MIT GLOBAL CHANGE FORUM XXIV

1

• Climate Change and Policy Modelling Assessment
• Impacts in Modelling
• Francesco Bosello

Dottorato
2
The typical structure of a IIA exercise
Environmental impacts
Economic Assessment
Social Economic impacts
Climatic drivers

• ? Agr. Prod.
• ? Health care expenditure
• ? Labour prod.
• ..

• ? flood. land
• ? desert. land
• ? crop yield
• ? mort./morb.
• ..

• ? Temp.
• ? Preci.
• ? SLR

• ? Tourism Flows
• ? Energy demand

3
Steps before the economic assessment
Quantify impacts
Translate them into meaningful economic variables
Choice of a convenient baseline on which
impacts can be imposed. Assess changes respect to
a no climate change scenarios
Static baselines ? status quo

• Dynamic baselines - - evolving according to
  exhogenous storylines (IPCC SRES)- - evolving
  according to endogenous mechanisms

4
IPCC and exhogenous storylines
A1 rapid economic growth and technological
dev.pm. Low population growth.
A2 heterogeneous world, preservation of local
id, economic growth but more fragmented
technological progr. High population growth.
B1 convergent world, low population growth,
development towards a high tech and service
society. Emphasis on sustainability.
B2 like B1, but with more emphasis on local
solution.
Source IPCC, Climate Change 2001, The
Scientific Basis
5
The scenario issue
IPCC approach emissions scenarios stem from
exogenous storylines proposed by/incorporated in
a set of soft-linked models.
Problematic for hard linked models to replicate
those storylines as the storyline is
endogenously embedded in fact it is the model
itself
Replicating soft linked emissions with hard
link models may gt unrealistic economic
assumptions alternatively using model-consistent
economic assumptions may gt different emissions
paths!
The same problem with model comparison and
harmonization
Crucial role of the baseline ? it determines the
impact
6
Quantifying impacts (1) Sea Level Rise, some
literature
Low land in coastal countries with elevation lt
5m. (Source EEA, 2005)
7
Quantifying impacts (1) Sea Level Rise, some
literature
Land loss in 2085. Source Nicholls 2007
Population living in coastal flood plain in 2080.
Nicholls (2004)
SLR impacts (1 m.) in selected EU countries
8
Quantifying impacts (1) Sea Level Rise, some
literature
9
Quantifying impacts. Sea level rise _at_ ICES
Combining Areas at risk ? Basis is the 1993
Global Vulnerability Analysis by Delft
Hydraulics and Land Loss ? Nicholls and
Leatherman (1995).
Data set 1 Sq. Km. of land lost due to erosion,
if there is no protection for different SLR
scenarios. Country detail.
Aggregated for the regions of interest,
calculated in 2050 for 25 cm of SLR
10
Quantifying impacts (2) Health, some literature
Possible Changes in the Distribution of Death
Rates from Heat Related Mortality in Europe
2000 to A2 Scenario 2100, based on the climate
signal alone.
Source PESETA project (2007) at CEC (2007)
11
Quantifying impacts (2) Health, some literature
12
Quantifying impacts (2) Health, some literature
13
Quantifying impacts (2) Health _at_ ICES
Change in
Morbidity (n of years diseased)
Mortality (n of deceases)
Health Care Expenditure
Due to Climate Change (?T)
Calculated for five classes of diseases -
Malaria, - Schistosomiasis, - Dengue, -
Diarrhoea, - Cardiovascular and Respiratory.
Meta Analysis
14
Quantifying impacts (3) Health _at_ ICES
Change in base mortality additional n of
deceased people Examples
Vector Borne Diseases
Diarrhoea
Cardio Vascular
Applied to UP gt 65
15
Quantifying impacts (4) Health _at_ ICES
Additional years of life diseased
Additional Mortality
Additional Health Care Expenditure
Additional Health Exp. VBDDiarr.
Additional Health Exp. Cv and Resp.
From the literature
16
Quantifying impacts (4) Health _at_ ICES
Additional Mortality (1000) in 2050 for 0.93C
wrt 2000 (static baseline)
Additional years of life diseased in 2050 for
0.93C wrt 2000 (static baseline)
17
Quantifying impacts (4) Health _at_ ICES
Additional Health Care Expenditure ? LEVELS
Is split between public and private additional
expenditure ? LEVELS (using WHO 2003)
These then calculated as of GDP consistent with
the original database (Tol) ?
The is reported to GTAP GDP ? LEVELS consitent
with GTAP GDP
These levels are calculated in of GTAP public
and private demand for Non Market Services ?
shocks in change
18
Quantifying impacts (4) Health _at_ ICES
Final impacts on labour productivity and health
care expenditure as shocks for the ICES model
(1.5C wrt 1980-1999 average)
old baseline static model
19
Quantifying impacts (5) -- Energy
Climate Change affects energy demand through
changes in temperature

• Heating effect higher temperatures in cold
  seasons lead to a lower demand for energy for
  heating purposes
• Cooling effect higher temperatures in warm
  seasons lead to a higher demand for energy for
  cooling purposes

Both effects are likely to weight differently at
different geographical locations ? Hot countries
vs Cold countries
Econometric investigation on panel data performed
to identify the elasticity of energy demand to
temperature
20
Quantifying impacts (6) -- Energy
Data

• 31 countries (OECD and non-) from 1978 to 2000
• The dataset includes
• Real GDP per capita (IEA)
• Residential demand for oil products, electricity
  and gas (IEA)
• Fuel prices (IEA)
• Seasonal Temperature (Hadley Center UEA ? High
  Resolution Gridded Dataset)
• Balanced panel with the following observations
• - Electricity 550 (T 22 N 25)
• - Natural gas 418 (T 19 N 22)
• - Oil products 418 (T 19 N 22)

21
Quantifying impacts (7) -- Energy
Cluster analysis used to identify temperature
clusters

• GROUP 1 mild
• Austria, Belgium, Denmark, France, Germany,
  Ireland, Luxembourg, Netherlands, New Zealand,
  Switzerland, Greece, Hungary, Italy, Japan,
  Korea, Portugal, South Africa, Spain, Turkey,
  United Kingdom, United States
• GROUP 2 hot
• Australia, India, Indonesia, Mexico, Thailand,
  Venezuela
• GROUP 3 cold
• Canada, Finland, Norway, Sweden.

22
Quantifying impacts (8) -- Energy

• Cooling effect for electricity is present in hot
  and mild countries in summer and spring
• Heating effect for all fuels in winter and
  mid-seasons

23
Quantifying impacts (9) Tourism, some literature
Europe Changes in Tourism Climate Index (climate
attractiveness) 2071-2100 rt 1961-1990 A2
scenario
Green gt Increased climatic attractivenessRed
gt reduced climatic attractiveness
Source PESETA project (2007) at CEC (2007)
24
Quantifying impacts (9) Tourism _at_ ICES
Using a World tourism Model (HTM13, Tol et al.,
2005)
Which assesses changes in domestic and
international tourist flows with a country detail

The model is calibrated on 1995 data and explains
tourism flows with population, income,
temperature, coastal lenghts, travel distance.
25
International Arrivals
An example for Italy( changes wrt no climate
change)
Domestic Tourist Trips
Total Tourism Demand
26
Formulas for tourism
27
Quantifying impacts (12) Agriculture, some
literature
Source IPCC, (2007)
28
Quantifying impacts (10) -- Agriculture

• Rosenzweig and Hillel (1998) report detailed
  results from an internally consistent set of crop
  modelling studies
• Wheat, maize, rice, soybean
• Australia, Brazil, Canada, China, Egypt, France,
  India, Japan, Pakistan, Uruguay, USSR, USA
• 3 GCMs with and without CO2 fertilisation
• 3 levels of adaptation

Data extended to the regions of the economic
model and to different climate change scenarios ?
main yield drivers regional T and CO2
concentration ? parameterization as reported by
Tol (2002).
29
Quantifying impacts (10) -- Agriculture
Source Rosenzweigh and Hillel, (1998)
30
Quantifying impacts Agriculture _at_ ICES
31
Quantifying impacts Agriculture _at_ ICES
Changes in agricultural productivity, without
adaptation for 1.5C increase and 600 ppm in 2050
r.t. 1980-1999 average
32
How to introduce these impacts into a CGE
Sketching the structure of ICES
Database (xx.HAR)
Key parameters (xx.PAR)
Instructions which variables are exogenous and
which endogenous (closure)
The model equations (xx.TAB)
Command File (xx.CMF)
Output in Levels (xx.UPD)
Output in change (xx.SOL)
33
How to introduce these impacts into a CGE
The nature of the impact
Supply side impacts ? on stocks or productivity
? (e.g. health ? labour productivity, agriculture
? land productivity, sea level rise ? land stock)
They affects variables which are typically
exhogenous, easy to accommodate ? direct inputs
to the command file
Demand side impacts ? changes in preferences ?
(e.g. health ? health care demand, energy ?
energy demand, tourism ? recreational services
demand)
They affect variables which are typically
endogenous, this is a tricky issue
34
Equation PRIVDEGYCOM private consumption
demands for energy commodities (HT 46)
(all,i,EGYCOM)(all,r,REG) qp(i,r) - pop(r)
adsp(i,r) sumk,TRAD_COMM, EP(i,k,r)pp(k,r)
EY(i,r)yp(r) - pop(r)Equation
PRIVDNEGYCOM private consumption demands for
non-energy commodities (HT 46) (all,i,NEGYCOM)(a
ll,r,REG) qp(i,r) - pop(r) adsnec(r)
sumk,TRAD_COMM, EP(i,k,r)pp(k,r)
EY(i,r)yp(r) - pop(r)Equation NEWBUDGET
eplicit budget costraint (all,r,REG)
INCOME(r)y(r) sum(i,TRAD_COMM,
VPA(i,r)(pp(i,r)qp(i,r))
VGA(i,r)(pg(i,r)qg(i,r)))
SAVE(r)(psave(r)qsave(r))

Explaining demand-side shock modeling
Equation PRIVDMNDS private consumption demands
for composite commodities (HT 46)
(all,i,TRAD_COMM)(all,r,REG) qp(i,r) -
pop(r) sumk,TRAD_COMM, EP(i,k,r)pp(k,r)
EY(i,r)yp(r) - pop(r)
35
Variable (all,i,MASER_COMM)(all,s,REG)apd(i,s)
private cons. dem. shock parameter for market
services in reg. r Equation PHLDDMAS private
consumption demand for market services. (HT 48)
(all,i,MASER_COMM)(all,s,REG) qpd(i,s)
apd(i,s)qp(i,s) ESUBD(i) pp(i,s) -
ppd(i,s)Variable (all,s,REG)apdC(s)
private cons. dem. shock parameter for all non
market in reg. r Equation PHLDDNMAS priv.
cons. demand for for all trad comm but market
services. (HT 48) (all,i,NOMASER_COMM)(all,s,REG
) qpd(i,s) apdC(s) qp(i,s) ESUBD(i)
pp(i,s) - ppd(i,s)Equation NEWBUDGET
eplicit budget costraint (all,r,REG)sum(i,TRAD
_COMM, VPA(i,r)(pp(i,r)qp(i,r)))
sum(i,TRAD_COMM, VIPA(i,r)(ppm(i,r)qpm(i,r)))
sum(i,TRAD_COMM, VDPA(i,r)(ppd(i,r)qpd(i,r))
)
Explaining demand-side shock modeling
36
An IIA exercise example
The model ? static recursive dyn CGE
12 Regions USA United States NEWEURO Eastern
EU OLDEURO EU 15 KOSAU Korea, S.
Africa CAJANZ Canada, Japan, New Zealand TE
Transitional Economies MENA Middle East and
North Africa SSA Sub Saharan Africa SASIA India
and South Asia CHINA China EASIA East
Asia LACA Latin and Central America
17 Sectors Rice Wheat Cereal Crops Vegetable
Fruits Animals Forestry Fishing Coal Oil Gas Oil
Products Electricity Water Energy Intensive
industries Other industries Market
Services Non-Market Services
Used for investi-gations on transi- tional
dyna-mics
37
An IIA exercise example
The baseline asumptions changes 2001-2050
38
An IIA exercise example
The baseline results
39
CC impacts
1.5º C temperature increase in 2050 wrt 1980-1999
average ( change wrt baseline)
40
Results
41
Comparison with the existing literature
In 2050 Damage 0.3 of world (2050) GDP 352
billions US 2001
Source IPCC, 2007 FAR
42
Comparison with the existing literature
/tC
Survey di 108 stime (Tol, 2005)
314 Stern
261 prtplt1
93 tutti
51 prtp1
50 pr
16 prtp 3
Intervalli di confidenza al 67
43
Results, static vs dynamic
44
The sectoral picture
Climate change impacts on production in 2050 (
change wrt base)
GDP
-0.12 0.36 -0.06 -0.43 0.42 0.12 -0.89 -0.63 -1.80 -0.18 -0.91 -0.57
45
Caveats in interpreting the results
The climate scenario issue (uncertainty on the
possible temperature increase)
The Impact scenario issue (no irreversibility
and or catastrophic events)
The economic scenario issue the geographical
scale, transitional dynamics and frictions in
substitution.
The economic variable represented stock vs flows
(GDP as a welfare measure)
46
Stock vs flows, the case of sea level rise
Source Tol (2001)
47
Stock vs flows, the case of sea-level rise
The implicit value of land ( per km2)
48
Building damage functions
A standard approach
In a more or less sophisticated way, parameters
of a given damage function, whose functional form
is chosen with some ad hoc properties, are
calibrated such that in a given time with a given
temperature the total damage reaches a given
level expressed as () loss of potential GDP.
This amounts to
Assume exogenously the link between damage and
temperature (linear, quadratic, cubic)
A more or less additive procedure in the
estimation of total damage
49
Examples of damage functions
Nordhaus and Yang (1996)
Nordhaus and Boyer (1999 -)
Manne and Richels (1996 -)
Peck and Teisberg (1992 -)
50
An example CCDF calibration in RICE 2007
Source Nordhaus (2007), lab notes on RICE 2007
51
An example CCDF calibration in RICE 2007
Source Nordhaus (2007), lab notes on RICE 2007
52
An example CCDF calibration in RICE 2007
Source Nordhaus (2007), lab notes on RICE 2007
53
An alternative methodology. Tol
Source Tol, (2002)
54
Using (static) CGE to calibrate the damage
function
Quantify all impacts for different ?Ts
Plug them together into the CGE
Estimate the parameters of the implicit
regional damage functions
The main advantage of this procedure is to
consider autonomous adaptations and thus impact
interactions.
55
Damages and damage coefficients
56
A new calibration
Recall the RICE 99 (and subsequent) damage
function
57
New damages by temperature
58
New damages by region
59
New emission path
60
Open questions
Is it legitimate to use a static model to
calibrate a CC damage function?
Is it legitimate to use a flow-based model to
calibrate a CC damage function?
Is it legitimate to use a market-based model to
calibrate a CC damage function?