Sie sind auf Seite 1von 23

Int Tax Public Finance

DOI 10.1007/s10797-016-9401-8

Common labor market, attachment and spillovers in a


large metropolis
Emilson Caputo Delfino Silva1 Vander Mendes Lucas2

Springer Science+Business Media New York 2016

Abstract In this paper, we extend the home-attachment model to a setting with multiple (more than two) jurisdictions and consider non-cooperative policy making for
provision of different types of metropolitan public goods in the presence of a common
labor market. Migration and working place choices are independent. We show that the
optimal redistributive policy implemented by a central authority always yields equalization of private consumption levels across jurisdictions. This result holds whether or
not policy makers are able to anticipate migration responses to their policy choices. In
the decentralized leadership games, jurisdictions make choices that fully internalize
externalities.
Keywords
leadership

Common labor market Home attachment Spillovers Decentralized

JEL Classification

C72 D62 H23 H42 H77 H87 R3 R5

1 Introduction
In many nations, one can easily find evidence of large common labor markets
in operation in major metropolitan areas. There are good examples in Brazil

Emilson Caputo Delfino Silva


emilson@ualberta.ca
Vander Mendes Lucas
vlucas@unb.br

Department of Marketing, Business Economics and Law, University of Alberta, Edmonton,


AB T6G 2R6, Canada

Department of Economics, University of Brasilia, Brasilia, DF 70919-970, Brazil

123

E. C. D. Silva, V. M. Lucas

(So Paulo), Canada (Vancouver), France (Paris), Japan (Tokyo), South Korea
(Seoul), the USA (New York and Washington, D.C.) and UK (London). Some
metropolitan areas have metropolitan governments. Metro Vancouver provides a
good example (see, e.g., http://www.metrovancouver.org/about/aboutuspublications/
BoardStrategicPlan2015-2018.pdf). It contains 23 members, being 21 municipalities,
1 treaty First Nation and 1 electoral area. A Board of Directors is the governing body
of Metro Vancouver. Its core areas are the provision of drinking water, wastewater
treatment, solid waste management, affordable housing and transportation and parks.
It also regulates air quality and urban growth.
Following Wildasin (1991, p. 758), we characterize a common labor market as
one in which at least some portion of the work force is able to switch from jobs in
one jurisdiction to jobs in another jurisdiction (within the time frame of the analysis,
and possibly at a nonzero cost). In this paper, we consider a large metropolitan
economy, with many jurisdictions (local governments) and a central, metropolitan,
government (like Metro Vancouver). In this metropolitan economy, individuals may
work in jurisdictions other than the ones in which they reside. Individuals are attached
to home.
Common labor markets involve substantive flows of commuters and should yield
a high degree of wage convergence across jurisdictions, particularly for occupations that face fierce competition between residents and commuters. Canadas 2011
National Household Survey (NHS) (see, e.g., https://www12.statcan.gc.ca/nhs-enm/
2011/as-sa/99-012-x/99-012-x2011003_1-eng.cfm) reveals that nearly 15.4 million
Canadians commuted to work, while 1.1 million worked mostly at home. On average,
commuters spent 25.4 min traveling to work. Many commuters worked in jurisdictions other than the ones in which they resided. For example, 46.1 % of commuters
who lived in Laval traveled to work in the municipality of Montreal, and 36.1 % of
commuters who lived in Burnaby traveled to work in the municipality of Vancouver.
Large commuting flows are observable in the USA. According to Agrawal and
Hoyt (2014), 75 million people reside and work in multiple-state MSAs in the USA.
Figure 1 displays the two largest commuting networks in this nation in 2011. Note
that New York, Connecticut, New Jersey and Pennsylvania are all connected directly
or indirectly with each otherNew York and New Jersey are commuting hubs
and Connecticut and Pennsylvania are commuting spokes. District of Columbia,
Maryland and Virginia are all directly connected with each other.
We build a model that is similar in many respects (especially with regard to the interjurisdictional consumption spillovers) to the one utilized in Wellisch (1994), which
demonstrates that non-cooperative jurisdictional policies are inefficient.1 Wellisch
1 Mansoorian and Myers (1993) and Wellisch (1994) demonstrate that decentralized policy is efficient in

the presence of home attachment if regional governments make transfers to each other and provide public
goods whose consumption benefits are enjoyed by residents only. Caplan et al. (2000) show that regional
authorities make efficient contributions to a federal pure public good in the presence of home attachment
if the central government redistributes interregional income after it observes the regional governments
contributions (regional and central governments play a decentralized leadership game). The linearity
in the aggregation technology produces the necessary incentive equivalence, which yields their efficiency
result. This fact is made clear by subsequent work (see, e.g., Silva and Yamaguchi 2010; Caplan and Silva
2011), in which the assumption that the federal public good is pure is relaxed. These papers demonstrate

123

Common labor market, attachment and spillovers in a large

Fig. 1 Top commuting flows in the USA2011 (US Census Bureau 2011)

(1994) assumes that labor markets are jurisdictional, ignoring the possibility that
workers supply labor in a jurisdiction in which they do not reside. Unlike Wellisch
(1994), we show that non-cooperative jurisdictional policies are efficient in the presence of a common labor market and of a benevolent supra-jurisdictional (metropolitan)
authority that cares about redistribution when jurisdictional and metropolitan policies
fully account for migration responses and the metropolitan authority redistributes
income between jurisdictions after it observes the choices made by the jurisdictions. Jurisdictional and metropolitan authorities play a sequential game in which
the metropolitan authority is a followeri.e., they play a decentralized leadership
game. We also show that the subgame-perfect equilibrium to this game coincides with
the subgame-perfect equilibrium for an alternative decentralized leadership game in
which jurisdictional and metropolitan authorities are unable to commit to public policies vis--vis the imperfectly mobile population of consumers. In other words, the
subgame-perfect equilibrium for the decentralized leadership game played by jurisdictional and metropolitan authorities is efficient whether or not these players take
migration choices as given. An immediate policy implication of our results is the need
of a central authority at the metropolitan level to implement intra-jurisdictional income
transfers, since the equalization of private consumption levels across jurisdictions provides incentives for the jurisdictional governments to behave efficiently.
Aoyama and Silva (2010) builds a model after Wellisch (1994) and shows that the
subgame-perfect equilibrium for a decentralized leadership game with interregional
spillovers is socially efficient under two particular circumstances: first, if regional
and central governments payoff functions are Rawlsian, and second, if the regional
governments payoff functions are quasi-linear and the central governments payoff function is a Bergson-Samuelson transformation of regional welfare levels. In
both cases, the centers income redistribution policy yields perfect incentive equivalence, which motivates the regional governments to make choices that internalize
interregional spillovers. More recently, Boadway et al. (2013) also shows that the
Footnote 1 continued
that, all else equal, the subgame-perfect equilibrium for the decentralized leadership game is not socially
optimal. Aoyama and Silva (2013) also builds a model after Wellisch (1994) in which regional governments
possess Rawlsian payoff functions and demonstrate that the decentralized non-cooperative equilibrium is
efficient in the presence of home attachment and interregional spillovers.

123

E. C. D. Silva, V. M. Lucas

subgame-perfect equilibrium for a decentralized leadership game, with ex post interregional income transfers implemented by a utilitarian central government and in which
there are interregional externalities other than those implied by a pure federal public
good, is socially optimal. However, this result depends crucially on their modeling
assumption that the regional welfare functions are concave transformations of quasilinear utilities (i.e., one of the two cases examined in Aoyama and Silva 2010). In such
a case, the centers optimal redistributive policy equates welfare levels across regions,
producing perfect incentive equivalence and efficiency. As discussed in Aoyama and
Silva (2010), it is straightforward to show that if the modeling assumption that regional
utilities represent quasi-linear preferences is relaxed and the welfare function is characterized by diminishing marginal utility of income, the efficiency results of Boadway
et al. (2013) do not hold in general.
Boadway et al. (2013) shows that regional governments make choices that internalize interregional externalities when migration occurs prior to policy decisions.
Boadway et al. (2013) also shows that the labor market allocations will typically be
inefficient, since the regionally independent labor markets will generally be characterized by different marginal labor productivities. In our model, the regional marginal
productivities of labor are necessarily equal, since this is a natural implication of a
common labor marketfirms in all jurisdictions choose labor quantities in order to
equate the marginal productivity of labor to the common wage rate.
Starting with Wildasin (1991), the fiscal federalism literature (see, e.g., Wildasin
(1991), Wellisch and Wildasin (1996), Rothstein and Hoover (2006) and Wilson
(2007)) has examined fiscal and redistributive implications of government policies
in the presence of a common labor market. Among other important results, Wildasin
(1991) demonstrates that a central authority may provide grants to lower- level governments that induce the latter to behave efficiently in the implementation of redistributive
schemes. Wellisch and Wildasin (1996) study the fiscal impacts of immigration in a
federation characterized by decentralized tax and redistributive policy making. They
show that immigration creates fiscal externalities and that these externalities can be
fully internalized if the central government implements an interregional system of
fiscal grants. Wilson (2007) also examines fiscal and welfare effects of immigration
control policies. Among other important results, he shows that a host country may benefit from attracting a large flow of non-impoverished workers from another country in
the presence of a common labor market. Unlike this literature, we do not consider either
decentralized income redistribution or immigration. Jurisdictions contribute to multiple types of metropolitan public goods: (i) They share contributions to a metropolitan
public good whose aggregation technology is not restricted to summation (see, e.g.,
Cornes 1993; Silva 2014), and (ii) they provide at least one public good that generates consumption spillover benefits to all other jurisdictions (see, e.g., Wellisch 1994;
Aoyama and Silva 2010).
In line with most of the fiscal federalism literature, we first consider an economy with two jurisdictions. This allows us to clearly demonstrate that our model is
a hybrid of models used in the common-labor-market literature and in the homeattachment literature. We later extend the model to allow for many jurisdictions. As
far as we know, we are the first ones to build a home-attachment model with spillovers
and a common labor market for an economy featuring a federation-like (hierarchi-

123

Common labor market, attachment and spillovers in a large

cal) governmental structure containing more than two jurisdictions. We postulate that
nature determines each individuals attachment benefits prior to the commencement
of all decision making. Each individual receives a message from nature that tells
him/her that he/she enjoys attachment benefits for a predetermined pair of jurisdictions
only. The individual does not enjoy any attachment benefit with respect to jurisdictions
other than the two jurisdictions that nature selects for him/her. The assumption that
each individual derives attachment benefits for two jurisdictions only does not seem
to be too demanding, as it is unlikely that in practice individuals enjoy attachment
benefits for several jurisdictions. Home attachment is typically determined by culture,
family or social customs.
Natures selection procedure implies that the total population is arbitrarily partitioned into home-attachment groups. These groups are not necessarily of the same size.
This asymmetric aspect of the model enables us to obtain a non-cooperative equilibrium in which jurisdictions have populations of different sizes. This certainly has a
very strong empirical appeal. We show that the main results for the small economy
(with two jurisdictions) also hold in the large economy.
We organize the remainder of the paper as follows. Section 2 introduces the model
with two jurisdictions. In Sects. 2.1 and 2.2, we examine socially optimal allocations
under jurisdictional and common labor markets. This enables us to clearly point out the
restrictive effects on the social optimum associated with the conventional assumption
that labor markets are jurisdictional, since we compare the social optimum under
jurisdictional labor markets with the social optimum under a common labor market.
This strategy also allows us to provide intuition for our equilibrium results in Sect. 2.3.
Sect. 2.4 extends the basic model by allowing interjurisdictional commuting costs.
Section 3 extends the basic model to one in which the federal economy contains many
jurisdictions. Section 4 concludes the paper.

2 Model with two regions


Consider a small economy with two jurisdictions, indexed by j, j = 1, 2. In this economy, individuals supply labor to competitive firms that produce a numeraire good. We
assume that the other inputs used by the firms (e.g., land and capital) are held fixed.
There are M j 2 firms in jurisdiction j. All firms use the same technology. This
technology is represented by a production function, F (.), which is concave (characterized by constant returns to scale), increasing in all arguments, with diminishing
marginal products, and
satisfies
the
 which


 standard Inada conditions (i.e., all inputs are
essential).2 Let f j lm j F lm j ; zm j denote the reduced form for the production
function used by a representative firm,3 m, m = 1, . . . , M j , in jurisdiction j, where
lm j is the quantity of labor hired and zm j is the vector of inputs that are held fixed. For
simplicity and to keep things as symmetrical as possible, we assume that zm j = z j
for all m = 1, . . . , M j . For concreteness, we also assume that the jurisdictions are
asymmetric: Region 1 possesses more of each of the fixed inputs than region 2 (i.e.,
2 This assumption guarantees that, in any equilibrium, all jurisdictions are populated.
3 In what follows, we use superscripts to denote functions.

123

E. C. D. Silva, V. M. Lucas

z1 >> z2 and M1 M2 ). The jurisdictional asymmetry provides the impetus for


interjurisdictional redistribution in the analysis below.
In order to make clear the role that a common labor market plays in the allocation of
resources in our economy, we shall consider two settings: one with segmented (jurisdictional) labor markets and another with a common labor market.4 With jurisdictional
labor markets, firms within a jurisdiction must hire labor from the jurisdictions competitive labor market. In this case, an individuals choice of a jurisdiction for residence
also corresponds to this individuals choice of a jurisdiction for work. With a common
labor market, firms are able to hire individuals who reside in either jurisdictionan
individuals choice of a jurisdiction for residence does not limit this individuals choice
of a jurisdiction for work.
Let w j denote the market wage in jurisdiction j. With a common labor market,
w j = w, j = 1, 2. Assume throughout that the price of the numeraire good is equal
to 1. The representative firmin jurisdiction
j chooses a nonnegative amount of labor

input, lm j , to maximize f j lm j w j lm j . Assuming an interior solution, the firstorder condition informs us that the firm hires labor up to the point where the marginal
j
product of labor is just equal to the marginal cost of labor, flm d f j /dlm j = w j , m =




j
1, . . . , M j , j = 1, 2. Hence, we obtain lm w j = l j w j , m = 1, . . . , M j and
j = 1, 2. In words, all firms within a jurisdiction demand the same quantity of labor.
With jurisdictional labor markets, the market-clearing conditions are
 
j = 1, 2,
L j w j = n j l,
(1a)
 
 
where L j w j M j l j w j and n j and l > 0 are, respectively, jurisdiction js
population of workers (and consumers) and the total amount of labor units that each
worker can supply in the market. We assume that each worker supplies all of his
available labor units in the market inelastically provided that w j > 0. Without loss of
generality, we assume that l = 1 in what follows. The market-clearing conditions (1a)
allows us to define the jurisdictional wages as implicit functions of jurisdictional labor
markets characteristics, w j = w j n j , M j , j = 1, 2. Henceforth, we shall ignore
 
the fixed numbers of firms and simply write w j = w j n j , j = 1, 2.
With a common labor market, the market-clearing condition is
L 1 (w) + L 2 (w) = N ,

(1b)

4 Following the literature on home attachment, we ignore intra-jurisdictional commuting costs throughout.
(Since average travel time from home to work in Canada is 25.4 min and this figure includes interjurisdictional travel time, it seems that the average intra-jurisdictional commuting cost in Canada is not very high.
Nonetheless, neglecting it certainly implies a limitation of the analysis.) We consider the effects promoted
by interjurisdictional commuting costs in Sect. 2.4. In future work, we plan to consider intra-jurisdictional
commuting costs as well. To properly model intra-jurisdictional commuting costs, one would have to build
a spatial model (see, e.g., Brueckner and Kim 2003; Brueckner and Helsley 2011; Agrawal and Hoyt
2014) where the distance between residence and working places would be taken into account. The model
would also need to account for the location of ones house and differing property and income tax payments.
In equilibrium, assuming standard housing size and quality and no amenity differences across locations, all
individuals would have to be indifferent between commuting costs and property and income tax payments
(which would be an increasing function of the housing value) within any region and across regions (see,
for example, Brueckner and Helsley 2011; Agrawal and Hoyt 2014).

123

Common labor market, attachment and spillovers in a large

where L j (w) M j l j (w), j = 1, 2, and N = n 1 +n 2 is the total population of workers in the economy. The market-clearing condition (1b) allows us to define the market
wage as an implicit function of the labor market characteristics, w = w (M1 , M2 , N ).
Therefore, the critical difference between the two settings is the fact that, under jurisdictional labor markets, the jurisdictional wages depend on the (endogenous) sizes
of the jurisdictional populations, whereas under a common labor market the common
wage does not depend upon the sizes of the jurisdictional populations. For the case
of a common labor market, we shall ignore the fixed parameters and simply write the
market wage as w.
All consumers in the economy have identical preferences with respect to the consumption goods. The representative consumer in jurisdiction j derives utility from
consumption of x j units of numeraire good, G 1 units of a public good provided
in jurisdiction 1, G 2 units of a public good provided in jurisdiction 2 and Q units
of a metropolitan public good produced with contributions from both jurisdictions.
We assume that Q =  (q1 , q2 ), where  (.) is a concave, continuous and differentiable function, increasing in both arguments, and q j is the contribution made by
jurisdiction j.5 The representative consumer in jurisdiction j obtains utility from
consumption of private
and public

 goods according to the following concave utility
function, u j = u x j , G 1 , G 2 , Q . In most of what follows, we assume that the marj
ginal utility from private consumption is diminishing and lim x j 0 u x = .6 Note
j

that u x u j / x j , j = 1, 2. The form of this utility function is identical to the


form of the utility function in Wellisch (1994), except for the presence of the last
argument.7
The representative consumer in jurisdiction j has i j units of market income (wage
and profit share) plus j /n j units of income that he/she receives (if positive) or pays
(if negative) as an income transfer controlled by the metropolitan government. The
consumer spends x j +t j units of this income to finance her private and public consumption levels, where t j is the amount of tax that the consumer pays to the jurisdictional
government. This consumers budget constraint yields
x j = i j t j + j /n j ,

j = 1, 2.

(1c)

The representative consumer in jurisdiction j earns market income from supplying


labor in the market and from being a shareholder in all firms located in the jurisdiction.
Under jurisdictional labor markets, we have
   
 
 
 
j
j
j
n j , j = 1, 2, (1d)
i n j = w n j + M j n j /n j = M j f l j w j n j
j

5 Examples include  (q , q ) = q q 1 and  (q , q ) = q + (1 ) q , 0 < < 1.


1 2
1 2
1
2
1 2
6 This assumption guarantees that x > 0, j, in the social optima and equilibria examined in this paper.
j


7 If the utility function for the representative individual in jurisdiction j is u x , G , j = 1, 2, instead,
j
j

there are no public goods that yield consumption spillovers. In such a case, the model would be similar
to the one examined by Mansoorian and Myers (1993). Decentralized policy yields efficient provision of
jurisdictional public goods provided the jurisdictions make transfers to each other.

123

E. C. D. Silva, V. M. Lucas

 
   
    
w j n j l j w j n j is the profit earned by the
where j n j f j l j w j n j
representative firm in jurisdiction j. Under a common labor market, we have
 
i j n j = w + M j j (w)/n j ,

j = 1, 2,

(1e)



where j (w) f j l j (w) wl j (w).
The metropolitan government controls the interjurisdictional income transfers. We
assume that 1 + 2 = 0; that is, the income transfers are purely redistributive. Since
the representative consumer in jurisdiction j pays a tax equal to t j units of income to
her jurisdictional government for the provision of public goods, the budget constraint
faced by jurisdictional government j is
n jtj = G j + qj,

j = 1, 2,

(1f)

where, for simplicity, we assume that it takes one unit of numeraire good to produce
one unit of each type of public good.
Consumers are free to choose their jurisdiction of residence. However, we consider situations where consumers are attached to jurisdictions. Attachment is captured
through an idiosyncratic attachment benefit function. Let n [0, N ] denote a consumer in the economy. This individual gets an attachment benefit equal to a (N n) if
she resides in jurisdiction 1 and an attachment benefit equal to an if she resides in jurisdiction 2, where a > 0 is the attachment intensity. Hence, the total utility individual n
derives from residing in jurisdiction 1 is u (x1 , G 1 , G 2 , Q)+a (N n), while the total
utility this individual derives from residing in jurisdiction 2 is u (x2 , G 1 , G 2 , Q) + an.
In the migration equilibrium, there is an individual, n 1 , who is indifferent between
residing in jurisdiction 1 and residing in jurisdiction 2:
u (x1 , G 1 , G 2 , Q) + a (N n 1 ) = u (x2 , G 1 , G 2 , Q) + an 1 .

(1g)

2.1 Social optimum under regional labor markets


We first derive the social optimum for the setting with jurisdictional labor markets. Let
n
N
S 1 = 0 1 [u (x1 , G 1 , G 2 , Q) + a (N n)]dn and S 2 = n 1 [u (x2 , G 1 , G 2 , Q) + an]
dn be the welfare levels enjoyed by jurisdictions 1 and 2, respectively. We assume that
the social planner is utilitarian. Hence,


n 1 
S = S 1 + S 2 = n 1 u (x1 , G 1 , G 2 , Q) + a N

 2 n 
2
.
+ n 2 u (x2 , G 1 , G 2 , Q) + a N
2

(2a)

We assume that the social planner takes the migration response function (derived from
the migration equilibrium condition)
into account
when it chooses taxes, transfers


and public expenditure levels, t j , j , G j , q j j=1,2 , to maximize the social welfare
function (2a). To obtain the migration response function, first combine Eqs. (1c), (1d)

123

Common labor market, attachment and spillovers in a large

and (1f) to rewrite the budget constraint for the representative resident in jurisdiction
j as follows:
   


 
j
j
j
n j , j = 1, 2.
+ j G j qj
x n j, j, G j,qj = Mj f l w n j
j

(2b)
Now combine Eqs. (1g) and (2b) to rewrite the migration equilibrium condition:


u x 1 (n 1 , 1 , G 1 , q1 ) , G 1 , G 2 , Q + a (N n 1 )


= u x 2 (N n 1 , 1 , G 2 , q2 ) , G 1 , G 2 , Q + an 1 .

(2c)

In writing (2c), we already account for the facts that 2 = 1 and n 2 = N n 1 . Equation (2c) implicitly defines the migration response function n 1 (1 , G 1 , G 2 , q1 , q2 ).
Inserting this function into (2c) and differentiating the implied expression with respect
to each argument yields the marginal migration response functions:
 



1
2
u 1x xn1 + u 2x xn2 2a ,
= u x /n 1 + u x /n 2

(2d)




1
u 1x xn1 + u 2x xn2 2a ,
+ u x /n 1

(2e)





n 1G 2 = u 2G 2 u 1G 2 u 2x /n 2
u 1x xn1 + u 2x xn2 2a ,

(2f)

n 11


n 1G 1 = u 2G 1 u 1G 1


 



2
1
1
u 1x xn1 + u 2x xn2 2a ,
= q1 u Q u Q + u x /n 1

(2g)


 



n q12 = q2 u 2Q u 1Q u 2x /n 2
u 1x xn1 + u 2x xn2 2a ,

(2h)

n q11

where xn x j /n j , u G 1 u j /G 1 , u G 2 u j /G 2 and q j /q j , j =


1, 2. As usual, we assume that u 1x xn1 + u 2x xn2 2a < 0. This condition assures us that
the migration equilibrium is stable. The marginal migration responses (2d)(2h) are
intuitive. They capture the incremental incentive (or disincentive) of establishing residence in jurisdiction 1 as a particular policy variable changes. Consider, for example,
condition (2d). It shows that, ceteris paribus, the population in jurisdiction 1 increases
as the level of interjurisdictional transfer received by the representative resident rises.
Aggregating the individual budget constraints (2d), we can obtain the economywide resource constraint:
2 





n k (.) x k n k (.) , k , G k , qk + G k + qk X k n k (.) = 0,
(2i)
k=1


  


where X k n k (.) Mk f k l k w k n k (.)
is the total amount of numeraire good
(i.e., income) produced in jurisdiction k. In writing condition (2i), we account for the

123

E. C. D. Silva, V. M. Lucas

 

facts that 1 + 2 = 0 and Mk l k w k n k (.) = n k (.), k = 1, 2. Hence, the economywide resource constraint is satisfied whenever the individual budget constraints (2d)
and the interjurisdictional income transfer constraint, 1 + 2 = 0, are satisfied.
The social planner chooses {1 , 2 , G 1 , G 2 , q1 , q2 }to maximize:
 




n k (.)
n k (.) u x k n k (.) , k , G k , qk , G 1 , G 2 ,  (q1 , q2 ) + a N
2
k=1
(2j)

subject to 1 + 2 = 0 and 2k=1 n k (.) = N . The social welfare function (2j) corresponds to (2a) when one accounts for the migration response function and the
individual budget constraints. Letting 2 = 1 and n 2 (.) = N n 1 (.), the first-order
conditions yield
2





u 1x 1 + n 11 (w1 x1 ) = u 2x 1 + n 11 (w2 x2 ) ,
u 2G j
+
n
2 2 = 1,
u 1x
u
 x

1
2
uQ
uQ
n 1 1 + n 2 2 q j = 1,
ux
ux
n1

u 1G j

(2k)

j = 1, 2,

(2l)

j = 1, 2.

(2m)

Equation (2k) informs us that the social planners redistributive policy equates the
social marginal utilities of income. The social marginal utility of income in jurisdiction
j
j is the sum of the individual marginal utility of income enjoyed by a resident, u x ,
and the extra marginal utility of income that this individual enjoys as a product of

j 
the marginal migration response, n 11 u x w j x j . The latter term corresponds to the
marginal regional benefit (or cost) associated with the expansion (or subtraction) of
the jurisdictional population. Utilizing Eq. (2d), we can rewrite Eq. (2k) as follows:


N u 1x u 2x [(w1 x1 ) (w2 x2 )] = 2an 1 n 2 u 2x u 1x .

(2n)

The literature refers to Eq. (2n) as the socially efficient marginal migration condition.
For future reference and comparison, note that u 2x = u 1x if and only if w1 = w2 . This
is true because u 2x = u 1x if and only if x1 = x2 , which occurs if and only if w1 = w2 .
Otherwise, the left-hand side of condition (2n) is not zero (positive if w1 > w2 or
negative if w1 < w2 ). With jurisdictional labor markets and jurisdictional asymmetry
with respect to the fixed inputs available in each jurisdiction (i.e., z1 = z2 ), the
jurisdictional wages are not equal in general. Hence, the socially efficient allocation
under jurisdictional labor markets does not generally yield u 2x = u 1x and x1 = x2 .
Equations (2l) and
(2m) are
the Samuelson conditions for the optimal provision
of the public goods, G j , q j j=1,2 . We obtain each Samuelson condition from each
public goods first-order condition, and by using Eqs. (2d), (2k) and the relevant
marginal migration response functioni.e., (2e) for G 1 , (2f) for G 2 , (2g) for q1 and
(2h) for q2 .

123

Common labor market, attachment and spillovers in a large

2.2 Social optima under a common labor market


As in the previous section, we first derive the socially optimal allocation when the
social planner accounts for the migration response function. Combining (1c), (1e) and
(1f), we can write the individual private consumption level in region j as follows:




n j,
x j n j , j , G j , q j , w = w + M j j (w) + j G j q j

j = 1, 2. (3a)

Equation (3a) differs from Eq. (2b) due to the fact that w is not a function of the
regional population sizes. Plugging the functions (3a) into the migration equilibrium
condition (1g), we can rewrite the latter as follows:


u x 1 (n 1 , 1 , G 1 , q1 ) , G 1 , G 2 , Q + a (N 2n 1 )


= u x 2 (N n 1 , 1 , G 2 , q2 ) , G 1 , G 2 , Q .

(3b)

Equation (3b) enables us to implicitly define n 1 (1 , G 1 , G 2 , q1 , q2 ), the migration


response function.8 Plugging n 1 (1 , G 1 , G 2 , q1 , q2 ) into (3b) and differentiating the
implied expression with respect to each policy variable yields the marginal migration
response functions:

   

u 1x xn1 + u 2x xn2 2a ,
n 11 = u 1x /n 1 + u 2x n 2

(3c)





n 1G 1 = u 2G 1 u 1G 1 + u 1x /n 1
u 1x xn1 + u 2x xn2 2a ,

(3d)





2
1
2
u 1x xn1 + u 2x xn2 2a ,
= u G 2 u G 2 u x /n 2

(3e)


 



n q11 = q1 u 2Q u 1Q + u 1x /n 1
u 1x xn1 + u 2x xn2 2a ,

(3f)


 



2
1
2
u 1x xn1 + u 2x xn2 2a ,
= q2 u Q u Q u x /n 2

(3g)

n 1G 2

n q12

Equations (3c3g) have the same format as Eqs. (2d2h). Two systems of equations
differ only with respect to the different impacts caused by jurisdictional versus common
wages.
The social planner chooses {1 , 2 , G 1 , G 2 , q1 , q2 } to maximize
8 We use the symbol on the functions in this section in order to distinguish them from their counterparts

in the previous section.

123

E. C. D. Silva, V. M. Lucas


 



n k (.)
n k (.) u x k n k (.) , k , G k , qk , G 1 , G 2 ,  (q1 , q2 ) + a N
2
k=1
(3h)

subject to 1 + 2 = 0 and 2k=1 n k (.) = N . Letting 2 = 1 and n 2 (.) = N
n 1 (.), the first-order conditions yield the Samuelson conditions (2l) and (2m) and the
following condition that governs the optimal policy for income redistribution across
jurisdictions:
2





u 1x 1 + n 11 (w x1 ) = u 2x 1 + n 11 (w x2 ) N u 1x u 2x (x2 x1 )


= 2an 1 n 2 u 2x u 1x .

(3i)

We claim that Eq. (3i) implies that x1 = x2 in the social optimum. Suppose not. Let
x1 > x2 . Then, u 1x < u 2x because u x x < 0. But, this violates Eq. (3i) because the lefthand side is negative and the right-hand side is positive. A similar argument applies if
x2 > x1 . Hence, we must have x1 = x2 and subsequently u 1x = u 2x .
But, if x1 = x2 , we obtain n 1 = n 2 = N /2 from (1b) and n 1 + n 2 = N . The
following proposition summarizes the important results we have obtained so far:
Proposition 1 Under a common labor market, if the social planners problem fully
accounts for migration responses to the choices of private and public expenditure
levels, the socially optimal allocation involves x1 = x2 , u 1x = u 2x and n 1 = n 2 = N /2.
Unlike the socially optimal allocation under jurisdictional labor markets, the socially
optimal allocation under a common labor market always yields equalization of marginal utilities of income and individual consumption levels for the private good. This
occurs because, in equilibrium, the labor market produces a wage that is common in
both jurisdictions. It must also be noted that the socially optimal level under a common labor market is at least as large as the socially optimal level under jurisdictional
labor markets because the marginal products of labor are necessarily equal across
jurisdictions under a common labor market.
We now demonstrate that the social optimum remains unchanged if the social planner is unable to account for the migration responses when it chooses the private and
public
expenditure
levels. To seethis formally,

  suppose that the social planner chooses


j , G j , q j j=1,2 to maximize 2k=1 n k u x k (n k , k , G k , qk ) , G 1 , G 2 ,  (q1 , q2 )




+a N n2k subject to 1 + 2 = 0, taking {n 1 , n 2 } as given. Letting 2 = 1 ,
the first-order conditions yield the Samuelson conditions (2l) and (2m) and the optimal income redistribution policy (2p). Hence, in the (constrained) social optimum,
x1 = x2 , u 1x = u 2x and n 1 = n 2 = N /2.
In sum, the social optimum is again the symmetric Pareto optimal allocation.
Proposition 2 In the presence of a common labor market, the social optimum corresponds to the symmetric Pareto optimal allocation whether or not the social planner
is able to anticipate the migration responses to its policy choices.
Proposition 2 is important because it informs us that under a common labor market
the socially optimal allocation is unique and symmetric. This follows because the

123

Common labor market, attachment and spillovers in a large

distribution of population across jurisdictions does not affect the common market
wage.
2.3 Decentralized leadership
We now consider a setting in which the economy is characterized by decentralized leadership. We focus on the setting with a common labor market, since this
is the novelty of this paper. Later, we consider the effects of restricting the labor
market, by assuming that each firm must hire labor from the region in which it is
located.
The jurisdictional governments choose their public good contributions in full anticipation of the metropolitan governments interjurisdictional redistribution mechanism.
The decentralized leadership game has two stages, namely simultaneous choices of
public good contributions in the first stage followed by the metropolitan governments
choices of interjurisdictional income transfers in the second stage. In addition, we
initially assume that the metropolitan government chooses its redistribution policy
accounting for the migration responses to its choices. The payoffs for the jurisdictional governments are the jurisdictional welfare functions. The payoff for the central
government is the social welfare function. The equilibrium concept is subgame perfection.
Consider the second stage of the game. The problem facing the metropolitan government is equivalent to the problem that the social planner faces when it accounts
for the migration responses, except for the fact that the central government does not
control the public good expenditure levels. Thus, the conditions that characterize the
solution in the second stage are (3i) and 1 + 2 = 0. By Proposition 1, we know that
x1 = x2 = x, u 1x = u 2x and n 1 = n 2 = N /2. Remember that 1 + 2 = 0 implies that
the economy-wide resource constraint is satisfied. Hence, we can utilize this constraint
to obtain:


2

N,
(4a)
x (G 1 , G 2 , q1 , q2 ) = I
(G k + qk )
k=1

where I = N w + M1 1 (w) + M2 2 (w) = M1 f 1 (w) + M2 f 2 (w) = X 1 (w) +


X 2 (w) is the total income produced in the economy. Equation (4a) informs us how
the metropolitan government allocates individual private consumption in response to
the jurisdictional governments choices of public expenditures.9
Plugging n j = N /2 and (4a) into the payoff for jurisdictional government j, we
have
 


 3a N
N
u x (G 1 , G 2 , q1 , q2 ) , G 1 , G 2 , Q +
.
(4b)
2
4
Each jurisdictional government wishes to maximize thepayoff function (4b)hence,
each jurisdictional government wishes to maximize u x (G 1 , G 2 , q1 , q2 ) , G 1 , G 2 ,


 

M2 2 (w)G 2 q2 M1 1 (w) G 1 q1 /2
to express the metropolitan governments best response function in the second stage. Plugging this function
into conditions (3a) and utilizing the fact that n 1 = n 2 = N /2 yields Eq. (4a).

9 Alternatively, one can use 1 (G , G , q , q ) =


1
2 1 2

123

E. C. D. Silva, V. M. Lucas

Q). Since the jurisdictional objective functions are identical, there is perfect incentive
In the first stage, jurisdictional government j chooses nonnegative

equivalence.
G j , q j to maximize (4b) subject to Q =  (q1 , q2 ), taking the choices of the other
jurisdictional government as given. It is straightforward to show that the first- order
conditions for the problems faced by the jurisdictional governments yield Eqs. (2l)
and (2m), where n 1 = n 2 = N /2.
Proposition 3 The subgame-perfect equilibrium for the decentralized leadership
game where the metropolitan government redistributes income between jurisdictions
taking into account the migration responses to its policy choices corresponds to the
symmetric Pareto optimal allocation.
Proposition 3 is remarkable in light of the negative results in Wellisch (1994).
Our model has two key departures from Wellischs model: (i) the presence of a common labor market, and (ii) ex post income redistribution undertaken by the central
government. As discussed above, a common labor market implies that the economys income is independent from the population distribution across jurisdictions.
This, in turn, implies that the economy-wide resource constraint is also independent from the population distribution across jurisdictions, a fact that yields the result
that the centers optimal income redistribution policy equates marginal utilities of
income across jurisdictions. As the jurisdictional governments make their choices
in full anticipation of the centers redistributive policy and also knowing that the
population will be divided equally across jurisdictions, they find it optimal to make
choices that fully internalize all interjurisdictional externalities. They make contributions at levels that satisfy the Samuelson conditions for optimal provision of public
goods.
A common labor market is necessary for the efficiency of the subgame-perfect
equilibrium for the decentralized leadership game. To see this, consider the decentralized leadership game under jurisdictional labor markets. In this case, the metropolitan
governments optimal redistribution policy satisfies condition (2n) and 1 + 2 = 0.
Plug 2 = 1 into condition (2n) and utilize the implied equation to implicitly define
the central governments best response function, 1 (G 1 , G 2 , q1 , q2 ). In the first stage,
jurisdictional government 1 chooses nonnegative{G 1 , q1 } to maximize

 



n 1 (.)
n 1 (.) u x 1 n 1 (.) , 1 (.) , G 1 , q1 , G 1 , G 2 ,  (q1 , q2 ) + a N
,
2
(4c)


where n 1 (.) n 1 1 (G 1 , G 2 , q1 , q2 ) , G 1 , G 2 , q1 , q2 , taking the choices of government 2 as given. Similarly, government 2 chooses nonnegative {G 2 , q2 } to maximize
 




n 2 (.)
n 2 (.) u x 2 n 2 (.) , 2 (.) , G 2 , q2 , G 1 , G 2 ,  (q1 , q2 ) + a N
,
2
(4d)


where n 2 (.) n 2 2 (G 1 , G 2 , q1 , q2 ) , G 1 , G 2 , q1 , q2 = N n 1 (.), taking the
choices of government 1 as given. Intuitively, each government makes choices that do
not fully internalize externalities in general because the objective functions (4c) and
(4d) are not identical; that is, the objective functions are neither the same nor multiples

123

Common labor market, attachment and spillovers in a large

of each other. Hence, the governments face different incentives and each government
makes choices that do not fully account for the benefits produced by the provision of
their public goods to residents of the other jurisdiction.
It is now important to note that the subgame-perfect equilibrium for the decentralized leadership game remains the same even if the governments are unable to
anticipate the migration responses to their choices. To see this, assume that the jurisdictional governments and the center take the population distribution {n 1 , n 2 } as given
when they make their choices. In the second stage, the center chooses transfers in
order to maximize welfare subject to (2b). Its optimal choices yield u 1x = u 2x as in
the problem solved by the social planner under similar circumstances in the previous
section. Hence, x1 = x2 = x and n 1 = n 2 = N /2. Then, we get Eq. (4a) and the
jurisdictional payoff functions (4b). The first-order conditions in the first stage yield
Eqs. (2m) and (2n). In sum:
Proposition 4 In the presence of a common labor market, the centers optimal redistributive policy equates marginal utilities of income across jurisdictions whether or
not the center is able to anticipate the migration responses to its redistributive policy.
As an immediate consequence, the jurisdictions find it optimal to make choices that
internalize all externalities in the first stage. The subgame- perfect equilibrium for the
decentralized leadership game corresponds to the symmetric Pareto optimal allocation whether or not the governments are able to anticipate the migration responses to
their policy choices.
Proposition 4 is important because it reveals that the efficiency of the subgameperfect equilibrium for the decentralized leadership game does not depend on the
timing of the migration equilibrium. Again, this fact follows from the common labor
market structure and the ex post income redistribution policy implemented by the
metropolitan government.
2.4 Interjurisdictional commuting costs
The previous analysis ignored commuting costs. In this subsection, we continue assuming that workers who reside and work in the same jurisdiction do not face commuting
costs. This is in line with the standard assumption of the home-attachment literature
that workers do not face intra-jurisdictional commuting costs. However, we now allow
for interjurisdictional commuting costs. We demonstrate that the efficient results of the
decentralized leadership game under a common labor market hold in the presence of
interjurisdictional commuting costs if there is a jurisdictional governments program
to reimburse workers for unpaid traveling expenses.10 Henceforth, we assume that the
governments take the jurisdictional populations as given.
Remember that z1 >>z2 and M1 M2 . The total quantity of labor demanded in
jurisdiction 1, L 1 (w) M1l 1 (w), is always larger than the total quantity demanded
10 Norway, for example, allows commuters from other European Economic Area countries to deduct commuting costs in their income taxes. See https://www.skatteetaten.no/en/
International-pages/If-you-work-in-Norway-you-need-to/Norwegian-employer/Norwegian-employer/
Brochures/Information-for-foreign-employees-Deductions-for-commuters/?chapter=6650.

123

E. C. D. Silva, V. M. Lucas

of labor in jurisdiction 2, L 2 (w) M2 l 2 (w), for a common market wage, w. The


Inada conditions guarantee that L 2 (w) > 0. If, in equilibrium, n 1 = n 2 , as in
the previous section, (some) workers in jurisdiction 2 must supply labor in jurisdiction 1. Assume that all workers who reside in jurisdiction 1 supply work in this
jurisdiction and that each resident of jurisdiction 2 supplies a fraction of labor in
jurisdiction 1.11 Suppose that the commuting cost faced by the representative resident of region 2 per unit of labor supplied in jurisdiction 1 is equal to c > 0.
Hence, this individuals commuting cost is c. Since workers who reside in jurisdiction 2 do not incur commuting costs when they supply labor in jurisdiction 2,
the representative resident in jurisdiction 2 is indifferent between working in jurisdiction 2 and jurisdiction 1 if and only if he/she receives a compensation equal to
c.
Suppose that the jurisdictional government 2 reimburses each of its jurisdictions
residents/workers with an
 to c for the commuting cost. The total cost of
 amount equal
the program is n 2 c = L 1 (w) n 1 c. The governments budget balance condition


in jurisdiction 2 is now n 2 t2 = G 2 + q2 + L 1 (w) n 1 c.
2

The
resident
  yields x (n 2 , 2 , G 2 , q2 , w) =

 budget constraint for the representative
w+ M2 2 (w)+2 G 2 q2 L 1 (w) n 1 c /n 2 once (1e) and government 2s
1

budget balance condition are taken into account. In region 1, x (n 1 , 1 , G 1 , q1 , w) =




j
w + M1 1 (w) + 1 G 1 q1 /n 1 , as before. Plugging x (.), j = 1, 2, into
the social welfare function and maximizing the implied expression with respect to
{1 , 2 } subject to 1 + 2 = 0 yields u 1x = u 2x in the second stage of the game.
  1
 
2  j

Hence, x (G 1 , G 2 , q1 , q2 ) =

q
X

L
N

N
/2
(w)
(w)
j
j
j=1
is the per-capita consumption of private good as function of the public good expenditures implied by the metropolitan governments optimal policy, where n 1 =
j maximizes
n 2 = N /2. In the first stage of the
 game, jurisdictional government


u x (G 1 , G 2 , q1 , q2 ) , G 1 , G 2 , Q by choosing nonnegative G j , q j subject to
Q =  (q1 , q2 ), taking the choices of the other jurisdictional government as given.
This implies that each jurisdictional government makes choices that fully internalize externalities. Their choices yield the Samuelson conditions (2l) and (2m), where
n 1 = n 2 = N /2. In sum:
Proposition 5 Under a common labor market and in the presence of interjurisdictional commuting costs, the subgame perfect equilibrium for the decentralized
leadership game in which the governments take the jurisdictional populations as given
is socially optimal.
The reimbursement program in jurisdiction 2 provides incentives to the workers
who reside in this jurisdiction to supply labor in jurisdiction 1. Even though jurisdictional government 2 finances the cost of the program through jurisdictional taxes, the
optimal redistribution policy adopted by the metropolitan government redistributes
11 This assumption is for expositional convenience. Assuming that some workers who reside in jurisdiction
2 work full time in region 1, while others work full time in jurisdiction 2 does not modify the qualitative
results of this section.

123

Common labor market, attachment and spillovers in a large

this cost as well as the extra income that jurisdiction 1 produces. Private consumptions levels are equated across jurisdictions. In sum, interjurisdictional commuting
costs do not affect the jurisdictional governments incentives to internalize externalities.12

3 Large economy under a common labor market


Consider now a large economy with J > 2 jurisdictions. For the most part, the model
is the same as before except that j = 1, . . . , J . In particular, we assume that: (i)
Each jurisdiction has a jurisdictional government; (ii) each jurisdictional government
is utilitarian and cares about the welfare of its residents only; and (iii) there is a central
government in the metropolitan area, which is utilitarian and cares about the welfare
of all residents in the economy. To facilitate the exposition, we assume that there are
no commuting costs.
Firms compete for labor in a common labor market. The individuals who work in
a particular jurisdiction may reside in any jurisdiction of the economy. Let w denote
the wage in the labor market. The representative
  firm in jurisdiction j chooses an
amount of labor input, lm j , to maximize f j lm j wlm j . The first-order condition is
j
flm = w, m = 1, . . . , M j , j = 1, . . . , J . Thus, lm j (w) = l j (w), m = 1, . . . , M j
and j = 1, . . . , J . All firms within a jurisdiction demand the same quantity of labor.
The clearing condition for the labor market is
J

j=1

M j l j (w) =

L j (w) = N ,

(5a)

j=1


where N = Jj=1 n j . Equation (5a) enables us to define the market-clearing wage as
a function of the vector of the numbers of jurisdictional firms, total labor force and the
maximum of labor units each worker is able to offer in the market, w = w (M, N ),
for M = (M1 , . . . , M J ).
The representative consumer in jurisdiction j derives utility from consumption
of x j units of numeraire good, G k units of a public good provided in jurisdiction
k, k = 1, . . . , J , and Q units of a public good that is produced from contributions
from all jurisdictions, Q =  (q), where q = (q1 , . . . , q J ) and  (.) is a continuous,
differentiable concave transformation of jurisdictional contributions, qk , k =
 1, . . . , J,
and increasing in all arguments. This consumers utility function is u x j , G, Q ,
j = 1, . . . , J , where G = (G 1 , . . . , G J ). The properties of this utility function are
the same as before.
We now extend the attachment model to J > 2 jurisdictions. In doing this, we
assume that any individual is free to establish his/her residence in any jurisdiction, but
derives attachment benefits from establishing residence in two jurisdictions only. These
attachment benefits are exogenously assigned to each individual by nature (prior to the
12 Clearly, if the metropolitan government is unable to implement income transfers, incentive equivalence

fails and the jurisdictions do not have incentives to internalize externalities. As a result, jurisdiction 2 may
not find it desirable to adopt the reimbursement program.

123

E. C. D. Silva, V. M. Lucas

beginning of the game examined here). Although any individual selects his/her jurisdiction of residence from a menu containing J jurisdictions (namely, all jurisdictions
in the economy), in equilibrium, he/she will prefer to reside in one of the jurisdictions
in which he/she derives attachment benefits because the total utility enjoyed by any
individual in at least one of these jurisdictions will be higher than the total utility
that he/she enjoys in a jurisdiction in which he/she does not derive any attachment
benefit.
Assume that nature partitions the entire population, N , into population groups,
Nr,s > 0, where r = 1, . . . , J 1 and s = r + 1, . . . , J denote jurisdictions in
 1
the economy and rJ=1
Nr,r +1 = N . Individuals assigned to a particular population
group derive attachment benefits from residing in either of the two jurisdictions that
characterize the group. Consider the individuals who
are assigned to population group

N1,h , h = 2, . . . , J .13 An individual n (1,h) 0, N1,h gets utility u (x1 , G, Q) +


a N1,h n (1,h) from residing in jurisdiction 1 and utility u (x h , G, Q) + an (1,h)
from residing in jurisdiction h. In equilibrium, we have


u (x1 , G, Q) + a N1,h n 1,h = u (x h , G, Q) + an 1,h , h = 2, . . . , J,

(5b)

where n 1,h is the individual who is indifferent between residing in jurisdiction 1 and
residing in jurisdiction h, h = 2, . . . , J . Letting n h,1 N1,h n 1,h , h = 2, . . . , J ,
Eq. (5b) can be rewritten as follows:

u (x1 , G, Q) + an h,1 = u (x h , G, Q) + an 1,h , h = 2, . . . , J.

(5c)

Now, consider the individuals who are assigned by nature to population group Nr,h for


where r {2, . . . , J 1} and h r + 1, . . . , J . Let
a pair of jurisdictions r and h,


r,h
n
0, Nr,h denote an individual who is attached to either jurisdiction r or h and
 

r,h
who gets utility u (xr , G, Q) + a Nr,h n
from residing in jurisdiction r and


 
r,h

Letting n denote the individual


from residing in jurisdiction h.
r,h
who is indifferent between residing in jurisdiction r or h and letting n h,r
Nr,h n r,h ,
we can write the following migration equilibrium equations:

u x h , G, Q +an




u (xr , G, Q) + an h,r
= u x h , G, Q + an r,h , r = 2, . . . , J 1, h = r + 1, . . . , J.
(5d)

13 By symmetry, N
h,1 = N1,h , h = 2, . . . , J .

123

Common labor market, attachment and spillovers in a large

The migration equilibrium is characterized by Eqs. (5c) and (5d).14 Letting n j denote
J
the population in jurisdiction j, j = 1, . . . , J , we have n 1 =
n 1,h , nr =
J
h=2
J

nr,1 + h=
j=1 n j .
r +1 n r,h , r = 2, . . . , J 1, h = r + 1, . . . , J and N =
3.1 Decentralized leadership for a large economy under a common labor market
Consider a setting where the large economy features decentralized
leadership. The
 J  n 1,h
1
payoff for the government in jurisdiction 1 is S = h=2 0 [u (x1 , G, Q) + a



N1,h n 1,h dn 1,h . This can be simplified to yield
S = n 1 u (x1 , G, Q) + a
1

 J


n 1,h N1,h

h=2


n 1,h 

.
2

The payoff for the government in jurisdiction r, r = 2, . . . , J 1, is

J



n
r,h
.
nr,h Nr,h
Sr = nr u (xr , G, Q) + a
2

(6a)

(6b)

h=r

The payoff for the metropolitan government is the sum of (6a) and (6b)

J
J





n j,h 
.
S=
n j,h N j,h
n j u x j , G, Q + a

(6c)

h= j

j=1

We can now derive the economy-wide resource constraint for the large economy. We
obtain this constraint by aggregating individual and governmental budget constraints
while also aggregating individual market incomes. The budget constraint facing an
individual who resides in jurisdiction j is again x j + t j = i j + j , j = 1, . . . , J .
Equations (1d) and (1f) that describe the budget constraint faced by the government
in jurisdiction j and the market income for an individual who resides in jurisdiction
j, respectively, hold in the large economy for j = 1, . . . , J . The interjurisdictional
income redistribution constraint faced by the metropolitan government is now
J

n j j = 0.

(6d)

j=1

14 As discussed in the text, every individual also considers the benefits of establishing residence in any of the

J 2 regions where he or she does not enjoy an attachment benefit. Hence, there is a larger number (indeed,
a continuum) of potential conditions that may characterize the migration equilibrium. However, as we will
show below, the full set of conditions that describe every individuals residential choice between a region
where he or she enjoys an attachment benefit and a region where he or she does not derive an attachment
benefit can be ignored. In equilibrium, the utility from residing in any region in which an individual does
not enjoy attachment benefit is strictly lower than the utility that this individual obtains from residing in the
preferred choice between the two regions in which he or she obtains attachment benefits.

123

E. C. D. Silva, V. M. Lucas

Hence, aggregating individual and government budget constraints and accounting for
Eq. (6d), we obtain
J



n j x j + G j + q j = I (M, N ) ,
(6e)
j=1

where I (M, N ) is the total income produced in the economy. It is again important to
note that the total income produced in the economy does not depend on the population
distribution across jurisdictions:
I (M, N ) =

Mj f



l j (w (M, N )) .

(6f)

j=1

Equation (6f) is an immediate implication of the common labor market, since all
workers earn the same wage in spite of the jurisdiction in which they work.
In what follows, we consider a decentralized leadership game in which the metropolitan government and the jurisdictions take migration decisions as given when they
make their choices. Taking migration decisions as given, jurisdictional and metropolitan governments play
a two-stage
game. Jurisdictional government j, j = 1, . . . , J ,

chooses nonnegative G j , q j to maximize its payoff function, taking the choices
of all other jurisdictional governments as given in the first stage. Having observed
{G, q}, where q = (q1 , . . . , q J ), the metropolitan government chooses {x1 , . . . , x J }
in the second stage to maximize (6c).
Consider the second stage. The centers first-order conditions yield
u 1x = u hx , h = 2, . . . , J,

(6g)

in addition to the economy-wide resource constraint, (6e). Conditions (6g) yield x1 =


x h for h = 2, . . . , J . Letting x = x1 = = x J , the economy-wide resource
constraint yields

x (G, Q) = I (M, N )

N.
Gj Q

(6h)

j=1

Consider now the first stage. Accounting for (6h), jurisdictional government 1 wishes
to maximize
 J



n 1,h 
n 1 u (x (G, Q) , G, Q) + a
n 1,h N1,h
.
(6i)
2
h=2

Also accounting for (6h), jurisdictional government r , r = 2, . . . , J wishes to maximize

J



n
r,h
.
(6j)
nr,h Nr,h
nr u (x (G, Q) , G, Q) + a
2
h=r

123

Common labor market, attachment and spillovers in a large

Taking migration decisions as given, the jurisdictional governments must also take
the jurisdictional population distribution as given. This implies that each jurisdictional government in the first stage makes choices to maximize u (x (G, Q) , G, Q), a
common objective function. Hence, they make choices that in the aggregate yield the
Samuelson conditions for efficient contributions to public goods. This fact yields the
following important result:
Proposition 6 In the presence of a common labor market and attachment for a large
economy, all jurisdictional governments find it desirable to provide contributions
to metropolitan public goods (with shared provision or not) that satisfy the Paretoefficient conditions for provision of such goods. The metropolitan government finds
it optimal to equalize private consumption levels across jurisdictions. In sum, the
subgame-perfect equilibrium for the decentralized leadership game is Pareto-efficient.
Proposition 6 has important policy implications. The working hypothesis that there
is a common labor market in the economy is easy to accept if workers who choose to
work in jurisdictions other than their jurisdictions of residence do not have to travel
long distances. Thus, for a fixed geographic area that constitutes the metropolitan area,
the larger the number of jurisdictions, the larger will be the commuting flows across
jurisdictions. In a context in which the economy is large and the existence of a common
labor market is very plausible, Proposition 6 informs us that attachment to jurisdictions
may not be an impediment for efficiency or equity. Metropolitan public goods may
be efficiently provided, and each individual in the economy may have access to the
same basket of private and public goods. Further, since the exogenously determined
population groups, Nr,s , are not necessarily symmetric, the efficient and equitable
allocation implied by Proposition 5 is quite consistent with an uneven population
distribution in equilibrium.
The fact that the subgame-perfect equilibrium for the decentralized leadership game
does not imply that the population distribution is symmetric can be demonstrated
with a simple example. Suppose that
 J = 3, N1,2 = 40, N1,3 = 40, N2,3 = 20.
=
N
+
N
In
equilibrium,
n
1,2
1,3 /2 = 40, n 2 = N1,2 + N2,3 /2 = 30, n 3 =
 1

N1,3 + N2,3 /2 = 30.
Note that no individual finds it optimal to move to a jurisdiction in which he or
she does not enjoy an attachment benefit. Since utilities of consumption are equalized in the subgame-perfect equilibrium, all individuals enjoy a net strictly positive
utility premium if they reside in their most preferred jurisdiction relative to all other
jurisdictions in which he or she does not derive attachment benefits.

4 Conclusion
In this paper, we extend the home-attachment model to a setting with more than two
jurisdictions and consider non-cooperative policy making for provision of different
types of federal public goods in the presence of a common labor market. Our model
is a hybrid of standard home-attachment and common-labor-market models in the
fiscal federalism literature. We show that in the presence of a common labor market, the optimal redistributive policy implemented by the central government always

123

E. C. D. Silva, V. M. Lucas

yields equalization of private consumption levels across jurisdictions. This result holds
despite the centers ability of anticipating migration responses to its income transfer
choices. Anticipating equalization of private consumption levels across jurisdictions,
all jurisdictional governments have incentives to make contributions at levels that fully
internalize externalities.
Our results have interesting policy implications. Unlike Wellisch (1994), we show
that home attachment is not necessarily an impediment for efficient behavior at the
jurisdictional government level. This finding is largely due to the existence of a common labor market and a central, metropolitan, government endowed with power to
implement income transfers across all jurisdictions. The presence of a common labor
market separates the decision of where one supplies labor from the decision of where
one establishes his/her residence. An individual can reside in a jurisdiction in which
he/she derives attachment benefits and work in some other jurisdiction in which
he/she does not derive any attachment benefit. This type of behavior seems to provide
an empirically supportable behavioral hypothesis for observable residential and job
choices in Canada and the USA, where there are significant commuting flows. One
also finds evidence for the existence of metropolitan governments, such as the Metro
Vancouver. The policy prescription here is to endow such governments with powers
to implement interjurisdictional income transfers.
One must recognize the limitations of the analysis. The model is deliberately built
after the model in Wellisch (1994) in order to show that its negative results can be
reversed once one accounts for a common labor market and interjurisdictional income
transfers promoted by a central authority. However, such a model negates the possibility that an individual obtains more utility from provision of a public good in his/her
jurisdiction than an individual who resides in a different jurisdiction. If one extends the
model in order to allow for this possibility and keeps all the other modeling assumptions intact, the income redistribution policy implemented by the center would not
equate private consumption across jurisdictions, in general. As a result, one would not
obtain incentive equivalence in general. Depending on the types of the consumption
spillovers, the center may be able to restore incentive equivalence, but only if it is
endowed with additional policy instruments (see, e.g., Silva and Yamaguchi 2010;
Caplan and Silva 2011).
Acknowledgments The authors thank the reviewer and editor for comments and suggestions that
improved the paper. Vander Lucas also thanks CNPq (Brazil) and FAP-DF (Brasilia) for financial support.

References
Agrawal, D. R. & Hoyt, W. H. (2014). State-tax differentials, cross-border commuting and commuting
times in multi-state metropolitan areas. CESifo working paper no. 4852.
Aoyama, N., & Silva, E. C. D. (2010). Equitable and efficient federal structures with decentralized leadership,
spillovers and attachment of heterogeneous labor. Journal of Public Economic Theory, 12, 323343.
Aoyama, N., & Silva, E. C. D. (2013). Regional cohesion maintenance, spillovers and imperfect mobility.
Finanz Archiv/Public Finance Analysis, 70, 116127.
Boadway, R., Song, Z., & Tremblay, J.-F. (2013). Non-cooperative pollution control in an inter-jurisdictional
setting. Regional Science and Urban Economics, 43, 783796.

123

Common labor market, attachment and spillovers in a large


Brueckner, J. K., & Kim, H.-A. (2003). Urban sprawl and the property tax. International Tax and Public
Finance, 10, 523.
Brueckner, J. K., & Helsley, R. W. (2011). Sprawl and blight. Journal of Urban Economics, 69, 205213.
Caplan, A. J., Cornes, R. C., & Silva, E. C. D. (2000). Pure public goods and income redistribution in a
federation with decentralized leadership and imperfect labor mobility. Journal of Public Economics,
77, 265284.
Caplan, A. J., & Silva, E. C. D. (2011). Impure public goods, matching grant rates and income redistribution
in a federation with decentralized leadership and imperfect labor mobility. International Tax and Public
Finance, 18, 322336.
Cornes, R. C. (1993). Dyke maintenance and other stories: Some neglected types of public goods. Quarterly
Journal of Economics, 108, 259271.
Mansoorian, A., & Myers, G. M. (1993). Attachment to home and efficient purchases of population in a
fiscal externality economy. Journal of Public Economics, 52, 117132.
Rothstein, P., & Hoover, G. (2006). Group welfare and the formation of a common labor market: Some
global results. International Tax and Public Finance, 13, 323.
Silva, E. C. D. (2014). Selective decentralized leadership. Journal of Urban Economics, 83, 15.
Silva, E. C. D., & Yamaguchi, C. (2010). Interregional competition, spillovers, and attachment in a federation. Journal of Urban Economics, 67, 219225.
US Census Bureau. (2011). Out-of-state and long commutes, Table 7. American community survey reports.
Wellisch, D. (1994). Interregional spillovers in the presence of perfect and imperfect household mobility.
Journal of Public Economics, 55, 167184.
Wellisch, D., & Wildasin, D. E. (1996). Decentralized income redistribution and immigration. European
Economic Review, 40, 187217.
Wildasin, D. E. (1991). Income redistribution in a common labor market. American Economic Review, 81,
757774.
Wilson, J. D. (2007). The welfare state versus the common labor market: Which to dismantle? CESifo
Economic Studies, 53, 618636.

123

Das könnte Ihnen auch gefallen