Transcript Diversification

Natural Resources, Energy
Supply and Economic
Growth:
What Does Diversification
Achieve?
Bankole Fred Olayele
Background/Motivation
 Economic
structure
growth depends on economic
 Resource-based
 Diversification
western Canadian economy
seen from two perspectives:
o Long-term growth strategy that can help mitigate
unforeseen problems in the event of structural
economic changes?
o Costly and unnecessary form of government
intervention?
2
Diversification
A
recurring theme in public
policy debates
 Cure
to the “resource
curse” challenge?
 Benefits
well known; one
puzzle lingers
Puzzle: Diversification helps some to succeed where
others fail!
Inconclusive empirical evidence on the resource-diversitygrowth nexus.
3
Why Canada and the US?
 Similar, yet
distinct,
resource endowments,
technology,
demographics and
institutions.
 Two
federations with
two or more orders of
government acting
directly.; flexible
regional economic
policy making.
 Ideal
for panel data
analysis!
4
Methodology
No single explicit framework
Different measures and concepts ; depends largely on the
theoretical foundation explored.
Popular models include: ec. dev. theory, portfolio theory,
regional business cycle theory, trade models, location
theory, economic base theory and input-output analysis.
Our approach

Compare regional employment distribution across
industries with the national average.

Sectoral composition of national employment is dynamic;
this defines the limits of diversification.

A region’s employment is taken to be more specialized
(or less diversified) than that of the “parent” nation.
5
Diversity Indices
Entropy
Index (ENT)
Hirschman-Herfindahl Index(HHI)
Absolute Ogive Index (AGV)
Quadratic Ogive Index (QGV)
Krugman Index (KRUG)
Notes:
1. Indices
highly sensitive to the number of industries used.
2. Four and six broad categories for goods- and servicesproducing sectors.
3. Strategy helps achieve greater data aggregation.
4. Also overcomes missing data issues.
6
Variables/Data
Variables
 Economic growth: per capita real GDP
 Natural resources/energy supply: mining as a share of GDP
 Economic diversity: five diversity indices
 Human capital stock: educational attainment
 Physical capital stock: gross capital formation under PIM
Employment data
 Labour Force Survey (Statistics Canada)
 Current Population Survey (Bureau of Labour Statistics)
GDP/EXR data
 Regional Economic Accounts (US Bureau of Economic Analysis)
 Provincial Economic Accounts (Statistics Canada)
 Rates and Statistics (Bank of Canada)
Summary
 Annual panel data set; eight three-year intervals from 1987 to 2010
 All 60 jurisdictions;1987-97 based on SIC,1998-2010 based on NAICS.
7
Sectors of Interest
a)
Agriculture, forestry, fishing, and hunting
b)
Mining
c)
Construction
d)
Manufacturing
e)
Wholesale trade
f)
Retail trade
g)
Transportation, warehousing and utilities
h)
Finance, insurance, real estate, rental and leasing
i)
All other services, except public administration
j)
Government and government enterprises
Notes: The mining sector comprises of establishments primarily engaged in extracting
naturally occurring minerals. These can be solids, such as coal and ores; liquids, such
as crude petroleum; and gases, such as natural gas. Natural resource-energy supply
nexus needs further clarification. An alternative variable less prone to productivity
biases is sectoral GDP distribution. However, GDP itself is likely susceptible to
measurement errors and exchange rate biases.
8
Mining Share of Production (2010)
9
Mining Share of Production (1990-2010)
10
Properties of Indices
Indices
Absolute/Relative
Reverse Order
Upper Bound
Lower Bound
Decomposability
Absolute
No
Unity
1/N
Yes
Absolute
No
Zero
No
Absolute
No
Zero
No
Absolute
Yes
Zero
Yes
Relative
No
Zero
No
HHI
AGV
QGV
ENT
Natural log of N
KRUG
Notes: The reference level for absolute measures is the equal distribution of
employment shares across all industries; relative specialization measures are based
on the average economic structure of the jurisdictions. For HHI, the index
increases with the degree of specialization, and reaches its upper limit of 1 when a
region is specialized in only one industry. In that case, the lowest level of
specialization is indicated by 1/N i.e. the lowest degree of specialization indicated
by an equiproportional employment share for each industry. Successively higher
values of the indices imply successively lower degrees of diversity; the only
11
exception to this rule being the Entropy index.
Ranking: US States
Panel A
Five Most Diverse
Entropy
Absolute Ogive
Quadratic Ogive
HHI
Krugman
Wyoming
Rhode Island
Wyoming
Wyoming
Missouri
Alaska
Delaware
North Dakota
North Dakota
Georgia
North Dakota
Wyoming
Iowa
Iowa
Minnesota
Texas
Iowa
South Dakota
South Dakota
California
Arkansas
Arkansas
Oklahoma
Oklahoma
Oregon
Rhode Island
Hawaii
Massachusetts
Massachusetts
Wyoming
Massachusetts
New York
Nevada
Rhode Island
Alaska
Hawaii
New Mexico
New York
Nevada
North Dakota
New York
Maryland
Rhode Island
New York
Mississippi
Nevada
Nevada
Florida
Maryland
West Virginia
Highest Index Value
1.89
0.94
2.07
0.31
0.35
Lowest Index Value
1.58
0.67
0.83
0.18
0.04
Panel B
Five Least Diverse
12
Ranking: Canadian Provinces
Entropy
Absolute Ogive
Quadratic Ogive
HHI
Krugman
Saskatchewan
Saskatchewan
Saskatchewan
Saskatchewan
Ontario
Alberta
Alberta
Alberta
Alberta
Manitoba
Manitoba
Manitoba
Manitoba
Prince Edward
Quebec
New Brunswick
New Brunswick
Prince Edward
Manitoba
British Columbia
Ontario
Ontario
New Brunswick
New Brunswick
New Brunswick
Quebec
Prince Edward
Newfoundland
Newfoundland
Nova Scotia
British Columbia
British Columbia
Ontario
Ontario
Alberta
Newfoundland
Quebec
Quebec
Quebec
Newfoundland
Prince Edward
Newfoundland
British Columbia
British Columbia
Saskatchewan
Nova Scotia
Nova Scotia
Nova Scotia
Nova Scotia
Prince Edward
Highest Index Value
1.94
0.86
1.73
0.27
0.21
Lowest Index Value
1.68
0.65
1.12
0.21
0.06
13
Model Specification

14
Two-Step System GMM Results
Dependent Variable
lnRGDP
lnRGDP
lnRGDP
lnRGDP
lnRGDP
Diversity Index Used
Herfindahl
Absolute Ogive
Quadratic Ogive
Entropy
Krugman
Log of lagged RGDP
0.630**
0.943***
0.448**
0.720***
0.700**
[0.260]
[0.322]
[0.031]
[0.252]
[0.308]
1.275
0.055
1.382*
2.482
-0.135
[2.120]
[1.735]
[0.803]
[5.376]
[0.395]
0.070
-0.019
0.070
-0.379
-0.073
[0.469]
[0.091]
[0.092]
[0.330]
[0.060]
Diversity
Natural resources
Diversity x natural resources
0.007
-0.143
0.073
0.685
-0.980
[0.290]
[0.367]
[0.157]
[0.512]
[0.953]
-0.264
-0.335
-0.474**
-0.031
-0.199
[0.338]
[0.286]
[0.221]
[0.227]
[0.497]
-0.134
-0.439
-0.076
-0.107
-0.297
[0.170]
[0.378]
[0.240]
[0.454]
[0.619]
6.116
0.104
6.118
1.457
2.087
[6.000]
[4.046]
[2.671]
[1.598]
[4.238]
Year dummies
Yes
Yes
Yes
Yes
Yes
Year fixed effects
No
No
No
No
No
Number of observations
360
360
360
360
360
Number of jurisdictions
60
60
60
60
60
Number of instruments
16
16
16
16
16
Number of lags used
6
6
6
6
6
Sargan Test (p value)
0.003
0.056
0.249
0.266
0.821
Hansen Test (p value)
0.003
0.020
0.264
0.016
0.672
Arellano-Bond AR(1) ( p-value)
0.358
0.430
0.007
0.254
0.561
Arellano-Bond AR(2) ( p-value)
0.726
0.157
0.077
0.314
0.019
Capital stock
Educational attainment
Constant
Specification Tests
Note: All estimations based on the Windmeijer’s (2005) finite sample correction to the
standard errors.
15
Resource-Diversity-Growth Nexus

16
Alternative Diversity Measures
Dependent Variable
lnRGDP
lnRGDP
lnRGDP
lnRGDP
lnRGDP
Diversity Index Used
Herfindahl
Absolute Ogive
Quadratic Ogive
Entropy
Krugman
0.924***
0.633**
0.718***
0.755***
0.555**
[0.257]
[0.259]
[0.266]
[0.128]
[0.241]
-1.567
1.200*
-0.011
-2.703**
0.549
[1.229]
[0.615]
[0.717]
[1.334]
[0.348]
Lagged log GDP
Diversity
Natural resources
Diversity x natural resources
Capital stock
Educational attainment
Constant
-0.097
0.036
-0.014
0.068
0.072
[0.384]
[0.050]
[0.070]
[0.109]
[0.113]
-0.004
-0.026
-0.056
-0.047
0.072
[0.218]
[0.069]
[0.075]
[0.157]
[0.071]
-1.040**
0.256
-0.130
-0.075
-0.016
[0.431]
[0.276]
[0.412]
[0.272]
[0.222]
-1.236**
0.028
-0.355
-0.328
0.153
[0.527]
[0.475]
[0.559]
[0.216]
[0.491]
-4.727
4.693
2.512
4.137**
5.621
[5.155]
[3.610]
[3.862]
[1.850]
[3.544]
Year dummies
Yes
Yes
Yes
Yes
Yes
Year fixed effects
No
No
No
No
No
Number of observations
360
360
360
360
360
Number of jurisdictions
60
60
60
60
60
Number of instruments
16
16
16
16
16
Number of lags used
6
6
6
6
6
Sargan Test (p value)
0.244
0.188
0.013
0.196
0.426
Hansen Test (p value)
0.270
0.479
0.029
0.058
0.245
Arellano-Bond Test for AR(1) ( p-value)
0.074
0.155
0.132
0.172
0.416
Arellano-Bond Test for AR(2) ( p-value)
0.886
0.082
0.098
0.011
0.000
Specification Tests
Notes: All estimations based on the Windmeijer’s (2005) finite sample correction to the
standard errors. We also model all five diversity indices as strictly exogenous, IV-style,
regressors.
17
Conclusions
 All
five indices are quite arbitrary because both the
absolute and relative specialization measures are arbitrary.
 Results
suggest the growth-promoting stance of economic
diversity.
 The
GMM framework does not allow us to test the
resource curse proposition. Same with the interactive effect
of diversity on resources.
 Jurisdictions
with KRUG value less than 0.209 will suffer
from the curse; those above will not.
 Conclusion
qualified due to endogeneity not addressed by
the fixed effects technique employed.
18
Future Work
 Spatial
autocorrelation effects among regions would
be critical in explaining any link between diversity
and growth.
 Pede
(2013) concludes that spatial econometrics
provides a framework for the true factors at the
origin of spillovers to be modeled by geographical
distance.
 Future
work will consider applying spatial
econometric techniques.
 Among
other things, this strategy will add
robustness by offering a basis for comparison with
the few DPD-based studies out there.
19