Extreme events: causes and consequences

1

• Extreme events causes and consequences
• WP4 Economic impacts of extremes

2
1. The need for models with endogenous dynamics
Stéphane Hallegatte, CIRED and ENM,
Météo-France M. Ghil, ENS and UCLA P. Dumas, LMD,
CNRS J.-C. Hourcade, CIRED, CNRS Andreas Groth,
LMD, CNRS
3
The  real cycle and the endogenous cycle
theories

• In the real cycle theory, business cycles and
  economic fluctuations arise from exogenous real
  shocks like changes in productivity, changes in
  energy prices or fiscal shocks
• Aside from these exogenous shocks, the economic
  system is stable
• All markets are at equilibrium, no non-voluntary
  unemployment
• No action on the economy (e.g. recovery policy)
  is desirable
• In EBC models, cyclical behavior originates from
  endogenous instabilities in the economic system
• Several instabilities have been proposed
• Profitability-investment instability
• Delays in investment
• Income distribution
• Actions on the economy can be positive

4
NEDyM (Non-equilibrium Dynamic Model)

• Represents an economy with one producer, one
  consumer, onegoods used both to consume and
  invest.
• Developed from the Solow (1956) model, in which
  all equilibriumconstraints are replaced by
  dynamic relationships with adjustment delays.
• NEDyM equilibrium is neo-classical and identical
  to the Solow models one. If parameters are
  changing slowly, NEDyM has the same trajectories
  than the Solow model.
• Because of market adjustment delays, NEDyM
  dynamics exhibit keynesian features along
  transients trajectory in response to shocks.
• NEDyM exhibits endogenous business cycles

5
Hopf bifurcation from stable equilibrium, to
limit cycle, to chaotic behavior
6
ainv 1.7
ainv 2.5
7
ainv 10
ainv 20
8
Endogenous dynamics an alternative explanation
for business cycles
9
Endogenous business cycle

• Business cycles originate from the
  profitinvestment relationship
• (oscillations with a 5.5-year period)
• higher profits gt more investments gt larger
  demand gt higher profits
• Business cycles are constrained by three
  processes
• Increase in labor costs when employment is high
• Constraints in production and the consequent
  inflation in goods prices when demand increases
  too rapidly
• Financial constraints on investment.
• Calibration (and validation) of endogenous cycle
  models is particularly difficult.
• Carl Chiarella (from the University of Sydney)
  spent 2 months at ENS to exchange on this point.
• Patrice Dumas and Andreas Groth are working on
  this issue (they will present later on today).

10
Catastrophes and economic situation
A vulnerability paradox a disaster that affects
an economy during the expansion phase of its
business cycle causes more damages than if it
touches it during its recession phase.
Recession
Expansion
Business cycle
11
Catastrophes and economic situation
A vulnerability paradox a disaster that affects
an economy during the expansion phase of its
business cycle causes more damages than if it
touches it during its recession phase.
Recession
Expansion
Business cycle
Economic losses due to a disaster, as a function
of the pre-existing economic situation
Limited losses if the disaster affects an economy
in recession
12
Catastrophes and economic situation
A vulnerability paradox a disaster that affects
an economy during the expansion phase of its
business cycle causes more damages than if it
touches it during its recession phase.
Recession
Expansion
Business cycle
Economic losses due to a disaster, as a function
of the pre-existing economic situation
Large losses if the disaster affects an economy
in expansion
13
2. The importance of reconstruction dynamics a
detailed analysis of the 2004 and 2005 hurricane
seasons in Florida
A RMS work with Auguste Boissonnade, Marc-Etienne
Schlumberger and Robert Muir-Wood (RMS)
14
14 FL Cities Selected for the Quantification of
Historical Economic Demand Surge
The 2004 and 2005 seasons in Florida
15
Model Principles

• Price responds to market conditions in the
  construction sector
• Price (p) increases when demand (DL) is larger
  than capacity (C) in county i
• Incentive to work in the construction sector
• Ci increases when pi increases as more workers
  shift toward the construction sector
• Incentive to move to the most affected countries
• If pj gt pi dijcij - then workers move from
  county i to county j

16
Observed and modeled Price Increases in Florida
17
3. The need to consider indirect impacts in more
detailed models
Stéphane Hallegatte, CIRED and ENM,
Météo-France Fanny Henriet, CIRED Barbara
Coluzzi, CERES, ENS Michael Ghil, CERES, ENS
18
Assessment of production losses in Louisiana due
to the landfall of Katrina
We start from (quite uncertain) estimates of
damages in 15 sectors.
19
Disaster consequences in economic modelsThe
Katrina case
Production changes in 15 sectors in Louisiana, in
the Katrinas aftermath, using the ARIO
(Adaptive Regional Input-Output) Model.

• An input-output model of the local economy, with
  15 sectors that interact with each other.
• We investigate the interplay of
• the decrease in production capacity due to
  damages
• the increase in demand due to reconstruction needs

20
Disaster consequences in economic modelsThe
Katrina case
Production changes in 15 sectors in Louisiana, in
the Katrinas aftermath, using the ARIO
(Adaptive Regional Input-Output) Model.
Boom and constraints in the construction sector
Direct losses 107b Production
losses 23b Housing service losses 19b Total
losses 149b
Cross-sectoral propagations
Decrease in production capacity
21
Imperfect results, but orders of magnitude are
consistent with data
22
The nonlinear relationship between direct losses
and total losses
The boom in the reconstruction sector is limited
by economic and technical constraints, and by
damages, leading to a nonlinearity. In
Louisiana, this nonlinearity appears when direct
losses exceed 50 billion.
23
Imperfect results suggest ideas for future
research
Value added losses per sector, data and model
24
Imperfect results suggest ideas for future
research
Underestimation in sectors with small businesses
Value added losses per sector, data and model
25
Conclusions and perspectives a hierarchy of
models

• An highly idealized model shows that extreme
  event consequences depend on the economy that is
  affected
• Its ability to carry out the reconstruction
• Its situation when the extreme strikes.
• A more detailed analysis of the reconstruction
  process after the 2004 and 2005 hurricane season
  in Florida supports this finding.
• Calibration is an issue, data-assimilation
  methods are developed by Patrice Dumas and
  Andreas Groth.
• 2. Assessing total impacts requires a more
  detailed model a Regional Input-Output model at
  the sector level (ARIO) is proposed to do so.
• The model shows that total impacts increase
  nonlinearly with direct losses.
• But the model is found to underestimate impacts
  in sectors with many small businesses (and many
  bankruptcies).
• Modeling these mechanisms requires a more
  detailed model, at the business level
• more work in under way with Fanny Henriet and
  Barbara Coluzzi to model the economy as a network
  of businesses.