Supporting Material is available for this work. For more information, follow the link from
the Table of Contents to "Accessing Supporting Material".
Table of Contents
Go Back
Affordable Housing and Urban Sprawl Policy Choices
in York County, Maine:
A System Dynamics Approach
John Voyer
School of Business
University of Southern Maine
96 Falmouth Street, Box 9300
Portland, ME 04104-9300
207-780-4597 phone / 207-780-4662 fax
voyer@usm.maine.edu
Abstract
York County, Maine, is filled with typical New England home rule towns (they have high control
over local decisions). It is experiencing a lack of affordable housing and too much urban
sprawl. After a thorough examination of the situation in a typical York County town, including a
causal loop diagram based on interviews with knowledgeable informants, a system dynamics
model is used to test five possible policies for the town: the status quo (large house lots and
modest construction levels), smaller house lots, a cap on construction, increased construction,
and smaller house lots and increased construction. The policy testing shows that the “status
quo” is not a terrible policy, that the “smaller lots” has some things to recommend it, but that
the combination of smaller lots and increased construction gives the best mix of outcomes—
better housing affordability and less urban sprawl. Implications for policy and future modeling
are discussed.
Keywords: Affordable housing; urban sprawl; Maine; municipal public policy
INTRODUCTION
Founded in 1636, York County in Maine is one of the oldest counties in the United States
(York County, Maine 2004). It is located in extreme southwestern Maine in a coastal region
bordered by New Hampshire to the west and southwest, the Atlantic Ocean to the east, and the
Ossipee and Saco rivers to the north (Encyclopaedia Britannica 2004; see map in Figure 1). Its
southernmost town, Kittery, is approximately 60 miles (96 kilometers) north of Boston,
Massachusetts. The county has a land area of 991 square miles (2567 square kilometers), its
estimated population in 2001 was 192,700, and its rate of home ownership in 2000 was 72.6%
(U.S. Census Bureau 2004c). York County’s estimated median household income in 2000 was
$46,081, ranking it first among Maine’s counties (U.S. Census Bureau 2004a) and well above the
estimated state median of $37,589, and even above the U.S. estimated median of $41,990 (U.S.
Census Bureau 2004b).
By many measures York County appears to be affluent and relatively problem-free.
However, there is a lurking problem—lack of affordable housing. “In the summer of 2002 the
National Association of Homebuilders proclaimed the Portsmouth-Kittery [New Hampshire]
area—which includes Berwick, Eliot, Kittery, South
Berwick, and York [in York County, Maine]—as one of
the ten least affordable regions in the country—and the
only one in the top ten outside of California” (Maine State
Housing Authority 2002, p. 12). In 2003, Maine ranked
9th in the United States in the rate of house price
appreciation, and York County, with its proximity to
Boston, no doubt was responsible for much of that
ranking (Office of Federal Housing Enterprise Oversight
2003, p. 9). Related to the affordability problem is one of
“sprawl,” or housing development that is perceived to be
excessive. This is related because the high price of
houses leads prospective homeowners to build further and
further out from existing development. In 1940, York
County was more than half rural. By 2000, the entire
county was either suburban/urban or “emerging suburb”
(Maine State Planning Office 2004).
The state of Maine has had an affordable housing
Aroostook
Figure 1 Map of Maine's Counties statute on its books since 1989. In addition, a statute
passed that same year requires each town or city in Maine
(Note York Country at lower left)
Source: to prepare a Comprehensive Land Use Plan, one of whose
http://www.state.me.us/sos/kids/government/co provisions requires that least ten percent of “new
unties.htm . . . .
residential development” over any given five-year period
be affordable housing. A provision added in to that
statute in 2001 is intended to prevent urban sprawl. Unlike most other regions of the United
States, counties in New England states like Maine provide a limited number of services—
typically sheriffs departments (a sort of “county police”), courts and jails (Richert 2003).
Typically, implementation of affordable housing and growth management policies, which appear
to have been ineffective in York County, is left to municipal governments.
Of the 29 municipalities in York County, only two are cities that have city councils, city
managers and mayors. The rest are towns, which make major policy decisions using annual
town meetings, where all eligible voters turn out to decide ballot questions on the spot. In
between these town meetings, elected townspeople called selectmen gather to make decisions,
which are then implemented by town managers and their staffs. Policy decisions about housing
are made by town meetings, but the opinions of selectmen carry weight. Evidence shows that
two typical policies carried out by towns in York County were caps on housing permits and large
house lots (O’Hara 1997). We will see that these policies have significant implications both for
affordability of housing and growth management (or mismanagement).
RESEARCH OBJECTIVES AND METHODS
In early 2003, a non-profit group called York County Affordable Housing asked for a
system dynamics model to study the affordable housing problem in the county. The aim was for
the model to be used by Boards of Selectmen in the various towns in York County, to help them
get a better feel for the consequences of their policy decisions. The model discussed in the
present paper was the result of that request. The purpose of the model is to see if there are any
policies that can be followed by York County towns that might ameliorate the affordable housing
situation.
Three experts in York County’s housing situation made themselves available for
interviewing. One was the Executive Director of York County Affordable Housing, one was a
top staff member from another non-profit that was directly involved in providing affordable
housing to towns in the county, and one was a county-based banker familiar with lending to
agencies involved in providing affordable housing. Using their expertise and available data from
the state and county, these informants provided reference modes (discussed in the following
subsection) that highlighted the problems mentioned earlier.
Income and median house price reference modes are based on full county data. The
population reference mode is based on an “average town” in York County. Because the purpose
of the model was to assist selectmen in individual towns, it too is based on an average town. The
interviewees also developed a causal loop diagram, which will be discussed in the results section
later. Lastly, a system dynamics model was developed and was used for policy testing. Because
of time constraints, the model examines only single-family housing in the typical town; modeling
of rental units will come in the future. This is not a serious deficiency, however, since in 1999
and 2000 Maine ranked /ast among the fifty states and the District of Columbia in the rate of
multifamily construction (O’Hara 2001, p. 8). While this is undoubtedly part of York County’s
problem, and perhaps part of the solution, its investigation must await future modeling.
Reference Modes
Figure 2 shows the reference modes for population, income and median house price.
Town population and median income both tracked in a moderate upward trend during the period
from 1989 to 2002. Median house price, by contrast, reflected the volatility typical of housing
markets, dipping severely from 1990 to 1992, dipping moderately from 1992 to 1995, rising
moderately from 1995 to 1999, and finally ending the period by rising significantly from 1999 to
2002. Figure 3 shows the county’s “affordability index”—the ratio of median income to the
income needed to live in affordable housing. Affordability in the county rose steadily from 1989
to 1994, and then declined steadily from 1994 to 2002, with the decline appearing to accelerate
near the end of the period. It was this lengthy and sharpening decline in affordability that led to
the modeling request from York County Affordable Housing.
10,000 people
60,000 dollar’ Year
200,000. dollar
8,000. people
40,000. dolkur/Year
140,000. dollar
6,000 people
20,000. dollar/Year
80,000 dollar
1989 199119931995 1997 1999-2001
Time (Year)
Town Population peopk:
Median Income dolku/Year
Median House Price dollar
Figure 2 Reference Modes for Town Population, Median Income and Median House Price
15
1.25
1
0.75
05
19891991 1993 1995 1997 1999-2001
Time (Year)
Affordability index Dn
Figure 3 Reference Mode for Affordabili
RESULTS
Causal Loop Diagram
The causal loop diagram obtained from the clients was extensive, and not all of it was
modeled. The part that was modeled is shown in Figure 4. (The parts not modeled dealt with
town budgets and the effects of affordable housing on town schools. These portions of the
causal loop diagram await future modeling.) The diagram has two balancing loops and one
reinforcing loop.
Balancing loop B1, “Price brake on in-migration,” says that as the median price of a
house rises, migration into the town will decline, reducing the number of households and the
occupancy rate, which reduces demand and therefore the median house price. This loop shows
how people balance demand with supply by “voting with their feet.” If housing prices are too
high in a town, people will not move into that town, demand will drop, and ultimately its house
prices will stabilize.
Balancing loop B2, “Market meets housing need,” is the flip side of loop B1. As house
prices rise, contractors are encouraged to construct more house units. Ultimately, this reduces
the occupancy rate, and prices stabilize. This loop shows how contractors balance supply with
demand by entering a town’s market with new construction. As with any market, all this entry
will eventually decline as prices stabilize and the motivation to enter diminishes.
If the two balancing loops in this diagram are classic examples of microeconomic
behavior, the reinforcing loop R1, “Out-migration leads to crowding,” is perhaps an example of
the deleterious effects of situations like the one in which the typical York County town finds
itself. This loops says that as prices rise, people do not migrate into the town. This stabilizes the
number of households, dampening demand and discouraging house construction. This leads to a
higher occupancy rate with its concomitant higher prices, further population stagnation, and so
forth. This obviously contributes to less-affordable housing, but it might be added that a side
effect of this does not directly affect the town, but certainly affects the county: people migrate to,
and houses are built in, towns that are further on the periphery of the county’s developed area,
leading to greater overall sprawl.
Average acres per
+ house unit Median income
Percentage of town )
used by house units - P+
+ Occupancy Affordability index
ea
Price brake on
in-migration +
House units Households
Median house
Out ) leads YN Town rice
to crowding ; D
population ~_
8)
Construction
Market meets
housing need
Interest rates
Figure 4 York County Causal Loop Diagram
Policy-Testing Results from Model
Using this causal loop diagram and relevant data, a 114-equation system dynamics model
was created. The model was calibrated to data, and as shown in figures 5 through 9, it produced
values that were reasonably close to the actual data for median income, population, houses,
households and occupancy for the years 1989 to 2002.
60,000
50,000
40,000
30,000
20,000
1989 1991—«1993:1995.«1997 19992001
Time (Year)
Data
‘Simulated
dollar’Year
dollar/Year
Figure 5 Comparison of Actual to Simulated Median Income
10,000
9,000
8,000
7,000
6,000
1989-1991. «1993 «199519971999 2001
Time (Year)
Data
Simulated
peopk:
peopk:
Figure 6 Comparison of Actual to Simulated Town Population
6,000
5,000
4,000 BaeeeeS? ane
3,000
2,000
19891991 1993 199519971999 2001
Time (Year)
Data house
‘Simulated house
Figure 7 Comparison of Actual to Simulated House Units
4,000
3,500
3,000
2,500
2,000
19891991 1993 199519971999 2001
Time (Year)
Data house
‘Simulated house
Figure 8 Comparison of Actual to Simulated Households
07
06
1989 1991 1993 1995 1997 1999 2001
Time (Year)
Data Dnml
Simulated Dnml
Figure 9 Comparison of Actual to Simulated Occupancy
The model was then used to test the effects of five policies, each of which would be
started in 2005 and run to 2013:
1. The status quo. This policy keeps the house lots at 2 acres (.8 hectares) and
maintains construction at 50 houses per year.
2. Smaller lots. This maintains construction at 50 houses per year, but allows lots
as small as .5 acre (.2 hectares).
3. Construction cap. This keeps house lots at 2 acres but cuts allowable
construction to 25 houses per year.
4. Increased construction. This keeps house lots at 2 acres but allows construction
to rise to 75 houses per year.
5. Smaller lots and increased construction. This policy reduces the lot size to .5
acres and increases construction to 75 houses per year.
Results of these policy tests are shown in Figures 10, 11 and 12. Figure 10 shows the effects of
all five policies on sprawl. Figures 11 and 12 show the effects for median house price and
affordability, respectively. (The “status quo” and “smaller lots” policies are combined in the two
figures, as are the “increased construction” and “smaller lots and increased construction”
policies, since their effects on median house price and affordability are similar. Only the
“construction cap” policy is unique in its effect on price and affordability.)
Status quo. This policy results in a mix of good and bad outcomes. One good outcome
is shown in Figure 11, which shows how this policy results in a lower median house price.
Figure 12 shows a similar good outcome for affordability, which improves from about .6 to
about 1.9. However, these are offset by the bad outcome shown in Figure 10, which shows that
the status quo policy results in more sprawl, with over 47% of the town covered by house units.
60
45
30
15
0
1989 1993 1997 2001 2005 2009 2013
Year
Status quo Dil
Smaller lots Dal
Construction cap Dil
Incteased construction Dil
Smaller lots AND increased construction. ———_—$_$$ $$$ $$$ Din!
Figure 10 Percentage of Town Covered by House Lots Under Various Policies
300,000
225,000
150,000
75,000
0
1989 1993, 1997 2001 2005 «2009-2013
Year
Status quo OR smaller lots dollar
Increased construction OR Smaller lots and increased construction ———— dollar
Construction cap dollar
Figure 11 Median House Price Under Various Policies
NR
-1
1989 1993 1997 2001 2005 2009 2013
Year
Status quo OR smaller lots, Dnnl
Increased construction OR Smaller lots and increased construction Dn
Construction cap Dnml
Figure 12 Affordability Index Under Various Policies
Smaller lots. Figure 10 shows one good outcome from this policy—less sprawl, the least
of all the policies tested. The results for median house price (Figure 11) and affordability
(Figure 12) are identical for those of the “status quo” policy. So this policy has some merit—it
raises affordability and reduces sprawl. Implementing this policy would be similar to the many
“cluster housing” policy suggestions that have been made in recent years (e.g., O’Hara 1997).
Construction cap. By limiting the number of house units constructed in a town, this
policy definitely reduces sprawl, actually more so than the “status quo” or “increased
construction” policies (Figure 10). Unfortunately, as Figures 11 and 12 show this policy has the
least favorable effect on median house price and affordability index. The very thing that helps
with the sprawl problem—a limit on the number of houses—produces a shortage of supply that
raises prices. This very common policy is really not a very good one from the standpoint of
affordability.
Increased construction. It is probably not surprising that this policy has the second-
worst effect on sprawl (Figure 10), since it would entail a higher number of large-lot house units.
The upside of the policy is that its increase in housing supply results in lower prices
(Figure 11) and a better affordability index (Figure 12). This may be similar to what happens in
York County towns that are on the periphery of the developed area. They get a surge of
construction because they have relatively low prices, but doing so contributes to sprawl. This
policy is helpful from the standpoint of affordability, but is probably not a truly viable policy
because of its effect on sprawl.
Smaller lots and increased construction. This policy moderates sprawl, but not quite
as much as the “smaller lots” policy alone (Figure 10), since it results in more house units being
constructed. It is as effective as the “increased construction” policy at reducing the median price
(Figure 11) and improving the affordability index (Figure 12). This policy seems optimal, in that
it achieves both policy objectives—it reduces price (by increasing the supply of housing) and it
manages sprawl the most effectively (by having the increased construction be on smaller lots).
DISCUSSION AND CONCLUSION
The present paper illustrates one thing for sure—if York County, Maine, is to solve the
twin problems of lack of affordable housing and sprawl, Boards of Selectmen in its towns must
abandon their traditional policies, which typically feature large lot sizes and occasional
construction caps. The caps can alleviate sprawl, and perhaps reduce pressure on town services,
but at the cost of a lower amount of affordable housing. Increased construction can help the
affordable housing problem, but the large lot sizes exacerbate sprawl. While it is beyond the
scope of the model used in the present paper, there is reason to believe that caps also exacerbate
sprawl regionally, when the resulting high house prices encourage prospective homeowners to
build in neighboring towns. What is needed is a two-pronged policy of increased construction
and smaller lots.
One objection that is frequently voiced by Boards of Selectmen is that increased
construction will lead to excessive population in the town. But Figure 13 shows that the policy
recommended here would, when compared to the “status quo” policy, increase the town’s
population by only about 60 households or approximately 150 residents. Another frequent
objection is that houses on smaller lots will decrease property values. But there is evidence that
even attached apartments can support house prices (Nelson and Bell 2003, p. 6), so it seems
unlikely that well-planned subdivisions of small-lot houses would be detrimental. As Nelson and
Bell (2003) argue, well-planned affordable housing can increase a town’s housing choices and
make its housing more desirable, and may even increase the pool of buyers for more expensive
housing.
Another lesson of the present paper’s analysis is that York County, and indeed perhaps the entire
state of Maine, needs to take a more regional approach to these problems. Unfortunately, this
will be difficult to implement, since Maine has a very strong “home rule” ethic (Richert 2003)
and a centuries-long tradition of local control (Bouchard 2003). (Home rule is the term for the
state allowing the towns to have a high amount of control over local policy decisions.) That is,
even when there are deleterious effects, Maine’s people and their politicians tend to prefer
decisions made within small jurisdictions. As Richert puts it:
[The New England town’s] belief that home rule is not merely a principle of
governance but the armor that keeps the external forces of change at bay is
unshaken by the realities around them. Home rule in today’s small political
jurisdictions packs plenty of political power but, with respect to the regional
forces washing over towns, it is an illusion. (Richert 2003, pp. 2-3.)
As an example, a special state task force published a report in 2000 that called for changes in
Maine statutes to compel smaller lot sizes in certain circumstances (Maine Task Force to Study
Growth Management 2000). The recommendation has yet to be adopted. The point is that
Boards of Selectmen in York County do have some things under their control. They can allow
more construction in their towns, with smaller lot sizes. If selectmen can overcome their natural
political inclination to view problems like affordable housing and sprawl as someone else’s, this
policy will help solve these problems, both in the town itself and for the entire county. In that
sense, the present paper provides support for policy making at the town level.
6,000 house
12,000. people
5,000, house
10,500 people
4,000 house
9,000. people
3,000, house
7,500 people
2,000 house
6,000 people
19891991 —*1993 1995
Households Status quo
197
199)
2001
Year
2003
2005
2007
2009
201
“Households Smaller lis and increased construction
Population Status quo
Population Smaller lots and increased construction
2013
hse
base
people
people
Figure 13 Households and Population Under “Status Quo”
and “Small Lots and Increased Construction” Policies
Further research based on this model could also yield useful results. One useful
extension would be modeling of multi-family housing units, which, even though they make up
less than 30% of York County’s housing, play an important role in affordable housing.
Examination of this housing segment would be particularly informative given Maine’s position
at the bottom of the ranks of states constructing multi-family housing. Another useful line of
future research would be to become more fine-grained in the examination of types of house units
in the typical York County town, along the lines of what Forrester did in his Urban Dynamics
model (Forrester 1969).
Even without these potential refinements, the model in its present state shows that towns
in York County, Maine, do have some policy choices at their disposal. In particular, it shows
that smaller lot sizes coupled with increased construction would have minimal negative effects
on sprawl and quite positive effects on the amount of affordable housing. It will be interesting to
see what effect the insights generated from this model will have on Boards of Selectmen in the
county.
REFERENCES
Bouchard, Kelley. 2003 “Regionalization confronts tradition of local control.” Maine Sunday
Telegram, June 29, 2003. Retrieved January 27, 2004, from web _ site
http://news.mainetoday.com/indepth/taxreform/030629regional.shtml
Encyclopaedia Britannica. 2004 “York.” Retrieved January 22, 2004, from Encyclopedia
Britannica Online. http://www.search.eb.com/eb/article?eu=96013
Forrester, Jay. 1969 Urban Dynamics. Portland, Oregon: Productivity Press.
Maine State Housing Authority. 2002 The State of Maine’s Housing 2002. Augusta, Maine.
Maine State Planning Office. 2004 “Expansion of Development Southern Maine, 1940-2050.”
Retrieved on January 23, 2004 from web site
http://www.state.me.us/spo/landuse/techassist/expansion/region2.php
Maine Task Force to Study Growth Management. 2000 “Final Report.” Augusta, Maine.
Nelson, Arthur C. and Bell, Carol A. 2003 “Let’s Get Efficient About Affordability.” Housing
Facts and Findings 5,1: 3-7.
Office of Federal Housing Enterprise Oversight (OFHEO). 2003 “Third Quarter 2003 House
Price Index.” News Release, December 1, 2003, Washington, DC.
O’Hara, Francis. 1997 “The Cost of Sprawl.” Maine State Planning Office, Augusta, Maine.
O’Hara, Francis. 2001 “Houses, Jobs and Maine People: 2001.” 2001 Report to the Maine
Governor’s Affordable Housing Conference, Sept. 10, 2001.
Richert, Evan. 2003 “Regionalism, New England Style.” Choices (Maine Center for Economic
Policy) 9,4: 1-6.
U.S. Census Bureau. 2004a “Small Area Income and Poverty Estimates 2000 State and County
FTP Files: Maine data.” Retrieved on January 23, 2004 from web site,
http://www.census.gov/housing/saipe/estmod00/est00_ME.dat
U.S. Census Bureau. 2004b “Small Area Income and Poverty Estimates 2000 State and County
FTP Files: U.S. data.” Retrieved on January 23, 2004 from web site,
http://www.census.gov/housing/saipe/estmod00/est00US.dat
U.S. Census Bureau. 2004c “State and Country Quick Facts: York County, Maine.” Retrieved
on January 23, 2004, from web site,
http://quickfacts.census.gov/qfd/states/23/2303 1.html
York County, Maine. 2004 “Historic York County.” Retrieved January 23, 2004, from web
site, http://www.co.york.me.us/history/
Back to the
CC BY-NC-SA 4.0