Modelling infrastructure and migration in a NEG model

1

• Modelling infrastructure and migration in a NEG
  model
• Luigi Marattin
• University of Bologna
• TERA FINAL CONFERENCE
• Ghent - 26th August 2008

2
PRESENTATIONS PLAN

• 1. THE NEG APPROACH
• 2. THE MODEL
• 3. RESULTS AND POLICY IMPLICATIONS

3
1. THE NEG APPROACH

• New Economic Geography (Krugman 1991)
• It attempts to provide answers to one question
• What leads the process of concentration /
  dispersion of economic activities in a given
  geographical territory?

4

• The ultimate question is
• not so much why a particular industry ends up
  in a particular place, but rather why
  manufacturing activities in general end up in a
  core region(s), leaving the other (s)
  relatively underdeveloped (periphery).
• INDUSTRIAL vs AGRICULTURAL regions
• Heart of EU structural policy problem

5
THREE FACTORS

• SHARE OF MANUFACTURING IN REGIONAL INCOME
  (determining the relative weight and importance
  of the crucial sector)
• DEGREES OF ECONOMIES OF SCALE (the higher it is,
  the higher the need for concentration so to
  exploit cost advantages)
• TRANSPORT COSTS (the higher they are, the less
  incentive to concentration, since it is
  convenient to locate close to their final market)

6
THE FORMAL FRAMEWORK

• TWO SECTORS
• Agricultural perfect competition, constant
  returns to scale, homogenous good traded
  costlessly.
• Manufacturing monopolistic competition,
  increasing returns to scale, transport costs.
• TWO REGIONS
• Labour force fixed in the short run, but
  responding to real wage movements in the
  long-run.
• MICROFOUNDED OPTIMIZING BEHAVIOUR.

7

• The arising of core-periphery pattern depends on
  three exogenous parameters
• 1) the share of demand going to manufacturing
• 2) the elasticity of substitution / degree of
  economies to scale
• 3) transport costs
• THEY ARE ALL EXOGENOUS.
• Sooner or later we always bump into something
  exogenous, especially in economic modelling.
• But number 3 is particularly unsatisfying,
  especially if coupled with the absence of public
  sector.

8
2. MARATTIN (2006) MODEL

• INSERT PUBLIC SECTOR INTO THE STANDARD FRAMEWORK
• - distortionary taxation on firms
• - public investment in infrastructure
• TRADE OFF Costs / benefits of public expenditure
• WHAT ARE THE EFFECTS OF THESE INNOVATIONS ON THE
  MODELS PREDICTION ABOUT AGGLOMERATION PATTERNS?

9
THE NOVELTY AND A WARNING

• The methodology and the solution methods relies
  heavily on Fingleton (2005), which simulated the
  standards Krugman 1991 model on Uk.
• That is virtually the only empirical application
  of a (standard) theoretical NEG model.
• Our attempt applies a more sophisticated model to
  six (regional or sub-regional) study areas.
• Policy results should be handled rather
  carefully.

10
THE FINAL SYSTEM OF EQUATIONS
11
Simulations

• Policy experiments on taxation
• Tax rate in the urban region 2
• 4
• Tax rate in the rural region 2
• 4

12

• Policy experiments on labour supply
• Labour supply in the urban region 5
• 
  5 (intra)
• 
  15
• 
  15 (intra)
• Labour supply in the rural region 5
• 
  15

13
THE STUDY AREAS
MEMBER STATE URBAN AREA RURAL AREA
ITALY Ferrara Basso Ferrarese
GREECE Heraklion Archanes
LATVIA Resekne/Daugavpilis Latgale region
CZECH REP. Ostrava Bruntal
FINLAND Joensuu North Karelia
SCOTLAND Inverness East Highlands
14

• Model is calibrated on real data collected by
  partners.
• Structural parameters are calibrated as in
  Fingleton (2005) or as standard in related
  literature.
• Results show the quantitative effects on
• 1) Income differentials
• 2) Price differentials
• 3) Nominal wage differentials
• 4) Real wage differentials
• We show only 1) and 4).

15
FORCES AT WORK

• 1)Home market effect nominal wage is higher in
  the bigger region
• 2) Competition effect nominal wage is lower in
  the bigger region
• 3) Price index effect prices are lower in the
  bigger region
• (1) and (3) pushes towards divergence
• (2) pushes towards convergece
• OUR MODIFICATION (distorsionary taxation to
  finance infrastracture) MESSES EVERYTHING UP,
  TWICE.

16
WHY TWICE?

• 1) BEING DISTORTIONARY, ALTERS NOMINAL AND REAL
  VARIABLES
• 2) ITS REVENUE IS USED TO REDUCE TRANSPORT COSTS,
  WHICH IN TURN IS ANOTHER FORCE PUSHING TOWARDS
  CONVERGENCE/DIVERGENCE.
• SO TRADITIONAL NEG RESULTS CAN BE SIGNIFICANTLY
  MODIFIED.

17
3.RESULTS AND POLICY

• 3 a RESULTS FOR TAXATION
• 3 b RESULTS FOR MIGRATION

18
ITALY - TAX
ITALY - TAX ?Y1 ?Y2 ??
t1 2 -1.12 / 12.97
t1 4 -2.19 / 23.37
t2 2 / -1.10 -1.71
t2 4 / -2.12 -9.58
19
GREECE - TAX
GREECE - TAX ?Y1 ?Y2 ??
t1 2 -1.17 / 12.29
t1 4 -2.31 / 25.82
t2 2 0.05 -0.18 -9.26
t2 4 0.1 -0.36 -17.54
20
LATVIA - TAX
LATVIA - TAX ?Y1 ?Y2 ??
t1 2 -1.15 / 19.72
t1 4 -2.25 / 41.17
t2 2 / -1.04 -18.32
t2 4 / -2.04 -34.98
21
CZECH REP. - TAX
CZ.REP - TAX ?Y1 ?Y2 ??
t1 2 -1.06 / 36.26
t1 4 -2.06 / 75.56
t2 2 / -0.8 -34.12
t2 4 / -1.61 -65.12
22
FINLAND - TAX
FINLAND - TAX ?Y1 ?Y2 ??
t1 2 -0.84 / 18.37
t1 4 -1.66 / 38.53
t2 2 / -0.76 -17.04
t2 4 / -1.49 -32.63
23
SCOTLAND - TAX
SCOTLAND - TAX ?Y1 ?Y2 ??
t1 2 -1.07 / 31.74
t1 4 -2.11 / 66.04
t2 2 / -1.24 -30.73
t2 4 / -2.43 -58.72
24
Sum-up for taxation
?Y ITALY GRE CZ.R FIN LATV SCOT
t1 2 -1.12 -1.14 -1.06 -0.84 -1.15 -1.07
t1 4 -2.19 -2.23 -2.06 -1.66 -2.25 -2.11
t2 2 1.10 0.13 0.8 0.76 1.04 1.24
t2 4 2.12 0.26 1.61 1.49 2.04 2.43
25
POLICY IMPLICATIONS

• Financing infrastracture via taxation in U
  affects negatively the income of U and does not
  significantly affect income in R.
• Income differentials reduce.
• Net effects (NEG three traditional ones
  distorsionary taxation) pushes towards divergence.

26

• Financing infrastracture via taxation in the R
  affects the income of R and does not affect
  income in U (exceptionGreece)
• Income differentials increase.
• Net effects pushes towards convergence, but
  with a less extent.
• HOW TO FINANCE INFRASTRACTURE BETWEEN URBAN AND
  RURAL REGION?
• Tax the urban income differential reduce, but
  push for divergence.
• Tax the rural income differential increase, but
  it is a help for convergence.

27
ITALY - LAB
ITALY-LAB. ?Y1 ?Y2 ??
L1 5 0 -3.99 13.99
L1 5intra 0 -8.27 21.85
L1 15 0 -11.03 27.50
L1 15intra 0 -22.6 57.35
L2 5 0 4.18 -1.12
L2 15 0 12.99 -10.01
28
LATVIA - LAB
LAT-LAB. ?Y1 ?Y2 ??
L1 5 0 -3.58 12.96
L1 5intra 0 -10.23 41.29
L1 15 0 -9.79 41.25
L1 15intra 0 -28.01 160.51
L2 5 0 3.75 -12.24
L2 15 0 11.27 -33.25
29
CZ.REP. - LAB
CZ.R.-LAB. ?Y1 ?Y2 ??
L1 5 0 -3 23.49
L1 5intra 0 -12.43 15.89
L1 15 0 -8.24 70.88
L1 15intra 0 -34.01 57.33
L2 5 0 3.17 -21.97
L2 15 0 9.50 -60.01
30
FINLAND - LAB
FINL-LAB. ?Y1 ?Y2 ??
L1 5 0 -3.91 13.13
L1 5intra 0 -10.09 38.24
L1 15 0 -9.89 37.36
L1 15intra 0 -27.13 143.78
L2 5 0 3.38 -10.4
L2 15 0 10.73 -30.3
31
SCOTLAND - LAB
SCOT.-LAB. ?Y1 ?Y2 ??
L1 5 0 -3.77 21.33
L1 5intra 0 -9.73 60.31
L1 15 0 -10.31 64.64
L1 15intra 0 -26.64 225.92
L2 5 0 3.96 -20.37
L2 15 0 11.87 -54.65
32
Sum-up for migration
?Y ITALY CZ.R FIN LATV SCOT
L1 5 3.99 3 3.91 3.58 3.77
L1 5intra 8.27 12.43 10.09 10.23 9.73
L1 15 11.03 8.24 9.89 9.79 10.31
L1 15intra 22.6 34.01 27.13 28.01 26.64
L2 5 -4.18 -3.17 -3.38 -3.75 -3.96
L2 15 -12.99 -9.50 -10.73 -11.27 -11.87
33
POLICY IMPLICATIONS

• Increase in the size of U increases the income
  differentials, much more if it occurs at the
  expense of R (internal migration)
• It promotes the cumulative effect for divergence.
• Specularly, increasing the size of R reduces the
  income differentials and pushes for convergence.

34
CONCLUSIONS

• 1) Allowing for endogenous provisions of
  infrastructure (via distiortionary taxation) can
  significantly affect traditional NEG conclusions
• 2) In the six study areas, we achieve
  qualitatively identical and quantitatively
  not-too-dissimilar results.
• 3) The choice of who bears the burden of
  infrastructure building is relevant for income
  differentials and convergence/divergence.
• 4) Effects on real wage differentialts must, in
  real life, be mitigated because simulations were
  run under the assumption that ALL tax revenue
  goes into infrastructure building.