# RECORD c0c750e0ff9e1fcc59888eee28b0b7208f559964
# TITLE: Essays on housing markets and economic *growth
# YEAR: 2010.0  VENUE: nan  DOI: 10.33915/etd.4652
# AUTHORS: Arzu Şen
# PAGES: 126  CANDIDATE TABLE PAGES (1-based): [35, 36, 56, 59, 60, 68, 88, 92, 93, 94, 95, 96, 104, 105]
# ABSTRACT: Essays on Housing Markets and Economic Growth Arzu Sen This dissertation is composed of three essays examining housing markets and economic growth, while carefully considering the role of space. The first chapter serves as an introduction and briefly discusses how each essay contributes to the literature. The second chapter estimates the relationship between residential property values and proximity to coal mines in Monongalia County, West Virginia. The study utilizes a spatial hedonic price model complemented with GIS techniques to estimate the marginal willingness to pay for properties surrounding coal mines. Study findings indicate that proximity to a coal mine translates negatively into property values. The third chapter is related to spatial county growth. Growth theories do not fully specify the prominent factors underlying the data-generating process for growth regressions. Introduction of space in growth regressions further complicates the estimations by adding uncertainty regarding the use of an appropriate spatial weight matrix and spatial regression specification. This study applies Markov Chain Monte Carlo model composition with the Bayesian model averaging methodology on a sample of U.S. counties, to deal with model uncertainty in spatial growth regressions. This study reports model averaged estimates to resolve the uncertainty pertaining to the determinants of U.S. county growth. The fourth chapter is related to the resource curse. An extensive literature has examined the presence and the possible causal mechanisms of a resource curse using a resource dependence indicator. However, a strand of the literature argues that switching from relative measures of resource abundance to absolute measures of resource abundance makes the resource curse disappear across countries. This study contributes to this strand of literature by examining whether coal abundance is a curse or a blessing for county economic growth, using both an absolute and a relative measure of resource abundance. Unlike previous research on resource curse, this study employs spatial county growth regressions to account for spatial dependence. Study findings suggest that introducing spatial dependence into growth regressions changes the results from non-spatial models drastically. When measured as a relative variable, coal dependence has a significant positive direct impact on own-county growth, and positive spillovers on related counties’ growth. When measured as an absolute variab


===== PAGE 1 (score 1, numeric-paren 0, table-words 0, var-words 1) =====

Graduate Theses, Dissertations, and Problem Reports (ETD) 
2010 
Essays on housing markets and economic *growth 
Essays on housing markets and economic *growth 
Arzu Sen 
West Virginia University 
Follow this and additional works at: https://researchrepository.wvu.edu/etd 
Recommended Citation 
Recommended Citation 
Sen, Arzu, "Essays on housing markets and economic *growth" (2010). Graduate Theses, Dissertations, 
and Problem Reports (ETD). 4652. 
https://researchrepository.wvu.edu/etd/4652 
This Dissertation is protected by copyright and/or related rights. It has been brought to you by the The Research 
Repository @ WVU with permission from the rights-holder(s). You are free to use this Dissertation in any way that is 
permitted by the copyright and related rights legislation that applies to your use. For other uses you must obtain 
permission from the rights-holder(s) directly, unless additional rights are indicated by a Creative Commons license 
in the record and/ or on the work itself. This Dissertation has been accepted for inclusion in WVU Graduate Theses, 
Dissertations, and Problem Reports (ETD) collection by an authorized administrator of The Research Repository @ 
WVU. For more information, please contact researchrepository@mail.wvu.edu. 


===== PAGE 35 (score 147, numeric-paren 114, table-words 1, var-words 15) =====

Table 2.4: Direct, Indirect, and Total Effects Estimates
Variable
Direct Effects
Indirect 
Effects
Total Effects
Coal Dummy_0.75
-0.154***
        (0.00)
        -0.008***
          (0.00) 
-0.162***
(0.00)
Coal Dummy_0.75~1.00
 -0.105***
(0.00)
     -0.005**
             (0.02) 
-0.110**
(0.02)
Coal Dummy_1.00~1.25
    - 0.093**
              (0.01) 
     -0.005**
 (0.03)
-0.099**
(0.01)
Coal Dummy_1.25~1.50
-0.020
(0.57)
-0.001
(0.62)
-0.021
(0.57)
Building size
0.214***
(0.00)
0.011*
(0.06)
0.225***
(0.00)
Building size squared
-0.020***
(0.00)
-0.001*
(0.09)
-0.021***
(0.00)
Lot size 
0.004**
(0.03)
0.000
(0.17)
0.004**
(0.03)
Lot size squared
-0.000
(0.18)
-0.000
(0.30)
-0.000
(0.18)
Age
-0.105
(0.16)
-0.005
(0.30)
-0.110
(0.17)
Age squared
0.028
(0.67)
0.001
(0.71)
0.075
(0.36)
Excellent Condition
0.459***
(0.00)
0.023*
(0.06)
0.482***
(0.00)
Good Condition
0.337***
(0.00)
0.017*
(0.08)
0.355***
(0.00)
Average Condition
0.249***
(0.00)
0.013
(0.11)
0.262***
(0.00)
Fair Condition
-0.026
(0.75)
-0.001
(0.78)
-0.027
(0.75)
Bedrooms
0.010
(0.30)
0.000
(0.39)
0.010
(0.30)
Bathrooms
     0.029**
 (0.01)
0.001
(0.14)
    0.030**
(0.01)
Heating
0.044
(0.33)
0.002
(0.43)
0.046
(0.33)
Fireplace
0.028*
(0.06)
0.001
(0.19)
0.028*
(0.06)
Full Basement
-0.010
(0.62)
-0.000
(0.66)
-0.011
(0.62)
Part Basement
-0.004
(0.86)
0.000
(0.88)
-0.004
(0.86)
No Basement
0.049*
(0.06)
0.003
(0.20)
0.052*
(0.06)
Urbanized Area Dummy 
0.058***
(0.00)
0.003
(0.10)
0.061***
(0.00)
Lake Dummy
0.102***
(0.00)
0.020***
(0.00)
0.122***
(0.00)
Median Income
-0.004
(0.54)
-0.000
(0.60)
-0.005
(0.54)
White
0.074 
(0.62)
0.004
(0.66)
0.078
(0.62)
College Graduates
-0.030
(0.66)
-0.001
(0.72)
-0.032
(0.66)
27


===== PAGE 36 (score 20, numeric-paren 5, table-words 3, var-words 8) =====

Poverty
-0.036
(0.56)
-0.002
(0.63)
-0.036
(0.56)
R-squared
                  0.50
ρ
                  0.21***
                 (0.00)
Table Notes: Numbers in parentheses are p-values based on heteroskedasticity-consistent standard errors, 
and ***, **, and * represent statistical significance at the 99%, 95%, and 90% level, respectively.
The variables of focus in this study are the coal mine buffer variables. The model 
predicts  that  a  negative  buffer  variable  estimate  is  associated  with  disamenity 
characteristics of close proximity to coal mines since the parameter implies lower house 
prices at the mean, all else constant. On the other hand, a positive estimate points to 
amenity impacts of coal mines. The results support the disamenity impacts of coal mines, 
even after controlling for neighborhood variables. The direct effects estimates show the 
average percentage increase in house sales prices resulting from a one unit change in 
various house-specific explanatory variable characteristics. In particular, the direct effect 
of being located within ¾ mile of a coal mine is estimated as -0.15, implying that at the 
mean, being located close to a coal mine reduces own-house price by 15 percent, all else 
constant. Purchasing a house located farther from a coal mine is associated with a lower 
discount. All else constant, houses located within ¾-1 mile and 1-1¼ miles of a coal mine 
experience 11 percent and 10 percent lower average own-house prices respectively. 
Beyond 1¼ miles, the proximity effects are no longer significant.   
Among the variables found to be significant in influencing house prices are house 
condition variables, building size, lot size, number of bathrooms, the presence of a 
fireplace,  the  urbanized  area,  and  the  lake  dummies.  In Table  4,  the  direct  effect 
magnitude of 0.46 for the excellent house condition implies that the mean house sales 
price is 46 percent higher compared with a house in poor condition, all else constant. 
28


===== PAGE 37 (score 11, numeric-paren 0, table-words 1, var-words 9) =====

Building  size  is  another  important  factor  influencing  house  prices.  A  direct  effect 
magnitude of 0.21 implies that a thousand square feet increase in house size results in 21 
percent increase in average house sales prices. Type of basement, number of bedrooms, 
the presence of a heating system, and neighborhood variables fail to explain the variation 
in house prices. 
Indirect effects estimates presented in the above table represent spatial spillovers 
and  capture  neighboring-house  effects  associated  with  changes  in  own-house  level 
characteristics. LeSage and Pace (2009) explain that indirect effects estimates can be 
interpreted in two different ways. The first interpretation is named as the average total 
impact  on  an  observation.  This  interpretation  captures  how  a  change  in  the  own-
explanatory variable characteristics of all houses by some constant amount would change 
the price of a typical house. The second interpretation is named as the  average total 
impact from an observation, which captures the cumulative impact of a change in own-
house explanatory variable characteristics averaged over all other houses.
Positive indirect effect estimates are indicative of positive spatial externalities, 
whereas negative estimates can be considered as negative spatial externalities. Using the 
second interpretation, the indirect impact magnitude of -0.008 points to negative spillover 
effects of being located within ¾ mile of a coal mine on neighboring-house prices. The 
impact estimates for spillover effects associated with houses near coal mines disappear 
beyond 1¼ miles. House condition variables and building size are significant sources of 
positive spillovers. For instance, compared with a house in poor condition, an excellent 
condition of own-house increases the sales price of neighboring houses by 2 percent.    
29


===== PAGE 56 (score 17, numeric-paren 0, table-words 2, var-words 8) =====

Table 3.3: Descriptive Statistics
Variable
Mean
Std. Dev.
Min
Max
Dependent Variable
15,075.30
3,223.56
5,910.95
45,621.42
Initial Income 
12,638.88
2,884.86
3,793.31
30,290.42
Agriculture, Forestry, Fisheries and 
Mining Employment 
1,272.82
2,202.38
6
64,060
Construction Emp. 
1,814.40
5,531.72
1
154,612
Manufacturing Emp. 
7,040.13
26,224.56
1
884,139
Transportation, Communication and 
other Public Utilities Emp. 
2,234.80
8,447.25
2
248,416
Wholesale and Retail Emp. 
6,328.98
22,380.66
2
700,108
Finance, Insurance and Real Estate Emp. 
1,866.13
8,344.95
1
249,271
Business and Repair Services Emp. 
1,285.49
5,925.64
1
203,265
Personal, Entertainment and 
Recreational Services Emp. 
1,274.45
5,208.80
1
195,217
Health Services Emp. 
2,296.69
8,238.12
1
250,413
Educational Services Emp. 
2,655.39
8,274.17
3
250,293
Other Professional and Related Services 
Emp. 
1,312.12
5,471.20
1
170,887
Federal Government Emp. 
1,114.11
4,047.05
1
91,087
State Government Emp. 
1,406.19
3,763.21
1
82,658
Local Government Emp. 
2,680.50
9,387.05
12
297,074
Self Employment 
2,139.83
6,189.09
11
235,754
Poverty 
8,862.83
  31,421.4
6 
1
1,003,390
Human Capital 
11,558.57
45,110.56
3.91
1,383,289
Population Density 
182.26
1,562.40
0.14
62,674.59
Area 
980.75
1,320.91
22.79
20,174.72
NAIX
0.05
2.30
-6.4
11.17
College town Dummy
0.02
0.133414
0
1
48


===== PAGE 57 (score 7, numeric-paren 0, table-words 0, var-words 7) =====

 It  is  commonly  acknowledged  that  human  capital  is  a key determinant  of 
economic growth. Following LeSage and Fischer (2008), human capital is measured by 
the skill of the workforce as measured by the log of the number of people with a college 
degree. Following Fingleton (2001), logged population density as well as logged area is 
used. Regions with higher population density would capture agglomeration effects on 
economic growth. Moreover, a poverty variable, which is the logged population below 
poverty line is included.   
As another control, a Natural Amenity Index (NAIX) created by McGranahan 
(1999) is employed.  NAIX is calculated from standardized mean values of climate 
measures  (January  temperature,  January  days  of  sun,  July  temperature  and  July 
humidity), topographic variation and water area as a proportion of county area (see: 
http://www.ers.usda.gov/Data/NaturalAmenities). A higher NAIX score implies higher 
amenities. 
 Following Higgins et al. (2006), a college town dummy variable is created to 
control for higher incomes in counties which may result from holding a considerable 
amount  of  advanced  degree  holders.  For  all  colleges  and  universities  with  a  total 
enrollment of at least 10,000 students, the ratio of number of students enrolled to 
county’s 1980 population is calculated. The county’s dummy is assigned a value of 1 if 
its enrollment to population ratio is at least 0.2 and a value of 0 otherwise.  
Finally, this study does not incorporate state dummies. Inclusion of state dummies 
results with many of the explanatory variables representing dummy variables, producing 
an invertability problem14. 
14 See LeSage (1999) for details on invertability problem arising from dummy variables. 
49


===== PAGE 59 (score 132, numeric-paren 106, table-words 2, var-words 9) =====

Table 3.4: Direct, Indirect and Total Effects Estimates with Contiguity Weight 
Matrix
Variables
Direct Effects
Indirect Effects
Total Effects
Initial Income
   0.5874***
(0.00)
-0.0197
(0.63)
     0.5677***
(0.00)
Agriculture, Forestry, 
Fisheries, Mining Emp.
   -0.0133***
(0.00)
   -0.0227**
(0.01)
     -0.0360***
(0.00)
Construction Emp.
0.0045
(0.37)
-0.0217
(0.11)
-0.0172
(0.23)
Manufacturing Emp.
-0.0049
(0.10)
0.0032
(0.64)
-0.0017
(0.80)
Transportation, 
Communication and other 
Public Utilities Emp.
0.0040
(0.33)
  -0.0363**
(0.01)
  -0.0322**
(0.03)
Wholesale and Retail Trade 
Emp.
0.0019
(0.82)
 0.0591*
(0.05)
 0.0609*
(0.08)
Finance, Insurance and Real 
Estate Emp.
   0.0248***
(0.00)
0.0165
(0.34)
  0.0413**
(0.03)
Business and Repair 
Services Emp.
   0.0145***
(0.00)
0.0057
(0.69)
0.0201
(0.20)
Personal, Entertainment and 
Recreation Services Emp.
  0.0095**
(0.03)
0.0181
(0.22)
 0.0276*
(0.09)
Health Services Emp.
  0.0098**
(0.01)
0.0164
(0.17)
 0.0262*
(0.05)
Educational Services Emp.
   -0.0339***
(0.00)
0.0431
(0.10)
 0.0091*
(0.09)
Other Professional and 
Related Services Emp.
0.0052
(0.20)
-0.0030
(0.84)
0.0022
(0.90)
Federal Government 
Employment
0.0025
(0.33)
   0.0211***
(0.00)
    0.0236***
(0.00)
State Government 
Employment
   -0.0131***
(0.00)
0.0058
(0.51)
      -0.0073
(0.43)
Local Government 
Employment
 -0.0141**
(0.04)
0.0004
(0.98)
      -0.0137
(0.53)
Self Employment
   0.0803***
(0.00)
-0.0063
(0.64)
    0.0740***
(0.00)
Poverty
   -0.0256***
(0.00)
0.0144
(0.20)
      -0.0112
(0.15)
Human Capital
   0.1133***
(0.00)
  -0.0397*
(0.08)
    0.0736***
(0.00)
College town Dummy
   -0.0194***
(0.00)
    -0.0652***
(0.00)
      -0.0846***
(0.00)
Population Density
   -0.1850***
(0.00)
-0.0481
(0.18)
    -0.2331***
(0.00)
Area
   -0.1967***
(0.00)
   -0.0823**
(0.03)
    -0.2788***
(0.00)
NAIX
0.0002
(0.60)
   -0.0021**
(0.03)
      -0.0018**
(0.03)
R-squared
0.8255
ρ
0.5315***
(0.00)
Ambiguous estimates across different specifications are highlighted with bold fonts.  
p-levels are reported in parenthesis. 
***Significant at the 99% level
**Significant at the 95% level
*Significant at the 90% level
51


===== PAGE 60 (score 125, numeric-paren 99, table-words 2, var-words 9) =====

Table 3.5: Direct, Indirect and Total Effects Estimates with 14 Nearest Neighbors
Variables
Direct Effects
Indirect Effects
Total Effects
Initial Income
    0.5920***
(0.00)
-0.0105
(0.88)
    0.5815***
(0.00)
Agriculture, Forestry, 
Fisheries, Mining Emp.
   -0.0131***
(0.00)
-0.0219
(0.10)
    -0.0350***
(0.00)
Construction Emp.
0.0057
(0.24)
        -0.0084
(0.70)
-0.0028
(0.90)
Manufacturing Emp.
      -0.0043
(0.13)
0.0093
(0.39)
0.0049
(0.65)
Transportation, 
Communication and other 
Public Utilities Emp.
0.0058
(0.16)
 0.0429*
(0.08)
-0.0371
(0.15)
Wholesale and Retail Trade 
Emp.
0.0053
(0.52)
0.0798
(0.19)
0.0746
(0.24)
Finance, Insurance and Real 
Estate Emp.
   0.0256***
(0.00)
 0.0593*
(0.09)
   0.0850**
(0.02)
Business and Repair 
Services Emp.
   0.0199***
(0.00)
-0.0364
(0.19)
-0.0165
(0.57)
Personal, Entertainment and 
Recreation Services Emp.
0.0052
(0.21)
0.0204
(0.45)
0.0256
(0.36)
Health Services Emp.
   0.0090**
(0.01)
0.0213
(0.35)
0.0303
(0.20)
Educational Services Emp.
   -0.0347***
(0.00)
  0.1015**
(0.04)
 0.0668*
(0.06)
Other Professional and 
Related Services Emp.
0.0032
(0.44)
        -0.0071
(0.81)
      -0.0039
(0.90)
Federal Government 
Employment
0.0020
(0.42)
   0.0397***
(0.00)
    0.0416***
(0.00)
State Government 
Employment
   -0.0143***
(0.00)
0.0025
(0.87)
-0.0118
(0.46)
Local Government 
Employment
      -0.0134*
(0.05)
-0.0329
(0.36)
-0.0462
(0.21)
Self Employment
    0.0798***
(0.00)
-0.0077
(0.73)
    0.0722***
(0.00)
Poverty
   -0.0280***
(0.00)
0.0165
(0.21)
-0.0115
(0.31)
Human Capital
    0.1142***
(0.00)
 -0.0707*
(0.07)
0.0434
(0.28)
College town Dummy
   -0.0181***
(0.00)
    -0.1443***
(0.00)
   -0.1623***
(0.00)
Population Density
   -0.1771***
(0.00)
-0.0920
(0.14)
   -0.2691***
(0.00)
Area
   -0.1904***
(0.00)
 -0.1194*
(0.07)
   -0.3099***
(0.00)
NAIX
0.0006
(0.20)
 -0.0028*
(0.05)
-0.0022
(0.11)
R-squared
0.8185
ρ
0.6517***
(0.00)
Ambiguous estimates across different specifications are highlighted with bold fonts.  
p-levels are reported in parenthesis. 
***Significant at the 99% level
**Significant at the 95% level
*Significant at the 90% level
52


===== PAGE 61 (score 11, numeric-paren 0, table-words 0, var-words 11) =====

In  both  model  specifications,  own-county  employment  in  the  agricultural 
industries and educational services are negatively related to own-county growth. Own-
county employment in finance, insurance and real estate, business and repair services, 
and health services have a positive impact on own-county growth. One ambiguity arising 
from  different  weight matrix  specifications  pertains  to estimates  of employment  in 
wholesale and retail trade and entertainment and recreational services. Both estimates 
from the model with contiguity weight matrix specification are significant at 90 percent 
level, whereas estimates from the other model specification are not significant.  
Across two model specifications, state and local government employment in own-
county  has  a  significant  negative  impact  on  own-county  growth,  suggesting  an 
overexpansion of the state government. Interestingly, federal government in own-county 
does not hinder own-county growth. Self employment  and human capital  have the 
expected  positive  impacts  on own-county growth, with relatively large  magnitudes. 
Poverty is associated with lower income levels. Population density, which captures the 
agglomeration effects, has a strong negative impact on own-county growth. This result 
reflects the congestion effects and disamenities connected to large cities and denser areas. 
Another result of interest is that college towns are associated with lower income levels, 
capturing the impact of disproportionate number of students residing in college towns. 
Indirect  effects  estimates,  presented  in  the  above  tables,  represent  spatial 
spillovers and capture related-county effects associated with changes in own-county level 
characteristics. LeSage and Pace (2009) explain that indirect effects estimates can be 
interpreted in two different ways. The first interpretation is named as the average total 
impact on an observation. This interpretation captures how a change in the initial level of 
53


===== PAGE 68 (score 122, numeric-paren 98, table-words 1, var-words 9) =====

Table 3.7: Model Averaged Estimates
Variables
Direct Effects
Indirect Effects
Total Effects
Initial Income
   0.6199***
(0.00)
-0.0212
(0.81)
   0.5987***
(0.00)
Agriculture, Forestry, 
Fisheries, Mining Emp.
   -0.0153***
(0.00)
-0.0188
(0.13)
   -0.0341***
(0.00)
Construction Emp.
0.0083
(0.39)
-0.0132
(0.57)
-0.0049
(0.87)
Manufacturing Emp.
  -0.0074**
(0.03)
0.0053
(0.41)
  -0.0021**
(0.04)
Transportation, 
Communication and other 
Public Utilities Emp.
0.0045
(0.30)
-0.0699
(0.15)
-0.0654
(0.17)
Wholesale and Retail Trade 
Emp.
0.0025
(0.80)
   0.0472**
(0.03)
 0.0497*
(0.05)
Finance, Insurance and Real 
Estate Emp.
   0.0261***
(0.00)
0.0324
(0.29)
  0.0585**
(0.03)
Business and Repair 
Services Emp.
   0.0156***
(0.00)
0.0087
(0.58)
0.0243
(0.68)
Personal, Entertainment and 
Recreation Services Emp.
0.0066
(0.16)
0.0231
(0.50)
0.0297
(0.40)
Health Services Emp.
0.0104
(0.14)
0.0267
(0.42)
0.0371
(0.25)
Educational Services Emp.
   -0.0419***
(0.00)
  0.1172**
(0.02)
  0.0753**
(0.04)
Other Professional and 
Related Services Emp.
0.0045
(0.38)
-0.0028
(0.83)
0.0017
(0.88)
Federal Government 
Employment
0.0031
(0.42)
   0.0489***
(0.00)
0.0520
(0.10)
State Government 
Employment
   -0.0194***
(0.00)
0.0065
(0.49)
-0.0129
(0.45)
Local Government 
Employment
-0.0150
(0.11)
-0.0359
(0.39)
-0.0509
(0.20)
Self Employment
   0.0962***
(0.00)
-0.0051
(0.52)
   0.0911***
(0.00)
Poverty
   -0.0324***
(0.00)
0.0179
(0.21)
-0.0145
(0.26)
Human Capital
   0.1197***
(0.00)
  -0.0764**
(0.01)
   0.0433*** 
(0.00)
College town Dummy
   -0.0172***
(0.00)
   -0.0776***
(0.00)
   -0.0948***
(0.00)
Population Density
   -0.1902***
(0.00)
 -0.1069*
(0.07)
   -0.2971***
(0.00)
Area
   -0.1994***
(0.00)
  -0.0941**
(0.03)
   -0.2935***
(0.00)
NAIX
0.0002
(0.61)
 -0.0033*
(0.08)
-0.0031
(0.13)
p-levels are reported in parenthesis. 
***Significant at the 99% level
**Significant at the 95% level
*Significant at the 90% level
60


===== PAGE 69 (score 5, numeric-paren 0, table-words 0, var-words 5) =====

3.5 Conclusion
Spatial growth regression models produce estimates and inferences that are conditional 
on both the particular weight matrix used to specify the spatial dependence and the 
explanatory variables included in the growth regression. Employing different spatial 
weight matrices often lead to ambiguous results in spatial econometrics. This study 
applies  Monte  Carlo  Markov  Chain model  composition  MC3 with  Bayesian  model 
averaging  on  a  sample  of  U.S.  counties  to  deal  with  model  uncertainty in spatial 
regressions  and  produce  model  averaged  estimates  and  inferences.  The  study  first 
displays the results based on two spatial weight matrices, contiguity and optimal nearest 
neighbors,  and  discusses  the  ambiguity  resulting  from  different  weight  matrix 
specifications. Later, model averaged estimates are reported to resolve the uncertainty 
pertaining to the determinants of U.S. county growth. 
 
The Bayesian solution to model uncertainty clarifies the ambiguous estimates 
across different specifications. For instance, the model averaged estimates reveal that 
human capital imposes a positive direct impact and negative spillover effects on related 
counties, and it has a positively significant total impact. The methodology also proves 
fruitful as it clarifies the ambiguity pertaining to the significance of employment in 
agricultural  services,  wholesale  and  retail  trade,  transportation  and  communication 
services, finance and real estate, entertainment and recreational services, health services, 
and natural amenities index. 
61


===== PAGE 88 (score 18, numeric-paren 0, table-words 2, var-words 8) =====

Table 4.3: Descriptive Statistics
Variable
Mean
Std. Dev.
Min
Max
Dependent Variable
15,075.30
3,223.56
5,910.95
45,621.42
Coal production per capita in USD 
346.17
3,163.68
1
104,192.60
Coal production per income in USD 
0.03
0.25
0
10.11
Initial Income 
12,638.88
2,884.86
3,793.31
30,290.42
Agriculture, Forestry, Fisheries and 
Mining Employment 
1,272.82
2,202.38
6
64,060
Construction Emp. 
1,814.40
5,531.72
1
154,612
Manufacturing Emp. 
7,040.13
26,224.56
1
884,139
Transportation, Communication and 
other Public Utilities Emp. 
2,234.80
8,447.25
2
248,416
Wholesale and Retail Emp. 
6,328.98
22,380.66
2
700,108
Finance, Insurance and Real Estate Emp. 
1,866.13
8,344.95
1
249,271
Business and Repair Services Emp. 
1,285.49
5,925.64
1
203,265
Personal, Entertainment and 
Recreational Services Emp. 
1,274.45
5,208.80
1
195,217
Health Services Emp. 
2,296.69
8,238.12
1
250,413
Educational Services Emp. 
2,655.39
8,274.17
3
250,293
Other Professional and Related Services 
Emp. 
1,312.12
5,471.20
1
170,887
Federal Government Emp. 
1,114.11
4,047.05
1
91,087
State Government Emp. 
1,406.19
3,763.21
1
82,658
Local Government Emp. 
2,680.50
9,387.05
12
297,074
Self Employment 
2,139.83
6,189.09
11
235,754
Poverty 
8,862.83
31,421.46
1
1,003,390
Human Capital 
11,558.57
45,110.56
3.91
1,383,289
Population Density 
182.26
1,562.40
0.14
62,674.59
Area 
980.75
1,320.91
22.79
20,174.72
NAIX
0.05
2.29
-6.4
11.17
College town Dummy
0.02
0.13
0
1
80


===== PAGE 89 (score 13, numeric-paren 0, table-words 1, var-words 11) =====

The study utilizes a large dataset controlling for as many growth determinants as 
possible. All variables are measured as initial period values (i.e. 1980 values) to avoid 
endogeneity problems. Initial period level of income, one common variable included in 
growth regressions, is included to control for conditional convergence. This study uses 
per capita personal income data obtained from the Bureau of Economic Analysis (BEA) 
and expressed in 1990 U.S. dollars, using the CPI Research Series deflator. BEA income 
data is collected according to place of work and adjusted for place of residence, to make 
the income data compatible with Census data. The dependent variable is the income level 
rather than the growth rate, measured by the log of average per capita income for the 
period 1980-1999. 
This study controls for industry mix by incorporating 11 employment-by-industry 
variables21. The logged levels of employment are used, following Lopez-Bazo  et al. 
(2004) and LeSage and Fischer (2008). Following Higgins et al. (2006), the size of the 
public sector is controlled by employing the logged levels of employment in federal, state 
and local governments. Additionally, logged levels of self employed are included to 
obtain its marginal effect. 
The economic growth literature has often identified human capital as an important 
growth determinant. This study follows LeSage and Fischer (2008) and measures human 
capital by the log of the number of people with a college degree. Following Fingleton 
(2001), logged population density as well as logged area is used. Agglomeration effects 
on  economic  growth  may  be  captured  by  regions  with  higher  population  density. 
21 The industry groups are agriculture, forestry, fisheries and mining; construction; manufacturing of 
durables and nondurables; transportation, communication and other public utilities; wholesale and retail; 
finance, insurance and real estate; business and repair services; personal, entertainment and recreation 
services; health services; educational services, other professional and related services.
81


===== PAGE 92 (score 135, numeric-paren 108, table-words 2, var-words 9) =====

Table 4.4: Direct, Indirect and Total Effects Estimates with Contiguity Weight 
Matrix and an Absolute Measure of Coal Abundance
Variable
Direct Effects
Indirect Effects
Total Effects
Coal Production per Capita 
($) 
 0.0001
(0.88)
-0.0015
(0.43)
-0.0014
(0.46)
Initial Income
   0.5882***
(0.00)
-0.0165
(0.69)
   0.5717***
(0.00)
Agriculture, Forestry, 
Fisheries, Mining Emp.
   -0.0133***
(0.00)
  -0.0212**
(0.01)
   -0.0345***
(0.00)
Construction Emp.
0.0045
(0.36)
-0.0222
(0.11)
-0.0177
(0.22)
Manufacturing Emp.
 -0.0051*
(0.09)
0.0024
(0.74)
-0.0027
(0.10)
Transportation, 
Communication and other 
Public Utilities Emp.
0.0043
(0.31)
  -0.0336**
(0.02)
 -0.0293*
(0.05)
Wholesale and Retail Trade 
Emp.
0.0015
(0.86)
 0.0564*
(0.07)
 0.0579*
(0.09)
Finance, Insurance and Real 
Estate Emp.
   0.0244***
(0.00)
0.0148
(0.40)
  0.0392**
(0.04)
Business and Repair 
Services Emp.
   0.0145***
(0.00)
0.0040
(0.78)
0.0185
(0.23)
Personal, Entertainment and 
Recreation Services Emp.
  0.0096**
(0.02)
0.0183
(0.20)
 0.0279*
(0.07)
Health Services Emp.
  0.0098**
(0.01)
0.0172
(0.16)
  0.0270**
(0.04)
Educational Services Emp.
   -0.0339***
(0.00)
0.0425
(0.11)
0.0086
(0.77)
Other Professional and 
Related Services Emp.
0.0053
(0.20)
-0.0026
(0.86)
0.0027
(0.87)
Federal Government 
Employment
0.0025
(0.33)
   0.0207***
(0.00)
   0.0232***
(0.00)
State Government 
Employment
   -0.0130***
(0.00)
0.0056
(0.51)
-0.0074
(0.41)
Local Government 
Employment
  -0.0144**
(0.04)
-0.0030
(0.88)
-0.0174
(0.42)
Self Employment
   0.0802***
(0.00)
-0.0080
(0.56)
   0.0723***
(0.00)
Poverty
   -0.0255***
(0.00)
0.0141
(0.21)
-0.0114
(0.14)
Human Capital
   0.1132***
(0.00)
-0.0340
(0.08)
   0.0732***
(0.00)
College town Dummy
   -0.0193***
(0.00)
   -0.0638***
(0.00)
   -0.0831***
(0.00)
Population Density
   -0.1839***
(0.00)
-0.0394
(0.29)
   -0.2233***
(0.00)
Area
   -0.1955***
(0.00)
 -0.0742*
(0.06)
   -0.2697***
(0.00)
NAIX
0.0003
(0.59)
  -0.0020**
(0.03)
  -0.0017**
(0.04)
R-squared
0.8257
ρ
0.5297***
(0.0000)
p-levels are reported in parenthesis. 
***Significant at the 99% level
**Significant at the 95% level
*Significant at the 90% level
84


===== PAGE 93 (score 139, numeric-paren 112, table-words 2, var-words 9) =====

Table 4.5: Direct, Indirect and Total Effects Estimates with Contiguity Weight 
Matrix and a Relative Measure of Coal Dependence
Variable
Direct Effects
Indirect Effects
Total Effects
Coal Production per Income 
($)
   0.0079***
(0.00)
0.0081
(0.50)
0.0161
(0.22)
Initial Income
   0.5849***
(0.00)
-0.0224
(0.57)
   0.5625***
(0.00)
Agriculture, Forestry, 
Fisheries, Mining Emp.
   -0.0142***
(0.00)
   -0.0238***
(0.00)
   -0.0379***
(0.00)
Construction Emp.
0.0040
(0.41)
 -0.0239*
(0.08)
-0.0199
(0.16)
Manufacturing Emp.
-0.0034
(0.27)
0.0042
(0.55)
 0.0008*
(0.09)
Transportation, 
Communication and other 
Public Utilities Emp.
0.0033
(0.42)
  -0.0371**
(0.01)
  -0.0338**
(0.03)
Wholesale and Retail Trade 
Emp.
0.0042
(0.62)
  0.0621**
(0.04)
 0.0662*
(0.05)
Finance, Insurance and Real 
Estate Emp.
   0.0251***
(0.00)
0.0170
(0.32)
  0.0421**
(0.02)
Business and Repair 
Services Emp.
   0.0154***
(0.00)
0.0087
(0.53)
0.0241
(0.12)
Personal, Entertainment and 
Recreation Services Emp.
  0.0092**
(0.03)
0.0186
(0.20)
 0.0278*
(0.08)
Health Services Emp.
   0.0100***
(0.00)
0.0163
(0.18)
 0.0263*
(0.05)
Educational Services Emp.
   -0.0348***
(0.00)
0.0419
(0.10)
0.0071
(0.80)
Other Professional and 
Related Services Emp.
 0.0050*
(0.07)
-0.0041
(0.79)
0.0009
(0.96)
Federal Government 
Employment
0.0032
(0.21)
   0.0207***
(0.00)
   0.0239***
(0.00)
State Government 
Employment
   -0.0123***
(0.00)
0.0063
(0.46)
-0.0060
(0.50)
Local Government 
Employment
 -0.0131*
(0.09)
0.0038
(0.86)
-0.0093
(0.67)
Self Employment
   0.0815***
(0.00)
-0.0051
(0.70)
   0.0764***
(0.00)
Poverty
   -0.0258***
(0.00)
0.0146
(0.20)
-0.0111
(0.14)
Human Capital
   0.1139***
(0.00)
 -0.0398*
(0.07)
   0.0741***
(0.00)
College town Dummy
   -0.0196***
(0.00)
   -0.0657***
(0.00)
   -0.0853***
(0.00)
Population Density
   -0.1916***
(0.00)
-0.0551
(0.14)
   -0.2468***
(0.00)
Area
   -0.2027***
(0.00)
  -0.0889**
(0.02)
   -0.2916***
(0.00)
NAIX
0.0003
(0.53)
  -0.0021**
(0.03)
   -0.0018**
(0.03)
R-squared
0.8258
ρ
0.5300***
(0.0000)
p-levels are reported in parenthesis. 
***Significant at the 99% level
**Significant at the 95% level
*Significant at the 90% level
85


===== PAGE 94 (score 127, numeric-paren 100, table-words 2, var-words 9) =====

Table 4.6: Direct, Indirect and Total Effects Estimates with 14 Nearest Neighbors 
and an Absolute Measure of Coal Abundance
Variable
Direct Effects
Indirect Effects
Total Effects
Coal Production per Capita 
($)
-0.0001
(0.92)
-0.0022
(0.50)
-0.0022
(0.48)
Initial Income
   0.5917***
(0.00)
-0.0055
(0.93)
   0.5862***
(0.00)
Agriculture, Forestry, 
Fisheries, Mining Emp.
   -0.0127***
(0.00)
-0.0195
(0.15)
  -0.0322**
(0.02)
Construction Emp.
0.0056
(0.25)
-0.0085
(0.70)
-0.0030
(0.90)
Manufacturing Emp.
  -0.0044**
(0.03)
0.0073
(0.50)
 0.0029*
(0.09)
Transportation, 
Communication and other 
Public Utilities Emp.
0.0062
(0.13)
-0.0385
(0.13)
-0.0322
(0.22)
Wholesale and Retail Trade 
Emp.
-0.0048
(0.55)
0.0786
(0.20)
0.0737
(0.25)
Finance, Insurance and Real 
Estate Emp.
   0.0254***
(0.00)
0.0565
(0.10)
  0.0819**
(0.02)
Business and Repair 
Services Emp.
   0.0196***
(0.00)
-0.0403
(0.15)
-0.0207
(0.47)
Personal, Entertainment and 
Recreation Services Emp.
0.0053
(0.21)
0.0197
(0.46)
0.0250
(0.37)
Health Services Emp.
  0.0093**
(0.01)
0.0241
(0.28)
0.0334
(0.15)
Educational Services Emp.
   -0.0345***
(0.00)
  0.1054**
(0.03)
0.0708
(0.17)
Other Professional and 
Related Services Emp.
0.0032
(0.43)
-0.0070
(0.81)
-0.0038
(0.90)
Federal Government 
Employment
0.0020
(0.42)
   0.0396***
(0.00)
   0.0416***
(0.00)
State Government 
Employment
   -0.0146***
(0.00)
0.0009
(0.95)
-0.0136
(0.41)
Local Government 
Employment
 -0.0133*
(0.08)
-0.0390
(0.29)
-0.0522
(0.16)
Self Employment
   0.0794***
(0.00)
-0.0115
(0.61)
   0.0678***
(0.00)
Poverty
   -0.0277***
(0.00)
0.0159
(0.23)
-0.0118
(0.30)
Human Capital
   0.1141***
(0.00)
 -0.0730*
(0.06)
0.0410
(0.30)
College town Dummy
   -0.0179***
(0.00)
   -0.1428***
(0.00)
   -0.1607***
(0.00)
Population Density
   -0.1773***
(0.00)
-0.0797
(0.22)
   -0.2570***
(0.00)
Area
   -0.1906***
(0.00)
-0.1093
(0.11)
   -0.2999***
(0.00)
NAIX
0.0007
(0.18)
 -0.0025*
(0.08)
-0.0019
(0.18)
R-squared
0.8186
ρ
0.6494***
(0.00)
p-levels are reported in parenthesis. 
***Significant at the 99% level
**Significant at the 95% level
*Significant at the 90% level
86


===== PAGE 95 (score 135, numeric-paren 108, table-words 2, var-words 9) =====

Table 4.7: Direct, Indirect and Total Effects Estimates with 14 Nearest Neighbors 
and a Relative Measure of Coal Dependence
Variable
Direct Effects
Indirect Effects
Total Effects
Coal Production per Income 
($)
   0.0077***
(0.00)
 0.0452*
(0.06)
  0.0529**
(0.03)
Initial Income
   0.5878***
(0.00)
-0.0230
(0.73)
   0.5648***
(0.00)
Agriculture, Forestry, 
Fisheries, Mining Emp.
   -0.0140***
(0.00)
 -0.0254*
(0.07)
  -0.0395**
(0.00)
Construction Emp.
0.0050
(0.23)
-0.0130
(0.56)
-0.0077
(0.73)
Manufacturing Emp.
-0.0027
(0.34)
0.0157
(0.15)
0.0129
(0.14)
Transportation, 
Communication and other 
Public Utilities Emp.
0.0050
(0.23)
 -0.0487*
(0.05)
 -0.0437*
(0.09)
Wholesale and Retail Trade 
Emp.
-0.0034
(0.68)
 0.1019*
(0.09)
0.0884
(0.16)
Finance, Insurance and Real 
Estate Emp.
   0.0261***
(0.00)
 0.0642*
(0.06)
  0.0903**
(0.01)
Business and Repair 
Services Emp.
   0.0207***
(0.00)
-0.0245
(0.37)
-0.0038
(0.89)
Personal, Entertainment and 
Recreation Services Emp.
0.0049
(0.25)
0.0219
(0.42)
0.0268
(0.34)
Health Services Emp.
  0.0094**
(0.01)
0.0213
(0.34)
0.0307
(0.18)
Educational Services Emp.
   -0.0357***
(0.00)
 0.0933*
(0.06)
0.0575
(0.27)
Other Professional and 
Related Services Emp.
0.0029
(0.47)
-0.0121
(0.68)
-0.0093
(0.77)
Federal Government 
Employment
0.0026
(0.29)
     0.0397***
(0.00)
   0.0422***
(0.00)
State Government 
Employment
   -0.0136***
(0.00)
0.0064
(0.69)
-0.0071
(0.67)
Local Government 
Employment
-0.0116
(0.10)
-0.0179
(0.62)
-0.0295
(0.43)
Self Employment
   0.0819***
(0.00)
-0.0042
(0.85)
   0.0777***
(0.00)
Poverty
   -0.0279***
(0.00)
0.0170
(0.19)
-0.0109
(0.33)
Human Capital
   0.1147***
(0.00)
 -0.0683*
(0.08)
0.0464
(0.25)
College town Dummy
   -0.0183***
(0.00)
    -0.1466***
(0.00)
   -0.1649***
(0.00)
Population Density
   -0.1843***
(0.00)
 -0.1297*
(0.05)
   -0.3140***
(0.00)
Area
   -0.1974***
(0.00)
  -0.1538**
(0.03)
   -0.3512***
(0.00)
NAIX
0.0007
(0.19)
  -0.0030**
(0.03)
 -0.0023*
(0.08)
R-squared
0.8188
ρ
0.6519***
(0.00)
p-levels are reported in parenthesis. 
***Significant at the 99% level
**Significant at the 95% level
*Significant at the 90% level
87


===== PAGE 96 (score 108, numeric-paren 86, table-words 2, var-words 9) =====

Table 4.8: OLS Results with an Absolute and a Relative Measure of Coal 
Abundance
Variable
OLS(1)
OLS(2)
Coal Production per Capita 
($) 
  0.0020**
(0.01)
-
Coal Production per Income 
($) 
-
  -0.0036**
(0.02)
Initial Income
   0.4387***
(0.00)
   0.4369***
(0.00)
Agriculture, Forestry, 
Fisheries, Mining Emp.
   -0.0180***
(0.00)
   -0.0203***
(0.00)
Construction Emp.
  -0.0147**
(0.01)
   -0.0155***
(0.00)
Manufacturing Emp.
   -0.0104***
(0.00)
   -0.0084***
(0.00)
Transportation, 
Communication and other 
Public Utilities Emp.
-0.0033
(0.51)
-0.0049
(0.33)
Wholesale and Retail Trade 
Emp.
  -0.0226**
(0.02)
  -0.0202**
(0.04)
Finance, Insurance and Real 
Estate Emp.
   0.0351***
(0.00)
   0.0360***
(0.00)
Business and Repair 
Services Emp.
  0.0113**
(0.01)
  0.0127**
(0.01)
Personal, Entertainment and 
Recreation Services Emp.
   0.0258***
(0.00)
   0.0259***
(0.00)
Health Services Emp.
 -0.0076*
(0.06)
 -0.0078*
(0.05)
Educational Services Emp.
   -0.0277***
(0.00)
   -0.0291***
(0.00)
Other Professional and 
Related Services Emp.
  0.0125**
(0.01)
  0.0125**
(0.01)
Federal Government 
Employment
-0.0008
(0.78)
-0.0007
(0.81)
State Government 
Employment
   -0.0117***
(0.00)
   -0.0112***
(0.000)
Local Government 
Employment
-0.0101
(0.20)
-0.0133
(0.10)
Self Employment
   0.0849***
(0.00)
   0.0884***
(0.00)
Poverty
   -0.0387***
(0.00)
   -0.1836***
(0.00)
Human Capital
   0.1100***
(0.00)
   0.1124***
(0.00)
College town Dummy
   -0.0299***
(0.00)
   -0.0304***
(0.00)
Population Density
   -0.1528***
(0.00)
   -0.1644***
(0.00)
Area
   -0.1945***
(0.00)
   -0.2053***
(0.00)
NAIX
   -0.0017***
(0.00)
   -0.0017***
(0.00)
R-squared
0.8287
0.8284
p-levels are reported in parenthesis. 
***Significant at the 99% level
**Significant at the 95% level
*Significant at the 90% level
88


===== PAGE 97 (score 7, numeric-paren 0, table-words 2, var-words 3) =====

Indirect effects estimates presented in the above tables represent spatial spillovers 
and  capture  related-county  effects  associated  with  changes  in  own-county  level 
characteristics. LeSage and Pace (2009) explain that indirect effects estimates can be 
interpreted in two different ways. The first interpretation is named as the average total 
impact on an observation. This interpretation captures how a change in the initial level of 
explanatory variable characteristics of all counties by some constant amount would 
change the income level of a typical county. The second interpretation is named as the 
average total impact from an observation, which captures the cumulative impact of a 
change in own-county initial level of explanatory variable characteristics averaged over 
all other regions. The counties that are closely related to own-county would experience 
greater  income  effects, while the ones that are loosely related  would have smaller 
impacts. 
Total effect estimates are the sum of direct and indirect estimates. As depicted in 
Table 4.5, the positive magnitude of total impact of initial income level suggests that 
higher income levels lead to higher current levels of county income. 
Direct, indirect and total effects estimates of the absolute coal abundance variable 
are not significant in either model specification. On the other hand, the direct effects 
estimates of the relative coal dependence variable are positive and significant in both 
specifications. These results are in stark contrast to the OLS regression results displayed 
on Table 4.8. The OLS results suggest a negative significant relationship between the 
relative  coal  dependence  variable  and  county  growth,  and  a  positive  insignificant 
relationship between the absolute coal abundance variable and county growth. The OLS 
results are in line with the strand of cross-country growth literature that suggests that 
89


===== PAGE 104 (score 125, numeric-paren 101, table-words 1, var-words 9) =====

Table 4.10: Model Averaged Estimates with Absolute Coal Abundance Measure
Variables
Direct Effects
Indirect Effects
Total Effects
Coal Production per Capita 
($)
-0.0001
(0.91)
-0.0024
(0.52)
-0.0025
(0.49)
Initial Income
   0.6186***
(0.00)
-0.0199
(0.81)
   0.5987***
(0.00)
Agriculture, Forestry, 
Fisheries, Mining Emp.
   -0.0151***
(0.00)
-0.0191
(0.14)
   -0.0342***
(0.00)
Construction Emp.
0.0078
(0.39)
-0.0139
(0.56)
-0.0061
(0.85)
Manufacturing Emp.
  -0.0071**
(0.04)
0.0060
(0.44)
 -0.0011*
(0.05)
Transportation, 
Communication and other 
Public Utilities Emp.
0.0052
(0.32)
-0.0681
(0.14)
-0.0629
(0.16)
Wholesale and Retail Trade 
Emp.
0.0026
(0.80)
  0.0464**
(0.03)
  0.0490**
(0.04)
Finance, Insurance and Real 
Estate Emp.
   0.0262***
(0.00)
0.0339
(0.28)
  0.0601**
(0.02)
Business and Repair 
Services Emp.
   0.0158***
(0.00)
0.0084
(0.55)
0.0242
(0.67)
Personal, Entertainment and 
Recreation Services Emp.
0.0068
(0.17)
0.0237
(0.50)
0.0305
(0.40)
Health Services Emp.
0.0098
(0.14)
0.0261
(0.41)
0.0359
(0.25)
Educational Services Emp.
   -0.0415***
(0.00)
  0.1188**
(0.02)
  0.0773**
(0.04)
Other Professional and 
Related Services Emp.
0.0047
(0.38)
-0.0029
(0.83)
0.0018
(0.87)
Federal Government 
Employment
0.0029
(0.31)
  0.0481**
(0.01)
0.0510
(0.11)
State Government 
Employment
   -0.0193***
(0.00)
0.0063
(0.49)
-0.0130
(0.46)
Local Government 
Employment
-0.0153
(0.11)
-0.0361
(0.39)
-0.0514
(0.20)
Self Employment
   0.0960***
(0.00)
-0.0048
(0.52)
   0.0912***
(0.00)
Poverty
   -0.0329***
(0.00)
0.0187
(0.21)
-0.0142
(0.26)
Human Capital
   0.1195***
(0.00)
  -0.0768**
(0.01)
   0.0427*** 
(0.00)
College town Dummy
   -0.0169***
(0.00)
   -0.0762***
(0.00)
   -0.0931***
(0.00)
Population Density
   -0.1905***
(0.00)
 -0.1071*
(0.07)
   -0.2976***
(0.00)
Area
   -0.1990***
(0.00)
  -0.0937**
(0.03)
   -0.2927***
(0.00)
NAIX
0.0002
(0.61)
 -0.0030*
(0.09)
-0.0032
(0.13)
p-levels are reported in parenthesis. 
***Significant at the 99% level
**Significant at the 95% level
*Significant at the 90% level
96


===== PAGE 105 (score 128, numeric-paren 104, table-words 1, var-words 9) =====

Table 4.11: Model Averaged Estimates with Relative Coal Dependence Measure
Variables
Direct Effects
Indirect Effects
Total Effects
Coal Production per Income 
($)
   0.0092***
(0.00)
 0.0465*
(0.05)
  0.0557**
(0.03)
Initial Income
   0.6181***
(0.00)
-0.0183
(0.79)
   0.5998***
(0.00)
Agriculture, Forestry, 
Fisheries, Mining Emp.
   -0.0154***
(0.00)
-0.0193
(0.14)
   -0.0347***
(0.00)
Construction Emp.
0.0081
(0.39)
-0.0141
(0.56)
-0.0060
(0.85)
Manufacturing Emp.
  -0.0076**
(0.03)
0.0063
(0.44)
 -0.0016*
(0.05)
Transportation, 
Communication and other 
Public Utilities Emp.
0.0055
(0.32)
-0.0691
(0.15)
-0.0636
(0.16)
Wholesale and Retail Trade 
Emp.
0.0027
(0.79)
  0.0462**
(0.03)
  0.0489**
(0.04)
Finance, Insurance and Real 
Estate Emp.
   0.0259***
(0.00)
0.0328
(0.30)
  0.0587**
(0.03)
Business and Repair 
Services Emp.
   0.0159***
(0.00)
0.0091
(0.59)
0.0250
(0.70)
Personal, Entertainment and 
Recreation Services Emp.
0.0058
(0.16)
0.0221
(0.50)
0.0279
(0.41)
Health Services Emp.
0.0100
(0.14)
0.0266
(0.42)
0.0366
(0.25)
Educational Services Emp.
   -0.0420***
(0.00)
  0.1185**
(0.02)
  0.0765**
(0.04)
Other Professional and 
Related Services Emp.
0.0048
(0.38)
-0.0031
(0.82)
0.0017
(0.88)
Federal Government 
Employment
0.0028
(0.30)
  0.0477**
(0.01)
0.0505
(0.11)
State Government 
Employment
   -0.0193***
(0.00)
0.0061
(0.49)
-0.0132
(0.46)
Local Government 
Employment
-0.0155
(0.11)
-0.0364
(0.38)
-0.0519
(0.21)
Self Employment
   0.0962***
(0.00)
-0.0050
(0.52)
   0.0912***
(0.00)
Poverty
   -0.0328***
(0.00)
0.0185
(0.21)
-0.0143
(0.26)
Human Capital
   0.1195***
(0.00)
  -0.0770**
(0.01)
   0.0425*** 
(0.00)
College town Dummy
   -0.0171***
(0.00)
   -0.0759***
(0.00)
   -0.0930***
(0.00)
Population Density
   -0.1904***
(0.00)
 -0.1070*
(0.07)
   -0.2974***
(0.00)
Area
   -0.1995***
(0.00)
  -0.0944**
(0.03)
   -0.2939***
(0.00)
NAIX
0.0003
(0.61)
 -0.0031*
(0.08)
-0.0028
(0.13)
p-levels are reported in parenthesis. 
***Significant at the 99% level
**Significant at the 95% level
*Significant at the 90% level
97


===== PAGE 106 (score 9, numeric-paren 0, table-words 0, var-words 9) =====

State government employment in own-county has a significant negative impact on 
own-county growth, suggesting an overexpansion of the state government. Unlike non-
spatial models that find a negative association between federal government employment 
and growth, the results of the spatial growth regressions suggest that federal government 
employment  in  own-county  do  not  hinder  own-county  growth.  Indeed,  although 
insignificant, the sign of the federal government employment estimate switches from 
negative to positive with the introduction of spatial dependence. 
Self  employment  and  poverty  have  positive  and  negative  direct  effects, 
respectively, and no significant spillover effects. Human capital imposes a positive direct 
impact and negative spillover effects, and its total impact is positive and significant at 99 
percent level. Natural amenities index does not affect growth. Population density, which 
captures the agglomeration effects, has a strong negative impact on own-county growth. 
This result reflects the congestion effects and disamenities connected to large cities and 
denser areas. Another result of interest is that college towns are associated with lower 
income levels, capturing the impact of disproportionate number of students residing in 
college towns.       
4.4 Conclusion
The negative association between resource abundance and poor economic performance 
has captured the attention of a wide range of audience, including academics, policy 
makers  and  international organizations. While the presence and the possible causal 
mechanisms of a resource curse have been extensively tested across countries, only a 
limited literature has extended the analysis using U.S. data. This study fills the gap in the 
98