public goods, taxes, and takings
TRANSCRIPT
International Review of Law and Economics 28 (2008) 287–293
Contents lists available at ScienceDirect
International Review of Law and Economics
Public goods, taxes, and takings
Thomas J. MiceliDepartment of Economics, University of Connecticut, Storrs, CT 06269-1063, United States
a r t i c l e i n f o
JEL classification:H41K11R52
Keywords:
a b s t r a c t
Blume, Rubinfeld, and Shapiro [Blume, L., Rubinfeld, D., & Shapiro, P. (1984). The taking of law: Whenshould compensation be paid? Quarterly Journal of Economics, 99, 71–92] first showed that compensationfor takings can lead to a moral hazard problem that results in overinvestment in land suitable for publicuse. To the contrary, this paper shows that when compensation is financed by a proportional property tax,the compensation rule is irrelevant regarding the level of investment landowners make in their property,
Compensation for takingsEminent domainMoral hazardP
as well as the amount of land they authorize the government to acquire, both of which will be efficient.Intuitively, landowners recognize the equivalence of taxes and takings in budgetary terms, causing thedistortionary effects of compensation and property taxation to cancel each other out through the balanced
1
(rotamgi
qipbgltlac
e
aaK
aiEt“stpptctop
iirla
0d
ublic goods budget condition.
. Introduction
Since the seminal paper by Blume, Rubinfeld, and Shapiro (BRS)1984), economic models of eminent domain have struggled toeconcile their “no-compensation result,” which is a consequencef the moral hazard problem associated with compensation, withhe constitutional requirement of “just compensation.” A frequentrgument in favor of compensation is the claim that the govern-ent will take too much land if it is not forced to pay for it. A
overnment that acts in this way is sometimes said to have “fiscalllusion.”
A noteworthy example of this approach to the compensationuestion is the paper by Fischel and Shapiro (1989),1 in whichndividuals, acting from behind a veil of ignorance, choose the com-ensation rule without knowing how many, or which, parcels wille taken for public use. In this setting, the authors show that, if theovernment is assumed to maximize the welfare of the majority ofandowners (rather than overall welfare), then partial compensa-ion is optimal, reflecting the trade-off between the tendency forandowners to overinvest under full compensation (moral hazard),
nd the risk of excessive government seizure of land under zeroompensation (fiscal illusion).A different tack in the search for an optimal compensation rulemerges from Richard Epstein’s controversial assertion that taxes
E-mail address: [email protected] Other papers in this vein include Miceli and Segerson (1994), Hermalin (1995),
nd Nosal (2001). Another argument for compensation is to provide insurancegainst takings risk to risk averse landowners. See Blume and Rubinfeld (1984) andaplow (1986).
wdsbfr“
ahf
144-8188/$ – see front matter © 2008 Elsevier Inc. All rights reserved.oi:10.1016/j.irle.2008.07.007
© 2008 Elsevier Inc. All rights reserved.
re themselves a form of taking in the sense that they, like tak-ngs, represent coercive transfers of resources to the government.pstein acknowledges, however, that if the tax revenue is usedo provide a public good, then the resulting benefits representin-kind” compensation of the taxpayer, thus satisfying the con-titutional requirement (Epstein, 1985, p. 196). It follows, though,hat any taxpayers who do not receive benefits in line with their taxayments are in some sense “undercompensated.” Fischel (1995a,p. 210–211) has observed that this undercompensation parallelshat suffered by actual takings victims who receive market valueompensation for their land (the legal definition of just compensa-ion), which almost certainly falls short of the minimum amountwners would ask in a consensual sale (i.e. their true reservationrices).
The parallel between these two forms of undercompensations further revealed by the public budget, which requires that anyncrease in compensation for the benefit of takings victims rep-esents a cost that must be borne by taxpayers. This insight hased Fischel to conjecture that market value compensation emergeds the legal definition of just compensation because constitutionriters, acting from behind a veil of ignorance, recognized theirual roles as taxpayers and potential takings victims, and thereforeought a way of “balancing the undercompensation of the market-ased rule against the greater loss incurred by higher taxes andoregoing public works that full, consensual compensation wouldequire” (Fischel, 1995a, p. 211). We will refer to this claim as the
Epstein-Fischel conjecture.”The purpose of this paper is to evaluate these competingpproaches to the compensation question—namely, the moralazard-fiscal illusion trade-off embodied by the Blume, Rubin-
eld, and Shapiro model, and the Epstein-Fischel conjecture—in a
2 Law a
chtrtc
aetssmotitctstitat
spttiS
2
(lstsatqGoffw
shiatv
tra
ata
tmwthi
jSr(btdncdB
eCucnsop
vcpiv
t
wic
t
Itt
i
U
L
U
W
88 T.J. Miceli / International Review of
oherent framework that simultaneously examines the choice ofow much land will taken for public use, and how much capi-al landowners will invest in their land, given the compensationule. Since both decisions can be couched in constitutional-choiceerms, we adopt the Fischel and Shapiro framework with two crucialhanges.
First, we assume that individuals’ valuations of the public goodnd reservation prices for their land are unobservable to the gov-rnment. This rules out the use of benefit (or Lindahl) taxes forhe public good, and also requires the use of an objective mea-ure of compensation for takings (for example, fair market value orome fixed fraction thereof). Second, we assume that the govern-ent finances compensation with a property tax that is assessed
n the market value of a given property, as opposed to a lump sumax. While this would seem to introduce an additional source ofnefficiency into the model (given that a proportional tax is distor-ionary while a lump sum tax is not), this turns out not to be thease. In fact, we show that under certain conditions, the propertyax distortion exactly offsets the moral hazard effect of compen-ation, thereby producing a kind of irrelevance result regardinghe compensation rule. Specifically, landowners make the efficientnvestment choice, and also allocate the efficient amount of lando public use (public choice considerations aside), regardless of themount of compensation. These conclusions are in sharp contrasto the existing literature.
The remainder of the paper is organized as follows. Section 2ets up the basic model. Section 3 examines the impact of the com-ensation rule on the investment choice of landowners. Section 4hen asks how the compensation rule affects the level of takingshat landowners will authorize when acting from behind a veil ofgnorance regarding which parcels will actually be taken. Finally,ection 5 offers conclusions and draws implications for takings law.
. The model
The basic model follows that developed by Fischel and ShapiroFS) (1989). Consider a government that will provide a public goodike a park or highway requiring the assembly of land. In particular,uppose there are a total of n parcels of land, s < n of which will beaken at random for purposes of providing the public good.2 (Forimplicity, we assume that land is the sole input.) Thus, the prob-bility that any given parcel will be taken is given by p = s/n, whilehe probability that it will not be taken is 1 − p = (n − s)/n. Let theuantity of the public good be G(s), where G′ > 0. We assume thatis a local public good in the sense that, once it is provided, it is
nly available to residents whose land is not taken and who there-ore remain in the community.3 Let vi(G) be individual i’s valuationunction of G, where v′
i> 0, v′ ′
i < 0. To conserve on notation, weill henceforth write vi(s) instead of vi(G(s)).
We assume that the v′is are unobservable to the government,
o i’s contribution to the cost of G cannot depend on his orer valuation. This rules out of any form of benefit tax. We also
gnore demand-revealing tax schemes as impractical. Instead, wessume that the government imposes a uniform property tax onhe assessed value of each parcel. Let A represents this assessedalue. Thus, the tax payment of each individual whose land is not
2 The fact that the parcels are taken randomly suggests that they are scatteredhroughout the jurisdiction, which seems inconsistent with the idea of assembly. Theandomness assumption is valid, however, if the location of the project is randomnd is made public, along with all parcels to be taken, at a single point in time.3 This assumption is inessential for our results. Fischel and Shapiro (1989) actually
ssume that the public good is pure in that even residents whose land is taken receivehe benefits. We note the effect of this alternative specification in footnotes whereppropriate.
s
E
Fen
aic
nd Economics 28 (2008) 287–293
aken is tA, where t is the tax rate to be determined below. Thisethod of assessing taxes represents a critical distinction from FS,ho assumed that taxes were lump sum rather than depending on
he actual assessed value of the property. The method of taxationere is more reflective of actual practice, but also has important
mplications to be discussed below.In contrast to the assessed value, we assume that the true (sub-
ective) values of the parcels are unobserved by the government.pecifically, let V be the value each owner attaches to his land. Thisepresents the amount that he would sell it for in a consensualmarket) transaction, where we assume that V ≥ A. The distinctionetween the fair market value of a piece of land and its subjec-ive value to the owner is crucial because it is what makes eminentomain a non-consensual purchase. This distinction, however, isot often made in the economic literature on takings.4 To facilitateomparison with the existing literature, we initially make the stan-ard assumption that both V and A are the same for all landowners.elow, we examine the implications of relaxing this assumption.
Finally, let C(A) be the amount of compensation paid by the gov-rnment to each landowner whose land is taken. We do not writeas a function of V because, as noted, the owner’s true valuation isnobservable to the government and hence cannot form a basis forompensation. The only other restrictions we place on C are that it ison-negative (C ≥ 0) and non-decreasing in A (C′ ≥ 0). An importantpecial case is C(A) = A, which reflects the traditional legal definitionf just compensation as being equal to the fair market value of thearcel (which we assume is the same as its assessed value).
A crucial element of the model is the public budget, which pro-ides the link between the tax assessed on untaken parcels and theompensation paid to owners of taken parcels. Assuming that theublic good, G, is the only public expenditure, that the land input
s the only cost of G, and that all landowners have equal assessedalues, we can write the balanced budget condition as
(n − s)A = sC(A), (1)
here the left-hand side is total revenue and the right-hand sides total cost. Using the definition of p above, we can rewrite thisondition as
= p
1 − p
C(A)A
. (2)
t follows that if C(A) = A (fair market value compensation),= p/(1 − p), while, at the other extreme, if C(A) = 0 (zero compensa-ion), t = 0.
Given this specification, we can write the realized utility of anndividual whose land is not taken as
Ni = V − tA + vi(s). (3)
ikewise, the realized utility of an individual whose land is taken is
Ti = C(A). (4)
eighting these expressions by the appropriate probabilities andumming yields expected utility:
Ui = pUTi + (1 − p)UN
i = pC(A) + (1 − p)(V − tA) + (1 − p)vi(s). (5)
inally, note that the utility of all individuals in the absence of gov-rnment action is U0
i= V for all i. That is, there is no risk of a taking,
o tax liability, but also no public good.5
4 Exceptions include Knetsch and Borcherding (1979) and Fischel (1995b).5 Of course, such a world does not exist since the risk of a taking is pervasive. Inmulti-jurisdiction setting, however, there may be a differential risk, for example
f some states or local governments are more prone to take or regulate land. In thisase, the land market presumably would capitalize this spatially varying risk into
Law a
3
apctp
lAiia
E
w
tlilwpaa
w(
(
D
tfi
[
NTrBtFr
p
lisoocrbi
rb
wsp
l
cvs
PColo
tziaelairbftp
clalFdtrtm
3
ntr
T.J. Miceli / International Review of
. Landowner investment and moral hazard
Since the original article by Blume et al. (1984), the economicnalysis of compensation for takings has focused on how the com-ensation rule affects landowner incentives to make irreversibleapital investments in their property. In this setting, BRS derivedhe famous no-compensation result based on the moral hazardroblem associated with full compensation.
We introduce landowner investment into the above model byetting the assessed and true valuations of landowner i be given by(xi) and V(xi), respectively, where xi, is the dollar amount that i
nvests in capital improvements. We assume that both A and V arencreasing, strictly concave functions of x, and that V′(x) ≥ A′(x) forll x. In this case, expected utility in (5) becomes
Ui = pC(A(xi)) + (1 − p)[V(xi) − tA(xi)] + (1 − p)vi(s) − xi, (6)
here all other variables are defined as above.As usual, the timing of the landowners’ investments relative to
he government’s taking decision is crucial. BRS and all subsequentiterature in this area have assumed that landowners make theirnvestment choices before the government decides which particu-ar parcels to take, and those investments then become sunk. Thus,
hen landowners make their choices, they each face the samerobability of a taking, p. Consequently, their optimal, as well asctual, levels of investment will be the same, given equal valuationnd assessment functions.
In this setting we first consider the socially optimal x,hich maximizes the expected net value of a given property,
1 − p)V(x) − x. The first-order condition is
1 − p)V ′(x) = 1. (7)
enote the resulting level of investment x*.The actual choice of xi by landowner i will instead maximize (6),
aking as given the tax rate and compensation rule.6 This yields therst-order condition7
pC ′ − (1 − p)t] A′(xi) + (1 − p)V ′(xi) = 1. (8)
ote that (8) differs from (7) by the first term on the left-hand side.he question is, under what conditions will this extra term vanish,esulting in the efficient level of investment? Note first that in theRS model, this extra term consists only of the C′ term, which yields
he result that C′ = 0 for efficiency.8 The same result emerges in theS model since they treat landowner i’s tax payment as lump sumather than as being assessed on A(xi).In evaluating the extra term in (8), we follow FS and let the com-ensation rule takes the form C = ˛A, where ˛ is a non-negative
and prices, thereby providing a kind of implicit compensation for purchasers of landn high-risk jurisdictions (Fischel & Shapiro, 1988, pp. 287–289). On a related point,ee Glaeser, Gyourko, and Saks (2005), who consider zoning regulations when devel-pers have an outside option, and Ghosh (1997), who incorporates the possibilityf exit into the FS (1989) model. Land market capitalization would not, however,ompensate landowners for the common background risk of a taking that they faceegardless of where they locate. The model in the current paper focuses on this risky implicitly assuming that the probability of a taking is the same everywhere. Thats, there is no “exit” option.
6 The assumption that the tax rate is fixed from the landowner’s perspectiveeflects the large number of properties in the jurisdiction. Thus, no single ownerelieves he can affect t by his choice of xi .7 In the case where G is a pure public good, the vi(s) term in (6) would not beeighted by 1–p. Since s is treated as a parameter, however, this term does not
how up in (8). Thus, neither the optimal nor the actual choices of xi depend on thearticular assumption regarding G.8 The no-compensation result therefore really says that compensation must be
ump sum (that is, C′ = 0), a special case of which is C = 0.
uoeerm
a
rvpo
ih
haTpa
nd Economics 28 (2008) 287–293 289
onstant.9 That is, compensation is proportional to the assessedalue of the property. We can now state the main result of thisection.
roposition 1. Assume that the compensation rule takes the form(A) = ˛A, and that taxes are uniformly assessed on the market valuef individual properties. Then there exists a symmetric Nash equi-ibrium in which all landowners choose the socially efficient levelf investment regardless of the value of ˛.10
The proof is provided in Appendix A. It follows from this resulthat awarding market value compensation for takings (˛ = 1), orero compensation (˛ = 0), will both induce efficient investment. Its important to emphasize that this result is not a consequence ofny particular assumption about the objective function of the gov-rnment. Rather, it reflects the trade-off arising from the fact thatandowners are simultaneously taxpayers and potential victims of
taking and thus automatically internalize the marginal cost ofnvesting in their land through the public budget constraint.11 As aesult, the moral hazard problem vanishes, and with it, the trade-offetween moral hazard and fiscal illusion. Compensation can there-ore be determined solely for the purpose of preventing excessiveakings, assigning takings risk optimally, or any other purpose. Weursue this point further in Section 5.
Finally, consider the role of the property tax in deriving thisonclusion. Note in particular that because the first term on theeft-hand side of (8) equals zero in equilibrium, the property taxnd compensation distortions exactly offset with respect to theandowner’s choice of x. The reason that this did not occur inS is that they modeled the tax as lump sum, and hence, non-istortionary.12 Thus, only the compensation distortion (the C’erm) remained. Here, in contrast, the tax, like the compensationule, is proportional to the market value of the land, thus leadingo offsetting distortions, which, by the balanced budget condition,
ust be exactly equal.13
.1. The impact of heterogeneous properties
To this point, we have assumed that all properties are homoge-eous in order to highlight the contribution of the current analysiso the existing literature. In this section, we note the impact ofelaxing this assumption. Specifically, let Vi(xi) be owner i’s val-ation function, which differs across owners. The primary effectf this change is that landowners will now generally invest differ-nt amounts in their land, resulting in different assessed values inquilibrium. (This will be true even if the assessment function, A(·),
emains the same for all properties, given that the realized assess-ents, A(xi), will vary.) With differing assessments, the balanced9 Note that ˛ is not restricted to be less than or equal to one. Thus, it allowsbove-market compensation (see the discussion in Section 5).10 We do not investigate the possibility of asymmetric Nash equilibria.11 Nosal’s (2001) model contains the same trade-off, but in contrast to the cur-ent model, he constrains the compensation rule to be equal to the average marketalue of all landowners’ properties, rather than depending on the actual value of theroperties that are taken. Thus, his main conclusion, which is a special case of thatbtained here, is that market value compensation is efficient in equilibrium.12 Specifically, the individual tax payments depended on the equilibrium level ofnvestment by all landowners, rather than on the amount each owner invested inis or her own property, as is assumed here.13 Hamilton (1975) derived a similar result in a very different context. In particular,e showed that in the Tiebout model of local public good provision, property taxesre not distortionary if communities are homogeneous in terms of taxable property.he reason is that, in a homogeneous community, the public budget transforms theroperty tax into a benefit tax for public goods. As a result, no free riding is possible,nd consequently, landowners’ choices of housing are not distorted by the tax.
2 Law a
b
t
wp
pAtiuchfiltb
t
wi
t
wtd
ccr
E
wrt(
IAbgmi
Ntaaa
ldemT
t
w
iiltthgp
dulttati
Pvi
poplrig1fdce
4
fitmWh
ttras
(1 − p)(v′ − t′A) = p′[V + v − tA − C(A)]. (14)
90 T.J. Miceli / International Review of
udget condition is no longer given by (1), but instead is∑j ∈ N
Aj =∑j ∈ T
Cj, (9)
here N is the set of untaken parcels, and T is the set of takenarcels. (To conserve on notation, we let Ai ≡ A(xi) and Ci ≡ C(A(xi)).)
To illustrate the impact of this change, suppose there are threeroperties in the jurisdiction with assessed values of A = {A1, A2,3}, and suppose that one property is taken at random.14 Thus,here are three possible outcomes, depending on which propertys taken. Consider the impact of this taking risk on the expectedtility of owner 1. If his property is the one taken, he will receiveompensation of ˛A1 and he pays no taxes. If, on the other hand,is land is not taken, he will pay taxes in proportion to A1 to helpnance compensation for the taking victim, but his expected tax
iability will differ, depending on which of the other two parcels isaken. To see why, note that if owner 2’s land is taken, the balancedudget condition is
(A1 + A3) = ˛A2, (10)
hich implies that t = ˛A2/(A1 + A3). Alternatively, if owner 3’s lands taken, the balanced budget condition is
(A1 + A2) = ˛A3, (11)
hich implies that t = ˛A3/(A1 + A2). The tax rates differ in thewo cases because the constituents of the two sets, N and T, areifferent.15
From owner 1’s perspective at the time he chooses x1, each par-el is subject to the same risk of being taken (i.e. p = 1/3). Thus,ombining the above cases, we can write his expected utility (cor-esponding to (5)) as
U1 =(
13
)(˛A1) −
[(13
)(˛A2
A1 + A3
)+
(13
)(˛A3
A1 + A2
)]A1
+(
23
)(V1 + v1(s)), (12)
here the term in square brackets is the “effective” tax rate. Theesulting “extra term” in the first-order condition for x1, reflectinghe offsetting compensation and tax distortions, is16
13
)˛
[1 −
(A2
A1 + A3+ A3
A1 + A2
)]A′
1(x1). (13)
f all properties had equal assessed values, then2/(A1 + A3) = A3/(A1 + A2) = 1/2 and the term vanishes as requiredy Proposition 1. However, it should be clear that this will notenerally be true with heterogeneous properties. Instead, the termay be positive or negative, depending on the situation of the
ndividual landowner.To illustrate, consider a landowner with a low property value.
ote that he will face a relatively high tax rate when his land is not
aken because inclusion of his property in set N lowers the aver-ge tax base, while exclusion of his property from set T raises theverage cost of compensation. Thus, he will expect to pay more onverage in taxes when his land is not taken than he will receive14 For simplicity, we continue to assume that parcels are taken randomly, regard-ess of the level of investment. If instead the government conditioned its takingecision partially on the amount owners had invested in their land (favoring, forxample, those with less investment), then the model would become substantiallyore complicated as owners would take this factor into account when choosing x.
he basic conclusions, however, would not be affected.15 In contrast, if all properties had the same assessed value, the tax rate would be= ˛/2 in both (all) cases.16 I assume, as above, that the landowner treats the computed tax rate as givenhen choosing x1, even though in reality, his choice will affect the tax base.
Nb
othb
i(Ha
nd Economics 28 (2008) 287–293
n compensation when it is taken. As a result, the bracketed termn (13) will tend to be negative, causing him to underinvest in hisand. The reverse is true for high-value property owners, who willend to overinvest because they expect to pay less in taxes thanhey receive in compensation. In effect, low-value owners subsidizeigh-value owners through the public budget, though the aggre-ate effect across all owners must be zero. A numerical example isrovided in Appendix B.
Another way to say this is that, while the compensation and taxistortions must cancel in aggregate, they do not cancel for individ-al owners. Consequently, the irrelevance result of Proposition 1 no
onger holds as some landowners underinvest (those for whom theax distortion dominates) and some overinvest (those for whomhe compensation distortion dominates). Efficient investment byll landowners can only be achieved if ˛ = 0 – that is, if compensa-ion is zero – for in that case, there is no public budget to distortnvestment decisions. Thus, we have
roposition 2. When landowners differ in their individual landaluations, the only compensation rule that induces efficientnvestment by all landowners is ˛ = 0, or zero compensation.
Two final points are worth making. First, in contrast to BRS,ositive, lump sum compensation will not achieve the efficientutcome in the current setting because the distortion arising fromroportional taxation would remain (unless taxes were also made
ump sum, as in FS). Thus, Proposition 2 is a true no-compensationesult. Second, when compensation is positive, the extent of thenefficiency will decrease as the jurisdiction becomes more homo-eneous. Thus, in a stratified Hamilton-Tiebout world (Hamilton,975), the irrelevance result of Proposition 1 remains largely validor intra-jurisdictional takings. However, in heterogeneous juris-ictions only zero compensation eliminates moral hazard, in whichase the possibility of excessive government takings (fiscal illusion)merges as a serious threat. We turn to that issue in the next section.
. The choice of public spending
To this point, we have treated the level of the public good asxed in order to focus on the landowner’s investment choice. Inhis section, we consider the choice of s, assuming that landowners
ake this decision before the specific parcels to be taken are known.hile this assumption may or may not be realistic,17 it provides a
euristic device for evaluating the Epstein-Fischel conjecture.Given that landowners choose s from behind a veil of ignorance,
hey are all in the same initial position and will therefore chooseheir desired level of s to maximize (5),18 taking the compensationule as given. (They will, however, recognize the dependence of pnd t on s.) Thus, taking the derivative of (5) with respect to s andetting it equal to zero yields19
i i
ote that the left-hand side represents individual i’s marginal netenefit from the public good. If there were no risk of taking, this
17 See the discussion by Fischel (1995a, pp. 204–205).18 In contrast, a majoritarian government would choose s to maximize the utilityf landowners whose land is not taken, given by (3). This is the assumption FS makeo derive their partial compensation rule. As we showed in the previous section,owever, when the moral hazard problem goes away, there is no longer a trade-offetween moral hazard and fiscal illusion.19 In this section, we treat x as fixed and the same for all owners, and hence suppresst notationally. Also note that if G were a pure public good, the left-hand side of14) would be [v′
i− (1 − p)t′A] and the right-hand side would be p′[V − tA − C(A)].
owever, nothing in what follows depends qualitatively on whether G is a local orpure public good.
Law a
tfplbe(
rwwiCas
tVc(topg
iiitp(p
v(rtdtf
(
Nddttsei
tt
ti
pteot(
Pmritt
ifeiiit“gler
5
hrt“aepmchlbmf
epeef(tc
b
T.J. Miceli / International Review of
erm would be set equal to zero to determine individual i’s demandor s. Since individuals differ in their marginal valuations but allay the same taxes (given equal A’s), this term will vary across
andowners. Thus, the equilibrium s would have to be determinedy majority vote (or some other method for aggregating prefer-nces), which may or may not result in the efficient level of sAtkinson & Stiglitz, 1980, pp. 507–509).
In terms of individual valuations, the resulting equilibrium couldesult in too much or too little s from i’s perspective. In the casehere it is too much (i.e. where v′
i< t′A), i overpays at the margin,
hich corresponds to Epstein’s case where the taxpayer receivesnsufficient in-kind compensation for his coercive tax payment.onversely, in the case where s is too small (i.e. v′
i> t′A), i underpays
nd thus, according to Epstein’s logic, receives too much compen-ation. (Individual i is therefore a free rider in this case.)
Now consider the right-hand side of (14), which representshe impact of the taking risk on i’s demand for s. Assuming that+ vi − tA − C(A) > 0, this term represents the expected under-
ompensation from a taking.20 Since p′ > 0, the right-hand side of14) is the expected marginal cost to landowner i of authorizing theaking of one more parcel in the form of an increased risk that hiswn property will be taken and insufficient compensation will beaid. The taking risk therefore reduces the demand for the publicood, all else equal.
The fact that the two sides of (14) are equal at the optimumllustrates the point that landowners, acting from behind a veil ofgnorance, will recognize the symmetry between taxes and tak-ngs, given their dual roles as taxpayers and potential victims of aaking.21 Note that the compensation rule is crucial here, since tax-ayers would prefer it to be low in order to reduce their tax liabilityi.e. it enters (14) through the t′ term), while takings victims wouldrefer it to be high to prevent undercompensation.
Now recall the Epstein-Fischel conjecture, which says that indi-iduals would choose market value compensation – that is, C(A) = Aor ˛ = 1) – as a way of balancing these offsetting interests. Whileecognizing that this amount would likely undercompensate vic-ims of a taking, landowners balance that concern against theiresire to limit their liability as taxpayers. To evaluate this conjec-ure, we use the balanced budget condition in (1) and (2), and theact that p = s/n, to rewrite (14) as
1 − p)v′i = V + vi
n. (15)
ote that, while this condition continues to reflect varyingemands for s across individuals, these demands are indepen-ent of the compensation rule. Instead, they reflect the individual’srue expected marginal benefit, and expected marginal cost, of aaking.22 Thus, any inefficiency in the determination of s will be
olely due to public choice factors (i.e. problems of aggregating pref-rences), rather than the amount of compensation or the mannern which it is calculated. We summarize this conclusion as follows.20 Note that this “undercompensation” incorporates any fiscal “surplus” or “deficit”hat landowner i would receive if his land were not taken. This reflects the assump-ion that G is a local public good.21 Another way to say this is that the two sides of (14) illustrate the interaction ofhe free rider problem (left-hand side) and the holdout problem (right-hand side)n determining the allocation of land to public use. See Cohen (1991).22 In particular, note that the left-hand side is the benefit of one more unit of theublic good, weighted by the probability of landowner i remaining in the jurisdic-ion, while the right-hand side is the expected cost of one more taking, which isqual to the opportunity cost of landowner i’s parcel, V, plus the foregone benefitf the public good, vi , weighted by the probability that i’s parcel will be the oneaken (1/n). Note that in the case of homogeneous valuations of the public good,15) becomes (n − s)v′ = V + v, which is a modified Samuelson condition.
fttlkaaTta
r
M1
nd Economics 28 (2008) 287–293 291
roposition 3. Assume that taxes are uniformly assessed on thearket value of individual properties and that landowners autho-
ize the number of parcels to be taken from behind a veil ofgnorance regarding which particular parcels will be taken. Thenhe number of parcels taken will be independent of the compensa-ion rule.
Although this result invalidates the Epstein-Fischel conjecturen the literal sense that it does not yield a determinate valueor C, it nevertheless confirms the basic insight from which itmerged—namely, that individuals, acting from behind a veil ofgnorance, would have accounted for both sides of the budget ledgern determining the proper measure of just compensation for tak-ngs. Further, this basic idea provides the common thread linkinghe analysis in this and the preceding sections: namely, that theequivalence” of taxes and takings as embodied in the public bud-et renders the compensation rule irrelevant, both with regard toandowner investment (at least for the case of homogeneous prop-rties), and the determination of public spending, under a wideange of conditions.
. Conclusions and implications for takings law
Since the seminal work by Blume et al. (1984), economistsave been concerned with the impact of the compensationule on land use incentives. Much of the subsequent litera-ure has attempted to provide counterarguments to their famousno-compensation result” by arguing that, despite the moral haz-rd problem associated with compensation, there are offsettingconomic justifications for the constitutional mandate of “just com-ensation.” The analysis in this paper has shown that this debateay have been unnecessary. In particular, using a constitutional-
hoice framework first developed by Fischel and Shapiro (1989), weave shown that the compensation rule is irrelevant regarding the
evel of investment landowners make in their property (which wille efficient), and the amount of land they authorize the govern-ent to take (which may be efficient, depending on public choice
actors).The intuition for this result is that landowners recognize the
quivalence of taxes and takings in budgetary terms, a notion firstroposed by Epstein (1985). In particular, they recognize that thexpected benefits of a more generous compensation rule will bexactly offset by the resulting increase in expected tax liability. Inormal terms, the distortionary effects of the compensation rulemoral hazard) and property taxation cancel each other out throughhe balanced budget condition. The idea is similar to the macroe-onomic concept of Ricardian equivalence.
How do these conclusions bear on actual takings law? Spurredy the BRS result and its aftermath, economists have primarilyocused on the moral hazard-fiscal illusion trade-off in answeringhe question of when compensation should be paid. Since it is set-led law that compensation is required for physical invasions,23 thisiterature is most applicable to regulatory takings cases, where theey question is whether compensation is ever due for governmentctions that reduce the value of a piece of property without actu-
lly taking title to it (Fischel, 1995a; Miceli & Segerson, 1994).24he results in this paper suggest that, under certain conditions,his debate is moot. That is, consideration of land use incentives,nd possibly the extent of government seizure (or regulation) of
23 See, for example, Loretto v. Teleprompter Manhattan CATV, 423 N.E.2d 320 (1981),eversed 458 U.S. 419 (1982).24 Key cases include Mugler v. Kansas (123 U.S. 623, 1887); Pennsylvania Coal v.ahon (260 U.S. 393, 1922); and Lucas v. South Carolina Coastal Council (505 U.S.
003, 1992).
2 Law a
lst
tibhtlofooic
cTitpjdeateCtdtstsw
A
r
A
t
t
wltl
C
Rp
a(
I
So
˛
TU
Ctar
A
hvptoettb
t
Tt1
t
TWyfi
Ttsc−Tv
R
92 T.J. Miceli / International Review of
and, can be largely ignored in determining when compensationhould be paid.25 Instead, emphasis can be shifted to factors otherhan efficiency, such as fairness or distributive justice.
This conclusion is in accord with Epstein’s (1985, p. 115) viewhat “the compensation requirement of the eminent domain clauses as much concerned with the distribution of gains and lossesetween persons as with their aggregate amount.” Epstein notes,owever, that compensation need not always be monetary to satisfyhis goal; it can also be “in-kind.” For example, a zoning or nuisanceaw that imposes a cost on a particular landowner by depriving himf certain harmful uses of his property simultaneously protects himrom similar harm caused by others. In this sense, the regulationffers reciprocal benefits that generally make all landowners betterff. This logic provides some justification of the historical unwill-ngness of courts to award compensation in most regulatory takingsases.
The reciprocal benefits argument, however, is less convincing forases where the benefits from the taking are narrowly distributed.his concern is especially pressing in public use (or “private tak-ngs”) cases, like Kelo v. New London,26 where the taken land isargeted for a redevelopment project whose benefits are largelyrivate.27 According to Epstein, the use of eminent domain is only
ustifiable in these situations if efforts are made to achieve a broaderistribution of the benefits. To illustrate his point, he uses thexample of the nineteenth century New Hampshire Mill Act, whichuthorized would-be mill builders to flood upstream land withouthe owner’s consent, provided that the mill builders paid the own-rs 150% of market value as compensation (Epstein, 1985, p. 174).learly the 50% premium was an effort to compensate owners forheir subjective values, and in so doing, to achieve a more equitableistribution of gains and losses from the taking. Although the par-icular amount of the premium is arbitrary, Epstein’s point is thatome amount above market value is called for. The conclusions inhis paper suggest that the debate about how much compensationhould be paid to landowners in private takings cases can proceedithout concern for the efficiency of the takings decision.
cknowledgements
I acknowledge the helpful comments of Avery Katz and twoeviewers.
ppendix A
Proof of Proposition 1Suppose that all landowners besides i choose x*. In the event
hat i’s land is not taken, he faces a tax rate equal to
= s˛A(x∗)(n − s − 1)A(x∗) + A(xi)
, (A1)
here the numerator is total compensation paid to owners whoseand is taken, and the denominator is the total assessed value ofhose properties (including i’s) that are not taken. In contrast, if i’s
and is taken, he will receive compensation equal to(A(xi)) = ˛A(xi). (A2)
25 But see Bernheim and Bagwell (1988), who offer a critique of models embodyingicardian equivalence, especially as regards the resulting irrelevance of governmentolicy.26 125 S.Ct. 2655, 545 U.S. 469 (2005). Also see Berman v. Parker, 348 U.S. 26 (1954)nd Poletown Neighborhood Council v. City of Detriot, 304 N.W.2d 455, 410 Mich. 6161981).27 On the economics of public use, see Merrill (1986) and Kelly (2006).
A
B
B
B
CE
F
nd Economics 28 (2008) 287–293
t follows from (A2) that
∂C
∂xi= C ′A′(xi) = ˛A′(xi). (A3)
ubstituting (A1) and (A3) into the first term on the left-hand sidef (8) yields
A′(xi)[
p − (1 − p)sA(x∗)
(n − s − 1)A(x∗) + A(xi)
](A4)
he question is under what conditions this extra term equals zero.sing p = s/n and 1 − p = (n − s)/n, we can rewrite (A4) as
˛sA′(xi)(n − s − 1)A(x∗) + A(xi)
[A(xi) − A(x∗)] (A5)
learly, xi = x* causes this term to vanish, in which case (8) reduceso (7). Thus, x* is i’s best response to xj = x* for all j /= i. Further,ssuming that EUi is strictly concave in xi, xi = x* is i’s unique bestesponse. Q.E.D.
ppendix B
This appendix presents a numerical illustration of theeterogeneous-property example in the text. Suppose the assessedalues of the three parcels are given by A = {$100, $200, $300}. Onearcel is taken at random, for which the owner receives compensa-ion equal to Ci = ˛Ai. Consider owner 1’s calculation of the impactf the takings risk on his expected wealth. If his land is taken, hexpects to receive compensation equal to $100˛ and he pays noaxes. Alternatively, if owner 2’s land is taken, owners 1 and 3 payaxes to finance compensation for 2 as determined by the balancedudget condition:
($100 + $300) = $200˛. (A6)
he resulting tax rate is t = ˛/2, which produces tax liability of($100) = $50˛ for owner 1. Finally, if owner 3’s land is taken, ownersand 2 pay taxes to finance compensation according to
($100 + $200) = $300˛. (A7)
he resulting tax rate is t = ˛, making owner 1’s tax liability $100˛.eighting each of the possible outcomes by 1/3 and summing
ields the expected net impact on 1’s wealth (corresponding to therst two terms on the right-hand side of (13)):
13
($100˛) − 13
($50˛) − 13
($100˛) = −$50˛
3. (A8)
he fact that this quantity is negative implies that owner 1 expectso pay more in taxes than he will receive in compensation as a con-equence of the taking risk. He therefore underinvests. A similaralculation shows that the impact on owner 2’s expected wealth is$40˛/3, while the impact on owner 3’s expected wealth is $90˛/3.hus, owner 2, like owner 1, underinvests, while owner 3 overin-ests. Note finally that the sum of the net effects is zero, as required.
eferences
tkinson, A., & Stiglitz, J. (1980). Lectures on public economics. New York: McGraw-Hill.
ernheim, B., & Bagwell, K. (1988). Is everything neutral? Journal of Political Economy,96, 308–338.
lume, L., & Rubinfeld, D. (1984). Compensation for takings: An economic analysis.California Law Review, 72, 569–628.
lume, L., Rubinfeld, D., & Shapiro, P. (1984). The taking of law: When should com-
pensation be paid? Quarterly Journal of Economics, 99, 71–92.ohen, L. (1991). Holdouts and free riders. Journal of Legal Studies, 20, 351–362.pstein, R. (1985). Takings: Private property and the power of eminent domain. Cam-
bridge, MA: Harvard University Press.ischel, W. (1995a). Regulatory takings: Law, economics, and politics. Cambridge, MA:
Harvard University Press.
Law a
F
F
F
G
G
H
H
K
K
K
T.J. Miceli / International Review of
ischel, W. (1995b). The offer/ask disparity and just compensation for takings: Aconstitutional choice approach. International Review of Law and Economics, 9,115–128.
ischel, W., & Shapiro, P. (1988). Takings, insurance, and Michelman: Comments oneconomic interpretations of ‘just compensation’ law. Journal of Legal Studies, 17,269–293.
ischel, W., & Shapiro, P. (1989). A constitutional choice model of compensation fortakings. International Review of Law and Economics, 9, 115–128.
hosh, S. (1997). Takings, the exit option and just compensation. International Reviewof Law and Economics, 17, 157–176.
laeser, E., Gyourko, J., & Saks, R. (2005). Why have housing prices gone up? AmericanEconomic Review, 95, 329–333.
amilton, B. (1975). Zoning and property taxes in a system of local governments.Urban Studies, 12, 205–211.
MM
N
nd Economics 28 (2008) 287–293 293
ermalin, B. (1995). An economic analysis of takings. Journal of Law, Economics, andOrganization, 11, 64–86.
aplow, L. (1986). An economic analysis of legal transitions. Harvard Law Review, 99,509–617.
elly, D. (2006). The “public use” requirement in eminent domain law: A ratio-nale based on secret purchases and private influence. Cornell Law Review, 92,1–65.
netsch, J., & Borcherding, T. (1979). Expropriation of private property and the basis
for compensation. University of Toronto Law Journal, 29, 237–252.errill, T. (1986). The economics of public use. Cornell Law Review, 72, 61–116.iceli, T., & Segerson, K. (1994). Regulatory takings: When should compensation be
paid? Journal of Legal Studies, 23, 749–776.osal, E. (2001). The taking of land: Market value compensation should be paid.
Journal of Public Economics, 82, 431–443.