determinants of land use change on the southern cumberland
TRANSCRIPT
Determinants of Land Use Change on the Southern
Cumberland Plateau
Matthew Lane, Kevin Willis
0
Robert Gottfried, Douglass Williams
R - 8 2 9 8 0 2 0 1 -
1
� drivers of this land use change in our study region
The purposes of our STAR grant
�
impacts �
to understand the socioeconomic
to explore possible future trends in land use and their potential environmental
to examine the effects of possible policy responses
2
� l
� ili
� l
;
� icy responses.
Deve oping a spatial socioeconomic model of change in land use for the period 1980-2003;
We are doing this by:
Integrating this model w th our research on bird diversity as a function of land cover and a water qua ty proxy to understand the socioeconomic processes bringing about environmental change in the region;
Using this understanding and the model to assess potentia future impacts on birds and water quality of likely socioeconomic events or trends
Exploring the impacts of possible pol
3
Overview of presentation
• • • • • • •
•
Focus on homebuilding and LULC change Review of the literature Theoretical framework The case study area The data Logit analysis of housing Does homebuilding lead to conversion of native forest - Discussion Conclusion
4
•
•
•
• landowner strategy
•
Urban areas are affecting the Plateau via:
Exurban expansion/permanent homes
Second home growth
Increasing demand for paper
These interact with changes in corporate
Political process
5
Why Are Homes Being Built?
• retirement and second homes
•
•
• Crossville
•
•
Growing demand by baby boomers for
Growing population in the Southeast
Florida rebound
Escaping Chattanooga, McMinnville,
Stock market of the 90’s
Lower interest rates
6
Review of literature
• – –
explicit
• –
•
•
Noneconomic models can incorporate economic variables fail to make underlying socioeconomic processes
Economic models lack of fine scale data led to larger scale analyses inappropriate for supporting many ecological models
Share of land uses at some aggregate level such as county or state; e.g, Alig and Healy, 1987; Hardie and Parks, 1997; Hardie, et al., 2000; Miller and Platinga, 1999; Platinga,1996; Stavins and Jaffe, 1990; Wu and Segerson, 1995
7
•
•
•
Finer scale economic analyses
Location of deforestation – parcel level (Cropper, Puri and Griffiths, 2001)
Land cover transitions in two watersheds – cell-based (Turner, Wear and Flamm, 1996)
Probability of land use change for San Francisco Bay and Sacramento – cell-based (Landis, 1995 and Landis and Zhang, 1998)
8
Most to this study
•
residential
•
relevant
Probability of parcel conversion in Patuxent watershed near Washington, DC and Baltimore for one period of time (Bockstael. 1996; Costanza et al, 1996; Geogehegan, Wainger and Bockstael, 1997)
- Uses probit to obtain transition probability of conversion to
- Cannot provide overall quantity of parcels converted, only their location
Optimal timing of land use change of parcels in Patuxent watershed for one period (Irwin and Bockstael, 2000; Irwin and Bockstael, 2002)
-Uses hazard model that provides overall quantity of conversion - Cannot examine economic drivers of conversion over time
9
This presentation:
• of parcels ten acres and over (4,752 parcels)
• the first home on a parcel
• • •
characteristics, and economic drivers
Uses logit to examine residential conversion
Residential conversion defined by building
Examines a long period, 1980-2003 Large rural study area: 616,000 acres Combines site characteristics, owner
10
Theoretical Framework
Ctij t t , Ot, It, S*et t , S*Ot, S*It t t, 97*Ot
where: et c NtOtIt
Land moves to its most highly valued use Land market is imperfect – lack of information on new land uses
In reduced form:
= m (e , c, N , S*c, S*N , 97*e , 97*c, 97*N, 97*I),
= exogenous economic variables at time t = parcel characteristics = land use in time t of neighboring parcels = owner characteristics in time t = cost of inputs to conversion process in time t
S = large parcel (0,1) 97= period 1997-2003 (0,1)
11
Effect of neighboring parcels
•
• changing land market
Externalities, both positive and negative
Information flows in an imperfect and
12
The Case Study Area
13
14
Geographic characteristics:
� plateau surface of 7 southern counties of central Tennessee
� 616,000 acres
�
� i
�
� l
�
contains some of the largest remaining privately owned, contiguous temperate deciduous forest in North America
Ecological characteristics:
predominantly oak and hickory – mast of mature oak canopy is a keystone resource w thin the food web of the ecosystem
headwaters of some of the most biologically diverse freshwater stream systems in the world
important neotropicamigratory bird habitat
one of the most diverse woody plant ecosystems in the eastern United States
15
�
Social
� l
�
� l
land is largely privately held
characteristics:
relatively high levels of unemp oyment and poverty
inmigration to study area counties has exceeded outmigration since 1983-84
ow levels of education
16
Land Cover, 1980-2003All Plateau
Land Cover 1980-2003
600000
500000
400000
300000
200000
100000
0
i
Logged
Pi l i
Mi ine
Ac
res
Nat ve Forest
Grass
ne P antat on
xed P
1980 1990 1997 2000 2003
Year
17
1980 2003
Within our study area, 20% of the forest being managed for hardwoods was converted to other uses.
18
19
0
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
Sal
Sales of Raw Land by Sales Size
2000
4000
6000
8000
10000
12000
14000
16000
18000
Year
Acr
es Sales less than 500 acres
es 500 acres of more
Sales of Raw Land by Sales Size
20
lin
Houses per year in each time period
0.00
2.00
4.00
6.00
8.00
10.00
12.00
14.00
16.00
1980-1989 1990-1996 1997-1999 2000-2004
Periods
Ra
te
Bledsoe Frank Grundy Marion Sequatchie Van Buren Warren
Houses per year in each time period
21
Homebuilding Over Time by 2000 Owner TypeHomebuilding over time by 2000 owner
800
H o
me
s b
uilt
pe
r y
ea
r 700
600
500
400
300
200
100
0
ls
Absentees
Loca
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Year
22
23
Percentage of new homes owned by absentees
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Year
Percent of New Homes Owned by Absentees
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
0.16
Pe
rce
nt
0
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
Sal
Sales of Raw Land by Sales Size
2000
4000
6000
8000
10000
12000
14000
16000
18000
Year
Ac
res Sales less than 500 acres
es 500 acres of more
Sales of Raw Land by Sales Size
24
19781980
19821984
19861988
19901992
19941996
19982000
2002
I
AVERAGE RAW LAND PRICE BY SALE SIZE
0.0000
200.0000
400.0000
600.0000
800.0000
1000.0000
1200.0000
1400.0000
1600.0000
1800.0000
YEAR
A V
ER
A G
E R
E A
L P
R C
E P
ER
A C
R E
200+Acres
0-199 Acres
All Raw Sales
25
Annual Rate of Change of Land Cover, All plateau
0
Ag/
l
-15000
-10000
-5000
5000
10000
1980 to 1990 1990 to 1997 1997 to 2000 2000 to 2003
Periods
Ac
res
pe
r y
ea
r
Native Forest
Logged
res
Pine P antation
26
Annual Rates of Conversion Parcels>10 acres
0
500
1000
1500
2000
2500
3000
3500
4000
4500
5000
1997-00
i
f i
f /
Pi
1980-90 1990-97 2000-03
Per od
A c
r es
p e
r ye
ar
Native orest to p ne
Native orest to ag res
ne to ag/res
27
Native Forest Conversion to:
35% Pine Plantation
42% Agricultural / Residential
22% Cleared / Unclassified
1980 - 2003
28
Summary
• while continuing (despite the pine beetle!), has slowed
• grass) has surpassed pine and increased rapidly
• are on the landscape
The conversion of native forest to pine,
Conversion to “ag/res” (predominantly
Why? May need to know who big players
29
0
20
40
60
80
1 2 3 4 5 6 7 8 9
LAND DISTRIBUTION BY DECILES
8.0444
4.3643 2.8504 1.9694 1.4239 1.0924 0.8441 0.6740 0.5145
78.2227
100
10
DECILE
% O
F T
OT
AL
LA
ND
HE
LD
30
TN
38%
LAND DISTRIBUTION BY OWNER LOCATION (INCLUDES GOVERNMENT)
19%
LOCAL 43%
OUT SIDE
31
3%
7%
DISTRIBUTION OF ACRES BY OWNER TYPE
Other
Individual 53%
Gov ernment
Timber 18%
Business 19%
32
0
20000
40000
60000
80000
100000
120000
Bowater In
c
State
of TN
USXCorp
.
Huber JM
Mea
dConta
inerb
oard
Inc
Souther
nPin
ePlan
tatio
ns
Wern
erLum
berCo
Flatb
ushLLC
Holland
War
e
Univer
sity
of the
South
Pine Pl i l
Top 10 Forest owners in SAA study area
Forest antat on Logged/C eared Other
Acres
33
Land Company Divestments, 1997-2004 (15% of landscape) Possible Bowater Divestments in Blue (13% of landscape)
34
The Data
• 1980, 1990, 1997, 2000 and 2003
•
• • •
Remotely sensed land use/land cover data for
Tax map for 2000/2001 for parcels 10 acres or greater; “holes” in tax map Associated tax data for 1999-2003 Large sample of sales data for 1980-2003 Sales data for all transactions of timber companies and other holdings of 1000 or more acres from 2000-2004, including Huber sales since 1998
35
36
Tax map and “holes”
The Logit Analysis
37
Variables in the Model
? ?
-
/+
+ -? -? ? +
+ + + + + +
? +
j ?
+ --+ -
--
Variable Description Expected Sign HBUILT House built on parcel during the year Na “COUNTY” Dummy for counties BUFFER%MR Percent of land within a radius of 0.1 or 1km in
mixed pine or reservoir BUFFER%PINE Percent of land within a radius of 0.1 or 1km in
pine plantation BUFFER%AR Percent of land within a radius of 0.1 or 1km in
agriculture or residential (grassy shrub) BLUFF_FRONTAGE Parcel located on bluff (0,1) SIZELT90 Size of parcel less than 90 acres (acres) PERC_NF % of parcel in native forest PERC_PP Percentage of parcel in pine plantation PERC_OT Percentage of parcel in grass (“other”) PAVED Parcel on a paved road NEIGHB_AVG Average value of houses on parcels within 1km
of a parcel WATER Parcel has public water (0,1) SEWER Parcel has public sewer (0,1) GAS Parcel has natural gas (0,1) ELEC Parcel has electricity (0,1) NEARBY_HOUSES Number of houses on parcels within 1km DISTWHOLES Distance to nearest parcel with a house, counting
areas with parcels <10 acres as having houses LNDISTCITY Log of distance to nearest city (meters) ADJ_PROT_AREA Parcel adjacent to a protect area (0,1) DIST_ROAD Straight line distance to nearest ma or road
(meters) BUSOWNER Parcel owner is a business (0,1) TIMBEROWNER Parcel owner is a timber company (0,1) MORTGAGE_RATE Mortgage rate on a home POP_SE_CHANGE Change in population in SE United States MS_CONST_COST Marshall Swift Construction Cost Index (deflated
by CPI) WILSHIRE Wilshire stock index UNEMPLOYRATE_TN Unemployment rate in Tennessee Interaction terms *GE90: parcels 90 acres or more
*1997: for period 1997 or later
38
l it = =
i2 = = 0.0786
l icient i
i
i
Logit Results Margina effects from log Number of obs 83603
chi2(47) 678.58 Prob > ch 0.0000
Log Likelihood = -3722.6869 Pseudo R2
Dependent Variable: HBUILT Independent Variab es Coeff Std. Error P>z Independent Variables Coeff cient Std. Error P>z
Franklin County 0.000121 0.000864 0.889 NEIGHB_AVG 0.000000 0.000000 0.051 Grundy County -0.000515 0.000640 0.421 WATER 0.001030 0.000423 0.015 Mar on County -0.000778 0.000758 0.305 SEWER 0.001008 0.000942 0.285 Sequatchie County 0.001233 0.000719 0.086 SEWER1997 0.006281 0.001364 0.000 Van Buren -0.001002 0.000663 0.131 GAS 0.002663 0.001150 0.021 Warren County -0.000556 0.001293 0.667 ELEC 0.003151 0.000491 0.000 Mar on1997 -0.005347 0.001219 0.000 ELECGE90 0.002413 0.000951 0.011 Warren1997 -0.005179 0.002912 0.075 NEARBY_HOUSES 0.000159 0.000030 0.000 BUFFER%PP, 0.1km -0.002569 0.003307 0.437 DISTWHOLES -0.000004 0.000001 0.000 BUFFER%AR, 0.1km -0.004524 0.001335 0.001 DISTWHOLESGE90 0.000004 0.000001 0.000 BUFFER%MR, 0.1km 0.009134 0.004756 0.055 LNDISTCITY 0.000082 0.000052 0.117 BLUFF_FRONTAGE 0.001218 0.000558 0.029 ADJ_PROT_AREA -0.001533 0.001070 0.152 BLUFF_FRONTAGEGE90 -0.002546 0.001270 0.045 DIST_ROAD 0.000002 0.000000 0.001 SIZELT90 0.003314 0.001030 0.001 DIST_ROADGE90 -0.000007 0.000002 0.001 PERC_PP 0.004418 0.003932 0.261 TIMBEROWNER -0.008699 0.004080 0.033 PERC_PP1997 -0.025573 0.011881 0.031 BUSOWNER -0.005015 0.001646 0.002 PERC_PPGE90 -0.037976 0.018229 0.037 BUSOWNER1997 0.004477 0.002094 0.033 PERC_OT 0.006117 0.001042 0.000 MORTGAGE_RATE -0.000027 0.000125 0.832 PERC_OT1997 -0.004692 0.001173 0.000 POP_SE_ 0.000000 0.000000 0.908 PERC_OTGE90 -0.008936 0.009235 0.333 MS_CONST_COST -0.002824 0.007049 0.689 PERC_PM -0.013242 0.011720 0.258 WILSHIRE 0.000000 0.000000 0.000 PERC_PM1997 0.004166 0.012936 0.747 UNEMPLOYRATE_TN -0.000215 0.000176 0.222 PERC_PMGE90 0.015426 0.011932 0.196 CONSTANT -0.027481 0.005116 0.000 PAVED -0.000868 0.000402 0.031
39
Factors influencing probability of home construction Signifi
l
l l
l
cant variables with positive effect
BUFFER%MR, 0.1km (relative to native forest) BLUFF_FRONTAGE, SMALL PARCELS SIZELT90 PERC_AR>PERC_AR1997 (relative to native forest) NEIGHB_AVG WATER SEWER, after 1997 GAS ELEC, particular y after 1997 NEARBY_HOUSES LNDISTCITY DIST_ROAD, small parcels WILSHIRE
Significant variables with negative effect
BUFFER%AR, 0.1km >BUFFER%PP, 0.1KM (relative to native forest) BLUFF_FRONTAGE, LARGE PARCELS PERC_PP1997 PERC_PP, LARGE PARCELS PAVED DISTWHOLES for smal parcels DIST_ROAD, arge parcels TIMBEROWNER (less like y than business owner) BUSOWNER, particularly before 1997
40
Homebuilding and Forest Conversion
• forest
•
•
• acres
All new homes on parcels<100 acres have
Average size 20 acres total of forest
Majority also have some grass/shrub
Average grass/shrub per parcel is about 10
41
• 1990-2000 cleared native forest for grass/shrub
•
• 1997-2000 no home construction caused logging or grass/shrub
• not be causing conversion to grass/shrub. Either developers bought cleared land or sold it cleared to people who then built.
About 20% of home construction 1980-90 and
Average grass/shrubclearing was 9 and 13 acres in the two periods
CONCLUSION: Home construction itself may
42
Anecdotal Evidence
• build
• and start over
•
•
• Buren County
Many people clear some land when they
People “don’t know what they have”, clear
People putting in hobby horse farms/farms
Cultural attitudes – Florida, NY, urban
Growth of grass for pasture and hay – Van
43
Conclusions
•
•
•
• some of these hypotheses
Conversion of native forest to pine is slowing; may stop – Bowater’s agreement (?), pine beetle Divestment may be allowing pent up demand for pasture/hay to be supplied – may be the primary cause of conversion to grass/shrub Growing demand for second and retirement homes by baby boomers may cause some, but less, clearing Multinomial probit analysis will explore at least
44
Conclusions and Issues
suggested by the theoretical framework •Homebuilding is following the pattern
•It appears to be due to permanent home growth (commuters and retirees) and second homes •Homebuilding may not be causing forest clearing as much as agricultural expansion •Problem of “splinters” needs to be resolved
45
Any comments or questions?
46