
Thank You!
43
Refs
M King J Pub Ecs 1983
P Dixon EP 2009
J Dreze and N Stern
Stern, Handbook of Public Ecs
C Dinwiddy and F Teal, CUP 1996
G Jenkins and Kuo CJE
C Stuart, AER 1984
P Forsyth JTEP 1980
Abelson EP 2011
Bureau of Transport Economics 1999
Dwyer, Forsyth Spurr, TM, 2004
Boardman et al 2010
Eddington Melbourne East West Study 2008
y
, 2004
Layman,
44
Download