Вы находитесь на странице: 1из 20

SOAR Modeling Team Progress Report

November 8, 2014
Outline
Section
A
1
2
a
b
3
a
b
c
d
e
f
g
h
i
4
5
B
1
2
3
4
C
1
2
3
4

Topic
Achievements to date
Description of data for SOAR model v1
SOAR model v1 structure
Modules structure
Spatial features
Initial model results
Local economy module
Government expenditures module
Local economy variables statistics summary
Labor market module
Labor market variables statistics summary
Housing market module
Housing market variables statistics summary
Demographic module
Demographic variables statistics summary
Potential Scenarios
Additional data needed for scenario analyses
November December plans
Continue model estimation
Choose scenarios
Conduct literature review on scenario
Design scenario
January February plans
Scenario literature review
Scenario data collection
Scenario estimation
Begin excel sheet development

A. Achievements to date:
1. Description of data for SOAR model v1
SOAR model v1 is being estimated with Cross Sectional data for 2007. More recent data is not
available from ACS and Census.
The data collection efforts for the SOAR grant project were divided into two stages. The first
stage involve secondary data from readily available sources. The second set of data require
more extensive work to digitize and will be delivered and reported upon in the coming weeks.
Some of the second set of data will be added to the model in SOAR v2 and others will replace
data series currently used in SOAR v1. All data collected are county level data for the counties in
Kentucky, as well as for those counties in states bordering Kentucky which are contiguous with
the SOAR region. Most data are collected for as many years as feasible in order to allow future
time-series analysis and projection of trends.
The first set of collected data contained data from Federal government including the Bureau of
Labor Statistics (BLS), the Longitudinal Employer-Household Dynamics (LEHD) produced the U.S.
Census Bureau, the Bureau of Economic Analysis (BEA), and the Area Health Resource File which
combines current and historical data from multiple governmental sources (including BLS and
Census). In addition, the first set includes data from Kentucky auditor reports of county
government accounts. Data were also collected from the 2007 Census of Governments reported
by the U.S. Census Bureau. The Census data will be used to provide additional detail about
countries where necessary. The data series collected, the coverage (time and geography),
sources and instructions for accessing the data are described in the table below.
The Area Health Resource File yielded data on population with second jobs, population below
the poverty line, some per capita income statistics, area in square miles, median house value,
total population, male population, female population, the number of births, the number of
deaths, number of occupied housing units, total labor force statistics, and number of people
employed. These data are compiled, sourced, and noted within the AHRF Access Database file,
which can be downloaded at http://ahrf.hrsa.gov/
From the United States Census Bureau, we sourced data on retail sales, total owner occupied
housing units, total business establishments, area in square miles, median house value,
dependent population, and occupied housing.
The 2007 Census of Governments provided data on revenue of utility services, other revenue,
sale of property, rents, utility expenditures, total charges as a source of revenue, parks and
recreation expenditures, administrative and general government expenditures, sewer and solid
waste expenditures, welfare, housing, and public welfare expenditures, health and hospitals
expenditures, fire protection expenditures, police and judicial expenditures, transportation
expenditures, and elementary, secondary, and higher education expenditures. This data can be
found at https://www.census.gov/govs/cog/historical_data_2007.html.
The U.S. Census Bureau was also a sourced for percentage of population below the poverty line
and the number of total establishments.
2

Crimes per 1000 people in a county were collected from the National Archive of Criminal Justice
Data.
From the Longitudinal Employer-Household Dynamics dataset, statistics for employment,
external employment, in-commuters, and out-commuters were collected.
2. SOAR Model v1 structure:
The SOAR model has a modular structure. Equations in each module are estimated as a
simultaneous system. Modules are estimated separately and related to each other in a recursive
fashion.
a. Modules:
i. Local economy module
1. Local business:
1) Retail Sales
2) Total business establishments
3) Retail establishments
4) FIRE establishments
5) Service establishments
2. Local government revenues:
1) Property Value
2) Total charges
3) Intergovernmental revenue
4) Other revenues
5) Utility Revenue
3. Local government expenditures:
1) Education expenditures
2) Health expenditures
3) Transportation expenditures
4) Police, protective etc. expenditure
5) Fire expenditures
6) Parks and recreation expenditures
7) Public welfare expenditures
8) Sanitation and solid waste expenditure
9) Administration expenditures
10) Utility expenditures
ii. Demography module:
1. Population
2. School age population
3. Dependent population
4. Population below poverty line
3

iii. Housing market module:


1. Total occupied housing units
2. Owner occupied housing units
3. Median house value
iv. Labor market
1. Labor Force
2. Female labor force
3. In-commuters
4. Out-commuters
5. Second jobs
b. Spatial features:
i. Spatial expansion method:
The spatial expansion is used in the Labor Force module. This technique involves
the use of an interaction terms between county area (square miles) and other
explanatory variables to account for systematic variation of parameters due to
the size of the observational unit. As the size of jurisdiction increases, the size of
the labor force and employment increase but the proportion of the labor force
identified as in-commuters and out-commuters decreases. This phenomenon
allows the direct estimation of average commuting distances over space.
ii. Spatial econometric procedures: dependence of two sources: Spatial
econometrics have been used to account for spatially correlated dependent
variables and error terms.
iii. Contiguity-based Spatial Weight Matrix:
The SOAR model is constructed using spatial econometrics which explicitly
measures the effect of spatial spillover effects. This method involves the use of
a spatial weights matrix. In the SOAR model, contiguity is used as the basis for
the spatial weight matrix.
The centroid of each county has been collected (longitude and latitude) to
estimate the average distance between counties. All Kentucky counties are
divided into the bordering/neighboring counties (48) and the inland counties
(71).

3. Initial Model Results


Several regressions of cross-sectional data from 2007 have been run for the labor market,
demography and housing market modules. Coefficients are from the spatial error model with
total per capita income and property value values, retail sales, education and health
expenditures and culture and recreational expenditures in thousands of dollars.
a. Local Economy module
OLS
Model

Property Value

Retail Sales

Total business
establishments

Retail
establishments

Variables
Intercept
total personal income
(Population *per capita
income) tpi
employment (emp)
out-commuters
(COM_OUT)
Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
Intercept
toal personal income
(Population *per capita
income) tpi
employment (emp)
in-commuters (COM_IN)
Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
intercept
total personal income
(Population *per capita
income) tpi
AREA
employment (emp)
Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
intercept
total personal income
(Population *per capita
income) tpi
employment (emp)
Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)

Coefficient
p-value
estimate
75794.1400

R-squared
0.0000
0.298

-41.8646
3.8410

0.0000
0.0000

0.2613

0.3570

-69.1830

0.0050

-0.0428
0.0328
0.0015

0.4810
0.0000
0.5440

-100.7100

0.0180

0.4600
0.2080
0.0190

0.0000
0.7600
0.0000

0.5690

-0.0040
0.0105

0.97

0.993

0.9748

0.8240
0.0000

b. Government Expenditures module

Model

OLS

Variables
estimates

Culture and
recreation
expenditures

intercept
total personal income
(Population *per capita
income) tpi
dependent population
(pop_depen)
AREA
Lambda (1.362<lambda<1)

p-value
-28.939

0.953

1.463

0.026

-0.3105

0.024

0.1043

0.946

R-squared
0.0328

Rho (-1.362<rho<1)
c. Local economy variables statistics summary:

d. Labor market module


Model
Labor Force

Variables

Intercept
employment (emp)
area
area*employment (a_emp)
external job (cemp)
Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
Women labor Intercept
force
population number (pop_num)
labor force (lf)
Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
In-commuters Intercept
employment (emp)
area
area*employment (a_emp)
external job (cemp)
Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
OutIntercept
commuters
labor force (lf)
area
area*employment (a_emp)
external job (cemp)
Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
Second jobs Intercept
toal personal income (Population
*percapita income) tpi
female labor force (lf_f)
employment (emp)
Lambda (.<lambda<20.044)
Rho (-1.362<rho<1)
Spatial Tests
Spatial error:
Moran's I
La grange multiplier
Robust Lagrange multiplier
Spatial lag:
Lagrange multiplier
Robust Lagrange multiplier

Estimates
-185.183
1.044
0.898
0.00001
0.0003
-0.300
-0.006
-581.640
0.022
0.458
-0.079
0.019
1333989.000
0.395
-14.163
0.00017
0.009
0.536

p-value
R-squared
0.098
0.999
0.000
0.006
0.264
0.416
0.512
0.480
0.000
0.999
0.113
0.000
0.836
0.583
0.631
0.86
0.000
0.041
0.523
0.125
0.084

-6640.757
0.607
12.057
-0.001
0.037
0.776
-0.244
-34.1390
-0.125

0.079
0.000
0.079
0.000
0.000
0.000
0.457
0.000
0.000

0.042
-0.001
-1.797
0.428

0.000
0.747
0.117
0.131

0.207
0.040
0.021

0.836
0.841
0.885

0.958
0.939

0.328
0.333

0.637

0.966

e. Labor market variables statistics summary:

f.
Model+B2:F27
Total
occupied
housing units

Housing market module:


Variables
Intercept
population number (pop_num)
employment (emp)

Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
Intercept
Owner
employment (emp)
occupied
total personal income
housing units (Population *percapita income)
tpi
Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
Intercept
Median House percapita income (pci_val)
Value
total establishments (tot_est)
poor
Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
Spatial Tests
Spatial error:
Moran's I
La grange multiplier
Robust Lagrange multiplier
Spatial lag:
Lagrange multiplier
Robust Lagrange multiplier

estimates
-805.028
0.350
0.172

OLS
p-value

R-squared
0.000
0.998
0.001
0.000

0.198
-0.137
2613.844
0.106

0.628
0.000
0.001
0.046

4906.760
0.735
-0.119
12533.370
2.829
18.053
-3.407
0.903
0.903

0.000
0.000
0.297
0.649
0.000
0.001
0.001
0.000
0.000

7.055
30.578
2.840

0.000
0.000
0.092

32.979
5.241

0.000
0.022

0.992

0.498

g. Housing Market variables statistics summary:

10

h. Demographic module:
Model

Variables

12851.190
1.953

0.128
0.000

Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
Intercept
population number (pop_num)

0.781
-0.144
-93.255
0.264

0.000
0.011
0.328
0.000

Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
Intercept
Dependent
population number (pop_num)
population

-0.024
0.052
-146.209
0.198

0.954
0.257
0.034
0.000

-0.130
0.024
1486.184

0.765
0.619
0.312

949.954

0.111

0.517

0.000

0.872
-0.408

0.000
0.058

6.386
26.150
46.120

0.000
0.000
0.000

2.402
22.372

0.121
0.000

Population

School age
population

Population
below
poverty

Intercept
labor force (lf)

estimates

OLS
p-value

Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)
Intercept
total personal income (Population
*per capita income) tpi
dependent population
(pop_depen)
Lambda (-1.362<lambda<1)
Rho (-1.362<rho<1)

Spatial
Tests

Spatial error:
Moran's I
Lagrange multiplier
Robust Lagrange multiplier
Spatial lag:
Lagrange multiplier
Robust Lagrange multiplier

R-squared
0.996

0.997

0.997

0.952

11

i.

Demography variables statistics summary:

12

4. Potential scenarios:
a. Encourage value added development within region:
i. Increase the diversity of opportunity for employment
ii. Increase the opportunities for small business
iii. What types of employment would be created?
iv. Investment in financial capital, human capital, social capital and public built
capital
b. Small business expansion (entrepreneurship);
i. Increase the diversity of opportunity for employment
ii. Increase the opportunities for small business
iii. Directly improve the quality of life of residents
iv. Increase the attractiveness of the region to families
v. What types of employment would be created?
vi. Investment in financial capital, human capital, social capital and public built
capital
c. Tourism development
i. Increase the opportunities for small business
ii. Increase the diversity of employment opportunities
iii. Increase the quality of life for residents
iv. Investment in human capital, social capital, public built capital, natural capital
d. Agri-tourism
i. Increase the opportunities for small business
ii. Increase the diversity of employment opportunities
iii. Increase the quality of life for residents
iv. Investment in human capital, social capital, public built capital, natural capital
e. Parkway development
i. Regional benefits and costs
ii. Distributional impacts across the region
iii. What types of employment would be created?
iv. Investments in public built capital, financial capital, human capital, social capital
f.

Conversion of reclaimed surface mines to other types of land use


i. Economic impacts of alternative land uses
ii. Development tradeoffs (e.g. forestry / tourism)
iii. Infrastructure needs
iv. What types of employment would be created?
v. Investment in natural capital, public and private built capital, financial capital
13

g. Regional industrial parks


i. Regional benefits and costs
ii. Increase the opportunities for small business
iii. Distributional impacts across the region
iv. What types of employment would be created?
v. Investments in public built capital, financial capital, human capital
h. Emerging regional growth hubs (in addition to Pikeville):
i. Impacts on employment and shopping patterns
ii. Impacts on residential choice
iii. What types of employment would be created?
iv. Investments in human capital, social capital, public built capital
i.

Impact of Commuting patterns within region and with the outside:


i. Impacts on employment and shopping patterns
ii. Impacts on residential choice
iii. Investments in human capital, social capital, public built capital

j.

Transportation improvements
i. Future transportation trends and volumes
ii. Impact on economic attractiveness
iii. Impacts on commuting patterns
iv. Impact on residential choices and land use
v. Increase in quality of life
vi. What types of employment would be created?
vii. Investments in public and private built capital, natural capital, human capital,
social capital

k. Impact of Medicaid expansion


i. Impact on health workforce
ii. Demand for health care provider and possible shortages
iii. Impact on child care needs
iv. Investments in human capital, social capital, public and private built capital

l.

High SSI (supplemental security income) utilization rates:


i. Impact on labor force participation
ii. Impact on labor demand
iii. Investments in human capital, social capital, public and private built capital

14

Scenarios that could be studied over time but require extensive research to develop the
scenario:
a. Increase Educational attainment
iv. Determine the value of increased college and technical degree
v. Determine how employers would be attracted to a more educated workforce
vi. Investment in human capital, public built capital
b. Increased use of broadband in Education
vii. Increase educational attainment of population
viii. Improve quality of education
ix. Investment in human capital, public built capital
c. Align education, workforce training and economic development strategies
x. Determine how employers would be attracted to a more educated and skilled
workforce
xi. Investment in human capital, public built capital
d. Investment in early childhood development & education
xii. Determine the impact on labor force participation, entrepreneurship
xiii. Determine how employers would be attracted to a more educated workforce
xiv. Investment in human capital, social capital, public built capital
e. Leadership development
xv. Development of soft employment skills
xvi. Investment in human capital and social capital
f.

Market the work ethic


i. Determine aptitude of regional workforce
ii. Investment in soft employment skills
iii. Investment in human capital and social capital

g. Substance abuse treatment


i. Transportation barriers and needs
ii. Needed facilities
iii. Workforce needs
iv. Investments in human capital, social capital, public built capital

15

5. Additional data needed for scenario analyses:


a. Small business, tourism, industrial part scenarios
i. Small business birth and death rates, size, growth rates, pay scales.
ii. Opportunities analysis for small businesses
iii. Labor and skill requirements of small businesses
b. Impacts of industrial parks and reclaimed mine lands
i.
Land values by category
ii.
Land development costs
iii.
Infrastructure needs and costs
iv.
Attractiveness of region to business
v.
Labor and skill demands
c. Transportation and commuting scenarios
i.
Current commuting patterns
ii.
Current commercial traffic patterns
iii.
Condition of transportation infrastructure
iv.
Right of way cost,
v.
Construction costs
vi.
Traffic accidents data
d. Health care expenditures expansion
i. Necessary public investments
ii. Necessary private investments
iii. Current health of population
iv. Relationship between health and productivity, health care demand,
e. Substance abuse treatment facility
i.
Infrastructure needs
ii.
Workforce needs
iii.
Potential demand for service
iv.
Cost of treatment
v.
Determination of best locations for facilities
vi.
Transportation needs
vii.
Impact on labor force participation and productivity
viii.
Impact on costs of other public services (courts, police, health clinics)
B. November December Plans:
1. Continue model estimation
a. Finish Ordinary Least Squares estimation for all modules.
b. Estimation of Two Stage Least Square, spatial econometric model
Run Morans I test to test for spatial dependence, spatial autocorrelation.
16

c. Start estimation of spatial lag 3 Stage Lease Square system


This methodology accounts for simultaneity bias which arises in systems of structural
equation models when a dependent variable is correlated with another equations error
term. It accounts for the correlated errors between equations which improves the
efficiency of the parameter estimates.
2. Choose scenarios
3. Conduct literature review on scenario
4. Design scenarios

C. January - February Plans:


1. Finish model estimation
a. Estimate the spatial lag 3 Stage Lease Square estimation of the final version of the
model
b. Assess coefficients and adjust as necessary
2. Scenario literature review
3. Scenario data collection
4. Scenario estimation and write up
5. Excel sheet development

17

Variable Name

Variable Description

Time Series

Obtained

Link

JOB2

Second Jobs

2002-2011

AHRF

POOR

Population Below Poverty Line

1993, 1995, 1997-2011

AHRF

AHRF Database Application


Population tab, Poverty Statistics (be sure to select most
recent as well as historical data, which can be listed a little
further down)

PCI

Per Capita Income

1975-2010

AHRF

Population tab > Income (include most recent and all historical
data, which are listed further down)

AREA

Area in Sq Miles

2000, 2010

AHRF

Environment tab, Land Area (check for historical data which


can be listed further down)

HOWN_VAL

Median House Value

1980, 1990, 2000

AHRF

Population tab, Decennial Census

POP

Population (total, male,


female)

2000-2011

AHRF

Population tab, Population (include most recent as well as historical


data, which are further down the list)

Births/Deaths

1970-2010

AHRF

Population tab, Births/Deaths (include most recent as well as historical


data, which are further down the list)

AHRF

Population tab, Decennial Census

Deaths

1970-2010

HOUSE_OCU

Housing

1980, 1990, 2000

PVAL

Property Values

EMP

Total Labor Force and Number


Employed

1980-2011

PCI

Per Capita Income

1970-2011

LF

Total Labor Force

1990-2013

RSAL
OWN_OCU

Retail Sales
Total Owner Occupied Housing
Units

AREA

Area in Sq Miles

HOWN_VAL

Median House Value

AHRF
AHRF
BEA - Select 'Local Area Personal
income and Employment' then
'Peronal income, per capita
personal income, and population'
then 'county' then 'Kentucky' then
'all counties' then 'all years' and
download the desired format file.
BLS.gov - Under the Heading
"Annual Average" compiled from
the yearly county tables containing
all county data nationwide
2007

Population tab, Employment (include most recent as well as historical


data, which are further down the list)

http://www.bea.gov/iTable/iTable.cfm?reqid=70&step=1&isuri=1&acrdn
=5#reqid=70&step=1&isuri=1

http://www.bls.gov/lau/#tables

Census County Quickfacts

http://quickfacts.census.gov/qfd/states/21/21001.html

Census County Quickfacts

http://quickfacts.census.gov/qfd/states/21/21001.html

2000

Census County Quickfacts

http://quickfacts.census.gov/qfd/states/21/21001.html

Census County Quickfacts

http://quickfacts.census.gov/qfd/states/21/21001.html

Census County Quickfacts

http://quickfacts.census.gov/qfd/states/21/21001.html

Census County Quickfacts

http://quickfacts.census.gov/qfd/states/21/21001.html

Census County Quickfacts

http://quickfacts.census.gov/qfd/states/21/21001.html

2008-2012

POP_DEPEN

Dependent Population

HOUSE_OCU

Total Occupied Housing Units

2010
1998-2012 (as a single
observation)
2010-2013 (as a single
observation)
2010-2013 (as a single
observation)

REV_OTH

Other Revenue, Sale of


Property, Rents and

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local

REV_UTL

Revenue of Utility Services

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local

Total Business Establishments

18

EXP_UTL

Utility Expend

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local

CHARGES

Total Charges as Source of


Revenue

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local

EXP_PRK

Parks and Rec Expend

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local

EXP_ADM

adm and gen gov expenditures

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local

EXP_SAN

Sewer and Solid Waste Expend

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local

EXP_WEL

Welfare, Housing, and Public


Welfare Expend

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local

EXP_HLT

Health and Hosp Expend

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local

EXP_FIR

Fire protection Expend

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local

EXP_POL

Police and Judicial Expend

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local

EXP_TRN

Transportation Expend

2007-2010

Census of Governments

https://www.census.gov/govs/local/historical_data_2007
.html#state_local

EXP_EDU

elem, secondary, higher ed


expend

Census of Governments

https://www.census.gov/govs/local/historical_data_2007.
html#state_local
http://www.census.gov/econ/cbp/download/02_data/

2007

Establishments

2003-2011

Compiled from the individual year


sheets for all counties in the U.S.

EMP

Employment

2002-2011

LEHD

COM_IN

# incommuters

2002-2011

LEHD

COM_OUT

# outcommuters

2002-2011

LEHD

CEMP

External Employment

2002-2011

LEHD

CRIMT

Crimes per 1,000 Population

1991-2012

POOR

Population Below Poverty Line

2000-2012

National Archive of Criminal


Justice Data
Percent of People Below the
Poverty Line in the past 12 months
(for whom poverty status is
determined), 2008-2012 ACS 5 year estimates

http://www.icpsr.umich.edu/icpsrweb/NACJD/series/57/studie
s?archive=NACJD&q=county-level

http://factfinder2.census.gov/faces/tableservices/jsf/pages/pro
ductview.xhtml?pid=ACS_12_5YR_GCT1701.ST05&prodType=table

19

20

Вам также может понравиться