Full text of Review (Federal Reserve Bank of Dallas) : May 1990
The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
BANK FEDERAL RESERVE BAN K OF DALLAS May 1990 • • conOIillC eVlew CVICW conOIDlC ANew Monetary Aggregate F. Koenig Koenig and Evan F. Thomas B. 8. Fomby Reducing u.s. Oil-Import Dependence: A Tarift Tariff, Subsidy} Subsidy, or Gasoline Tax? Mine K. K. VOcel Vucel and Carol Dahl 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 Oallas Rob,rt H. Bopin "'_n OwII_... t:JI/G William M. W.II. e. a.t._.,( _.- f ........" - - " ' " co< H.rv.y Rollnblum S-~a~'" G.r.ld P. O'DriICOII, Jr. vui'tft_"'" .w......DwwItII(Jt~~ -- W. Michael COlt Yu","_"'" Stephen P. A. Brown "'""11/11 \flu "'_I WId SoI>oar!tIlIl(>7!jJl Economi$U Nalronal and InteffiaflOOal John I{ HIli Evan F Koerul! Robert T Clall Cala S lOWl'l Kennelh M EmetV Josect1 H Haslall ltnda C Hunt81 MartAWyme RegIOfliJI and fnergy W,'I.am C Gruben Robert W GII""l!I' MII'Ie K Yucel Keith R PhillIps La... L laylor F,ooa 0 Stga11a Editon Rhonda Hams Diana W Palmer Virginia M Rogers The: fCt:lflOffllC RlMew i.! ~Is/'I«I by the FUr.1 f\eseNe BatIk 01 0.1,. The..eM!< presstd are thaW of !he lIIJ!hors ard do not rrfIect!hl! pcJSIhons 01 !/'Ie ftdelal I\e$er\oe IYrlk of DalIIS or !he ftderJl ReseM _,ty System SublCnpllOOS Ife Md~ tr" 01 cIwve PIaa sen:I "",,*u fO' .. ng~ and ..vbpie<OPl' ~O(IIS. bid oSSUeI. and Id· 6"e$S to!he!'\bloc; Af!,," (IepartmIrrt ~11IeseM! 8ri. of o.IIM. S~10011 k, 0.1. ''''''981 IU. T_75222, (21'/651-6289 Andes may blllflfl/1l*1 CII1 the conotl(lll tNt 1M R;Ua! II CtedlIId .tId !he ReseIItII 0.~ .. prO'llded ..,,,11\, QlP\'ot !he pdlialIB'! COIIlH1lng the !!plnted II\f!efIaI Contents Page 1 ANew Monetary Aggregate Evan F. Koenig and Thomas 8. Fomby During the past two dccldes, financial innovations have proceeded at a r.tpid pace. These innovations h;wc altered the liquidity of some assets relative to th::11 of others. As a resull , traditional measures of the money supply may have become less reliable as measures of household [iqll idity. Even sophisticated measures of the money supply, such as the D ivisia moneta!), aggregates, do not adeq uately adjust for the effects o f changes in the payments technology. Koenig and romby propose a measure of the money supply that avoids some of the sho rtcomings of existing monetary aggregates The behavior of the new measure suggests that monetary policy during the late 1970s and early 1980s was s uhstantially less expansionary than (he corresponding traditional and Divisia aggregates might lead o ne (0 helieve. Page 17 Reducing u.s. Oil-Import Dependence: A Tariff, Subsidy, or Gasoline Tax? Mine K. VOcel and Carol Dahl l ow oil prices and rising oil imports have caused growing concern about U.S. vulnerability to Oil-supply shocks. Mine K. Yucel and Carol Dahl devise a measure of vulnerability and use it to compare three poliCies that have been proposed to reduce U.S. vulncr;:lbililY 10 o il-supply disruptions: a 25-percc nt oi l-import ta riff, a S5-per-barrel suhsidy to domestic oil producers, and an increase in the gasoline !;IX from 9 cents 10 25 cents per 8:1I10n. Yucel and Dahl fi nd that the tariff would make the United States less vulnerable to disruptions. By increasing both consumer and producer prices, the tariff lowers consumpLio n while encouraging domesLic production. TIle increased gasoline tax could either lower o r raise vulnerability. If domestic supply is not very responsive to price c hanges, the gasoline tax inc reases vulne rability. If domestic supply is responsive to price c ha nges, the gasoline tax reduces vulnerability. The subsidy encoul"J.ges increased consumption and production, leadi ng to a faster depletion of the resource base. Hence, the subsidy would make thc Unitcd States more vulnel"J.bJe to o il-supply shocks. EVan F. Koenig Thomas 8 . Fomby Seniof Economist and PolICY Advisor Federal Reserve Bank of Dallas Associate ProfeSSOf 01 Economics Southern MethOdist University Consultant Federal Reserve Bank 01 Dallas ANew Monetary Aggregate n 1982 the F<..'{\erdllk.~t!I"\·t! Hoard. cil in).,( Ihe "actll:1I :lnd porcntial dTt:cts o n M I of ongoing l'h:ll1ge.~ in fin:tndal techno logy and the Wl':lter :l\'ai lahili tY of a wide \'arit:ry of Illolll'y-likl' imtrument .. :lIld near-monies: dt.:·emph:I... i;o:t,.'{\ the J\H mont'l;!ry aAAfcg:nc as a guide to polKY' SIIll!1:lrly. in empirical fcse;trch on the dema nd fOf money and on the rel:ll io nship lX'twl-'Cn growth in the money :-.upply and growth in o utput or prin.'s. m:ln}' t.:conomht:-. Il:Ive fou nd it helpfu l to introdUeL' a shift into tht.: p ubli.<;hed 1\ [ I :-.erit.: .... This shifl is typically placed at or ne:lr l he 1X').(inning or 19tH. coincident with the n:l\ ionw ide Introduct ion of nCAOIi:lble order of withdr.lwal (NOW ) I aC('l)tlnT:-.,~ lInle",.. ~O\X' :.!ccounl balance., are pt.:rftXI .... uh... tiuues for (".Ish and dl'mand dt..'posil<;. sim p ly adding the three as.sel.. To}(t:lher to obl:tin :1 mea. . urc of lr.msactio n balanct.:s (whidl is <.'on\'cntion:11 pracTice) is no mort.· appropri:lll' th:1I1 adding Ihe nllmber of automobile.... to the nlllnbn of hk'yde.~ . trains. bo;1\s , blISL'S, and ai rpbnc s 10 oblain a measure of The siock of transportaTion cCjllipment. A beller approat:h i.~ to Weight thc d ifferent assels by the tf<msaclion sen'ice . . Ihey yield-jllst as in me<Uiuring the.: .,tock of Ir::tn.<;p<lfI:llion equipment . one would want to wcight different types of equipment d iffere ntl ~' . depending o n thdr carrying c..lpacil Y and .'> IX'(."", [n the constn1Ction of:1 moneta ry aggreg:ut:. a \'arict)' of asset-'weighling ....chcmes haw l)cen proposed. J\ lost have at least som e int uili \'C apJX·a!. The Didsia :lpproach. however . ...Iands 0111 both for the inTellectu:11 rigor w il h which it has bel'n dcn:lopt.-d and for the.: :lItcntion it has fL' l'ci\·t.'{l.' The Divisi:l approach l-onCenfr,ltcs on Economic R~itw ~May 1990 measuring tllc rute ojgrowlb of the 1Il0ncy supply r.llhe r lh:m its level. This aggreg~lI e gro wth r.lle i..., t:lken 10 he :111 avcr.1ge of rht! growlh r.1\("S of individu:.t l as:-.cls. Tbe Imvt!r the r:lIe of ime.:rcst p:lid o n an :t.... seL Ihe gfe~lI c r is tht;' weight Ihat m()\'el1lcnts in the ;L-;.set rccclvc in the Divi:-.ia meaMlre of the r.lle of change of the mo ney supply. The idt!:t i ... Ih:lI if people remain willing 10 hold :In a.... ~ ( oncring ;\ low pecunia!)' r.lIe o f return. they must considcr the :Issel t!spt::ci:llly lIseful a.... a me:.!ns of payment. Accordingl y. ch:mge... i n ho lding... of assets bearing :1 low rail' of return p l<lY a more important role in delenn ining movemenb in aggregate IrJ.nsaction balances than do cha nges in ho ldings of high-yielding a"''>I;:IS The authors 'Msh /0 /flank Ed,m A Adams for superb research aSS'5Umce and Cara 5 Lown. John K ~#iI. and Kenne/h M Emery for helpful Commenls Mone/al}' PolICy Repor/ submil/ed /0 CO<lgress on Febru· ary 10. 1982. as publ,shed In /fIBFederai Reserve au letm. MarCh 1982 The F8d6raJ Rcserve abandoned Mllargel· ngen/uelyin 1987 (Heller 1988) See 8CIX A for definlllOflS 01 M / and rela/ed mone\aty iJ!XIfCQ8/es SH. for example, Darby. Mascaro. and Marlow (1989). Rasche (1987J. 5ronfI and T/IorrnOfl (1'J87) and Board 01 Gover"",s (1982) Wilham Bame". WIth Oilier$. has au/herod a long serIes of papers on OWlsia rrl<)(Ielary aggrf1{Jalion Sec. lor examp/6. Barnell and Spondl (1'J82) and UameN. Offenbacher. and Spmdl (1984) Altema/lves 10 OoVlSla aggrer;a11Ofl are descflbed by Spmdt (1985) and Rotemberg (1989) The Components and Definitions of Money As defined by the Federal Reserve Board, M1-A was the sum of currency held by the nonbank public and demand deposits at commercial banks (other than those due to depository institutions, the U.S. government, and foreign banks and official institutions), less cash items in the process of collection and the Federal Reserve float. M1 is the sum of the items that were in M I -A plus travelers checks of nonbank issuers and other checkable deposits (OCOs). OCOs consist of negoliable order of withdrawal (NOW) and automatic transfer service (ATS ) accounts at depository institutions , credit union share draft accounts, and demand deposits at thrift institutions. Beginning in January 1983, OCDs also included Super NOW accounts. Super NOW accounts paid more generous interest to depositors than did regular NOW accounts but had more stringent minimum balance requirements. After March 1986, with theelimination of Interest rate ceilings on OCD accounts, the distinction between regular NOW accounts and Super NOW accounts disappeared . Thc Divi o;i;1 imkx o( growl h in 11K m o ney .~ uppl y h;I'" two highly :ttlr.tet t\C 1<":;ll u res. Fi rst. cons truction or the index d(le~ n( II rL·qllin.: de tailed kllo\\ k'dgL' o r thL: conl l'li>ution \';Irious as~eh make 10 t he hll~ in~ :I ncl o;e ll lll~ of .W)(xl~ and The definitions used in this article are somewhat different from those used by the Federal Reserve Board. Thus, M1 A here refers to the sum of currency held by the non bank , nonthrift public; travelers checks of nonbank issuers; and household demand deposits. Business and other nonhousehold demand deposits are excluded Irom the analysis, on the grounds that the motives for holding such deposits are substantially differ· ent from the motives for holding household deposits (Mahoney 1988). Further, only households have the option of opening NOW accounts. Similarly, for the period during which a distinction must be made , the OCD account variable here , N, is a weighted sum of Super NOW and non-Super NOW OCD balances. Non-Super NOW OCD balances are given full weight. while Super NOW balances are multiplied by the ratio of the opportunity cost of holding Super NOW balances to the opportunity cost of holding non-Super NOW balances. -Household M1 balances~ are the sum of MIA and N. .~L·f\ in:s: t hel't.: ;II'L' nl 1 fu nel io na1 f()"lll~ tl) ~pL'dfy . no parailletcr.... (0 L:.~lilllah.: .1 S...:cond . the D ivi~ ia imkx i ~ :U..'t'uralL·: for :ln }' gi\'<.:.'n pay tl1C1llS t\'Chnolo)-.:r. thc irl(kx':-o l1l()n.' lll ...:nt~ (ully rdkct ; ll1 d l' h an~L:'" in Ir;l n ~aL'l i(Jf) '-L·1'\·it·L'::;. ' T ilL' acn ll:l('~ 01 thl' Oh l:-o m inde.x i .~ con tinupon t he a~' lIlllpIi O n lilat Ihl' pa} ments tec hnology. whatl'\'er ib rOnll. i ... uncha nging. The ~I..·nt • /I IS. howeWf. necessNY NI ooe /u'Iow.• Pl'oOt, whctl .asMts . . 10 t» IfIC4Id«J If! ttw ~Ie IJI)(J lhallhe ~t..w..tyllClmtJgel'''''''In'''COITp:lI'''''assets (so /hal a $Il71CIIlanwuS doubhng cI aN Ccmponenl assets resu'IS 1fI" douOltflrJ 01 /!Ie aggregate} I Only CO'lI'nllouS-t me [)'v'SUI itggrflfJ<ll<Of! .S fuRy accurare 2 .nae, OrvrsJ8 ()IOvtCICS it /fl"d-order lhe /fue aggregale (o.e...erl 1976) /n d,scrcle Mle 11>6 IIpptC»IITlaIO'lIO l (rO~ and 19l'l<1:-. hO\\\..·\ ·l'r ....;1\\ thL' 'pread of new typ....~ of 1r.llh.U.:liOll .ll count___ known col/ecthL"ly.h other l hed.... tbk· dep().~l'" (OCD... }-and of inno\;niol1' Cll kL' alltomatc:d Il'lk'r m:tchint:s. 1l10n\..·~ Illa rket 1ll1l1U:iI l und ~l('l·()llllh . .mel 1ll011t:y 111;l rkel depo'lt aU'ollllto;) dut bcilil .ttL·d th...: tr.msroo-r of \\l'ahh 11110 ~Hld (Jut o r tr.\fl~act i()n h:lianc...:s. !'t--de ntl Kt"SC o 't' Bank or Oallas making s uch balances more e:fficicnl. The evidem ch:mges in payments technology during the p'lst two decades suggest that altt..'rnati\'c~ to Divisia aggregation arc worth exploring. In this article, w<.' descrihc.: one such alternative approach-an approac h that, li ke the Divisia index. makes use of m<lrkct inle:r<.'sl r..Ltes to weight the contributions of individual assets bUI which, uHl ike the Divi.s i;1 inde:x. m;lkes at least SOI11<.' provision for !'>hifts in the payme nts tedmology." We find that the Divisia meilSUIl.:! of the mo ney supply. hecauS<.' its tonsl mctioll fails to m .. ke: proper allO\vilnce fo r Ih<.' inl'fe:.sing aV;lil.. bility of OCD accounts, significantly dist ort.~ the pattern of growth of trans:Ktion ha b rKes in the late 1970s :Ind early 1980s. NOI surprisingly, in order to relax along one dimension Ihe assumptions that unde:rlie the Divisi:. approach. we are forced 10 make more stringem as... urnptions in other areas. In pa rticula r. we find it necessary to assllme that Ihe p:lymenls lechnology ha!i a spt!cific functio nal form. We :oI lso find it nc,."cessary [ 0 estimate one of the paranU;'h::rs of this technology econometric.llly [n the fo llowing scttlon. we review the r.Jtion:lle for Divisia aggregatio n. After noting the limitations of the Divi!'>ia approach. we propose an ;Ilte rnative aggregation proced ure. Our procedufC yields a mo netary aggre:gate: thm . dming the late 1970s :md early 1980s, is much less st'n!'>ilive to Illov<.'mems in OCD habnccs than is either the Simple-s um or the Divisia me:lsure o f household ~ Il hal:l nces. Results arc rohust with respect to scveral possible cha nges in our assumptions. The concluding section of the .. rticle discusses the need for fun her research. Divisia aggregation R\!callthat the r..Lle of change: o f th\! Divisia index of Ihe money supply is a weighted avemge of the rates of change of individual as.'>t!lS, with greatest weight placed on the changes in as..,els offering the [owe!'>t rates of return Suppose. for cXllmple, that P.. represents the interest:1 ho usehold loses hy holding a dollar as cash or ir1 a demand deposit account for a unit tim\! period instead of invc:..'sting the dollar in a hond or other high-yielding security (BccauS<.' cash .. nd demand deposits pay no interest. P.. is ...imply equal [0 the Ero nomic Revlew _ MilY 1990 interest nltc on honds) Sim ilarly. suppose Ihat Pn represents the interest lost when a household ke:ep... a dollllr in an OCD account instead of in bonds. Then, if 0 denotes the Divisia measure of household M I halances and b it.<; ra te of change, where M IA is the sum of C'.tsh and ho usehold demand deposit'>: N denotes OCD account balances; and P.I' defint.."XI as p,,(M l AID) + P.(NlD), is a meaSure o f the average cost of holding Divisia money Equation I can he rearnmged to obtai n where and According to equation 2, the percentage rate of growth of the Divisia aggregate is a weighted avemge of the percentage niles of growth of its individual component assets. The weights, s.. and s., arc the shares of total "expenditu res" o n transaction services thai are devoted 10 each as.~et. Only readily ohscrvable price and quantity info rmation is requi red to calculate lJlD by using eq uation 2.It can be shown that , as long as the payments lechnology is li nea rl y homogcnL"Ous (so that ;1 doubling of all assets doubles the flow of transaction services) and is fi xed. movements in VI D will exactly match those in transaction services. (Sec Box B) , The new aggregate makes s/lowarlce lot shihs If) technology rnallend 10 favot ontI form 01 transacl1(lll bal8nce over Snolhcr. bul II does nol make lJ//oWttfIce lot changes If) IOClvlology lila! have a vrnform mf/venr;e Qr'IlIle effi<:oency oIalilfansaC!1()(I balanc:es , Dala on pril;e and quantify ,rs ,~aila/)/e only ar dl$(;rele lIme Ifllervals ConseqU8t'lIIy. If1lXflCl~. , dlSt:relfi1.11tT1e apptox:mal1(lll10equation 21S vsed 10 r;onslT\.'C/ rile OrVlSIS 'rlfJregste , BoxB The Accuracy of the Divlsls Aggregate A payments technology with the property that a doubling of all assets results in a doubling of financial services is said to be ~homogeneous of degree one ,M or "linearly homogeneous. Suppose that transaction services , M, can bewrittenasa linearly homogeneous function, tP(M1A,N ), of cash , demand deposit, and OC D balances. Then the Divisia aggregate will accurately reflect movements in M. In particular, DID, calculated as in equation 2, will equal MIM al every point in time . R To see that this equality holds, note first that by direct differentiation, M1M = (M1A~.1 ~)M;A I M1A +(N¢"I ¢)N I N. where ¢. and ¢In are the partial derivatives of 4'( , ) with respect to M1 A and N. Because tP( , ) is linearly homogeneous, however, ¢(MI A,N ) equals ~.(M1 A . N ) M1A + ~"(M1A . N ) Ntor all M1 A and N. (This property of linearly homogeneous functions is called Euler's Law.) Hence. M1M = [M1A~. / (M1A¢, + N~")jM;A I M1A + [ N~" 1(M1 A¢, + N¢")jN1N. Ul."t::IU"..· (x:[) ;1(U)lIIlI" p :lr illll.'r"',,1 " h ill.' cl ..h :lI1d d",II1;lIld dl.'P0:-'II:. do nol ./! ... gn...·atcr than p~. :lnd Ihl.' Di,·bl:1 nll.·:t:.lIrl.' of';h .., pcrc..~nt- If M I deno/,.,s me sum of cash noosehold demand depas'l, IIndOCDtJ.1lances IhenMI/M1 ~ W.lMiAIM1A). w(NJNI whe<e "'. lSdCfNledasMIAI(M1A • N)and w, IS defmooas N'(MIA • N)Mdequals 1 - w, SlftlC/Calgebrtl esrabl,shes fhat s. IS grealet Ihan wo,f andOflly l' P, IS (JIBtller than R • Suppose, finally, that households allocate their money between M1 A and Nto minimize the cost, in forgone interest, of attaining a given level of transaction services, M. Then the marginal rate of substitution between M1 A and N will equal the ratio of the price of M1 A to the price of N: ' .'4',, :: P,,1P,,, It follows that Ml A9./(MlA4'. + N¢" l:: p. M1A 1(P. M1A + p"N) '" s. and , si milarl y, that N¢J" /(MlA¢/I + N¢J,,) '" p"N l(p/lM1A + p"N ) = sn. Therefore, M1M=5, (MiA I M1A)+5" (N 1N) =0 1D. as claimed. Suppose , now , that M equals 9 (mMl AnN) , where m and n are parameters that vary through time as financial innovations occur. Following the steps outlined above, one can readily show that MI M = sa(M1A I M1 A +m l m) +Sn(N I N+h f n) = 0 10 + 5,(tiI 1m) + 5" (1i! n). Equation 4 , in the text , is obtained in the special case in which m equals (1 - a) .. Jt ..... ·) and n equals a .7'(..... . 1. Thus, financial innovation is capable of destroying the accuracy of the Divisia aggregate. .l~'" ).!rl)"th r:n ..· of Ihe: Illon<;:y "uppl)' kqll:llion 2) pUl" I'clal l\'e!r ).!r"':ltcr \H.'lght on g1"l)wth in ebh .111d dcm.lI1d d,,·po:-.ib Ih:1I1 dOt;''' "lIupk-:-'UIll ;JR).!l"l·g.ltl(m. " Ike:lll"", OCD b:I!:l nll'" h.I\ .., h\.:~ n n~illg :11 0\ r;I~ll.·r Ix·r(,,·l1!agl:.' fate thoBl 1.";1,,11 ;111<1 dl.'lll,llld dcpo-..it lubm·..·" (Ch;u1 I ). lhl" dil'f... rencl' III rl'i;Jli\1:.' \\'l'lghh illl pl i..'s tha t th ... Illoncy :'lIpply gf(J\\th 1~1l"'.:l~ Clkubtt'd by t ilt: Did"l;! pn)l·l.· dllt'l.'. h;I' gt:ner.dly hevll below thl.' gr{)\\I il ralt.' of th ..· "i11l p lc -" LlIl1 IlW;I:-.llfl' of huu:.t;'hold ,\11 hal- Federal Rese n 'e Ibn k or Oa llas Chart 1 Chart 2 OGD Account Balances as a Share of Household M 1 Balances Alternative Measures of the Rate of Growth of Household Transaction Balances Perce nt Percenl change from 12 monlhs before 50 25 ... 20 " " " 30 5 20 o, ~----------~~----------- " 5 o ~~~~~-__~---1973 1975 1977 1979 1981 1983 19 85 1987 -101-______= 1973 1975 1977 "" _ _______________ _ 1979 198 1 1983 1985 1987 SOURCE Of PRIMARY OATA SOI.JRCE OF PRIMARY DATA Board 01 Governors, Fe<ieral Reserve System Board 01 Governors, Federal Reserve System :lOces. On the o th\:r hand. Ix:callsc the rapidly growing OeD h.lianl"c:-, arc gi\'cn at least sOl1le \\"l'ight in it:. const rUl"lion. Ihe growth mIt;' of Iht,: Divisia agg rC~: H e ha ... cxn:cdcd that of M II!. In pr:lclieC'. the Divisi:1 :Iggregalc has beh:lvc d mllch morc like hou:oci1old 1\11 th:1O like M IA. (Sec Chart 2..) An alternative aggregation procedure To implement Ollr alternat ive aggreg:l!ion procedur(', w e ne(:d to a....·;umL· Ihat the p" yments tcchno logy ha:. " sptX'ifi (' fu nctio n:ll form . We will assume. in particu lar. Ilwl cash, demand deposits. and OCD balarKL'S (;ombinc to yield tranS:K't ion sL·rviccs. M , ;!ccording 10 the fo rmula OCD halances. at the m;lr~in. relatIve to corresponding scr"icc.~ ~tickk-'{I hy cash and by demand dcpo.~it s.IC The p:lr;IIll\:lt.!r 0". o n thl! o lher hand, measures how nearly intl!rchangeablc the I\\'O types of :\S:OCI babnces arc. It is called Ihe -elasticity of suhstitution" IX:lwecn M lA and N. When 0" equals zero. there is no possihility of ..ubstitUling OCD habnres fo r (.':I~ h o r demand deposit babnces: Ihe assels must IX! us(.-d in fixed proportio ns and w ill mow together. in locksto..:p In contr;lst. MIA :lnd N become perfcct :>ubslitlltc:-. as 0" approaches infinity. and eq uatio n 3 rl!duces to M .. [(I - a ) MIA + aNJ Only when, simultaneously, cr is infinite and ex equals onl!- ha lf is simple-sum aggregation of OCD ba la nces with cash a nd dema nd de posit 0-1 M = [ (l - a )MIA(o I) " +aN() 0 " ] {o-'I . where a and 0" are nonneg:H i\'c parameters. with a less th:1O o r equal to I The functional form on the right-hand sidl! of equation 3 is quilt;' flt;'xihk and has been lIsed in :>l!wr.11 cmpirical sludit;'s of Ihe demand for mo ney." If a equals ze ro, OCD halances yield no tr.msaction services. a nd equ:llion 3 reduces to .11 - MIA The larger is a , lhe greater are the tr.ms;lction services yielded by f..conomic Rt:v ie ...· _ May 1990 • See. lor elfample Barnell (1980. (981) and Chetly (1969) or &:me fifllJnclal 1fIfI(lY811OnS. such as the tt'I/foducbOfl 01 IWlamaled 18I/er maclw1ocs. /'I1CIO'Ie)' market mutual /vndS. and money market depo$Jt 8CCOt.H'lfS. will/end 10 K>Crease the marglfllJl ptoductnnly 01 a I types 01 1fan5aC/1Of1 balances. IJ/'lj/ctmly The af}f}feglJ//OfI procodute we develop fIere ~ not capeOle 01 ClIDturlflg the effec/s of such dlnOViI· /.Q1S NCI/fIfIf ale /hey captured. however. by the OMsla and $¥TIpIe·wm ll{X}rcgates , balances ju~lifk--d. OnJr when a or a is equal to zero is it appropriate to ignore OCD halances and use M lA as a measuft: of {r.msaction services Divisia aggregation, in (on\r.I~l , is valid for ally a ;lnd a, provided they arc timc-inV;1riant. Unfortunately, the parJmeter a likely did not remain constant as OCD ;I(counts spread grddually across the n:llioo. After all, during most of the decade of the 1970s, thl: only way in which a household located in , S;ly, Texas could even use ;1n OCD ;lccmmt was through the mail. II The cost and inconvenicnce ;1.~s(X' iated with making OCO account dcposits and withdrawals must have dropped sha rply for such households in 1978when commercial banks were allowcd 10 offer automatic tr..msfcr service (ATS) accounts-and again in 1981 - whcn NOW and ATS accounts 1x.'(:<ltllC leg;llly av,libhlt: ,Il all Tcxas depository institutions. But lowt:r marginal dcposit and withdrawal costs arc equivalL"nt to incrcas<---d marginal transaction serviccs (Orr 1970; Milbourne 1986). So, a - the rdativc weight placed on OeD ba lances in the household liquidity function-very likely incre:lS~:.'(1. A timt:-varying a drives a wedgt: hetween the growth r..l1es of ,If and I): (4) f,flM=DI D+[a l (l-a)](a-,,,)u / u. where v is defin<..x1 ;IS a/( I - a), so tlmt ul v equals [1 / ( 1 - a)1it! a . (Scc Box B for a derivalion.) Thus, whether in(.'fcases in a raise o r lower the growth rate of M relativc to that o f the Divisia index depends both on whether a is greater than o r less than unity and o n whether the current value of a is greatcr than or less than s". If a is birge, so [hat Nand M JA are close substitutes, then :1 small ch,lIlge in the price of one of the :ISSCIS, relativc to the price of the other, ought to induce a large reallocation of tmnsaction balances away fro m the asset that has become rclatively more expensive. Conversely, if a is smati, then even a iargt: change in P,. relative to PIt ought to have very little impact on the allOC'dtion of trans..ction ha1:lnccs. Forma lly, household cost minimization implies (5) In(N I MIA)= aln( v)- aln (p,./ P.). Equation 5 <-,10 be differenced, then estimated econometrically, treating .dln(v) as a random disturbance with nonzero mean and a as an unknown parJmeh:r. In carrying out this estimation, we obWined a value of 0.2 141 for a. with standard error 0.1606. (Details are given in (he Appendix.) After (J is cstimated, equation 5 can be turned around and solved for U ::IS a function of the price ratio, p,/ p". and quantity ralio, NIM I A. Equivalently, onc ca r' express U as a function of NI MIA and the ratio of expenditure shares, s Is . " " (6) 1,--).- U= (sJ s.. )( N I M IA ) We have already seen that thl! quantity ratio, N/AI lA, was increasing during the tate 19705 and early 19805 and th,lI the estimated value of the daslicity of substitution, a, is less than unity. Chart 3 demonstrJles that sIt has been declining and, hence. that the mtjo SJ SII has been increasChart 3 Expenditures on Cash and Demand Deposit Balances as a Share of Total Expenditures on Household M 1 Balances Percent "'r -----. 95 90 85 80 75 70 " Before NI:Jvembef 1978. n/8fes/-beanngc/leckJtlg 8CCO!IflfS ..".." llvadablo only '" the New England starc-s Tho early fIIs/Of'y oIl'lrer6s /-otIarKIQ c/IecJOOg accounrs is d iSCUSsed 111 Gibson (1975J 6 .5~--------------------------'-1973 1975 1977 1979 198 1 1983 1985 1987 SOURCE OF PRI MARY DATA ' Board 01 Governors. Federal Reserve SY$rem Fedf: ral Reserve Ban k o f Oalla$ i n ~. Frolll 1.:<l U:lli(1I) 6. Ih...·n. il !"ollow~ Il w l V has n ... t.: ...s.~lril}" 1X'1.:11 lrl.: ndi n ~ upward I n olher \\oro .... Ihl.: m :l r).!in:11 lI~tn ~:K ti o n s...n ·K ...S on ...:It.:d hy OCO hal:i rK"'s ha\·...• 1X'1.:1l ri ... mg n.: l:l li\·...· to thoS<.: offt.:r...d hy C:1.~h and ( il.:l11and ( k- ro~it h:I!aI1(t:~ A,!.:ain. I hlS finding llIak ... ~ ....... n.~·: :1' oeD :K(;Ollnt.~ han' Ix:(;0111 '" :1\":li l:l hk at mor... and Illort..' h ;lIlks :Ind thn!"! inst itutions. hou",(;holds b:I\·... found il innl':.N ngl y nJnH'n IL'nl 10 lhl..' Oe D h:II:I11(;L'S 10 fi n:uK...• thL"i r pllf(: h a~ ...·.~ To j.(L't :I 1ll0 1"1..' p ......("Is ...• notio n of how v ha", Ix-L'n lx-h:I\'ing throllgh timl· . \\..... I I.~\.'(I equation 6 \() cakul:ttl' th ...· IXlll1 of v impliL'd hy l h...· t.:stim:lt ... d \·:llu ... of a Th ...· log:lrithm of thi.... path is plOl1l'd in Ch:lr! -I. Th ... j.!... n",·r:tI t,.... nckncy for V to innl·;ls...· through th ..., 1970s and It)H()s is apparl'nt l'articlIbrly rapid ).!1"owth in V o{.:cllrr... d in 1')76. when \0\,\1 : 1 "'T()lI n t.~ Ill·ClIll ...· :1\':libhk throll~houl N...·\\· En,L:l:md : in bll' 1t)7H. Whl..'ll NOW' :1(TOunt.... heCUll ... kg;! 1 in :\...·w York and ATS :lc('ounIS \\·...·r.... inl rodUH'd : :1I1d in l'ady 1t)~1. \\'h"" n '\OW ac("( )li nts Ix·clm.... a\·:II!all!l.: n:lt i(lIlwidl..' From t h...• dd"inition of v. a is (:q ual to v'( I + IJ) I-knn... equat ion 6 (':m Ix- us..."(i to clllumat ... a fnlm "'"(Illation 3. yiddinj.( a formula for .11 that d"'IX-Il(b only on a and r....adil y ohs"'I"cd p ric.... ~I nd qu:mtily dat:l : I\ .~ a ;lppro;Kh....s ar...· lx-rkcI infinity. so that OCD h aianc...s ~uhstitllle .. for Gtsh and demand de· po;...it balatK....s. cqu;IIio n 7 b""('omes !II - [p,, /I p" + " )1,1/ I I I + II' /(" + " )IN. It r ... Ill:l ins tn.lI.... th:1I .I /IA :lI1d ,\ Il~o\'e ~O,L:L·l'{l",·r. in lo(·kstep. w hen a eq uab z",·ro. so that. as lx-fori.:, it is :tppropri:ul..' to u .. . e .11 I tI ;1" :t n ll':I.~ urr.: of liquidity ....... n:ices in this C:lS\.·. \X/ e \\"ill ed l Ih ... 1ll0Ile[:IIY aggr... glll ... ohtain ...d from <.:qll:ltion 7 \vh <': 11 a is ., ct cqllal to its l:. . [i111aled valuL' Ih.... -CES aggregat .....· as eqll;ltiol1 7 i... d<.:rh'... d llnd ...,]" IIle assumplion that tht..'rc is a C()[lstanl l'bslidty () f suhstituti(lI1 hl.:lwl.:cn asS(..'ls. Ch:u1 'i-lik ... Chart 2- rlots :.tnnll:tl ratcs of ).t["()\\ III of thl" I)ivisi:l aggr...).tatt..' simpll..'-....ulll hous... hold 1\ 11 h:lI:IIK ...S. and !IT I A In addit io n . it plols till" grO\\"lh r,H ... of tIll" CES a,L:grl..'g:lt ...· eakuI:u",'d hy usi n).! tlli.: v;l llll;' of a ohtained from > ...",tim:lt ing "'qu:ll iotl '3, 0.214 1 In Chan -). til ... l'xt",'nt to which j.(I'( )wth in the CE..'i :1,L:grl..'gatc par-i llds th:1\ o f ,II lA during the bl ... 1970", and I.:arly 19HOs is slrikm,L:. In 198 1. for eX:Il11 p ll..'. th ... growth mtt.'s of til<' C I'"~'i aggn:gatt' Chart 5 Chan 4 Effects 01 Shifting Technology on the Marginal Productivity of N Relative to the Marginal Productivity of M1 A CES Agg regate and Other Measures of the Rate of Growth of Household Transaction Balances Percent change Irom 12 monlhs before o 5 " 25 ,. Al l 15 10 ,. - 25 30 '5 L,=--:=c-C~--:=C-~---C=--:~-C~-1973 1975 1977 1979 1981 1983 1985 198 7 SOU RCE Of PRIMARY OATA Boa rd 04 Governors. Feeler.1 ReS8fVe S~Slem Eco nomic Kc:vicw - !tf:1 )' 1990 5 o ~------------~f------------5 -10 1-_ _ _ _ _ _ _ _ _ _~--------1973 1975 1977 1979 198 1 1983 1985 1987 SOURCE OF PRIMARY DATA Boafd o f Goverr"lOls. Feeteral Reserve System. 7 and .11 IAlxllh rlllmm~L whil~ tho.-.c (If s impll. . ~\l1\1 MJ and Di \' i~i;t i\'Il M tr~~ Du ring 1983 a nd 19Wt. simi larly. when till' growlh r:.Hl'S of silllpkMIIIl {\11 and I)i vi~ia .\·11 hOlh In.:nd downward . gnl\\lil ralL'~ of the CES ;lggreg:ne ;mel .11 111 arc.: ~l:nl'rallr on tht· inc~t!>C E\'cn aftL'r 1 ~ 1. diffl:rI.: n CL'~ IX'I\\ L'Cn the CF.5 series :md Dh'isl;! 'iCril.:s rem;lin ( illant it'll i\'d~' . if n Ol alwars qualtt:ltiH: ly. .-.ignifk:'lnt. Chart 6 Sensitivity of the CES Aggregate to Alternative Values of the Elasticity of Substitution Percent change trom 12 months belore Sensitivity analysis In (·on.~tnt l·t in).t our propost·c\ new aggn.:~:He. WL' assum ...·d that Ihl.: elasticity of subst inn io n 111.:t\\'e~n oeD balances and c,ls h and d~tlland depo.,il h;iI,ll1n's is constant. I low realistic is Ihis ;t",slll11ption~ If the s ubstitutability between OCD habn(X'S :111d tile lllorc traditional 1r.1I1.'i:It;tion balanl'l~.~ has. indt:ed. changed. the shifl most likely o (X lIrr..."( 1 in 1981. when NOW accounts h(:came ;lvaibhlL'Il;ltionwide Au·ordingly. \\" rCL" ;(illlalL'd l'q\I;l\ion 5. a!1owin/>: I h~ pre- I98 1 :lI1d po~ t - l 9HO \·,lluL· ... of (J 10 differ No ... ignificam shift in the v:thlL' of a \~;tS detl..'cle(\.l! H ow important is it th:1I (J he precisely L~st i mah.:d~ To addre~., thi ... question. we llsl..'d t'quation 7 to clkul:l1 ~ the time p:llh of Ih" growth mte of thl' CL"i aAArL'gale for three values of the el:lslidl)' of :'\lh~l itution : (J - 0.0535 (one stambrd error '}('/ml" tltl..' val ue of (J obt:lincd from estimating equ,lIion 'i). (J - 0 214 1 (the ,'alue of (J ohtaim.:d fnml C'slima t in~ equalion 5), and (J o j7'i7 (one st:lIlc!ard error above the valu..: of (J o ht:linL'd frorn ..:~tim:tting cqtl:!tio n 5). As Chall 6 ma kes d ear. in any given oneye:II' period. thL' v;llw..: of acan indeed h:t\·c a substantial impact On the eSlimated growlh ralC of tll\: Illoney ~llrpl}'. In mid-19tH. for cxamplc. there is ;t d ilTerence of marl' than 6 percentage points he tween the annual growth r.Lte of the mOllq' ... upply Glkul:tted a:,sllming a equlIls 001}; and rill' corresponding growrh nHI..' calcll bt ...""{1 asslimin/>: (J equ:tls 03747. What i.-. most strik ing and also mOM reassuring. how~\'~r. is Ihal Ihe l/lla/iwlil'e lx:havior of Ihe CES aggregale is me d" ' ercncfl bf/rweenlhe pte . 196 1 9fld pasl 1980 el8s· be 0 (J237. WIth a stannard errOl 01 0 2635 11C'11l!S 0( su()snMIOf' WIIS eslmBled to • 1973 1975 1977 1979 1981 1983 1985 1987 SOURCE OF PRIMARY DATA: Soald of Governors. Federal R_rYe Syslem. \'ery mud] Ihe ~ame reg.'lrdlC's.~ of Ihe value of a . In tltl.' Divbia appr<xlCh 10 monda!)' aggreg:lIion , which :t:.... ume... that Ill(' pay men!.~ t(.·dmolORr i~ (,;onst:lIlt. C\'C!)' change in relalive as.-.ct ho ld ings nm ... t he au ributcd to a ch:tnge in rdative a ....'iC1 prices. Our estim:lle of equation S indiC".!tcs. hOw...·vI..T. that mo:.t of the \':Iria tion in Ihe nllio of OCD balance~ to cash :lnd dcmand dq>osit b"lan(·I..'." is unrdat ...'(i to relm ivc p rice Chllll,gc:S It i.~ pos.~ ih le. though. that relative price changes have not lx:en appropriately mcasurC'd In particular, it is I X)s~iblc Ih:1I measuring p" as the (\iffeft.:nCt.: hetWL'C1l Ihe r:Jt {~ of rcturn on bo nds and the ..IlC of rl'turn on O CD hal:11lces unc\l.!rMale." the avcr:I).!e co... t of holding OCD balances in rx:riods duri n~ which OCD act'oLlnts were not av ai l:tbl~ in all st:l1es. A mort.' re:llistic :tppro:tch might be to think of Pff as det'lining as it Ix'(:ame lega l for b:tnks in mOfL' and more states to offt:r OCD ,lCcounts 10 their c uSIOllier.. To de tennine how sensitive o ur aggregate is to IX) ....~ihle lIl i~mcas urc ment of rcl:tti\'e prices, we ddined a nC'w opportunity cost for OCD halances. p;'. eq ual 10 ;l weighted ;Ivcnlge of p" :tnd p,, ' The weighl :l\I;lchcd to P.. was set equal to lhc pro portion of the u .s population living in Slales wh..:re hank... cou ld kg:t lly offer the prinCipal type ft:dera l Reserve Ban k o r Dallas of OCD account-lhe NOW account. The weight attached to p.. was, s imilarly, set equal to the proponion of the U.S. population living in states where NOW accounts were illegal. Thus, falls as NOW accounts become available in additional states. A plot of the CES aggregate constructed by using was found to be: virtually indistinguishable from II plot of the CES aggregate :lS constructed origi na lly, usingp~. Apparently, then, the behavior of the Cl-::S aggregate is insensitive 10 plausihle changes in tht: measurement of relative asset pnces. p; p; Concluding remarks Ch:mges over lime in the r.llio o f the public's OCD babnces to its holdings of cash and demand deposits cannot be adequately explained by changcs in relative prices. This finding suggests that thcrc have been significant shifts in the technology by which households obtain transaction services from cash, demand deposits, and OCD balances. Simple-sum and Divisia monetary aggrq:ates, which do nO( make allowance for changes in the payments technology, may therefore give a misleading pict ure of the direction of monetary policy. ntis article has outlined a new aggregation procedure-a procedure providing specifically for financial innovations that may tend to fa vor one asset over another. Applied to data on Cash, demand deposits, and oeD balances, the procedure yields a monetary aggregate that behaves much like the household counterpart of MI-A through the late 1970s and early 1980s (as OCD accounts oc"'C3me more prevalent) and very differently from the household counterparts of si mpl ~-sum o r Divisia r.·l l. The next step, a critical one, is LO determine whether the CES aggregate has any closer relationship to innation and real economic activity than do the Divisia and simple-sum aggregates. In this regard, it is encour.lging that studies examining the connection between money, prices, and omput during the late 1970s and early 1980s have generally found that MI-A has a closer connection 10 innation and real economic activity than does M I. ]j .. SH. IOf tlJlample. Darby, Mascaro. tiM Marlow (1989). Roley ('985). Haltlt (1964). 8M W6IYIUlgef (1964) Economic R"kw-May 1990 • Appendix Econometric Specification and Estimation of Equation 5 The first-order conditions for maximizing household utility suggest the specification (A. 1) In(N I MIA), =Gln( v,) - uln(p" I P. ),. Assuming that v is an unobserved random variable with a positive domain. equation A.l becomes a statistical relationship in which the elasticity of substitution (a) is a parameter to be estimated from time series observations on (NlMl A) and (Pn IP. ). Estimation of a in (A. 1) can, however. be problematic. First, the stationarity of the error term, aln(v), ;s in question and, second, there is reason to believe that (NlM1 A) and (P,,tp. ) are jointly endogenous. To address the first issue. the time series In(NlMl A) and In(p,/p, ) were plotted and examined for nonstationarity. The relative holdings variable. ln(NlM1A). is an increasing function of time and . thus, appears to be nonstationary in the mean. The autocorrelation function of In(NlM1 A) is slowly damping. thus reinforcing the visual pattern of nonstationarity in the mean. Differencing In(NlM1 A) to the form .1ln(NlM1 A) stabilizes the mean and produces a quickly damping autocorrelation function. The variance profile of .1ln(NI Ml A) is roughly of the grouped-heteroscedasticity form (Fomby, Hill, and Johnson 1984, 174-76), the groups being defined for the periods January 1973-December 1978, January 1979-December 1981. and January 1982-May 1988. There also appears to have been some intervention (outlier) behavior in the data at points of change in NOW account availability. To formalize a judgment on the nonstationarity of In(NlM1 A), unit root tests of the Phillips-Perron type (Phillips and Perron I. 1988) were performed. These tests explicitly allow for quite general heterogeneity in the series (the present grouped heteroscedasticity and interventions being a special case). The tests strongly supported the null hypothesis of a unit rool. In a similar manner, the relative holding costs variable In (Pn1pa) was investigated for possible nonstationarity. A plot of the data indicates a "slow-turning" behavior similar to that which would be exhibited by a random walk without drift. The autocorrelation function of In(p,/p. ) is, as expected. slowly damping. Differencing the data generates an autocorrelation function that damps rapidly. The relative holding costs variable after differencing , Llln(p,,1p. ), exhibits nonstalionarity in the variance of the grouped form during the same periods as the grouped heteroscedasticity of .1ln(NlMt A) and , also, exhibits interventions near dates of deregulatory activity. Again running unit root tests of the Phillips-Perron type, it appears that. after proper adjuslment for the heterogeneity arising from grouped heteroscedasticity and deregulatory interventions. the stationary form of the relative holding costs is .1ln(pnlp~). Even though In(NlM! A) and In(Pn IP. ) are both apparently first-order nonstationary, a cointegrating relationship could exist between them. Should cointegration exist. least squares estimation of equation A.l would provide a consistent estimate of (J (Stock 1987). When the cointegration tests of Engle and Yoo (1987) were applied to our data, however. the null hypothesis of cointegration was strongly rejected. Given the apparent first-order nonstationarity of In(NlM1A) and In(Pn ' p.) and the I'wc ral Kc."K n ·1: IkInk o r Da U;tS apparent absence of cointegration between these variables , we conclude that the difference form of equation A. 1 is likely to lead to an amenable statistical specification with meanstationary errors-namely, a.1ln(u,). Though mean stationarity in the regression errors may have been achieved by differencing the variables In(NlMl A) and In(p./Pa ) , there still remains the problem of jOint endogeneity between the regression variables .1ln{NIMl A) and dln (p" Ip,.), as well as the heterogeneity of the errors a.1ln(u,). With the mean of the errors a.1ln(o,) taken to be nonzero , the initial regression equation was (A .2) d In(N I M1 A), = p, - P, d In(p, I p, ), + 810 11 + 820,2 + 830 ,3 + 84 0 14 °/ +£/. + 85 5 In equation A.2 , p, represents the nonzero mean of adln(o,), /32 is defined as G , E, is defined as GJln{ lI,) - /31' and add itive dumm ies 0 '1 through D,s allow for structural change at points of NOW account deregulation. In particular, O/1 = { OI'Omnol:07312 ·lOf740landtnefNft8( O12::: { Olrom73olI07602 11(11" 76 03 and lhelei'le< ments considered were (individually) the lagged values of L1In(p,,1p.), dln (NI M1 A), and dln(p. ) in conjunction with the constant term and the additive dumm ies 0 ,1' ... , 0 ,5. Auxiliary regressions of the right-hand-side endogenous variable, .dln(p"lp. ), were run on up to four lags of a potential instrument (always in conjunction with the constant term and the additive dummies). Using the adjusted R 2 as a criterion for selecting lag length as wel l as for discriminating between potential instruments, it was found that the best set of instruments was the first and second lags of .1ln{p" 1Pa )' together with the constant term and the additive dummies. Because the number of suggested instruments exceeds the number of right-handside variables in equation A.2 , we have the case of a surplus number of instruments. Of course , with the availability of a surplus number of instruments, efficient estimation is achieved through using nonspherical IV estimation , which appropriately adjusts for the heterogeneity in the errors (Bowden and Turkington 1984, 74). Nevertheless , the spherical IV estimator will produce consistent residuals, which can provide the basis for properly specifying the var i anc~ovariance structure of the I.., for use in subsequent nonsphericaJ estimation. The prelim inary consistent residuals, [I = (dln(NlM1A),+ p,dln(p,lp. ),- 0,0,, ... - b50 IS }' generated by using the spherical IV estimates, besides reflecting the grouped heteroscedasticity associated with the regression variables Aln(NI Ml A) and .dln(Pn IPs ), also indicated substantial temporary interventions. (See Vandaele 1983 for a discussion of intervention modeling and the various possible types of interventions that frequently occur in time series data.) The interventions occurred on the dates 73:01 , 74:01 , 76:03, 78:1" 80:01 , and 81 :01 , coinciding with the successive dates of deregulation of NOW p, 011078 10 O13 = { Olrom73 110,18 I I and therea' ler 01\07912 O/4 = { Olrom73 110r 80.0 lapdthereal1er Olrom73olI08012 O/5 = ( IfOf8tOlandtherea'ter Still , £ is possibly heteroscedastic, given the observed grouped heteroscedasticity in the variables dln(NlMl A) and dln(p, Ip )., To obtain conSistent residuals £, from equation A.2 , spherical instrumental variable (IV) estimation was implemented (Bowden and Turkington 1984, 68-69) . The instru- . II accounts. From this preliminary residual analysis, a final regression equation was chosen of the form (A.3) ,,In(NIM1A), =p,-p" lIn(p"lp.), + y, ,I" + + Y'3 / + y'2/1-1.1 2.' Y22 / ,-1.2 + Y2,1,2 + 123'/-2.2 + Y3,"3 + Y32 /- 1.3 ' + Y33 / ,-2.3 + 141'/4 + Y42 /_1.4 ' + r43'1-2.4 + ys,I/ S + YS2 / -l.5 + Y53 /1-2.5 + ,Sl'tS '63' + YS2 /1-,.6 + 1-2,6 + 0,0/1 + 820 12 + 03 0 13 + 04 0/ 4 + 05 0 15 +£/. The only difference between (A.2) and (A.3) now is the Inclusion of temporary inter· vention dummies I". .. " along with the first and second lags ot each, the lags included to account for the inertia in the temporary interventions that is indicated by an analysis of the residuals. Evidently, the money market ad· justments to NOW account deregulation were not instantaneous. The intervention dummy I" takes the value 1 for 73:01 and 0 otherwise. and are likewise The dummies defined according to the intervention dates 74 :01,76 :03,78:11,80:01, and 81 :01. ',6 "2' ',3' "4' "5' " "6 To verify the heteroscedasticity in its error term, equation A.3 was estimated by using the spherical IV estimator while augmenting the previous list of instruments with the intervention dummies and their first two lags as welL Again the resulting residuals were examined. As expected . the intervention dummies and their lags adjusted tor the intervention behavior at the IV residuals (driving the residuals at the corresponding intervention dates to zero), and the grouped heteroscedasticity still remained. Auxiliary regressions involving the nonintervention residuals on their lagged values did not indicate significant autocorrelation. Thus, the heterogeneity in the errors (after adjusting tor interventions) apparently is of the "pure" groupedheteroscedasticity form. In a final attempt at achieving efficient instrumental variable estimation , we chose to use a nonspherical estimator (Bowden and Turkington 1984,69, eq. 3.3) and to model lhe variance-covariance structure of the regression error of (A.3) as being of the groupedheteroscedasticityform. The instruments used were the same as those used in the previous spherical IV estimation. Given the assumed validity of equation A.3 and the specified grouped heteroscedasticity, it follows from the Generalized Gauss-Markov Theorem (Bowden and Turl<· ington 1984, 74) that the nonspherical IV estimator is asymptotically more eHicient than the spherical IV estimator. The nonspherical IV point estimates 01 the parameters of equation A.3 and their respective t statistics (in parentheses) are as follows : F«kul Kcsc .....·<: 'la nk of Oallas lIln(NlM1A), ' 0.01813 - 0.21409l1ln(p",p.), (1.8674) (1.3333) The basic monthly data were obtained from the Federal Reserve Board of Governors and span the period January 1973-May 1988. + 0.231571/ 1 + 0.076601 (8.2063) (2.0497) "' .• + 0.0370911-2.1 - 0.037521'2 (1.3184) (- 1.1108) - 0.0437011-1.2 - 0.00606/1-2.2 (- 1.2833) (-0.2052) + 0.07080113 + 0.0898511-1.3 (2.6385) (3 .3366) + 0.0350211-2.3 + 0.18897114 (1 .3068) (4.7413) + 0.184721"".4 + 0.16253/1-2.4 (5.0113) (1.8171) - 0.192321'5 - 0.01299/1-1.5 (-2.1336) (-0.1441) - 0.14996/1-2.5 + 0.336571'6 (-1.6436) (3.9293) + 0.20069/1-1.6 + 0.04383/1-26 (2.3291 ) (0.5099) + 0.02550011 - 0 .006890'2 (2.3326) (-0 .9354) + 0.01865013 + 0.01100014 (0 .7080) - 0.052880'5 + (-1 .8405) Economic K..""·ic.... - M ay 1990 (0.2860) 2/, ., References Barne n , William A. ( 198()), ~ EconoTllic Monetary Aggregates: An Application of Index Number and Aggregation Theory ,~ jotlmal 0/ Econometrics 14 (September): 11-48. - - - ( 1981), COllsumer DeI/wild (l nd l.abor Supply: Goods, MOlletary Assets, and Time (Amsterdam: North-Ho lland Puhlishing Company) Barnell , William A. , Edward K Offenhacher, anti Paul A. Spind! (1984), "The New Divisia Monet;)ry Aggregates ," j ot/mal of Political Economy 92 ( DC<"cmber): 1049-85. Barnell, William A., :tnd P.tul A. Spindt (982), Divisia MOllef(/l)' Aggregates: Compilalioll , Da/a. (llId lIis/oncal Behavior, Staff Srudit!s, no. 116 (Washington, DC: Board of Governo rs of the Federal Reserve System). Board of Governors of the Fede r..!1 Reserve System (1982), "~'I onelary Policy Repon to Congress. ~ Fede ral Rese,.ve i3lllletill68 (March); 125-34. Bowden. Roger j. , and Darrell A. Turkington (1 984). Instmmenftll Variables (C unbridge: Cambridge University Press). Reserve Dank of Boston New Eng/and Econ omic Review, j a nuary!February, 17-26. The Money-GN P Link : AsHafer, R. W. (1')84), M sessing Alternative Transaction Measures," Feder... 1 Reserve Bank of 51. Louis Review, !\'!arch, 19-27. Heller, H. Robert (988), "Implementing Monetary Policy,~ I-ederal Reserve Bulle/ill 74 Quly): 419-29. Mahoney, Patrick I. (988), ~Thc Recent Behavio r of De mand Deposi t s,~ Federal Reserve Bulletin 74 (April): 195-208. Milbourne, Ross (1986), "Financial Innovation and the Demand fo r Liquid A sset s, ~ joumal of A'loney, Credit, alld Banking 18 (November): 506-J J. Orr, Daniel (1970). Cash Management a nd the Demand for Money (New York: Praeger Publ ishers). Phillips, Peler C. B., and Pie rre Perron (1988), ~Testing for a Unit Root in Time Series Regression ," Biome/rika 75 Qune): 335-46. Chetty, V. Karuppan (1 ')69). MOn Measuring the Nearness of Ncar-Moneys," AmeriC(1II Economic Review 59 (lune): 270-8l. Rasche, Roben H. (1987), ~ MI-Veloc ity and Money-Demand Functio ns: Do Stable Relationships E xist?~ Canlf!8io-Rochester Conference Series all Public Policy 27:9-88. Darby, Michael R., Angelo R. Mascaro , and Mich;hd L Marlow (989), ''The Empirical Reliability o f Monetary Aggregates as Indicators: 1983- 1987."' Ecollomic II/ql/iry 27 (October): Ro ley, V. Vance (985), "Mo ney Demand Predictability, " joun/al of Money, Credit, and Banking 17 (November, pl. 2): 611-41. 555-85. Dicwen , W. E. (976), "Exact and Superlative Index Nllmbe rs . ~ j oun/al of ECOllometrics 4 (May): 11 5-45 Engle, Ro be rt F., and Byung Sam Yoo ( 987), ~ Foreca st ing and Testing in Co-integrJ.t<.'ci Systcms: jOllrual of };.(;onometrics 35 (May, Annals Issue): 143-59. Rotemberg, julio j. ( 1989), "Monetary Aggregates and Their Uses: A Comment on Barnett, Hinnich and Yue's 'Monetary Policy with the Exact Theoretical Rational Expectations Monetary Aggregates'" (Paper prepared for the 14th Annual Economic Policy Conference sponsored by the Federal Reserve Bank of 51. Louis. November). Fomby, Thomas B . R. Caner Hill , and Stanley R. j ohnson ( 1984). Advanced };.(;01l0metric Methods (New York: Springer-Verlag). Spindt, Paul A. (985), ~ M oney Is What Money Does: Mone tary Aggregation and the Equation of Exchange, ~ jotlma/ of Political Economy 93 (February): 175-204. Gioson, Katharine (975), ~The E'lrly Hislory and Iniliallmpact of NOW Accounts," Federal Slock, James H. (987), ~ Asymptotic Properties of Least Squares Estimators of Cointegrating " Federal Reserve Bank of DalJaoi Vectors, Econometrica 55 (September): H 103>-56. Stone, Courtenay c., and Daniel L. Thornton (987), "Solving the 1980s' Velocity Puzzle: A Progress Report, ~ Federal Reserve Bank o f St. Louis Review, August/September, ;-23. Vandaele, Walter ( 1983), Applied Time Series a nd Box-Jenki ns Models (Orlando, Fla .: Academic Press). Wenninger, John (984), "The MI-GNP Relatio nship: A Component Approoc h , ~ Federd.1 Reserve Bank of New York Quarterly Review, Autumn , 6- 15. EconomJc Review-May 1990 " Mine K. VOcel Carol Dahl Senior Economist Federal Reserve Bank 01 Dallas Associate Professor Louisiana Stale UniverSity Reducing U.S. Oil-Import Dependence: ATariff, Subsidy, or Gasoline Tax? ntTC;I .,~·d oil import:- Ita\ l' GH I:-..:d gl'O\\ ing yL·:II·~. I.ow oil priL:L'''; l'r)COll r;rgt:d nll1~umption. dl'LTl'a~l'd dOrrre~liL: oil pro· dunion, and It'{! to higher irllport~ BecaUM: ll.S. or! l'iL-kb arc rciatin'!y m;Hurl', ,Ill' (h:dinc in durn(',,1\' produl'fioll and l" p lor;Hion \\'il1 n>nlinllC, nt.':ltin,g ;r morl' "l'riou, oil-import prohlt:1ll ill Ihl' fullll'l' A di"rupuoll of oil I1l1pOl"b o r ,harp inlTl':r"l" III world oil prin'~ could adn:"I'St:ly :rlfl'('1 11ll' l ' ~ l'Conorn~ 111t.'rdof\:, "ollle oh.-.ern:r:hdil'\ l' Illl' L'nrll'{l SI:IIL', ,hould n:<llll'e import:h~ fcduL:i ng l'On""ulllplion :rnd pfOtL'{'linl-\ till' (\11I11l'., tic oil indlblry, Thrl'l' rlll'a,un,'" ha\l'lx'1..'n "u~c"ll'd 10 10\\'1;'1' U.S uri Import.,: o il -import fl'e~, "uh!'oidlcs 10 dOrlll':-lit' oil pmdul'ef!'o, and;tn in('rt:a~1;' in Iht: ga-'(llim' 1,1'1: I Initially. Ihl·-.c: pollne . . \\oul(\ 1I)\\t:r ('ul'1'('nl lInport ... by eithl'r rl'ducing con'lImption Of inCl'L':hin).t dOI11t"~tic pro· dll('liot1 Thc tariff would InLTl':hl' pfoduction and tlt:cn.:ase n)l1~I\lllPti(Jrl. till' . . . ub~idy \\oukl innt:a..;c hOlh pro(\ul'lion :md C011,""I11Pliol). and 11lL' )o:;I.~olil)e tax would n: dll('(' hOlh production ;mc! nm . . . umptlon . BeGlu~ tht' l()n~ 'll'nn d'k'ch of Iht.'Sl· polidc!-o may diffcr from thl' .~h(jr1-tl·rnl cfTt'cts, Wt' dl'\'i:-cd;r mca .. urc of t·.S. nilncr;lhitity to oil,upply .,h(X'k.., and 1r;1('L'd Ihi" nll':t,,>Ufe O\'t"r linlL'. \,\'(" nl!l"idcred thrn' ";fl\-'cilk ptllit'ie~ di!'ol'u!-o,';l'd h~' polil'ymakcr"-:r 2":i-pl'lu'nl im porl t;lriff. a S')flL'r·llarrl'! "uh.. idy to dome"til' oil producer:-. :lnd ;rn in<.:rc:l!-oe in the ).(:t!-oolim' lax from 9 cent:- 10 1":i ccnl.~ pl'r ~allon-;I nd dcll'nninl.'{l which policy \\ould m.lkc Ihe l 'nill'{l !'>Iall':- lL'a!'ot \ ulnt'r::rhlc to oil':-lIprl~' disruplion:- ha"l'd on ollr llle:I.,>Ufe of I umn.:rn in rCL:l'n, F.conomJc Rt,,\'!cw - M ay 1')90 nrinerabililY . III :In l'ar!il' l papr:r ('{lied and D:lhl 191-19). wc C\'a lualr:d thc t'o"ts of Ihesl' polkit:...; in Illl' ab..cn<:c 01':1 disruption In lhl.~ :l rtil'k-. \\'l' <ll'tennillc the henefit.... of the p()lidt:~ in Ihe crSl' of;t di . . . n..rplion. A vulnerability measure \"'e nll';l:-lIfCd Illl' l llllll'{l SI:IIC'>' \·ulner.thility oil·.'u pply "IIo<:k . . h~ Ih(.' ch:l1lge in l't'onornit' "elf;rre resulting from ;r l'h:Ulgl' in import . . , Wht.'fl' 11Y.4f(/rt' b dcfinl'{l :\ . . (.'onMrl11ef plus prodtKt'f :-lIfplu ..., The llle:\"Ufe c.tkulates the C()~I of ;tn oilsupply ~h(X'k for t':lt'h Yl';rr in a 50-yeilr lillle hori' zon. Thus. the co,1 for l';ll'h },e;lr is Ihc Ch:1I1 ~C in \\l'lf;1rt: if lhe lii:-rtlplioll had occurred in [h:1I p:lrticubr year \'ih- cl lnll;l[ed the vulnl'r::rhilily rnC;l.~url' undef l·tlffl'lll policy and wilh each of the proposl:'d p01i('il-'~ 10 delerminc wh~lht:r tilt" new poli<:ie!'o \\Ollle! I'l·duct, the CO"tS of an oil10 We WOIJld "ke 10 man/( Sreph(}(l P A Brown. John K H,II, Ke,m R Ph,II,ps and Mark Frf)fICh lor helpful convnents and d'SCUSSIOfIS and Dav,d NBnns lot excellent research assislance rile US Senate Fmance CctrmItt!!B protJOSed an oil· Jt1100II fH II 011 IInpOI'ts WCfO prOfOC/6(J 10 exceed 50 per. ccnloldlJmand (L8ngIey 1981} Pl'CSldenIGeotge81JSh.1Il hss csmp8'f1'1. suggested a S1Jl)Ss(1y 10 011 and gas produc , ers ReprCsertlill've DanteI Roslenkotl'sk, rf!COfflfT>ende( ncrCBSUlg lhe CUIT(}(l/ 9o(;f1f11 C~ctS6 tar 011 gasobne by al leaSi 15 cents pet gallon (Blmba<X1l 1990. 1989} Federal ResCfV9 C/laIlman Alan arfJCfIspan also argued mal a f}BSolJM liJX WOUld do 'less harm' than 0If1ef /eWc$' ($loot '989) .7 supply {h:-ruptltlll ! Tht: 1lli,.·;L,urt: of \'ldn ...·I:d)ilily dqx:nd, o n 11K' dOIllt::-tit· pri!.:...· of 01 1. tht: ~Iur...' of illlport,.. III do1Il...·,..tit· {·on,umptlon . ;md till' da'liciu...·:- of supply :.nd (kmand durin).: till' di~nlpllon A 10\\ im port .~hart: \\ould m;lk ...· thi,.· l ·nih..'d ~1 ;Lt ...·, k:-., \·ulnt:r;.hlt: to :10 o il-:-upply ,hoek. :md a hi).:h lIupon :-han.· \\()uld lllakL' 111'..' t 'nit ...·d Slalt:' m ol"\.' \,u lnt:r.lhk \·ulnt:l:th,l ilY j, It:.,:- \\"1lL'n {km:md :md supply l'1;1~lidIK':- art: hl).:h 1\ high (k-mand t:b~lil"ity 1l1t::!n, th;1\ do.~t: :-uh."tllul<::- for oil :11\: :1\ aibhlt: 10 con.,umt:1"'o if ;. :-u ppl)' di,ru plion ocem:-. 1\ high ... uppl y da:-lil"il)" illlplit:.~ Ihal dO!m::-lic pnKll1c...-r:could r...'spond ljl1it·kly 10 a ..... ,]1ply di:-nlplion by innt::I... ing d(lllll·sli{· (Iii pr<J{iUCli(1I1 In 1)(1111 C;IS...·S. :1 Illajol" r""(itKtion in for...·ign imports would he k-,.~ . .u ...lly for 11](.' lInil",'d S[:ucs. B',:C:IU't: Wi,.' cdnl lale till' drL'eh o f ;1 t . . . mpora ry diMuplion. \n· u ......... sholl-run ",'bsli("t[it:.~ in Iht: mt:a,urL' How Ihe policies a ffect vulne rability !'()Iit it::- Ih.u (·han).:...' d()ml'~tic nm~utl1pli(Jn. dom . . . :-Il(· Pf(){.iu'·llon. or Iht: 10n).:-1\m supply t:bstieilY in :1 ).:h·. . . " yt:ar :llfo.."(·1 ndnl!r.lhlliIY tn Ih:n y ...·a r To Glkul:tw tht: n llm·r:.th illl} lllL':I.... ur...· \\ t: Uli lizi,.,(l tht: r ...•... lI lt ' of our fOrL'l"a"lill).: m(xld. \\ hldl Gdcubt ...,... OIlIpUt p:lIl" for Iht: Or).::lIl iz:tIlOn of P l'lro!...·UIll Expt.)11ing COlln lrit'.s (OPEC) and 111(' l ·nilt.:d Slalt:', rh l~ Iht: prict: p:1I11 of oi l for the Ihft't: roli(:i. . . ... and for :I hast: C I ......' (.~('i' I /)e I)(n· rilled ·· W·ol"/d Oil ,I/odel") \'iI ...' itknl il"it.:d lilt: h:I.~L· (";I'l!:t~ (·UITl'nl policy. tlK'n impJ...'Illt:nll'd Illl' Ihrt:t.: policil.'.' 10 dt:[t:1"1l1ill(.' how lilL'Y ch;lIlgL· Ih. . . pIke :md Olltpul p:IIIl ... For . . . :[(.:h policy. \\... . . pelformt'd a s~'nsiliviIY a!l;IJy.~i .. using a \·a fiL·I Y of ~h()rt - rlln dem:lIld :tnd supply t:1:t~lkitil!."" T IK' sh(Jn- run pri,·c elaslid[it:.~ We mod,/,ed IIle measure lOt /lJe Ihtef) pollocs 10 1flC/vae subSIdIes 8Ild lax and lar,ff fCo;erlUe5 The appen(Ju. PIe- sents Iflos procC<Jute Notdh81.lS ( ' 980, iJIId Hog/lfl (1981' a S5(ll)ll mat the shott· run elastICity'S OI"le-lenrh me Iong·run elastlClly Btown and PnoIIops (1 989, hod a lac/Ot 01 one·seveom We C()(>dvclec a sensolfV'ty 811afysls uSIng laclOts langrog /rom Otl&-lIatf 10 ,. one·t"'f!fIllCtn The lesulls reponed ., rtlIS al/tcle ale With a laclOl 01 OI1C-lenth World 011 Model We use a dynamic optimal control model set in a partial equilibrium framework to simulate time paths for oil prices, oil production , and consumption. We model OPEC as a dominant firm facing U.S. total demand for oil, less U,S. domestic production and non -OPEC imports to the United States. Domestic producers are profit-maximizing price takers in the crude oil market. Both the United States and OPEC own reserves and maximize the present value of profits over a 50-year time horizon. We simulate the paths of the variables using an isoelastic demand function for crude oil with a price elasticity of -{).9 and an income elasticity of 0.8. The demand function was derived from domestic demand for products and normalized around 1987 demand. The cost functions lor OPEC and the United States were also normalized around 1987 production costs for these regions. We assumed that U.S. income would grow 2.5 percent per year. The discount rate was 5 percent. t For lhemathemat>eal &p9C,he:allOll oIlhe problem and the spec,foe: Iunc:honal IOfrTlS . see Vocel and Oahl (1989) of dt:mand u"'L·d III our Cl il"ublion of I Ill· \·llln ...·r:Ih ility I11C:I.'-llrl' ar . . . lakl'n 10 Ix· \':triOll ~ fraclion, of th ...· long-ru n t!""ma nd L'b:-liciI Y. a., .'-l11-t}:c'fL·d by til\." t:conOlllK .S lil ...·r:llur...· , Tht:,t: t:b~liL"itil!.s do not vary a\."l"O,:-- poitcii,.·, Enmomi...." ).:t:nl'r:.III} find Ih ........ !tot1-nm supplr ...·b .. linly o f oil to lx' \.i,.'I). low. ranging IX·I\\ ...·L·n () :lnd () I . IIllplyin~ a \ cry ,mall olHplll rl'''' I")()1l~ 10;1 prit't: dl:tngc (Ilo~:11l 1981. :Ind Bro:.thll:tll :uld " o~,tn 19WH. Tht, ...mall OUIPU! rt:~p(Jn.sc.: 1"\.: ... \111:- from \ .... J!. h i~h Ct p:lt·iIY uliliz:llion r:11 ...,... in 011 pr(XIUClion [n til\." ( .;lX' of:1 di,fuplion in Ill1 pOn:-. hO\\ L'\·t:r, thL' prie~' inl'fl':lst: i:- Fcdcr:al Rcse rve Bank of l)aUas :111 un~XI)cl1e<11'oh()11-tcrnl \\ lIldfa ll for till" proonc mit-tlt1 i.:xpt.'ct donK·stit· produ('l-r.. to dr.lw from pm at~ !'Iockpile!' 10 t:lke :ldv:tlll:lt-te.: of Ihe.: tl'1ll1'X)r.lry price incrca1>!..' Ht.·nCl'. the !'iupplr da,>tidty I.:ould Ix: fa irly hiAh,' \'(Ie clkul:ltt.·d the \\dfare co... t ... of a .. upply shock u .. mg ;t \'ariety of .. hOt1-rl.ln "upplr d:t~t ici tk:-. r.tnglllg fmlll 001 to.t Thc ...,.: da .. ticil ies arlO ohl:tlllL'd hy I~kin~ \':triO\l" fr,II,:l1Im .. of the.: Hlpply d:t ..tidtil·" Inherent in till' .. irnui:lt ion ... Base casco L'mk'r l'\lITCIlI policy. oil prin:.. ... tart at $20 per h;lrrl'l (1Il 19~ - dollars) and rbl'lo SKo~6 pCI' h;md in ";0 Yl'ars (CharI )), Simuhlll·d U.S produ(·tion cb:l inl's fairly ...Iccply d uring I Ill" "iO-yc;tr linK' horizon (() a houl 19 pcrcent of its initia l len:1. whik O PEC protilll.'tion is f,lirly .~t:lhk (Chari 2) Fin:t] OPEC production is on ly <.)2 pl'rn:nt of i1.~ ini ti:ll Icn;:1. rl· ... ult ing fro111 OPEC's \'c ry high 1e"l'l of rc ...elYC." ('oupk'd wilh rl'l:ttivdr 1()\\ and ... Iahk pr(x]uclion co .. t-. Thl' u.s im]'Xm .. I!;tfe incfl,::tS!.:.·o; steadily th roughout Ihe -;O-year tinK' horizon 10 9·, l)Crccnl duc~r In .. uch a (·:hC. or nm"'llillption :t~ (J , ~ rCM.·f'\'l·~ :lI1d output dwindle and prod lin ion nl.~h 'o():lf 1111;" welfare co.. t of an Oll-Sllppl)' (h.. mpl ion. h:lsed on our Hllnl·r.lhihtr IllC:.... Ufl·. abo innl·:t"<;·... Ihrol1~h time du(' to IIlL' dedinc tn dOllll· .. tic p roduction :tntl ,ht· inn!..';I' ...· in IlHlxms (Ch:u1 .H Ch:Uh " and 5 sho\\' Ihl' pnc(' p;tth:; and dOIllC'li(' production p:lIh:-. \\'ill1 thl' h:l"C ca:-c :lnd Ihc Ihrce IX)itc-ics analyzed i n till." anidc Chart 1 Price Path lor the Base Case Pll ce IU S dollars) '00 Chart 2 Output Paths for the Base Case ProductlOl1. bollton ba.rels pet year , OPEC 2 United State s oL-~=======: 1987 1997 2007 2017 2027 2037 Tariff. 111(; 2-;-jX!rCl:11I tariff increa'>l'1> thl' dome''''ic pri(c of oil hy appro.ximaldy 7 )X'rct'nt to Ihl' carly }'l·:Ir.-.~ T be h ighcr pricc1> lead 10 hi~hcr U.S production. resulting in an earl ier dcpktion of Ihe l'esouTl'c. h i~ her produclion costs. and suh"<;'qllenlly a lower produo ion r.IIC in Ihe fUlllrc. O'·er.lll. lOl al dl'>COllntcd U.s pr<xlucer profits ovcr the -;0 ye:lr.-. increase hy 118 pt:.·r<.'t'nt. OPEC. howe\'l'r. i ... wor.-.c off becau:)t; it ahsorbs a br~t" r ... h:lrc 01 the 1:trill wilh 10lal discounll'd profits falling hy 29.2 jX!n·cnt. By Tcdudng t:on .. umption and incrt':lsing dOIllCSlic Imxlu{·llon. thl' lariff decre:l1>es illlpon sha res Inili:11 IIUpOI1 ... hares wilh thi: ta riff :lfl' lower th,11l with th!: b:l.~c ca:-e for about 40 years (Chart 6), As de.:pletion dfccts reduce dOI11l'stic production nl';l!' Ihe end of Ihl,; lime hori zon , Ihc eo .. • Tho Umle(J SlalC!; llisa has !;/OCkp!lcs of 011 m /he SflaleglC F'rllroH!um RC!;CfW If tho Icdcral governmenl drew 011 from me StrategIC PeI.cHum Reserve dutmg an 0II-mp0t7 a,srupJlOfI. lho 100ai dotne!;/IC !;VppIy re!;p(J(l5e would be .e/8t",ely more elastIC 20 oL---________________________ 1987 1997 2007 Economic RevieW_ May 1990 2017 2027 2037 • In our model 01 m.oilmatket t'ICJiI-OP£C mpotlS of oil are IIcpt CO'lSlltnl III tf'/eIf 1967 levels If non-OPEC imporl!; were II IuncllOfl 01 prICe. /tIefl a U S 18"" WOUld 1e8d /0 samewha/ fItgher domestIC prICes 8fId a smaller decline in world oil ""e(]S " Chart 3 Vulnerability to Oil-Supply Disruptions, Base Case Scenario Vulnerabdll~ MeaSUfe 900 800 700 600 500 ElastIC Supply ' 00 300 200 "0 f9 87 1997 2007 2017 2027 2037 d t.'cn:;I ~(; If) impon ,~h: l rc:-; hecomt;:s mort;: modest. tari ff .11-.0 lowt:,"" the \ 01:11 o il-impon h il l 1X'C:IlI!'>(.' il de(,:n.:a:-L'''' !hL' \\ o rld p rin' of oil Wlw n a di,ruption in o il i m pon:- (X'("UP.> , ho(h (un!'>u llIcr ;md p rod llCcr p rin's i ncrt.':I~. If dOIllt.·... ti(· p r<X h K C P.> are vcry unfcspon:-in: 10 the n it i n oIl Illllx lO _ tha t i:-. if the :-hon-nm :-.upply is inda:-t K--n m sll me~ bear mo~t o f the price inn"'-:I'-\.'. PnxllK cr.- arc hell\!r o ff \h ;m constllllt:r.. after :1 disruption 1)(!C:HI ~e pn:x1ucer:-. TCe<:I\'(' a h igher pr iet.' for IhL'ir n.:b !ivd }' unch:Hlg/;.'<.! kVL'1 o f out p ut T he highn price:-. :lIld lower d omt:stic consumption \Villi Ihe 1,lriff dam pen Ihe loss i n CO!1SUlllcrs' .sllrpll.l .~ duc 10 the sud den disruption in oil su ppli\!s. T hc o ve l~ tl l wdbr\! co.~ t o f Ihc o ilsu p pl y :-;hod.;. ;t.' ctlcub1Cd by Ollr vul nt:r:thi l ity IllL': I.~ UI\· , is lower wilh thc I;tri ff th:1I1 wi th the hast: ca:-e. On Ihe otlier lwnd , if do mesIie su pply is n :ry (: la:-;Tic and respo nsive 10 a cut i n imports. d()[llt::-;tic Oillput L'xp<T nd:. 10 fill the ga p crt:.':lIed h y rt:duced impons, and consumer p rices risL' l iul;.: . TIlt." Chart 4 Effective Consumer Oil Prices, Expressed as a Ratio to the Base Case Prices 1.25 1.20 115 1 ..==:::::::::::: 10 ~"~~~'~11__________~==" t.051 _______, .. .~::::::::::::::~::;:::: t.CO t Base Case The ptescnl varve calc..-arlOf1 a5S(.mEI$ an eaval ptooatJil. Ily 01 an OI/-$uppI)' d<$lUpl1Of1 fey each year .wl lhe 5().year 95 tme hotl.lOtl 90 A downward $hdt r!lal IS fIO/ parsllellO the /IlI/,SI demano funcll()tl would lead 10 SIJC/) a resu/f fey 20 rL' ... ultinl-t in w tin ~ in producers· ...u rp l us and small los.x·:-. in con:-;ulller.-· , urplu ... Al-t:.li n , thL' wdf:m: ("{):-t i:-; 1;.::-..... with the l ariff tha n \\ il h 1111;.' h:I,t: (';I:-'L', alt hough IhL' difference l)(!twcen the two li me patlh I~ ...OI11L'Wh:1I lc:-.~ Ihan in the inda-;til' l':'h e. Tilt: lotal p fesent ,·:11(11..' of thL' vulnL'mhility mL':r"'ll ft: O' L'r the enlire IJlne horizon r:- always 1L':-:- w ith tilL' tariff than w ith the hast: (';1'><.' ." The d,,'crea ~L' TO tht: "ul n\!rab ility Illca ... Uft' range:- from I I 'i rx.'rn'nt in thc case o f "('ry ind a:-tic Mlpply to 2 I pl..'rccnl \,·hl.:l1 :-u pply h highly d aslic Subs idy. A S-;-pcr-barrcl :-.ubsidy \0 t!omc:-;lic o il PIl KhKL'TS lowt;:rs prices in the .~h()rt tL'nn hut 1~ li .'-\.' ., lo ng-It.'rm p rice:-; for oil. Tilt: :-uhsid y IO\'·L'rs the d fe('live co,~t s To o il pnx lu("\!r:-;. 1hus d r; I.~ l i (,·a l ly in(TL'asing in iti:11 U, S, oil p n Kl uCli( m . T hL' role of thL' suh-;idy :IS a ca t ~l lyst \ 0 d OIllL'stic p roduction (1;.:t.·..e:l.,es with timc ht:("au .~e thL' suh:-;id y j.., (·OT1st:lnt and iI.' d isco untcd value dcc rL':'ls<:s I hrough (illlL', A., WiTh the ta riff. incr\!:lsed pr<x1ucl ion k a(b to L':l rlief dep ictio n of rc ...o u n::es. h igher (·o~ [ s for l '.s pnxlucers, and cu n ailed p r<Klun ion in 1aH.'f years Alt ho ugh sim ilar to the t:triff C:bL'. thl' :-llh:-Id y m agnific ... these d lt.'ct:-.. Stl rpri,~ ingly . tIlL' ...uh:-Itl y also i ncrea:-;es O PEC p nxl u('lion in Ihe l'arly ye<TfS. Innc:l -..cd Li.S p rodu('tio n in tilL' C:l rly yL':II~ n.'dut:l'''' O PEC's m arket sharc. i ncrt.'a:-.('s O PEC':-; (kllland d asticilY, and results in higllL'r O PEC pmdUt""l10n - a monopolist Subsidy 851--_,----,-_ _ _,----,-_ __ 1987 1997 2007 2017 2027 Fedcl1Il Rcsco 'c Bank 2037 or Dallas Chart 5 U,S, Production , Expressed as a Ratio to the Base Case '" Gasoline T'1t lO': F~2 Base e.se 90 Tan!! 80 7',L,c---~=c---ccc:----c-:=:::::c:~-c" 1987 1997 2007 20 17 2027 2037 \hhollgh th ..· ,uh"id~ 11:,,, 11ll' It}\\',,',,1 irnport ,han: 0 1 th ..· Il1rn' poli(lt·, 1'£11' thl.: fir-I [hr......: ~ ,,':Ir:-o, ,mlx)rt "har..·" q Ul d..I~ h..Tom ...• high ... !' wilh I h.., "uh"l(h Ih;ln \1 ,Ih "'llh"'r Ill\.' 1;lril)' o r Ih ..• g:l"olin ..· 1.1' 11K' "ul h,(h ..'lln,u!':I).:..·" ("(llhlllllplion h~ IO'\I,'ru1g (,OIl'lHlll.'r pri(I.'" Th ... in«...;,'>1.· i ll dOllll'''It..' "\Ippl~ mor.., 111;111 off..... ,,, Ih ... in<'TI.':" I.· in (on:-OUIllPlioll in Ihl.' "'arly ~"';I r'. Ih ..: rl.'by lo" ... ring imlxllh. 11011 "'I l'r. l11('r,,';I'I." In 1'101II ('on.,u l11 ]1' ion and :-upply k'ad [0 a qUit. k ... r dl.'plt.'lion of Ih ... dom""lic IV'><lUn ... h,t...., .1Ilt.! h igh ... r illlport ... in rUlli1"\.' r ..';lr, TIll' h)I,,1 (("11.1' o i l import" i:- ;,1,0 high... r \11111 th .. · 'lIh:-oidy 1)l'CltI'... 01':1 higlll'r \OIt.IIIH: t.f IItll)(1t1'" TIl\.' :o.1I11." til 111\\ ,,'r, Ih .. · nJ,,1 ()f:lll (Jil-'\Ippl~ di'J'UPlioll for a :o.hnrl lilll ..· if ,hon-run :o.u ppl y h ind:l"lil' Th ... 101\\.:r i lllport ,h:IIL'.' 111 the l'arir Y... :II·" hl'lp (ll.'''-'''..';I'I.· Ih .., 11I1!l\.'r:lhilit\' rll ...asu r..: As i mpOrt:o. Ill(rl.'a'..· III I:ltl.'r Y"':l r, . hOI\ I:\'I.;r. oi lsllpplr di'rUplicl1l' Ix:nml\.' co:o.tl~. 'I'll ..· magll itu(k' 01 t he 10.... in I.·Olhlllll ..·I.... 'lirplu' j" higlH:r \\'il h tIll: sllh.,i<h' h,,'{';HI'l' of imn:a, ..'<1 <lollll'"ic ('( m,umptioll 1klKl'. ;1 dl:l.'i'l.';i'''· III imptJlb "'qual 10 Ihl.: uriff ('; ' '1.' 1\ ,II 1~"\lh ill hl~hl.:'r ('OIl'UIlK'r lo,~c', ;lIld Ihc (,(.,,1 (11'.1 dl'l'llp' ion. h"'l.:d 0 11 (lUI' \\ll ner.,h ilit\ 1Il\.':I,ur..' 1\ ti l hI.: hl~IH:r \\.11 11 th ... 'uh'id~ than II illl Ill ..' t:,rif1 OOIllt.:'11l prll·..·, \1 ill 1101 IIltrt:'a'... 'lIh'I.11l[i;llI~ if demand I' \"'ry d .I~ li t. and ('OlbUIlK'r Economic Kt:\'lc w - M :I)' tm IO'.' l'" \\ill Ix: \I.:r~ '1ll:11l \'("itl1 Ill\.' lar).!e i nLTea:-.1.: in dOIll""'K OUtput. Ill\.' :-OUlbld ,... , p:lid 10 domt'!'otil prod m .... r-. II ill inu-c:l't.: !-!1'.."ll ly. :ll1d Ihl.: policy will ,tI\\;I\'" IX' more co",tlr [h;lll Ihe h:I'I.' l':'~<" III the ,,' 1 l'lll t.f :111 ()il 'lipply dl'rupti()Il 1'11 ..' !()(;iI pr(o'.~1.'1lI \:tlu ..' (If Iht: I'ulm:r.lbilily 1I11_'a.,ur<.: j, ;,I\\:tys higlll'r ilun tilt.: lM'1: (-;1'>1.' \11111 Ih ... ~U l hid~. r,,·g;l rd k ..., I I I d:l.,ti('ili ..', Gasoline lax_ An illlrt:;I'>I.· in th ..• ).,:asoline 1:1:\ from 9 t.:l'nI:-o [0 1; n.:Il'" pl.:r gallon drivt's :1 III.'d).(1.: l X't\\'t'C1l p roduc ..... :111..1 ClUl'llllll.·r pnn;:s, II ilh I ll\.' i lll..n::I'", III ('On'l lIlll.'r pri('t:~ 1:l r).,:t:r Ih:ln Ill.... dl.·crt."I"t.: In produccr priCt.:~ Bl'CltI.'e th .... di,.,lOUIlI .... d l.tllIl' of Ill ..· prcKlm·..'r:o.· I.IX lia hility d"'l'I'''';I'o.:.. lI'ilh lim ..... [II ..· gasolin ... I:IX in dfec\ pO'lpon .... ' ! ,S prodll ~'liol1 T IHI:-o. dOIllI.·.' lic pro, dud ion tll'dint::-. only in I II .... l.'a d)' y...:,r,." Til.... g:t,>ol int.: I:IX 101\\'1':- illlPlJll " hare.' o n :r I ll.... \.'nlirl.· l illlt.: horil.on TIll' I:IX r..·dlll'l:'" both ('( 1II'Ulll pI IOI1 and d()IllI.·:o.lit I)f(Klu""l ioll h~ iIKr..'a .. ing tht.: COlhu nwr pril.'\.· ;lnd dt.:LTt.:a:o.mg [hI.: pf<KllIc ..'r price Ikclll"l' Ih ..· la~ I')(J'l polle.' produ('\ion. 1I111)()11 ,h;l r~" II ilh Ihl: .l!.L'olint.: tax arc nOI Ihl' 10\1l'... 1 Ll1 Ih..· ..':\rly 11.',1r-. Frol11 (he Ihird ~1:;lr on, bo\\t.:\·..·r. I Ill' ~a .. olinl: 1;1,'1; ha' [h .... 10\1'(;;"1 illllxm .,h:lre of .tllT hl: Ix)li('ll". JIll hiding Iht' ha"", Ch..' LOI\ IIUpt'" ~h.II\::-. hO\I..:I ..'r, do nOT 11"'('",:0.~;I ril~' IIllpl)' ;t r.."<.lucl ioll III d,'rl' plloll ('O:-Oh Ch;trl - ,hem., tit;lt \\'llI_:n ;o.hol1-rll n " upplr b quile ine- Chart 6 Share of Impons in Domestic Consumption , Expressed as a Ratio to the Base Case Impon Share 1 04 Subsidy 1.02 Gasoline Tax. 92 l---------:----:c':::--~ 1987 1997 2007 2017 2027 2037 21 Table 1 How the Policies Change Vulnerability: Total Present Value of the Vulnerability Measure Expressed as Percentage Change From the Base Case Short·Run Supply ElastictUes· Tariff .01 .06 .&0 .15 1.5 2.0 3.0 -11 .5 - 11 ,3 -10.9 - 9.4 - 7.0 -5.2 - 2.4 Subsidy 3.2 4.5 6.5 12.3 17.3 19.5 22.0 Gasoline Tax • .7 7.1 4.' -.3 -3.6 -4.7 -5 .• · long·run elastIcities vary between poliCIes and over time The short ·run elasticl~e, shown above a.eaverages tor lhe base case. The actualelasticitlllS vary in proportIon 10 long· run supply elaslicWes lor Bach Ollhe Ihrea polIcies . b ..tic (;u"Ound 0 .06>. Ihl.: l-t:.,;olml.: 1;1,. "hid) 1()\\I.:r.-. 1I11f>< )n .. harL·:' Ihn)lIglJ(JUllhL' "iO- ~l·ar tlllll.: horil.on . Cll} Ix: \\"Ol~l" th.Ul Ihl.: h:I"l' ca"L' \\ Ilh rl.::'I'lt.:Cl 10 d i"ruptioll (·o ~t:. If :.hol1 -rlIn "upply i" :1~,.llIlll'd to I'll.' \ I.:r~ · llllrl.:"ponsiq:: to priCI.: ch;lngl''' I III I hl' l~lI1gl.: of 0 to 0.6). ;t di.o;ruplio!1 in il11POlb I">n'ol11l.:." co."II~ wl\h Ihl.: g:l.,olill1.: lax. 1,0\\ dOllll""IIC production in 11ll' l':u'ly yl':tr~ (·ollplt.:<] \\'jlh l1nrL'~I)()lbl\1.: Mlppl~ 1ll:l kl''' thl' l "ni!(:d Sl;tll.:.~ Tl101"(: \ lIlnL"I~lblt: 10 ~llP pl y"Il<Kk:. If " hon-fun :'\l ppl~ i~ Tl1on: 1"L':.pon:.i\'1.: to pnn: ch,mgl'''. howL·n'r. 11lL' di~rtlrll()n in 1111pOlh dlX:" nOI IlllTl.:;I"L' nm"llIllL'r prll"l"" \ 'lTY I11 l1ch T hL" 10\\"l.:r impo rt "h:lrl''' "lI,d 1()\\L'r CO II "lllllpllOn w!l1l Ihl' g;\ ,~()linL" I:c\ lX:l"(Hlll.: Illl" d0111111;\11I raL'!o1".". and \"llhll' I~lhilily dCLTl'; I .~l"S \\ IIIl Illl' g:lsoIi IlL' l:IX. As Tahk 1 :-.hows, 11K" lolal prl'SL'!l1 \'al lll' of vll]nl'I~lh llil r wilh l ilt.: ga:-.olinL· l ax dq"X:nd.~ \"l'l!" l11uch Oil lill;' sh o!1-run L'b~lK" i ly of ~llpp l }' 1iL'- GHI,.c domt.::.IK prt.J(luClioll i!'> lo\\cr wit h IIlL' gaso· linL' 1;1-':, Ihl.: pn x lul'n:'>' rt.:"I)(m:-'L' \0 thL' d i"rupl ion in 11ll1X)lh IX'(:o1lle:-. a dl'ci .. i\'L' fanor. ThL' g,,"~o l inL' 1:1'1; 1llakL'~ Iht.: l'l1 itcd Stall'..; morl' nl l n{'I~lhk if ~h()r1'l"un ." upply i.~ \'lTy 1J1L'Ia~l i<". fh tht.: ~hort-run :-.llppl)' 1'1;I.~tlnly incrl'asL·:'>. hl)\\'I.:\L"r. I Ill" :'::I-"1)liI1L' l a.-.: IO\\'L'r.-. dbnlplion U)"lS . Summary or poUcies, Chan." 7 and H ." how till.' r:tnkHlg." of Ihl' thrl'L' pol idl''' wilh l"L'·~f)(:cl 10 Ihl" b:I'l" {'ase O\'l'f timL' Till" 1:lrill" dl'arl}' II Illakt.:, till" l "l1 lll'd St alL".~ Ie"" \'ul lll'r:l bk 10 0 11 "lIppl~ dl'fliP110lh \\ hL"tlln 11ll" ,.hon -run :,upply rl"I)(ln~L" i" L'1.I'IIL' Of inl'b"lk. The "ulhldy {'ould IO\\l"r di,rup"on ("O!'>I~ in IhL' :.horl IL'!"I11 If "upply \\.I~ \L"r~ I1K"b'lK" OIIll"rwbl',llll' l"(),,1 1)1':1 diSruption i" Illghl'" \\1lh thl' sllh~idr 111:11' \\illl I lll' /):1:'>(' GI.~l' ,"'l111il:l1"iy. 11K' ga"olinL' I:LX IO\\l"I'" dbl'uplion Chart 7 Vulnerability to Oil-Supply Disruptions , Inelastic Short-Run Supply (c.::0,06) . Expressed as a Ratio to the Base Case Vulnerab.bly Measu re 'IS J-_ ...:Ga::::SOline TaJ 110 105 'OO L-__~~______-.~~=I Base Case __ ~ 95 oor~==::;;;:::~~===:~: ~:':'"~~~:;: .5.~ 1987 1997 2007 2017 2027 2037 Feder a l Rcsc: .....c Sa n k o f Oall;os Conclusion Chan 8 Vulnerability to Oil-Supply Disruptions, Elastic Short-Run Supply (E.",1 .5) , Expressed as a Ratio to the Base Case " 0 , 30,]-______________ "0 "0 1.00 trB~'~.~.~C~.~.~.;;~::==:::;~::::::~~:;::: .90 Taut! Gasohn. Till 80"---------------------------1987 1997 2007 20 17 2027 2037 CO:o.l"i for about 20 rC;:lf"S if s uppl y is very eklstic. Yet. the g:lsoline tax increase."i disll.lption costs if suppl y is very unrt.'sponsive to price changes. Table 1 shows the total present value of o ur vulner.lbilily measure over SO rears with the diffen.:nt policies and different supply e b sticities The suh~ id }' d early make"i Ihe U n it(.~ States more vulnel"<lhle to oil-s upply ..hocks Within the nmge of shun -run supply elasticities cited in the economics liter.llurc (0-0.1). tht: ~psol ine tax is costly as wdl. Ollr vulncmbiliry measure indicates thaI the l:lriff is the most effective tool against supply disruptions. Ilowever, if short-run su pply elasticities :Ire higher, the cost of a disnlption may be lessened by the gasoline t:IX. Th<: g:lsoline tax could even dominate the tariff if short-run supply is very \:!lastic. Economic Review-May 1990 Low oil prices ;and rising oil imports ha.ve increased policymakers' sentime nts for protecting the oil and g"dS ind ustry. In this a rticle, we 1..'0111· pared the effl.."'<."!." of Ihree possible measures to reduce o il impon... : :1 25-pcrcent oil-import tariff. a 55-per-barrel suhsidy to domestic o il producers, and an increase in the gasoline tax from 9 cents to 25 cents per gallon. One objective o f these policies is to n.:dm:e U.S. vulnerahility to oil-supply shocks by lowering the coumry·s import dependence through dl..'Creased consumption, increased production, or both. If policymakers want 10 ;lChie\'e import independence by reducing consumption, bOlh the tariff and the gasolint: tax attain Ih:1I goalthrollgh increast:s in oil prices. Thl' subsidy dl'<:reascs impC)l1 shares through inc reased production. bll\ only in the short nm. Based o n our measure of vulnerability , the . . ubsidy to dOl11l'sti<: producers will make the Uniwd States mo re vulnerJble 10 Oil-supply disruptions than lmdcr cu rrent policy The t.a.riff, on the other hand. lo wers the cost of futu re o ilsupply disruptions. By increasing bolh consumer and producer prices_ the tariff lowers consumption while e ncoumging domestic production The present ,·alue or (otal vulnerdhility is consistently lower with the tariff than with the base G IS<:. The gasoline lax would make the United States more vulnerJhle to oil-supply disruptions if short-nm supply is very inelastic and domestic producers could nol respond to the cut in im· ports Howevl'r, if domeslic producers could incro.!ase domestic output by fully utilizing available Gl pacily and by dntwing down priv;Hc ..tockpiies, the gasoli ne tax could low(;!r vulnerahility. Tn the extreme C;lse o f very elastic shari-run supply, the cost of a disruption with (he gasoline tax could be lower than with the tariff. " Appendix Measure of Vulnerability In our microeconomic framework, the change in welfare caused by a supply disruption is the sum 01 losses or gains in consumers' and producers' surplus . At the margin, this change can be expressed as the change in imports multiplied by the infinitesimal change in price, that IS, (1 ) (6) VM is a function of imports, the price of oil, and the elasticities of supply and demand. One can similarly calculate the following modifications of VM when the different policies are in effect. d W : - MdP : qdP - OdP, Ad Valorem Taril /: where q is domestic production, 0 is domestic consumption , M is oil imports, and dP is the price change . (2) M : O(P ) - ",P ) (3) dM= OpdP-qp dP (4) dP : dM Op-qp (7) Mp (1 + r)(" q- £,0 ) + rP"', where T is the tariff rale, PJ is the domestic price of oil, and Pw is the world price of oil. Per- Unit Subsidy : ( 8) VM - MpDp S + SF p Dq S £ , q p O_c0 Qps ' Op and qp are the partial derivatives of 0 and q with respect to pnce . Substituting into dW where 5 is the per-unit subsidy, po is the demand price, and PSis the supply price of oil. and rearranging , we obtain the vu lnerability measure: Gasoli ne Tax : (5) dW M VM = - : . dM qp - Op VM can be expressed in terms of elasticities, Z.. rJ VM = (9) VM _ MP'P' - '/FoP Q - £, qpO_e0 Op s ' where y is the gasoline lax. 1'c;til" r ... 1 H~·st:n·c lbnk "f D:tllas Re fe rences (Ca m~ Birnbaum, Jeffrey I L (1990), MRostenkowski Backs Boos! in Gas Tax, Would Use Defense Cuts to Trim Dcfic i!' ~ \Vall Slreeljollmal, January 3. cd. David A. D<..-ese and Joseph S. Nye - - - (989), MRevenue Hunt: Wilh Tax Rise Talk langley, Mo nica (1987), "Senate Rejects Plan to Keep Oil Imports From Exceeding 50% of Dom(.'Slic Usc,· \Vall 5treetjoumal, July 2. Reviving, Legislalo rs Line Up Candidates,· \Vall Slreeljoumal, January S. Broadman, I larry G . (986), ~The Social Cost of Imported Oil ,~ E"ergy Policy 14(3): 242- 52. - - - . and William W. Hogan (988), ·'Is an Oil Tariff Justified? An American Debate: II. The Numbers Say Yes,MEllergyjollmaI9(3): 7- 29. IJrown, Stephen P.A., :tnd Keith R. Phillips (1989), "Oil Demand :mcl Prices in Ihl! 19905," Federa l Reserve Bank of Dallas Economic Review, January, 1-8. Hogan, William W. ( 981), "Import Management and Oil Emergencies,' in Energy and Secun'ly, Econ o mic Review_ M ay 1990 bridge, Mass.: Ballinger Publishing Co.): 26 1-301. Nordhaus, W . (980), "The Energy Crisis and Macroeco no mic Policy,M Energyjoumal 1(1): 11-19. Stout, Hilary (1989), "Fed Chairman Says Gas Tax Would Be Least Hannful l.evy to Shrink Ddicil, " \Vall Srreetjoll nw/, February 3. Yuce!, Mine K., and Carol Dahl (989), "A Dyna mic Comparison of an O il Tariff, a Prod ucer Subsidy and:t Gasoline Tax ," Federal Reserve Bank of Dallas Research Paper no. 8912 (Dallas, August). "