Mapping a Future Low Carbon Economy

Dirk van Seventer and Rob Davies

MAPS ECONLAB 3:

“PROVIDING ROBUST SOCIO-ECONOMIC INFORMATION OF SCENARIOS & NATIONAL CONTRIBUTIONS”

Cape Town November 5 and 6, 2014

Outline

Motivation

Structural Decomposition Analysis

Introducing new products

Motivation

Bruno asked how we might go about constructing a SAM representing a low energy economy at some date in the future

Our response: you have CGE models which you use to look at policy impacts

run them and generate the future SAM

also – predicting far into the future is inherently impossible

so easy and impossible task

Motivation

But then thought and argued and realisedwhat if CGE model does not arrive at dream future?- where can we do more?

some vision of future useful for gauging progress- multiple targets

process of trying the impossible is instructive

So presenting some incoherent and simplistic thoughtsnot sure if either useful or original

But hope it stimulates some thought

Stylised view of CGE modellingWe all know, but to set the stage

Stylised view of CGE modelling

Post-shock SAM could be produced by static or dynamic model

formal or intuitive model

Just want ‘before’ and ‘after’

Represents a new structure

What drives the changes?

Can use decomposition to think about this

Aggregate emissions decomposition

Kaya identity

Useful for decomposing sources of change in emissions in the past

Also useful thinking about targets and constraints for futureIf - population growth rate is exogenous in medium run- growth rate of income per capita has some lower (desired) bound

then improving energy cleanliness and efficiency provide only room for manoeuvre to reach RBS target

But in a multi-sector/economy wide model there is also possibility of compositional effect

Can change ratios by changing technology AND by changing sectoral structure

Decomposition of CGE results

Structural Decomposition Analysis looks at change of:

Technical change in energy use/intensity (exogenous driver)

Economic growth (exogenous driver)

Shifts in industry composition (endogenous)

Shifts in industry composition may be missing link

Help understand why model may not get “where we want to be”

Could include:- Introduction of new products (currently unknown)

- May require rebalancing end year SAM (entropy method)

Broad Objective of Decomposition Analysis

Decompose change in output into contributions from:

Technical Change- Industry linkages

Change in Final Demand- Change in Levels of Final Demand

- Change in the Mix of Final Demand

- Change in the Distribution of Final Demand

Decompose change in Energy Use

Same as above

Additional: change in Energy Intensity- Can be expanded to other variables (employment, emissions)

Examples

But … also some algebra

-

Some More Algebra

(1)

Let and and substitute into (1)

(2)

Re-arrange

(3)

Average weighting using base year and final year shares

(4)

Technology (Industry Composition)

Change

Final Demand Change

Results of exampleDemo at Speed Dating??

What do the results show?

Further Decomposition of F

(5)

In which & are vectors of expenditure shares

…and and total expenditures (scalars)

Now, follow the same decomposition principles

(6)

Substitute into previous results

eqn (7)FD Level Change

FD Mix Change

FD Mix Change –

New Products

Adding Energy (8)

With &: matrices eng/output ratios on the main diagonal

If : difference in eng/output ratios on the main diagonal

Then: decomposition along same lines:

(9)

Energy Input Effect

Technology Effect FD Effect

Energy Intensity Multipliers

Direct and Indirect

Energy Use Per 1 Unit of

Further Decomposition

possible: FD Mix Change – New

Products

Extensions

Previous examples could be extended to

various types of energy

emissions

Can apply to SAMs

industries, commodities, types of households, types of factors

Needs SAM in constant prices to get proper technical coefficients

Can apply to Hybrid SAMs

Energy (or other) non-monetary units

Monetary units

Can do similar decomposition simultaneously- Energy use

- Output

Lessons from SDA

Shows sources of change in Energy (and emissions)

Not only from growth and improved energy efficiency

But also from structural changeinter-industry linkages

final demand structure

Could be used to summarise major sources of changes from CGE analysis

Might show the ‘gap’ in achieving the RBS ‘dream SAM’

But only deals with existing industries and commodities

What about new industries and products?

New Industries and Products

“Dream SAM” will have new industries/products

Inherently unknowable

But maybe we can think about the characteristics they need to have to help ‘fill the gap’

Relevant characteristics:energy intensity

cost structure

linkages

Old New FD Total

Old A C

New D B

VA

Total

Impact of new industries will depend on:

B – cost structures, energy use and linkages within the new cluster

C – retarding effect of the old on the new

D – progressive impact of the new on the old

Our initial thinking before constructing numerical model

A Stylised Example

A : cluster of old industries

B : cluster of new industries

C : backward linkages from new to old

D : forward linkages from new to old

Took our previous 3x3 IO Table

Added new industriescannot do that without changing size of economy, so halved old economy and put in roughly similar sized new cluster

Constructing a toy model proved difficult, but some lessons

Constructing a numerical example

Old 1 Old 2 Old 3 New 1 New 2 New 3 F XOld 1 10 20 25 - - - 45 100Old 2 15 5 30 - - - 30 80Old 3 30 40 5 - - - 25 100New 1 - - - - - - 0 0New 2 - - - - - - 0 0New 3 - - - - - - 0 0V 45 15 40 0 0 0 0 100X 100 80 100 0 0 0 100

New industries use some of Old 3 “Dirty Energy”

Problem: where do the supplies from old to new come from?growth in output of old?

switch from final demand of old?

switch from intermediate supplies to old?

Adding backward linkages from new to old

Constructing a numerical example

Old 1 Old 2 Old 3 New 1 New 2 New 3 F XOld 1 5.00 10.00 12.50 - - - 22.50 50.00 Old 2 7.50 2.50 15.00 - - - 15.00 40.00 Old 3 15.00 20.00 2.50 0.60 1.00 0.15 12.50 51.75 New 1 - - - 6.00 7.50 12.50 25.00 51.00 New 2 - - - 12.00 5.50 10.00 17.50 45.00 New 3 - - - 5.40 9.00 1.35 13.00 28.75 V 22.50 7.50 21.75 27.00 22.00 4.75 105.50 X 50.00 40.00 51.75 51.00 45.00 28.75 105.50

Old industries replace some dirty energy with clean New 3

Reduces energy use in economy

Completed filling in linkages between ‘non-energy’ old and new

Adding forward linkages from new to old

Constructing a numerical example

Old 1 Old 2 Old 3 New 1 New 2 New 3 F XOld 1 5.00 10.00 12.50 - - - 22.50 50.00 Old 2 7.50 2.50 15.00 - - - 15.00 40.00 Old 3 13.50 18.00 2.25 0.60 1.00 0.15 12.50 48.00 New 1 - - - 6.00 7.50 12.50 25.00 51.00 New 2 - - - 12.00 5.50 10.00 17.50 45.00 New 3 1.50 2.00 0.25 5.40 9.00 1.35 13.00 32.50 V 22.50 7.50 18.00 27.00 22.00 8.50 105.50 X 50.00 40.00 48.00 51.00 45.00 32.50 105.50

When we have a dirty cluster and a clean cluster, the total energy use depends on the total energy produced by each. The linkages do not matter

But the impact of an increase in final demand does depend on the linkage structure including linkages between non-energy sectors, because of multipliers.

Could do a multiplier decomposition (not SDA) for ‘within’ and ‘between’ cluster effects

General lessons from constructing numerical example

Concluding Thoughts

Building a dream SAM that accurately captures a low energy future is inherently impossible

Nonetheless might be a useful exercise‘Balancing constraint’ of SAM constrains our guesses

Put different pieces of an imagined future into a consistent whole

If we do build a dream SAM, might provoke thoughts about the policy levers needed to get there

More insights might come from more realistic SAM – without the simplistic balancing mechanisms in our toy model

will need some statistical balancing procedure

But remember models are simplifications to help us understand issues, so dream SAM could be simple

not trying to describe the future, but to think about it

End

Play it again, SAM

Introducing New Products

Why?: to get closer to “where we want to be”

Unconstrained problem

What do we know? Products must be:

More energy efficient

Sufficiently high value added generation

But do not have policy levers in models

That control introduction of new products

See example