Transcript ECON 146 - Week #6

Urban and Regional Economics
Prof. Clark
ECON 246
Weeks 5 and 6
Discussion of Growth Papers

Bartik
– Addresses the question of who benefits
from regional growth

Noll and Zimbalist
– Does the building of stadiums promote
economic growth
A Brief Overview of Central
Place Theory
Size distribution of U.S. cities
Pop Range
» >12.8 million
» 6.4 - 12.8 million
» 3.2 - 6.4 million
» 1.6 - 3.2 million
» 800k - 1.6 million
» 400k - 800k
» 200k - 400k
» 100k - 200k
» 50k - 100k
Number of Areas
1
2
4
14
19
33
52
99
172
Systems of Cities

Considers market areas
– Focus is on distribution of goods within
market
– Derive market shapes in competitive
market structure

Cities shaped by markets for various
goods and services
– Look at market sizes
– Look at number of markets
Define Market Areas

Producer assumptions
–
–
–
–

Producers serve geographic areas.
Producers have same technology
ubiquitous inputs
No agglomeration in shopping
Consumer assumptions
– Consumers evenly distributed over space
– Buyers must travel to store to buy goods (constant
travel costs per mile)
– Consumers care about the net price
Market Area for Firm



Net price=store price +
transport cost
Consumers living closer
to market pay lower net
prices.
Market area defined by
cost of home production
• x miles in this case


Net Price Graphically
Net Price
Cost
of HP
Store
Price
Suppose monopolists
carve up a region
x
0
x
Distance from Market Center
Monopoly Markets
Non-Competing Market Areas
$
Monopolist 1
Monopolist 2
Monopolist 3 Monopolist 4
Distance from Market Centers
Market Areas
Monopolist
1
Monopolist
2
Monopolist
3
Monopolist
4
Monopolist
5
Monoplist
6
Monopolist
7
Monopolist
8
Introducing Competition



Assume
monopolists are
making pure
economic profits
Is this a stable
situation?
If not, where would
firms enter?
MonopolistMonopolistMonopolistMonopolist
1
2
3
4
Monopolist Monoplist MonopolistMonopolist
5
6
7
8
Firm Entry Drives Out Profits
Monopolist
1
Monopolist
2
Monopolist
3
Monopolist
4
Monopolist
5
Monoplist
6
Monopolist
7
Monopolist
8
Firms enter here
Eventual Market Shape





All profits driven
out
Market structure is
monopolist comp.
No areas unserved
Note: Actual shape
of market is open to
debate.
Introduce other
markets
Different sized markets
Market sizes differ
according to the scale
economies and
density of demand.
There are many small
markets and fewer
large markets.
Cities simply reflect Collections
of Markets

Maybe only one market for top level
plays
– Located in NYC, and spans entire nation

Maybe 4 markets for large international
airports
– LA, NY, Chicago, Atlanta

There are many markets for gas stations
How Realistic is this?

Are assumptions satisfied?
– What does this imply about value of model?


Does model predict well?
Rank-size rule:
• Rank*Size=constant
• Statistical regularity in some regions
• Doesn’t seem to hold in U.S.
» Population more evenly distributed over space than this
suggests.
» Possible reasons?
The Regional IO model
The regional IO model is based on an
accounting identity that states: The sum
of all inputs must equal the sum of all
outputs.
 Assuming:

– accurate accounting of all sectors
– accurate account of for all the transactions
between sectors and outside economy

Then identity should hold!
Overview of How Model Works




Step 1: Model identifies sectors in the
regional economy, and then sets up a
transactions table to evaluate resource flows
between these sectors.
Step 2: From transactions table, coefficients
of technical coefficients can be inferred.
Step 3: Derive demand relationships.
Step 4: Shocks in external or final demand
are mapped to each sector.
Step 1: Transactions Table
Output Sold To
Inputs
Manuf. Service Trade Households Exports Gross
Supplied by
Output
-----------------------------------------------------------------------------Manuf.
6
4
10
0
20
40
Service
5
8
2
25
10
50
Trade
0
0
0
30
0
30
Local L,K,D
14
33
8
0
0
55
Imports
15
5
10
0
--30
Total Inputs
40
50
30
55
30
• This gives indication of intersectoral interdependencies.
Step 2: From Transactions Table
to Technical Coefficients
Determine how much of the total value
of inputs for the sector was spent on
any given output.
 Divide the column by the total input
value for that column.

Technical Coefficients Table
Manuf.
Service
Trade
Households
-----------------------------------------------------------------------------Manuf.
0.15=6/40 0.08=4/50 0.33=10/30 0.00=0/30
Service
0.125=5/40 0.16=8/50 0.067=2/30 0.455=25/55
Trade
0.00=0/40 0.00=0/50 0.00=0/30 0.545=30/55
Local VA 0.35=14/40 0.66=33/50 0.26=8/30 0.00=0/55
Imports
0.375=15/40 0.10=5/50 0.33=10/30 0.00=0/55
Total Inputs
40
50
30
55
Column Interpretation: How inputs are used in the sector.
For $1 of Manuf., you use $0.15 of Manuf., $0.125 of Service,
$0 of Trade, $0.35 of Local inputs, and $0.375 of Imports.
Technical Coefficients Table
Manuf.
Service
Trade
Households
-----------------------------------------------------------------------------Manuf.
0.15=60/40 0.08=4/50 0.33=10/30 0.00=0/30
Service
0.125=5/40 0.16=8/50 0.067=2/30 0.455=25/55
Trade
0.00=0/40 0.00=0/50 0.00=0/30 0.545=30/55
Local VA 0.35=14/40 0.66=33/50 0.26=8/30 0.00=0/55
Imports
0.375=15/40 0.10=5/50 0.33=10/30 0.00=0/55
Total Inputs
40
50
30
55
Row Interpretation: Who buys output
Manuf. Demand = 0.15*M+0.08*S+0.33*T+0*Y(income)+XM
XM is known as final or exogenous demand.
Step 3: Derive Demand Equations






M=0.15*M + 0.08*S + 0.33*T + 0.00*Y + XM
S=0.125*M + 0.16*S + 0.067*T +0.455*Y + XS
T= 0.00*M + 0.00*S + 0.00*T +0.545*Y + 0
Y= 0.15*M + 0.66*S + 0.267*T + 0.00 *Y + 0
Endogenous variables: M, S, T and Y are
determined inside this system of equations: (i.e.,
we have 4 equations and 4 unknowns)
Exogenous Variables: XM, XS, (XT=0 in this case)
(XY=0 since this is local value added) are
determined outside this system.
Question: If we solved for M, S, T and Y
given current values of exports, what
solution would we get?
M=40, S=50, T=30 and D=55
Can Derive Local Multipliers
(Manipulate so each sector depends only on X)
M=
S =
Y =
T =


1.613*XM
1.141*XM
1.542*XM
0.880*XM
+
+
+
+
0.772*XS
3.236*XS
2.413*XS
1.316*XS
Thus, multipliers no longer constant for
all sectors!
Step 4: Mapping out influence of
Disturbances

Suppose exports change:
– Then you have a new set of four equations and four
unknowns to solve simultaneously.

Technical coefficients don’t change.
– What does this assume about input substitutability?

Get new endogenous levels of demand, as a result
of the external shock.
– Allows you to get idea of interdependencies between
sectors and how growth in one sector effects other
sectors.
Limitations

This is still a demand-based model.
– It does not allow for supply effects.

Implicitly assuming constant wage (i.e., horizontal
supply).
– Why?

SR model
– Assumes constant multipliers
» LR vs. SR assumption?
– No substitution available.
» LR vs. SR?
Limitations - continued

Regional limitations
– Difficult to get local transactions tables
– National proxies must be used but they
may be inappropriate.
– Regional technical coefficients may change
more rapidly than national coefficients.
Extensions of this approach
Over the last 20 years, this model has been
refined substantially.
 There are ways to deal with supply issues.
 There are also ways to allow isoquants to
be smooth (i.e., allow inputs to be
substituted in production).

Two popular models:
IMPLAN and REMI
IMPLAN model is pure IO model
 REMI model is commercially available
hybrid model.

–
–
–
–
–
Developed by George Treyz at U. Mass.
Has an econometric and an IO component.
Does incorporate supply effects.
Widely used by policy makers.
Look at demo
Regional Econometric Modeling

These are constructed differently than IO or
Export-Base Models.
– Can incorporate both supply and demand factors.
Model is based on model-builders beliefs about
how the urban economy works.
 Relationships are typically estimated using
local, regional and national data.

Regional Econometric Models:
Overview
Roger Bolton
Journal of Regional Science, 1985,
Vol. 25 (4) pp. 495-519.
Not assigned but on reserve FYI
Very thorough review article
Article focuses on academic models which
have been developed in 1970’s and early
1980’s.
 Focuses on single-region models.
 Our focus on Sections 1-4 briefly, and 13-14.

– 5-12 give specific details on individual
components of models.

Keep this paper handy as a reference,
should you work in public policy.
Exogenous vs. Endogenous
Variables


Distinction between two types
Advantage of model with numerous endogenous
variables.
– Can model simultaneous (feedback) effects between
variables.
» e.g., increase in demand may put upward wage pressure in the
sector, and ultimately lead to inmigration.

Disadvantage
– Difficult to estimate due to data limitations.
Level of Aggregation

Single-region model
– May develop model for Milwaukee, or
Southeastern Wisconsin.
– Everything else is considered ROW.
– No interdependencies between cities in region.

Multi-regional model
– May include metropolitan areas in Wisconsin
– Can include all metropolitan areas in state.
» Derive the interdependencies in great detail.
Single-Region Model
• From regional to
national is called
bottom-up structure
• From national to
regional is called
top-down structure
• Link between region
and the rest of world
(ROW) is frequently
unidirectional.
Regional
Model
National
Model
(ROW)
Bottom up influences
(i.e.,
) are
frequently negligible.
When Bottom-up links are not
negligible

One sector in region is dominant for
nation.
– e.g., Detroit and auto industry.

When single region is large.
– e.g., Suppose California is considered a single
region.

Region’s policies affect national markets
– e.g., California emission standards
Multi-Regional Models
Bottom-up component now more likely
to be important.
 Interregional feedback effects now
possible.
 Models get more complex.
 Lets look at Bolton’s Figure 2.

– We break it into components.
Multi-regional Models:
National components
Exogenous
National
Variables
Endogenous
National Vars.
(not regional sum)
Endogenous
National
(regional sum)
Multi-regional Models:
Adding Regional Components
Exogenous
National
Variables
Region 1
Model
Endogenous
National Vars.
(not regional sum)
Region 2
Model
Interregional Feedback Effects
Endogenous
National
(regional sum)
Region 3
Model
Multi-regional Models:
Top-down structure
Exogenous
National
Variables
Region 1
Model
Endogenous
National Vars.
(not regional sum)
Endogenous
National
(regional sum)
Region 2
Model
Region 3
Model
Multi-regional Models:
Bottom-up structure
Exogenous
National
Variables
Region 1
Model
Endogenous
National Vars.
(not regional sum)
Endogenous
National
(regional sum)
Region 2
Model
Region 3
Model
Which is theoretically preferred?
Differences between regional and
national models

National models based on National Income
identity:
Y=C+I+G+X-M

Data limitations prevent comparable regional
models.
– Components C, I, X, and M typically not available.

Regional income becomes sum of labor
earnings or industry output.
Other data limitations

Nonmanuf. output data less readily available.
– No investment data on nonmanuf. sector.

Nonlabor income is difficult to track from
region to region.
– i.e., returns on land and capital earned in one
region and spent in another.
– Thus, focus is on less mobile labor income.

Capital stock even in manuf. sector weak.
– No public capital stock included.
Models tend to be SR rather than
LR
Since models can’t deal well with
changes in industrial structure
due to investment, they tend to
be SR rather than LR.
Purposes

Models are often built for a specific purpose.
–
–
–
–

pure science (not typical)
forecasting
government revenue forecasting
policy analysis
Like other academic endeavors, models may not be
balanced.
– Tend to favor purpose of the modeler.
– Tests of forecast performance rarely done for long term
forecasts due to limited time-series.
Econometric Techniques

Frequently use OLS.

Sample sizes may be to small to take
advantage of 2SLS.
Data limitations may make identification a
problem
Monte Carlo studies suggest that the
simultaneous equation bias is small.


Advantages and Disadvantages

Advantages
– Flexibility
– Ability to model both supply and demand side of
economy.

Disadvantages
– Expensive to build
– Data constraints frequently lead to top-down even
when theory suggests bottom-up design.