Data requirements for ARIO economic assessment model

Data requirements for ARIO
economic assessment model
Aaron Gertz and Jim Davies
2012 Advanced Institute on Data for Coastal Cities at Risk
October 22, 2012

Economic impacts of flooding: overview
◮ Direct impacts (damage to capital stock):
◮ Buildings;
◮ Machines;
◮ Infrastructure (transportation, electricity, water);
◮ Indirect impacts:
◮ Intermediate inputs not available;
◮ Labour supply, electricity constrained;
◮ Demand for goods reduced.
◮ Other considerations:
◮ Price changes in aftermath of flood;
◮ Recovery process and duration;
◮ Adaptation: Store inventories, increase imports, etc.
◮ Insurance, borrowing, tax base.
◮ Model it! Build on Hallegatte (2008) ARIO model.

Outline
1. Brief overview of model
2. Input-output table
3. Damage to capital stock from flooding
4. Recovery
5. Optional: Prices, labour
6. Summary

Model overview
◮ Damage function converts flood level to capital stock
damage by sector.
◮ Each sector’s capacity to produce reduced based on
capital stock damage and capital-output ratio.
◮ Production is allocated between intermediate demand and
final demand (TFD) for each industry:
Y (i) =
A(i, j)Y (j)+LFD(i) + E(i) + HD(i) +
D(j, i)
j
j
Total Final Demand(TFD(i))
◮ TFD is comprised of local final demand, exports and
disaster damages to households and firms to be repaired.

Model overview - continued
◮ Since production capacity is constrained post-disaster,
there is imbalance between supply and demand.
◮ Iterative procedure used to determine new supply;
intermediate demand satisfied first, TFD demand equals
residual.
◮ Part of residual used to repair capital stock,
HD(i) +
j D(j, i), increasing next period production
capacity.
◮ Price effects enter through adjustment to LFD(i).
◮ Adaptation enters through increased imports and
overproduction.

Input-output table
◮ Different sectors of economy use inputs from each other,
for example (Miller & Blair, 2009):
Input
Output
Ag Mfg
Household
Consumption
Other
Consumption
Total
Output
Agriculture 150 500 50 300 1000
Manufacturing 200 100 400 1300 2000
Labour
services
300 500 50 150 1000
Other domestic
payments
325 800 300 250 1675
Imports 25 100 200 150 475
Total 1000 2000 1000 2150 6150
outlays
◮ Use to obtain matrix of technical coefficients, A.
◮ a11 = 0.15.

Input-output table - continued
◮ Input-output tables published by government agencies
often “rectangular” ⇒ Must transform to square.
◮ Several levels of aggregation (e.g. 3, 15 or 100 sectors).
◮ Damage function requirements limits degree of
disaggregation ⇒ use 15-sector level in model, perhaps
break out key sectors.
◮ Tables often at national of provincial/state level ⇒ May
need assumptions to create local table.
◮ Example: B.C. I-O table.
◮ I-O tables not published each year, but can typically get
output by sector annually.

Flood damage
◮ To assess economic impact of flooding, must determine
damage function D(i, j, l).
◮ That is, damage to capital stock in sector i of type j for
flood level l.
◮ Ex. Damage to machines in agricultural sector for 1 m
of flooding.
◮ Damage function could be obtained in two ways:
1. Using data from past floods;
2. Using standard formulas.

Flood damage - continued
◮ Next must convert capital stock damage to production
constraint ⇒ need capital-output ratios.
◮ If necessary, can use standard capital-output ratios.
◮ Production capacity reduced by
ˆD(i)
, where for given l,
κi VA(i)
ˆD(i) =
j D(i, j), κi is the capital-output ratio and
VA(i) is pre-flood “value added”.
◮ Value added is net output, i.e. not counting intermediate
purchases.

Recovery period
◮ In model, recovery is endogenous because output used to
reconstruct capital stock.
◮ However, adaptation affects speed of recovery ⇒
adaptation parameters play role in determining pace of
recovery.
◮ Two mechanisms of adaptation:
1. Overproduction;
2. Additional imports.
◮ Data on capacity to overproduce in aftermath of flood
(unlikely) or increase in imports (perhaps more likely)
would be helpful.

Prices and labour
◮ Input-output modelling neglects price effects on consumer
and firm choices, but in short run little scope for
substitution.
◮ However, there are some effects of prices even in short
run.
◮ E.g. Surge in demand for construction drives up demand
for labour in construction which drives up wages.
◮ In GE models, solve for price such that supply = demand.
For I-O models, must improvise.

Prices and labour - continued
◮ Model price effects as follows:
◮ Decrease in supply of goods ⇒ price of goods increases
⇒ demand for goods decreases.
◮ Income decreases/increases based on demand for labour
⇒ decrease/increase in demand for goods.
◮ Elasticity parameters determine magnitude of response.
◮ Price data from previous floods could help choose
parameters, but could use standard values.
◮ Labour: Depending on desired output from economic
module, may need labour (employment) data.

Summary of data requirements
◮ Essential data:
◮ Input-output table;
◮ Data/information to localize table;
◮ If table out of date, current output by sector;
◮ Capital stock damage from previous floods or damage
functions.
◮ Desirable data:
◮ Capital-output ratios by sector;
◮ Recovery time for previous floods;
◮ Additional imports during floods;
◮ Price/quantity changes during floods;
◮ Labour force?