The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
FEDERAL RESERVE RESERVE BANK BANK OF OF DALLAS fEDERAL Of FEDERAL Fourth Quarter 1992 • • conOllllC CVICW eVlew conOIlllC ConOInlC Do Interest In/erest Rates Rales Help Predict Inflation? Kenneth M, M. Emery and Evan F. Koenig Trade 7)'ade Policy 'l}'ade Polic)' and Intellectual Property Properly Protection: The 'l7Je North-South Nortb--SOutb Dispute William C C. Gruben Wilham Gruben Regional Wage u;(tge WClge Divergence and National Divergellce andNational Wage Inequality Keith R. Kerth R. A. Phillips Phillips This publication was digitized and made available by the Federal Reserve Bank of Dallas' Historical Library (FedHistory@dal.frb.org) Economic Review Federal Reserve Bank of Dallas Robe" O. MeT.." Jr. ,.......,cw~OIJar y...,J. Sat'l'll,io "'VGI~_OvI-",OI'IcIo H......, .......lum s-- i\cIJ~_'"'-.,....." W. Mic.... I CCllI Ib"'-_E.-~ Gt ,.ld P. O'Dtilcoll. Jr. Io'iw"-'- _~""" S1. phtn P. A. Bro_ ~ ~M_'" s..", It:cntwosI EcollOlllists Zsolt Becsi Roberl T CI,m JaMV Cuca KennelllM E~ RotJett W Gilmer David M Goold WIlIIaI'II C Gruben Joseph H Haslag EvanfKoeI'llll D'Ann M Peterwn Keith R PhIllips Fiona 0 S!gaIIa llrll Taylor John H Wetdl _AWymo Kev.n J Yeats Mine K YOc:eI Rn..n:h Assoc i.... l'I'ofHsor Nathan S Balte Southllm MeUtodisr tklMniry l'Iolessor ThomIs B Fomby Sou/hem Methodm Unrmslfy Professor &1ln Fraemall Umll8fSiryof rexas Professor Gregory W Hulfman Soolhem M8lhotiis/ University ProtesSCl Roy J Ruffin I.Inrversiry of Houston Professor PII'lQ Wang ~ Slate Unrverslty Editon: Rhonda Hams VW'gIll.a M Rogers ,t._ TIlt r - Rftww II putI/lsfIId by 1M ,...., ~ BanI: of DIIIIas P:III'euecI . . _ of IN 1IIl/'ICIf11nCl6G IIOll*tIUIiIy ..flllCt IN pcsIlIGlI al m. FecInI lies.- a.... oj DAs Of lilt ~ .. "-"'- s.u- ~.,,~_ot~ ~.-.:I~"~,,,...rupe. c:oDf ~ ........ _ _ _ dlqnD lilt I'IobIc: A.'Qn 0Iptnnt FedenI ~ Brid\ll!l.B. SIIbIln I:.~ T_1'S2'1l.UI 41!l22·52!>1 MIdIs NY lit ....... an ItwI UII'Iditlon 111M me _ os amm n '!he ~ "1J(MdId 'MIf'I. U1PI at TN lUlIallOIIlXIIUnI'o!Il!oe ~lIUed ~:en.I ~ Contents Page I Do interest Rates Help Predict Inflation? Kenneth M. Emery and Evan F. Koenig Accur<l.Ie forecasts of inflation are important \0 policymakers and \ 0 individuals who must make decis ions on the basis of expect:uions about the fut ure purchasing power of the dollar, Recent research on forecasting influion has shown that interest rates, by themselves, may provide useful infonnation about fUlUre inflation. In this article, Kenneth M. Emery and Evan F. Koenig investigate whether interest r.!.ICS contain infannation about future inflation beyond that fou nd in Irddilionai inflation-forecasting models. In other words, docs adding interest rates \0 traditional innation models e nhance the models' forecasting ability? Emery and Koenig find that incl ud ing imerest rates doe.'; Significantly improve the forecasting abili ty of trdditiona l models. They also find , in contrast (0 recent research focusing on the forecasting abiliry of interest rates in isolation , that the information content of interest rates d id not diminis h in the 1980s. Emery and Koenig point out thar whethe r the historical forecasting ability of interest ra tes can be e.xploited by policymakers is problematic. Be<7dUSe interest rates reflect expectations of future monetary policy, if the Fcderdl Reserve were to begin relying more o n interest rates as a guide to policy, the relationship between interest mtes and inflation wou ld likely change. Page 19 Trade Policy and inteilectu({I Property Protection: Tbe NOI'/b-Soulb Dispute Wi!tlam C Gruben News about foreign-made counterfeit products that range from pharmaceuticals to tennis shoes and about processes that have been copied without payment of royalt ies 10 patent holders has become common place. Many of these appropriations of intellectual propelty originate in developing countries, where intellectual propen y lawS and e nforcement have traditionally been less restrictive than in developed countries. 111ese d ifferences between the North-the developed countries thaI produce most intellectual property-and the South-the dcveloping countries that consume more than they producehave gener..lted much international friction. (Col1fl nued on rbe next page) Contents (Continu.ed/rom the Prwious page) 'frade Policy and In tellectllal Property Protection: 7be NOl'tl>-Soutb Dispute WIlliam c.Gruben Recently, however, some developing countries have begun to tighten these Jaws and their enforcement A few a nalysts have been quick \0 explain wh y ~ these countries are simply reacting to incre;lsingly tough U.S. pressu re. But lire they? Not all cou ntries unde r the most intense U.S. pressure have reacted to it, :t nd thcrc is a difference between those that have a nd those Ibm have not. The d ifference turns o ut to be these individua l countnes' own lmdc policies. Focusing primarily on the experiences or Latin American (nunlries, William C. Grulx-n shows how :mcl. why a nation's own trade policy innucnces Lts intclk-clual property Jaws and enforcement and w hy con\'~nt ional argument... about U.S. trade pressures may be only P'lrt of the Story. Gruben shows why developing countries that practice strong trode protectionism are mOliy.n ed toward wC'J k Inlellcctua l property protection, but thOSt: lhat have Hbcrolized lmdc omy fi nd strong inieUeclUa l propcny protection more attractive. Page 31 Regional Wage Divergence and National ~Ige Inequali(y KeiTh R Phillips Throughout much of the 1980s, W"Jge inequaHry incre-.lSed in the United States. Previous rescarch has found tha! a rise in earnings by educatio nal level and increased wage dispersion across occupatio ns wt"rt" im portant factors in Ihe rise in wage incqualiry. Researchers, however, have nuted Ihat much of the risc in wage inequality was left unexplai ned by Ihe demographic and industry factors that they examined . In this study. Keith R.. Phillips extends U1C analysis by examining the impact on wage inequality of a divergenc.:e in regi~1 wages that occurred during the 19BOs. The :wlhor finds th.·lt regional shocks, such as the recessions in the o il and f:mn belts. pushed wages lower in these below-average-wagt' area ... and incr~d.scd dcf('nsc spend ing pushed up wages in m:IOY l.tbovc-'I\'t!rage-wase areas. Wh ile these regiona l shocks i(lCrc;lM.'d overnll Mlge inequality. Phillips fi nds that region:tl wage di\'e~cnce a('COunted for onl y 2. J percent to 5 percent of the rt'iC in n:uio nal "\\'3ge inL'quality. Other factors, such as increased w-d.ge dispersion across eduC:lIjonal and IXCup:Hlonal groups :lOd :1 reduClJon in the male-female wag"-' g:lp. playoo larger rok·s. Kenneth M. Emery Evan F. Koenig Senior Economist Federal Reserve Bank 01Dallas Senior Economist and Pohcy Advisor Federal Reserve Bank 01 Dallas Do Interest Rates Help Predict Inflation? ht: primary responsd)ility of a <:~ntr;J1 bank is to prt:st:rve the va luc of its nation's currency. In (h ~ U nit~d States. thc Federal Ikse:rve System allc:mpls 10 mee t this rcsponsihility by pursuing 111onc:I:II)' policic:s IIlal prornOfe C:(.'( )o( )lllic gro~1 h without fueling inllati()[l . Achieving s ustained no ninll ;Hionary growth j.-; complic:llcd by the fact Ihal changes in mOllet;lry policy atTel'l the economy wilh : 1 hi}: Thus. the successful (:onduct of monetary poliq' requires rcli:lble foreGI.~ts of both infbtion and economic activity. A plot of Ihe threl'-month US. Treasury bill rate and subsequent qua neriy inll:llion suggests Ihal inlerc:!'>\ rales may tuntain infonllalion about future inll'ltion (Fip,lIIl! I ) Indeed. rt:ccnt research indit~ll ...s that lIloVemt'J\IS in imel1.:'st r.ltes are helpful in predicting innation (Ue rnanke 1990; Fama T Figure 1 Three-Month Treasury Bill Rate and CPI Inflation Percenl " " .. -- " " • • HWiRali • ,"',.... , ~, SOURCES OF PRIMARY OATA Board 01 Governors. Federal Reserve Sys:em; U S Sureau 01 LaI:lof Slal,SUCS Econ o mic Revie w - )' o u nh Qu ~ ne r 1?92 1990; Mishkin 1990a, 1990b; Frankel and Lown 1991). Interest mte movements may, therefore, pro- vide {he Fedcral l{eSCtve with information useful in cond ucting monel:IlY policy. In fact, some monet:uy policyma kers (johnson 1988) have :Iuvoca(ed a strategy for conducting policy b:lscd on monitoring scveral fin.mcial varia hies, induding interest r~Hes. While: the studies tiled above have addressed the question of wbetht'r interest rates, by themselves. hdp pn.:dic\ inflation, more relevant to policymaking is tbe qllestion of whether adding interest r..ues to tr'.Iditional models for forecasting inflation yields any improveme nt in the models' foreca~l inK ahility This ll nide examines whether intert.'SI r..ues provide information about innation beyond that containt."<.I in traditional models. For two re:ISOIlS we focus on inflation as measured by the consumer price index (CPO, First, existing studies or the predictive content of interest rates usc the consumer pricc index almost exclUSively. Secor1d. because they al~ not distoned by transitory shifts in the composition of output, flxcdweight indexes like the consume r price index give a more accumte picHire of ncar-te rm ch>lnges in Ihe COSI of living Ihan do variable-weight indexes. The first sect ion o f the :utide revicws the economic intuition undc:rlying thl.! connection between interest rates and inllation. Tht: second section pre~ents evidence of the power of interest lenes. by themselves. to pr(:ciict near-tenn inllation. Consistent with previous studies, we find Ihal although int... rest rates arc quite helpful in explaining movements in inll'llion during the 1960s and WSWlSh/O/tIankJohnV Oucs. OavrdM GouId. andJoseplt Ha~ tot hc/pIvi commcfIts and Mr.enne C Slack {Of /1 excel/em research assIstance Table 1 Inflation and Interest Rates MltrglNiI Significance Level DV 3M 3MFF IOY3M JOINT DV • • .000 .034 ,012 .000 3 A B. •• C. D. Number of Lags M 3M 733 .920 .059 296 3 .006 .000 .483 000 8 8 .146 .862 .671 .168 2 2 3MFF 10Y3M R' 6 5 899 618 5 _493 265 Sample periods A. 1959:1 - 1979;4 B_ 1980 1- 1991:2 C. 1959 1- 1979:4 o 1980 :1- 199 1:2 Definitions DV • dependent variable 10 the regression l't • change ioganthm 01 the consumer price Index. l!.n • change in l't 3M 3MFF lOY3M JOINT • • . . NOTE A ll Interest rate on three-month Treasury bills. dltterence between 3M and Ihe lederallunds rate. difference between the Interest rate oillell-year Treasury bonds and 3M ma/gmal Sl{lnlhcance level from a test 01 the hypotheSIS that the coelhcleflls 01 allinlerest rale variables are equal to zero. .ogressoons we re canled OUI usmg tile RATS regresSion pacto.age, V elSOOfl 3 0 1 9-0~, the l..";..ptm.tH1rr pO\\l'r ot l1 H l'rl'~t r; U l'~ lll:t rkl'dlr d ..:tl'r i()r.III.."~ 11) Ihl" 1 9H()~ T lll'thini ~ect i () n of the ,tn idc '1X'(:ilk, tl':ld illfln,1 1 inll:i l ioll fo n...--<::b tll1 g Ill{xkb ,tnd l 'X:lIll UK', \\ Ill'l l ll'r i ll('I(ld ing in tl:re., t 1: 11..:, i n IIll'''l' lllod ,,' I, n.:'Ulh III ;l(ldil iolla l e.x pb n:lloI"Y pOWL" f \'ill ' li nd th:lt ~ h {)rl Il Tlll i l1l l"fl"~1 1: l1 es do po.~Sl"s:, m aq,.:i n:d l-xpl:II1:llflrr j){I \\lT ill Ihl" prl.." 19HO sa m pll" I'UI"I IIl"ril \{lr..:. this i lll'l"l":I " l 'd e xpb nat( II"}' Ix)\\ ..: r d{x: .. I /( I! di... :tPPl":l f <h iring III ..: 19HO'... [ndel'<.l. \\'l" lind 111;11 lon).,:- tl'nll in tl"re."t r:lt e.... also hl '(;oml" ~ i ~n ili(,,;ln l d uri n).: tltl' 1')H{}o.. The fourth .......'i..:tion of Ihl' ;trlil'll" e:<.ulli nl"" 111<' rea l-linK' fo n':Chtin).: perfonll.I1l<:e of tIll,: r1\Il:o- t 1-l<'rll'r;11 of o ur inll;l\ lon nl(xk-b Fo u r H'r-.;on" o f tilt ... ).,:etlL"ral nuxld ;m: I,."on ..i<.il"n.:d -\,cr-.io,,:o- r1\o.:;m l 10 l"xpla in tile kH-1 o f int1:uion and t he t han}.:e in lnll:u lon. hoth \\'il h and \\ ilho\ll Ihe hl.:lp of in tl"rt--.l 1~l\e .... \X'\! fi nd that hettl'r intlat ion rtlrt'ch l ~ :Ifl' nhl:!i nnl from m odel .. t hat l-;.. p l:l lll Ihe cll:! n,L:'" in inlbtlo n , th:ln IrOIll 11Io"kb Ih:1I l" p Llln thl' k,,,,:l of mtbllt,n ,\ l o reO\lT, \\l' fi nd Ihat ,n ll'IV"l raI l'" impro\'t:' 0111' .! h ll l! ) 10 p lVdil"l IlI.',l r t,,'rrll t h:Hlge.. III inlb .. lioll \'\ l' co nd lld e h~ in l ..'.-p n:t ing 0 111' fin di n ~.~ :l lld d"l ll~," lIl g t lwir impli{-at ioll" [u r 1){Jlicy, The relationship hc twccn interest rates and inflatio n I n ,L:elll-rai. thl' l 'XIXTlt,:d real rl"t u rn on:1 loall \:qua l .. thl" n0ll1 i l\:11 rl'hlrn Il" ...... t he l'xpo.:cll 'd r. lll' of inlbl io rl T urllln,L: Ihh n:bl ion .. hip around, (I) \\ Iwr\.' I, I I" th ..- r1l:trkl" i ntere .. t r; l\e ;11 tim e 1 - 1. I~ I i .. lh..: ..:xpn ll'<.l fl'al mIen:.'" rail' at lime 1 - 1. and 1%"; i .. 11K' l ';"P",'{·I..:d r,I1 ..' o f IIllb tion l"lCt\\t't'n rx:ru KI 1 - I and 1"lCI'iod I. If till" n: po.:('tl'<.\ rca l Ft'lkral Iksc:n 'c: Hank of Dallas interest rate is constant, then movements in exIX'Cted inflation will be rCflC(1ed, one fo r one, in movemenLS in the m:lrkct intCl\."St rate. More gcncrdlly, as long as the expected re:11 interest mte is not l-'orrcJ:lled with the market interest mte. analyst.. can infer a I-percentage-point increase in expectatiort'i of inflation from a I-percentage-point increase in the market interest mtc. Of course, insofar as analysts arc able to find variables Ihat control for movements in the expcClL-d rea l interest rate. they will be able to improve the accurdCY of their estimates of expected inflation. Ilowever, even exact estimates of cxpected inflation are of lillie ust:: unless expcCled inflation is, in turn, an act'urme indicator of realized inflation. Simi1:Lrly, the expected challge in inflation is ( I') " - " =11, - , ,I-I" -<: ), where Jrl-l represents the realizl.-d rate of inllation between period t - 2 and period t - I and where (~l ~ IH - )'f'_I) is the ex post real rate of return on a loan made in period t - 2 that matures in period 1- I. If expected changes in the real interest rate are not ("Orrelat(.'(.] with changes in the market interest ratt.'-as will Ix: the case, for example, when the real intere~t rate follows a ra ndom walkthen a l-percentage-point increase in the market interest rdtc can be used to infer a I-percentagepoint increase in expectatio ns of the change in inflation. Variables correlated with the expected change in the rcal interest ("'dte will allow analysts to control for the ch:mge in the real rate and thereby imp rove their forecasts of the market expect:ltion of the change in inflation. This market expectation m:ly or may not be a good predictor of the actual cha nge in inflation. Before mov ing on to furthe r analysis of the additiona l predictive content of nominal interest r.lies. we briefly review the evidence that movements in nominal interest rates, by themselves, Cdn help prediCt future inflation or future changes in inflation. Infla tio n a nd inle rest rates: the e x isting evide n ce A st udy by Fama (1975) first showed that interest rolles might be useful for prediding inflation. For the period from 1953 through 1971 , Fama Economic Review -fourth Quarte .. l<)92 fou nd that the level of one- to six-month Treasury hill rates provided information about future inflation beyond that contained in lagged values of inflation Additionally, Fama 's results suggested that all the varialion in one- to six-month nominal Treasury bill mtes during this period was due to variation in expcct<..'(.] inflation, rather than variatio n in ex]X.'Cted real interest rates. Hess and Bicksler (1975) showed that Fama's finding o f a consta nt real rate o f interest was specific to the sam ple period he examined. However, Fama's other resuh--that market interest rates provide information about future inflation beyond Ihal obtained from lagged values of inflatioo- has been confirmed by subsequent stud ies (Failla and Gibbons 1982, 1984). Table I demonstrates the usefulness of interest mles in explaining inflation and changes in inflation. Line A, for example, summarizes results obtained when quarterly CPI-inflation data from the first quarter of 1959 through the fourth quarter of 1979 are regressed on three lagged values of inflation and several interest rate measures. l The interest rate measures arc the three-month Treasury bill rate, six lags o f the difference be[Ween the three-month Treasury bill rale and the fedeml funds rate, and five lags of the difference between the ten-year Treasury bond r'dtc and the threemonth Treasury bill mte.l If the expected real intcrest rate were constant over the 1959-79 ]X:riod (apart from white noisc) and if market , InflatJCfl is rTl6awred as lhe /ogarilnmic difference of the daIS series CPI-UX (U S BUleauo/Labor Slatistlcs). which cosls on a rental·oqulvaltmee basis Two tf18 rsgrossions 10 cap. ture tnelfllP8Ct of ff1e Nixon Administration wage and price tl98IS /loosing dummy van"aoles were included in dummy variable IS !.KIlty over the 8ClI.Jal (1971 .~1972·4)< and ffIc second durrmy variable is (NlItyover Ifle penod lITm8dtare/y follow.. ing the prICe conllol$ (1973. 1-1974 4) controls Tile ~'rst period of prICe controls , A11lfltBfesl rates are qvartBfty averages We used Akaike·s final predlcllOfl Bfror (FPE) cmenOl'1 fOdIoos8the apptopt"f. ate lag qfh f« 8adl nght-hand-S<de variable The FPE $lO/JSlte (AAaille 1969) measures a regressions ~ square prfKhc/tOfl.,or We used the FPE srafl$lic, along ...tn II methodology OIJI/lned by HsoM> (1981). fO dclermne the orticr In \01IotCII I8g-1cngrn selecrlOflS were made A maunum oI81ght IBgs was allowed expccl:nions of inflation differed from realized inflalion by only :1 white-noise error, then equa tion I would imply a coeHkient of unity on the three-momh Treasury bill mit.: and zero cocfftdents on all Olh~r right-h:U1d-sidt.: varbhh,-... [n fad . the cOt!fficient on the three-month T-l1ill r.t1e is es timatt.-'d prcdsely and is within one stand:1f"(1 error of unity, hut lagged inflation and the interest rale spread.. also have Significant explanato ry power for inflation ' Apparently, expos' real relUrns were corrdatt'd with both lagged iofliltioo and lagged interest rail' spn;:ads over the 1959-79 period, A nu mher of studies have sllggcsted thaI the relationship between interest r.t1es and inflation has broken down or, at ICilst. changed since 1979' This s hift in the intcrest rate-inflation relationship is illustr.lted on line 13 of Tahle 1, which presents r~Slllt,~ from :1 regressio n of inflation on lagged infl:lliol1 and sever.1 1 intl.!L'e.~t rate measures over the period from the first quartl.!r of 1980 through thl.' second quarter of 1991. Note that over this later sample pe riod , none of the inte rest rate measures help to explain movements in infl;lIi()n. Simi!:lr results arc o btained when one substitutes the cbange in inflation for the IeI'd of inflation in the rq:.:rcs....ion equations. DurinR the pre- l980 period , Ix)th the le ve\ of the three-month Treasury bill rate and the spread hetw<::cn the three-month Treasury hill r.ttc and the fooer.ll fuml.. r:lte have significant m:lrginai explanatory power for changes in inflation (line C). After 1979, thi.-. exp!:tnato ry ]X>wcr dis;lppears (line D). What :tn:ounts ror Ihe changed rdalionship betwl;:en m:lrket interest r.ttes and infl:llion since 1980? 13rooldly speaking, only tWD explanations are • The eSIIf7l8Ie<1 cocflrcl6nf ClfI lfIe 1/lI~f1IOfllh Treasury bill ,ate IS I 115. WIth II standard errOl 01 0215 • See. 101 example. CI8Ilda and Fncdman (1 984), Hwz.nga and MrsMlI'I ( 1986), Bemank8( /990), Fama( 1990). Mrsflktn ( 199Oa. I99OtJ). at>dJononancJMISIlIun (1991 J Thec/lange "' ,he,elatlOflS/ljp09lWHr1l(}reresr ,alCS 8fldlll/la/lOfl1$ also eMdenl ..... F.gute I • AIsoseeRomcrandRomet tl989)and Ball<e and Fomby (I991J • possihle Either the rdationship between :t<:tual inflation and cxpc..'Ctcd inflation or the com:b tion hetween movements in nominal interest r.ttcs and movC;'lllents in real intcrest r:lIes must have Ix:cn different during the 1980s compared with the 1970s, Evidence exi.-.ts to support eac h of these views (which. after all. arc not mutU:llly exclusive). TIlllS, for e xample, E";lns and Lewis (\99J) and R:lymond and Rich ( 1<)92) have presentt:d empirical resu lts suggeSiing that inflation policy is subje<1 to discrete ~ regimc s hifts." \ During the 1970s, according to these two studit:s, there may ha\"l;: hcen a systematiC tendcncy for reali7.ed inflation to exceed expectations of inn:nion. This app:trent hias in expectations wml ld he rational if. owr Ihis period, people had fdt that the Feder.tl Reserve might soon adopt a ,~tro n ger anti-infl:Llion:lry st:Lnce, Beginning in late 1979, according to ma ny ohservers, the federal Rcscrvc did indeed launch a campaign to bring down the r.tte of inflation. Therefon;:, the seeming hi:l); in inOat io nary expectations likely wa); eliminated. or even reverSt.'d in sign , afte r 1979. Thu);. :t plaus ible case can Ix' made for a SlKklcn ... hift in [he relationship hetwl."'Cn expeL1ed infl:llion ((he infl:l1ion incorpor.ltcd into market interest r.ues) :lnd realized inflation (the infl:l1ion we arc altempting to explain in o ur regression (:qllation .. ), The view thaI the correl:llio n betwcen nominal interest r.ue); :md re:11 interest r.ues may h:I\'1;: been altered has re<.:civcd indirect suppon from t:mpirical sludics d(X'umenting a shift in the rcl:llionship between nominal interest r.L1es a nd real economic activity- a shift that apparently occurred at alx mt the same time as the shift in the rclationship between nominal interest r.L1es :Ind infl:ltion Bernankc (19<)0) documents the cha nged rclationship octwccn interesl r:ltes :lnd real a<:tivity and uiscussl.!s financial innov:Llions that may have 1"lL1.:.'n respons ible for the change. Expected real rctU!l1S, though not dirct1ly ol"'ISClvable, :Ire thought 10 he closely rel;Ltl.-o 10 movement" in real economic activity (Ch:lpman 19<)1), In summary. exi... ting studies indi<::'lle thar over some ,..ample periods, interl."'St r.lll.'S ~md inteTCst mte spreads have provided informatio n useful in explaining or predicting ,uovements in inflation . Howe\'t'r. thcM.' l)tudies have only eX:II11int:<.1 whedler l11;LrKet intcrc... t r.ttc.."'); h:n'c explanatory or pn.'<.Iictivc power heyond Ih:lt cvident in the history federal Rescn 'c Bank of [)allas of innation itM:lf." In tht, analysis that follows. we adopt a lIIore stringl;:nt test of the u.'it-fulness of interest r.lles in prL"{lil1ing innation. In ]Xmicular. we ex:uninc whether markL1 interest ••lIes provide information ahout price movement" heyond the infonnation inc1udL-d in v:triahles tradition;llly ust.""(\ to forecast innation. &:sidcs inter!::.t r.lIe ..... Ihe variables we consider ;In: 1llC;ISurl'S of money gro"1h. Ihl! velocity of money, the rclativc price of energy, productivity-adJusted w;lges. and ()utpulmarket .!.lack. Wt: first eSlim:lIe empiriell modds that rely solely on the...;c non-inlere...t-r.lh: varlahles. thcn IC."I whcthcr :tdding inll'rest fates 10 tlh.: models ...,ij..:nifk:mtly implUves Iheir JX'rform:lOo.:. Sp ec ificatio n of trad itio n al inflatio n models r.,·lost e('Ollomists would :lj..:n:e that over tht: long tt:1"1n. inflation b <ktt:rmined hy the rate of Illont:y growth rcbti,,\.' 10 tht: l~lI(' of pott:ntial output growth. Tll\.'re is t'onsidt:r:thly ks." con."t:nsus alxlUt the fa(.:tors Th:lt innut:nn' ne:lr-Ienn movemenTS in inn:ltion. Two pamdi/-tllls. th\.' Illonetari:-.t :Ind Phillips-CUi"l' approadles. havc dominated the elllpiril'al Iiler.lIure on near-teml movements in innation In th\.' mom:tarist ]Xlr.ldigm. inn:nion i:. GIUS(.-d hy money-supply growlh in cxcess of money-dem:tnd grO\\·th-in othcr words. by -too much moncy chasing [(X) fcw g<XXI ... - In monctarist lll()(tcis. pa ... r mon!.:y wowth i... Ihe prim:lry dert..'rmin:Hu of nt..'ar·t!.:rm move' ments in innation '1111..' l'hillips-curvc p:lmdigm, in eontr.tst. attribules !lear-term mO"ements in innation to v;lri:ltions iI' execs... dcm:lnds on Ihe bl)()r or output !lwrkets. [n l'hillip~-(" ur\'e Illlxlels, variablt:.~ That proxy for tht: ~ap ])ctwCl'n current OUlput ;md the full -cmrloyment level of out put Pl:tY;H1 impOl1ant rolt..' in determining infl ation. ~/t..' e~timate V:lri:tnts of hoth mOlletarist :Ind Phillips-cur\'\.' mooels, :IS well as :I ~eneral model that draws upon both par.ldigms. Estimalin~ scverJ.l inflation modds allows liS 10 dt..'ten11ine whctllt..'r our rt..'."ull:. afe rohust wilh n.:sIX-""(;t to model specific:llion. Testing for rohustness i." import"nt, given the currcnI lack of consensus ahout whkh inflation p:lradigm i.. :.upc;rior. A monetaris t model of inflation , A typical monetari:.t mcxlcl of inn:llion 1:lkes thc fonn (2) Jr, = (I + L I"e + L. c,I\I", + L d F..con o m ic Rc 'ic w -Fo unh Qua rte r 1992 i where Jr is innation , M is money growth, E is an em:rgy·slux:k variable, and thc lowercase letters dl..'note parJ.mctcrs to he estimated. We use the logarithmil' differencc of the M2 monetary aggregate as OUT Olt..'aSlIrt: of money growth and usc the growth r..ue of tht..' producer prict! index for fuels and energy rc!ati\'c to the overall producer price index as our t..'nergy·shock variable.' All data are quarterly, mnnin/-: from 1959 through the second qU:1l1er of 1991 Beforc cquation 2 I.:an be estimated. several tedlOil'ill issllc:' n!.:ed to be addresS(.'d . The first I.:Onl'l'rnS :Ippropriate lag lengths for Ihc right. ha nd-side variables in the equation Choosing:1 ))1(xlt..'l with hlgs that a rc tex:) ~hort can result in hiased ct>limates, while choosing lags that arc too long ca n I"l'sult in inefficiency. We rclicd upon Akaike's ( 1969) final pr<.-didion crror (FPE) criterion to sekct the appropri:l1e lag ICI1gth for each righth;lOd· . >ide vari:lhle in our regrt..'ssions. A .-.ccond tt..'c hnic;ll issue is the stationarity of lhe rq.:re ...sors in L'qllation 2. If any of its righth:lI1d·side v:lriahles have nonlrivi:1 1 trends, then misleading conclu. . ions may rt.'Sult from hypoth<.."Sis (t..'stin/-: For this reason. Dickey-Fuller tests were u!>Cd to en:.ure th:l\ all right· hand-side variables enterl.-d in (2) arc sl:lIionary" These tests ~fJ.ve amhi~uollS re~lIlts fo r the :.tationa rity of inflation. 111ercfore. (2) was eSlimiited Iwice: once with inflation as the dependent variable and once with thc <:hange in inflation as the dependent variable 9 • Tile exceptIOn Is Bemsnk6 (/990). who also mcludcs a I1lstOfy 01 the U 5 Commercc ();;JpSrlmenr"s Index of lead· ing matestors We also exsmmcd fIlOo'lEIl/lfist models using the 51 Louis fIlOo'lEItllty base. ratfler than M2. and lO(Jnd flO qualitative dIfferences in lew/IS Separalely. we 8xamlflcd modets If! which the CflCf"gy·SIIOCk vllrlBbio was Ihe growth ralc 01 cons(.ll7l(ll"onergycos/s rclalNe to overa" conSIXnCf prices. rather lhan Ihcgrowthrate 01producer enetgyCOSlsre/all'lfi we found flO quahtallW) 10 overall prodlJcer pnces Aga.n. dlffer8f"lC6S III results A /IfI'I8 serlOs random varoablc IS said to be statlOflaty jf lIS dlstributoon does not dePMd on /me • Orflcrencl#'lf1 II random VIII/able lhal IS IlQ'lSla11Oflaty with a SIngle U/'IIt 1001 ensures SI/ItlOll/lfll)' , Table 2 Tradtflonallnflation Models A. Monetarist Model Marginal Number 01 lags L.evel DV E M DV E R' .000 000 .003 3 8 .833 .000 000 .002 2 8 433 DV , ~gniflcaoce a. Phillips-Curve Model Number 01 lags Marginal Slgnllicance Level DV E GAP V2 DV E .000 .000 .013 .303 3 8 000 000 .00 1 .006 2 8 DV .. • GAP 6 V2 R' 2 .829 2 .476 C. General Encompassing Model Marginal Significance Level DV .. , Number of Lags DV W E GAP V2 M DV W E 000 .0 18 000 .093 .194 001 3 6 8 000 .251 .000 .024 114 000 2 8 GAP V2 2 M R' 856 .473 Sample period : 1959.1 - 19912 Definitions ~ • change In the logarithm of the relative pnce of energy In the ptoducer pnc(llnde • . M • change In the logarithm of the M2 monetary aggregate GAP • difference between the logariThm of GNP and the Ioga nthm of potenllal GNP V? • logarithm 01 the velocity 01 M2. W • change in the Iogan thm of the ratio 01 compensation per hou r to output per hour In the nonfarm business seclor. NOTE: For lurther e( planations. S{le Table 1 Iksu lt.., ;m.." pn.·:-.... lu .... d 111 '1';11110.: 1. p:lrl A T ht" t"Xpl.u':tIOt)' po\\ ..:r of lIw 1ll01ll.·I;lri..,1 Il\(xkl.., I" quil l' hi~ h CIo.:;!rly. hoth Llgg l 'd m o ney g rowth ;md l a~nll'nl'r}:.y p rill'''' hl 'lp L'"pbi n Illl )\'l" .. Godfrey Bteusch Lagrangemull1pltet rests wef8 used '** 1or.,,1tJI le$l/IO ~ e..,.,..1t;r ~ n.tyHl<DOIIlIot ~.r.d«hllrF 6 /Q w. ~ • ...".". 01 ~""'('S 1estI .... ,1S/tWN$'~oI 1980. IIK'II I" in il1fl;uil Hl In Iwu hL'1 rq.~rL·." :-ioll i.. :-':I"i;ll corn:I:llio l1 or hL"l\·n l." kl·(b"liL"i ty 0 1 thl' l 'n'or" ;I pro h k m 1'1l11hl·n JlClll ·. tl~ td l l i ol1 .t1 , : ll·..h do nOI p .-o v id .... l' \ id l" n cl' 0 1' ;1 "lnlt·tur. II Im .:;lk in l'it hl'r mod el .•., A Ph i llips · c u rve m odel of i nn' llio n . 0 11l' m oli· \.lliOI1 for th ..: i'h illi p:- \Ilrn' [I1lb lion Ill(xld b prl".'>l·ntL·d hr 11.t1 I111a n . I'orln . ;lI1d SIII;11I (199 1). An ;lll I: I('li\ l' Il': ll llrL' Ilf IIW lr appn ):tl h h Ih:1 1 it m; lkl"" l'xplici t till" 11111'11.111:111(·.... o f monq' g rowth h lr ]l llt).Hl." l"I n In.,., l! .. In inll:H lIJ1l. \\ h ilL- :1 1II)\\mg ,,'X, ( km:ltu l for o u tp u t to in ll u .... l)("l' 11\0\"0;;:- l.''''' I't.··t krn l Itcsc rn ' Ha n k of Dallas ments in infl;nio n over the ncar term. Hallman et al. write inllation ;IS a function of ta~L"(1 inflation and of the gap lx.'1wccn the current price level, P, :md the long-nm equilihrium price level, P'. II The equil ibrium price le\'eI is defined as the price kvcl consistent with the current outstandin~ stock of M2, the aver.Jge vdocity of ~n (V2'), and the current value of potenti:l!. or fllll-employment. real output. u Fonnally, (3) p' . 1112+ 1'2' -Q', when: Q' is polenti;11 real output and :111 variables .m.: in logarithmic form A("(:ording to the e'luation of exchange, p - .~-T2 + 1'2 - Q, where 1'2 is the current vt:loclty of M2 and Q is the current level of output. Clearly, if the velocity of M2 te nds to revt..'r1 to 1'2' a nd if outpllt tends to reVe!"t to Q', then Ihe price kvd will converge to P' Ha llman. Porter, :md Sm:ll1 ;Irgue th:lt this convergence is describt.'"<i hy the equation (4 ) .111" = -b( p - P' ) where .111" is the change in inllation and b is a I1xed, positive p;tr:.uueter. From the equation of exch:mge, (4) can be rewritten a:-. M = -bI(V2 - V2") - (Q - Q')J sions. The Akaike criterion was ust..-d to sck><.:t lag lengths, ;lIld Dickey-Fuller tests were applied to ensure stat iona rity of the right·hand-side variables. Like l-IaUm:m, Poncr, and Small, we found that both M2 velocity and the output gap a re Slation;try . Hc{erosked:lslicity and serial correlation did nOl appear to be a problem. II Results are pn.>sentc..>O in Table 2, p:1n B. The explan:llory power of the inflation-level version of the Phillips-curve model, as measured by itsR1 , is sim ilar to Ih:lt of the corresponding monetari~t model. '111C infl ation-change version of the Phill ips-curve model, on the Olher hand , has somewhat gre:lter explanatory power than does thc corresponding monctarist model. Each righlhand-side vari:lb lc in the Phillips-curve model makes ;1 significant contriblltion to the explanation of the change in inflation. Every right-band-side varia hIe except the velocity of money makes a significant contribution to the exphlnation of the level of inllalion. A generaJ encompassing model of inflation. The fin:11 inllation model we estimate includes both monetarist and Pbillip...-curvc variables. ',he ., Stncllysoealllllg. IhcHal/mafH'orler-SmallderNalOlQOCS /tlfoogh onlY II P IS an If7IP/iCII defla/Ol for some measure of OUtpul As a practlC8l maller, fIowevcr. rrKNe- /Jf1!}fegat~ If! /he C(ltI$(KTI6I" prlCelfldex are fIIgflIy COITetaled ""'th those n f/loimpllCit gross r'lBtKJ(ltl/ product deltalor and mctlts or, more gener.llly, as (6) .111" = a - &1'2 + c(Q - Q'), gross OOmcSIIC prodUCI denator Consequently. 006 rmghl reasonably expect Ihal the set of llaflabies thaI helps explain movements In /he <mp!lCi/ deftators WOtJld atso flelp explain moV6fTlCflls in me CPI where (j is defined as /)1'2'. Equ:nions 5 and 6 art: equivalent if c equals h. We estimated not only :1 version of the Phillips-curve model with the ch:mge in inflation ;tS the dependent vari'IDle, as in t..'"quation 6, but also a version with the level of inllation as the dependent v:lriable In hath cases, lagged v:llues of the dependent variable werc allowed to enter as right-hand-side variables. We used gross national product as Ollr measure of output. Estimattc's of potcntial output camc from the Hoard of Governors of the FL'Cier.l1 Reserve System, As with the monetarist models, an energy-shock variable and dummy v;lriables for the Nixon wage and price cont rols were included in the regresEco no mic Review - FOllr1h Quarter 1992 " Mehra (1989) advocates an altsrna/we definition of the Iong·run equilibrium pncs 16V61 Mehra 's analysIs leads 10 an ;ntlab'on sQuatlOfl that Includos lfIe cllange In lTICfley's oppotluMy cost as a riaht·hand,slae variable Because changes in /h(!oppotlurllryCOSlof hOlding rrI(l()8y are highly correlaled woth Cll8f1Qes In Interest rates. Iv/ehra's analysIS provides e rallOflalc for inclVdmg interest rale measures in 1116 Klfla/KJf1 aquat.on thai 's an sMIched earilel in /tHs af/IC1e sttems/,ve 10 /he ratJonale " F tests yJe/d6d strongo'l1(ienCeBgartlSl thrl slructural slabi/· Ityof thePhdllps-ClXVefTlOd9iflS/lftlatedlll /eVeI$ formbul no tMdencesgsrnst structural slabiittyol /he rrw;JdeI estJmaled form Accordingly. etrnough _ present Issults lot bolt! forms 01 /he rr'ICJ(k!J/. WIt ~ araaler emohaSlS III our analysts on results obtarned from /he IIlflstlOfl-Changc III chIJnges "'."'''''''' 7 Table 3 Interest Rates and the Monetarist Model A. Inleresl Rale Variables 1959:1- 1979:4 Marginal SlgnlUcance Level DV 3M 3MFF I OY3M JOINT 1980:1-1991:2 Marginal Significance level Number of Lags 3M 3M 3MFF IOY3M Number of Lags 3MFF I OY3M JOINT , .032 .089 .1 48 .025 4 .165 .01 1 .043 .000 .dlt .122 .032 .468 .034 4 .978 .016 .000 .000 3M 3MFF IQY3M 4 3 2 3 2 B. Non·lnlerest·Rale Variables 1959:1-1991:2 Number 01 lags Marginal Significance level DV , DV E M DV .000 .000 .001 3 .000 .000 .001 llloti\'atio n lor L·.~til11: l tlll/-:: 1 /-:t·m."r.11 lIl(xlt:1 01 thi:typL' l"(>I11C'" frolll rL~'L'nt finu inl-\:-' lh:tt 1110d l·ls indll(Iing I)(>tll rll {l1lct ari ... t and Phillip:-'-i..:tlr.e l" h :Ir:I(le1"i.~1 ie...... i).:nilkant I ~' o utperfo rm n;l rnl\\"l'r modd :-. ( 11 :lslag and Ol.l11cnl 199 1; t\1chr,1 19(0). Ik.~ idl's lagged value:-. of Ihe cle(X'ndc:nl variallk' (cilllL"r Iht.' k'\cl of intl at ion or Ihe change i n infblion). 1ll(1)L'Y ).:rtlwt h. anclthe OUlpul gap, \\"0,.: include faggt.·d \':tilles or t\ 12 VdCKiIY. bg).:ed ellt.·r,!..:y-pril"e .~ hOt·k:-., ;I nd b).:gel! v: tlucs of :1 1I1t.·:I"Ure of bllor t"().~IS :IS right-hand side \':Iri:Ihlt:... I, Dumllly \ari:lhk's for Ihe )/ixon \\·: I).:C :md prkt.· uml1"oL" l"IJillplo.:lC Ihe model. Thl.·rt.· i .~ no e\·idl"I1t"t.· of :1 ... [rUt'tur.ll hn::ak in the ).:cncr:11 I1J( Xlt-1. re,t.:ardle... s o f \\ hl"till"r [h t.· mcxlel is fO rtllll1:l k ·d i n l enn... of I lle k' \'d of infllition or 1111.' " The labor-cOSl measure IS dchncd as lhe growth rale 01 Cc:mpensal.on per hour rstalNfl 10 OUtput per hOur Hl rile nQIl/atm /)(JSIIICSS SCCIC)( Mehta (1m} and Has /ag and O;!m(JIl/ ( I991} JI'Idude SIfflIIar I8bor-CCSI ~es <fI1hett ilnalyses • 2 E • • R' .876 .564 cha nge in inlblion TIII.: rt: ,... ;Ibo no c\ i(knn' of .....·rbl corrd:ll ion o r IIL·II.·ro"kt.·t!a"'lll"Il\". Ik... u1!s ;tfl.' prl.·... t.·lll1..." in T:lhk 2. pari C. Ikco rdi n ~ to thl! la hl~. Ihe gener.l l IIIc x\t.'1 cxpl:tirb :1 ...o llli:\\·h:u hi,t.:her propor1io ll of 11ll.· v:lri:ttic III in tilt.' kvcl of infl:lIion Ihan do the Jl)onetari."1 :lnd I'hillips CII I"\'\'! IIIcxkls. /\ 10:'>1 of Ihi" innt.·mi:n t:l l e., \pJan:ttory power C()!l1l"S from pnKll lctil·ily-adju:-.lt"d W:l).:CS Th t.: illflaliclrl-ch:ln,t.:t.' \·LT:-.iOI1 of Iht: ).:t: n t: l~ d modd pLTfonns 110 hi:ttLT 1h:1l1 dOi:... Ill ..· corrl.·:-.ponding l'hiltip:'>·l."l IlY..· model. 110\\ 1.'\'t..1". money ;.l!"O\\·t h hl"lps :-.igllific:l lllly in l"xpt:tining \':lrialion<; in hoth tht.' !t.-I·d of infbtion :Ind Ihi: I.·hangt: In inflation. :md Iht.· \d ocil )" \:lri:thli: i.~ no\\' in:-.ignifiGlnl in Ihe c\!al)g..·... rt.·,t.:ri: ....~ion In tlli: cllangl":O; rt.'~1"e'"ion. t tnlik t: l ilt." inflalion· !t-\'d fc,t.:rl.·:-..... ion. \\ ;lge gro,,·th I.·o nlrihult.,... l ink· l.·xpl:l n:I\{)1)· pO\\l"r. Additional inrorm:ltion content of the interest rate v.ariab les To l"x:l1nine tile marginal mf0I"111:1I101l t"C1rl1t.·1lI o f inl,,·rl.':-.I [";1Il"... fo r fuwrt.." inlbllon . wt.· :Iddi:d laggt:d valut..:s of Ih1"(:t..: intert:sl ralt: measurt..:s 10 the right-hand .~idt..:s of our innalion modds Thest..: nW:lSurcs wert..: the level of the lhrecmonth Treasury hill ratt..: (3M). lhe spread Ix:twet:n thl' Ihrl'l'-l11onth Treasury bill ralt: :lOd the ft:der:d funds rale (3MN'"), and Ihe spre:ld hetween Ihe tL'n-YL':lr Treasury bond rale :lIld Ihe thret:-monlh Trt..:asury hill rale (IOY3M). The three-month fatL' was included Ix:cause economic theory suggests Ih:]( the r:l!e incorporales markL'\ expectations of inn:llion O\'er the coming qual1er (comp;lft: wi1h equ;l1ion I). The spread between the ten-ye:tr bond r:lle and the three-month bill rate was indudL'd heclll.~e previous rese:lTch indicates th:1I thi.~ spread is dosely rebted to rutun: growth in n:al econol11ic :Ictivity and. hence. might conlrol for movements in expectt:d re;11 interest r.Hes (Bernanke 1990: Estrella :lIld Hardouvdis 1991). The spread Ix:tween the three-monTh Tn:asury hill rate and the fedl'r.ll funds r.lIe was included p:ully for compldt:nl'SS and p;lrtly hec;lust' tIll' fede ral funds rale is the intcrt..:st rate that is most nt:ady under the conlrol of Ihe mont'tary authority. In e:Kh n:gression t..:quation, both the constant IeI'm :md tht..: coefficients on the interest r:llo.;: variahles wen: alluwt:'d to differ between the prt,;1980 and po~t-1979 s:ltnplc Ix:riods.'~ This split W:1S introduced ht'GIU.'>C formal F tests indicaTl'd that thl.' codlkients on some interest r.:lle vari· abIes in lhe monL'larist :lIld Phillips-curve models s hifted Ix.'~inning in 1980. A split WllS :llJowed even in tIll.' genl'r.11 l11odel. w hich shO\ved no evidence of a structural .~hift. because w<;.· wanted to examine w hether interesl rates played :1 less import:ml rok in explaining movements in innation during th .... 19HOs than during the 1%0.~ and 1970s. (lkcall thai in regressions involving only interest r.lles and Ihl..' bgged level of infl:llion or ch:m~e in inll:ition. we found evidence thai the m:lr~inal explan:ltory power of interest I~Hes d..:dines alkr 1979.) Coefficients on non-iml.'Tl,'slr:lle varbhll.'.~ Weft· nmstr.lined 10 be ..:qu:1I across the entire 19'59-91 1wriod. Wt: found that in the pre-1980 period. thL' short t..:nd of the intereST r.lle yield olive has substantial marginal e.'<pbn:ltory pO\ver for [xltli Ihe level of intlation :lnd Ille change in innation. \\'hile the long end of the yield curve does not. In the IXlst-1979 period. holh short-term and longEconomic Rcvtew - Fou nh Quane r 1 ~2 tl:nn interest rates help to l'Xphlin movements in inflalion. Our general encomp:lssing model with inten:st r.Hes apparently domiml\es the intereslr.lte-all~mented monetar i.~t and Phillips-curve modds. nOI only in terms of ilS stmctur.d stability hut also in tenns of irs ability 10 explain innation. In the gcner.J1 modd, ;1 yield c urve that is steep at its long end lends to signal low, and declining, near-term inflation. Impact of interest rates on the monetarist model, According to Table 3, adding interest rdtcs to lhe monel;lrlst model of the kvd of innation l~lise:-; the J?! of Ihe regression L'qU:llion 10 0 88. Over the pre-1980 period. the three interest rale me:lsure." arc jointly significant at well under the ,>-percent lcveL Ovcr Ihe post-1979 period, the Joint .~ignilka nce lev\.'! f:111.'> 10 0.02 percent, indicating that the marginal explanalory power of interest rales increased during the 1980s. Most of th .... m~l rginal explanatory pow.... r of interest mtes over the elldy sample period is au:ounted for by Ihe level of Ihe three-monlh interest rate, though Ihe spread Ix:tween Ihe three-month rdte and the feJer.tI funds rate is alsO significant at the 10pt!rcent level. The spread Ix:twt;'en the ten-yea r and Ihree-month interest r.:nes-representi ng the slo lX: of Ihe long end of the yidd CUlvc---doeS not contribute Significant marginal explanatory power until the 1980s. l~ .. Lags on the norHn/crcst-rate va"abJcs were held fixed at lenglhs determmed by Akaike's FPE criterIOn 8S app/Jed 10 the monctilnst. f'hill,ps·curve. and general modets withoul m/crest rales The Akaikc cflterlOn was applied separately /0 /he prc- l980 and post·/919 Interest rato vllriables to determine the opllmal number ol/ags of these variables 10 be mcluded m lhe regression equal'ons .. One explanalron 101 th,srcsu/I - llnd 5JfTIIlar results reported later m the arllcle-can be prOVIded Wlthm the context 01 a model (such as /hat of Mehra 1989. 1990) m which /he opportum!y cost of hoIdmf} monoy IS ono determmant of Inflation During the 19605 and 19105. when deposit ralO ceilings wcre generally bmding. /he opportunity cost of holdmg monoy prmcipally vaned wi/h snort·term interest rales Once deposit rate ceilings were abolished, banks wero tree to change depoSl/ rales m rcsponse 10 move. ments in loan fates- whICh arc relalTVCIy Iong-Ierm- and the OpporlUMy cost of hoJding moncy became a funcl10rl of /he sproad between IDrIg·lerm and shott-term market mles 9 Table 4 Interest Rates and the Phillips-Curve Model A. Intern' Rete Variables 1959:1- 1979:4 Marginal S/gnHtclirtce Level DV 411 3M 3MFF 1980:1-1991 :2 Marginal 5'gnlflcance Level Number 01 Lags lOY3M JOINT .020 .018 .31 4 .011 .542 422 .774 413 3M 3MFF lOY3M 3 3M Number 01 Lags 3MFF lOY3M JOINT 4 .082 .778 .012 .011 2 .589 .026 .001 .00' 3M 3MFF IOY3M 2 3 B. Non-Inlerest-Rate Variables 1959:1 - 1991 :2 Marginal Significance Level DV DV • GAP E Number 01 Lags DV V2 .00' .000 .766 .055 3 8 000 000 .019 004 2 • ", GAP E 6 V2 R' 2 .856 2 567 Table 5 Interest Rates and the General Encompassing Model : Split-Sample Interest Rate Coefficients A. Interesl Rate Variables 1980 :1- 1991 :2 1959:1- 1979 :4 Marginai Significance Level DV , '" 3M 3MFF IOY3M Marginal Significance Level Number 01 Lags 3M 3MFF I OY3M JOINT .015 051 990 007 3 596 037 .700 093 4 3M 3MFF tOY3M .049 677 .002 013 Number 01 Lags JOINT .097 .00' 3M 3MFF lOY3M 000 3 2 00' 3 2 AI R' B. Non-interest-Rat. Variables 1959:1- 1991 :2 Marginal Significance Level DV DV W E GAP V2 AI DV W E ., 5B6 012 .000 628 .022 003 3 6 8 .000 286 000 00' 2 , 10 Number 01 Lags 62' 75' 8 GAP V2 2 .895 56' 1't.-dc l"3l Rese,,·., Il:i nk o f DaUas \I;'hl'n intcn..':-t r.Ul.... arc addl'd 10 monelanst ' ";Iriahles in an :lI1Clllpt tn explain cbal/p,es in inllation. the R! of thl' Illono.:tarist equation riS(:.~ frolll O. I,) to 0,6 Colll,t'th·L'iy. tht' inh::rest rate IIlI.:a,un,:..; hdp :..i~nitk:lIll ly in t"xpl:li ning (.'li:l n).,:•.:s III mtl:uion durin).,: hoth the l:;lrly and the Jatl' .~lIllple period, A, in thl' inll:ltion-h.:\'d n:gre:-:-Ion, thl' hulk of interl' .. t r:lle ... · l1larginal explan:ll0l)' PO'\'l'r during the t'arly S:1I11plc period t:ome..; from thl~ ., lIon end of the yield nllYC During the 19HOs. the.: long l'nd o f the yield n1l"\'c ;]1.'><1 1ll:lkcs :1 significant (.'ontrihuti()ll Impact of interest rates on the Phillips-curve model. Ik'sllil:- for the Phi ll ip:-'l'ur..-e model ('1'(//)/" 4) :Ire qu:dit:lllvdy ~illl j br to those for the Illondari:">t modd Illduding interest rates in I h<: I'hillip."-('tI!'\'e equation for tht: Ien:1 of intlation r:lisl'S thl:' I:'qu:ltion's R:' fl'Orll O.H3 to 086 and lai,l:'s thl:' Rl of till' Phillip:-'('ul'\'l' equation for tile cll;lIlJ.,:e in mtl:l1ion from 0 -tH 10 0 57. In thl:' lewis regre:-sion, o nly :-hort-tenll interl'~t r:llt'S h:lve :..ignifiGlIlt m:trHinal expl:m:lt()I)' power in 111<..' carty ~lIliple (X'riod, while it I." prindpally tht' :-Iopt: of the long l'nd of the yidd 1'1lI'\'t' th:1I hdp.~ expl:lin rnon:rnenls in intlation du rinJ.: the 1980s In tht, rt:gn.':.... ion for tht, ('h:tnge in inflation, none of thl' intl're:-ol r.lIe Illl'a:-olire' :tre st:lli:-otic:.ll1y signifil':tnt during Ihl' c:lrly :..:t111ple period. hut the slope:- of hoth the short end and thl' long end of the yidd nu'\'e ha\'e suhstantial expt:lIlalol)' power during the 19HOs. Impact of inter-est rates on the general cncompassing model. Induding interest r.tte .... in I I!I:' general cncomp:t..;~i\lg l1l(xld of intlation r:li:">t'S th<..· Rl of tlte le\·d.. rewession from 0.1-16 to 0 H9 and raise:.. Ihe /{! of the l'h:mges regression frolll 0.47 to 0 ;6. (Compare Table 5 with T:dlle 2.) [n bOlh the pre-19HO :l1ld the posl- 1979 segments of Ihe 1959-91 .~;lInplt- period, measure... of short-term intere't r.Hes help subst:m, li:ll1y in explaining movements in the levt'l of intl:lIion :md the change in intlation. In contra...;!. tile slope o f the long ..:nd of the yk'id clIrn: is of lillie help in explaining movements in the level of inll:llion or the change in intlalion u nti l afler 1979 Tahle 6 shows th:1I when the cocfficie.:nts of tht, intl'f('St rate variahles arc constr.linC'd to I'll..' ('()\lst:lnt o\'l;:r Ihe .~am plc period. all three intt're:-t r.IIC me:l:>llres 1Il:lk..: a st:llistical1y signifiemt l'ontriblltion to hoth thl! level-or-inflation Eco nomi c Rt.'Vlcw _ Fo urth Q u a rte r 1992 and the chan).,:l!-in-inll;lIion regressions Induding i nter..:'t r:lles among the riJ.:hth:tnd-siclc variahles in the regression equ:Hions for the geller.tl en(·ompa:-sinJ.: model h:l.s import:1I11 eff<..octs on Ihl' expl:matOlY power of hoth the o utput g:lp :lnd tlK' vdt)(:ilY of money. The expbnatory powl'r of the output gap for the level of inflat ion is wt':lken(:d hy the inclusion of interest r:lIe" ",hilt- tlt:\1 of \'eI(l("ity i:"> markedly strenWhened 1- (Comparl:' T:lhle:-., and 6 with T:lhle 2) In the inflation-dlan~e "ersion of Ihe gl'ner:11 ent:omp:t~:-in~ model. adding interest r.l«..-'S mark<"-'<.11), redu('es the explan:uOI)' power :lHrihulah1c to the Olltplll Hap. The vclocity vuiahle remains insignifkant Inflation aud the slope of t he yield cu rve_ Is a .~!e q) yield curve ;1 •..;ignal of high infl:ltion or of low infl:1!ion~ 1:- a :..Icep yield cUI,\,e:t Signal of rising inlbt ion or of f:tlling infl:llion? One l':ln determinc the lonJ.:-flm dTet:t of:l change in the Met:pness of the yield ( 'U TVe on predicted intlation. or on the predit'ted change in infllllion. hy using tlte ~Ul1l of thl! t:()effici<..'nts attached {O lagged v;llue:.. of the intert:st r.ue spre:ld. In our estim:lIe o f thl! I('ids version of the gener.tl encompa....... ing tn{)(lcl, this sum of cOt:fficients is 0.891 (with sland:lrd ermr 0,346) for the sprl'ad het\\'l.--cn the three-!llonth r.Ue and the feder.. 1 funds r.Jte :lnd is -0.403 (With standard error 0,175) fo r ,he sprelld Ix:tween the ten-year rate :Ind the thn..-'C-Illotllh r.IIl', Thus, a thrl.-'C-Illonth intercst 1~ltl! {1t:1I is hi~h relath'c to the feder.1 1 funds nile siHn:lb a bi}!.b ncar-term nlll' of inflation. whiil' :1 ten-yl'a r intt:reM r.ue {hat is high relative to lite tllft:e-mOf}lh r.1It: Signals a/oIl' near,tenn r.lH: of inflation. One: interpretation of theSe results is thaI hond ma rket p:tn i c:ipanl .~ eXptXI the Fedenl l Ikserve to r.l i .~c the federal funds r.ue qUickly when near-term in!lation thre:ltens to he high Bel':lusc Ibe thn:e'!llonth ra((: is mo re responsive 10 cxpected ncar'tenn movements in (h..: fede ral funds rate Ihan is tht: ten-ycar rate, the prospc(1 o f an increase in Ihe fedenll funds rare mises the " This 5181emenl 81so 8pphBS 10 me PhdIiPS-i;r,KVe model (Comparo 18b1o <t. p81t B, wrlh Table 2, part B) 11 Table 6 Interest Rates and the General Encompassing Model : Unj,jed·Sample Interest Rale Coefficients, 1959:1-1991:2 A. Intern l Rate Variab les Marginal Significance level DV , 3M 3MFF IOY3M JOINT 3M 3MFF .011 .004 .000 5 , 10Y3M 001 _0 17 .011 .002 001 2 2 2 , B. Non·lnt...... t·Ral. Variables Number of Lags Marginal Slgnlflcanc. Level DV w E GAP V2 M .02 1 .000 .000 .289 .002 .000 , .000 .196 000 .518 .134 000 2 DV , Ilm...1..· lllomh r;th: 1"...·blln.: !O h0111 til ...· ntrrL"1li fund ... inflal ion I cn d ... !o 1'1 ....... llllliV ... Iowly tn 11ll" l ·... li l1 u ' ~·( l l·l)dfici t'n '" :lu;R'hl'd to Lt.!.!,~l"ll \ .I h ll· ... o f Ih...· ... ptva d h":I\\"t:'t:'1l [ t:"n-~l': l r :lnd 111I'L"l"111onlh m 1crl· ... 1 1.11l'''' I'" ...(1.5().". \I 6 8 GAP V2 M 2 R' .891 '48 8 ThL' f(JIiII\\U1).,: "'clli(1Il .lll.lh/t: ... lh..' f(ln:c.I ... l lng ·11 1..: . I n.II ~'''' I''' comp ,l rL" ~hl' 1( 1!\."<.,I"'llIlg p.:lf,ml1:l1ln: or lllod..:l ... wilh tn ll·r..·... 1 r.t1l·" \\ rlh Ih l · 1(II\."CI'>I II1;': f ulllr..:. Illl' :-.t...1..·lx·f h !ht: IOllg l'nd o f th..: ~ n.:ld Ii [111.: E or in lit..: rq.:n.· ....... on ":X p IaUlUl).: (h(/I//.':I ·~ in ('UrIl' T im.... III ...· MIIII W Ix-rl(ll1l1:mcl' lit ..: ~..' rK·r:,ll·rKomp .l ...... in.~ 1Il()lkh or Illl ' lL'\ l ·1 0 1 nllbl ioll ~md thL' lh~Hlg ...· III mil,ll ion . r:tl..: .tnt! Ill ...· ll·n·\ ..... u· r.lh.: :IIltbUO Il . DV il il ...!.l m!.lr... 1 l 'rnlr n . I ';:-l , '\ ).:. 1111 .. 1 I)( ....... il)k inll'r- IX'rf()n1lan c ..· I)" Ill..: 1.-"rr"'·" IXl1ld l l1~ 1111 11.1...'1 ... \\"il I1I )ll \ illk·I"l·... \ 1":'11..: ... \\ 'l ' l o n finl' our ~lIt t:nt ion 10 ilk' gl·lk'r:11 mo(k·l ... Iwell hl' Ihl'Y ""Ill"o m p~ I """ Ilk' Ilhllll'\;lri"'l ;ll1(1 f' l lllllp", ntl"\l' r1 ll xtd, .lmllx·cIU'(' prt:I:1I1( 1) I ... 111.11 Il ll' F...·dt'r.t1I ~l· ...l·l"\l· I:-.Iilougill to ,IlL' Ul'L"ttkil'nt ... I ll) till' in ll'r\' ''1 I~IIL' \,lri.ll lII: .. In I hL' hI: morl: I tI'l'ly to d l'in.: dow n IH-; 11 1('rl11 tnleres! rates II hL'1l infbtiun:II)' pn.·...:-.ltrL\... :11"(' II :tning th:tn st:llIl...· :1("1"0 ... :-. th..: l.·nli."I..' "':l ll1plt' \\'111:11 .\!'L· tl L·I~ 1 1 1110(k ·h -un lit.. ...· I h\· o t l1l ·r ll1o<!l'h- a .....' inlbtiol1 :lI"}, prcsslIl"l·s an: hllilding O ut-o r-sa mple fo reca.!ots rile olll 01 ',Hl1plL' lor .......·; .... ! ir)).: l ·\\·ITI .......... \1 ...· {"(I1UlllU a rL' 0111..' qll.ll"tl-'r·.lhl"ld "'I;l\i\ fl .aT;:I "'\'- Ill' m llt lioll .1' gl·lll ·r.lll'd h~ till' gl,tll:'I: l ll'mornp:l,... in~ St!f> /tIcM..~('969 33} AIICm. ltlV<'Iy, .1 1..1,geSPfead ~ tW('CIlk/l>(l-/(,lm and shor/-lerm lI1/('tL'S/ ta'es may S.gnat.l low <1I't'f"I\WI'ly COSt 01 hOId,f'IQ IJU'kOY ....(1 f>8nCe a lOw """.-'/~' oIlJ1<)rlt"y A Iowveloc.lyColfl1(lf«'y lend$IO.o7lJ)I~,l It ... eq., Ur'(jm (r~~r'''q pr.ce Ie. .. , 'or anv do.',,·'.>.lrd t)t~$SU'e .In " j}<"i·4,.... "If! :"'$ ~ ' IJ'a"d ' ;;w-r <} .j' II /·"rr "'n 'ftOt"'Cy WP lJty 1.-_"" 1f:'_aroOf' ~eJcX,lr "1("',,,Io--,j ," iI St1O.-1rare ''91,', '1i1",1 IIP,I<' ~<" . ,b 01-11 '""I.on II"(.I'(l'$SI(It'I 1l Tt>e s ;;.reiJc, (>;n tI>e ie>.-ei mudd ... TtK' tI)rl"l.l"''''' .I f l' rt.,; t!·linll' in II\\..' ...L'n .. ..: 11 1.11 IIll'Y .If": g l 'llI.."l:Ikd u .. m).! only info rt ll:lt loll .I\ ;tibhk ;tll h .... I lIlll' o f ,Ill· IO!"l·C;t"'I'" FIII11ll'1'11l0rl·. Ih\ · P;U: ul1":(I.·r l·... tim.II ...... o f Ill ...· 111111.1...-1, .m: lIPd;lIl"<.1 \·:tC11 Ix·rit>lI,I ... 111..'\\ 1111<11111 : 111401) 1l<.:nmll·... .1\ ~lIl.thk· IIO\\l·\·l·r. tlt l · b)-.t -.t.nll"lllfl·... nllllt.· Illul.k l ... \\I.·fl· ......·k(·I ..'(1 h~ 1I'1Il)-.t d.II.1 hum thl' l.·nllI"l· '>;l lllpk-. ( "'f X·(· ificlll~. t h ... I.I).:' ll"'l.·d In Ilk' fll ......I'. I ...ling Table 7 Forecasting Performance of the General Encompassing Model WIthout Interest Rates With Interest Rates , Mean eHor -.278 -.255 - 235 -.130 Mean absolule efTor 1.445 1.175 1.303 1.211 Root·mean·square error 1.836 1465 1.697 1.574 NOTE: Based 01'1 one·penod·ahead forecaSIS from 1980:1 lhrough 1991:2 Ali $til1l$I,CS are mulllphed by 100~ for example. a mean error of -0 255 represents an erfor 0/ -0.255 percentage po,nt illOl.kl" ;Ir..: lil..: .".11110.: ;1." I Il()."": in TI IlI .... ()) T;lhk - pn...., ... nl .; for"'Cht .....:-;llll." I'<)r Ihe k\d o f ;111d lhl' d1;111).:V III inlblloll o h t;lint:d from til ..' ~"'IWr:11 ..·'ln ullll.1" .... II).: 111< I{I"'I" \\ it 11 :llld \\ ilh(),ll int,,·rc ..' 1~ 111.:'> Inl·\,.:I)" Cl"': . Ih ..· I'ol....et ... t lX'nod fun ... fWIII IIIl' fir... t q \l;lrtn of 19HO throtl).:h t ill' .....·u "ul qU;lnl'r i >f 1')1) I '\c Int: of Ill..· llll';m l'1Tor~ ;m : "'LlII"'Il(,;III~ difk,.... nt f nun 1t:1"0. indk;lIill,g Ih:11 IIK· lill"lx.l ... t~ .In•.' tlllhLt.'>I..'d l~cK)H ll,,·an -"qtl:lfI.: l'rn>l~ . 1"'" I/)\\l'r for 1I".... ("; I ... t ... lu .....>t.1 Oil n'gn.:~~ io ll '" in thl' dl:m.l..:l· 111 rrlll,lIion 111;111 lor lorl'C hh h:I"'l'd on rq.:f..·...... ion ... in I lIl' k\ ..: 1 C If inll:uioll. ,., T:l hk l{ pn...... o.:llh I lw I"l· ... IIIl ... of lofl11al o.:n · C0t11p:I ...... I11,e. 11.: ...1 ... \ 1"01' :111 l·,pl:lnati(11l of fon.:C''<1 ,.:I1«(Utlll,I ....... Ill,e.ll'''' .... ~L·t: 111l ·1){>x litkd ·· I ·'JI'I."GI ... 1 FI1(( ullp.l ...." in,g. p.I,g1..' l'i ) ' lllo.:"L· I..:... " ("( mfinn tll:H 11ll' inn:11 i( m·d 1:' 11).(1..' \ ·L'r... ic In ... (If 11K' ),:I..'I1L·l~1 1 Ill ( )(k l ~ idd ... i).:l1 i ri clIll I ~ 1110rl..' ;1,'('111';111..' forl..',·:I " l.'; of in lb· [i(m 111.111 d(l II I" , I UTC",p(Uldi n ),: intbli{)I1-k l 'd 1000r.... ("I ... Thu .... h>f\:cl ... t ...'ITIH" ... ()f t h ..· inlb,iol1L·II:I!1).!l' (11(JlkL... IIdp l'xpb m tl1\.' 1"0rn:I.... 1 l ...ror.... of Ih ..· mn:l1ioll-k\ d (11odd~ h ilt I1 CIt 1 i("..' \ l·r.;a (T:lh1<.- 1'1. lirK"~ \ :md 1\) COIll P:II"IIl),: lilL' ill11:lIiOIl("11:111),:"" 111(){k·I ... . tiK' rnoc.kl \\ itll im"''!"I..'''t r:III..'''' ...'nullllp.l .......I..' ... tIl ...· 11lI)(ld \\jllulut r:lll· .... 11111 IIII..' 1I11){ld \\ ithollt r: IIL· ... l:iih 10 l·lll·1l1l1r:t" ... IhL' lllod,,1 \\illi r:IIL· ... IT,th k H. linL' C) CO lb ... <tUl·nll~. th..: forl>t. .l'ling jll.:l'lnnH:IIK... o f th ...· l110 dd \\ Ilh illll..'r...·...1 r:lll·'" i" unal11hiJ,.:lIou ... ly "'''p•.: riof tn tll:11 of lIlt.' Illll{kl \1 nilolll inl ...r ...·"t 1: lIl'''' In p:mi("ub r. iOl("I"I.· ... t 1: 11 ..'" {{>nUII1 rllt"rln;UrOIi u ......l"ul III pn.·dicllng nl..'.l r-ll'1"l11 rn(I\\'llll'nl~ 111 inlblioll fu:onnrnk R~vi<:w - Fou nh Quano:r 1992 Discussion I'r",\·jo\l'" of th" 1".. •... l,;lI"(,'h 11:1:-0 :-ohml n Illat 1110.: :-;lopl..' illIL·n:..'~11~ lll· ~il·ld clln ...· can ."'0.:,,0.: ;I.... ;t u ...dlll pn,:dit'llll" (> f h mJ,.:-t ... nn changes in intl:.il i(lll. AlIL·mpt.... to I,;'XI I': I("1 in fonn:llion alxl\l1 11('(II~\('rl11 111()\L'111L'nt'> ill 111(bll(1Il fl"(1111 1I1II..'r......1 1~lte~ h:l\\..' prc.)\ ...·11 1(1 Ix: k~~ ""ICI.·c" ... ful-l..·"Ix:<."i:l l ly tilr Ih... 19HO~ FI jlrimi. 011 ... l11i1!hl h.I \'L· ... Xp ...'CIl·c,I [he m:l l",gin:l1 in fo1"m:Hiol1 (·0111 ...·111 (>1' int ... rl..' ... l r:l!I..'S to ..,hrink ,I'" ;I(I<lili(1I1.1 1 L·xpbrl.llory \·afi;l hk.; \\"1.'1"... indulinl Oil lill' ri,e.1l1 hand ... ilk of Iho.: inlblion rq.;r ..·....'01( III API); l rt·IlII~. 111 1\\ \'\ ... r. Ihl' \,:lrbhl t'~ Ih:ll migllt 11:1\ ...· lx'i..'l1 L'xfX'{'lnl 10 Ix: lllo:-o1 (:0111lWl ili\l' \\jlh 1ll;lrkL·III1tt.-.....·...II':ltl'.; :KI ll:llly help ti ll..' :11l:lh ... 1 10 c..:onl rol for u n o h :-<"'1"\:lhle IllO\'L'1l1t:nt.... in <:'xlX....ll'd real inIL·rt:...1 r.IlL' ...., T l1u ..;. :-;Upp k111l"n ti ng i lllnl·." 1 1~ 1 1 t'.'; \\ it h 1I:1(l itiol1:11 inl1ujol1-pro.:d iI..1 i ng \ :Iri:lhk ... 1ll:l kl· ... it ...·:lSil·r I(l <.'"I "let 11K' inf()rm al h ltl :lhoul I1I.·ar t<.:rm inlblion tllal i" ilid(kn in tht: \'i dd \.lll"\·(·. By wayol eotr!pafISOf) McNees (1992}rcporIS rhe mflarlOl1· forecaSI<ng perlIYmanceol1!ll ononyn'OU$ 'prOtTldlen/ loreC<lSfe('1ot lire perIOd /(om 11>8 flTSt qU.1f/Cf oIl980lt>rcvgJl the f"51 QUilIrcr 01 1992 nte mean efl(x. mean absolure em;x 8fl(1 fOOl.meat1-SOIJ<I(e emx lot /hJs flYccasler were 01 12 ,ma I 7percooifIgfJpotrIIS fCSpeclJveIy(McNees 1992 r,lt)/e 2! " Table 8 Encompassing Tests Independent Variable A. (K - ~) - Dependent Valiable ,- (Oil-OK) .. Oil -olt' B. C. (K - 1'I") - (OK-OJt") (01[ - &Il') - (4K - oJt") ,- .. f Statistic 2.710' .192 5.146" 01[ -or<! .696 OIt-oK 2.685" 4 1t -or<! .458 Definitions It - It' • forecast error from the level-ol -inflalion model wi lhout interest rates. 0.1[ - on' • forecast error from the change·in-inflalion modet without interest rales. I[ -'II! • forecast error from the level-of·inttatlon model with interest rates. 41l - oil! • forecast error from the change-in-inflatioo model with interest rales. - S!grMllcanl al the '-percer., level. In Ille (: xi'>linJ.l Ii\(:r.HUl"e. a . . ICcr yickl curvc siJ.ln:.ls rbin}: inn:!lion. I h: fl.', howc\'l.'r. short -h.:rIn r..IL·S Ihal :' n.' low reblh '\..· 10 IOIlI-:-I\..'rm rail.''> af'l' a .... i).:o;ll of !alllllg innalion on.·r Ih\..' nca f Il'nn Bond-m :.rkl.'l Ir.ll k·"" po ........ ihl) Ix·l ,l.'\'\..' I h.,1 policym:.ker--. are mor\..· inl'linl.'d \0 driH .' down short wrm intcn.:.... \ r:UI" .... ......Ieepclllll).: I Ile YII.: ld (·Ul'\'l.'. w hl"n till.' ne:lr-h.:nn outlook lor infb l ion i.... fa\,or:.hlc lil:1I1 \\'h ...·n Ihl' Ill"ar-Il"nn o Ullook for infl:tlion is unf;,v( )rahlc. 'l 'hi... inlC rprl't:l l illll I )f IHII" l'mpi rica l rl'."uli s woul d im pl y t ll:11 :1 shift in Iho..: motH;lary aUlho rit y's I"espon," e to chan).:es in 11ll' nl'ar-lI..'rm 1I111:.l lon outlook would 1c:.d [0 : 1 ,..hift in ollr p:I1":II11I"I\..·r \..·....[i1ll:11\..·. . In p: lrtil.~ llbT , if I hI..' F...·tk ·r.1I IkSl'lY\..' wen.: t o hc ).:in to n: ly m o l'\..' heavily on t he .... Iop\..· of thc yidd ('lIIYl' :I.... a g uide for mo nelary policy. Ix )liq'makl.'l"S mighl find that tile t:.~li11latt"d rel:uion.... hip hetwCf..:n tht: :-.Iope of t ile yield n ll'\·t.' :lml n\..';lf -ll'nn innal ilJll wou ld ("hangt: \Vhi k JIlll.'rt..'.~t r:llt:.... hiswrit":.lly have ("olll:linl'd informalion helpful in pr\..'lhn ing nc:u··ll.'l'm innat ion, Ih is histo ric:,1 n.:1:lI ion:-.i1 ip h no g uara ntec l ilal 1ll0Vl'ml'llf.~ in 111:.rkt:1 i nle r!: ... t r:ll cs Gi ll Ix: Llsed SLI("cL·..:-.fu lly to ).:ui(k poliq' f cdc ..." Rese rve' fbnk o r D" lbs Forecast Encompassing The idea underlying the Chong and Hendry (1986) forecast encompassing test is not complicated. Let y denote the variable being forecast, and let y : and y~ denote forecasts generated by two competing models. Consider the regression equation (B. 1) Y = cry: + (1- cr)y~ H , where E is a random error term . If cr does not contains useful informaequal zero , then tion lor forecasting y that is not contained in y ~ , and model 1 is said to "encompass" model 2. 11 a does not equal 1, then Y~ contains useful information for forecasting ythat is not and model 2 encompasses contained in model 1. If model 1 encompasses model 2 but model 2 does not encompass model 1 (that is, if cr equals 1), then model ! is clearly superior for forecasting purposes. II model 1 is encompassed bul is not encompassing (lhat is, if a equals zero) , then it is model 2 that has clear superiority. An easy way to test for forecast encom · passing is to estimate each of two rearrange· ments of equallOn B.l : y; y:, 18.1') y - y: =ll -alllY- y:H y - y;)) . . and (B. 1") y - y~= cr((y - y~ ) - (y - y : )]+E. (1 - cr) is significantly diHerent from zero, then a is significantly diNerent from unity , and model 2 encompasses model 1. Similarly , if the estimated value of a in equation B.l " differs significantly from zero, then model 1 encompasses model 2. If (1 - a) is significantly diHerent from zero but cr is not, then model 2 contains information about y beyond that contained in modell , while model 1 contains no information about y beyond that contained in model 2. In this case, model2 is clearly superior for forecasting purposes. Similarly, if a is significantly different from zero while (1 - cr) is not, then model 1 is clearly superior. In Table 8, we first test whether fore· casts of inflation derived from models of the change in inflation encompass forecasts of inflation obtained trom models of the level of inflation. (ImpliCit in any one-period-ahead forecast of the change in inflation is a oneperiod·ahead forecast of the level of inflation. ) The t statistics reported on lines A and B of Table 8 confirm that forecasts of inflation derived from the change-in ·inflation regressions are superior to forecasts obtained from the level·ol-inflation regressions . Finally, the t statistics reported on line C of Table 8 indicate that in predicting inflation , forecasts derived from change· in· inflation regressions with interest rates are superior to forecasts derived from change·in ·inflation regreSSions without interest rates. In equation B.l ', if the estimated value of " References Akaike, Hirotugu ( 1969), ~ Filling Autoregres...ive Models for Prediction,~ A mmls of tbe !flSfifute ofStatistical Matbematics 21 , no. 2: 243- 47. Fama , Eugene F., :lOd Michael R. Gibbons (1984), "A Comparison of Inflation Forccasts ,~ jOllma! of Monetary EconomiCS 13 (May): 327-48. Balke, Nathan S. , ;1Od Thomas B. Fomby (19')1), ~Large Shocks, Small Shocks, and Economic Fluctuations: Oudicrs in Macroeconomic Time Series,~ Feder.lI Reserve Bank of Dallas Research Paper no. 9101 (Dallas, Pebrua!),). - - - and - - - (1982), ~ 1nf1ation , Real Returns, and Capital Investment, '· joun/al of Monetary ECOllomics 9 (May): 297- 323. Bemanke, Ben S. (1990), "On the Predictive Power of Imerest Rales and Imerest Rate Spreads," New En8,ltmd Economic Review, Fede~1 1 Reserve Bank of Boston. November/ lx.x:cmbcr, Fmnkel , Jeffrey A. , and Carn S. Lown 0 990, "An Indicator of Future Innation Extracted from the Steepness o f the Interest Rate Yield Curve Along Its Entire Length," NRER Working Paper $erit:s, no. 375 1 (Cambridge, Mass.: National Bureau of Economic ReseMch , J une). 51-68. Chapman, David A. 09')1), "Real Bond Yields and Aggregate Activity" (University of Rochester, November, Photocopy). Chong, Yock Y.. and David F. He nd!), (986), ~ Economctric Evaluation of Linear MacroEconomic Models ,~ Review of £Conomic Studies 53 (Augusl): 67 1-90. Clarida, R. H., and Benjamin M. Friedman (984), "The Behavior of u .S. Short-Tcffil Interest Rates Sincc October 1979," jOllma! of Finance 39 (july), 671-82 . Estrella , Arturo, and Gikas A. Hardouvelis (991), ''The Term Structure as a Predictor of Real Economic Activity, " joumal of Finance 46 (June): 555-76. Evans, Martin D. D., and Karen K. Lewis (991), "Do Expected Shifts in Innation Policy Affect Real Rates?" (New York University, University of Pennsylvania, St:ptembcr, Photocopy). Fama, Eugene F. (1990), "Tenn-Structure Forecasts of Inte rest Rates, Innation, :lOd Real Re(ums,~ joumal of MOflelary &ollomics 25 (janua!)'): 59-76. - - - (1975), "Shon -Tenn Interest Rates as Predictors of Innation : America n Economic Review 65 (june): 269-82. " Granger, C. W. .J., a nd Paul Newbold ( 1986), Forecasting &OflOmfc 'lime Seri£'S, 2d cd (Orlando, Fla .: Acade mic Press). Hallman, Jeffrey .J., Richard O. Poner, a nd David It. Small (1991), "M2 Per Unit of POiemi.1.1 GNP as an Anchor for the Price l.evel," American Economic Review 81 (Septe mber): 84 1-58. Hasiag, Joseph II , and D'Ann M. Ozment (991), "Money Growth, Supply Shocks, and Inflation, ~ Federal Reserve Bank of l>dllas &onom ic Review, May, 1-17. Hess, Patrick). , and James L. Bicksle r (J 975), "Capital Asset Priccs Versus Time Series Models as Predictors of Innation: The Expected Real Rate of Interest and Market Efficiency," joumal of Financia! EcOl1omics 2 (December); 341-60. Hsiao, Che ng (981), "Autoregressive Modelling and Money-Income Causality Detectio n ,~ joun/al of MOllefary Economics 7 (janu:t!)'): 85- 106. Huizinga, John, ;md Frederic S. Mishkin (1986), "Moot'la!), Po licy Regime Shifts and the Unusual Behavior of Real Interest Rates ,~ CamegieRochester Conference Series OIl Public Policy 24: 231- 74. Johnson , Manuel H. (1988), "Cu rrent Perspectives on Mom:t:!!)' PoHcy/ Catojoun/a/8 (Fall) 253- 60. fcdc:ral R~rvc: Bank or Dallas jorion. Philippe, and Frederic Mishkin (199 0 , "A Multicountry Comparison of Tenn-Structure Fon.."C'ds(s al Long Ilorizons. jOlll1lal of Filllmcial Economics 29 a"larch); 59-BO. Mishkin, Frederic S. 0990:1), "The lnfoonatio n in thl;! Longer Maturity T erm StnJcl ure About FUlure Inflalion," Quarleriyjollmal of Economics 55 ( August): 8 15-28. i\kNl-'t'S. Stephen K (1992), "Ilow Large Are Economic ForcedSt Errors?M New Englantl ECOllomic Hel'ieu', Fl-"derdl Reserve Ba.nk of Boston , july/ August, 25-42. - -- (l990b), MWh(lt Docs the Teon Structure Tell Us Aoou t Futurl;! Inflmion?" jOtlmal of Monetary EcO llomics 25 (January): 77-95. M - - - (1989), " How Well Do Financial ~brkel'> Predict the Inflation Rate?~ New Englwld ECOliomic Review, FcdcrJ.1 Reserve Bank of Boston. September/ October, 31-46. Mdll~l. Y;lsh P. (1990), "Ikal Output and Unit Laixlr of Inflation," Federal lksclVt.: Ctn>!S ;1$ Prl-,<-li(1ors Bank of Richmond ECOllomic Review. july! Augu:-.t , 3 1- 39 Raymond. jennie, and Roben W Rich (992), ~Changes in Regime and the Behavior of Infl:Hion" (Auburn University, Vanderbilt U niversity, ,1:l nU:lI)', Pho tocopy). Romer, Ch ristin a D , and David II. Romer (989), " Does M ont.:t;tl)' Po licy Mauer? A New T est in the Spirit of Friedman and Sch wal1z," in NBER Macro(:'CO l lomics AII/Illal 1989, ed . O l ivier jean Blanchard and Stanley Fischer (Cambridge, Mass. : MIT Press), 12 1- 70. - - - (1 989), ~Coi ntcgr.nion and a Test of the Qllantity '11lt:!ol)' of Money,MF.::dcrJ.i Reserve Bank of Richmond Working Paper no. 89-2 ( Richmond, April), Economlc Rcvie w -Fou nh Quane r 19'92 17 William C, Gruben Senior Economist and Policy Advisor Federal Reserve Bank 01Dallas Trade Policy and Intellectual Property Protection: The North-South Dispute ill\'l:nIOr~ h:l\'l: a righl 10 profil from Ihl:ir invl:rllil)lb? Thl.: i11lplicalion~ (If Ihi:-; que:-;li{lIl h;lvl' l'\'okcd c(lntrnvcr~)' in r(!n~ nt ye:lr~. :md IlllKh of I Ill' di~pulc h:l." oCCllrnxl hClwl:l:'n devdoping and lkvehll1l"d countri(:!:-; Thl:' maj(lr pl'(ldllCl~rs 01' inldkClu;l1 rroPerly. Ih .... dcvdoped 11:l1ions. I}'picdly \\anl ~lrong int ..:llectual properly protl'Clion Ik\'doping n;l1ions. which Iradilion:rll y ('Ol1.'lUlH: more rntl'llcCl ua l properly th:lIl they prodllCl·. pRofl'r nOI 10 p:ry for it They oftl'l1 \\'a nt \\ l"akl'r pruil'ctron (See lile hox titkd - What Is Intl'lknual ProJX"rly~" ) Hl"n.·ntly. 110\\ L'\ -;()lIll' dC\'doping luuntri\.·s h:l\ l' IX'gun to tiglllen Ihl'ir proIL'l'lion of imd1cctll,ll propl'11)' Thi.' :111i('[t' di'>Cu~~'~ lilt, moti\'l'S hehind IhL'X' ('OllOlriL'S' recent ch:l11ge!'> :md 1'0.:':1.1.>;('., o n lllL' l"rx:rrl'Ill'l' of Latin AmeriC:1I1 COlllllries in Ihi' pr{)l'I':~' In larl-te p :1I1. the h:l!'>c'" for IhL·~· (·h:lngl.':-; il1\OI\L' tradl.· policy. I arWle Ihal t\\'o types of linklgl.'s hl.'I\\'l·I.·1l 11~ ldl' poliq' and inlellectual propt.'ny exist Fil};l. :IS 11:1' ofl .... n Ix~t.' n nOled, we:tk inll'lkCIlI:ri propt.'rly Ia\\ ., in devl.·loping Olunlries 11:1\'1:.' kd d;"vdoped ('ountril's to n.::ta liall' with sdectivl' barriers 10 Third WOI'ld exports. Some do.:\'eloping ('ountries ha\'I.' hegun to re ... pond by tight .... ning tho.:ir intdIl-du;11 p1'OPl:'rt}' I;l\\ .~ Uut therl' is l!!llCh that tl1i:-; ., l:lndard ;l rgll111 .... t11 (';Innot l'xpblll. i\ol all counIril·... lIndl'r tire highl's! pf(.'~.'lIft' 10 ch:mgl' Tlll'ir I :I\\'~ ha\ L' l·hangt.'d tht.'1ll Thl: dilkrl'l1c .... ho.:t\\'l:en ~·olln tn~'.'" tllat han;, not ch:rngl.'d and thoSt., that 11:1\ l' app..:ar, to rl'nl.,(·t ;1 SI.'Cond typl.: of linkagl' TIll' ........~:()nd type;, of linklge i!l\'o!\·I.-'S :r spt.·cific rdallOtl., h ip Ix·t\\'(;'en :1 l'ollmr)"~ tr.ldl.' polky and \\ hat it Ina~ pern.'i\·e :1... its Optilllulll imdlt.x·llI;tl pro]lI.·r1y n.:gilllL' When:l coun try Ifil.·s to lkvdop a r.l11ge of hOIllt., indu<;tries hy pro(I.'l1mg thC!ll from fOfl.'lgn l'omrx'tition, :t loo~ intdll:'l'lu:rl propt.·ny D (J L".. t:':on o m ic R..,v it:w _ Fo unh Qua n c r 1991 rq..:illll' m:.y ~ee111 rational. A ... I will expl:lin. getting tlrl' late...1 technology in :,o uch a country may not he ca ... y. Will1mll foreign comp ....liTioll , howl:'\'l:'r. industries 111:1Y not need the latesl tl:'chnology; wilhout intellectual property prolection 10 himil'r them. th~'y ca n get older lechnologil:'s for nothing. But many developing countries are shifting fro111 .' trong 1I';lde protectioni.~11l to more lilx:r.ll tr..dl.· and foreign investment poli cie~ and 10 an inl'fl';hing foclLs o n m:mufaclUring exp0rls For IIlt.'ir p roduci s 10 cornpl: te more intl'nsl'ly in IXlth d0111l·... tic and world market!'>. dt.:\'doping nations mllsl h:t'e their pro(illCl.:. on 11ll' latt.:...ltL'l·hnology. Whldl Ill..:)' C;ln :lCquire morl' e;l~ily wh..:n tht::}' proll.'l'I intdlecillal pmpt.'rty 1r1 ~UI1l. IwO of m y pnndpal a r~lIllle nts ;lrC thar Ir.l(it.' proll."CIioI1~,m r:ttion a1izl:'.' loo.'tt: intdlectu:11 propt·rty bW!i ;Ind that tl';lde 1iI)t'r.lli.' m r;lIion:lli/.;.·~ light laws. Why has Latin America preferred weak protection? During Ih..: first IWO dl.·C;ldes itftcr World War II. Mexico. Br:lzil. Argentina. and mo...t other COll111ril.·s in i.:rtin America followed trade lxllicies !(It.'llsl.'d on illl/XH1 slIbslill//iOIl. a policy thato.:n1:1ik:d the protection of local m:rrkt·t~ wilh hi.gh barril'rs to import!'> Most of these cOllntrie... ('(mlinUl:'d their :ldhl'rl'nC(' to import substitution polidl's in The 1')60.... hut h y the 1970s. Ihl.' dbtnbution of tr.rde ano Da.-.d AI Goold offered many heJpfui cormtefIIS as oIlhts attJClc I also oen· 81,led fromd'$ctI$SIOOS .... ,thAh$l;tnButler. Ed\IwIF E~sre<n. Joe W Prtts III. IJfId Rober' AI Sherwood Any errors tY Stephen P A Brown 0f'1lISSIClflS are ,_s my own t. What Is Intellectual Property? Intellectual property has two anributes. First , it is a tangible product of the intellectan invention, an idea, a product, or a process. Second , a nation's laws say that intellectual property can be owned; it may not be used by others without the owner's permission. In the most general sense , laws concerning intellectual property address fou r types : trade secrets, patents, trademarks, and copyrights. I Even though each of these four types is different, the lines separating what they protect are often blurred , In the United States, for example , a computer program may be protected either by copyright or palent , Despite these ambiguities, an understanding of the broad distinctions between such laws is useful, Among the alternatives in protecting in tellectual property , the patent is the most powerful, The patent is a temporary right to exclude others from using an invention. Patent laws grant this right for only a certain number of years, and it is not unusual for inventors to complain that a patent's duration is not long enoug h, The copyright is also a temporary right . It is the right to keep others from sell ing copies of one's creative expression . Pirating copies of copyrighted material-such as films, books, and computer programs-may be among the pol i<.:ies ;1I110Ilg till::."":: CO llntri<.:.~ had hegun to widen A fe w beg;m dr()n ~ to d evdop SOI11l' 1l1;l1lufau ll ring for l'xport. ' But for the 111()st p;1I1, l'\'l'n thl':--l' nat io ns pe rsi:-;tl'd wi th their impnr1 s l1h.~ti t u ti ()n polidl':-- It i:-. no coinckk nn .: [h:II , thro ugh this l'nti rl' pl'rtod ;Ind into Ihl' 19HO:--. Ihl'sl' count ril's a lso o lle rt;'d rl'bt in ::ly \\'l':lk :md ~poltr intclll,(·tua l pn )[ll'rt y pro tection. most publicized class of violations of intellectual property rights. A trade secret consists of otherwise legally unprotected confidential information that, in a firm's opinion , gives the firm a competitive edge , The trade secret, although possibly not covered or even subject to coverage by patent or copyright law , has demonstrable value ; a firm could legitimately claim damage if a departing employee left with the secret and transmitted it to a competitor. Last is the trademark. The trademark is a word or mark that identifies the sou rce of a good or service , The sale of clothing and ath letic shoes that display unauthorized trademarks isanotherofthe more publicizedclasses of intellectual property rights violations , In the United States , protection of trademarks and trade secrets lies principally with the states , as contrasted with the federal protection that dominates for copyrights and patents. , While tnese loul classes o! intellectual propeny law are the !OOl;l llnponant, there are orhers SheIWOOd ( r 990) adds a lilth etass laws tllal prOlect computer mask works: or chip de· soons [115 po$S<bIe 10 protect mask works b-,. patent. copynghl , o. t.ade socrel law, bullncrea$II'lQ fy. a separate body 01 law IS developrng 10. them lesser 11990) also Clles a t>tlh class' special laws that offer palenl · ~ke protec1lOO 10 cover bred plant vaflel>es l e sser does not Include mask wooIIs as a separata category IInl)ort s ubstitutio n rind i nte llectual lll"opCrty. l ~ lIin i\ml: ric I!1 ('mmtril-s co mmonly bast;:c\ im port ... uh~[ i tution ;md intd lt'ctua l p ro pl'l1y IX)lit:il's o n th~ :l rgu ml'l1I s (If Ibut I'reb isch ( 19')0 and 1959), who m :lint;lilll"(i that lilt, tCrlll:-- o f Irad !:: Wl.'rl' turninl-\ ;Ig;tilh l na tio n .. w ho:--\:' c hic f e xports \\l' rl' raw m; lt c rb l~, :nld Hl f:I\'or of e xpo rte rs of m:mufactures J> rc h i ~('h ("o un ..dt'd co nti nlled export.~ of r:I \\ n1:Itl.·ri:rl:-. to gl'l hl·r w ith thl.' dc \'\,"lo pm .... n t o f til.· m;t lllifacturing l'Ol pa("it)' ta rgct\:'d 1(m::rrd [tw ho me m:l rkc l Th i.. :--tr:llq-l), wou ld not o nl y 10:--[('( d L'\ d o p m l'lll. hut ;t1sn .__ ;1\ e fo rt'i AIl exl h:rngL' FI)II,)\\ in).: p ~bi'o(· h . .\lexi("( ), Ur:r zil, i\ 1}~L'nti ll ;t. am i "'()lIll' "'lI1;tlkr coumril's p ro tl'ClL'd d,,,,,,: . Amotlg me most OOloeeaO/e such coon/TIes ...,ere BtaZtl and KQlea 20 t'l'tkr.1i Rl."SCl"n: llank or Oall:1S manufacturcn; with high tariffs and mher barriers. The protection of their markets gave moS! manuf:lCturcr::. in ,hese countries lillie incemive to hu y the btt-'St technology because trade harriers barred many of the foreign producers who h:ld il from competing in the developing countries' markets :myway.' Prebisch and his followers counsdl!d weak intellectua l property protection, so that the local developing countlY producers <..·ould avail them.selves of technology without the discomfort of paying for it. ~ i'res::.ure:-. against .~t ronK intellectual property protection were not only high :md widespread in these countries, hut also efft:·ctivc. Local manufacturers might not have heen ahle 10 get ehe vcry btest technology hut, for what they ("Ollid g~t, the price was righ!. Ahhou~h Latin American and other developing COLlrltril.:s have traditionally offered weak intd!l!<.."il1al property protection , they h;Lve offered some. The difference between the motivations of nm . . uming countries and the moriv:llions of producing countries has m;lt!e developing-country protection spony. A country can v;uy it" degree of intelle(1llal pro]X'rf)' protection among individual pnxluJ.:ts ;Ind processes Developing COUniricS' pro{IXtion lIlay have fr.ldit ionally heen Spolly in order to encour;lge innovations the country otherwise could not get; spD(ty protection enahled Ihe J.:ount!)' to avoid paying for innovations that would occur ;lIlyway. For ~x:ltnple, where an intelk-'Ctu:tl-proJX·rtyI.:onsuming country's distribution of demand.. for innovations differs from the producing country's di:-ot ribution of demands for innovations, the consuming J.:oulll ry is motivated to protect (Diwan and H(xlrik 1991). Suppose the consuming country ch:m;mds innovations that arc: very differ~nr from innovations the producing country demand..,. The producing country may still innovate pmdu("!s the con:-oulTIing country W:lOtS. but only if Ihe consuming country protects intelleCtual propt'rty. If the consuming country has a polio epid~mic and tht;.· producing country does nO{, the pnx lucing country 1ll;IY still invenl a ('ure for polio. if the consuming country protcrts the CU fe ' SUPIX>SC a polio epidemic hits both countrit-"'S. In this C ISC. tile prodllcing country and consuming count!)' have identic.1I preference:-, so the consuming COUlllry has Ie ....... incentive to pmlcCI. The producing country may still inno"~lIe whm the Eco nomic Kl"Vlc."·w - Fo unh Quaner 1992 consuming country wants hcC'.llIse firms in the producing country can at least benefit from innovating when they sell the product at home. Other arguments for weak inteUectu.al property protection. The case for weak intelleaual property protl.:aion does not rely solely o n the ;Ihility 10 acqu irc innovations without paying for them. Strong intellcctual property protection may :Ilso pt!nllit monopolislic abu.sc,:.;. Under some condition... , :1 monopolist may accumulate patents to preserve tht: monopoly a nd Ihen allow the patents to "slt:e p ~ so as to deter e ntry inlo an induslry (GilI)t;'rt and Newbcry 1982). Chin and Grossman (1990) cite cases in which, for innovation-consuming countries like the developing nations, the cost of monopolization morc than offsets the contribution that stronger intellcc."iua l propeny protection can make in stimulating cost-sa ving inn ova l ions. ~ , This roiarlOfl$hlp borween /fade prolCCt/Oll'sm and the de- mand for oow/ec1lllOlogy has perSlSIcd UsIng oata /rom a 1981 SUt'VCyoi moto than3.tlXlBrazilialllirms. Braga and Wlltnore (199/' /ou(l(J thatlirms' ~ oIredlnol- OW IhlOtJ9h rc5fNJfCII and deveIopmcn/ and IIieII likelihood 01 pl.lfc:haSlflg fore.gn /echnology were both negalrve/y rela/ed /0 rhe degrco /0 which lhil<r IfIdustfl6S were protec/cd Ircm /oI6ffJfI compe/ll/OII J FrY discllSSlOflS 01 /lie early ar~ts 01 Prcblsc/1 and hts /oNowefs. sco$hMwood (1900. 173-74)and Gad/:lawar>c/ Rldlards (1988,21) ShcIwoodno/8S thaI Preblsch. not ICn(} belOfQhls death on 1966. re/fICtcd these earl/Cf arguments Otxong the perIOd I 8fII cIIscussong. howellllf. /he conSlS- /ency of weak mtellectual property laws With impofr subsliIv/lOll was perceived as high When Preblsch f;nal/ydld rum af}ams/ tho notJOfl of weak intellectual property protection. he aloo lumed against the /<ica 01 Import svbs/itutlOf) • /I 's mtcrestmg /0 nota that Fusch/ok, m a paper that oIfers a plan 101 t'9ntcnmg Btu,'" In/ellectual properly policies. Sl.lQges/s changes //leI arecomp/ele/yconsistent With these arguments -The onlroducflOll 01 full (prodUCI and process, patent protccl/Oll ITIIQ/lt lead /0 market dJSFUpllOfl and. po$Slbly. IfICrBascs IT) ckug prICeS Wllhool clear benefits A poSSJOIe t/SfISIIJOfIaI step might be /tie development 01 a new cktJgs would be "posI"ve hst' 01 c/sslJaSeS for which palentBOIe, such as t/opIC8I dlseascs highly prevs/enl '" 8uwJ (C/lagas dISease. schtS/OSOmaSls. maJarra. elc r (Fosch/8/< 1990. 64, • Hefe. tho def}t1N! 10 which /lie COflSUfTWIg country is moll· va/ed not 10 ptOlecl IS. IT) parr. Inversely re/alcd 10 lhe (elatlve SIlt! olliS market " From the pc~pcctive of net innovationconsuming countries that want to e nco llr.lge inno\·;\tion at ho me . an additional a rgument as'... inst sHong intclicctu:11 propeny prott:ct ion involves the instilutional ..fnlcturcs in which such property is produce<.! and dislribult."d. Vessuri (1990) notes that Imnsnatio ll:J1 compule r corpor3lio ns in Br.lzil were not intcrc:.te<J in developing or :1 bsorb in~ kr.ll technological efforts bcc'.lllM~ of the centr.llized nature of their own rese'.lrch and developmenl So, instt..'ad of protet1ing this intel1t,:ctual propeny strongly, 8mzil tried to foster local innovat io n by reserving a portio n o f tht: market fo r minicomputers, microcomputers, and their peripherals for domestic pr<xlucers. For Latin Ame rican and othe r consuming countril.'s, anothef afgume nt ,lgainst strong intellectual property laws is that enforcement eosts can be very high ( Primo Br.IEP 1990h). Whe n the bulk of protection is for foreign innovations. these enforccmc=nt co... ts lead to increas(.'(i royalty gains for foreigners and gre;lter roY;llty expenses for nationa l s.~ Foreigners do, in fact, hold the b ulk of palents that a~ ~gisten.."XI in developing countrit:'s. Not only might foreigners benefit disproportionately from lio:n..;ing and myaltit.'S, blll also from cost advantages. '111t:' introduction and enforcement of tougher intellectual property laws would inCrL'a.se prodm·tio n COSts for domeslic producers who had not ht.-cn p:lying fo r the intellC("tual property that they wcre using. These domestic producers might he s upphmted by the foreign finns that had origin:ltly produced the imellcctllal property on whil.:h the domestic firms' production processes were hased. Pir.Hes who continued to sel l products of the intellect, or use them in production proccsscs , withOlll paying thl,; producers would in ;lny Cl.~c= be run Olll of hus inc:.:ss. The changing role of Ihe Unilcd Siaies in intcrnalional trade liberalization In the 19HOs, new pressures Illolivated some developing COUnlrit:'s to cha nge bot h their tr.l.de • Accordll1{1 fa the OrganualJOflIot ECOfIOfflIC Cooperall(l(l SfId Dovclopmenl (GECD 1989). fJarlOfl8ls 01 dcve/aplag CCUlIne$ hcldorly I {NIfCMlfclClt15/ong paTents" Iheworlt:1 policies and their intellectu;11 pro(X!rty regimes. Some of the pressures involve events in the United States. From the end o f \'('orld War 11 until the late 1970s, the Unitc:.'<.i State~ adopted a relatively lihc.:.·r.lltr.lde regime based o n the idea of I1IlIlt;lateralism. Ihe idea th:l! trade agret:ments :lTnong m:my nations at the .same time work lx:st. The United States expressed this commitment through il<; support ~md use of the most /{//I()ret/ /latio/l d:llIse of the GenC!"'.JI Agreement on Tariffs and Tmde (GAlT). By this clause, any reciprocal tafiff re<.!ll(.1 ion negotiated between the United S\;Ites and :my of its trading partners applies to all U.S. tr.lding p:lrtne rs. However, under the GArr. the Cniled Staws also offered speci:ll tr.ldc privileges for developing countries through the Gcncr.dized System of Prefere nces (GS P). Through the GSP, the United States and other developed cmllltrie."; may waive duties on se lected imports from .selechx l developing I:oumrit:'s. Curre ntly, :1 number of products of the Mexican in-bond pbnts ( maqlli/culoms) flow ha<.·k to the Unite<.! States under the GSP. lly the 1980s, the Unit(.-d Sta tes had begun 10 hack well aW3y from tr.lde liber.J lis lll and had startt.'<.ito use GSP prOvisions :IS wC;lpons. Fruslmted with what it viewed as the con\;lmination of Illultil:ller.l.lism, the United States markedly inl:rcast.'<.i its use of cOl l11ten/ai/il18 actio/IS, suc h as r.lising tr.lde barriers it had lowcn..'<.i anti inI:reasing its filings of dumping c h:lrgcs \Vhatevcr else these a ction~ meant, they signak.'<..I a rise in U.S. protectionism . Beginning in 1984, tht.· mte o f :I<.u::ler.ltion lX!came more acute. One rOllte Ihe United Slates chose fo r th is escalation is p,lrticularly relevant to intellectual property protect ion in developing countries The U" ited States provides itsel f retaliatory n.:mcdies for "u nfair trade~ through Sect ion 30 1 of the US Tr...de Act of 1974 and Section 337 of Ihe Tariff Act of 1930. Seclio n 337 prolects U.S. firms from "unfair" t."Ompctition from impons. Se<:tion 301 offers more ]c:.-eway; it allows the United States to ael as'... insl inadt.""quatc prott.'Ction of its intdlectual property in other countries, even if Ihe vio latio ns do not involve products imponed into the United St~l tes . In 1984, the U.S. Congress passt:'d a Tr.lde :Jnd Tariff A(.1 that strengthened oplions under Fc:d e.,.,1 R~,","4e Bank of Dallas S~ct ion 301. Th~ 1984 act ex plicitly desiRnates wt'ak intdlectual property protection in a country ;IS grounds for withdrawing trade concessions extended to that country under the Generalized System of Prderences. Next, Congrt..'ss pa.'i.sed [he Omnihus Trade and Competitiveness Act of 19HB. Thi.~ act requires Ihe United States Trade Repfesentative 10 specify timetables for investigation of actions and 10 identify countries that have inadequate intellectual property regimes. In sum. tlH:se me;lsun:s exp;mded options the Unitt:d St;l1es could use to pressu re countries to ch:mge their intdlectual property protection Congress adck.--d provisions to signal to U.s. officials thai they had hetler use these new tools. Mody (1990.219), commenting on Ihe 1984 Trade and TarilT ALt, notes that '"four cOllmries, Korea, i'I "It::xico, BrJzil, and Thai land, have been affected by this legislation."' With tIll' 1988 act, the list lengthened consider:.lhly. Foreign debt, trade Hberalization, and intellectual property reform l';tr1. :I!though not all. of the increase in U.S protectionism constituted retaliation for inc rease.~ in Third World protectionism during the 19HOs.During the 1970s. sewral Third World countries had chosen to t:'xpand their development progr:.lms by resorting to foreign loans. These cOllntries borrowed under the assumption that the prices of their exp()rt.~, \vhich had been high during much of lhe 1970s, would remain high enough to allow repayment. \Vhen [he prices of many developing countries' princip:l! exports fe ll hard in the early 1980s. a world debt crisis ensued To revive their s:lgging terms of tr:.lCle a nd to earn ha rd currency to pay their external obligations. some developing countries commenced the saml' beggar-thy-neighIXlf policies that the developed countries h;1(1 used during rhe Great Deprl'ssion. Th;1\ is. they increased protectionism . In the face of m:lssive trade retali:l\ion from the Unitl.--d St:nes. however, some countries changL'<.1 cours(:: and ope ned their own economies. If Mexico's economy, or Korea's, became more open, how could the United Stmes, with its ongOing dedar.nions of commitment to free trade, remain closed to thell1~ Of [en developing nations in the Western Hemisphere evaluated by Wil!i:lmson Eco no m k Revtew _ Founh Quancr 1992 0990.26), "only Peru and BrJzil remain locked in the old protectionist mode " By themselves, Ihese tr:.tde liberalizations offered a good deal of motivation for the increased intellectual property protection that occurred in Latin America in the late 19HOs a nd early 199Os. Tnlde liberalization , after all, means that foreign products incorpor,lIing and pr<xluced by the kllest and most inexpensive technology will now he competing at home with domestically made goods a nd services. If developing-country producers wish to replicate this latest technology, they will often have to pay for it. The latest technology is often harder to acquire by subterfuge than afe p rocesses that ha ve been longer in the market in some fonn. Moreover, the policies of a nonprotecti ng consuming country can affect its businesspersons' ability \0 purchase such technology, even if they :.tre willing to p:l.y for it. When dealing w ith firms in such countries, fo reign producers of technology are caLllious about selling it because the prospective buyers may violate purchasing agreements w ith impunity. Sherwood (1990) dtes anecdotal evidence in which a Brazi li;tn finn's employees h:.tve approached companies abroad to gain costeffective tedmology. However, the firm's negotiations with the foreign source often came to an a brupt end when the source learned o f Brazil"s we;lk protcction for many innov'ltions. Increased competition from foreign technologies may also convince a ch;ve1oping country to seck domestic innovations as well, ;lIld there is much evidence to suggest that weak intellectual propCrty protection discourages innovation. In a Brazilian survey, 80 percent of 377 finns said they would invest more in internal research and would improve tra ining for their employees if better legal proteaion were ava il:lble (Sherwood 1990). While Brazi lian survey evide nce suggests that the lack of protection discourages innovation, U.S. survey evidence (Ma nsfield 1986) suggests that protection stimulates innovation. ProtectlCJ(l,sm in the Unotcd States had been nsmg during the 19lOs as well. belOte/he dcveJopmg COUfl/nes began /0 ra,se theHS And, at le;L~t in the United Statcs, there also appc:.--ars to be a high social r:ile of retum to innovatio n, considembly higher th;tn the r::IIC of return 10 the innovator (fo.bnsficld ct al. 1977) Latin America's widening technological gap in the 1980s Whi lt: their trade liherdlii'..ations alone might well have motival",'d L.uin American countries to tighten their intc!lL,<:t u;l1 property protection , the pressures for slI(.:h pro(ect.ion increased in the 1900s as the h"'chnological gap betwet!n L:ltin America and thl! developed cOlL ruries widened . One reason this technological gap wident---d so much in the 19HOs is hec:lUse terms of 1'~lde shocks and <.Ie])! probkllls forced :I slump in Ullin American investment that lasted nlllch of tht! decade. New clpital goods wcr" not replaci ng 01<.1 and , as the :lver::lge age o f Latin American plants and eqUipment [-o..,e, the region's ca pital stock heGlnle in crea si n ~l y obsole te. 111e technological gap widened further ix.--causc, while investmem was plunging in LLtin Ame rica, the de\'c!opcd countries were e ntering a period of partil:lliarly r;lpid technological advance (B:l\:~ r 19(3). During this period. spending on research a nd de\'c!opment was growing m uch more rapi<.lly than gl'O:"" national pnxlU~1 in all major OECD coumries, and prodllct life cydes were hecoming shorter. Mon.-'()ver, while the productivity of knowledge was increasing relative to other input..;, the cost of gener:Hing it was also rising. Accordingly, knowledge became increasingly pri vatizl;'d (Mody 1990). In light of these ch:mgt:s, frischtak 0990, 62) nOles, "The tr;Jde-off ht:tween encourJging the d iffusion of existing technology thrOllgh unlicensed imitation and stimu];lting the cre:ltion of new technology becomes steeper over time· This la~t argument became particularly compclling in a Latin Ame ricm context during the 19HOs Latin American cou ntries' Mscvere loss of compctitiveness· was worsened hy their M inCOI_ padt)' to carl)' out the Str\lcturJI mtXiifications of an institutional char-Jcter that would h:lvt' e na hled them to develop sources from which they could c~ate :lnd disseminate new tedmologit!s· (Barherd 1990.92). In sum, at the smIle ti me that foreign debt problems had led m:tny devcloping countries to seek new avenues 10 ~arn foreign exchange, these I.:ollntries were suffering from incre:lsing technological backwardness. Moreover, it had become ohvious Ihat m:lny o f these n:u ions' dcteriol'dting competitivene:.s resulted , in part , from institutional problems that imped ed the development and trJnsft:r of technology In the new :nmosphcre of f;dling Latin AmeriC'J.n trade ha rrit:rs. with its rt:suhing increases in local compet itio n from imports. the incentives for technologically competitive domestic products lind forrm of production ha<.l riscn In any CISe, Ihe debt problem and techno logica l back\\':lrdne~ were important faclor:. tbat affected L<ttin America's ahility to rc."fX>nd to foreign compet ition, and they may also ha\'e been importa nt in changing developing nations' responsiveness to U.S, initiatives concerning Third World intellectual property protection. Has U.S, policy caused the l .."uin American move toward stronger inteUectual property prou..'Ction? • Gadbaw (/968. 284) nole$ 1/181 'In 1986 and 1987. Ihe RepuI:Jhc 01 Korea (Korea/ (){)Xtcd II1to law a scncs 01 reVISlOflS /0 ,IS Intellectual ptopflft)' regmt] /hal pt()Vlde a dramallc ellamp/e 01 U 5 success in fXon'IObI'I9l1llellfJcrual ptOptUty reform tnrough the COI'I'IbItl«J use 01 negoi181JOfl end the /hIeal 01 trade lelalla/1Oft - 1h8 U 5 ~I malll/alflS 1/181 Cflforcemenl remaons a problem. however (East As an ExcciJ~ve Reports 1990) In 1990. MexIco pUJposed sweepmg rclorm$ 0111$ ifllcIJcctua/ property laws 1/18/ CCr'Is/'/u/8d -I) vrrue/laundry /lSI oIlhe present law·s majorlnad6Qu8clCs-(Elfl$lem 1991, 134} In June 01 199/. MaIoco passed IfIOSB relom'ls 11110 law Consid ering th:lt the United 5t:llc,.; has not only empowered itself to inflict trdde retaliations upon countries that do not fully proh.:.'<:t intell~c llIal pro perly. hut also 10 impose sch...-dules hy which tbe U.s. Trade Repre~ntati\'e is sllppo..;cd ({) inflict the m. increased Third World compliance m i~ht be expected. Moreover. a prcponder:mce of Third World ch;ln~es in intellectual propeny protection h:l.'; occum:d since the pa.ssing of the United States' 1984 TrJde ;lIld Ta riff Act and of the Omnibus Fede ral Rcsc l"\le Ibnk of Dallas Trade and Competitiveness Act of 1988 Gadbaw and Richards (1988, 21) offer the mgument. writh.:n unck-r thc heading of "Tl.lde Pressure," that "only when the potenti:J1 loss of other economic benefits i.~ introduced into the <'-'<I.u:lIion have government.... of the nations studied concluded that il is in their interest to refoml their intellectual property protection regimes.~ \\' hile the putcntialloss of economic benefit.s surely motivates countries to strengthen their intellectual property Jaws. the perception of potential loss appears to I~ a more complicated phenomenon than Gadb:l\v and Richards' chara<.·terization of it. If U.S. t"lde pressure is the diret:t ancl preponderant reason developing countries have strengthened their intelleCTual property Jaw. what expbins the laggards! As an example, consider the four countries t-.kxly notes :IS :lffected by the 1984 Tl.lde and Tariff Act. By the end of 1990. Mexico and Korea had both greatly liberaliZl.'d their trade regimes (see above) ;tnd had takt.'n steps to markedly strengthen their intellectual property laws.~ while B"lzil :md Thailand had donL' neither. ~ The recalcitrance o f Brazil and Thailand'" does not mean U.s. trade pressures arc ineffective any more than the continuation of speeding in tht: United States means thm traffic laws are ineffective. But this recalcitrance docs suggest that other fa<.'{ ors are invoh'ed Here. a hro;ldcr definition of "rotential loss" than what G:ldbaw a nd Richards sel~m to he using is in order That is. cou ntries that impose b;lrriers to foreign competition may more fully serve themselves by making backw;ITd technology c hea r than hy allOWing the most modern technology 10 be expensive, But when «l.~ in the case of 1'.'lexico) they c hoose to open their economies to foreign competition, it is the loose intellectu;tl propelty regime that becomes more expensIve Thus, w hile U.S. policy encou rages countries to tighten their intellectual property regimes. the policy may be most effective with countries th;lt have chosen to open their economies to trade After all. these countries bave the strongest predisposition to tighte n their intellectu;tl property regllnes. 111 :Iny case. This argument has significant implications for the process by which U.S. policy actually affects other nation's intdlectu;ll property 1;1\\/s. This arguEl:onontic Revkw - Fourth Quarter 1992 menl suggests th:lt the dirC(."( ('ffeet of U.s. trade policic.:s ll lxm Third World intellectual property bws may not be as strong as the indillc"Ct effect, in which the u.s tl.lde policies simply influ ence other nation'S trade policies. That is, to lhe extent that U.S. trade policy motivates Third World countries \0 lower their Irade ba rrio.:rs, then U.S. policy indirectly motivates Third World cou ntries to tighten their inlt;.'lIectual property regimes. If trade libeml- • Although Braz,/ d id adopt copyright protection for softwale m 1987 and enforcfXi it (Sherwood 1990. 196). until very reccmly there have been lew Other cl1anges to Bmzii's intelleelual property regime since the ellrly 19705 In addition. -the BIOlZlban judicial sy~lem does nol appear to proVl(ic an ellectlVe dotetrentlO violations of intelieclUal propertyrights · (Frischlak /990. 13) In 1991 , theofliceof the U S Trade RepresentatIVe placed Brlll'il on its prlOfity walch list of areas "where the U S faces '5eflOUS problems' WIth mrcllec/ual prOperty protectlOfl or market access (Truell and Lachica 1991. A 16) However. Brazil has very iatelybegun to libera/ue moteol its lIade policieS. including those on some hlf}h·technology ,mports As oIli1SI Feb!U· al)'. Brazil began to allow Ihe flee mporl 01 hard and floppy disk dflves and digital II()Icemail eqUipment Addlhonal products '.11,11 be subje<;tto llberalizahon as 0/ October 29, 1992 Meanwhile. some tlghtemng 01 re/auxt intellectual properly protection has a/so begun 10 occur In June 1992. Brazil announced thaI if would begin 10 protect -{ranchlse agreemenls - that may mclude a combmatlOfl 0/ trademark usage. teehmcal serVICe assistance. and other types of lechnology lIans/or included m the agreemenlS With re· gard 10 Thailand, Schumann (/990. 164) no/es that -In· frmgement of intellectual property fights seems to be an ongomg ISSue despite efforts by some ASian countries to suppress It In 1988. the ASia-Pacific CounCIl 01 American Chambers of Commerce VlCW Thailand as lhe worSt offfHlder • She adds thlll -tho United Slams Trade Represen/alive (USTR) leporl Of) 'Special 3D!' 01 May 25. 1989. aerermmed Thailand and India as being leaders of 115 'priority watch list' • '" This recalCitrance has had liS costs As a result 01 noncom· pbance WIth US requirements. Thailand /osl GSP prlVl' leges, lesultmg m a 5'percentlo 100perCfHli imporl duly IncreaS90f) $165 mi/11Ofl of Thai exports The United States also imposed 1000percent dUlleS on $39 millKXI of Brazilian exports. even though these exports were net related to prodUCIS or services on which intellectual propertyin/ringe· mfHlt was alleged Moreover. the umrfXi Stales proposed but did nol implement punitive tariffS on $105 mil//Oll of Brazilian cxportS as a sanction against &azilian computer pOlicy ThiS lasl measure was withdrawn with the passage of the Brazilian sofrware copyflght law mentioned in the previolls !oomo/(:: As!or the other penel/(es, ·exporters in Thailand and BraZil have said they can Jove With the InCreased duties • (Mody 1990.224) iSl11 I11I)ti":ltl.:" intdkctual P(\)PI..:1'ty C\Jlberva!islll. thcn prc:-.... urel'> for trad~' lihe rali.~m ;lR' pre:-....urel'> fc)r light intellectual prc l peny pro!ec:tion . II [n this l 'onteXI , howe"er. it il'> impor1ant to rememher thaI !he event.. of the 19HOs in the Th ird Wortd ll1oti\';u~xl intdk'CllIal propcny protection in any G1Sl..', brin ,\meriel 'l'> inc: reasing tl.x·hnologil'al h;Jckw;m.lm::~~';,lllS~xl h~' a l 'omhin;ltioll of c.k-dining in\'e . . tmt:nl thl're. ;ll'l'der.ning techno[OgiGII advance .. . in the First World, and Latin AmeriGI 's ()\\'n inno"alion-c.liS(.:our.lging in..titutional ..t11.tl1Ul't' -shOll ld he a !'>trong mOli\':Jtor when nation.. . in the region Inu!'>t f( K'ul'> on manubcturcl'> eXfXlr1!-i to p:ly tlwir fort:ign dehts, It h, of l'ourl'>c . diffic.·ult to kno w if thl.'St' prcl'>sure!'> alone would ha\'t: Ix.ocll l'>ufficienl to motivate Third Wortd l'ountries to amend their intdk..:tu:d proJX'n y [aWl'>, But it i!-i unllsu:ll(o find 1'00lOtr1t."!'i th:ll h:l\'t.' lighlt.'nt.'(llhest.' laws when trade lihc..'r.lliz:ll ion had nOI O(:culTed as wdl " W'hIIe IfI(1lfT!p(XfaflCe 01 who'll / ClalfT! 10 /)e Inc connecllCt'l ocrween trado pobcJes and Ifltctlcdual property reglfflCS seems not /Q be fully apprecJa/oo. " C9rla/llly has oecn f8COpfll, OO Pnmo Staga (19900) notes /hal ' /116 fofergn deb! CflS'S oocreaSlOg pfllla:e capItal flows 10 oeve oping coonl.oes. t>ega1M1 eJlpsnences WI/h /he regul8rury approach. ourward-or18l1100 aeve.opmenl strateglCS. and me ongorrrg /ecNrOIOt}ICSI feV'OluIIOt'!' are $O"TW 01 fne poSS!OIe cJlp/Bt1abons fa( /he mote ht:>etal post(lfC adopIed by many doIrcIopIng counmes or! ItIlallcctual prCJPf!rfY- /empllaSlS _I Fcden.i Rcse",'~ Rank of Dallas References Baer, Werner (993), ··U.S-Latin American Tr.tde Relations: Past, Presenl. and Futurc ,~ in Free Trade in North America. Expanding Trade Jar Prosperity, edt Gcr.tld P O'Driscoll, Jr. (Boston: Kluwcr Academic Publishers), forthcommg. £nsl Asilm Executive Reports ( 1991), "Intellectual Property Rights in Korea: Legal Framework and Enforcemc nt Problcms--Part II : Trademarks, CopyrighLS, Sor[warc, ~ 13 (February): 18-20. - -- (1990), M lntellectual Property Rights in Barbera, Mattia (990), "Latin America's Place in World Trddc, ~ CEPAI. Review 4 1 (August). Korea: Leg:11 Framework and Enforcemenl Problems--Pan I: Patents: 12 (November): 13-5. )je ristain, Javier, and Ignacio Trigucros (990), "Mexico," in Latin American Adjustment: /-low Much lias Happened?ed. John Williamson (Washington, D.C. : Institute for International Economics), 154-68. c., and Larry N. Willmo re (991), M Technologicdl Impons and Technological Effort: An Amllysis of Their Determinants in Brazilian Firms, ~ 71)ejollmal of Industrial Economics 39 (June): 421-33. Braga, Helson Butler, Alison (1990), "The Trade-Related AspecLS of Iniellt.'CI.ual Property Righl'i: What Is at Stake," Federal Reserve Bank of SL Louis Review N'ovembcr-Dt.."CCmbcr, 34-16. Cardoso, Eliana A., and Daniel Dantas ( 990), " Brol lil,~ in Latil/ Americal/ Adjl4Slmelll: /-low Much Has Happelled?cd. John Williamson (Washington, D.C.: 1r1stitute for International Economics), 129-53. Chin, Judith c., and Gene M. Grossman (990), Property RighL'i and North-SOuth Trade," in '{be PolitiC(lf £Col/omy of Intema liollal Tmde, ed. Ronald W. Jones and Anne O. Krueger (C..mhridge, Mass; 13asil Blackwell), 90-107. ~ lntell ect ual Collins, Sus.'n M. (1 990), M I.cssens From Korean Economic Growth," America n Economic Revieu' 80 (May): 104-07. Diwan , Ishac. and Dani Rodrik (991). "PmenLS, Appropri:ltc Tt."'C hnology. and North-Soulh Trolde," jOtlmal of InlematiOlwl E."cOl1omics 30 ( February); 27-48. Economic Rcvicw_ Founh Quane r 1992 Einstei n, Edwin F. (991), "Promis ing Developments in Technolob'Y Tr-dnsfer a nd Intellecrual Property Protection in Mexico," in North American Free Trade: Proceedings of 0 Conferrm ce (Da llas: Fedeml Reserve Bank of Dallas), 123-37. Evenson, Robert E. (1990), "Survey of Empirical Stlldies,~ in SrrE'nglbening Protection of Ime/lechlaJ Property i1l Developing COIWln"es: A Surr..>iO' of,heLileramf"e, ed. Wolfgang E. Siebcck, World Bank Discussion Papers, no. 112 (Washington D.C.: The World Bank), 33-46. Frischtak, Claudio R. (1990), "The Protection of Intclle<rual Propcrty Rights and Industrial Technology Development in Brazil," in inlellectual Property Rights in Scien ce. Technology, and Economic Peifon1lwzce: l//ten/ariollal Compari50115. cds. Francis W. Rushing and C1role Ganz Brown (Boulder, Col.: Westview Press), 6 1-98. Gadbaw, R. Mich::tel ( 988), "Republic of Korea," in /nlel/('''Ctual ProjX!11y Righls: Global Consen sus, Global Conjlicl? eds. R. Michael Gadbaw and Timothy J. Richards (Bou lder, Col.: Westview Press), 272-3 10. - - - , and Timothy J. Richards (1 988), "Introduclion," in Intel/eell/ol Property Rights: Global Corlse1lSUS, Global Co"jliCI? cds. R. Michael Gadbaw and Timothy J. Richards (Boulder, Col.: Westview Press), 1-37. Gilbert . Rj., and 0 Ncwbcry (982), "Pre-emptive Patcnting and the Pcrsist~nce of Monopoly," American Ecollomic Revi£'w72 (June): 514-26. " Hoay. T.B. (1988), M Thc 1987 Sing.lpon: Copyrighl ACI and If:-> Economic Consequences ,~ APIA nolOJ.(.v. wtd Economic Peifonnance: IllIemational COnl/Xlri.m tlS. eds Francis W. Rushing IAsia-Pac ific I..tllJ),ers A-.sOCitltioll/jollmal, and C<1f()le Ganz Brown ( Bo ulder. Col.: Westview Pres."), 203-39 December. Ill-1 5 Laird. Sam. and Julio Nogucs (198$) , '1",,1<.11: Policies a nd the Ilighly Indcbtt:d Countrics. ~ IVorld &lIIk Economic Review 3 (May): 241 - 61 . Latin Americall MOllilor( 1991), -The Doors Arc Nunnenkamp, PCI~r, and Ugo Fas:.mo-Filho (986), ·'Manufactun..--d ExfX>rt Perfonnance of Brazil: Analysis of I)c:lenninant.. and ProspeClS." Kid Working Paper, no. 269 ( Kid , Gemlany: Kiel Institute of World Economics. St:ptembcr). Opening.- 8 (Scplcmher): 937. Lesser. William (l99(}). "An Overview of Intellectual Propcny Systems, " in 5trellgthelli l l& Proteclioll oj Illtelll!ctlwi Property ill Developillg COLlntries: A 5111vey oJthe Wemlll1r!, cd. Wolfgang E Sit:beck. World Bank Discussion Organi7_ation for Economic Coopcr:Jtion and Dcvclopmt:nt (989), "Economic Arguments for Protecting Inlcliectll;Ll Propcny Rights Efft.'{.'t ivcly:' TC/WP (88)70 (P1ITis: OECD). Prebisch. Raul (959). 11)e ECOllomic Developmenl Papers, no 112 (Washington. D c.: The World Bank), 5- 15. 0/ Latin Amen'c(/ ( /lid Its Prillcipal Problems (Nt.:w York: United N;IIions). Levin, Richard Co, Alvin K. Klcvorick, Richard R Nelson , and Sidney G. Winter (987), -Appropri:lIing the RctUn1S from Industrial Rcs<::arch and Oevclo pment ,~ Brookillgs Papers 011 Economic Activi~V ( 3 ). 783-H20 - -- (950), "Commercial Pol icy in the Underdcvdopt.'<.i Countries.·' 1be Americall EcoI/omic Review, Papt!fS and Proceedings of (he Americm Economics Asso ciation , 49 (r-.'Iay): 251-83. Maclaughlin. Janct II. , Tilllothy) . Richards. and Leigh A. Kenny (1988), "The Economic Significance of Piracy ,~ in Intellectual Property Rights: Global Consensus, Global Conflict? cd" R. Michael Gadbaw and Timothy J. Richards ( Boulder. Col : Westview Press). 89-108. Primo Braga , C:ulos Albcno 0990a), "Gu ida nce from Econo mic ' 111COI)',w in Stl't!llglhening Prolection oj IlIlel/!'CI/I(l1 propeny ill /JelJ(>lopjug COlmtri!'S: A SlIn~)' oJlhe Literature, cd Wolfgang E. Siel~ck , World Bank Discussion P<lPCrs. no. 112 (W;lshington . D.C. : The World Bank), 17- 32. Mansfield, Edwin 09R6), "Patents and Innovation: An Empirical Study," M(lI/agement Science, Februal)', 173-81 - For a nd Agai nst Intellectual Property Protection,·· in StYf!llgfhenl118 Profl.'Ctio ll o/Inte/lecfIlal Property ill Developing COllflfries: A Survey oJlhe l.itemtllre, ed. Wolfgang E. Siebeck, - - - (1985), "I low Rapidly Does Industrial Technology Leak OUI?"jollnllll oj Industl,al Economics 34 (Deceml.x:r): 21 7-23. - -- , John Rapopon, Anthony Romeo. $:J.mud Wagner. and Gl."'Orge Beardsley ( 1977). ~ Soc i al and Private Rates o f Return from Industrial Innovations.~ Qllarterlyjoumal oj Ecollomic.s 91 (""by); 221-40. fo.kxly. Ashoka (1990), ~ New Inlcmational Environmenl fo r Intellectual Propcny Righls,~ in Intellectllal Property RighfS ill Sciellce, Tt.'Ch- 2. - - (1990b), -The Developing Countl)' Case World Bank Discussion P:lpers. no. 112 (Washington, D.C. : The World Bank), 69-87. - "The Economics o f Ime[]ectual Propeny Rights and the GAlT: A View From the South .~ VanderiJilfjolinlaloJTra1lSnational Law 22, 243-64 - - (989). Schankcrman, M'lrk ( 991). - How V.tluable is Palent l'rol\!ction? Estimates by Technology Field Using Pate nt Re newal Dala ,~ NBER federal R~rve Bank of Dallas Work ing P:l per &:ries, no, 3780 (C<lmbridge, Mass: National Bureau of Economic Research, Ju ly). Tr.lde as Congres.o; GelS &1 to Vote o n Extending Negotiating Powers, ~ Wall Street Jou rnal, April 29, A16. United Nations 09(4), 'fbe Role of P(llents in th e Schumann. Gunda 0 99()), MEconomic Development and [ntelleclUal Propert y Prott."'Ctio n i n Southeast Asia: Korea , Taiwan, Singa[X>Te, and Thaibnd. ~ in /1lIellectlla l Property Rights in Sciel/ ce, Techll%gy, and Ecollomic Performalice: Illlen/Otio flal Comparisons, eds. Frnncis W . Rushing and Carole Gan z Brown (Bo ulder, Col.: Westview Press), 157- 202. Vcssu ri , II.M .C. ( 1990), "0 Invemamos 0 Erramos: The Power of Science in latin America," World Developme1lt 18 (November): 1543-1553. Sherwood, Robert M. ( 990), Intellectua l Property (lIId ECOllo mic Developme m ( Bo ulder, Col .: Westview Press). W il liamson, John ( 1990), n )e Progress of Policy Reform i1l Latin America (Washington, D.C. : Institute for Intem :Hiona l Eco nomics). Tmel1 , Pt:tcr, ,mel Eduardo I...;.\dlica (1991), "Politics and Policy: Administfution Get'> Tough on Economic Revkw-Founh. Qu.an cT t 992 Transfer of Technology to Developing COllntries, Report of the Secretary General (New York: Un it(.'<i Nations). Kenh R. PhillipS Economist Federal Reserve Bank 01 Dallas Regional Wage Divergence and National Wage Inequality n n:~~m year,... COIlCt'1ll ahout ri .. ing wagl.: :md in(:01111: inequality in the UnilL'd St:l\es has he~:n expressed in Ix)\h til ..: l11a ..... media and the aCldemic I 1iteratun,;' Tht: rm:cii;1 :0.101"1(:." han: stressed \hl' um:vo.:ll nalllre of Iht: ).!:lin.. in w;lgcs and inC0111l'S during Ihe 19HO'.... wilh rq)orls Ih:1I the gains have heen cOllcent rated al the l OP of the distrihution :ll1d Ihal 1110:-.1 Atncrkan.. have :>ccn linlt:, if :my, mcrease in eit her wagl.:' or irKomc,' The acadc mit' lill'r.l1url' on Ihe SUhje<.1 h:l.o; IWl'n fo(.'u5(..,<1 on tht, po...:--ihl!.' Cilises of the riSl': :H;('ountcd for on ly 2. 1 pcrc~m 10 5 percent of the ri.~~ in nalional wage int.'(!u:llity The regional dfe('( is similar in lI1agni\lldt: 10 f:Ktors rebling to the industrial structure but is rdativcly small in ('omp:lrison with r:Klof," n,:lating to education, occupation, gender, and agt:, Givt:n this result, tile regional ('OI1\'('rgena: that will likely t:lkc pbcc in the 1990s is unlikely 10 h:t\'e :1 significan t imp:lct on (}\'I:r.lll wage inequality in tht: United Stales, National wage inequality b hor-Mlppl)' fa clor:'-sudl as ch:mges in tht: l'dllcalion. :I).!C, :tnd j.Cl'mk·r of the work forn.:·:Ind lahor-demand f:I('lor.-.-such ;1:' changes in the industry :ll1d (Kl"Ilpati<)Jl IllIX , Although sc"er:.11 ret'ellt ,~t lldi\!,~ !la"e IXllr1lt.'d to :-0111(' pos:-ibk ca u,'>t.',~ ror the rise in inl'qualit}" much of Ihe riSt.' remain~ unexplained ~ In tllb :'llIdy, I '>I,:ek to :Idd to [he liter:.nun,: by addre!'>Sing ,h\.' qlle:-lion of how much of Ihe ri,~ in LS, \\';lgc inequality during the: 1980s can Ix: explained by :1 divt.'l',gl.'llt'l.' in \"a~es acms,~ rt;'~ions. SI.'\'er:.d ,~tudit.'s h:l\'e :-hO\\'1l Ihat :Iftcr II long pl.'li(xt oj" t'olH'ergcllce, fl.'gion:1l W:lgl'S diver~l.::d during thl.' 1980s, Thest.' ,"audie:- also indicale Ih;1t Ihl.' din:rgencl~ was likely tt.'rnpor:.lry :Inel th:n \vages an..' likely to ft:llIm tOl'onvcrgl:llce. [I' a siro ng link exbb helween the regional divergence ;lnd Ihe n:-.c in U.S, w:lge iIK'qualit}', thl'Tl a r('tum 10 ~gional t'oll\'ergl.'ncc in I he 19lXls ('ould put signifk:ml dowl\w:ud pre!>,Mlre on \\ aw: ineqllalilY, t flnd th:n th\.' r<.'),':lon:11 ~h<Kks that (Kt'llrfl,..d during th\.' 198(), pn x hlt'l'd a p:Hlem of w:lge d in:rgl:.'ncc th:\! t'orn.:I:ltt'(l do!>dy with the p:utcrn of nation:ll wage int.,<[uality In a mathem:ltic:'ll dl'l'omposition of n:uional wage ine<[uality , howe\'er. I find Ihal Ihe di\'er).,:t.'Ilt'e of regional w;lg~s Economic Rc.'it·w - founh Quan~r 1992 In a ~\'il'w or t he liter.lture on wage in(.'quality, Loveman and Tilly (988) find that most fC!>carchers agrL't' 11t:lt the inequality of boor earnings has inCfC:.lsl.'d !>inCt: the l;lte 19705, TIle ('auses or thl: increa~ , h()wt:\'er, are not well undersl(xxI A\.'cording to Lo\'el11:m and Tilly, rat'ton; such as dmnge.~ in gender c()mpc:)!>it io n , age diSlrihution, and industri:11 :lnd ()('('up:Hiona l mix explain only :I sma ll part of the risc.: in c:lrnings inequality I WIsh to lhank.)0s8ph H HasJ8!J. ZsoIf BeeS!, STephen P A Brown. aM Lor, L Taylor for helpfu/cammenls and James L lledges for exccJlcnl fCSlN3ft:h aSSlslance For example SH S'Dfh ( 1992) snd StCl(l (/992) • ..w.te Sludles have Iovnd e fiSC in both II>C'OfTIC and wage 1tlfICIU8h1y. the 1100 nlCil$(NCS 8re quite d,ffcrM Income is a//cc/oo by NWeS/menilflCOfl'lC 8fKJ /lans/ers Household tnCOfT/8l$afl«/ooby 18ITIIIys,zesndSlfuclure In""s study. 1100/0; solely 8/ pie/ax w8l1CS. 8fId, thus. my focus 's OIl me s/ruclureollabet.ncome Labet .irJcom6d09sllOl fellee/trle lOUIl n:omt or /he (]YfJfal/ wctfare oIlf1dMduals 31 In more a mor!.! r!.!c!.!nt study, Gmhb and Wilson (1992) find th:1t fa<.1ofS n:lating to t.'(lucation and occu patio n played important roles in :1 rise in wage im:.--<Iuality in the 1980s, while workt.'r agf...' and industry had relali\'dy small imp:lCts. Gmhh and Wilson find, howcver, that much of the ri..<;/.." in inl.'quality in the 1980s W:IS unexplaint.:d. 'nley concill(.k thai Ihe ~pre:ld o f wl_ges anos... l.-xJul'ation groups is likely to cont inue to in<:n:asc in the future. causing further upward pres~ure on W:lg!.! inL"'quality The authors Gluti(m that furtht'r :m:llysis is needed to find a mo rf...' complt'te expl:mation for tht! rise in w:lge inequality Bound and Johnson (1992) Iligh ligiit three major w ag!.! structure devdopI11e1lls in Iht' 1900s: an increase in wage di ffcrcnti:ll.~ hy t'd ucation, a decreast' in the gemk:r w:Ig!.! gap , and- l(lr workt.·rs without a college degree-:m increase in the aver.lge wage of older workers rda ti ve to younge r workers. Uound :md Johnson concludl..' thai the principal reason for the incr!.!;I.<;/.." in wage differential by t..'(lucalion :Lnd the (k."<:rl';i!'>l.' in the gendt'r differl'nti;!1 was a skillt'd-1al>or-hiaSl'd te(:hIllJI(lgica l changl..'. The :Iuthors found Hull' to explain the rela tive-wage dl.>crl..'ase of younl-( workers with low t.'duc:lIional all:li nment TIlt' authors indiGLlr.: that wage differentia ls hy l.-due.ltion :I~ likdy to continue to increase, putting: further upward pre~ure on w:lge inequality Regional wage divergence One possible factor in risins:: national wage inequality that has received rdatively lillie :lttention is a divergence in wagr.:s :lCross fl'l-(io!ls. Browne (989) notes Ih:1\ "aftl'r three dr.:cades of J In /hIS SllJdy. wages afC fflCasured as annual wages and sal8fIflS 01 people 16 yeafsoI8(}eex older who worl<ed yearloooeJ. 1u1l-11fTIe ('(AFT). OF. more (XlJCJse/y. peopic who worked lilly "1)Cks ex mora durlllfllhc year and /hJIty·!we Of more hovfs durJT19 the week The dala afe laken from lhe Mafc;hCvrrem Popu/aIIOfl$urveys (CPS)oI me U S Bureau 01 /he Census 0uwIg me sl'UCfy perJOd. VAFT worl<ers Icplesenled 54 percenllO 6lJ percent 01 all sampled wor/o:. ers. and /he sal'1¥'ie SlZC 01 YRFT workers ranged from 41.T5()0fI 19711050.434"" 1989 AlfflOuQhavefage/Joorty wages recerwxJ durmr; /he yeaf !ex all workers afe (Xc-terred. rtus J(JtormallOfl is not avadabUJ Of) the CPS tapes gentle con\'ergem:e. re~ional per capil:1 incomes diverged sharply in the 19HOs ~ Sbe finds that ch:tnges in rdative wages Wl're the primary SOUTCC of c hange!'> in re~ional per (<Ipita income. If regional sh()ck.~, ~uch as the recessions in the oil and ran11 bell.. , caused wage," in low-wagl' afl..."..lS of tht' country to dedinc n:l:Ilivl' to the \\'e:rlthier :ue:I.... then this change could C..luse national wage inl.·qu:llity to increase. And if these shocks wl..'re tempor:.uy. then a rl'turn to regional nlllvergl'nce in the 1990.. (':outd put downward Pfl.'SSllfl' on wage inequality. Elx,rts ( 19H9) finds th:1I a d ivergence in rt'gional wagl's in the 19HOs likely was causcd hy n:gion-spt't'ific f:lt'tors, sll('h :IS the oil :1I1d fanll crisis, :lIld W:I ... not l11erdy thl.: uneven regional impact of changing n:tlion:ll p:ltll.:rtlS. For eX:Hllple. if thl.'re was a region:llconccntf:ltion of highly t.:duGIIl'd workt'rs, then the national rise in wages rt.:sulting frolll l'duGltional attainment cou ld h;lve re.'>llhed in rq.,:ional wagt.: divergence. EI)(;!rts shows that the rise in region:d w:lgc in(.'qu:llity gener-Illy "":IS not ('';IU:-.t.'(1 hy differences in regional shares of denl0W~lphic and industry factors. Clrlino ( 1992) u ......-:-. a liml'-serit.'s :Ippro::lch to delenllinc whether the n:gion:11 e:tmings divergence in the 19HOs W:IS c~u!'ol.!d b}' ()(;!rll1:1nent or tempor..lry shocks. He concludes that the divergence was most Hkdy (::Ju~ctl h y telllpor..lry ~ho(.'k~. such as the energy and agricullllr.ll shocks, and that the divergence docs not repn:sent :1 reversa l of the long-term trend in regional convergence. Given thl..' results of Eherts' and C;lrlino's studies , the region:tl wage divergl.:nce in the 1980s was likely tempor:rry and independer1t of naT ional demogr:.rphic :Iml indust!), f:lClOrs. T hese results indicate that if rel-(ional divergence in the 19ROs had:l strong impact on national W:lge i nequality, then a return to regional w.Ige convergence in the 1990s could pllt ~ igni lk.rnt clownw:lrd pressure on wage inl."'quality Describing U.s_ wage inequality and regional shocks As shown in Figure I . wage int-quality in the United States in<:n:ased throughout much of the 19HOs.' The ll1ea~llfe of wage int.'quality shown in Figun: 1 is the population-weighted 111eil statistic. (For more information ahout the C'..liculalion and Figure 1 Theil Measure of Wage Inequal ity '" '" '" ". '" ,.. "" ,., '" ,'" '" "" '~ ~~~~~~'--r~--r-~'--r-' .. 1977 1978 li79 1\180 11111 lN2 1933 1934 1 ge~ cr.lsht:d and rc:.1 estate markets tumbled. In contr:.lst with the oil :md farm a reas of the t:ollnlty. the defense dependent regions grew at a hC'.lhh)' pace during the 19ROs (FiRIIIT! 3).~ Between 1900 and 1987, nalional dt.'fcnse outlays, in nlll~tant dollars. increased :11 an :mnual r:.IIC of 5.3 pcrt:t'nl, and total defense employment incfC'Jsed :11 an annual r.lte of 4.2 percent. 'l1,e military buildup was centered in privale-st.-'(:tor companies thai proolLce military equipmcnl. Private-sector dt:fcnst: employment grcw 7.8 JX'rccnt annually, whilt: public-scctor mil itary personnel grew only 1.4 percent annu:lliy.9 The slrong surge in defense 1986 198 7 19M 1989 NOTE: This mea syre IS defined Irl aq yallon 21n lhe bo x Karoly (992) SllOl'l$ thaI len d,lIarem measures of wage SOU RCe OF PRIMARY DATA US Depal1ment of Commerce, SlJreay of ltle Census, Cu rrenl Populallon SlJrvey, March lapes, mCQIJIJ"'Y all increosoo durmg the 1980s For example. Blank (1985) finds 11101 -the income distribul.on narrows m I!rnes of gfowtll, pnmari/y because 01 large clasl,e'l,es of labor market mcome among POOr profX'rtit':-' of thi:-. mea ... urc of im.""qualiry, SI..'C the Ix)x titb.! '·lJ..;in~ till' Popul:nion-WeighKxI '1l1t:il 1\1t.-:I.->UTt,.' of Im:qllalit}',-) A:-. s hown in Figure I. wage int.-quality inaea ......d from 1978 10 1%7 and th~n dt."t'liru.."tl l1H! ri-.c in wage inequality ~hown here is consi.... tent with other me:ISUR-"S of wagt' int.·-quality.' TIll,,' rise i:-. wage int.'"qll:lliry in the early 19H(b i... at le:IM panly explained b}' the two rtx~ssions that o<:<:urrcd hetween 1980 al,d 19R2. I'rt.'vious work has shown Iha( w:lge inequ:llity (~tpically increascs durin).\ t:yd it::d downturns and dt:cr~ases during expan.~ions , ~ 'Il,e increase in inequality during the long expansion from 1983 to 19l'0 is more diffinllt to expbin The rise in waW': int."tluality. ho\\,evt:r, matches dosely w ith the economit: ch.:dine in the fann and oil st:ltes and (hc t:Ol'l'cspondin~ rise in the defense st:lles. As shown in Figl1l't: 2, employment in the f:trm :11,d oil ,..t'ltt:S gt:ner.dly grew faster (h:1I1 the national a\'cr:.lge during Ihe 1970s." But durinl-{ tht., 19HOs. tilc"c regions we:lkened sh:lrply High debt Ic"":!.", com hi ned \\ ith rising interest rates and railing farm cxpons. pushed thc f:tm, el'onomy into;1 n..'Cc~si()n Ih:lI persbtoo until 19H7: The I..'nl..'r~y :-.talc:- also t:ntcred a recession following a !X'ak in oil prices in t::Ir1y 1981. FollowinS-! some impro\'c1lll'r1t in 1984 ;lOd 1985, till' t:n~rgy slatcs s111Tcrt.'tI:l sharp hlow in 1986, when oil priccs F"cono m i<: Mevicw - Fo ur1h QUilncr 1992 heads 01 housc/lOld for whom both wages and hours mcrease during 8COi'IO'llOC upturns • Also, Bluestone and HamS()(> ( 1988) lind Ihal tne proporhOn 01 year-found fut/· Illne workers who Cllfned low wages IlaS a stiwsllCal/y Slgmflcant neQB!I'I(Irc/IJtronsfllP Willi cycllCalmovemcnls III gfOSS nal","l1/ p'00uc1 The farm Slates we dclrned es delMrlf} a/least 5 2 perccr>/ 01 lherr gross Slate prodvct If) 1977 from famwng (The nai/OfllllsVCtBge was 2 6 pefCOflt ) These Slates afe Ar/o:an$/IS, Idaflo, Iowa. KlfllSas. ~1InncSOla, MISSISSIPPI. Monlana, Neblaslt.lf. Nonll Dakota, South Dakota. and WISCOf)ssn TIle energy slates are d61ined as t/lose Illal W(1(Jld s11ffer CfIlPIOyrnenl Iosscs from If decline 10 energy prICes, as sflowrJ If) 8rown and 1/1/1 (1988) Tile energy slales afe Aillska. CoIorado,l OUtS/ana. NewMexlCO. OKlahoma. Texas, West Virginia. and WyomIng (Kansas and Norm Dakota /lISO sul/crC(! emp/QymCn1 losses 'rom an energy-prICe dcelme but wero not Incl!ld9(j Mre because mey were cons,dared pnmarrly agncullyral states) For a d,scussson of/he reasons for the larm recessiOO "., the early 1980s and ,IS sybscqucnt recovery lhat began III 1987. see Dr/lbcf1SlOIland Baf~ema (1990) T1ltt def_ Slales ware defrned as SlillCS (plus tnc DlstllCl 01 CoII¥nbta) /flat rn 1991 /lad ailaaSl3 percentol t/lerrworlc force employed If) tile pnvale defense Ifl(1uStry The defcnsttSlal8S arc ArIZOna, CabforfWII. CcnnccIJCt.Jl. MatyIand, MassachusetIS. V'lfgonia, find WBS/IIflgtcn • FIfJIIICS lot defense spend""f) and employment were /aken ffom pages 61, 229. and 230 01 U S Calgress_ Office T8ChrIoIogy Assasstnenl (1992) " 01 Figure 2 Economic Performance of Energy and Farm States Figure 3 Relative Economic Performance of Defense States p""",, Percent , , ,• , , ._.. o +---~----~~-----¥~---- -, -, " ~ .. ~ NOTE: The lines are eql,lal to the year -Qver·vear percentage growth employment In the larm an(! energv stales mlnl,lS lhe year· over·year percenlage 01 growth In U S employmenl In NOTE· The ~ne <5 equat to the year·over ·year percenlage growth in employment in the doefense stales mll'lus Ihe year·over·year parcenlage of growth in U S employment SOURCE: Bureau 01 Labor SlahslJCS. ESlabhhrnenr Survey SOURCE : Bureau of labor SlallsloCS. Eslabl<shment Survey spending boostl...'d employment and wagL's in ;treas of the country when: dL'f•.:nsc <'·OI\1 I);II1II:'S were located. The weak pe:rro rmance of the Emn and energy n:gions reduced their rt:btive W~I ~L'.s. whilL' the relative wages of the dcfen...;e :-o(ales roSe. ,\ ...; shown in Figure 4. from 19MI to 19H7 febtivL' wages in the energy states do.::dint.'d from 102.H percent of the national average to 97.7 percent, ~lncl relative wages in the farm ~ta l o.::s declined from 94.3 percent of the nation:ll ;lVer.lge 10 H9.3 percent. In contrast, wages in the defense slalt:s grew from 105.6 percenT of the n:nional average in 198110 lJ03 percent in 19l:S7 During the bte I980s. deh..'nse spending began to dedine. The end of the defense buildup had a detrimental eff\;.'(:l on employment growth and relati\'e wages in the ddt·nse states. as shown in Figures 3 and 4 During this period , a ~r.jdu;JI n.."Covery was occurring in the fann and oil regions of the country, :lIld cmployment growth impron."'<.I in these areas. As shown in Figure 5, the dispersion o f ;lver;lge wages acrns:. rt:gions increased sharply betw(.'Cn 1978 and 1987, and suhsequcntly fd l from 191-17 to 19K9 The di\'t:r~ence in regional wagL"S from 1971'110 191:17, ;In<.l the ... uhseqllcnt com·ergem·e, i:. con:.is!ent w ith first the increase, and tlll:n the <b:l ine:, in n:Hionai w:lge inequality in the 1i)HI)...;. 34 A decomposition of the rise in wage inequality While the timing of changes in region:11 w .. gc dispersion sugge.~ts a link to n:nion:ll wage ine4u:llity, a decomposition of national wage inequality shows that region:iI w;lge divergence had only :I .~mall direct imp:rct on the rise in nalional wage inequality. Although Iht: dispersion in average waJ.lCs :KroSS the four economic re:gions increased ...;harply hctwccn 1978 and 1987, :lIld from 1982 to 191-17 ( FtRlfre 5"), the effect of Ihe increased dispersion on national wage in\;.'quality \'"'3S s mall in IXlIh periods. The Thdi mt:asure of inequality <.0111 be decompoSl..-"(1 into fWO main effects: inequality Gluscd by inequ;ililY in avcmge wages :leros.'> groups and ine:qualit~, caused hy wage inequality ft.xk:ral Rt.'SCrvc Ban k o r Dallall within groups. Ooth of these effects (·an then he divided into a prim:lry cffct:1 and the cffeCI of changing shares. For example. inequalily thaI rc~ults from inequalilY of aver:lge wages across group:. (·ould inerease ('ven if Ihe eslim;Ilt.'<.I mean wage of each f,lroup remained consl:1.nt. This increase would occur if the employment shares shift such thaI the weight givcn to a high- or loww:lge Woup increast.'S and the weight given to an aver:.tge-wage group decre:lses. Similarly, O\'erall inequality could incrca'iC if wage incqualilY was constant within each group hut the employment s h:m: r()~e for :L high-inequality region :lIld (k'dinec! for:L low-inequ:ility region. The Theil me;lsure of inequality increased from 0 177 to 0 225 from 197H to 1987, from 0.199 100225 from 19H2 to 1987. and then ueercascd 10 o 199 in 1989. T:Lbie I shows the decomposition of these change.s in Ihe Theil into the Iwo main effc("\.s ;md then. within the.se m:tin efft.·cls, thaI pall that results frorn ch:mging employment shares. A... shown in the top half of Table I. between 197~ and 1987 the n:ltional 1'11<.:il incfCaseu by 004HI. ,lIld the change in the dispers ion of a\'er.lge wage~ aero....... region... only a(·(·ounlt."(1 for 0.00 1 (or 2.1 percent) of Ill\..' in(Tt.:a:.e. Almost ;lllthis ch:mge was hroughl on hy W;lgt.: ch:tnge~. not by changing employmem Figure 4 RelatIve Regional Wages (Regional Wage Divided by U.S. Wage) '" '" ,,. ,,. ". ... .. ,~ , ~ .......... ..... ..... .. ..... ". --- .............. " ...... ....... "u~~~~~~~~~~~~·~··· ~_s-. ' lin ,1178 ,lin ,9110 '911, ,9112 1983 ''''' l!i8S 1_ 19$1 1988 1989 SOURCE OF PRi MARV DATA Bureau o1labof Sta\lSbcs. Es,atJloshment St.>rYey Economic Rl'VieW _ founh Quancr 1992 Figure 5 Dispersion of Wages Ac ross Economic Regions -..... . " ......" .....". ro" ...."" .." "" 1---, "" w,,+-~~~~~~-~~~-~~~ 1916 1979 11180 1981 11112 1983 HI84 ,985 1986 1987 1988 1989 NOTE : The meaSlire 01 dIsperSIon show he re is defined In equation 5 on !he box SOURCE OF PRIMARV DATA Bureau of Labor StaMt]cs . Establishment Su rvey share:-:. Excluding the rccessionary periods of the late 1970:. and early 198Os, increases in thl: variation of ;lv(.'r.lgl.' wages anoss e(.'onomic rl:gions :.Iill accounts only for 2.7 percent of the increase in national wage di~pcrsion. 1\·lost of the risc: in wage in''qualiIY is left unexplained by the divergence in average wages acros.... re}!ion.". Incrclls(.'d wlIge dispersion was moslly a result of inneased wage dis]Xrsion within regions. The increase within regions gives little information ahout the causcs of the risc in wage inequality. As shown in T:lble 2, wage ineqwt lity, as llH::asllred by tht' regional Theil statistic, rose in all four regions from 1978 to 1987 and from 1982 to 1987. During the period from 1982 10 1987. inequality in the farm a nd oil states increased at a significantly fa ster p:lce than in the defense slates and al .1 moder:ltely faster pace tha n in the renl<lining states. While Ibe shocks 10 Ihe fMm. oil, and uefense regions likely played ;l major role in the diver~cncc of regional wages, olher factors likely also played a role. To account for other faclors , I also look at the efft.'"C1 o n n:ilional wage inequalilY of the wage divcrgence acros... all fifty states and the DiMricl of Columbia. As shown in the hellom half of Tahle 1, the ovcrdll Cfft.'<..1 does nOI in- " Table 1 Decomposition of Theil' Measure of National Wage Inequalily Decomposition b y Eco nomic Regi ons' 1978-87 1982-87 1987~ .048' ,0260 - 0256 Change in inequalily 01 wages across regIOns (Equals A + 8) _0010 .0007 -0003 A Due to change in weight given to each region .00005 00003 .00000 B. Due 10 change in wages 00098 00069 - ,00030 .0471 ,0253 - .0254 -.00005 - ,00040 ,00002 .04709 02572 -.02539 1978-87 1982-87 1987-a9 .048' 0260 -0256 Change In Inequality of wages across states (EqualS A + B) .0016 .0013 A Due to change In weight given to each state .00005 .00009 .000<l0 B. Due to change in wages .00158 00121 -.00040 .0464 .0247 -.0252 C. Due to change in weight given to each slate .00 122 .00002 - .00008 o Due to change In state ,04522 02474 - .02513 Change In Theil (Equals 1 + 2) 2. Change in inequality within regions (Equals C + 0) C. Due to change in weight given to each reg'on 0, Due to change In regional Inequality measures Decomposition by Siaies' Change In Theil (Equals 1 + 2) 2. Change in Inequality within states (Equals C + 0) Inequality measures The Thetl SIaIISIOC Increased 21,' pe'ce<"I! "om 1978 to 1981, I/rom 0,1n to 0 225) 13 I pe'09flt t,om 1982 to 1987I"om 0 199 10 o 225). and declined II ,5 percent from 1987 to 1989 (trom 0 225 to 0 199) For m(ue ,nfOfmillllon on the ca~lallon and 0ec0rnp0s.11Ofl of the The' l. see the box , The lour ec»oomoc r&9'O'1S are the energy Slates. the farm stales. the defense states. and all other stales (including the Orstnct of Columboa) For more Inlo,ma~lon on the slates Included in each regoon. _footnotes 6 and 7 IllIIM1 text ft-"'dcr:1I Rt-"S4.' n ,c Ij:lllk "r ();lII:o" Table 2 Average Wages and Wage Inequality' Within Regions '98' 1978 Regions Wages Energy 14,287 13,424 15,625 14,475 Farm Defense 0100' '987 Theil Wetgnt Wages Theil Weight Wages Theil Weight .'98 11.9 22,4 55.7 24,785 22,655 27,993 24.807 12,0 .176 ,215 .208 .205 ,189 .250 9.9 20.200 18.962 22.020 19,724 13,2 '91 ,242 .221 .215 24.1 169 , Data a re lor year· rouod , lurf,t,me worllers. The Ttl(!!r measure of rnequahty 9' 23.4 ... 0 9.3 "' 5 IS described 10 the bO_ SOURCE OF PRIMA RY DATA U 5 Depanment of Commerce . Bureau ot the Census, Currenl PopulatIOn Survey , MiliCh tapes { f\',I"\' .. i~ n ilk;l nll~ \\ illl 111\' I. l r).,:"' f h rL,:II, do\\ n. O nl~ :t.~ IK 'I'\'l 'nl (If Ill ... net lIH f\ ·:t .. V III n'lti o n ;l1 invqll,lhl~ IWI\\\'l'11 I<)- /'i .Intl I ')I'C (';111 h •.: dirL'dl\ al1nl ltlll'd 11) .In In, r\·,I"l·(1 dl~IK'r" I ')11 In .1\ L'1:lgL' \\:Igl'" ,Inl)''':-- "1.11 \'" Excl u d i ng t ill' \'{·'l·.... i(lI1.try r'll.:rilllt.. ( If Illl' I.l1l· I ()- C):-' ;Intl L':ld~ 1')H(h. IIl l f l :. I:-" ... 6 p "'H"l'nt 0 1 till' irllTl.';i"l' A H'(h l('l ioll in lilt: ·" I,.. in lhl' \ .lri,1I10n "f \\ ,Igl" ,II n "1.lll'" .. Ii ll onl~ ,llUllIIlI" for:; r'll.:r".:1l1 (If ll1l' int'rl':I"l' III 11,111011;1 1 \\:I.l:L' d l"rM,:r,ulIl ()n(l' ,1).:.11 11 . IIlL' IIKn.::I:--\'d \\ :tgl' dl'JX'l'-lOIl \\:1' !llo" h .1 rl" lI ft 01 incn.:;r-.\ ·d \\ :I).:l' dl"JX'1'-1I11l \\ullin ,UI\'.. T he ri'l' in \\,1).:,: inl'qu:rl · il~ \\ ilhill ,LUl " \\.1'- \\ idl:"pr\·,rd. \\ 1111 \\ ,Igl' )l ll'l lll , l lil~ n .. inJ.: in fo rt y-Iollr ..I;IIL" 1'1'0 111 J') - K to II)S- ,In ri i ll lh ll1\,wH:n "1;11\ '" 1'1'0 111 f')Xl1 0 191';'1h ..· dl\ \ 'rg..'I1('\' 0 1 \\ .lgl·" ,Kro,," ..1.11 ..', pl.l~ cd a "Ill,tlkr rok in Il,Hional \\':lgc Ir lL'q llalil~ Ih;ll1 d id "l'I \'r;1I ( 1IItt 'f f;l\'lo l'''. :--lIch ;b ;1 rbt: i n !Ill' \ ;lri; II)('t: (,f \\ : I g~' .. :I<TI J:--." g n IlIp.. d:l...... int:d hy c , iJ l( ;ll i"Il , (In up:u II 111 :lI1d ,f .~(· Thl' c OcCI of ,!.!t:nri('1' LK lnr .. \\, I ~ ;l1 ..() "lnl n~l' l' lll; m the 1t:).ti,)Jl: tf 1:1l'1(11''', IlUI in 11 1(' "]lP().. il l· d il'\ 'l'I j(l[) A rl'dul'ti(lll in t ill' \\.I l!t: j.l: lp IX'l wl't'n 11I1'1\ :lnd \\[lll1l'n act ed II I Il'du( t, \\.Igl· l! lL'qualil~. ~'I 1'1l 1\'h ilL' IIt<-' .. ha rl ' 0 / \\tl l11l.' 11 in .IIL' \\(lr k fOrt'l' grt:\\ 1'11\' rl'gional 1:.'11..-l l \\,(' ' II n iLlr Dilly 10 IIt;! 1 of l it.,: dl:l!\).tln g in dlh tri.rl " l n ll IUrl: o f t ill' \lor\.. force \ ~ ~ ho\\ n III T.lhk- :\. "'dIlCliioll .11lL! on:up.ltlorul i.r('lor'" "\'l'lll 10 h,11 l' pl.ly ..-<I the b rg(' ''( rol l' III 11lL' li'-4,.' rn \\ agl' In ..:q u;ll ll \ ,. hUIlI 1,,,,,110 19:-1-. :rn Il1In,':I",,' ill I ll\' d i"pcr-it)1l of 1\,lgl" ,I("n),..; \'dUl.llI0Il,rl group.. \\;1" rl'''J'l0n .. ib k- lor 115 p ..: r ll'nl (1f til\' 1lL'1 inul·:l"I.' in Iltl' Tlll'il. ,rn IIKr,:;I",' F...·o n .. "' .... Rc" k w - " .. urlh Qu:orll'r 19'91 .I efl ).... ' 1<"(·I IP:lIIIIII.11 g1'(Jup:-- \\ , 1" rl: ..;ptm .. lll!L- fo r 1- - lX'I'l'l' n t of tire imTr.:a:--l: : ;1Il HlU..·;t ,,_· ;I('I'OSS ;tgl' gn 'lll'~ \\ .1" lv,p<m .. ihk- 1,)1' X K IX 'l'n'n l: and an ilu r\';I'\.' ,1('1'0 .... Illdu:--lril::-- \\ .1" fl ' ''lx lIl .. ih k' for di" Ix,.... ion o f \\,'gl '" ,Kro:-.... ),:t:nlk'r group :-- n:dlleed Ihl' 1lL'1 1\l( 1'l';l'-l' ttl 11ll' T bl,.1 h\ I ~ H pl'rc<-'nl .\ .... IUl\\ n in 111L' th ird l"olurlln' of'l':lhl\-.. :l nd 5. 11.lg<-' in('ljU,l ll lr d,:dinnl q uilt: .. Iurply frolll II)S- 10 Ii)SI) The ('o n n 'r),:clll't:' in \\ :Igl '." ;I('H'......1;lll'" l h :1 1 occu rred f f( )1Il 1<)1{- 10 JI)I->'<) :\C('()unll'd for o n l~ I (l pt.· ..l t:llI o f Iht: (k'din ..: in Il:u io ll:rl \\,:Il-tl: inl.'qu;t l ity dUl'ltll-t I llat p.,:n o d ThL' 1110 :--1 impo lI;l nt f;I\'!Of :l fft:'l'ling I Ill: d~d inl' in \\ :tgl' In ...qu,l lity \\,1." , f conli nui ng dcdi nl' in t ill;' 1l1,lk'-fcm,d c \\";t.l:e g ap 1\ Ikd ll1l: in l it..: r..:l:l1in : w;lg\'" (II Ill:I1l:lgl,.-b l . prol-..: ..... i(ln:r l . Ic d tnictl , <l nd ' ';lk.. \\'Ol'ktTS rl'dllc<-'d lilt: OCl't lp:rli()n:tI w ;lgc ," prl':ld . 1\ hieh :11.. \ I Ic(1 1( 1 k'''~ ()\'cl: 111 1\ age inl'qu ;l li l ~ Til \,' d i .. pl:r.. io n " I' \\ agcs :tnn..... l:d llCallo n:lf !-(roup:-- wa s l':-....l·r111;tll y ul1cll;l11gl'd Irolll I ')X7 10 J I)S() Tfle docol>l/JOS'lJOnS ptOH(J(J-:J 111 Tables I an-:J 3 SJmPly plr:NrOO clUes as 10 wfJJcn ('C.(}I'I(lIThC and demogtapll.c /ac/(XS may n.we Dlilyed ;J10/6lnrlSJfJ9 II-age rnequahly The deCom,:)O$llO'lIQflO£eS""P"'lanr .nle fllC rIOfIS bct~n (dC' lors sucn as 0CCtI031f(}fl and edvcallOf' and 'I IS /TlUS '!ilpptOO<..J'e 10 lora/the percent e~fJliJ<ned by a 8 factors " Table 3 Decomposition of Theil Measure of National Wage Inequality By Demographic and Industry Demand Factors' Education 1978--87 1982-a7 1987-39 .0481 02<0 -.0256 .0131 .0058 -.0001 - .00018 - .00156 - .00049 .01331 ,00732 .00035 0349 ,0203 C Due to change In wel(lht given \0 each group -. 00016 -. 00022 .00009 0 , Due to change In group Inequality measures ,03510 .02051 - 02543 1978--87 1982-87 1987-89 0481 0260 -. 0256 Change In Inequality 01 wages across age groups (Equals A ~ B) ,00.)4 .0023 -. 0023 A Due \0 change m weight given \0 each group - .00325 -. 00085 - .00070 B. Due to change in wages ,00663 .00313 -, 00 156 .0447 .0238 -, 0234 - .00056 -. 00021 .00075 04525 .02397 - 02413 Change tf1 Theil (Equals 1 + 2) Change in inequality of wages across educational groups (Equals A + B) A Due to change in weight given to each group B. Due to change In wages 2. Change In inequality within groups (Equals C + D) -.0255 Age Change In Theil (Equals 1 + 2) 2. Change in inequality within age groups (Equals C .. 0) C. Due to change m weight given to each group o Due to change m group mequality measures , For <Seton-hOM ot demogr.;tphoc and IfIduslry groups. see Ihe Appeod,x Dala are lor year·round, lul~me W{)rI<~ SOURCE OF PRIMARY DATA, U 5 Depanmeru 01 CommerCE!. Bureau Of the Census. Current Populatoon Survey. March tapes (Contmued on lhe rlfI.K1 page) Table J-Continued Decomposition of Theil Measure of National Wage Inequality By Demographic and Industry Demand Factors Gender Change in Theil (Equals 1 + 2) I, Change in inequality of wages across gender (Equals A + B) 1978-87 1982-87 1987-a9 .0481 .0260 -.0256 -.0077 - .0049 -0040 A. Due to change in weight given 10 each group 00147 00040 -_00000 B. Due 10 change in wages - .00921 - .00528 - .00460 .0558 .0309 -.0210 -.00211 -.00064 -.00001 .05792 .03156 -.02102 1978-87 1982-87 1987--89 .0481 0260 -.0256 Change In inequahty 01 wages across occupations (Equals A • B) .0078 0046 -.0033 A. Due to change in weight given to each group - .00147 -.00033 ,00014 B. Due to change in wages .00926 .00428 -.00340 .0403 .0214 -.0224 C. Due to change In weight given to each group .00864 .00144 .00154 D Due to change in group mequality measures 03164 .02000 -.02393 2. Change in inequality w thin gender groups (Equals C + OJ C, Due to change in weight given to each group o Due to change in group Inequality measures Occupat ion Change 10 Theil (Equals 1 + 2) 2 Change in Inequality Within occupations (Equals C + OJ (Continued on the next page) r: .....nnmic Rt" ';t'W- fou rth Q U""t' r I'.N! " Table 3-Continued Decomposition of Theil Measure of National Wage Inequality By Demographic and Industry Demand Factors Industry 1978-87 1982~7 1987~9 Change In Theil (Equals 1 + 2) .0481 .0""' -.0256 1. Change in Inequality of wages across industries (Equals A + B) .0020 .0012 -.0022 .00064 .00029 .00022 .00136 .00089 -.00242 .0461 0249 -.0234 C. Due to change In weight given to each group .00539 .00188 .00124 D. Due to change In group Inequality measures .04071 .02298 -.02469 A. Due to change in weight given to each group B. Due to change in wages 2. Change in inequality within industry groups (Equals C l' 0) Summary a nd implicatio ns Thro u~ hOlI[ much o f lh ...' 19HOs. w:lge inequality ill(:rt:!:l.....:d i ll 11K' Un ited St:llt'S Prn-ioll:-' n :.'iL'arch h:l:-. fOlmd tllal ;1 ri:-.l..' i n 1.::lrnings h y ...·(itIGllion:ll it.:\'d ;Ind iIKrl':I"",'" wago.: dispersion anoss occupations \\,,,,'re lI11pOnant factOI':' in til(' 1'i...C in \\'ag ...~ inl'qu:di' y Ik.....·ardter:-.. I1O\\'e\cr. 11;1\'l' no1t:d 111:11 much of t ill' riM: i n \\,:I~e inequ:l lily \\';I ,~ left uno:pbinl,:d hr the dL'lIIographic :Uld indu!'>try f:K'tor:-. tll:1I tl1l'}1 L'x:lI11 ined In thi,.. ,.. tudy. I Itavt: fll rtlll: rni till' :111:IIY,"I .. hy l'xamining tht: i ntpact on waRt: inl'qll:llil)' of:l di,'t:rgt:no: in rq.:ion al wa~t:~ that ()('l'um,:d d urinj.! the 19HO:-. I find that rq~lo n:d .. hc)(:k.... !'> uc h a.s tin: n.:n:!'> .. ions i n tilL' oil and farlll hd t ... pushcd wagc!'> IO\\'l:'r in Iht;.·M:' 1'll:low-:I\·\:r:lgo.:-\\,:Ij.!1.' :Irca!'>. and increased dcfcn ..<.· .. po.: ndin).: ptl .. hl:'d lip \\ agOO'!'> in alx)\ c-;t"t:r:I~l'- \\ :I).:l· ,lfl":I!'> \,\'h l k' ill(.'''l' rl.:"gional .. h(X'k:-; innea ..ed o\'l.'r;l1l \\ a).:<.· int:qu:dity. I find l 11al fCRion:l l \\age d i\'erj.!...·!ll·...• acnlll l1l<.·d fo r o nly 2 I pcrn'nt to :; perl'cnt of tiK' n .....· in n:lliollal W:I).:l· inc qua lity Ot hl;'1' f:I(:l or... "lKh ;1:-. inneasl;'d \\'a).:c dl"rx:I':'IOn ann..!'> I.'dul·alional ;Ind OCCUp:ltiona I group~ and:1 rl.'dul'lion i n llll' ma1c-fcmalt: \\a).:1.' g;lp. phl}l'd 1:l f gl'f ro1c!'> Altl l()llgll ren:nt "Illdil.''' .. uggL'... 1 , 11:11 regional \\:Iges arL' likely 10 return IOl·O!1\l'fg...·IK"L·. re:-'lI!ts of thb swdr ,"\lgge.'~1 Ihal Ih ..., l'Ol1\o.:rgl.'lKe \\ ill li l\cl y h;I\·1.' only:1 :-111:111 iI11P:KI on 11;llion:11 wage in ...·qu:dily Tllt: flJlufl' dir..."ction of \\ :Igl' inl.'qu:I l ily i .. lI!K 1c:I!'. A )(I"owin),: demand I'm high-sk i lkd \\ork<.'r!'> (Without ;l klfgl' i ncrca!'>l.' in till' ,..tlppl y of "ki lll,d worker!'» would PUI upw:ml pre..... urt: on \\,:I).:l' lIleqll;ll iIY. whik a ('Olllinu<''l1 dedinl:' in the III:II<.'-fcmale wa!-,!t: gap \\ould pUi downward Pfl''':-ll rl' o n w:lgt;.· inl'qll:llity To !-,!L'I :1 ho.,:II<.·r ide::! of Ihe flltllr<." di ....:.'l't ion of n:llion:11 \\ ag<.· inc::qllalit)'. f1ll1 11 ...·r \\orl.: need.. III ho.: don\.' 10 fu ll) llmk"r:.tand till' faclor.. :llTecti nR \\ .1)..:...· 1Il"''lIU:lliIY FC<k r.1i RCSC ...·l· 6:l nk of Dalla:> Using the Population-Weighted Theil Measure of Inequality If the wages of each person were multiplied by some positive constant, the degree of inequality would remain the same_ In this paper, I use a measure of inequality due to Theil (1967, 126-27)_ As shown in Horrigan (1991 ), this measure can be written simply as (1) T= Ips,li ps,), , ' \ IS, where ps, is the population share of the ith person , iSI is the income share of the i th person , and P is the number of people in the sample. Thus, the greater is the difference between individual 's income share and their population share , the greater is the Theil statistic. As brought out by Haslag, Russell, and Slottje (1989), no single inequality measure has been proved superior to the others . As described in Shorrocks (1980), however, the population -weighted Theil is one of a limited number of additively decomposable inequality measures that satisfy three important properties : A transfer of earnings from a richer to a poorer individual reduces the value of the measure ; Shorrocks finds that among the limited number of additively decomposable inequality measures that fit these criteria, the population-weighted Theil is the most satisfactory because its decomposition is unambiguous. In calculating the Theil with CPS data, several problems must be addressed . The first is that each observation is not weighted equally. I address this problem by adjusting the Theil measure to account for unequal weights. As shown in Horrigan (1991) , the adjusted Theil is calculated as " I~ln ( Z, ) (2) T = ln,u N where " ' Z, I~ (3) ,u = ' N , and (4) If there are r groups with n individuals and each group has an identical distribution , then aggregating the groups would result in an inequality measure that would be equal to the measure for each of the individual groups; and F~"nomk K.,.,icw _ I'oun h Quancr 1 9~~ ~ is equal to the CPS sampling weight for individual i, and Z, is the wage and salary earnings of individual i. (Continued on the next page) ., Using the Population-Weighted Theil Measure of Inequality-Continued The second problem in using the CPS data is that wages that exceed a set level are not given in the sample . Although the truncated part is usually only between 0.5 percent and 1.5 percent of the sample, it nonetheless can represent an important downward bias in the Theil. To account for these high-wage earners , a Pareto distribution was estimated by state, excluding the lowest-paid 60 percent of the work force. I An average top-coded wage for each state was calculated by using the estimated distribution of the top-coded individuals , and each top-coded person was then assigned the average wage. Once a wage was estimated for each of the top·coded individuals and a nalional Theil statistic was calculated using the C PS weights, then the Theil was decomposed first into the contribution of wage differences between subgroups : (5) 1 N Jl G 8 =- "L,N log - , g g J1. g where Nand jJ are defined in equations 3 and 4. The contribution due to inequality within each subgroup 9 '" 1,... , G is then defined as (6) where T is the Theil for each group. The aggregate Theil measure is then equal to (7) Pasl research on 11>8 d.su.b\Jtlons 01 wages and .rocome has Iouod that lhe P8fe1O d.su,b\Jbon Ms _~ al ll>8 ~ppef 81ld 01 the OtslflbutlOl'l For .~ample. see Sll19 h and Maddala (1976) and Crarner (1971) Appendix Group Definitions 1. Age A. 16-19 B. 20-29 C. 30-39 D. 4Q-64 E. 65+ 2. Education, A. 0-8 B. 9-11 C. 12 D. 13-15 E. 16 F. 17+ Years Completed Elementary School or Less Some High School High School Graduate Some College College Graduate Post·Coliege Graduate 3. Gender A. Female B. Male 4. Industry A. Agriculture, Forestry, and Fisheries B. Mining C. Construction D. Durable Goods Manufacturing E. Nondurable Goods Manufacturing F. Transportation , Communication, and Public Utilities I~ no mic Hl""icw - f .. unh Qlla n \' r 19')2 G. Trade H. Finance, Insurance, and Real Estate I. Business and Repair Services J. Personal Services K. Entertainment and Recreational Services l. Professional Services M. Public Administration 5. Occupation A. Executive, Administrative, and Managerial 8 . Professional C. Technicians and Related Support O. Sales Occupations E. Administrative Support F. Private Household G. Protective Service H. Farming, Forestry, and Fishing I. Precision Production, Craft , and Repair J . Machine Operators, Assemblers, and Inspectors K. Transportation and Material Moving L. Handlers, Equipment Cleaners, Helpers, and Laborers References Blank, Rebecca (1985), ~Disaggregat i ng the Effects of Economic Growth on the Distribution of Income ,~ Inst itute of Research on Poverty Discussion Paper no. 780-85 (Princeton, N.J. : Department of Economics and Woodrow Wilson School o f Public and Intem:ltional Affairs, Princeton University, September). Bluestone, Barry, and Bennett Harrison (988), "The Growth o f Low.W'lge Employment: 1 963-S6,~ America/l ECOllomic AssocialiOIl Papers and Proceedings 78 (M.IY): 124- 28. Bound, John, and George j ohnson ( 1992), "Cha nges in the Structure o f Wages in the 1980's: An Evaluatio n of A1temative Explanations," Amen'can Heol/omic Review 82 (June): 371- 92. Brown, S.P.A., and John K. Hill (1988), MLower Oil Prices and State Emp loyment,~ Contempormy Policy Issues 7 (July): 60-68. Haslag, joseph H. , William It Russell, and Daniel Slottie ( 1989), Macroeconomic Activity and Income Inequality in the United States, vol. 55 in Contemporary Studies in Economic and Financial AnalysiS (Greenwich, Conn.: JAI Press, Inc.). Horrigan, Michael (991), "The Effect of Truncation Bias on Measures of Inequality~ (Paper presented at the Western Economic Association mt.-ctings, Seattle, Washington, june 29- July 3). Karoly, Lynn A. (992), "Cha nges in the Distribution o f Individual Earnings in the United States: 1967- 1986," Review of EcOllOmics and Stalistics74 (Fcbruary): 107- 15. Loveman , Gary W., and Chris Tilly (988), "Good Jobs or Bad j obs: What Docs the Evidence Say?" Federal Reserve Bank of Boston New Elig/alld Economic Review, January/February, 4&-Q5 Browne, Lynn E. (989), '"Shifting Regional Fortunes: The Wheel Turns,· Federal Reserye Bank of Boston New England Economic Review, May/June, 27-40. Shorrocks, A.F. ( 1980). "'The Class of Add itively Decomposable Inequality Measures,~ Econometrica 48 (April); 6 13-25. Carl ino, Ger-.dd A. (1992), "Arc Regional Per Capita Earnings Diverging?" Feder-d.1 Reserve Bank of Philadelphia Business Review, March/April , 3- 12. Singh, S. K., and G. S. Maddala (976), ~A Function for Size Distribution of Incomes," Econometrica 44 (September): 963-70. Cramer, J. S. (1971), Empirical Econometrics (Amsterdam: North- Holla nd Publishing Company). Smith , Lee ( 992), MAre You Heuer Off?" Fortune, February 24, 39- 46. Drdocnstou, Ma rk, and Abn D. Barkema (990), ·· U.S. Agriculture Charts a New Course for the 1990s," Fedcml Reserve Bank of Kansas City r.collom ic Review, JanuarylFehnlary, 32- 49. Stei n, Herbert ( 1992), "Refl ect ions on the Top 1 Percent ," If/all Srreetjoumal, May 16. A14 Eberts, Randall W. (1989), M i\ccounting for the Recent Divergence in Regional Wage Differe nli a ls, ~ Fedcml Reserve Bank of Cleveland Economic Review, Third Quarter, 11 - 26. Grubb, W. Norton, and Robert H. Wilson (992), "'111e Effects of l)emogrdphic and Labor Market Trends on Wage and Salary Inequality, 19671988, ~ MOllthly Laoor Review, forthcoming. Theil, Henri (967), £COllomics and In/onnalion Theory, vol. 7 in Studies i ll Mathematical and JHanagerial Economics (Amsterdam : NonhHolland Publishing Compa ny). U.S. Congress, Office of Technology Assessment (1992), A/ler the Cold War.: Living If/itb Lourer De/enS(! Spendillg, OTA-ITE-524 (Washington , D.C. : U.S. Government Printing Office), f ebruary. Federal Reserve Bank of Dallas 1992 Index Economic Review First Quarter n\~ C:lM' of the Missmg M2 job" V. Dum Monetary Policy in ;t Small Opcn Economy: The Case of Singapore 101m I I Wood RegionJI Effc(:1s of LihcrJlizcd Agricul tural Tr.lde Fiolltl D. Siga l/a Second Quarter The Compar,l(ivc,' Growth Performance of the U.S. Economy in Iht! PoStwar Period Mark A, Wymle Free Tr.HJc Ag~mcnt:. and the Credibility of Trade RefomlS IJtlt'id M . Gould Quantifying Management's Role m tiank Survival "/00"'0$ F. Siems Third Quarter Money and Output: Corre];ulon or Causality? Scott Frr!f!moll 1.0::10 Grow1h and Loon Quality: Some Preliminary Evidence from Texas Banks Raben T. Clair When WiJllhe United States Grow Out of tL'i Foreign Debt.? jolm K. Hili l;ourth Quarter Do Interest Rates Help Predict infl:J.t ion? KelllJet/J AI Emely lInti E/Xm F. Kocnig Tmde Policy and [nldlectu:l] Prvpcny PfO{cnion: TIle W'iII/am C Gnlwn ~orth-Soulh Dispute Regional W:Igt.! Divergence and "'a!ional Wage Inequ:lltly Keirh R Phillips I'as l IS-,UC" uf Ihe £Collomic Reule/w are available free of chlrge hy wrillng to the Puhlic Affairs Depart- ment. Fc.."dcr.ll Reserve &mk of Dallas. SbliOn K.. Dalla.<:. TX 75222. o r by calling (214) 922-5257.