Regional Science and Urban Economics 96 (2022) 103802

Contents lists available at ScienceDirect

Regional Science and Urban Economics

journal homepage: www.elsevier.com/locate/regec

Urban planning policies and the cost of living in large cities

William Larson a , Anthony Yezer b , Weihua Zhao c, ∗
a
Federal Housing Finance Agency (FHFA), USA
b Department of Economics, George Washington University, USA
c
Department of Economics, University of Louisville, USA

A R T I C L E I N F O A B S T R A C T

JEL classification: Using a numerical simulation model to provide a sterile laboratory for studying the long-run effects of both land
R30 use and transportation policies, this paper offers two main findings. First, we argue that land use regulations
R31 have relatively small effects on the cost of labor in large cities due to location substitution by housing producers
R38
and households. Second, we show the compensating differential paid to workers in growing cities is invariant
with respect to land use regulation. In the long run, the vast majority of the costs of land use regulations are due
Keywords: to changes to the cost of commuting rather than housing.
Zoning regulation
Housing cost
Urban transportation
Standard urban model

1. Introduction siderable work remains to be done in pinning down the likely size of
the potential gains.”
Substantial attention has been given to the effect of land use and In this paper, we make three main contributions. First, we argue
transportation policies on the supply of labor in large cities. These that much of the prior literature mis-attributes the high cost of labor
policies have both costs and benefits, affecting housing affordability, to land use regulation. Using the canonical monocentric city model, we
access to employment, and urban amenities. Recent research, including show that the simple geometry of cities combined with the ability of
Herkenhoff et al. (2018) and Hsieh and Moretti (2019), among others, households to commute to different locations makes the long-run effects
has argued that a regulation-induced increase in housing costs in highly of all but the most stringent and broadly applied regulations relatively
productive cities has affected the supply of labor in these locations. benign. Our results suggest the long-run effects of various regulations
This, in turn, alters the national allocation of labor and capital, with on the cost of labor are substantially smaller than Hsieh and Moretti
consequent negative effects on aggregate output and economic growth. (2019), ranging from between 0.4% and 1.9%, or at most, $1142 per
Hsieh and Moretti (2019) estimate that land use regulations reduced the worker. The reason for this lower estimate is location substitution in the
growth rate of aggregate output by 36% between 1964 and 2009, reduc- vein of Monte et al. (2018). Regulation in one part of the city simply
ing GDP by 3.7% and average annual earnings per worker by $3685. causes building in another part of the city whose residents commute
Based on these papers and the importance of the subject area, Glaeser further.
and Gyourko (2018) issued a call to arms in the Journal of Economic Second, we show that the change in the compensating differential
Perspectives (page 25): “We view any gain that involves adding several required to attract more workers is invariant with respect to regula-
percent to GDP as quite sizeable and worth pursuing. But clearly, con- tion, including FAR restrictions or limits to land for residential use.1 In
order to increase the number of workers in a city by 10%, the long-run

∗ Corresponding author.
E-mail address: weihua.zhao@louisville.edu (W. Zhao).
1
Many common land use regulations can be expressed as limits on the ratio of interior space to land, or the floor area ratio (FAR), including both height limits
and minimum lot size zoning. Of course, FAR need have no effect on a city if it permits greater density than would arise under laissez-faire operation of the land
market. Following Brueckner and Singh (2020), the extent of FAR regulation is related to the difference between the permitted FAR and the FAR under laissez-faire.
The spatial distribution of FAR regulation within the city is shown to be very consequential and commonly used measures of house price change are found to be
problematic due to the locational effect of regulation.

https://doi.org/10.1016/j.regsciurbeco.2022.103802
Received 7 May 2021; Received in revised form 26 April 2022; Accepted 12 May 2022
Available online 20 May 2022
0166-0462/© 2022 Elsevier B.V. All rights reserved.
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

wage need only increase by 1.0%, giving a population-wage elasticity of exogenously impose land use policies in the city. Alternative models
about 10. While households commute further on more congested roads are available that can be tightly fit to an actual, real-world city, but
than before, adaptation by households and housing producers achieves these do not facilitate simulation of long-run equilibria in a city under
a new equilibrium with little need for an increase in wages, regardless either laissez-faire or specific patterns of regulation. Actual cities are
of the stringency or location of regulation. determined as much by market forces as historical accident and path-
The third contribution is to provide isolated estimates of the effects dependence, and we wish to avoid these issues and understand the long-
of transportation restrictions. Because there is currently no adequate run consequences of regulation, including endogenous changes in traffic
measure of the performance of urban transportation networks, empir- congestion.3
ical estimation of the effects of independent effects of transportation We would also like to emphasize that the model results represent the
planning on the cost of housing and labor in cities has not been possi- largest possible effects on earnings and labor supply, despite already
ble. The numerical simulations conducted here allow these transporta- being substantially smaller than Hsieh and Moretti (2019). Regula-
tion effects to be isolated from other planning policies and their effects tions that enhance city amenity would produce smaller earnings and
to be estimated. Similar to land use regulations, transportation restric- larger house price effects through application of the Rosen (1974)-
tions increase the wage necessary to attract labor. But unlike land use Roback (1982) model, and our regulations are implemented assuming
restrictions, the efficiency of the transportation network affects the rate no change in amenity.4 Such amenity effects may be particularly impor-
of change in the compensating differential necessary to attract addi- tant for housing and land use regulation.5 Next, the demand for housing
tional workers to the city. is modeled as a demand for interior space which is the direct object of
Our approach is to impose land use and transportation regulations in FAR regulation. However, developers can augment the utility of inte-
a controlled theoretical laboratory in which the long-run effects of spe- rior space by providing design enhancements, effectively mitigating the
cific land use and transportation planning policies on housing and com- effect of regulation on the ability to supply housing services. We explic-
muting costs, and hence on labor supply to cities, can be isolated. Cru- itly ignore the possibility of producing quality enhancements to inte-
cially, we are able to impose exogenous regulations of different types rior space because we wish to provide an upper bound estimate of the
on different locations in the city. We find the effect of regulations on effects of FAR regulation. Ignoring both the potential utility of lower
the cost of labor has two main channels. First, there is the housing density and adjustments to space quality both tend to overstate the
cost channel, which has been exhaustively studied in seminal papers by effects of FAR regulation on the supply of labor to the city.6 Finally, our
Segal and Srinivasan (1985), Malpezzi (1996), Mayer and Somerville model is necessarily restrictive in the means of adjustment to a long-run
(2000), Green et al. (2005), Glaeser et al. (2005), Quigley and Raphael urban equilibrium. We assume an exogenous decentralized employment
(2005), Saiz (2010), and Turner et al. (2014). These papers estimate gradient and only one mode of transportation, and neither changes in
models of the relation between indexes of land use planning regulations response to regulation in our model. Relaxing these assumptions would
and the price of housing, construction of new units, and/or the price cause the effects of regulation to fall further as adjustment occurs across
elasticity of housing supply. The empirical results all indicate a signifi- other dimensions.
cant negative (positive) relation between the strength of regulation and There is a fundamental interaction between land use regulation and
housing supply (housing price). The second channel is through chang- commuting, making it essential to consider both housing and trans-
ing commuting costs, featured prominently in Timothy and Wheaton portation costs in analysis of labor costs in cities. Regulation reducing
(2001), Monte et al. (2018), and the related “quality-of-life” literature central densities can actually lower average house prices as low cost
of Albouy and Lue (2015) and Albouy et al. (2016) and others. We are units at the fringe of the city expand in place of high cost inner city
able to model both of these channels simultaneously, while also allow- units. But when commuting costs are included in the cost of living cal-
ing consumption of housing to change in response to changes to relative culation, it is clear such a planning intervention raises the cost of labor.
prices. As will become apparent, the transportation channel is essential In the long run, the inefficient spatial distribution of households sub-
to consider alongside the housing costs to understand and quantify the stantially raises the effect of regulation, particularly where commuting
effects of regulation. is expensive due to congestion. We find that in response to an exogenous
Our model is straightforward and well-studied, and we view this as
a strength. We augment the theoretical monocentric city model with
spatially dispersed work locations, endogenous traffic congestion, and
restrictions imposed by residential building regulations, land use codes, 3
In addition, the politics and path-dependence of zoning has been the object
and transportation planning policies. All housing units—both renter of substantial research. In Chicago (our baseline calibration target) the politics
and owner-occupied—are considered and prices are based on rent per of zoning are discussed at length in Schwieterman and Caspall (2006) and used
square foot of interior space rather than unit price, thus allowing us to for structural empirical identification by Khan (2020). The classical statement
consider the intensive margin of housing as advocated by Baum-Snow on the endogeneity of land use restrictions is discussed in Shertzer et al. (2017)
and Han (2019). We relax the assumption of Cobb-Douglass functions who trace the arguments from Hoyt’s (1933) observation that “whenever there
as recommended by Glaeser and Gyourko (2018), instead using CES for is any possibility of a higher use for any block or parcel of land than the one
for which it is zoned, it is not very difficult to have it zoned for the higher use,
both utility and production. Solutions to this model give the long-run
as the five thousand amendments to the zoning law testify.” Wallace (1988)
effects of these various development restrictions on the price of hous-
and Munneke (2005) provide updated empirical evidence that Hoyt’s view is
ing and the wage required to attract labor from the representative firm’s still correct. Both Ihlanfeldt et al. (2007) and Brueckner and Singh (2020) dis-
perspective. The regulations in the model are assumed to take the form cuss the endogeneity problem in detail, and Molloy (2020) provides a useful
of limitations on density, on the fraction of land available for housing, summary of the challenges of empirically estimating the effects of zoning on
and/or on transportation systems of a large city. Numerical solutions housing costs, specifically preferences.
4
are provided due to the complexity of the model, as is standard in this See Albouy (2016) for an empirical demonstration of this concept.
5
literature (see Muth, 1975, Bertaud and Brueckner, 2005, and Borck, One limitation of the approach taken here is that it assumes that there are
2016, for several examples).2 no positive amenities associated with planning. In the case of transportation
The primary rationales for our choice of model are its fidelity to infrastructure, this may not be serious but height and density regulations are
often justified based on aesthetic grounds. This is also a problem in the empir-
the treatment of space in the urban area and ability to selectively and
ical literature which assumes that higher house prices are not due to amenity
effects. To the extent that land use planning has such positive effects, the labor
supply effects of these building restrictions can be overstated if not properly
2
Alternative models considered include the RELU-TRAN model of Anas and accounted for.
6
Liu (2007) and Anas (2015) and the “Berlin” model of Ahlfeldt et al. (2015). We thank an anonymous referee for this point.

2
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

city-wide FAR regulation limiting housing density to half of that which In addition, many other conditions that influence the spatial form of
would be built under laissez-faire in a large, congested U.S. city, labor the city, including technology, are changing. Local governments influ-
costs needed to maintain city size rise by 1.9%, the highest land-use ence the spatial pattern of development through a variety of decisions
regulation-induced increase in labor costs among our policy scenarios. other than zoning. There are also historical development patterns that
This particular scenario is illustrative of the need to measure both com- have substantial effects on cities, some positive, e.g. cultural attrac-
muting and housing costs together. Regulation pushes building to the tions and others negative, e.g. superfund sites. The current spatial pat-
suburbs where the land available for development increases exponen- tern of development in large cities is the result of complex histori-
tially, mitigating housing supply reductions and actually lowering hous- cal forces. Sorting out the partial effect of recent planning policies on
ing expenditures, but dramatically increasing the cost of transportation. these observed spatial patterns is a major challenge for any empirical
The next section of this paper discusses the challenges of measuring approach or for any model calibrated to the actual parameters observed
the relation between land use and transportation regulations and city in the current city.
labor supply. Then the form of the simulation model is presented. Next A second challenge is the difficulty of characterizing planning
model calibration is discussed. The fifth section presents and discusses a restrictions. Even if these are regarded as exogenous decisions, plan-
variety of simulation results in which the laissez-faire city is contrasted ning maps of cities are complex documents. Land use and transportation
with various regulated cities. Further implications of these results for planning are usually integrated as building height limits vary directly
city size and growth are found in the sixth section, with conclusions with the width of streets in order to preserve light and air circulation.
and implications following. Currently there are no detailed or even city level indexes of the con-
straints imposed by transportation planners and those for land use plan-
2. Modeling challenges and choices ning policies tend to be city or neighborhood-level averages. Thus it is
currently very difficult to isolate the individual effects of land use versus
Before moving onto the theoretical model used to perform our coun- transportation planning policies because of measurement problems and
terfactual experiments, it is useful to take a step back to understand the interaction between measures. It is also difficult to characterize the
why this is our preferred approach. There are a litany of challenges in precise location of city-level planning measures or development restric-
estimating causal effects empirically within a city or in a regional sys- tions. Relations such as those shown in Fig. 1 are common in the liter-
tem. While the prior literature has made great strides in understand- ature. This figure shows that commute times vary directly with topo-
ing causal drivers of housing and labor costs within these systems, graphic and regulatory restrictions to development. These city-level
issues remain. To begin, we discuss three of the empirical challenges measures do not distinguish between restrictions near the CBD ver-
researchers face that can be addressed using a computable general equi- sus the periphery (e.g. center-city height limits versus periphery quasi-
librium model. Then, we move onto our choice of theoretical modeling greenbelts). Thus, while relations between regulatory and topographic
framework, as there are several that are available in the literature. restrictions may provide useful stylized facts, it is difficult to distinguish
The first empirical challenge in attempting to determine the effects empirically between the effects of topography, land use planning, and
of land use and transportation regulations on city labor supply and transportation planning, as we will show.
growth is the timing and diversity issue. There is a substantial lag The third challenge is to be consistent with constraints that the
between policy and city redevelopment during which any city is sub- geometry of cities impose on the possible effects of planning policies
ject to many alternative changes, both endogenous and exogenous. The on the supply of housing. Consider a very simplified city in which hous-
lag is so long that planning policies likely change substantially during ing is produced with a single input, one unit of land so that housing
the interval between initial implementation and full effect on the city.

Fig. 1. Topographic interruptions, land use regulations, and commute times.

Source: Figures (a) and (b) use the 2019 ACS 1-year estimates for CBSA commute times. Figure (a) data on topographic interruptions are from
Saiz (2010). Figure (b) data on land use regulations are from Gyourko et al. (2021). Slopes are estimated using OLS with robust standard errors
in parentheses. There are 159 obs in figure (a) and 169 obs in figure (b).

3
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

rent equals land rent and city households equal city area.7 It is easy to The current form of Chicago is the result of past policies, technologies,
demonstrate that the elasticity of housing supply does not depend on political division, and government regulation. The density of population
the topographical conditions or land use decisions that determine the has been influenced by problematic provision of public services, zoning,
fraction of land used for housing 𝜃 as is commonly assumed in the lit- crime, 13 superfund sites, and other departures from conditions for eco-
erature. In contrast, it can also be shown that a fall in transportation nomic efficiency. For example, Shertzer et al. (2016) demonstrate that
cost increases the rise in the city limit caused by a given rise in house current land use patterns in Chicago have been substantially influenced
price. Therefore, the elasticity of area or housing supply with respect to by the comprehensive zoning ordinance of 1923.
house price varies inversely with transportation cost. Thus even in this The model we develop in this paper is based on research by Alonso
simple model with linear commuting costs and housing produced with (1964), Mills (1967), and Muth (1969). While the canonical rendition
a single land input, a fall in transportation cost should raise the elastic- of the model has existed for some time, recent extensions have demon-
ity of land supply with respect to changes in housing price but changes strated its usefulness in addressing a host of 21st century urban issues.
in the fraction of land planners allocate to housing have no elasticity For instance, the model has been extended to study the effects of differ-
effect. Put another way, this simple illustration based on city geome- ent policies on urban spatial structure, including height limits (Bertaud
try and Muth’s equation indicates that some planning regulations, i.e. and Brueckner, 2005; Borck, 2016), greenbelts (Larson et al., 2012),
transportation decisions, influence the elasticity of housing supply and city size (Larson and Yezer, 2015), telework (Larson and Zhao, 2017),
others, i.e. the fraction of land used for housing, do not change the and autonomous vehicles (Larson and Zhao, 2020). The present model
housing supply elasticity. This does not mean that land use and plan- is designed to incorporate zoning regulation and transportation plan-
ning do not change the number of housing units in a given city but ning into the standard urban model to study the effects of these policies
rather that only transportation costs change the elasticity of housing on urban density, the wage rate, and interactions and elasticities with
supply. This concept, which runs counter to Saiz (2010) and much of respect to city size.
the empirical literature that relates planning regulation to the elastic- The numerical simulation model solved here creates a laissez fair
ity of housing supply, is confirmed and extended by the more detailed version of Chicago that has no regulation or restrictions to develop-
model we present in the next section. ment other than the presence of Lake Michigan, although we use actual
Having noted some of the empirical challenges that can be overcome Chicago labor earnings and population as part of the baseline calibra-
using theoretical approaches, our focus moves to the selection of a the- tion. This setup provides a clean environment in which to estimate the
oretical model to be used. The object of study is the effect of a variety of relation between a laissez-faire city and one subject to selected exoge-
land use and transportation policies implemented in different portions nous planning innovations.8 Given the goal of providing a sterile labo-
of a large city on the compensating variation needed to attract work- ratory in which to model the effects of various planning regulations on a
ers to that city. This requires an open city model in which workers are city the size of Chicago, the modeling choice made here is appropriate.
attracted subject to an iso-utility condition using a flexible utility func-
tion. The goal of the modeling exercise is to generate and then perturb 3. Model structure
with regulatory interventions a laissez-faire model of a large city.
To address these challenges, we use an open city numerical sim- Our theoretical model is based on the classic monocentric city model
ulation version of the standard urban model (SUM) with endogenous but includes spatially dispersed employment. The city occupies a fea-
congestion in commuting on a fixed transportation grid and spatially tureless plane, with no geological or regulatory features that would
dispersed employment. Alternate approaches to modeling the effects of inhibit development. The city is uniform at every radius, allowing char-
planning on cities could have been chosen for this exercise. For exam- acteristics to be expressed in radial terms as a function of the distance
ple, excellent calibrated models by Ahlfeldt et al. (2015) (the “Berlin from the CBD, k. There are three regions in the city, a central business
Wall” model), and Anas (2015) (“RELU-TRAN” CGE) are available and district (CBD) between 0 < k < kCBD which has only employment, a
the latter was even implemented for Chicago. These calibrated models residential district between kCBD ≤ k ≤ 𝜅 that has both employment
have hundreds of parameters estimated based on data from the actual and households, and an agricultural hinterland where k > 𝜅 with no
city being studied. Consequently, they are superior when it comes to (urban) households or employment. Housing producers and households
replicating the current configuration of that city, and these models have receive a reservation profit and level of utility, respectively, at every
been useful to help understand the effects of hypothetical congestion location inside the city.9 The city is open and people are able to migrate
fees and telework expansion, to name two examples (Anas and Hira- costlessly within or across cities.
matsu, 2013; Delventhal et al., 2021).
However, replication of Chicago is not the goal of this research.
8
Instead the goal is to produce a laissez-faire city which is not regu- We view our model as a laboratory experiment as opposed to a field experi-
lated and then impose controlled planning innovations on that city. ment. If the goal of this research were to determine the likely effect of changed
planning on Chicago as it exists today, it would not be the model of choice.
Such changes in planning would be imposed on the current version of the city
which differs substantially from laissez fair because its development has been
7
Following the standard urban model (SUM), Muth’s (1969) equation holds distorted by historical, political, and other forces. It would be a major chal-
so that, dr∕dk = −t∕h, where r is the rental price of housing, t is transportation lenge to take any model calibrated to data from Chicago today and use it to
cost per unit distance and h is housing or equivalently land consumption. This generate the city that would have developed without distortions to laissez-faire.
relation is based on the requirement that households be compensated for higher Indeed, even estimating the housing production function is a challenge because
commuting cost, tdk, with lower housing rent, hdr. Applying this to the deter- actual production is based on height and setback regulations along with his-
mination of the city boundary at k∗ , rising rents justify an increment in city torical preservation so that current unregulated production possibilities are not
radius equal to dk∗ ∕dr = −(h∕t). Thus the relation between a given increase observed in the data. The transportation system in this model accommodates
in rent at the city edge and the change in city limit, k∗ , varies inversely with traffic efficiently given land inputs as opposed to one calibrated based on actual
transportation cost, and directly with housing consumption. This implies that commuting flows and times in the city today.
d2 k∗ ∕drdt > 0. Consider a city with land area available for housing determined 9
It is also possible to model the urban spatial structure of employment as
by A = 𝜃𝜋 k2 , where 𝜃 is the fraction of land, due either to topography or endogenous (e.g. Fujita and Ogawa, 1982; Lucas and Rossi-Hansberg, 2002)
alternative land use, available for housing, and k∗ is the city radius. Taking Our strategy involves assuming the CBD acts as the primary–but not the
logs yields ln A = ln𝜃 + ln 𝜋 + 2 ln k∗ . It is clear that the elasticity of housing only–source of employment in the city. Employment outside the CBD exists
land supply with respect to k∗ is d ln A∕d ln k∗ = 2 and because d2 k∗ ∕drd𝜃 = 0 exogenously in our model. This decentralized employment offers a lower wage
it follows that the elasticity of land supply with respect to r does not depend on to compensate for shorter commutes. Accordingly employment can exist any-
𝜃. where within the city without altering the main findings.

4
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

There are three main improvements relative to the earlier genera- Households maximize utility by choosing how much transportation cost
tion of models. First, we include decentralized employment, similar to they are willing to bear and how much numeraire and housing to con-
Larson et al. (2012), based on gradients found in McMillen (2004). We sume, all of which vary by location.
impose this as exogenous following Thurston and Yezer (1994). Second, The number of households in the city is N, which is equal to the
we include endogenous traffic congestion similar to Muth (1975). This integral of the density of households from the CBD to the edge of the
uses the common “Bureau of Public Roads” congestion function, where city at radius k.
vehicle speed is based on the traffic flowing through a particular loca-
k
tion. Finally, we layer into the model exogenous land use regulations N= 2𝜋𝜃 kD(k)dk (5)
and road provision. These additional features add to the richness of the ∫kCBD
model while maintaining its key features.

3.4. Cost of commuting

3.1. Employment
Commuting occurs in one of two ways. For CBD employment, com-
Both earnings and employment are exogenous in the city. Employ- muting takes place on a congestible road network, incurring costs
ment is divided between the CBD (ECBD ) where it is uniformly dis- related to owning and operating a vehicle as well as time opportunity
tributed, and the residential district (ERD ) where it is subject to a neg- costs. For local employment, commuters only incur a fixed time cost.12
ative exponential gradient, following McMillen (2004). This gradient For CBD commuters, annual commuting costs, T(k), for a household
cuts off at an radius kE , representing the beginning of the portion of the living at radius k include fixed costs of owning an operating an auto-
residential district where there is no employment. Total employment in mobile m0 (e.g. insurance, licensing), variable costs related to distance
the city is thus traveled (e.g. vehicle depreciation), nonlinear gasoline costs with price
kE per gallon of pg , and non-linear time-cost of commuting which is 𝜏 frac-
E = ECBD + ERD = ECBD + E(kCBD )e−gk dk (1) tion of the wage rate. All workers commute to the CBD via automobile.
∫kCBD

k
1
3.2. Housing production T (k) =m0 + m1 k + pg d𝜅
∫0 G(V (M (𝜅)))

Housing H at distance k from the CBD, is produced by profit- k

1
+ 𝜏W d𝜅 (6)
maximizing firms, combining structure S and land inputs L under a con- ∫0 V (M (𝜅))
stant returns to scale technology according to a CES production function
Both fuel and time costs are related to the velocity of the automobile
with an elasticity of substitution between inputs of 1∕(1 − 𝜌).
at various locations in the city, which is in turn related to the ratio
[ ]1∕𝜌
H (k) = A 𝛼1 S(k)𝜌 + 𝛼2 L(k)𝜌 (2) of traffic volume to roads. Following the common “Bureau of Public
Roads” specification of the congestion function, velocity is expressed as
Structure inputs are perfectly elastically supplied, but aggregate land
1
input is fixed at each radius as the fraction of land available for res- V (k ) = (7)
a + bM (k)c
idential development, 𝜃 .10 We assume an elasticity of substitution of
0.75 following Muth (1975) and Altmann and DeSalvo (1981) and treat where M (k) = N ⃖⃖⃗(k)∕R(k), and a, b, and c are congestion parameters and
A, 𝛼 1 , and 𝛼 2 as parameters to be calibrated.11 ⃖⃖⃗(k)∕R(k) is the ratio of traffic passing through annulus k to roads.
N
It is assumed that fraction of land area allocated to roads is uniform
3.3. Households following Muth (1975), Larson and Yezer (2015), and others in this
literature; accordingly we assume R(k) is a constant fraction 𝜓 of land
All households are identical and consume two goods, rental housing area in each annulus. The traffic volume at radius k, N ⃖⃖⃗(k), is calculated
h and a numeraire consumption good y, under a CES utility function. as the sum of the commuting workers living at or beyond radius k.
[ ]1∕𝜂
U = 𝛽1 y𝜂 + 𝛽2 h𝜂 (3) k
⃖⃖⃗(k) =
N 𝜖 2𝜋𝜃 kD(𝜅)d𝜅 (8)
∫k
𝛽 1 and 𝛽 2 are related to consumption shares between the two argu-
ments, and 1∕(1 − 𝜂) represents the constant elasticity of substitution For local commuters, the isoutility condition holds so that CBD and
between housing and the numeraire good. We assume an elasticity of local workers are indifferent between their location of residence and/or
substitution of 0.75 following the range of estimates in Albouy et al. location of employment. Local commuters face a fixed cost of commut-
(2016), then calibrate 𝛽 2 as a relative measure, setting 𝛽 1 = 1. House- ing t. This gives local household wages wL (k) = w − 𝜖 T (k) + t. Under
hold expenditure is divided among the numeraire good, y(k), housing this expression, local wages fall with CBD transportation costs.
purchases, r(k)h(k), and total transportation costs given by the product
of workers per household, 𝜖 , and transportation costs per worker, T(k). 3.5. Model solution
w = y (k ) + r (k )h (k ) + 𝜖 T (k ) (4)
The solution method follows Muth (1975), Arnott and MacKinnon
(1977), Altmann and DeSalvo (1981), and McDonald (2009). The sys-
10
tem of equations described above can be reduced to one with two simul-
This model ignores the role of maintenance, rehabilitation and durability of
taneous differential equations with initial values. After a solution is
structures in housing production.
11 There has been some recent debate on the size of this parameter, with Epple

et al. (2010) finding the elasticity for single-family housing near unity. But
they also find the elasticity for commercial real estate to be substantially less
12
than unity. Our housing production function spans both single family and large This fixed cost could represent a mix of pecuniary and time costs for local
multifamily structure types, suggesting the use of a parameter less than one. buses or a cost of biking or walking to work, to name two examples. Other mass
Reinforcing this choice is the empirical finding of Davis et al. (2021) that land transit is not present in the model. Rail-based mass-transit is difficult to rescale
shares fall with distance from the CBD, implying a Cobb-Douglas production with changes to urban form and density, in contrast to roads, and we leave this
function is inappropriate. for further research.

5
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

obtained, the remaining gradients can be found recursively. The two- marginal commuting costs are from the American Automobile Associ-
equation system of nonlinear differential equations includes marginal ation. The time cost of commuting is set to 50% of the wage follow-
commuting costs and the household density at radius k. ing findings of Small et al. (2007).13 The reservation agricultural land
rental price per acre per year is $2000.
⎡ dT (k) ⎤ ⎡m + p 1
+ 𝜏w
1 ⎤ Results of the simulation calibration are shown in the final column
⎢ dk ⎥ = ⎢ 1 g
G(V (M (k))) V (M (k)) ⎥ (9)
⎢ dN (k) ⎥ ⎢ ⎥ of Table 2. The baseline city consists of 3,216,798 households, with an
⎣ dk ⎦ ⎣ 2𝜋𝜃 kD(T (k)) ⎦ assumed CBD radius of 2 miles, and a total developed radius, including
with initial values the CBD, of 36 miles. Median income is set at $75,158 per year in the
[ ] simulation. Commutes are slightly shorter in the simulation, at 29.9 min
[ ⎡ ] 1 1 ⎤ per trip compared to city average of 32.6 min. Average unit size is
⎢m0 + kCBD m1 + pg
T (kCBD ) + 𝜏w ⎥
=⎢ G(vlow ) vlow ⎥ about 1806 sq. ft. compared to 1799 in Chicago, with an average single-
N (kCBD ) ⎢ 0 ⎥ family lot size of 0.25 acres in the both Chicago and 0.22 acres in the
⎣ ⎦
simulation. With the exception of the city radius, the simulation is fairly
After solving this system, it is possible to derive house prices, housing close to the observed city.
demand, land prices and structure/land ratios as a function of com- Fig. 3 illustrates the within-city simulation outputs. Price gradi-
muting costs and housing unit density, following Altmann and DeSalvo ents fall with distance following Muth’s equation relating transporta-
(1981). tion costs to house prices. Traffic congestion causes vehicle velocity to
There are three necessary conditions that must be met to achieve fall approaching the CBD. Accordingly, commuting time is a concave
an open-city equilibrium. First, the land rent at the edge of the city function of distance. Density is a convex function of distance from the
must be equal to the agricultural land rent pL (k) = paL , and second, the CBD. Fig. 4 shows the employment and population density for both
number of workers in the city must be equal to the number of jobs avail- the baseline and FAR restriction scenario (to be discussed later). The
able 𝜖 N = E. If either of these equilibrium conditions is not met, the employment density is below population density at each radius, result-
simulation is re-initialized and solved again until subsequent iterations ing in a (weakly) increasing commuter flow in each annulus nearer to
achieve an equilibrium solution. the CBD. Commute times are visually similar to those of actual con-
Once these two necessary conditions are met, there is a third neces- gested commute times shown in Fig. 2, panel (c).
sary condition required to close the model. The city is “open” as is typi- It is possible to evaluate the city calibration by observing city size
cal in the familiar regional iso-utility framework. Yet, in order to deter- elasticities. When the city population changes, several variables can
mine the effects of imposing regulations on housing cost, it is necessary be evaluated to determine whether the simulation replicates the pat-
to hold population constant in the city. The solution is for firms to pay tern of city growth observed empirically. Using each decennial census
workers a compensating wage differential so the reservation utility is between 1970 and 2000, a simple reduced form (i.e. long-run) model is
achieved and the population remains constant. Households are free to estimated for two city-level variables—(logs of) household income and
relocate across cities, but choose not to because w adjusts endogenously home value:
to achieve the reservation utility.
yit = at + 𝛽 × populationit + 𝜖it (10)

4. Calibration In the case of income, this corresponds to a standard urban wage pre-
mium regression, with estimates of 𝛽 interpreted as an elasticity of
The calibration of the numerical urban simulation model is based on earnings with respect to (log) population. As discussed previously, and
parameters found from the literature and its output may be compared outlined in detail in Glaeser and Mare (2001), this specification has a
to characteristics of Chicago. The Chicago urbanized area is selected host of issues. In particular, aggregation bias, omitted variable bias, and
as calibration target due to its city size and low regulatory barriers. endogenous sorting of household types due to city-level amenity and
Gyourko et al. (2008) report a low value for Chicago in the Wharton productivity differences confound this specification. Furthermore the
Residential Land Use Regulatory Index. In addition, Chicago’s urban- model simulates growth under laissez-faire development. Nonetheless,
ized area population ranked 3rd among all urbanized areas in the US. this model produces estimates which should give a rough indication of
A large city was selected because of the concern that planning is par- empirical elasticities with respect to population.
ticularly consequential in raising living costs in large congested cities. Table 3 presents estimates of Equation (10). The elasticities of
Chicago will not match the simulation model for a variety of reasons. income and home value are 0.09 and 0.13, respectively. In the simu-
First, it is not currently in long-run equilibrium. For instance, Helms lation model, these elasticities are 0.10 and 0.08 (see Table 2). Given
(2003) reports on a steady pattern of gentrification in older areas with that the estimated elasticities are developed using data from a cross-
good transportation access. Furthermore, spatial variation in the qual- section of cities and the model is calibrated specifically to Chicago, this
ity of public services, violent crime, and other aspects of local amenity level of agreement is remarkably close.14
have distorted the development of the city away from laissez-faire.
According to Saiz (2010), only 60% of Chicago’s circular area is
available for development due to the geographical constraints imposed
by Lake Michigan and other features. This gives rise to a simulated 13
This parameter has wide variation in the literature depending on the income
city with the geometry shown in Fig. 2, which shows the actual city of of person, reason for the journey, and the mode of transportation. Small et al.
Chicago alongside a stylized monocentric city. The simulated city has (2007, page 53) state “we conclude that the value of time for personal jour-
a CBD, a residential district, and an agricultural hinterland with an arc neys varies widely by circumstance, usually between 20% and 90% of the gross
angle of 216◦ (60% of 360◦ ) to reflect the fraction of the circular area wage rate and averaging around 50%.” In the numerical simulation literature,
occupied by water. Bertaud and Brueckner (2005) use a value of 60% of the wage rate. We set
ours to 50% and perform a robustness test which shows that the wage income-
Parameter calibration is performed following the literature. The
population elasticity is robust to variation in the time-cost of commuting (see
parameter values are shown in Table 1. The housing and utility parame-
appendix Table A2).
ters are close to those found in Altmann and DeSalvo (1981) and Albouy 14
It should also be noted that these estimates likely suffer from a “spatial
et al. (2016), which gives elasticities of substitution between both struc- composition effect” as planning regulations change the location of the mean
ture and land inputs and housing and numeraire consumption of about and median housing unit in the city. Thus regulation changes both the location-
0.75. Land shares to housing and roads are similar to Muth (1975), specific house price and the distribution of housing across locations. See Contat
as well as the speed parameters in the congestion function. Fixed and and Larson (2021) for a thorough investigation of this issue.

6
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

Fig. 2. Chicago, actual and simulated.

Source: Figure (a) is from Google Maps. Figure (c) is calculated in R using Mapbox and 2019 Census block data, for all census
blocks in the Chicago-Naperville (IL, WI, IN) CBSA, within 30 straight-line miles from the Cloud Gate sculpture in Millennium
Park.

5. Results plane to 0.2 (−40%). This is a common procedure in simulation models.

The effect of further restrictions due to zoning regulations, is generated
Three main classes of counterfactual scenarios are considered in this by lowering 𝜃 below 0.2.
section. Each alters one aspect of the baseline laissez-faire city intro- Second, building restrictions on FAR are imposed consistent with
duced previously and generates general equilibrium effects on an open the prevailing view that planners, or at least the political system within
city. The three types of restrictions considered include a reduction in the which they operate, consider market forces. The FAR at each location
supply of land for housing, a restriction of the floor-area ratio (FAR), associated with the laissez-faire model solution sets the standard for
and reduced land use for commuting roads. These changes are discussed unregulated development. Then, at locations selected within the city,
in turn. that FAR is reduced by a given percentage. Thus the FAR regulation
First, geographic land use restrictions, whether due to topography, is expressed as an area-specific fraction of the laissez-faire structure
zoning, or a combination of the two, are most straightforward. Indeed, density.
the baseline simulation includes an adjustment for the fact that Lake Finally, transportation system restrictions are based on depar-
Michigan cuts off almost half of the land area that would be available tures from the infrastructure that generates the laissez-faire city. Over
if Chicago were located on a featureless plane. The fraction of land selected exogenous segments of the journey to work, restrictions on
available for housing in each annulus, 𝜃 , is adjusted for the presence of transportation infrastructure are imposed by reducing the fraction of
the lake by lowering 𝜃 from the normal level of 0.33 on a featureless land available for roads used for commuting transport. The consequent

7
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

Table 1
Simulation parameters.
Parameter Baseline Value Description Source
City Income and Size
W 75,158 Annual earnings American Community Survey (2019)
N 3,216,798 Households American Community Survey (2019)
𝜖 1.25 Workers per household Assumed
Housing Production
1∕(1 − 𝜌) 0.75 Elasticity of substitution Muth (1975), Altmann and DeSalvo (1981)
𝛼1 1 Numeraire parameter
𝛼2 0.2 Land parameter Muth (1975); Altmann and DeSalvo (1981); Davis et al. (2021)
A 0.09 Technology parameter Calibrated
Household Utility
1∕(1 − 𝜂) 0.75 Elasticity of substitution Albouy et al. (2016)
𝛽1 1 Numeraire parameter Numeraire
𝛽2 0.11 Housing parameter American Community Survey (2019), Calibrated
Land Use
𝜃 0.33 Fraction of land used for housing Muth (1975) uses 31.8%
𝜓 0.25 Fraction of land used for roads Muth (1975) uses 19.7%
kCBD 2 Radius of the CBD Assumed
paL 2000 Reservation agricultural land rent per acre Davis et al. (2021)
Transportation
vlow 12 Minimum commuting speed Calibrated
vhigh 65 Maximum commuting speed Calibrated
c 1.1 Parameter in speed function Calibrated
𝜏 0.5 Commuting time cost fraction of income Small et al. (2007)
pg 2.5 Auto fuel cost per gallon Assumed
m0 2124 Auto cost per year Association (2019)
m1 0.222 Auto depreciation & maint. per mile Association (2019)

Note: This table presents the parameters used in the calibration of the model. Model values are as noted, and are either equal (or
approximate to) values in existing data or literature, or calibrated to achieve model estimates that resemble the real-world Chicago
urbanized area.

Table 2
Calibration.
City Characteristics Chicago Urbanized Area Simulation Source
Occupied Units 3,216,798 3,216,798 American Community Survey (2019)
Lot Size (acres, 1–4 unit mean) 0.25 0.22 Davis et al. (2021)
Land share of value (1–4 unit mean) 0.30 0.19 Davis et al. (2021)
Housing Unit Size (sqft, median) 1799 1806 American Housing Survey (2019)
City Radius (miles) 36.0 35.7 2010 Urbanized area map
Time to work (minutes, mean) 32.6 29.9 American Community Survey (2019)
Income (median) 75,158 75,158 American Community Survey (2019)

City Size Elasticities Decade Panel Simulation

Median Income 0.09 0.10
Median Home Value 0.13 –
Housing Rent – 0.08

Notes: This table presents values from the Chicago urbanized area from the American Community Sur-
vey/Decennial Census, American Housing Survey, and Davis et al. (2021), alongside simulation output from
the calibrated model. The empirical city size elasticities are from Table 3. The major point of inconsistency is in
the land share. This is due to the average depreciation of structures 𝛿 in real-world cities, giving land shares of
PL L∕(S(1 − 𝛿) + PL L) rather than PL L∕(S + PL L), where PL is the land price, L is land quantity, S is the quantity
of structure with a numeraire price.

increase in congestion slows vehicle speeds through those regions. The computation of prices regardless of tenure status and holds unit qual-
effects of transportation regulation are accordingly very non-linear ity constant by measuring price per square foot. The size and nature of
because they are based increases on congestion. Restrictions imposed the misleading implications produced by the spatial composition effect
some distance from the CBD have less effect because, although sub- are evident when changes in the mean, median, and near CBD price are
stantial land is diverted away from transportation, the initial levels of compared. Price changes should certainly be measured holding location
congestion are low and hence the increase in congestion is modest. constant but even then they can mislead regarding the compensating
The individual effects of planning restrictions are discussed below. variation in earnings because of changing commuting cost. Both land
Because there is a concern with the influence of regulations on labor use and FAR regulations alter the location of the mean and median
supply, special attention is given to the effects on the compensating housing unit so that these prices reflect both the change in price at a
variation in earnings, compared to laissez-faire, required to hold pop- given location and the change in location of the mean or median hous-
ulation of the open city constant. For those interested in consequences ing unit.
for the housing market, the change in the price and quantity of housing Because much of the empirical literature has attempted to estimate
consumed near the edge of the CBD and effects on median and mean the effects of geographic restrictions on the elasticity of housing supply,
housing price per square foot are presented. Note the model allows a series of exercises are simulated in which earnings are increased suf-

8
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

Fig. 3. Baseline and far restriction simulations.

Notes: The figures present gradients from four numerical solutions of the model. The baseline model is calibrated with respect
to Chicago. The number ranges indicate models where the floor-area ratio (FAR) is restricted to 1/2 the baseline FAR within
the given range.

ficiently to raise city population by 10%. The elasticity of city size with constant. Land available for transportation at each radius is not changed
respect to the price and quantity of housing consumed near the CBD are in this exercise and the city predictably sprawls. Restricting land for
then noted under laissez-faire. The effect of regulation on these elastici- housing raises density, with the city experiencing approximately equal
ties is then computed as the difference in the elasticity under regulation proportions of size and density increases as land used for housing falls
and laissez-faire. As noted above, these differences are often very small while households per unit land rises. Note that the increase in earnings
because, in the SUM, land use and FAR restrictions create differences in required to compensate workers for living in the city is quite modest,
levels rather than differences in elasticities. only 1.3% even in the case of the reduction to 0.16. While the city area
rises by 14.5% in this case, average commuting time to work only rises
5.1. Effects of restrictions on the fraction of land used for housing 3.3% because of the increase in density. The fall in the average size
of housing units is also less than 1.8% in response to the 3.4% rise in
Topography already limits the availability of land for housing in housing price. Overall these results indicate that changes in 𝜃 , whether
Chicago so that the initial value of 𝜃 in the laissez-faire simulations of due to topography or zoning restrictions, have a relatively small effect
the city is 0.20, calculated as the product of a 33% land share for hous- on the ability of the city to attract workers even while effects on city
ing and the 60% of the radial area as land. The top portion of Table 4 form are significant. This is consistent with the observation that the
shows the departure from laissez-faire when 𝜃 is successively reduced to largest U.S. cities have grown in spite of having substantial topographic
0.18 and 0.16, with earnings rising sufficiently to maintain population barriers imposed by water features.

9
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

Fig. 4. Population and employment densities.

Notes: The figure presents population and employ-
ment density gradients from two numerical solutions
of the model. The baseline model is calibrated with
respect to Chicago. The FAR restriction model has 1/2
of the baseline FAR between 0 and 5 miles from the
CBD. The employment density gradient is exogenous
and identical in both.

Table 3 that empirical testing for the effects of variation in 𝜃 due to either
Estimates of city size elasticities. topography or regulation should be based on differences in the level
of housing prices or earnings rather than on differences in the elasticity
Model: Dependent variable: log (Column Variable)
of earnings or house prices with respect to these variables. As noted
[1] [2]
Median HH Median Home
earlier in Section 2 discussion, land use regulation changes the level of
Income (log) Value (log) earnings required to compensate workers but not the elasticity of that
Population (log) 0.0867∗∗∗ 0.126∗∗∗
compensation with respect to city size.
[0.00434] [0.00851]

Decade Fixed Effects Yes Yes

5.2. Effects of restrictions on FAR

Observations 1324 1324 Only restrictions on FAR that are binding on the housing mar-
R-squared 0.937 0.843
ket have real effects and more binding regulations should have larger
Notes: Robust standard errors in brackets. ∗∗∗ p < 0.01, effects. The precise location where the regulation is imposed should
∗∗ p < 0.05, ∗ p < 0.1. The sample includes a balanced decadal
also matter. One advantage of the simulation is that it is possible to
panel of 331 MSAs (1998 definitions) between 1970 and 2000 standardize the location and degree to which FAR regulation is binding.
from the Decennial Census. Decade fixed effects are included in
In the simulations, the upward boundary of the FAR regulation is scaled
both specifications.
to be 50% of the laissez-faire density.16 In any location where FAR reg-
ulation is not imposed, the housing market is allowed to respond fully
to the housing price that prevails in that location. This allows housing
The bottom portion of Table 4 shows the effect of restrictions on res-
density to rise in areas which are not regulated producing a potentially
idential land for the elasticity of wages and house prices with respect
irregular pattern in both housing density and land rents. Of course, the
to the number of housing units. A wage change sufficient to achieve
compensating variation in earnings is well-defined because earnings are
a 10% increase in households is imposed on the simulation with 𝜃
measured in the CBD and are based on both housing and commuting
set respectively at 0.2, 0.18, and 0.16. The elasticity of earnings with
cost.
respect to households under laissez-faire is 0.10, and the elasticity of
Fig. 3 shows the relation between the laissez-faire city and three
housing space supply is 0.91 because housing space per household falls
alternative patterns of FAR restrictions in an open city where earnings
slightly.15 The next two columns of Table 4 demonstrate that differ-
are adjusted to hold population constant. In each case FAR is cut to half
ences in 𝜃 have little effect on any of these elasticities. This suggests
the laissez-faire level over a restricted part of the city. The three cases

15 16
These results are all standard properties of the neoclassical SUM. See Liu Brueckner and Singh (2020) measure the extent of FAR restriction through
(2018) for both a general proof that, in an open city SUM, average housing con- the effect on land prices and find that an FAR restriction of 50% is an upper
sumption falls with rising CBD wages and an empirical test that demonstrates limit even in cities like New York. They further find that large FAR restrictions
this result holds in the data. are typically only found near the city center.

10
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

Table 4 to the increasing amount of land affected by the different policies. For
Reduced land for housing. example, the compensating variation in earnings is 0.4%, 1.0%, and
1.4% as the restriction moves from the [0, 5] through the [5, 15] inter-
Simulation Model Baseline City Size
val, while the housing price at the CBD rises by 1.5%, 4.0%, and 5.4%,
Baseline Land for Housing
and city area increases by 4.5%, 11.0%, and 13.9%. Average density
−10% −20%
falls uniformly as the quantity of land regulated rises.
[1] [2] [3]
The results illustrate dramatically the points made earlier regard-
Unit Size (average) 1806 −0.8% −1.8%
ing the problematic nature of the relation between FAR regulation and
Unit Size (median) 1778 −1.0% −2.0%
Unit Size (CBD) 1438 −1.3% −2.8% housing price movements due to the spatial composition effect. The
Land Price (per acre, avg.) 21,223 6.2% 14.3% mean and median housing rent both fall when FAR near the CBD is
Land Price (per acre, med.) 31,363 8.2% 18.0% restricted and rise when the FAR restriction is moved toward the sub-
Land Price (per acre, CBD) 143,838 8.5% 19.0% urbs. This illustrates that FAR restrictions can either raise or lower
House Price (per sq. ft., avg.) 8.67 1.5% 3.4%
House Price (per sq. ft., med.) 8.82 1.7% 3.7%
aggregate measures of the price of housing units depending on where
House Price (per sq. ft., CBD) 12.92 2.4% 5.2% the restriction is enforced. The presumption that FAR restrictions have a
Housing Stock (sq. ft., millions) 5806 −0.8% −1.8% positive effect on a singular “house price” appears to rest on the assump-
Housing Expenditure (per month, avg.) 1304 0.7% 1.5% tion that these restrictions are most binding in the suburbs and do not
FAR (CBD) 0.89 4.4% 9.7%
restrict densities significantly near the CBD. Strict imposition of limits
Residential Density (hh per sq. mile) 1339 −6.4% −12.7%
City Area (sq. miles) 2389 6.9% 14.5% on FARs near the city center raises densities in the outer portions of the
Time to Work (avg.) 29.9 1.5% 3.3% city where housing prices are lower. This can result in a lower average
Wage Income 75,158 0.6% 1.3% housing price even if prices rise throughout the city. This spatial com-
Simulation Model Elasticity WRT City Size position effect also alters commuting flows as seen in the 8th panel of
Baseline Land for Housing Fig. 2, affecting commuting costs at each radius which are incorporated
Everywhere −10% −20% in the required compensating variation in earnings.
[1] [2] [3] The results in Table 5 illustrate that, while FAR regulation has a pos-
Unit Size (average) −0.08 −0.09 −0.09 itive effect on wages as the area influenced moves from [0, 5] to [5,15],
Unit Size (median) −0.10 −0.10 −0.10 the effect on average housing prices is nearly zero when the regulation
Unit Size (CBD) −0.17 −0.17 −0.17
is imposed on the [5,15] interval due to the spatial composition effect,
Land Price (per acre, avg.) 0.64 0.67 0.68
Land Price (per acre, med.) 0.82 0.85 0.85 while it is actually negative when FAR restriction begins at the edge of
Land Price (per acre, CBD) 1.10 1.12 1.11 the CBD in the other two alternatives. Similarly the effect of zoning on
House Price (per sq. ft., avg.) 0.16 0.16 0.17 total square footage supplied is positive when FAR is imposed near the
House Price (per sq. ft., med.) 0.17 0.18 0.19 city center and only negative when the regulation is concentrated in the
House Price (per sq. ft., CBD) 0.31 0.32 0.32
Housing Stock (sq. ft., millions) 0.91 0.91 0.90
[5, 15] ring. Put another way, these results indicate that discussion of
Housing Expenditure (per month, avg.) 0.07 0.07 0.08 a general relation between FAR regulation and either the average price
FAR (CBD) 0.57 0.58 0.57 or quantity of housing supplied is a vexed question because the answer
Residential Density (hh per sq. mile) 0.41 0.43 0.45 depends entirely on the location of regulation. The simulation produces
City Area (sq. miles) 0.57 0.55 0.53
a straightforward answer to questions about the labor market effects
Time to Work (avg.) 0.63 0.62 0.61
Wage Income 0.10 0.11 0.11 because it measures the compensating variation in wages, including
both housing and commuting cost, needed to hold population constant.
Notes: This table presents simulation output from the baseline model and
Those effects range from 0.4% to 1.9% for the examples considered
three different counterfactual scenarios. Column 1 represents the baseline
here. These are upper-bound estimates because of the assumption that
model which is calibrated with respect to a laissez-faire rendition of Chicago.
Models 2 and 3 represent model output under counterfactual simulation
the reduction of building density has no amenity effect, and there are
parameters representing restrictions to land available for housing, 𝜃 . no possible compensating quality improvements available.
The bottom portion of Table 5 shows the elasticity of various city
characteristics with respect to changes in city size. As was the case
involve FAR restrictions on the intervals [0,5], [0,10] and [5,15] miles with Table 4 where 𝜃 was changing, these elasticities do not vary sig-
from the edge of the CBD which itself has a radius of 2 miles.17 Note nificantly with the nature of FAR restrictions except in the case of the
in the first panel of the figure that FAR restriction shifts the housing mean house price where elasticities vary substantially because of the
price profile vertically because it has no effect on Muth’s equation. This measurement problems associated with the spatial composition effect.
contrasts with the land price and housing density effects where FAR Specifically the earnings elasticity of size is exactly 0.10 and elasticity
lowers land values and density where is it imposed and raises them of interior space supply is about 0.91 regardless of FAR regulation. As
elsewhere. While there are massive effects on the physical form of the was the case with variation in the fraction of land used for residences,
city (e.g. density falls substantially), increases to city size conditional on the effect of FAR regulation is on the level of housing prices rather than
a given FAR regulation cause land used and commute times to increase the elasticity of price change in response to city growth.18
at similar rates.
Table 5 compares the land use, commuting, and house price effects
of the three alternate FAR restrictions. Some effects are proportional

18
Two different experiments were performed to supplement the exercise in
Table 5. In the first exercise, the effects of regulations reducing laissez-faire
17 A city with an inner-region FAR restriction resembles present-day Wash- FAR by 50% was examined over the intervals [0, 5], [0, 10], and [5, 10] in
ington, DC, which has a height limit and restrictive historical designations order to illustrate the difference in effects on CBD housing price and unit size
throughout the District, which occupies the CBD and several radial miles of versus the effects on average price and unit size. However, in this exercise, the
housing. Outside the district, but still within the metropolitan area, regulations amount of land regulated increases. In separate calculations, the intervals [0,
are relaxed, and correspondingly, maximum density is much higher (see Ross- 5], [5, 7.8], and [7.8, 10] produce a pure locational effect holding the quantity
lyn, VA, Silver Spring, MD, and Rockville, MD). Other cities in this class include of land regulated constant. The same patterns of sign reversal are found in
major European cities such as Dublin, Amsterdam, Paris, and London (to name this exercise but the sizes of the differences are attenuated because land area
several) with large historic center-cities. regulated is held constant.

11
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

Table 5
Density restriction.
Simulation Model Baseline City Size
Baseline Floor-Area Ratio −50%
0-5 miles 0-10 miles 5-15 miles 0-15 miles
[1] [4] [5] [6] [7]
Unit Size (average) 1806 0.5% 1.0% 0.2% 1.1%
Unit Size (median) 1778 0.4% 1.1% 1.4% 1.8%
Unit Size (CBD) 1438 −0.8% −2.1% −2.9% −3.9%
Land Price (per acre, avg.) 21,223 −1.2% −1.0% 1.4% −0.4%
Land Price (per acre, med.) 31,363 −3.5% −8.6% −10.8% −14.1%
Land Price (per acre, CBD) 143,838 −20.9% −16.6% 19.9% −10.6%
House Price (per sq. ft., avg.) 8.67 −1.0% −1.8% −0.1% −1.9%
House Price (per sq. ft., med.) 8.82 −0.8% −1.9% −2.4% −3.2%
House Price (per sq. ft., CBD) 12.92 1.5% 4.0% 5.4% 7.5%
Housing Stock (sq. ft., millions) 5806 0.7% 1.1% 0.2% 1.3%
Housing Expenditure (per month, avg.) 1304 −0.5% −0.8% 0.1% −0.8%
FAR (CBD) 0.89 −50.0% −50.0% 10.1% −50.0%
Residential Density (hh per sq. mile) 1339 −4.3% −9.9% −12.2% −16.6%
City Area (sq. miles) 2389 4.5% 11.0% 13.9% 20.0%
Time to Work (avg.) 29.9 4.3% 9.4% 7.9% 14.7%
Wage Income 75,158 0.4% 1.0% 1.4% 1.9%
Simulation Model Elasticity WRT City Size
Baseline Floor-Area Ratio −50%
0-5 miles 0-10 miles 5-15 miles 0-15 miles
[1] [4] [5] [6] [7]
Unit Size (average) −0.08 −0.09 −0.09 −0.08 −0.08
Unit Size (median) −0.10 −0.10 −0.09 −0.09 −0.09
Unit Size (CBD) −0.17 −0.17 −0.16 −0.16 −0.16
Land Price (per acre, avg.) 0.64 0.70 0.69 0.62 0.68
Land Price (per acre, med.) 0.82 0.83 0.82 0.81 0.80
Land Price (per acre, CBD) 1.10 1.09 1.07 1.04 1.04
House Price (per sq. ft., avg.) 0.16 0.16 0.16 0.15 0.15
House Price (per sq. ft., med.) 0.17 0.17 0.17 0.17 0.16
House Price (per sq. ft., CBD) 0.31 0.31 0.31 0.30 0.30
Housing Stock (sq. ft., millions) 0.91 0.90 0.90 0.92 0.91
Housing Expenditure (per month, avg.) 0.07 0.07 0.07 0.07 0.07
FAR (CBD) 0.57 0.57 0.57 0.54 0.57
Residential Density (hh per sq. mile) 0.41 0.48 0.49 0.44 0.51
City Area (sq. miles) 0.57 0.50 0.49 0.53 0.47
Time to Work (avg.) 0.63 0.60 0.58 0.59 0.55
Wage Income 0.10 0.10 0.10 0.10 0.10
Notes: This table presents simulation output from the baseline model and three different counterfactual
scenarios. Column 1 represents the baseline model which is calibrated with respect to a laissez-faire rendi-
tion of Chicago. Models 4 through 7 represent model output under counterfactual simulation parameters
representing restrictions to density. This is operationalized by taking the baseline density at the radius
range in the column header, dividing it by two, and setting this as the maximum density for the scenario.

5.3. Effects of restricting transportation infrastructure 0.3%. In this case, average house prices rise only 0.8% although the
average unit moves closer to the CBD. This result is due to the increase
Because the simulation model allows the highways to become con- in the transportation cost per mile, which causes the house price gra-
gested, slowing traffic velocity, commuting costs are sensitive to the dient to become much steeper. In the case of transportation restric-
fraction of land allocated to roads, which is 25% in the baseline sim- tions, the effects on labor supply are captured in a straightforward way
ulation. The shock to transportation infrastructure implemented here through the compensating income differential. The housing cost effects
is accomplished by reducing the fraction of land to 16.7%, or a 33% are complicated by the shift in the slope of the price gradient along
decrease. This reduction in land used for roads is subsumed into other with the spatial composition effect on the location of the average and
non-residential uses in the city. Table 6 compares the city under the median housing units.
25% allocation of land to transportation throughout the city with a The bottom part of Table 6 shows that restrictions on transportation
restriction on infrastructure down to 16.7% of land area. These restric- infrastructure have a dramatic effect on the elasticities of earnings and
tions are implemented at different distances from the CBD. The most other city characteristics with respect to city population. In the base-
extreme scenario reduces land for transportation throughout the city line case, the elasticity of earnings with respect to city population is
while other alternatives reduce land to 16.7% on the [0, 1] and [1, 2] 0.10. Transportation restrictions dramatically increase the elasticity of
mile intervals from the edge of the CBD. The most obvious result of earnings with respect to labor supply, by 40% when roads are reduced
the exercise is that the reductions in the first mile from the CBD are everywhere, and 20% when reduced only in the innermost one-mile
most consequential. For example, the earnings effect of infrastructure annulus. There is also a significant change in the elasticity of central
restriction over the entire city is 2.5% and for the first mile only it is population density as transportation restrictions increase the slope of
1.9%. Restricting transportation infrastructure makes the city denser the housing price gradient, resulting in a more compact city. This result
and smaller as housing supply contracts and prices rise. of the simulation model is consistent with both the simple model of the
This creates the false impression of a negative housing supply effect. effects of city geometry presented earlier in this paper and the general
In the case of a uniform fall in land for transportation, total supply falls theoretical demonstration in Broxterman and Liu (2019) that restrict-

12
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

Table 6
Reduced roads.
Simulation Model Baseline City Size
Baseline −33% Roads
Everywhere 0-1 Miles from CBD 1-2 Miles from CBD
[1] [8] [9] [10]
Unit Size (average) 1806 −0.4% −0.1% 0.0%
Unit Size (median) 1778 −0.4% −0.2% 0.0%
Unit Size (CBD) 1438 −2.5% −1.1% −0.6%
Land Price (per acre, avg.) 21,223 2.4% 0.4% −0.4%
Land Price (per acre, med.) 31,363 3.5% 1.5% 0.0%
Land Price (per acre, CBD) 143,838 17.0% 7.3% 3.6%
House Price (per sq. ft., avg.) 8.67 0.8% 0.2% 0.0%
House Price (per sq. ft., med.) 8.82 0.8% 0.3% 0.0%
House Price (per sq. ft., CBD) 12.92 4.7% 2.1% 1.0%
Housing Stock (sq. ft., millions) 5806 −0.3% 0.0% 0.1%
Housing Expenditure (per month, avg.) 1304 0.4% 0.1% 0.0%
FAR (CBD) 0.89 8.7% 3.8% 1.9%
Residential Density (hh per sq. mile) 1339 1.1% 0.0% −0.6%
City Area (sq. miles) 2389 −1.1% 0.0% 0.6%
Time to Work (avg.) 29.9 24.8% 19.7% 2.4%
Wage Income 75,158 2.5% 1.9% 0.3%
Simulation Model Elasticity WRT City Size
Baseline −33% Roads
Everywhere 0-1 Miles from CBD 1-2 Miles from CBD
[1] [8] [9] [10]
Unit Size (average) −0.08 −0.10 −0.09 −0.09
Unit Size (median) −0.10 −0.10 −0.10 −0.10
Unit Size (CBD) −0.17 −0.20 −0.18 −0.17
Land Price (per acre, avg.) 0.64 0.77 0.71 0.69
Land Price (per acre, med.) 0.82 0.83 0.84 0.82
Land Price (per acre, CBD) 1.10 1.28 1.16 1.14
House Price (per sq. ft., avg.) 0.16 0.18 0.17 0.16
House Price (per sq. ft., med.) 0.17 0.18 0.18 0.17
House Price (per sq. ft., CBD) 0.31 0.37 0.33 0.32
Housing Stock (sq. ft., millions) 0.91 0.88 0.90 0.89
Housing Expenditure (per month, avg.) 0.07 0.08 0.08 0.07
FAR (CBD) 0.57 0.66 0.60 0.59
Residential Density (hh per sq. mile) 0.41 0.52 0.47 0.47
City Area (sq. miles) 0.57 0.46 0.51 0.51
Time to Work (avg.) 0.63 0.71 0.70 0.63
Wage Income 0.10 0.14 0.12 0.11
Notes: This table presents simulation output from the baseline model and three different counterfactual sce-
narios. Column 1 represents the baseline model which is calibrated with respect to a laissez-faire rendition of
Chicago. Models 8 through 10 represent model output under counterfactual simulation parameters represent-
ing restrictions to road networks.

ing transportation raises the housing price and lowers the elasticity of Column 2 (model 11) gives a city where the road land area is solved
housing supply in a general form of the standard urban model. to give an equivalent wage income increase as Column 3. Column 3
(model 4) is identical to Table 5, column 2, and gives a city with
restricted FAR of −50% of the baseline model. This shows a FAR restric-
5.4. Comparing land use and transportation effects tion of −50% of lassez-faire between 0 and 5 miles from the CBD has an
equivalent compensating differential with a −7% restriction on roads.19
Given that the land use and road network restrictions we place on Thus, the FAR limit imposes the same labor cost as removing 1 out
the city are somewhat arbitrary, it is difficult to compare the costs of of every 14 road lanes in a city with no height limit. In this simple
each to one another. However, we can perform two exercises to estab- illustration, a dramatic halving of center-city density is equivalent to a
lish relative effects. First, taking the compensating differential for a barely-visible reduction in lanes of travel.
land use restriction (FAR or available land) as given, we can solve for On the other hand, what if an urban planner wished to have a FAR
the land area for roads necessary to generate an equivalent necessary limit but keep labor costs low? Our model shows a FAR reduction can
wage increase. Second, it is difficult to improve on laissez-faire out- be met with an increase in roads to lower the compensating differential.
comes using land use regulations, with some limited exceptions (Lar- Column 4 (model 12) shows the same −50% FAR restriction between 0
son and Yezer, 2015; Borck, 2016). However, it is possible to add to and 5 miles from the CBD can be offset by an increase in roads of 8%,
the city’s land area for roads beyond the baseline values given in our resulting in zero compensating differential. This road-compensated city
model to lower the compensating differential. Knowing this, we can add
roads to a model with a land use restriction to establish the quantity
of transportation network investment required to offset the otherwise 19
Note that the city with fewer roads is identical in almost all respects to
necessary compensating wage differential. Table 7 shows solutions for the baseline city, except with higher incomes, and longer travel times. This is
the baseline model and six scenarios. The fist three scenarios involve the classic Duranton and Turner (2011) result that higher/lower road capacity
FAR restrictions and equivalencies, and the last three involve residen- simply results in a larger/smaller city with similar levels of congestion; holding
tial land area reductions and equivalencies. population constant, wage increases simply subsidize congested travel.

13
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

Table 7
Comparisons of various restrictions.
Simulation Model [1] [11] [4] [12] [13] [2] [14]
Parameters
FAR -50%, 0–5 miles No No Yes Yes No No No
Land for Housing −10% (𝜃) No No No No No Yes Yes
Land for Roads (𝜓) 0.250 0.232 0.250 0.271 0.225 0.250 0.286

Output
Unit Size (average) 1806 0.0% 0.5% 0.6% 0.0% −0.8% −0.8%
Unit Size (median) 1778 0.0% 0.4% 0.4% −0.1% −1.0% −0.8%
Unit Size (CBD) 1438 −0.4% −0.8% −0.4% −0.6% −1.3% −0.7%
Land Price (per acre, avg.) 21,223 −0.4% −1.2% −0.8% −0.3% 6.2% 6.0%
Land Price (per acre, med.) 31,363 0.0% −3.5% −3.5% 1.0% 8.2% 7.1%
Land Price (per acre, CBD) 143,838 2.4% −20.9% −22.2% 3.8% 8.5% 4.5%
House Price (per sq. ft., avg.) 8.67 0.0% −1.0% −1.0% 0.1% 1.5% 1.4%
House Price (per sq. ft., med.) 8.82 0.0% −0.8% −0.8% 0.2% 1.7% 1.5%
House Price (per sq. ft., CBD) 12.92 0.7% 1.5% 0.8% 1.1% 2.4% 1.3%
Housing Stock (sq. ft., millions) 5806 0.1% 0.7% 0.6% 0.1% −0.8% −0.8%
Housing Expenditure (per month, avg.) 1304 0.0% −0.5% −0.5% 0.0% 0.7% 0.6%
FAR (CBD) 0.89 1.3% −50.0% −50.0% 2.0% 4.4% 2.4%
Residential Density (hh per sq. mile) 1339 −0.6% −4.3% −3.8% −0.6% −6.4% −6.4%
City Area (sq. miles) 2389 0.6% 4.5% 4.0% 0.6% 6.9% 6.9%
Time to Work (avg.) 29.9 3.9% 4.3% 0.3% 6.2% 1.5% −4.6%
Wage Income 75,158 0.4% 0.4% 0.0% 0.6% 0.6% 0.0%

Notes: This table presents simulation output from the baseline model and six different counterfactual scenarios, all
at the baseline city size. Column 1 represents the baseline model which is calibrated with respect to a laissez-faire
rendition of Chicago. Column 2 (model 11) gives a city where the road land area is solved to give an equivalent wage
income increase as Column 3. Column 3 (model 4) is identical to Table 5, column 2, and gives a city with restricted
FAR of −50% of the baseline model. This shows such a FAR restriction has an equivalent compensating differential
with a city with −7% roads. Column 4 (model 12) shows the FAR restriction can be offset by an increase in roads of
8%, resulting in a compensating differential of 0. A similar exercise is performed for the land share for housing in the
final three columns.

is quite different than a lassez-faire city with a standard road allocation, i.e. the effect on the urban wage premium needed to expand the regu-
with variation in CBD densities, travel times, and the urban area. One lated city compared to laissez-faire; and 2), an estimate of the extent to
important caveat: in this theoretical exercise, we maintain the assump- which the population of a Chicago-sized city would expand if laissez-
tion that transportation infrastructure is funded externally with respect faire were substituted for each regulatory environment. While these pre-
to the city, so relative cost-effectiveness questions remain. dictions are based on a long-run model and thus have obvious limita-
A similar approach can be used to demonstrate the tradeoff between tions, they are robust to alternative model parameterizations (see tables
open space provision, which restricts land for housing, and transporta- A1 and A.2 for several exercises) and are therefore instructive of signs
tion infrastructure. Column 5 (model 13) gives a city with 22.5% roads and relative magnitudes one might expect when doing empirical work.
(a 10% reduction vs the baseline) which requires a compensating differ- To begin, the simulation results in Tables 4–6 allow us to compute
ential of 0.6% of the baseline earnings. This is the same compensating the effect of regulation on the urban wage premium. The effects of reg-
differential as a 10% reduction in land area for residential develop- ulations on the marginal inverse urban wage premium and the marginal
ment, as shown in column 6 (model 2). This implies that an additional elasticity of household population with respect to earnings are provided
acre of land removed roads has the same effect on required wages as directly in the tables. This is termed a marginal effect because it arises
1.7 acres taken from housing. On the other hand, if an urban planner in a response to a 10% marginal rise in population. Given the nature of
wished to reserve 10% of the land area of the city that would otherwise non-linearity in the model, particularly in transportation, the earnings
be put to residential use and offset this loss in welfare with an increase requirement to produce 10% growth will be an increasing function of
in transportation infrastructure, column 7 (model 14) shows an addi- city size. Under laissez-faire, the elasticity of city size with respect to
tional 14.4% increase in roads is necessary. In this case, households earnings change, again measured based on a 10% marginal increment
face higher housing costs and smaller homes, and are forced further of households above 3.5 million, is 0.1.
from the CBD, but substantially faster commutes eliminate any welfare It is common to refer to the inverse of this elasticity, 10, as the
loss. In any case, it should be noted that the outcome values described urban wage premium or as the percentage increase in population gen-
are averages, and effects of land use regulation and transportation net- erated by a one percent rise in earnings. In the case of our laissez-
work provision vary with the precise location of the change in zoning faire Chicago, the marginal increase in households associated with a
or land use within the city. one percent rise in earnings, $751 in this case, is 321,000 households
or almost 1,000,000 individuals. The effects of land use restrictions on
6. Discussion this marginal elasticity should not be surprising given the previous dis-
cussion. The marginal elasticity of labor supply or the urban wage pre-
The relatively straightforward model presented in this paper offers mium is not changed significantly by either type of planning restriction.
several surprising findings. Additionally, we present some suggestions Again, as expected based on theory, transportation restrictions raise the
for empirical researchers who wish to heed the lessons from this model marginal elasticity substantially. The general reduction in road capac-
as they seek to quantify the effects of regulation on the cost of living in ity to 16.6% of land area raises the marginal elasticity by 40% to 0.14.
large cities. The effect is most pronounced when the restriction is just outside the
Our model offers an alternative way of estimating 1), the direct CBD where the marginal elasticity increases to 0.12 and falls rapidly to
effect on the marginal elasticity of population with respect to earnings, 0.11 when the restriction is implemented between 1 and 2 miles outside

14
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

the CBD. While the general direction of these effects could be expected from residential to agricultural. Relaxation of FAR limits will cause land
based on a simple analytic model, the magnitudes are the result of the prices to tend to rise in those locations, and fall in previously unregu-
nonlinearity in the simulation model. lated locations. Additional roads lower land prices in all locations where
We now turn to the second question: how much larger would our there is currently residential development.
simulated Chicago-sized city be if we begin with a regulated base- What lessons do these estimates offer empirical researchers? Recall
line and then restore laissez-faire? The answer to this question first empirical studies including Segal and Srinivasan (1985), Malpezzi
requires computing the compensating variation in earnings associated (1996), Mayer and Somerville (2000), Green et al. (2005), Glaeser et
with replacing laissez-faire with each type of regulation The effects on al. (2005), Quigley and Raphael (2005), Saiz (2010), and Turner et al.
earnings needed to achieve a city the size of Chicago under regulation (2014) that have estimated models of the relation between indexes of
compared to prices and earnings under laissez-faire are shown in the land use planning regulations and the price of housing, construction
upper half of Tables 4–6. Then these increases in earnings can be mul- of new units, and/or the price elasticity of housing supply using an
tiplied by the marginal elasticity of household population with respect empirical two-step method. Because the urban wage premium is highly
to earnings, i.e. by 10, to get the estimated rise in city population under sensitive to transportation regulation, it would be prudent to include
laissez-faire compared to each type of planning regulation. measures of infrastructure costs, such as those pioneered by Brooks and
Reducing land available for housing by 10 and 20 percent raises Liscow (2019), in empirical models. Second, it is crucial for researchers
earnings (via the compensating differential) by 0.6% and 1.3%, imply- to consider increases in commuter transportation costs in cost-of-living
ing a rise in household population of 6% and 13% if the regulation measures (Timothy and Wheaton, 2001; Albouy and Lue, 2015; Albouy
were removed. These are small increases that could be counteracted by et al., 2016) and when considering the urban wage premium. While
amenity effects of added open space. Lowering FAR in various parts there is a positive relation between regulation and commuting times in
of the city raises earnings from 0.4% to 1.9%, or $240 to $1142 per our models presented here, which align with the simple empirical cor-
worker, compared to the estimate of Hsieh and Moretti (2019) of $3685 relations shown in Fig. 1, this need not be so. There are many types
per worker. These amounts would raise population by 4%–19% and of regulations that could potentially reduce travel times; for example,
certainly in the case of massive FAR regulation there is potential to the imposition of an urban growth boundary (Brueckner, 2007; Larson
lower population by more than any reasonable estimate of compen- and Yezer, 2015). Finally, the spatial composition of homes changes
sating amenity effects could offset. Effects of limited restrictions on with regulation. Accordingly, measures of changes in mean or median
highways prompt compensating earnings increases required to main- housing costs are imperfect measures when attempting to assess the
tain population of 2.5% when all highways are reduced, 0.3% when effects of regulation. Rather, there is a need for price indices holding
the reduction is only implemented 1–2 miles from the CBD, and 1.9% spatial composition constant (Contat and Larson, 2021) or highly dis-
if the restriction is just outside the CBD respectively. The 2.5% earn- aggregated studies (Baum-Snow and Han, 2019) in order to assess the
ings effect implies a reduction in population compared to laissez-faire effects of planning policies on housing costs.
of 25%. This massive effect is likely not offset by any amenity gain from
reducing land for highways. Accordingly it appears that congestion cre- 7. Conclusion
ated by limiting transportation capacity can have a substantial effect on
the ability of large cities to attract labor. Using a theoretical monocentric city model to estimate the effects
Finally, it should be noted that while these effects of regulation on of planning restrictions on the price of housing and labor reveals that
earnings and population may seem large in our main scenarios, they are research questions involving effects of urban planning are complex. The
small compared to the stringency of regulations necessary to require current focus on measuring the effects of regulations on the own price
the compensating differential of Hsieh and Moretti (2019). To achieve elasticity of housing supply and using the two-step approach to gener-
a compensating differential of $3685 per worker, one of the following ating implications for labor supply to cities is problematic for two main
needs to occur: FAR must fall by 92% versus laissez faire, land area for reasons: labor supply depends on differences in cost of living which is
roads must fall from 25% to 15%, or the land used for housing must the sum of housing and commuting cost, and regulations change both
fall from 35% to 13% (see appendix Table A3 for details). That these the price and location of housing, making it difficult to estimate a true
draconian restrictions each cause such a small compensating differential elasticity of housing supply. The numerical simulation model developed
is a remarkable illustration of location substitution in the case of FAR or here handles both of these difficulties.
residential land reductions: land available for development expands at a Given that the object of this research is to first generate a laissez-
rate equal to the radius of the city, squared. Thus, even stark restrictions faire city and then to implement individual planning restrictions on that
can be met with modest increases to the city radius. The costs of this city, the calibration of the numerical simulation model should be based,
expansion in land area and reduced density at each location is mostly an insofar as possible, on relations taken from the general literature gov-
increase in transportation costs. In the case of a reduction in roads, the erning the utility function, housing production function, highway con-
price gradients steepen and density increases, with the compensating gestion, and cost of commuting. Detailed calibration to characteristics
differential roughly maintaining average levels of housing consumption of Chicago today would involve fitting to functions that are constrained
though there is some substitution towards the numeraire good. by historical accident, political decisions, and other idiosyncratic fac-
Who wins under land use regulation? Our model can give some basic tors. For example, the laissez-faire simulation produces a city that is
predictions regarding the relative beneficiaries. In general, it is easier substantially more compact than the actual city presumably reflecting
to consider relaxing a regulation rather than imposing one, because it distortions in the current development of the city.
is difficult to lower FAR or remove land from residential development We develop a model that shows the primary effect of changes in
once it has been built. From the outset, the representative firm will land use and FAR regulations is on levels of house prices and earnings
benefit in all cases by capturing the compensating differential (other and not on the elasticity of these variables with respect to city size as
assumptions are possible but they require alternative model solutions), has been found in the literature. These regulations also influence the
renters would be indifferent by achieving the reservation utility, and required level of earnings but not the elasticity of city size with respect
owners of housing structure would be indifferent in the long-run due to earnings. Large increases in open space have little effect on the sup-
to depreciation. The only agents in the model able to achieve long- ply price of housing or the compensating variation in earnings needed
run benefits or losses are the absentee landowners of both residential to attract households. FAR or density restrictions can raise housing cost
and agricultural land. Adding within-city developable land lowers land and earnings significantly, but only when implemented throughout the
values substantially in all locations, and shifts land use at the periphery city. If implemented in particular locations, in most cases development

15
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

is simply substituted elsewhere with surprisingly low levels of ineffi- Data Curation, Writing - Original Draft, Writing - Review & Editing,
ciency. Visualization, Supervision, Project administration. Larson contributed
Unlike land use and density regulations, transportation restrictions to this research in his personal capacity and not as part of his official
increase the urban wage premium. Restricting land for highways is duties at the Federal Housing Finance Agency. The analysis and con-
particularly consequential when the restriction occurs near the CBD. clusions are those of the authors alone and should not be represented
Given the differences in labor productivity between larger and smaller or interpreted as conveying an official Federal Housing Finance Agency
cities from which workers might be drawn, transportation infrastruc- position, policy, analysis, opinion or endorsement. Any errors or omis-
ture improvements in large cities may induce population reallocation sions are the sole responsibility of the authors.
that produces significant increases to the value of output, even if infras-
tructure does not reduce equilibrium congestion (Duranton and Turner, Declaration of competing interest
2011).
We also demonstrate how planning restrictions distort the spatial The authors whose names are listed immediately below certify that
development of the city. For example, FAR restrictions lower density they have NO affiliations with or involvement in any organization or
and land value in areas where they are binding and raise them in areas entity with any financial interest (such as honoraria; educational grants;
that are not regulated. The simulation allows regulatory effects to be participation in speakers’ bureaus; membership, employment, consul-
framed in terms of departures from laissez-faire development so that tancies, stock ownership, or other equity interest; and expert testi-
the extent to which regulations are binding can be controlled. The old mony or patent-licensing arrangements), or non-financial interest (such
metaphor of a balloon which, when squeezed in one place, is distorted as personal or professional relationships, affiliations, knowledge or
in another dimension certainly appears to fit the effects of planning on beliefs) in the subject matter or materials discussed in this manuscript.
cities.
Regulations that alter the spatial distribution of housing also change Table A.1
Elasticity of Substitution in Housing Production, Robustness
the cost of commuting by altering commuting flows at each location in
the city. The overall effect on the compensating variation needed to Baseline vs FAR Limit Scenario

attract labor to a large, congested city includes both the differential 𝜎 = 0.75 𝜎 = 0.65
cost of housing and the cost of commuting which changes both with Unit Size (CBD) −3.9% −3.8%
household location and the burden that location places on the highway Land Price (per acre, CBD) −10.6% −15.8%
House Price (per sq. ft., CBD) 7.5% 7.3%
system.
FAR (CBD) −50.0% −50.0%
Overall, we believe these theoretical findings will be of use to Wage Income 1.9% 1.8%
researchers seeking to empirically estimate the effects of land use and
Notes: This table presents compensating wage differentials from
transportation policies on the cost of labor in large cities. There are a FAR limits under different elasticities of substitution in housing
number of nuanced issues to consider that are often absent in empir- production (𝜎). Baseline models are solved with the elasticity of
ical studies, in particular, commuting costs, measures of transporta- substitution set to 0.75 and then 0.65, respectively. These baseline
tion infrastructure regulation or costs, and measures of composition- scenarios are then compared to scenarios giving a −50% restriction
to the lassez faire FAR from 0 to 15 miles from the CBD, with differ-
constant house price changes.
ences relative to the respective baseline model shown. CBD values
illustrate the similarity between baseline and scenario differences,
Author statement with the exception of the change in the CBD land price, which is
quite sensitive to the elasticity.
William Larson, Tony Yezer, and Weihua Zhao: Conceptualiza-
tion, Methodology, Software, Validation, Formal analysis, Investigation,

16
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

Table A.2
Time Cost of Commuting, Robustness
Value shown: Wage Income-Population Elasticity
Baseline Land for Housing −20% FAR -50%, 0–15 miles −33% Roads, 0–1 miles
𝜏 = 0.50 0.10 0.11 0.10 0.12
𝜏 = 0.45 0.10 0.10 0.10 0.12
Notes: This table presents compensating wage differentials for various scenarios under different time-costs of
commuting (𝜏). Scenarios shown in columns are solved with the time-cost of commuting set to 0.5 and then
0.45, respectively, for the baseline population. These scenarios are then compared to cities with an additional
10% population, with the necessary wage income-population elasticity shown. Elasticities are largely invariant
to the time-cost of commuting.

Table A.3
Comparison with Hsieh and Moretti (2019).
Simulation Model Baseline FAR Reduction Roads Reduction Land Reduction
Parameters
FAR 0–20 miles No Limit −92.14% No Limit No Limit
Land for Housing (𝜃) 0.350 0.350 0.350 0.129
Land for Roads (𝜓) 0.250 0.250 0.154 0.250

Output
Unit Size (average) 1806 3.9% −0.8% −6.6%
Unit Size (median) 1778 4.0% −0.7% −7.2%
Unit Size (CBD) 1438 −9.6% −4.7% −9.6%
Land Price (per acre, avg.) 21,223 −34.9% 5.9% 63.2%
Land Price (per acre, med.) 31,363 −28.5% 6.1% 79.2%
Land Price (per acre, CBD) 143,838 −77.5% 34.5% 84.0%
House Price (per sq. ft., avg.) 8.67 −6.8% 1.6% 13.3%
House Price (per sq. ft., med.) 8.82 −6.8% 1.3% 14.3%
House Price (per sq. ft., CBD) 12.92 20.3% 9.1% 20.3%
Housing Stock (sq. ft., millions) 5806 4.0% −0.8% −6.5%
Housing Expenditure (per month, avg.) 1304 −3.1% 0.8% 5.8%
FAR (CBD) 0.89 −92.1% 17.0% 37.5%
Residential Density (hh per sq. mile) 1339 −35.6% 3.4% −36.2%
City Area (sq. miles) 2389 55.5% −3.3% 57.0%
Time to Work (avg.) 29.9 40.7% 46.6% 11.6%
Wage Income 75,158 4.9% 4.9% 4.9%
Notes: This table presents simulation output from the baseline model and three different counterfactual scenarios, all
at the baseline city size. Column 1 represents the baseline model which is calibrated with respect to a laissez-faire
rendition of Chicago. The remainder of the columns give restrictions resulting in a change in wage income of $3, 685,
the amount found by Hsieh and Moretti (2019) as the annual lost earnings per worker due to land use regulations.

17
W. Larson, A. Yezer and W. Zhao Regional Science and Urban Economics 96 (2022) 103802

References Growth of Chicago to the Rise of its Land Values, 1830-1933. University of Chicago
Press, Chicago.
Ahlfeldt, G.M., Redding, S.J., Sturm, D.M., Wolf, N., 2015. The economics of density: Hsieh, C.-T., Moretti, E., 2019. Housing constraints and spatial misallocation. Am. Econ.
evidence from the Berlin Wall. Econometrica 83 (6), 2127–2189. J. Macroecon. 11 (2), 1–39.
Albouy, D., 2016. What are cities worth? land rents, local productivity, and the total Ihlanfeldt, K.R., et al., 2007. The effect of land use regulation on housing and land
value of amenities. Rev. Econ. Stat. 98 (3), 477–487. prices. J. Urban Econ. 61 (3), 420–435.
Albouy, D., Ehrlich, G., Liu, Y., 2016. Housing Demand, Cost-Of-Living Inequality, and Khan, A.R., 2020. Decentralized Zoning with Agglomeration Spillovers: Evidence from
the Affordability Crisis. National Bureau of Economic Research. Technical report. Aldermanic Privilege in Chicago. University of Illinois Working Paper.
Albouy, D., Lue, B., 2015. Driving to opportunity: local rents, wages, commuting, and Larson, W., Liu, F., Yezer, A., 2012. Energy footprint of the city: effects of urban land
sub-metropolitan quality of life. J. Urban Econ. 89, 74–92. use and transportation policies. J. Urban Econ. 72 (2–3), 147–159.
Alonso, W., 1964. Location and Land Use: toward a General Theory of Land Rent. Larson, W., Yezer, A., 2015. The energy implications of city size and density. J. Urban
Harvard University Press. Econ. 90, 35–49.
Altmann, J.L., DeSalvo, J.S., 1981. Tests and extensions of the Mills-Muth simulation Larson, W., Zhao, W., 2017. Telework: urban form, energy consumption, and greenhouse
model of urban residential land use. J. Reg. Sci. 21 (1), 1–21. gas implications. Econ. Inq. 55 (2), 714–735.
American Community Survey, 2019. 1-year Estimates. U.S. Census Bureau. Technical Larson, W., Zhao, W., 2020. Self-driving cars and the city: effects on sprawl, energy
report. consumption, and housing affordability. Reg. Sci. Urban Econ. 81, 103484.
American Housing Survey, 2019. Metropolitan Public Use File. Technical Report. U.S. Liu, Y., 2018. Estimating the elasticity of supply of housing space rather than units. Reg.
Department of Housing and Urban Development. Sci. Urban Econ. 68, 1–10.
Anas, A., 2015. Why are urban travel times so stable? J. Reg. Sci. 55 (2), 230–261. Lucas, R.E., Rossi-Hansberg, E., 2002. On the internal structure of cities. Econometrica
Anas, A., Hiramatsu, T., 2013. The economics of cordon tolling: general equilibrium and 70 (4), 1445–1476.
welfare analysis. Econ. Transport. 2 (1), 18–37. Malpezzi, S., 1996. Housing prices, externalities, and regulation in us metropolitan
Anas, A., Liu, Y., 2007. A regional economy, land use, and transportation model areas. J. Hous. Res. 209–241.
(RELU-TRAN ): formulation, algorithm design, and testing. J. Reg. Sci. 47 (3), Mayer, C.J., Somerville, C.T., 2000. Land use regulation and new construction. Reg. Sci.
415–455. Urban Econ. 30 (6), 639–662.
Arnott, R.J., MacKinnon, J.G., 1977. The effects of the property tax: a general McDonald, J.F., 2009. Calibration of a monocentric city model with mixed land use and
equilibrium simulation. J. Urban Econ. 4 (4), 389–407. congestion. Reg. Sci. Urban Econ. 39 (1), 90–96.
Association, A.A., 2019. Your Driving Costs. Technical Report. American Automobile McMillen, D.P., 2004. Employment densities, spatial autocorrelation, and subcenters in
Association. large metropolitan areas. J. Reg. Sci. 44 (2), 225–244.
Baum-Snow, N., Han, L., 2019. The Microgeography of Housing Supply. (Technical Mills, E.S., 1967. An aggregative model of resource allocation in a metropolitan area.
report). Am. Econ. Rev. 57 (2), 197–210.
Bertaud, A., Brueckner, J.K., 2005. Analyzing building-height restrictions: predicted Molloy, R., 2020. The effect of housing supply regulation on housing affordability: a
impacts and welfare costs. Reg. Sci. Urban Econ. 35 (2), 109–125. review. Reg. Sci. Urban Econ. 80, 103350 (Special Issue on Housing Affordability).
Borck, R., 2016. Will skyscrapers save the planet? building height limits and urban Monte, F., Redding, S.J., Rossi-Hansberg, E., 2018. Commuting, migration, and local
greenhouse gas emissions. Reg. Sci. Urban Econ. 58, 13–25. employment elasticities. Am. Econ. Rev. 108 (12), 3855–3890.
Brooks, L., Liscow, Z.D., 2019. Infrastructure Costs. Available at: SSRN 3428675. Munneke, H.J., 2005. Dynamics of the urban zoning structure: an empirical
Broxterman, D., Liu, Y., 2019. What Is the Relation between Land Use Controls and investigation of zoning change. J. Urban Econ. 58 (3), 455–473.
Housing Prices. SSRN Working Paper. Muth, R., 1969. Cities and Housing: the Spatial Pattern of Urban Residential Land Use.
Brueckner, J.K., 2007. Urban growth boundaries: an effective second-best remedy for University of Chicago Press.
unpriced traffic congestion? J. Hous. Econ. 16 (3), 263–273. Muth, R.F., 1975. Numerical solution of urban residential land-use models. J. Urban
Brueckner, J.K., Singh, R., 2020. Stringency of land-use regulation: building heights in Econ. 2 (4), 307–332.
US cities. J. Urban Econ. 103239. Quigley, J.M., Raphael, S., 2005. Regulation and the high cost of housing in California.
Contat, J., Larson, W.D., 2021. Transaction Composition and House Price Index Am. Econ. Rev. 95 (2), 323–328.
Measurement: Evidence from a Repeat-Sales Aggregation Index. FHFA Working Roback, J., 1982. Wages, rents, and the quality of life. J. Polit. Econ. 90 (6), 1257–1278.
Paper Series. Rosen, S., 1974. Hedonic prices and implicit markets: product differentiation in pure
Davis, M.A., Larson, W.D., Oliner, S.D., Shui, J., 2021. The price of residential land for competition. J. Polit. Econ. 82 (1), 34–55.
counties, ZIP codes, and census tracts in the United States. J. Monetary Econ. 118, Saiz, A., 2010. The geographic determinants of housing supply. Q. J. Econ. 125 (3),
413–431. 1253–1296.
Delventhal, M.J., Kwon, E., Parkhomenko, A., 2021. How do cities change when we Schwieterman, J.P., Caspall, D.M., 2006. The Politics of Place: a History of Zoning in
work from home? J. Urban Econ. 127 103331. Chicago. Lake Claremont Press.
Duranton, G., Turner, M.A., 2011. The fundamental law of road congestion: evidence Segal, D., Srinivasan, P., 1985. The impact of suburban growth restrictions on us
from US cities. Am. Econ. Rev. 101 (6), 2616–2652. housing price inflation, 1975–1978. Urban Geogr. 6 (1), 14–26.
Epple, D., Gordon, B., Sieg, H., 2010. A new approach to estimating the production Shertzer, A., Twinam, T., Walsh, R.P., 2016. Race, ethnicity, and discriminatory zoning.
function for housing. Am. Econ. Rev. 100 (3), 905–924. Am. Econ. J. Appl. Econ. 8 (3), 217–246.
Fujita, M., Ogawa, H., 1982. Multiple equilibria and structural transition of Shertzer, A., Twinam, T., Walsh, R.P., 2017. Zoning and the Economic Geography of
non-monocentric urban configurations. Reg. Sci. Urban Econ. 12 (2), 161–196. Cities. Mimeo.
Glaeser, E., Gyourko, J., 2018. The economic implications of housing supply. J. Econ. Small, K.A., Verhoef, E.T., Lindsey, R., 2007. The Economics of Urban Transportation.
Perspect. 32 (1), 3–30. Routledge.
Glaeser, E.L., Gyourko, J., Saks, R.E., 2005. Urban growth and housing supply. J. Econ. Thurston, L., Yezer, A.M.J., 1994. Causality in the suburbanization of population and
Geogr. 6 (1), 71–89. employment. J. Urban Econ. 35 (1), 105–118.
Glaeser, E.L., Mare, D.C., 2001. Cities and skills. J. Labor Econ. 19 (2), 316–342. Timothy, D., Wheaton, W.C., 2001. Intra-urban wage variation, employment location,
Green, R.K., Malpezzi, S., Mayo, S.K., 2005. Metropolitan-specific estimates of the price and commuting times. J. Urban Econ. 50 (2), 338–366.
elasticity of supply of housing, and their sources. Am. Econ. Rev. 95 (2), 334–339. Turner, M.A., Haughwout, A., van der Klaauw, W., 2014. Land use regulation and
Gyourko, J., Hartley, J.S., Krimmel, J., 2021. The local residential land use regulatory welfare. Econometrica 82 (4), 1341–1403.
environment across us housing markets: evidence from a new wharton index. J. Wallace, N.E., 1988. The market effects of zoning undeveloped land: does zoning follow
Urban Econ. 124, 103337. the market? J. Urban Econ. 23 (3), 307–326.
Gyourko, J., Saiz, A., Summers, A., 2008. A new measure of the local regulatory
environment for housing markets: the Wharton Residential Land Use Regulatory
Index. Urban Stud. 45 (3), 693–729.
Helms, A.C., 2003. Understanding gentrification: an empirical analysis of the Senior Economist, Federal Housing Finance Agency (FHFA). Larson contributed to this
determinants of urban housing renovation. J. Urban Econ. 54 (3), 474–498. research in his personal capacity and not as part of his official duties at the Federal
Herkenhoff, K.F., Ohanian, L.E., Prescott, E.C., 2018. Tarnishing the Golden and Empire Housing Finance Agency. The analysis and conclusions are those of the authors alone and
states: land-use restrictions and the US economic slowdown. J. Monetary Econ. 93, should not be represented or interpreted as conveying an official Federal Housing Finance
89–109. Agency position, policy, analysis, opinion or endorsement. Any errors or omissions are the
Hoyt, H., 1933. One Hundred Years of Land Values in Chicago: the Relationship of the sole responsibility of the authors.

18