Why Haven t Regonal Wages Converged? Jason L. Savng Federal Reserve Bank of Dallas Research Department Workng Paper 1711 https://do.org/10.24149/wp1711
Why Haven t Regonal Wages Converged? By Jason L. Savng Senor Research Economst and Advsor, Federal Reserve Bank of Dallas* December 18, 2017 Regonal wage convergence has long been predcted across the Unted States as barrers to factor moblty have fallen, yet there s lttle evdence (apart from a bref perod n the 1970s and 1980s) that convergence has actually occurred. Why not? I reexamne ths ssue by developng a model n whch fscal polcy dfferences across states endogenously mpact labor supply across jursdctons. I fnd that states whose safety nets are relatvely generous wll tend to drve out workers, rasng wages for those who reman whle also promptng net outmgraton to less generous states. Ths suggests that regonal wage convergence requres not only free factor moblty but also the coordnaton of fscal polcy across jursdctons. (JEL H73. Keywords: regonal wage convergence, fscal federalsm.) * The vews expressed here are those of the author and do not necessarly reflect those of the Federal Reserve Bank of Dallas or the Federal Reserve System. I thank Alex Rchter, Mchael Plante, John Duca, Leonard Nakamura, Mchael Wess, and semnar partcpants at the Federal Reserve Bank of Dallas. I also thank conference partcpants at the Southern Economc Assocaton conference n Tampa, Florda for constructve comments. Any remanng errors are my own. 1
Introducton As economc ntegraton proceeds n a federaton, one outcome predcted by socal scence research s the convergence of wages across jursdctons. Ths convergence occurs for several reasons ncludng the movement of captal, the mgraton of people, and often the greater prevalence of laws and regulatons from the center vs-à-vs member states. To be sure, persstent dfferences n culture, language, and other factors may slow ths convergence even wthn countres (say, comparng southern and northern Italy) but the expectaton s that convergence wll happen over a suffcently lengthy long run 1. One mght expect the Unted States to be among the federatons whose member state wages have made the most progress toward convergence. After all, the country has snce at least the Cvl War developed a common natonal dentty, wth a common currency and a generally spoken language, and has for the last half-century at least offered convenent nterstate hghways for exceptonally rapd movement across jursdctons. And at least untl the md-1980s, there was some reason to beleve these phenomena had coupled wth globalzaton and technologcal change to produce a powerful trend toward regonal wage convergence, both wthn the U.S. and between the U.S. and Mexco [Hanson 1996 and 1997, Robertson 2000]. Yet ths trend was not sustaned, ether n the Unted States [Phllps 1992] or across the nternatonal border [Gandolf et al 2014]. Explanatons provded for the lack of convergence 1 See Magrn (2004) for an excellent summary of the economcs of regonal convergence as well as some of the dffcultes n evaluatng whether such convergence has occurred. 2
nclude dfferng propenstes to mgrate across educaton and ncome levels [Mshra 2007], dfferentals n regonal exposure to world trade [Autor et al 2013], and resdual barrers to captal and good movements that may be larger than prevously thought [Robertson 2005]. Whle varyng somewhat n ther partculars, however, each of these explanatons assumes regonal wages would converge f only barrers to moblty could be removed and/or ndustral compostons costlessly reshuffled. In ths paper, I present an alternatve explanaton for the lack of wage convergence that focuses on dfferences n regonal fscal polcy. I take as my startng pont the assumpton that the factors of producton are costlessly moble across jursdcton, abstractng away n partcular from the barrers to labor moblty that are sad to mpede convergence. I then ask whether one would expect regonal wage convergence to occur. If t can be shown that convergence would occur n the absence of barrers, then polcymakers who wsh to pursue convergence mght reasonably engage n measures to mtgate those barrers, whether they be more generous permt systems for cross-border mgraton, larger fscal transfers that would rase (lower) after-tax wages n poorer (rcher) jursdctons, or fewer labor-market structures lke employer-provded health care that lock n workers to partcular jobs n partcular jursdctons. On the other hand, f convergence would not occur even n a world wthout barrers or frctons, then current efforts (extensve n some countres/regons) to brng about convergence by fosterng factor moblty may perhaps need to be rethought. Beyond the relevance to these partcular programs, the queston of regonal wage convergence bears on the effcent dvson of responsbltes across levels of government. The fscal federalsm lterature begun by Oates [1973] outlnes a wde swathe of government servces that are best provded at the state or local level, though there are sgnfcant emprcal ssues wth 3
how to properly categorze them and warnngs that mproper categorzaton could swamp any potentals gans from decentralzaton [Prud homme 1995]. Balanced aganst ths wth others extendng the analyss or even suggestng a robust federal system can help reduce overall government spendng n a federaton [Rodden 2003] or even preserve a market economy wthn the federaton as a whole [Wengast 1995, Qan and Wengast 1997]. Whle ths paper does not drectly address these ssues, a fndng that regonal wage convergence s affected by fscal decentralzaton would suggest some level of cauton regardng that decentralzaton may be warranted. The man respect n whch fscal polcy wll dffer n my model s the safety-net level jursdctons select. When ndvduals are free to mgrate across states n response to dfferentals n wages and welfare benefts, states that offer generous welfare benefts wll attract recpents and repel workers. 2 Ths phenomenon wll n turn rase prevalng wages n hghbeneft jursdctons whle reducng them n low-beneft jursdctons, ensurng regonal wage convergence won t and can t occur n a world where jursdctons can pursue ther own ndependent fscal polces. How Are Beneft Levels Related to Wages? Puttng ths model n context requres frst dscussng how tradtonal work vewed the nterplay between benefts and wages. Tradtonal labor and publc fnance models began wth 2 There s an extensve emprcal lterature on the extent to whch welfare beneft levels nduce recpent mgraton, a thess most famously formulated n Peterson and Rom [1989,1990]. Baley [2005] provdes an excellent summary of the evdence to date, concludng such mgraton exsts but s not a domnant factor n moblty decsons. 4
the proposton that ndvduals are altrustc wth decreasng margnal utlty of ncome [e.g. Gramlch and Laren 1984]. Accordng to ths explanaton, areas wth hgher prevalng wages would tend to offer more generous safety nets, not because they are more compassonate but smply because they have a greater avalablty of funds. Smlarly, states wth lower prevalng wages would tend to offer sparser socal protectons even f equally altrustc. Alternatvely, economsts such as Varan [1990] vewed state safety net levels as stemmng manly from the desre for socal nsurance. Accordng to ths lne of reasonng, hgh-wage workers are more wllng to part wth any gven amount of ncome, and can therefore be expected to gve more to the poor, manly as a hedge aganst the possblty of nvoluntary unemployment. Ether way, the relatonshp between benefts and wages would be straghtforward, wth hgher-wage jursdctons offerng more generous safety nets because hgher-wage ndvduals are more nclned to provde them [Mofftt et al 1998]. Ths nsght led n turn to the natural expectaton that, as wages converged, electorates would naturally wsh to rase ther state safety nets as well, wth welfare beneft levels ultmately equalzng across jursdctons n lne wth wage convergence. Insghts garnered from the chartable contrbuton lterature were key to ths expectaton, showng for example that hgh-wage workers make larger chartable contrbutons to the poor [Feldsten and Clotfelter 1976, Steurle 1987]. Wth ever-greater economc ntegraton n the US fosterng factor moblty, t would only be a matter of tme before then-poor jursdctons mproved ther safety nets as resdents of those states saw mprovements n ther wages. For ths analyss to hold, t must be the case that preferences over beneft levels are an ncreasng functon of wages. But do hgher-wage ndvduals actually favor more generous safety nets than ther lower-wage counterparts? Whle t s true that they make larger chartable 5
contrbutons, there are mportant ways n whch chartable contrbutons dffer from government-run welfare programs. Frst, those who contrbute to charty can choose the set of people to whom they gve and the amount gven, whle those who Acontrbute@ to the government can do nether. More mportantly, ndvduals bear the full cost of ther chartable contrbutons but only a small porton of the cost of government welfare programs, whch suggests that beneft-level preferences may be determned as much by one=s wllngness to redstrbute the ncome of others as by the margnal utlty of one=s own ncome. Fnally, the presence of progressve taxaton substantally ncreases the margnal cost of welfare beneft provson for hgh-wage workers whle reducng or even elmnatng t for low-wage workers. Taken together, then, t s not clear a pror that one should expect to see hgh-wage ndvduals favorng relatvely generous socal nsurance programs. Indeed, survey evdence from the Unted States s nconsstent wth ths expectaton. For example, a survey by the Pew Research Center found that 59% of ndvduals n the lowest ncome quartle favored greater expendtures on welfare recpents versus whereas only 29% of ndvduals n the hghest ncome quartle who expressed smlar vews. An NPR/Kaser/Kennedy School survey reached qualtatvely smlar fndngs, though wth a smaller gap. 3 Whle the avalable evdence does not defntely fnd a negatve relatonshp between ncome and one s preferred beneft level, t s certanly not consstent wth a postve relatonshp. To be sure, these and other surveys ask mperfectly worded questons and do not carefully dstngush between varous classes of welfare programs, so no sngle survey should be taken as defntve evdence on ths pont. But the overall weght of the evdence cuts aganst the 3 Ths survey was taken n 2001 and provdes no data on the vews of upper-ncome Amercans. 6
argument that hgh-wage ndvduals naturally support hgher beneft levels, whch I take as a startng pont for the work that follows. If wages do not drve beneft levels, though, how can the observed relatonshp between benefts and wages be explaned? In the model gven here, generous safety nets endogenously cause hgh wages through the channel of mgraton. More specfcally, when ndvduals are free to mgrate across states n response to dfferentals n wages and welfare benefts, states that offer generous welfare benefts wll attract recpents and repel workers. 4 Ths mgraton of workers rases the prevalng wage n hgh-beneft states and lowers the prevalng wage n low-beneft states, consstent wth the emprcally observed postve relatonshp between safety nets and wages n the Unted States. And the nature and drecton of the causal relatonshp turns out to have mportant mplcatons for regonal, and natonal, fscal polcy decsons. Model and Results Interjursdctonal models whch permt mgraton often assume that ndvduals dffer n some exogenous earnngs parameter and mgrate n response to dfferentals n the prce of housng [e.g. Epple and Romer 1991]. More recent work utlzes a neo-rcardan framework n whch each ndvdual receves both an endogenously determned wage and some welfare beneft level offered by hs jursdcton of resdence and mgrates n order to maxmze ther sum [e.g. Wldasn 1991]. Each of these models presents problems n an attempt to explan the wage- 4 There s an extensve emprcal lterature on the extent to whch welfare beneft levels nduce recpent mgraton, a thess most famously formulated n Peterson and Rom [1989,1990]. Baley [2005] provdes an excellent summary of the evdence to date, concludng such mgraton exsts but s not a domnant factor n moblty decsons. The mpact on workers s largely unexplored at ths pont. 7
beneft correlaton n the Unted States, where labor markets are relatvely flexble and relatvely few ndvduals receve both benefts and wages. In ths secton, I develop a model that combnes the wage flexblty of Amercan labor markets wth wage-welfare exclusvty. When wages and welfare beneft levels dffer across jursdctons, ndvduals have an ncentve to Avote wth ther feet@ and mgrate toward jursdctons that offer hgher wages or more generous welfare benefts. Abstractng away from any costs assocated wth mgraton, ndvduals wll move untl nterjursdctonal dfferences n well-beng dsappear. 5 In order for ths to occur, there must be some mechansm through whch the net ncomes of workers and recpents s congestble. 6 In the case of workers, decreasng margnal productvty of labor provdes a natural avenue for ths congeston to operate: wages fall as the number of workers rse. In the case of recpents, the answer s less clear because welfare beneft levels are set by statute and do not respond (at least not drectly) to changes n the recpent populaton. 7 For ths reason, labor markets cannot be the sole means by whch ndvdual well-beng s equlbrated across jursdctons. I shall assume n ths paper that both recpents and workers value locatonal amentes as well as welfare benefts and wages. It s well known that amentes affect the mgraton decsons of ndvduals [Rosen 1979, Graves and Lnneman 1979]. In fact, emprcal work suggests that 5 In general, the qualtatve conclusons of ths paper hold when moblty s costly, although nterjursdctonal dfferentals smaller than the cost of moblty would persst n equlbrum. 6 If ths were not the case, the system would equlbrate at a "corner soluton" n whch each type of moble ndvdual would resde n exactly one jursdcton. Such a system would offer scant theoretcal or emprcal nsght. 7 It mght stll be possble that poltcans respond to such mgraton by alterng welfare beneft levels, but the emprcal evdence for ths s at best mxed [Mofftt 1992]. 8
regonal wage dfferentals are largely a functon of amenty dfferentals [Roback 1982]. Whle some amentes are ndependent of populaton sze, most are n some way subject to congeston effects. Examples nclude beaches, publc parks, and even the crme rate, whch ncreases wth ncreasng populaton densty. As long as locatonal amentes are at least at the margn congestble, they can provde the means by whch recpent net ncome equlbrates across jursdctons. Assume two types of ndvduals n the model, workers and recpents, who are costlessly moble across a fxed number of jursdctons. 8 Let each jursdcton be endowed wth a Rcardan producton functon f(e) for the numerare commodty, where I s the jursdcton of nterest and e s the number of ndvduals who are employed n the jursdcton. 9 Ths producton functon s monotoncally ncreasng and concave n the number of workers. Each jursdcton s also endowed wth an amentes functon a(u+e), where u s the number of ndvduals who resde n the jursdcton but are unemployed. The amentes functon s assumed to be, at least at the margn, decreasng n ts argument. Workers are endowed wth one unt of labor whch must be used to produce the numerare commodty. Each worker receves from hs employer the margnal value of labor n producton of the numerare, and no resdent who s able to work may be excluded from employment. Recpents, on the other hand, receve subsdes from ther local jursdcton nstead 8 Although I assume that all recpents and workers are moble, such an assumpton s not crucal for the analyss. It should be noted that, when some ndvduals are mmoble, the mplcaton of changes n welfare polcy may dffer consderably for moble and mmoble ndvduals. 9 The producton functon need not dffer across jursdctons. However, the model s completely general wth regard to employer human captal and jursdctonal technology, both of whch could generate nterjursdctonal dfferences n the producton functon. 9
of wages from an employer. These subsdes are dentcal across all recpents n a gven jursdcton, and no unemployed resdent may be excluded from the subsdy. For the sake of smplcty, suppose that subsdes are provded by taxaton of fxed captal, the owners of whch clam all resdual profts and fnance all redstrbuton. Further, suppose that each type of moble ndvdual (recpents and workers) attempt to maxmze ther net ncome. Let the benefts pad to recpents from a gven jursdcton be denoted b and assume competton so that the wage n a gven jursdcton s w f e workers and recpents respectvely n jursdcton I s gven by. Then the net ncome of Y w a u e E Y b a u e U (1) Snce both workers and recpents are costlessly moble across jursdctons, equlbrum net ncome for both must be dentcal across all jursdctons so that, w a u e w a u e j j j j b a u e b a u e j j j j, j (2) Denote the equlbrum level of net ncome for workers and recpents respectvely as Y E and Y U. Fnally, let M be the total number of moble ndvduals n the system. Snce these ndvduals are dvded nto two mutually exclusve types, t must be the case that and u e M (3) u e U E (4) 10
Now, denote the equlbrum net ncome dfferental between workers and recpents by Δ. Snce the equlbrum values of Y E and Y U are constant across jursdctons, the equlbrum value of Δ s also constant across jursdctons and s gven by E U Y Y w b w b, j (5) j j A rearrangement of ths equaton yelds the relatonshp wk=δ+bk œk between welfare beneft levels and wages. To explore whether wages and benefts are related, suppose that jursdcton I chooses to ncrease ts subsdy to welfare recpents. An ncrease n redstrbuton by jursdcton I ncreases the net ncome of recpents n jursdcton I. It also creates an ncome dfferental between recpents n jursdcton I and recpents n all other jursdctons, whch n turn provdes opportuntes for mgratonal arbtrage. Solvng mplctly for employment n a gven jursdcton whch mples e 0,. yelds e 1 E U w Y Y b Implctly dfferentatng (4) wth respect to b, we have E b e b k. Snce E s fxed, ek 0 b e e k k 0 b E U E U Y Y b Y Y bk 0 e e k k b b E U Y Y 0 e e b 11
Rearrangng terms, we obtan E U Y Y e b e k 0. (6) Ths dervatve, whch I shall denote by the symbol, measures the margnal mpact of a change n redstrbuton on the dstrbuton of ncome and s always negatve. Snce 0,, the mgraton of workers nduced by a change n b s gven by e b e b j 1 e 0 e 0 j (7) These equatons demonstrate the spllover effect for workers: addtonal redstrbuton on the part of jursdcton reduces the number of workers n jursdcton but rases the number of workers n all other jursdctons. Recpent effects may be obtaned n a smlar manner. Snce Y U b a u e must be the case thatu 1 U e a Y b dfferentatng wth respect to b yelds Snce U and E are fxed,, t. Substtutng ths equaton nto (3) and mplctly b U E a b 1 k 12
0 a b a 1 k 1 1 k k U U 1 1 1 a k k b b U Y b Y Y a b k a 1 1 ak a Therefore, t must be the case that the change n net ncome for recpents s gven by Y b U a 1 a 1 k 0. (8) Ths dervatve, whch I shall denote by the symbol π, measures the margnal mpact of a change n redstrbuton on the net ncome of recpents and s always postve. Snce 0,, the mgraton of recpents nduced by a change n b s gven by u b u j b 1 1 e 0 1 a a e 0 1 j j (9) These equatons demonstrate the spllover effect for recpents: addtonal redstrbuton on the part of jursdcton rases the number of recpents n jursdcton but lowers the number of recpents n all other jursdctons. These results suffce to descrbe the comparatve statcs assocated wth wages and benefts. In partcular, when a jursdcton chooses to offer a more generous level of welfare benefts, workers have an ncentve to mgrate toward other jursdctons. Ths mgraton of workers ncreases the scarcty of labor n the jursdcton of nterest and decreases the scarcty of 13
labor n all other jursdctons, whch n turn rases the prevalng wage n the jursdcton of nterest and lowers the prevalng wage n all other jursdctons. In equlbrum, the dfference between benefts and wages wll be exactly the same n every jursdcton, wth ncreased wages n a partcular jursdcton always accompaned by ncreased welfare benefts n that jursdcton. The mplcatons for wage convergence are both stark and mmedate. When states n a federaton are allowed to mantan dfferent levels of socal protecton, wages would not be expected to converge ether n the short or the long run. Moreover, n contrast to suggestons that greater factor moblty s needed to facltate wage convergence, such a development would actually nhbt wage convergence n ths framework because workers would then be better able to escape, and recpents better able to seek, hgh beneft jursdctons. Only when states offer the same level of socal protecton would wages be expected to converge across member states. Implcatons and Concluson At ts core, ths paper renterprets the relatonshp between safety nets and wages wth the member states of a federaton. Prevous work suggests that hgh-wage ndvduals wsh to gve more to the poor and that ths desre naturally leads hgh-beneft states to offer generous publc support for ther poor. But ths work mplctly assumes that ndvduals vew publcly provded welfare benefts as a natural extenson of ther prvate chartable contrbutons B and government revenue as a natural extenson of ther own pocketbooks. Because socal nsurance programs dffer from prvate charty n mportant ways, t need not be the case that ndvduals who make large chartable contrbutons wll necessarly also wsh to see a relatvely generous publcly 14
provded welfare system, whch s nconsstent wth a flow of causaton from jursdctonal wages to beneft levels. Ths paper provdes an alternatve explanaton n whch wages are endogenously determned, wth generous safety nets repellng workers and thereby rasng wages n those jursdctons as labor scarcty rases workers margnal products. Smlarly, low welfare beneft levels nduce worker nmgraton whch reduces wages n those jursdctons. In equlbrum, then, jursdctons that offer hgh (low) beneft levels can expect hgher (lower) wages to preval n ther jursdctons. Once that s understood, the mplcatons for regonal wage convergence become clear. In general, one would not expect regonal wage convergence to preval n a federaton whether or not there are sgnfcant barrers to factor moblty. For ths reason, polces to facltate factor moblty whether they take the form of mproved transportaton networks, broader access to fnancal markets, or greater transferablty of benefts from one job to the next would not be expected to brng about wage convergence even though those polces mght well have postve mplcatons for economc growth. And f regonal wage convergence dd preval across a federaton, then any change to safety-net polcy by a member state would dsrupt the convergence unless and untl that change to fscal polcy were rescnded. Smply put, regonal wage convergence cannot occur when ndvduals are free to mgrate and jursdctons are free to offer dfferent safety-net levels to ther resdents. These results contan both good and bad news for proponents of regonal wage convergence. If t s desred that regons have the fscal autonomy to offer the safety-net levels of ther choosng, ths s feasble, but t must be accepted that wages wll dffer across those jursdctons and do so n a predctable manner, wth wages n New York Cty and San Francsco 15
(for example) exceedng those of Houston and Atlanta whle mgraton patterns favor the latter two ctes over the former. On the other hand, f t s desred that regonal wage convergence happen, ths too s feasble, but only through a cedng of fscal autonomy that prevents jursdctons from offerng more (or less) generous safety nets than ther counterparts elsewhere. The central fndngs of ths paper therefore te nto both the well-establshed lterature on fscal federalsm and current debates over the extent to whch ctes or states should be allowed to pursue autonomous fscal polces 10. Past work has generally found safety nets to best functon as a jont state-federal endeavor, whch balances nter-regon equty wth the desre for smaller areas to ndulge ther preferences for progressve/regressve fscal polces. If regonal choces affect overall wage convergence, however, then those choces can mtgate or exacerbate nequalty across regons and potentally thwart federal desres to a greater extent than s commonly realzed. The ablty of jursdctons to affect ther own prevalng wages also matters n federatons that condton nter-regonal transfer payments on wages. Programs such as Medcad n the Unted States, provncal equalzaton payments n Canada, and certan development ad n the European Unon are desgned n such a way that larger payments are made (or more generous matchng rates offered) to lower-wage jursdctons. To the extent wages are endogenous to a jursdcton s fscal polces, however, t s unclear that regonal transfer payments would properly accomplsh the desred objectve of helpng a regonal converge. One mplcaton of ths paper, then, s perhaps to prompt a reassessment of the level of government at whch safety-net polcy should be optmally set. Fnally, the results of ths paper bear on ssues such as the desre to fght socal 10 See Shah (2006) for an excellent overvew of these ssues. 16
dumpng by states that offer low safety nets and supposedly scoop up workers across the rest of the federaton n so dong. Solutons n the European Unon and elsewhere have generally taken the form of socal chapters desgned to lmt member states fscal flexblty n the belef that more smlar fscal polces across the federaton would reduce worker outflows (smlar to unfar tax competton agreements that seek to reduce captal outflows). Whle addressng the desrablty of such agreements s beyond the scope of ths paper, ths paper suggests they do correctly frame the fundamental tradeoff as one that pts fscal autonomy aganst the desre to staunch the flow of labor and captal from one jursdcton to another. 17
Bblography Autor, Davd H., Davd Dorn, and Gordon H. Hanson. 2013. The Chna Syndrome: Local Labor Market Effects of Import Competton n the Unted States. Amercan Economc Revew 103: 2121-68. Baley, Mchael A. 2005. AWelfare and the Multfaceted Decson to Move.@ Amercan Poltcal Scence Revew 99: 125-135. Epple, Denns, and Thomas Romer. 1991. AMoblty and Redstrbuton.@ Journal of Poltcal Economy 99: 828-858. Feldsten, Martn, and Charles T. Clotfelter. 1976. ATax Incentves and Chartable Contrbutons n the Unted States: A Mcroeconometrc Analyss.@ Journal of Publc Economcs 5: 27-56. Gandolf, Davd, Tmothy Hallday, and Raymond Robertson. 2014. Globalzaton and Wage Convergence: Mexco and the Unted States. IZA Dscusson Paper 8254, avalable through SSRN at https://ssrn.com/abstract=2460157. Gramlch, Edward M., and Deborah S. Laren. 1984. AMgraton and Income Redstrbuton Responsbltes.@ Journal of Human Resources 19: 489-511. Hanson, Gordon H. 1996. Localzaton Economes, Vertcal Organzaton, and Trade. Amercan Economc Revew 86: 1266-1278. Hanson, Gordon H. 1997. Increasng Returns, Trade, and the Regonal Structure of Wages. Economc Journal 107: 113-133. Luttmer, Erzo. 1999. AGroup Loyalty and the Taste for Redstrbuton.@ Jont Center for Poverty Research Workng Paper 61. Magrn, Stefano. 2004. Regonal (D)convergence. In Handbook of Regonal and Urban Economcs, edted by J. Vernon Henderson and Jacques-Francos Thsse, pp. 2741 2796. Amsterdam: Elsever. Mshra, Prach. 2007. Emgraton and wages n source countres: Evdence from Mexco. Journal of Development Economcs 82: 180-199. Mofftt, Robert, Davd Rbar, and Mark Wlhelm. 1998. AThe Declne of Welfare Benefts n the U.S.: The Role of Wage Inequalty.@ Journal of Publc Economcs 68: 421-452. Orr, Larry L. 1976. AIncome Transfers as a Publc Good: An Applcaton to AFDC.@ Amercan Economc Revew 66: 359-371. Peterson, Paul E., and Mark C. Rom. 1989. AAmercan Federalsm, Welfare Polcy, and 18
Resdental Choces.@ Amercan Poltcal Scence Revew 83: 711-728. Peterson, Paul E., and Mark C. Rom. 1990. Welfare Magnets: A New Case for a Natonal Standard. Washngton DC: The Brookngs Insttuton. Phllps, Keth R. 1992. Regonal Wage Dvergence and Natonal Wage Inequalty. Economc Revew Federal Reserve Bank of Dallas, Fourth Quarter: 31-40. Prud'homme, Remy. 1995. The Dangers of Decentralzaton. World Bank Research Observer 10: 201 220. Qan, Yngy and Barry R. Wengast. 1997. Federalsm as a Commtment to Preservng Market Incentves. Journal of Economc Perspectves 11: 83 92. Roback, Jennfer. 1982. AWages, Rents, and the Qualty of Lfe.@ Journal of Poltcal Economy 90: 1257-1278. Robertson, Raymond. 2000. Wage Shocks and North Amercan Labor-Market Integraton. Amercan Economc Revew 90: 742-764. Robertson, Raymond. 2005. Has NAFTA Increased Labor Market Integraton between the Unted States and Mexco? The World Bank Economc Revew 19: 425-448. Rodden, Jonathan. 2003. Revvng Levathan: Fscal Federalsm and the Growth of Government. Internatonal Organzaton 57: 695 729. Rosen, Sherwn. 1979. AWage-Based Indexes of Urban Qualty of Lfe.@ In Current Issues n Urban Economcs, edted by Peter Meszkowsk and Mahlon Straszhem. Baltmore, MD: Johns Hopkns Press. Shah, A. 2006. A Practtoner's Gude to Intergovernmental Fscal Tansfers. In Intergovernmental Fscal Transfers: Prncples and Practce, edted by Robn Boadway and A. Shah, pp 1-53. Washngton, DC: The World Bank. Steuerle, C. Eugene. 1987. AChartable Gvng Patterns of the Wealthy.@ In T. Odendahl, ed., Amerca=s Wealthy and the Future of Foundatons. New York: The Foundaton Center. Tebout, Charles M. 1956. AA Pure Theory of Local Expendtures.@ Journal of Poltcal Economy 64: 416-424. Varan, Hal. 1980. ARedstrbutve Taxaton as Socal Insurance.@ Journal of Publc Economcs 14: 49-68. Wengast, Barry R. 1995. The Economc Role of Poltcal Insttutons: Market-Preservng Federalsm and Economc Development. Journal of Law, Economcs, and Organzaton 11: 1 31. 19
Wldasn, Davd E. 1991. AIncome Redstrbuton n a Common Labor Market.@ Amercan Economc Revew 81: 757-774. Wldasn, Davd E. 1992. ARelaxaton of Barrers to Factor Moblty and Income Redstrbuton.@ Publc Fnance/Fnance Publques 47: 216-230. 20