The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
Distance and Decline: The Case of
Petersburg, Virginia
WP 18-16
Raymond Owens
Federal Reserve Bank of Richmond
Santiago Pinto
Federal Reserve Bank of Richmond
Distance and Decline:
The Case of Petersburg, Virginia∗
Raymond Owens† & Santiago Pinto‡
October 17, 2018
Working Paper No. 18-16
Abstract
Petersburg, Virginia, prospered over two centuries as a center of production and trade. However,
the city experienced economic difficulties beginning in the 1980s as a large number of layoffs at
production plants in the area coincided with an erosion of retail trade in the city. Prolonged
economic decline followed. In contrast, somewhat similar shocks in other moderate-sized cities in
Virginia were followed by gradual economic recovery. We examine these differing outcomes and
offer an explanation that hinges on the proximity of Petersburg to its larger neighbor, the greater
Richmond area. We find evidence suggesting that after the job declines, higher-skilled residents
in Petersburg initially commuted to jobs nearer to Richmond, later relocating from Petersburg
toward Richmond–an option not readily available in the other Virginia cities considered. We
suggest that, as a result, Petersburg suffered a sharp decline in tax revenues and that municipal
costs could not be proportionately scaled down, leading to severe fiscal stress.
JEL Classification: R23, R40, R51
Keywords: spatial equilibrium, urban decline
∗
The views expressed herein are those of the author and are not necessarily those of the Federal Reserve Bank of
Richmond, or the Federal Reserve System. The authors would like to thank participants of the 2018 Annual Meeting
of the Virginia Association for Economists for valuable comments.
†
Federal Reserve Bank of Richmond; Raymond.Owens@rich.frb.org.
‡
Federal Reserve Bank of Richmond; Santiago.Pinto@rich.frb.org.
1
Introduction
Cities and towns arise and some subsequently disappear for a wide variety of reasons. Those that
arise may sit at an advantageous location or may be adjacent to resources. But some–such as mining
towns– may disappear soon after the ore runs out and their initial advantage disappears. Others
persist long after their initial advantage is lost (Bleakley and Lin (2012)). Petersburg, Virginia, falls
into the second category. Situated at the fall line of the Appomattox River, Petersburg’s location
represented the farthest point inland that colonial-era ships could navigate. The fall line was also
the shortest distance that relatively costly-to-transport goods produced in the interior had to travel
to be shipped by vessel. This combination led to the importance of Petersburg as an early trading
center.1
Because of its location, Petersburg rose to prominence as a colonial-era city and played a crucial
role in the Civil War. After the war, it served as a transportation hub and a regional trade center
for over a century. It remained a growing city until a few decades ago. In recent years, though, it
has received national attention for its dire fiscal situation. The city had, for example, some of its fire
equipment repossessed, and delayed maintenance led to the failure of its water system (Schneider
(September 6, 2016)). But even deeper issues plagued the city. Newspapers reported that the city’s
finances were being investigated by the Department of Justice in late 2016 (Buettner (October 13,
2016)).
How did Petersburg fall so far in recent decades when other, somewhat similar, cities in Virginia
generally have exhibited modest economic growth? One intriguing possibility is that technological
advances in transportation resulted in Petersburg becoming effectively too “close" to its more
economically vibrant neighbor Richmond to endure some of the negative economic shocks that other
similar cities have seemingly weathered.2 The role that effective distance may have played on the
ability of Petersburg to recover from local economic shocks will be examined in this paper.3
To understand Petersburg’s transition from economic prominence to its present fiscal difficulties,
we will briefly review Petersburg’s history and look at its economic structure in recent decades.
Next, a simple model of two cities is proposed and following that, a similar model for a standalone or
“isolated" city is proposed. The implications from these models are matched to Petersburg’s economic
features and demographics measures and similar measures in the control cities of Waynesboro and
Lynchburg.
1
Burnett et al. (2017) provides an extended description of the history of Petersburg, along with the challenges
faced by the city in recent times.
2
Throughout the paper we use the term “Richmond" loosely. Specifically, we refer to those localities within the
greater Richmond area, neighboring or very close to Petersburg, that have been experiencing significant economic
growth for the last fifty years, such as Chesterfield County and Henrico County, among others. Chesterfield County, just
north of Petersburg, separated only by the Appomattox River, has seen an increase in population from approximately
77,000 in 1970 to 340,000 in 2010. The population in Henrico County has more than doubled during the same period.
3
We use “effective" distance to denote the actual travel time between Petersburg and Richmond. Prior to I-95 (and
later I-295), the cities were linked by Rt. 1 and Rt. 301. These highways passed through both dense residential and
commercial areas between the cities. As such, traffic congestion as well as stop lights made travel slow. With I-95 and
I-295, travel time went down considerably, lessening the “effective" commuting distance for residents.
2
2
A Brief History of Petersburg
Like many fall line cities and towns in Virginia, Petersburg continued to exist and prosper long after
ships ceased being an important transportation mode. In part, the scale economies that developed
early in its history made the city an attractive connecting point for later, more efficient modes of
transportation, such as turnpikes, railroad lines, and eventually, interstate highways (Bleakley and
Lin (2012)). All of these subsequent forms of transportation passed through Petersburg, with both
railroads and highways intersecting there. Each new, more efficient mode of transportation enhanced
Petersburg’s connectivity with surrounding areas and added to the city’s locational durability but,
ironically, may have also ultimately contributed to Petersburg’s decline.
Along with transportation connections, Petersburg’s location along with its infrastructure and
population size made the city an attractive location to process and manufacture regionally produced
agricultural products. In the post-Civil War period, the city became a center of cigarette, furniture,
and textile production. Those industries, along with chemicals, trade, and services sustained
Petersburg’s economy through the 1970s.4
It was early in that decade when the Petersburg City Council, perhaps encouraged by the new
interstate highways intersecting there and the opening of retail malls in the city, made the decision
to annex about fourteen square miles of adjoining territory from neighboring Dinwiddie and Prince
George Counties. Following several court challenges, the annexation was approved and went into
effect on December 31, 1971. Petersburg officials believed the area had great potential for future
development and found the prospect for increased tax revenues attractive. While the city would
provide infrastructure and services for the annexed area, the existing population there was relatively
small, generating only modest tax revenues. The payoff to the city hinged on future substantial
development in the incorporated area, but that development never fully materialized. That would
ultimately have important local economic consequences, especially on the public finances of the
city.5
By the mid-to-late 1970s, though, the traditional industries that had sustained Petersburg’s
economy for nearly a century were changing. A pivotal event was the 1982 announcement by tobacco
producer Brown and Williamson that they would eliminate one-third of their 2800 employees
in the city. By 1985, the firm would completely shutter their antiquated Petersburg operation.
Adding to the city’s woes, Southpark Mall was constructed in the late 1980s north of Petersburg
in Colonial Heights. The new mall –located along I-95– could draw shoppers from the rapidly
developing southern Richmond suburbs, as well as those in Petersburg and areas south. Storeowners
in Petersburg saw the potential of the new mall and a number relocated there, leading to the closure
of the city’s Walnut Mall, which had opened in the 1960s, amid much optimism and anticipated
4
Section D in the Appendix, shows a timeline of Petersburg’s recent history.
We leave for future research a complete assessment of institutional framework that regulates the decision by
localities to annex, or, if the process fails, to “deannex" territory, and its economic implications. In fact, approximately
thirty-six states have deannexation laws, but such a mechanism is rarely used in practice (see Moreo et al. (2015)).
5
3
sales tax revenues for Petersburg. Originally, city leaders viewed Walnut Mall as a facility that
would serve a broad base of households throughout southside Virginia. However, Southpark Mall
would ultimately fulfill that role, bypassing revenue generation for Petersburg.
In many ways, the events unfolding in Petersburg were not unique to the city, as other small
cities in Virginia were exposed to the broader shifts in industry and retail taking place during
the early 1980s. In Waynesboro, for example, large firms such as DuPont, General Electric, and
Compton had been important local employers for many years. But the recession in the early 1980s
combined with global shifts in production weakened sales at these firms. Weaker sales, in turn,
translated into local layoffs numbering in the thousands from 1980 through 1983. By the middle of
the decade, however, jobs in the city and the immediate surrounding areas were growing on balance,
though DuPont continued to shed positions – many through early retirement.6 Craddock Terry, a
large shoe producer, was established in Lynchburg in 1888, and Lynchburg Foundry was founded
eight years later. Both laid off substantial portions of their workforces in the 1980s, especially in
the first half of the decade following the back-to-back economic recessions. In the first three months
of 1982, for example, these companies, along with General Electric, Babcock and Wilcox, and other
firms with a smaller local presence, laid off 6,740 workers. This was a severe blow to a town of
67,000 at the time.7
The negative shocks experienced by Petersburg, Waynesboro, and Lynchburg were typical of
those that hit small specialized towns and cities across America in the 1980s and 1990s.8 Cities
that were specialized were vulnerable to sectoral or industrial shocks as advances in technology
and globalization combined to alter both the form and location of production globally. But towns
and cities differed in their post-shock experiences. Some eventually rebounded, if slowly, from the
episodes, while others, such as Petersburg, did not. While a number of factors likely contributed to
the differing outcomes, we focus on one particular explanation: the distance to a larger city may
partly explain the extent to which small cities recover after experiencing a negative economic shock.
Our approach is to look at Petersburg, which lies about twenty miles south of Richmond, Virginia,
the state capital, with a Metropolitan Statistical Area (MSA) population of about 1.3 million people.
Petersburg lies within the Richmond MSA, though at the southern edge. We propose that a small
city that lies close enough, defined as easily commutable, to the core larger city will face hurdles in
its recovery from a negative economic shock that are stronger than those faced by a similar city that
6
See, for example, Nardi (March 15, 1987).
See, for example, Goodman (March 2, 1982).
8
The present work focuses on the economic and social challenges faced by small cities whose economic growth
has historically depended on a narrow set of economic activities (in other words, cities that have specialized in
the production of one or two goods) and that have been hit with a substantial negative economic shock. In these
specialized cities, employment is generally concentrated in a small number of relative large firms. Some research in
urban economics has focused on the relationship between the degree of city specialization, city size, and city growth.
For instance, Glaeser et al. (1992) found evidence that a diverse local productive structure (rather than a specialized)
is associated with strong local growth. Duranton and Puga (2000) show that smaller urban areas tend to be less
diverse in terms of their employment composition. However, the precise mechanism through which the local industry
composition affects urban growth is more complicated and less clearly understood. For a thorough and complete
discussion on this topic, see Henderson (2005) or Duranton and Puga (2014).
7
4
is isolated (“distant enough" from a larger city). Finally, we briefly explain how the local public
finances may reinforce the negative effects of the adverse economic shock and highlight the fact that
some decisions made by local officials, namely the annexation of a sizeable area by Petersburg in
the early 1970s, could have imposed additional financial burdens on the city’s finances.
The next section proposes a model of distance and fiscal impacts of a relatively small city that is
subject to a negative economic shock. The first model allows for the city to be located near “enough"
to a bigger city, a system of two cities. The second describes the spatial equilibrium when a city is
isolated from the rest. The next section aligns the model’s results, the outcomes expected both
from the interaction between the small and large cities and the isolated city scenario. Section 4
compares data on the evolution of economic and demographic indicators in Petersburg, Waynesboro,
and Lynchburg in recent decades. The objective of this section is to compare some stylized facts to
determine if these are consistent with the predictions of the models. The last section concludes.
3
The model
We consider a closed, linear urban area that has unit width, represented by the interval [0, x̄]. There
are two employment centers or cities: R and P . The central business district of city R (CBDR
hereafter) is located at 0, while the CBD of city P (CBDP ) is located at the other end x̄. Distance
from CBDR is denoted by x and distance from the CBDP by (x̄ − x). The distance x̄ between the
two employment centers is short enough so that it is feasible for residents at any location x in [0, x̄]
to commute to work in either CBDR or CBDP .
Individuals receive labor income from working at either CBDj with j = R, P . The population
consists of two types of individuals characterized by their skill level: a group of high-skilled
residents/workers, with population size Nh , and a group of low-skilled residents/wokers with
population size N` . High-skilled individuals earn higher wages than low-skilled individuals at every
location. Specifically, we assume that whR > whP > w`R > w`P .
Urban residents derive utility exclusively from the consumption of a composite nonland good,
denoted z. In other words, the utility of an urban resident is simply z. The price of z is normalized
to one. Individuals who commute from their residential location x to either CBDR or CBDP face a
commuting cost of t dollars per mile of travel. Residents directly consume a fixed amount of land qi ,
i = h, `. We assume that land consumption of high-skilled individuals qh is fixed at one unit, and
land consumption of low-skilled residents is q` = q < 1. Moreover, we assume that whj /qh > w`j /q` ,
which is consistent with a positive, but less than one, income elasticity of housing demand.9
Ultimately, individuals jointly decide their residential location and where to commute to work.
Specifically, the utility of an individual of type i who resides at location x and commutes to CBDj
9
Empirical evidence seems to indicate that this elasticity is between 0.3 and 0.9.
5
is
uji (x) = wij − t x 1{CBDR } + (x̄ − x) 1{CBDP } − rij qi ,
h
i
(1)
where 1{CBDj } is an indicator function that is equal to one when an individual who resides at
location x commutes to CBDj , and rij is the rent per unit of land paid by a type-i individual who
works at CBDj . For simplicity, land rent is set equal to zero when land is devoted to nonurban
purposes. We assume that urban area is large enough to accommodate the entire population, i.e., x̄
= Nh qh + N` q` = Nh + N` q.
3.1
Equilibrium: Definition
In an urban equilibrium model with two employment centers, individuals of each type must
be indifferent across all residential locations x ∈ [0, x̄] and between commuting destinations or
employment centers CBDR and CBDP . Specifically, for each type i = h, `, uji (x) = ūi , for
j = R, P . Since we are considering a closed urban equilibrium model, the utilities ūi are endogenously
determined. The highest land rent each individual of type i is willing to pay per unit of land at
location x is
wiR − tx − ūi
when commuting to CBDR ,
qi
wP − t(x̄ − x) − ūi
riP (x) = i
when commuting to CBDP .
qi
riR (x) =
(2)
(3)
Different types of spatial configurations may arise in equilibrium depending on the slopes and
intercepts of the bid-rent functions. Among other things, the equilibrium spatial configurations
depend on the slopes and intercepts of the bid-rent functions. First, note that the slopes of the
bid-rent functions are, in absolute value, equal to 1/qi . Since 1/qh < 1/q` , the bid-rent function of
high-skilled individuals is always flatter than the respective slope of low-skilled individuals. Second,
the intercepts of the curves, (wij − ūi )/qi , determine whether they cross at a location x ∈ (0, x̄)
or whether one curve is always above the other. We initially focus on an equilibrium of the type
illustrated in the top panel of Figure 1 in which type-` individuals reside close to the respective
CBDs (those at locations x ∈ [0, x̂R ] commute to CBDR , and those at locations x ∈ [x̂P , x̄]
commute to CBDP ), and type-h individuals tend to reside in the suburban areas (they reside at
locations x ∈ [x̂R , x̂P ], where residents in [x̂R , x∗ ] commute to CBDR , and residents in [x̂P , x̄]
commute to CBDP ).10
We formally define an equilibrium (as the one characterized in Figure 1) as a list of values {x̂R ,
x∗ , x̂P , ūh , ū` } that satisfies the following conditions:
1. r`R (x̂R ) = rhR (x̂R ),
10
So, for instance, an equilibrium in which individuals of type ` reside closer to the respective city centers, and
individuals of type h reside in the suburbs of the cities would be characterized by (w`j − ū` )/q` > (whj − ūh )/qh , in
addition to the fact that the bid-rent curve of type-h individuals is flatter than the bid-rent curve of type-` individuals.
Since whj /qh > w`j /q` , then, in equilibrium, the relative utilities satisfy (ū` /ūh ) < q` /qh = q.
6
2. rhR (x∗ ) = 0,
3. rhR (x∗ ) = rhP (x∗ ),
4. rhP (x̂P ) = r`P (x̂P ), and
5. x̂P − x̂R = Nh .
The locations x̂j , which define the borders between high- and low-skilled individuals in the residential
area surrounding CBDj , are determined by the intersections of the bid-rent functions of high- and
low-skilled individuals who commute to city R (condition (1)) or city P (condition (4)). Condition
(5) states that there should be enough land for all type-h individuals residing in the area. Location
x = x∗ separates high-skilled individuals who commute to CBDR from high-skilled individuals who
commute to CBDP (defined by (3)). Moreover, at x = x∗ , land rent, when land is used for residential
purposes, should be that same as land rent in the alternative (agricultural) use, assumed to be zero
(condition (2)). As mentioned earlier, we will focus on equilibria in which 0 ≤ x̂R ≤ x∗ ≤ x̂P ≤ x̄.
The market land rent at each location x is given by r (x) = max[r`R (x), rhR (x), rhP (x), r`P (x)], where
r (x) ≥ 0 for all x ∈ [0, x̄] and r (x∗ ) = 0.
3.2
Characterization of the equilibrium
Solving (1) - (5), we obtain
x̂R =
x̂P
=
x∗ =
ūh =
ū` =
x̄ − Nh q (whR − whP ) − (w`R − w`P )
−
,
2
2(1 − q )t
x̄ + Nh q (whR − whP ) − (w`R − w`P )
−
,
2
2(1 − q )t
x̄ whR − whP
+
,
2
2t
tx̄
whR + whP
− ,
2
2
w`R + w`P
t [x̄ − Nh (1 − q )]
−
.
2
2
(4)
(5)
(6)
(7)
(8)
Note that x̄ − Nh (1 − q ) > 0.11 Consider, initially, an equilibrium in which whR = whP = wh and
w`R = w`P = w` . Then,
x̄ − Nh P
x̄ + Nh ∗
x̄
tx̄
t [x̄ − Nh (1 − q )]
, x̂ =
, x = , ūh = whR − , ū` = w`R −
.
(9)
2
2
2
2
2
We assume that this initial equilibrium, more precisely x∗ = x̄/2, determines the border between
x̂R =
the “metropolitan statistical areas" (M SAs) that contain the respective employment centers CBDj .
With some abuse of notation, we denote these areas as M SAj , so the interval [0, x∗ ] defines the
M SAR and (x∗ , x̄] defines M SAP . Moreover, this border is assumed fixed hereafter.12
11
Since x̄ = Nh + N` q, then x̄ − Nh (1 − q ) = q (Nh + N` ) > 0.
An initial symmetric equilibrium is chosen to simplify the exposition. The conclusions do not qualitatively change
if wiR and wiP are initially different.
12
7
3.3
Negative productivity shock to CBDP
Next, suppose that CBDP is hit by a negative productivity shock that decreases wages of both
types of workers in the same proportion, so all types of workers who commute to CBDP receive
(1 − θ )wi , where 0 ≤ θ < 1.13 The new solution becomes
x̄ − Nh θ (qwh − w` )
x̄ + Nh θ (qwh − w` )
−
, x̂P =
−
,
2
2(1 − q )t
2
2(1 − q )t
(2 − θ )
tx̄
(2 − θ )
t [x̄ − Nh (1 − q )]
wh − , ū` =
w` −
.
2
2
2
2
x̂R =
ūh =
x∗ =
θwh x̄
+ ,
2t
2
(10)
(11)
We now examine how the equilibrium changes with the magnitude of the negative shock.14 Let Lji
denote the number of type-i individuals who work at CBDj , and Lj = Lj` + Ljh the total number of
workers at CBDj . Then,
LR = x̂R /q + (x∗ − x̂R )
| {z }
LR
`
|
{z
LR
h
and LP = (x̄ − x̂P )/q + (x̂P − x∗ ) .
}
|
{z
LP
`
}
|
{z
LP
h
(12)
}
From the comparative static results
∂ x̂R
∂ x̂P
(qwh − w` )
=
=−
< 0,
∂θ
∂θ
2(1 − q )t
∂x∗
w
= h > 0,
∂θ
2t
∂ ūi
wi
= − < 0,
∂θ
2
(13)
we can conclude the following. First, ∂LR /∂θ = −∂LP /∂θ > 0, so a shock that negatively affects
CBDP attracts more workers to CBDR and fewer to CBDP . Second, there is also a change in
R
the composition of commuters to each location. Since ∂LR
h /∂θ > 0 and ∂L` /∂θ < 0, the share of
high-skill workers increases and the share of low-skill workers decreases at CBDR . The opposite
effects take place at CBDP . Third, the resulting relocation of type-h individuals from CBDP to
CBDR displaces some type-` workers who used to reside in M SAR and commute to work to CBDR
before the shock. Fourth, if the negative shock in P becomes sufficiently large (in this case, if θ
is equal to some critical value θc ≡ [Nh (1 − q )t]/(wh − w` )), then only type-` workers will end up
commuting to CBDP , i.e., x∗ (θc ) = x̂R (θc )). And fifth, even though the shock initially affected
CBDP , the negative effect of the shock is propagated throughout the entire area, reflected in a
decrease in the utility of all residents. Specifically, ∂ ūi /∂θ = −wi /2 < 0.
The negative shock will also affect total production in each city and total land rents in the
respective M SAs. Production in each CBD is given by
R
Y R = w` L R
` + wh L h ,
Y P = (1 − θ )w`R LP` + (1 − θ )wh LPh ,
(14)
and total land rents in each M SA by
R
R =
Z x∗
0
max{r`R (x), rhR (x)},
P
R =
Z x̄
x∗
max{r`P (x; θ ), rhP (x; θ )},
(15)
where the land price gradients riR (x) and riP (x; θ ), defined in (2) and (3), respectively, are evaluated
13
Note that this is a shock that affects all workers in the city in the same way.
The exogenous variable of interest in the analysis is θ. In fact, the political boundary between the two M SAs is
determined by x∗ (θ ) when θ = 0.
14
8
at the equilibrium utilities ūh and ū` .15 Consider the impact of the shock on production:
∂Y R
∂θ
∂Y P
∂θ
∂ (Y R + Y P )
∂θ
(qwh − w` )2 wh2
+
> 0,
2q (1 − q )t
2t
∂Y R
w` N` + wh Nh
− (1 − 2θ )
= −
< 0,
2
∂θ
∂Y R
w N + wh Nh
+ 2θ
= − ` `
,
2
∂θ
=
(16)
(17)
(18)
where the inequality in (17) holds for θ < 1/2. Note that the effect on Y R + Y P consists of two
terms: the first one represents the immediate impact on production in CBDP , which is negative,
and the second one captures the positive effect that the relocation of workers has in CBDR . As a
result, the negative initial effect is, in the aggregate, partially compensated by the higher production
taking place in CBDR .
Graphically, the effects of the negative shock to CBDP are illustrated in Figure 1. The figure
depicts two spatial equilibria. The top panel presents the initial equilibrium, where the two cities
are identical and θ = θ0 = 0. As stated earlier, type-` individuals reside closer to the respective
CBDs (at locations x ∈ [0, x̂R ] and x ∈ [x̂P , x̄]), and type-h individuals reside in the suburbs of the
CBDs (at locations x ∈ [x̂R , x∗ ] and x ∈ [x∗ , x̂P ]). The bottom panel of Figure 1 shows the new
spatial equilibrium when a shock θ = θ1 > 0 negatively affects CBDP . As a result of the shock,
individuals tend to relocate: type-h individuals move away from CBDP toward CBDR , displacing
some of the type-` individuals from locations closer to CBDR to locations closer to CBDP . The
figure shows an extreme case in which the negative shock is so large (in this case, θ1 = θC , as defined
above) that only individuals of type ` end up commuting to the employment center in city CBDP .
In other words, high-skill employment at CBDP has been entirely replaced by low-skill positions.
15
The notation riP (x; θ ) means that income at CBDP is (1 − θ )wi . Also, note that total land rents before and after
the shock are calculated over the fixed intervals [0, x∗ ] and [x∗ , x̄], since the M SAs are assumed to be defined by the
initial situation, without the shock.
9
Figure 1: Urban equilibrium:
Before the shock, θ = 0.0 (top panel), and after the shock, θ = 0.1 (bottom panel)
To further characterize how the negative shock to CBDP affects the spatial equilibrium, we
construct a very simple numerical example based on Figure 1. Table 1 in Appendix A summarizes
the results of the numerical exercise for different values of θ, including the two cases examined
10
in the figures (θ0 = 0.0 and θ1 = 0.1), and other values of θ between them.16 In addition to the
equilibrium values of x̂R , x∗ , and x̂P , the table also includes total land rents in each M SA, RR and
RP , the total amount produced at each CBD, Y R and Y P , and equilibrium utilities ūi . First, total
land rents in M SAR increase, and total land rents in M SAP decrease, but in absolute value, the
former is substantially larger than the latter. Second, as expected, total production goes down in
CBDP and up in CBDR . However, the negative shock in CBDP has interesting qualitative effects.
Consider the changes taking place when θ increases from 0.00 to 0.10, i.e., a 10% negative shock.
Initially, production in each CBD is 2.70, so aggregate production Y R + Y P is 5.40. The immediate
effect of the shock (i.e., the short-run effect ignoring the spatial relocation of residents and labor)
is to reduce production in CBDP from 2.70 to 2.43 (10% reduction entirely due to the shock),
and production in CBDR is unaltered. Hence, aggregate output declines to 5.13 (5% decline). As
residents and workers relocate and commuting patterns change, local production will be affected
further. Specifically, Y R increases to 3.60 (33% increase), and Y P decreases to 1.62 (40% decline).
At the end, overall production Y R + Y P goes down to 5.22 (3.33% decline).
Finally, note that since we consider a closed system of cities, a negative shock to CBDP adversely
affects the utilities of all types of individuals residing in the area.17 However, even though the initial
negative productivity shock affects low- and high-skilled workers in CBDP in the same proportion
θ, low-skilled individuals end up experiencing a proportionally larger negative effect on utility. In
the example, while ūh declines in 6.25%, ū` declines in 6.90%.
3.4
Implications of the analysis
The key underlying assumption in the model presented here is that when economic activity declines
in one city, workers and residents in the area have the option to commute and eventually later
move toward a nearby and more vibrant employment center. This movement is amplified if the
costs of doing so are relatively low. Under these conditions then, what does our simple spatial
equilibrium model predict will happen in the region? We consider below the changes that would
take place following a negative shock at each employment center CBDj and in the surrounding
areas or M SAs.
According to the model, a negative shock in city P (i.e., a uniform shock that negatively affects
the productivity of all types of workers, low- and high- skilled, in the same proportion) is expected
to have the following effects:
1. Total employment at CBDP declines, and total employment at CBDR increases.
The bottom panel of Figure 1 is consistent with a value of θ1 = θC = 0.1.
If nothing else happens in the region, then it is expected that in the long run some individuals will leave the
area if alternative regions offer higher net-of-moving costs utilities. In a previous version of the present model, we
consider a situation in which the overall productivity in city P declines, and the productivity of type-h jobs increases
at CBDR . In this case, the utility of type-h workers may actually increase if the latter dominates the former effect
for this type of workers.
16
17
11
2. The composition of residents at each M SA and the composition of workers commuting to
each city change as follows:
• The proportion of high-skilled residents increases in M SAR .
• Similar changes are observed in the composition of workers commuting to each city: the
proportion of high-skilled workers commuting to CBDR increases, and the proportion of
low-skilled workers commuting to CBDP increases.
• A larger proportion of high-skilled workers who still reside in M SAP (recall that the
model predicts they will locate in the suburbs of city P ) will start commuting to
the more economically vibrant employment center CBDR . Eventually, if the negative
shock in CBDP is large enough (as shown in the bottom panel of Figure 1), then
all type-h individuals who reside in M SAP (represented in the graph by the segment
[x̂P (θ1 ) − x∗ (θ0 )) will commute to CBDR .
3. The negative shock on M SAP reduces production in CBDP and increases production in
CBDR . Some workers, in particular the most productive ones, are induced to relocate to
CBDR . As the workforce in CBDP becomes increasingly low-skilled, the production in
CBDP declines further, amplifying the initial negative impact. Overall, at the regional level,
the decline in production CBDP is partially compensated by the higher production taking
place in CBDR .
4. Land prices will tend to rise at all locations in M SAR and decline at most locations in
M SAP .18 Low- and high- skilled residents in M SAR face higher land prices, and low-skilled
residents in M SAP face lower land prices (this is also true, on average, for high-skilled residents
who remain in M SAP after the shock). Hence, low-skilled workers displaced from M SAR to
M SAP end up paying lower land prices, but those who remain in M SAR pay higher land
prices after the negative shock in CBDP .
5. Total land rents in M SAP , RP , will tend to decline, and total land rents in M SAR , RR , will
tend to increase.
3.5
Negative shock to CBDP and positive shock to type-h jobs at CBDR
In the case of Richmond and Petersburg, not only has Petersburg been declining, but it has been
doing so at the same time Richmond has been booming. We show in Appendix B how the equilibrium
values of {x̂R , x∗ , x̂P , ū` , ūh } change when we consider at the same time both a negative productivity
shock to CBDP and a positive productivity shock to CBDR . Specifically, we assume that wages
for low- and high-skilled workers in CBDP are (1 − θP )wi after the shock (the same as before), and
wages in CBDR increase to (1 + θR )wi after the shock, with θj > 0.
Specifically, the land rent function is lower after the shock at all locations x ∈ [x̂P (θ1 ), x̄] in M SAP . Note that
locations x ∈ [x∗ (θ0 ), x̂P (θ1 )] are inhabited by high-skilled residents who live at the edge of M SAP and commute to
CBDR .
18
12
The following conclusions can be drawn from this exercise. First, the positive shock to CBDR
has exactly the same effect on x̂R , x∗ , and x̂P as the negative shock to CBDP . In other words, the
equilibrium values of x̂R , x∗ , and x̂P are exactly the same when {θP , θR } = {θ̃, 0} and {θP , θR } =
{0, θ̃}. In fact, only changes in the sum of the two shocks θ̄ = θP + θR would affect the equilibrium
values of these variables. Second, the equilibrium utilities ū` and ūh depend on the difference
between the two shocks (θR − θP ). If the magnitude of the two shocks is the same, then equilibrium
utilities would not be affected. And third, for a given θP , the effect of the positive of shock to
CBDR is to push even more high-skilled workers to CBDR and M SAR (x̂R decreases, x∗ increases,
and x̂P decreases).
3.6
Reinforcing effects
The model above captures only a portion part of the economic effects of the negative shock to
CBDP . It should be emphasized that, as all these events unfold, other factors simultaneously take
place reinforcing and amplifying the initial negative effects of the shock. For instance, a decline in
land prices and the shift toward a predominantly low-skill workforce may adversely affect the local
public finances of city P . The resulting lower land prices and lower local production would translate
into lower tax revenues and lead to a lower quality of local public goods, such as schools, parks, and
other public infrastructure. The dimming economic prospects at M SAP due to the deteriorating
balance between taxes and local public goods would eventually induce additional residents to leave
the area. Presumably, the first workers to leave the area are mostly high-skilled workers, further
hampering the local economic situation. These feedback effects are not explicitly considered in the
present model but undoubtedly contribute to explain the overall decline of the region containing
city P .19
The recovery hurdles could by exacerbated by the existence of asymmetric fixed boundaries for
the city, that is, it can grow but not shrink when structural economic conditions change. Fixed
boundaries impose costs of operating the city that are “sticky" when population declines as a result
of the negative shock. Petersburg, for instance, riding a wave of optimism in the early 1970s,
annexed fourteen square miles of land from adjacent localities, envisioning an industrial park. Their
vision, however, never materialized. In fact, the annexed parcel had a relatively small population
for the large area, saddling the city with sizeable maintenance costs that were not balanced by
increased tax revenues. Such decisions made the city particularly vulnerable to shocks to its local
public finances.
3.7
Isolated city: Benchmark for comparison
How would the situation described above differ from that of a city that is also subject to a negative
shock but is completely isolated from other cities? More precisely, suppose that the distance from
19
In a different paper, we focus more precisely on the local public finance implications of urban decline.
13
any location in the M SA surrounding an isolated city, denoted by CBDS , to the next closest
employment center is sufficiently large so that commuting costs are prohibitively large. Moreover, if
residents wanted to move out of M SAS , they would need to incur sizable moving costs. We assume,
hence, that the city is closed and examine the implications of a negative shock using the standard
closed-city urban equilibrium model. The goal is to compare the impact of the shock to CBDS and
M SAS to the respective effects on CBDP and M SAP analyzed earlier. We assume, for a more
direct comparison, that CBDS is located at x = x̄.
In this case, the equilibrium is simply determined by
x̂S = x∗ + NhS ,
ūSh = wh (1 − θ ) − t(x̄ − x∗ ),
ūS` = w` (1 − θ ) − t[x̄ − x∗ − NhS (1 − q )],
(19)
where x∗S is the outer border of M SAS , and location x̂S separates the residential locations of the
two types of workers (i.e., type-h workers reside at locations x ∈ [x∗ , x̂S ] and type-h workers reside
at locations x ∈ [x̂S , x̄]). In order to compare the impact of the shock on M SAS and M SAP , we
assume the initial conditions of the two cities are identical. In other words, we assume that wages
wi and commuting costs t are the same, and NiS = Ni /2, so that when θ = 0, x∗ = x̄/2, x̂S = x̂P ,
and ūSi = ūi .
Clearly, the only impact of the negative shock on CBDS and M SAS is to reduce the equilibrium
utilities of both low- and high-skilled workers. However, total land rents are unaffected and the
decline in production is substantially smaller when CBDS experiences a negative shock of the same
kind and magnitude as CBDP .20 In the case of the isolated city, the amplifying effects discussed
earlier in the case of a negative shock to CBDP (triggered basically by the fact that high-skilled
workers would over time tend to disproportionately abandon the area) do not completely materialize
to the same degree because the negative effects are partially contained within the geographic area.
To some extent, by circumscribing the harmful effects of the negative productivity shock to the
immediate surrounding area of CBDS , it eventually becomes relatively easier for the city to recover
in times of prosperity.
4
Some Stylized Facts
A number of the implications from the models can be matched to data from Petersburg and the
greater Richmond area, in general. We also include, for comparison, a number of indicators of
economic performance for Waynesboro and Lynchburg. These cities are in some ways comparable
to Petersburg, but importantly, are not close to other economically vibrant MSAs, and we, thus,
considered them to be isolated.21
Since Y S = (1 − θ )w` N`S + (1 − θ )wh NhS , then ∂Y S /∂θ = − w` N`S + wh NhS , which essentially captures
the direct effect of the shock on production discussed in the previous case. Moreover, note that, evaluated at the
equilibrium utilities, the land rent functions are unchanged. It is straightforward to verify this by replacing ūi from
(19) into the land-price gradient (3), substituting wiP for (1 − θ )wi .
21
Lynchburg, Virginia is fifty-six miles from Roanoke, Virginia, the nearest larger city. Waynesboro, Virginia is
twenty-eight miles from Charlottesville, Virginia, also the nearest larger city. These distances are greater than those
20
14
Figure 2 shows that, in general, similar shocks hit all the three cities in the 1980s and 1990s.
From 1990 to 2015, for example, unemployment rates in the cities typically moved together, rising
around U.S. recessions and falling with expansions. Of the four cities, Petersburg posted a generally
higher unemployment rate, particularly around recessions, but it tended to converge toward the
unemployment rates of the other cities during expansions. Notably, the gap between Petersburg’s
unemployment rate and that of the other cities was wider following the Great Recession than after
previous recessions. In addition, Figure 3 displays the job numbers in the Virginia cities considered
in the present analysis. In the case of Petersburg, the figure shows a peak of employment in the early
1980s before its long steady decline. For Lynchburg and Waynesboro, the situation is somewhat
different. Despite going through a period of decline (attributed to the shocks described earlier),
employment in both cities not only recovered, but, by 2010, reached the highest levels since 1970.22
Differences are also apparent in the population trends of the cities as seen in Figure 4. Subsequent
to the economic shocks of the 1980s and 1990s, Petersburg’s population followed a downward
trend that persisted throughout the thirty-five-year period shown. Specifically, the population in
Petersburg declined from a peak of 41,000 in 1980 to 32,000 in 2015 (approximately a 22% decline in
population). In contrast, the populations of Lynchburg and Waynesboro rebounded after a period of
population decline. Lynchburg notably recovered from a small population loss observed during the
1980-2000 period, reaching 79,505 in the year 2015 (a 20% population increase compared to 1980).
The population in Waynesboro importantly declined during the 1980s, but also rapidly recovered,
surpassing by 2010 the population levels observed prior to 1980.
With declining population and higher unemployment plaguing the city, Petersburg was further
hampered by a changing demographic composition.23 On average, relative to the other cities
examined, Petersburg residents were less likely to have a high school diploma (Figure 5), and more
likely not to hold a Bachelor’s degree or higher (Figure 6). This is consistent with the model’s
prediction that high-skill residents of the city would relocate outside of Petersburg (toward better
job prospects), while those with lower skills were less likely to move (Duranton and Puga (2014)).
One plausible explanation is that the lower educational attainment in Petersburg may result from
higher-skill residents leaving the city. If higher-skill residents leave, they are more likely to be
working age and their departure would have implications for Petersburg’s residential age distribution.
The proportion of Petersburg residents that were age 65 or older was the second lowest among the
four cities in 1970, but second highest among the cities by 2010 (Figure 7). Note that an older
population is consistent with a lower labor force participation rate in Petersburg compared to other
between Petersburg and Richmond, but both Roanoke and Charlottesville are substantially smaller and perhaps less
diversified than Richmond, lessening the available job opportunities, and consequently the “pull", to residents from
the smaller cities after a negative shock.
22
For Waynesboro, the substantial layoffs in the early 1980s mentioned in Section 2 are not apparent in Figure 3
because that data are decennial, and jobs lost in Waynesboro after 1980 that were offset later in the decade do not
show up at the 1990 level. As a result, the 1990 reading will appear similar to the 1980 level.
23
While unemployment also tends to rise in Lynchburg and Waynesboro when economic conditions are weak,
residents in those cities do not seem to respond to the negative shock by overwhelmingly leaving their respective areas,
as opposed as to the activity seen in Petersburg.
15
Virginia cities and is likely a factor in the widening unemployment rate gap (as seen in Figure 2).
While these figures paint a picture of a city that has wider employment swings around business
cycles, is experiencing an aging population, and has falling educational attainment, the spatial
aspects of these changes remain unaddressed thus far. The following charts provide some evidence
that the flows of people and economic activity in and around Petersburg have been moving toward
its neighbor to the north, Richmond. Additionally, corresponding charts for Waynesboro and
Lynchburg do not have a nearby city and the patterns of flows appear to differ in those jurisdictions.
Figure 8 shows that the locations of high-income residential concentrations lessened in Petersburg
and tended to show up more prominently in areas north of the city (toward Richmond) over the
period examined. Further, concentrations of higher home values also migrated in a similar fashion,
mostly disappearing in Petersburg and arising north of the city –toward Richmond (Figures 9).24
This movement is not surprising in light of the income migrations as local income levels are typically
correlated with local housing prices.
The outmigration of higher-skilled workers left Petersburg with both a lower population and a
lower-skilled workforce subsequent to the closing of tobacco firms in the 1990s. For decades, the city
had relied on these firms for a major portion of the job base. Because these firms required workers
with lower-tier skill sets, their departure may have disadvantaged Petersburg’s ability to attract new
firms, many of which likely required higher skill sets. Compounding the problems facing the city,
potential workers with higher skills appear to have migrated toward Richmond, possibly leading
relocating firms to always choose Richmond for that city’s labor pool.
The data for both Lynchburg and Waynesboro suggest a different reaction to negative employment
shocks, with less clear-cut changes in geographic patterns. For instance, Figures 10 and 12 show a
less pronounced movement of high-skill departures, and Figures 11 and 13 reveal a muted change in
housing values. Specifically, within the city borders, both household income and home values did
not significantly change from 1980 to 2015.
Over time, the tax base of Petersburg faced downward pressures, even as the costs of running
the city could not easily adjust down commensurately partly because of the large land area the city
acquired in the 1971 annexation. Ultimately, facing mounting fiscal stress, Petersburg’s infrastructure
–including schools– declined, creating added headwinds to efforts aimed at reviving the city.
5
Conclusions
Following two centuries of general economic prosperity, Petersburg has experienced a prolonged
period of decline. A fixed-boundary city combined with a shrinking population may have left the
24
While generally the case, note that housing values also increased east of Petersburg, as seen in Figure 9. This
increase likely occurs because that area contains I-95, which connects Petersburg with Richmond, and households
located there are able to easily commute to Richmond, as are households located north of Petersburg. Though not
explicitly modelled here, the “effective" distance to job centers, basically determined by the time it takes to reach the
destination, is likely the relevant distance for households when they determine their residential locations.
16
city vulnerable to negative economic shocks as city officials faced “fixed" municipal costs in a context
of declining tax revenues. When large layoffs occurred beginning in the 1980s, the city appears to
have lost residents, especially higher-skilled residents, to the Richmond suburbs north of the city.
Additionally, a new regional shopping center in neighboring Colonial Heights drained the city of
retail tax revenues. These development led to a prolonged period of decline in the city.
But other Virginia cities also experienced substantial layoffs around the same time as Petersburg,
yet they did not decline to the same degree. The question is why? We model two scenarios. The
first incorporates two cities, one relatively economically vibrant and the other less so. We show that
a negative productivity shock to the less vibrant city will lead to an outflow of high-skill workers
to the more vibrant neighboring city along with higher-value homes. As tax revenues fall, the city
experiences fiscal decline, which amplifies and reinforces its decline.
We also model an isolated city in which a negative shock does not result in as large of an outflow
of high-skilled workers. In this setting, the city experiences a loss in aggregate utility for residents
but is in a better position to weather the shock and eventually return to a path of economic growth.
Evidence from several Virginia cities is consistent with the implications of the models. In
Petersburg, the period after the shocks saw high-income residents and higher home price areas
decrease in the city and increase in areas closer to Richmond. As higher-skilled workers left,
the population of Petersburg got older and less well-educated. In contrast, isolated cities that
experienced somewhat similar shocks showed less pronounced effects. We conclude that Petersburg
was a victim of being “too close" to Richmond, and as residents and the tax base left the city, an
inability to scale down city municipal costs led to the severe fiscal difficulties seen today.
References
Bleakley, Hoyt and Jeffrey Lin (2012). “Portage and Path Dependence”. The Quarterly Journal of
Economics, 127, pp. 587–644.
Buettner, Michael (October 13, 2016). “Special prosecutor to probe Petersburg’s finances”. The
Progress Index.
Burnett, Anne, Raymond E. Owens, and Santiago M. Pinto (2017). “The Rise and Decline of
Petersburg, Va”. Econ Focus, (4Q), pp. 28–32.
Duranton, Gilles and Diego Puga (2000). “Diversity and specialisation in cities: why, where and
when does it matter?” Urban studies, 37(3), pp. 533–555.
— (2014). “Chapter 5 - The Growth of Cities”. In: Handbook of Economic Growth. Ed. by Philippe
Aghion and Steven N. Durlauf. Vol. 2. Handbook of Economic Growth. Elsevier, pp. 781 –853.
Glaeser, Edward L. et al. (1992). “Growth in cities”. Journal of political economy, 100(6), pp. 1126–
1152.
Goodman, Sandra (March 2, 1982). “Recession Victim”. The Washington Post.
17
Henderson, J. Vernon (2005). “Urbanization and growth”. In: Handbook of economic growth. Vol. 1.
Elsevier, pp. 1543–1591.
Moreo, Bob et al. (2015). “Municipal Boundaries in Tennessee: Annexation and Growth Planning Policies after Public Chapter 707”. Report of the Tennessee Advisory Commission on
Intergovernmental Relations.
Nardi, Gail (March 15, 1987). “Pride Rekindled as Waynesboro Stokes Economy”. The Richmond
Times-Dispatch.
Schneider, Greg S. (September 6, 2016). “City on the brink: Petersburg can’t pay its bills, and time
is running out”. The Washington Post.
18
A
Numerical example: Negative productivity shock to CBDP
Parameter values: w` = 2, wh = 3, q = 5/6, t = 0.6, Nh = 1, N` = 1/q
B
Negative productivity shock to CBDP and positive productivity
shock to CBDR
Suppose that while CBDP is hit by a negative productivity shock θP > 0, CBDR experiences at
the same time a positive productivity shock θR > 0. Specifically, suppose that wages in CBDP are,
as before, (1 − θP )wi , and wages in CBDR are (1 + θR )wi , for i = `, h. The equilibrium values of
{x̂R , x∗ , x̂P , ū` , ūh } are now given by
x̄ − Nh θ̄ (qwh − w` )
−
,
2
2(1 − q )t
(2 + θ R − θ P )
tx̄
ūh =
wh − ,
2
2
where θ̄ = θ0 + θ1 .
x̂R =
θ̄wh x̄
x̄ + Nh θ̄ (qwh − w` )
+ , x̂P =
−
,
2t
2
2
2(1 − q )t
(2 + θ R − θ P )
t [x̄ − Nh (1 − q )]
ū` =
w` −
,
2
2
x∗ =
19
(20)
(21)
C
Isolated city: Benchmark for comparison
Parameter values: w` = 2, wh = 3, q = 5/6, t = 0.6, Nh = 1/2, N` = (1/q )/2.
20
D
Petersburg’s recent history: Critical events
1953
Named
"All American City"
1989
1991
Southpark Mall
opened
Walnut Mall
completely vacated
1971
1986
2009
Annexation
Failed annexation
Part of Dinwiddie Co, Prince George
Co; tripled geographic size of city,
added only 7,300 citizens
Failed attempt to annex large portion
of Prince George County
Beginning of
financial decline
1982
1985
2014
Brown & Williamson
eliminated 1/3 of the 2,800
employees
Brown & Williamson
closed operations in
Petersburg
City's bond rating
downgraded
21
From A+ to BBB
(S&P's lowest rating)
Evolution of various economic variables for different cities
Unemployment Rate
2.5
5
Percent
7.5 10 12.5 15
1970-2015
0
E
1970
1975
1980
1985
1990
1995
Year
Lynchburg
Waynesboro
2000
2005
2010
Petersburg
Richmond, Henrico, and Chesterfield
Sources: National Historical Geographic Information System (1970-1980) and Bureau of Labor Statistics (1990-2015).
Data is available for 1970, 1980, and annually starting in 1990.
Figure 2
22
2015
Employed Population (Thousands)
200
250
300
350
400
Richmond, Henrico, and Chesterfield
Petersburg, Lynchburg, and Waynesboro
0
10
20
30
40
1970-2010
1970
1980
1990
Year
Lynchburg
Waynesboro
2000
2010
Petersburg
Richmond, Henrico, and Chesterfield
Source: U.S. Census Bureau
Figure 3
Population (Thousands)
500
600
700
800
900
Richmond, Henrico, and Chesterfield
Petersburg, Lynchburg, and Waynesboro
0
20
40
60
80
100
1970-2015
1970
1975
1980
1985
Lynchburg
Waynesboro
1990 1995
Year
2000
2005
2010
2015
Petersburg
Richmond, Henrico, and Chesterfield
Source: U.S. Census Bureau
Figure 4
23
Percent of Population 25+ with Less Than High School Diploma
0
10
20
Percent
30 40
50
60
1970-2010
1970
1975
1980
1985
1990
Year
Lynchburg
Waynesboro
1995
2000
2005
2010
Petersburg
Richmond, Henrico, and Chesterfield
Source: National Historical Geographic Information System
Figure 5
Percent of Population 25+ with Bachelor's Degree or Higher
0
5
Percent
10 15 20 25 30 35
1970-2010
1970
1975
1980
1985
1990
Year
Lynchburg
Waynesboro
1995
2000
2005
Petersburg
Richmond, Henrico, and Chesterfield
Source: National Historical Geographic Information System
Figure 6
24
2010
Percent of Population Age 65 or Older
0
3
6
Percent
9
12
15
18
1970-2010
1970
1975
1980
1985
1990
Year
Lynchburg
Waynesboro
1995
2000
2005
Petersburg
Richmond, Henrico, and Chesterfield
Source: National Historical Geographic Information System
Figure 7
25
2010
E.1
E.1.1
Household income and housing prices
Petersburg
Census Tracts by Median Household Income in 1980, as Percent of Median Income in Petersburg
Petersburg and Surrounding Counties
Sources: U.S. Census Bureau and Longitudinal Tract Database
Legend
Petersburg City Limits
Pct. of Petersburg's Median HH Inc.
<100%
100-150%
>150%
Census Tracts by Median Household Income in 2015, as Percent of Median Income in Petersburg
Petersburg and Surrounding Counties
Source: U.S. Census Bureau
Legend
Petersburg City Limits
Pct. of Petersburg's Median HH Inc.
<100%
100-150%
>150%
Figure 8: Petersburg: Household income
26
Median Home Values in 1980, as Percent of Median Home Value in Petersburg
Petersburg and Surrounding Counties
Sources: U.S. Census Bureau and National Longitudinal Tract Database
Legend
Petersburg City Limits
Med. Val. as % of Med. Val. in Petersburg
<100%
100-150%
>150%
Median Home Values in 2015, as Percent of Median Home Value in Petersburg
Petersburg and Surrounding Counties
Sources: U.S. Census Bureau and National Longitudinal Tract Database
Legend
Interstate Highways
Petersburg City Limits
Med. Val. as % of Med. Val. in Petersburg
<100%
100-150%
>150%
Figure 9: Petersburg: Home values
27
E.1.2
Lynchburg
Median Household Income in 1980, As Percent of Median Household Income in Lynchburg
Lynchburg and Surrounding Counties
Sources: U.s. Census Bureau and National Longitudinal Tract Database
Legend
Lynchburg City Limits
Med. HHI as % of Med. HHI in Lynchburg
<100%
100-150%
>150%
Median Household Income in 2015, As Percent of Median Household Income in Lynchburg
Lynchburg and Surrounding Counties
Sources: U.s. Census Bureau and National Longitudinal Tract Database
Legend
Lynchburg City Limits
Med. HHII as % of Med. HHI in Lynchburg
<100%
100-150%
>150%
Figure 10: Lynchburg: Household income
28
Median Home Values in 1980, As Percent of Median Home Value in Lynchburg
Lynchburg and Surrounding Counties
Sources: U.s. Census Bureau and National Longitudinal Tract Database
Legend
Lynchburg City Limits
Lynchburg City Limits
Med. Val. as % Of Med. Val. in Lynchburg
<100%
100-150%
>150%
Median Home Values in 2015, As Percent of Median Home Values in Lynchburg
Lynchburg and Surrounding Counties
Sources: U.s. Census Bureau and National Longitudinal Tract Database
Legend
Lynchburg City Limits
Med. Val. as % of Med. Val. in Lynchburg
<100%
100-150%
>150%
Figure 11: Lynchburg: Home values
29
E.1.3
Waynesboro
Median Household Income in 1980, as Percent of Median Household Income in Waynesboro
Waynesboro and Surrounding Counties
Sources: U.S. Census Bureau and National Longitudinal Tract Database
Legend
Waynesboro City Limits
Med. HHI as % of Med. HHI in Waynesboro
<100%
100-150%
>150%
Median Household Income in 2015, as Percent of Median Household Income in Waynesboro
Waynesboro and Surrounding Counties
Sources: U.S. Census Bureau and National Longitudinal Tract Database
Legend
Waynesboro City Limits
Med. HHI as % of Med. HHI in Waynesboro
<100%
100-150%
>150%
Figure 12: Waynesboro: Household income
30
Median Home Values in 1980, as Percent of Median Home Value in Waynesboro
Waynesboro and Surrounding Counties
Sources: U.S. Census Bureau and National Longitudinal Tract Database
Legend
Waynesboro City Limits
Med. Val. as % of Med. Val. in Waynesboro
<100%
100-150%
>150%
Median Home Values in 2015, as Percent of Median Home Value in Waynesboro
Waynesboro and Surrounding Counties
Sources: U.S. Census Bureau and National Longitudinal Tract Database
Legend
Waynesboro City Limits
Med. Val. as % of Med. Val. in Waynesboro
<100%
100-150%
>150%
Figure 13: Waynesboro: Home values
31
@FedFRASER