Memo to Bruce Bartlett: Just Do the Math

Transcription

1 Memo to Bruce Bartlett: Just Do the Math By Davd G. Tuerck Davd G. Tuerck s char of the Department of Economcs and executve drector of the Beacon Hll Insttute at Suffolk Unversty n Boston. Tuerck led a team of economsts at the Beacon Hll Insttute n a seres of studes amed at determnng the economc effects of the FarTax natonal sales tax proposal. In ths artcle, he crtczes a recent Tax Notes artcle by Bruce Bartlett, enttled Why the FarTax Won t Work (Tax Notes, Dec. 24, 2007, p. 1241, Doc , 2007 TNT ). Tuerck dentfes sx mstakes made by Bartlett n attemptng to dentfy the dstrbutonal consequences of the FarTax. These mstakes, accordng to Tuerck, arse over Bartlett s understandng of the prce effects of the FarTax, hs nterpretaton of the purpose of the FarTax rebate, and hs assessment of the burden of the FarTax on government. The author wshes to thank Alfonso Sanchez- Penalver for hs nvaluable assstance. Copyrght 2008 Davd G. Tuerck. All rghts reserved. Table of Contents The Effectve Tax Rate Three Addtonal Mstakes Who s Beng Dshonest? Mstake Number Concluson Appendx In the November 13, 2006, ssue of Tax Notes, my coauthors and I publshed an artcle (the BHI/Kotlkoff study) n whch we hoped to resolve ssues related to the FarTax proposal, under whch the federal government would replace almost all exstng taxes wth a sales tax. 1 1 Paul Bachman, Jonathan Haughton, Laurence J. Kotlkoff, Alfonso Sanchez-Penalver, and Davd G. Tuerck, Taxng Sales Under the FarTax: What Rate Works? Tax Notes, Nov. 13, 2006, (Footnote contnued n next column.) Our artcle was amed largely at correctng and updatng fndngs reached by Wllam G. Gale n hs earler crtque of the FarTax. 2 We hoped that our artcle would elevate the dscusson of the FarTax, whch s ganng ncreased attenton n natonal poltcs. We also hoped that, by presentng the underlyng mathematcs n panstakng detal, we could dspel the confuson that had long lngered over the calculaton of the FarTax rate. Now, n a recent artcle, Bruce Bartlett launches an attack on the FarTax n whch he lapses nto the same confuson that we (and Gale before us) had attempted to dspel. 3 Bartlett sngles out our study for crtcsm, accusng us of duplcty n our efforts to work through the mathematcs nvolved. 4 In the process, he brngs hs readers back to square one n gettng thngs rght. Although some of Bartlett s arguments are substantve, much of hs artcle s drected at how people (proponents, voters, poltcans, analysts) perceve the FarTax. Perceptons are mportant, but they should be based on fact, rather than error. The problem s that Bartlett s artcle s strewn wth errors errors that he could have avoded smply by comprehendng what we and Gale had already put before hm. Although Bartlett covers many ssues and n the process makes many mstakes, I wll lmt my comments here to hs attempt to work through what he calls the dstrbutonal consequences of the FarTax. Ths s no easy task nsofar as Bartlett barely sketches the mathematcs that underles hs reasonng on ths ssue or any ssue. It means revstng many ssues already addressed n our artcle. It also means engagng n some mathematcs and, n effect, correctng Bartlett s mathematcs (or what hs mathematcs would show f shared wth us). It s mportant, nevertheless, to undertake ths task because, left unchallenged, Bartlett s artcle wll only sow further confuson. 5 Of Bartlett s mstakes, I wll focus on sx: p. 663, Doc , 2006 TNT Ths and other Beacon Hll Insttute studes may be found on the Web ste of Amercans for Far Taxaton (see and on the Insttute s home page, 2 Wllam G. Gale, The Natonal Sales Tax: What Would the Rate Have to Be? Tax Notes, May 16, 2005, p Bruce Bartlett, Why the FarTax Won t Work, Tax Notes, Dec. 24, 2007, p. 1241, Doc , 2007 TNT Id. at Also, Bartlett s vews are gettng ncreased attenton from the meda. See, e.g., Jonathan Wesman, Crtcsm Asde, Far- Tax Boosts Huckabee Campagn, The Washngton Post (Dec. 28, 2007), p. A04, avalable at wp-dyn/content/artcle/2007/12/27/ar html. See also Bran Mooney, In Spotlght, Far Taxers Push Cause, (Footnote contnued on next page.) TAX NOTES, February 4,

2 msstatng the calculaton of the effectve tax rate under the FarTax; concludng that the removal of exstng taxes n antcpaton of the FarTax would reduce prces; observng that mposton of the FarTax would rase prces by the nclusve rate, rather than the exclusve rate; mplyng that the FarTax rebate s ntended to compensate consumers for rsng prces; allegng duplcty n how BHI/Kotlkoff handles the rebate n calculatng the FarTax rate; and erroneously clamng that the FarTax would mpose a burden on the federal government and on state and local government. In proceedng, I wll assume that the reader knows the fundamentals, about whch more may be learned by readng the BHI/Kotlkoff study or consultng the Web ste of Amercans for Far Taxaton. The FarTax s ntended to be revenue neutral and mldly progressve. It would mantan exstng federal government programs n real terms whle also provdng a rebate to every household approxmately equal to what that household would pay n taxes f t were at the poverty level. It would be mposed on both personal and government consumpton expendtures. Its goals are to remove the bas of the exstng tax system aganst savng, expand the economy, and smplfy tax admnstraton and complance. One matter that comes up repeatedly n connecton wth the FarTax s the dstncton between an nclusve and an exclusve tax rate. The archtects of the FarTax have calculated that the tax rate would be 23 percent on an nclusve bass (whch, we fnd, s close to the correct rate) and 30 percent on an exclusve bass. It s mportant to understand the dstncton between those concepts. People usually thnk of sales taxes on a tax-exclusve bass. Consder an tem for whch the prce, exclusve of the sales tax, s $1 and the prce nclusve of the sales tax s $1.30, wth $0.30 gong to government as tax revenue. The tax-exclusve rate s 30 percent, computed as $0.30/ $1. But ths mples a tax-nclusve rate of 23 percent (= $0.30/$1.30). Generally, the relatonshp between the tax-exclusve and the tax-nclusve rate s gven by ths equaton: (1) t e = t 1 t where t e s the tax-exclusve rate and t s the tax-nclusve rate. Ths algebra s mportant because t remnds us that (gnorng other taxes such as excse taxes), the prce of a good, nclusve of the tax, always exceeds the prce, exclusve of the tax, by t e (30 percent under the FarTax). Lkewse, the prce exclusve of the tax s always t (23 percent) less than the prce, nclusve of the tax. Although ths algebra s well understood by most students of the ssue, Bartlett manages to get t wrong at several ponts n hs artcle. So let s consder what Bartlett has to say. The Effectve Tax Rate One of Bartlett s arguments has to do wth the effectve tax rate that ncome earners would pay under the FarTax. It s n makng ths argument that he makes hs frst mstake. Effectve tax rates are calculated to show the sacrfce mposed by a tax, gven that there s often a dfference between the statutory tax rate and the tax rate that measures ths sacrfce. Consder the followng: Joe currently has a gross ncome of $50,000. He gets a deducton aganst hs gross ncome of $10,000 so that hs taxable ncome s $40,000. If the statutory ncome tax rate s 25 percent, hs tax bll s $10,000, leavng hm wth $40,000 n after-tax ncome. But the effectve rate s not the same as the statutory rate. We compute hs effectve tax rate by expressng hs tax lablty as a fracton of hs gross ncome, whch yelds a rate of 20 percent. Alternatvely, we could set up the calculaton as follows: (2) Effectve Rate = Gross Tax Deducton From Gross Tax Gross Income Computng hs effectve rate usng ths equaton we get $12,500 $2,500 = 20%, as before. $50,000 The numerator n Equaton (2) s the dfference between the amount that the taxpayer would pay on hs earnngs f there were no deducton and the amount that he can deduct from hs tax lablty, gven that there s a deducton. We can thnk of the numerator as hs net tax payment. To look at the calculaton n a dfferent way, consder Joe s standard of lvng. Assume 50,000 wdgets are produced. If Joe buys only wdgets and f the prce of a wdget s $1, then, under the ncome tax, he gets to consume 80 percent, or 40,000 of these wdgets, leavng 20 percent, or 10,000, for the government. The goal of computng the effectve rate should be to determne how the tax, whatever t s, mpnges on hs standard of lvng. Here t mpnges on hs standard of lvng by allocatng to government 20 percent of the wdgets produced. Suppose then that the ncome tax s abolshed and that a sales tax s put n ts place. Assume the statutory sales tax rate s 25 percent (expressed on an exclusve bass). Also assume that there s no rebate, that the taxpayer s gross ncome remans unchanged, and that the prce of goods rses by 25 percent. 6 He spends $50,000 on consumpton, of whch 20 percent, or $10,000, s taxes. If we The Boston Globe, Jan. 1, 2008, avalable at boston.com/busness/taxes/artcles/2008/01/01/n_spotlght _far_taxers_push_cause/. 6 Note that ths s the tax-exclusve rate (0.25 = (0.20/1-0.20)). Below, we refer to the case n whch prces rse by the tax-exclusve rate as one of full monetary accommodaton. 640 TAX NOTES, February 4, 2008

3 apply these facts to Equaton (2), we get exactly the same result as before, that s, 20 percent. In ths nstance, there s no Deducton From Gross Tax, but the formula gves the correct answer, nsofar as Joe stll has to sacrfce 20 percent of all the wdgets produced. A wdget prevously cost $1 but now costs $1.25. Joe s $50,000 buys hm exactly 40,000 (= $50,000/1.25) wdgets, leavng the remanng 10,000, or 20 percent of the total, for the government to consume. 7 Now, to make the example more realstc, let the government offer a tax rebate. Ths wll requre a hgher statutory rate. But suppose that the rebate s just hgh enough to make t possble for Joe to mantan exactly the same standard of lvng that he enjoyed under the ncome tax and prevously under the FarTax wthout a rebate. Then, surely, we have to say that hs effectve rate would stll be 20 percent. To see that t would, assume that the government decdes to set the tax-exclusve rate at 30 percent and to provde a rebate of $2,000. Ths mples a tax-nclusve rate of percent, just as under the FarTax. 8 Joe now has $52,000 (the prevous gross ncome plus the rebate) to spend and pays a gross tax of $12,000 (= $52,000). Joe can stll buy 40,000 (= $52,000/1.30) wdgets, and the government stll gets 10,000 wdgets. 9 Thus we expect the effectve rate to reman unchanged. And, ndeed, t does. Subtractng the rebate from hs gross ncome, hs net tax bll s $10,000, whch, dvded by gross ncome, yelds an effectve rate, agan, of 20 percent. The effectve rate remans the same even though the statutory rate s hgher. Amercans for Far Taxaton, the object of much crtcsm from Bartlett, uses a smlar method, he ponts out, to compute effectve tax rates for hypothetcal taxpayers. 10 But Bartlett would change ths method. He would dvde the gross tax by the sum of earnngs and the rebate as follows: (3) Effectve Rate = Gross Tax Gross Income + Rebate Under the FarTax plus rebate, Equaton (3) would yeld an effectve rate of percent (= $12,000/ ($50,000 + $2,000)), whch, n ths nstance, s the same as 7 Note that, n ths example, as under the FarTax, the government pays the tax-nclusve prce, $1.25, for wdgets. The government gets $12,500 (= 50,000 $0.25) n tax revenue, whch t uses to buy ts 10,000 (= $12,500/1.25) wdgets. 8 We show how we get ths rate n the dscusson that follows. The example mples that the government has determned the famly consumpton allowance, dscussed below, to be $8,666 (= $2,000/0.2308), whch s bgger than the comparable number under the FarTax but used here for llustratve purposes. 9 The government collects $15,000 (= 50,000 $0.30) n tax revenue ($12,000 from Joe s consumpton and $3,000 from ts own consumpton) of whch t uses $2,000 for the rebate, leavng t wth $13,000 to buy wdgets at $1.30 apece. 10 Bartlett, supra note 3, at the statutory rate. Ths, he argues, s closer to what taxpayers thnk of n terms of the effectve rate. 11 Hs reasonng s doubly confused. Frst, he suggests that proponents of the FarTax use Equaton (2) rather than Equaton (3) because Equaton (2) yelds a lower rate. But suppose we appled Equaton (3) consstently to the ncome tax and the FarTax. Then t s the ncome tax that would yeld the hgher rate, at least for these examples. Usng Equaton (3), the rate under the ncome tax would be percent (= $12,500/ ($50,000 + $2,500)), whch s greater than the rate (20 percent) under the FarTax wthout the rebate and greater than the rate (23.08 percent) under the FarTax wth the rebate. Second, Bartlett s method makes t mpossble to compare the burden mposed by the FarTax wth the burden mposed by current law on an apples-to-apples bass. Bartlett apparently beleves that people thnk of the effectve rate n a way that makes ths comparson mpossble. But f they do, we should encourage them to thnk n terms of Equaton (2) rather than Bartlett s specfcaton. It seems counterproductve to provde numercal llustratons that could encourage readers to thnk only n the nonsenscal way Bartlett suggests. By tryng to renvent the calculaton of the effectve rate, Bartlett steers the reader nto a mstake of hs own makng hs frst mstake. 12 Three Addtonal Mstakes Bartlett crtczes two promnent FarTax supporters, Neal Boortz and John Lnder, for makng a muddle of ther argument concernng the effect of the FarTax on prces. 13 Bartlett crtczes Boortz and Lnder as follows: One problem wth analyzng the dstrbutonal consequences of the FarTax s that ts supporters sometmes argue that after-tax ncomes wll rse by enough to pay the hgher prces for goods and servces once the 23 percent s added to the prces people pay today. At other tmes, they argue that prces wll fall once ncome taxes currently embedded n prces are removed, mplyng a free lunch n whch everyone s better off and no one s worse off. Actually, t s a double free lunch because not only do you get to keep all the taxes currently wthheld and pay no more for goods and servces now, but you get the rebate as well. 14 Whle Bartlett s rght about Boortz and Lnder, he goes on to botch hs own analyss of the effect of the FarTax on prces. He wrtes: 11 Id. at In fact, Bartlett (supra note 3, at 1244) shows that he s trply confused when he suggests that the effectve rate, as defned n Equaton (3), assumes that taxpayers save the rebate, whereas the statutory rate assumes that they spend the rebate. In fact, nether calculaton has anythng to do wth what people spend or save but wth the sze of the rebate relatve to the other terms n hs (ncorrect) equaton. 13 Neal Boortz and John Lnder, The FarTax Book (New York: HarperCollns/Regan Books, 2005). 14 Bartlett, supra note 3, at TAX NOTES, February 4,

4 If the FarTax mposes a 23 percent tax on goods and servces, t looks as f t s largely a wash. Ther prces wll fall by 22 percent once all taxes are abolshed and the FarTax wll add about the same, so the fnal consumer cost wll be no hgher than t s now, at least on average. If ths s true, however, t s hard to understand why there needs to be a tax rebate to compensate for the burden of the tax, snce t appears as f there s no burden. 15 Then, n speculatng further on what mght happen to prces, he consders the possblty that prces mght rse, rather than fall: However, f prces rse by 23 percent to allow workers to avod cuttng ther wages, they aren t really better off. They have more dsposable ncome because of the abolton of wthholdng, but everythng costs more because of the nflaton necesstated by mposton of the FarTax. 16 Bartlett makes three mstakes here. Frst, he buys nto the very mstake that Boortz and Lnder made by sayng that prces would fall when current taxes are removed. In fact, the removal of current taxes would, n and of tself, leave prces unchanged. Second, he mples that, gven constant wages, prces would rse by the nclusve rate t when n fact they would rse by the exclusve rate t e. Thrd, he msconstrues the role of the rebate. In fact, the rebate has nothng to do wth what happens to prces. To see how Bartlett errs, we have to do some math. Assume that there are just two knds of consumpton, personal consumpton C and government consumpton G. Assume also that under current law, G s fnanced by revenue from a tax t nc mposed on ncome and that all after-tax ncome s consumed. There are no transfer payments. We then ntroduce two equatons: (4) G=t Y=t ( C+G) = Total Revenue nc where t nc s the rate at whch ncome s taxed and Y s producton n real dollars, and (5) MV=PY where M s the money supply, P s an ndex of market prces, and V s the velocty of money. For our purposes, t s convenent to rewrte (5) as (6) P = MV Y nc Consder a smple example. Let producton consst only of wdgets and let Y equal $1,000, M equal $1,000, and V equal 1. Thus P equals 1. In ths world, the prce pad by consumers (the market prce) s just equal to the prce receved by producers (the producer prce). We can set the market prce and the producer prce of a wdget at $1. The ncome tax t nc s 20 percent, so that government rases $200 n revenue, whch t uses to buy 200 wdgets. Consumers (who are also workers) are left wth $800 n after-tax ncome, whch they use to buy the remanng 800 wdgets. Now let s address Bartlett s second mstake. Consder what would happen f the government smply removed exstng taxes and, perforce, qut spendng. The straghtforward answer: Wth t nc equal to zero, and assumng no change n M, V, ory, consumer demand would rse to absorb all 1,000 wdgets, stll prced at $1 apece. 17 There would be no change n prces. To understand Bartlett s thrd mstake, suppose the government substtuted the FarTax for the ncome tax. Ths would requre t to compute a FarTax rate that, when appled to personal and government consumpton, would yeld just enough revenue to keep government purchases constant n real dollars. As above, desgnate ths rate as t when expressed on tax-nclusve bass and as t e, when expressed on a tax-exclusve bass. So what s t? Well, n ths example, 20 percent. In real terms, government consumpton G stll equals 20 percent of total consumpton C+Gand of total producton Y.Of the total number of wdgets produced, government stll gets 20 percent and consumers the remanng 80 percent. The FarTax rate has nothng to do wth what happens to prces, and what happens to prces has nothng to do wth the FarTax rate. To see why, consder the dfference between the market prce and the consumer prce of a good. The market prce s the prce the consumer pays, nclusve of the FarTax. The producer prce s the prce receved after he pays the FarTax. The relatonshp between the two prces s as follows: (7) PP = P 1 +t e where PP equals the producer prce. Before the FarTax, PP = P, because t e s zero. If, under the FarTax (and as t turns out n ths example), t s 20 percent, then t t e = = 25 percent and P must exceed PP by 25 1 t percent. But that s all. P may rse, fall, or reman the same after the FarTax s mposed, and whatever t does depends strctly on how the monetary authortes decde to adjust M, assumng that they decde to adjust t at all. Although the monetary authortes could brng about a fall n P by contractng M, there s no reason to beleve they would do so just because the FarTax has been nsttuted. As to what would happen, we can narrow down the possble outcomes. At one extreme, the monetary authortes could hold M and therefore P fxed, so that the 15 Id. at Id. at Of course, M, V, and Y mght well change. But t s approprate to assume that all three varables are constant wthn the statc framework assumed here. 642 TAX NOTES, February 4, 2008

5 market prce of a wdget remaned fxed at $1 and the producer prce fell to $0.80. At another extreme, the monetary authortes could ncrease M by 25 percent, the tax-exclusve rate, so that the producer prce of a wdget remaned fxed at $1 whle the consumer prce rose to $1.25. In ether event, the government would stll get ts 200 wdgets and consumers ther 800 wdgets. Now let s consder some addtonal algebra. Frst, assume that the monetary authortes chose to accommodate the FarTax by ncreasng the money supply by α, where 0 α t e, once the FarTax s enacted. When α = 0, we have the case of nonaccommodaton, and when α = t e, we have the case of full accommodaton. 18 Now assume, as Bartlett does, that the monetary authortes act n such a way as to keep gross wages, whch s to say producer prces, constant. To that end, they would have to set α equal to t e. Why? Because, as Equaton (7) shows, P would have to rse by t e so that PP would reman constant. Thus, prces rse by t e, not t. Ths corrects hs thrd mstake. Bartlett s fourth mstake s to confuse the purpose of the rebate wth what happens to prces. So let s assume that the federal government has determned there to be some amount B, whch represents the base for the rebate. B s the aggregate famly consumpton allowance roughly equal to the famly s poverty level. Operatonally, B s the aggregate of ndvdual household ncomes, set at whatever level government chooses to compensate households for ther taxes. For example, n the BHI/Kotlkoff study, we determned that the famly consumpton allowance for a household of four n whch there are two marred partners s $26,981. Under the FarTax, every household fttng ths descrpton would receve a rebate of $6,206, whatever ts ncome. There s a dfferent famly consumpton allowance for each type of household, dependng on whether the household s headed by marred partners or not and dependng on household sze. We got B by aggregatng the famly consumpton allowance across all of the varous household types. In what follows, we contnue to gnore varous other complextes, consdered n detal n our study. So now, we are ready to wrte down an equaton that permts us to solve for t : 18 The monetary authortes would have to consder how the degree of accommodaton, varyng from none to full, would affect the overall economy and how t would affect the wellbeng of varous groups such as retrees. As Bartlett ponts out, nonaccommodaton mght be mpractcal nsofar as workers would resst a reducton n ther gross wages even though ther real wages would reman unchanged. At the same tme, full accommodaton would rase prces and thus penalze retrees who had been payng ncome taxes on ther savngs all along. In the end, whatever the degree of accommodaton, there would be some transtonal effects to consder and perhaps amelorate. All that sad, the degree of accommodaton and any prce change that results has nothng to do wth ratonale for the rebate, the calculaton of the FarTax rate, or the burden of the FarTax on the purchasng power of consumers or government, fgured n the aggregate. (8) t( C+G )(1 + ) = ( G+tB )(1 + ), whch yelds (9) t = G C+G B As before, f there s no rebate and B = 0, then $200 t = = 20 percent. $800 + $200 If government sets the aggregate consumpton allowance B equal to, say, $200, then the rate would be hgher: $200 (10) t = = 25%. $800 + $200 $200 Why the rebate? The answer s that some households have less ncome than others. Suppose that there are 10 households sharng the 800 wdgets. Rch households mght get many (some would say a dsproportonate share) of the wdgets. Poor households mght get few or none. But f every household got a rebate of.25 $200 = $5.00 (n real dollars), t could have at least 10 5 wdgets. Because there are only 800 wdgets to go around and because every household now pays a tax of 25 percent, rather than 20 percent, rcher households would end up wth fewer wdgets than they would have had wthout the rebate, but poorer households would end up wth more. Put dfferently, the rebate reduces the effectve tax rate for low-ncome households and ncreases t for hghncome households. The dstrbutonal consequence of the rebate s thus to redstrbute ncome, specfcally, to reduce the effectve tax rate (correctly calculated) for the poor. It has nothng to do wth compensatng taxpayers for rsng prces or fallng ncomes. Thus, we have Bartlett s fourth mstake. Who s Beng Dshonest? The foregong example leads to Bartlett s ffth mstake. Referrng to the BHI/Kotlkoff study, Bartlett says: The cost of the tax rebate that would have to be pad s also cleverly dealt wth n the new study so as to mnmze ts budgetary mpact...[t]he study does not show government spendng rsng by the amount of the rebate....rather the cost of the rebate s dealt wth by reducng the tax base...the only purpose of dong t that way s to mantan the fcton that the rebate s a reducton n taxes rather than an ncrease n federal spendng. It would be more honest to do ths accountng by addng the rebate cost to the spendng sde of the budget. 19 Let s see who s beng dshonest here. Frst, as equaton (8) shows, we are, n fact, puttng the rebate (t B)n the budget; we add t to G to reflect the new spendng that the rebate necesstates. 19 Bartlett, supra note 3, at TAX NOTES, February 4,

6 Perhaps Bartlett would rewrte equaton (9) as t = G + Rebate. But that s the same as wrtng C+G t = G + tb, whch follows drectly from Equaton (8). C+G Apparently, Bartlett doesn t understand that t s necessary to take the further step of wrtng down Equaton (9), as BHI/Kotlkoff dd (n far greater detal), to solve for t. The BHI/Kotlkoff specfcaton has nothng to do wth anyone s zeal for the FarTax. Rather t has to do wth the problem of calculatng t when the rebate, whch nfluences the sze of t, also depends on the sze of t. Our method, whch s based on one used earler by Gale, solves ths problem. So Bartlett s ffth mstake s lettng hs own zeal for debunkng the FarTax nterfere wth hs attenton to the algebra. Mstake Number 6 Fnally, we have ths passage from Bartlett: The new study also mantans the fcton that the federal government would pay taxes to tself on ts own purchases and that the tax on those purchases would not ncrease spendng by the amount of the tax...thus they have effectvely assumed away one of the major problems of the FarTax that t wll rase prces by the amount of the tax. 20 Agan, as Equaton (8) shows, there s no presumpton that prces wll rse. That depends on the response of the monetary authortes. Or to put t algebracally, the expresson (1+α) appears on both sdes of the equaton and therefore cancels out. Furthermore, the assumpton that the government pays taxes on ts own purchases s not a fcton but a reflecton of the status quo. To see why, consder what happens, under current law, when the government buys a wdget. The government pays $1 to the producer, who then pays $0.20 n taxes, whch the government collects and uses to buy wdgets. In other words, for every dollar the government spends on wdgets t gets $0.20 back n taxes. The total tax take s $200, equal to the sum of $160 (= 800 $0.20) n taxes collected on producton for personal consumpton and $40 (= 200 $0.20) collected on producton for government consumpton. Now assume that the FarTax s put n place and that (to keep thngs smple) there s no rebate (B = 0). At one extreme, the market prce of the wdget would reman fxed at $1 and the producer prce would fall to $0.80. The government would pay $1 per wdget, of whch $0.20 would be taxes, just as before. At the opposte extreme, the monetary authortes would permt the market prce of wdgets to rse to $1.25, n whch case the government would pay $0.25 n taxes on each wdget that t buys. Nothng has changed n real terms. In ether case, the taxes that government pays tself are worth exactly one-ffth the cost of a wdget ($0.20/$1.00) or ($0.25/ $1.25). Whether market prces rse or reman the same has nothng to do wth the effect of the FarTax on the real value of government spendng. 20 Id. at Bartlett also fals to understand that t would make no dfference f government ddn t tax tself. Suppose the archtects of the FarTax had decded not to tax government, that s, not to nclude government spendng n the FarTax base. In effect ths would mean reprcng government to remove the taxes already pad on producton for government. Agan, nothng real would change. Ignorng the rebate, we rewrte equaton (8) as: (11) tc (1 + ) = G(1 t)(1 + ) whch agan gves us: G (12) t = = 20%. C+G 21 Thus, f government wanted to end the practce of payng tself taxes on current purchases, t could do so. Government would no longer have to pay the market prce for goods because t could now buy wdgets at the producer prce, whch would be 20 percent below market prce. Wth nonaccommodaton (α = 0) market prce would reman unchanged. But government would pay the producer prce ($0.80) and would need 200 $0.80 = $160 n tax revenue to contnue buyng 200 wdgets. Indvdual consumers would buy ther 800 wdgets at $1 apece, payng the needed $160 (= 800 $0.20) n tax revenue. Wth full accommodaton (α = t e =.25), market prce would rse by 25 percent and consumers would buy ther 800 wdgets at $1.25 a pece, payng a total of $200 (= 800 $0.25) n tax revenue. Government would use the $200 to buy 200 wdgets, as before, payng $1 (the producer prce) apece. Ths example shows (as Gale showed) that the FarTax rate has nothng to do wth whether government pays taxes to tself or not and (as BHI/Kotlkoff and Gale showed) whether market prces rse or not. Bartlett contnues to botch hs algebra n the followng passage: It cannot possbly make any sense for the Department of Defense to pay 23 percent more for a weapons system because the FarTax has been added. The revenue just goes to Treasury, another branch of government. Spendng and revenues are hgher to exactly the same extent; t s a wash, whch s why foregn countres don t do t. 22 Frst, a tax-nclusve FarTax rate of 23 percent, combned wth full monetary accommodaton by the Fed, would rase the prce of weapon systems by 30 percent, not 23 percent. Second, as shown above, t s an entrely arbtrary matter whether the Defense Department or any department of government ends up, n effect, payng taxes to tself. If, as Bartlett concedes, t s a wash, what, precsely, s the objecton? 21 Equaton (11) accounts for the fact that, n ths specfcaton, t s necessary to reduce the cost of government by t, gven that the government doesn t charge tself a tax. 22 Bartlett, supra note 3, at TAX NOTES, February 4, 2008

7 Bartlett s not content to msrepresent the mplcatons of the FarTax for federal government spendng but manages to do so also for state and local government spendng. Thus he says: The problem for state and local governments s worse. The FarTax smply rases ther spendng wthout smultaneously rasng ther revenue. Realstcally, ther only choce s to ncrease ther taxes to pay the FarTax on ther spendng.... Thus, to the extent the FarTax forces state and local governments to rase ther tax rates, t becomes a backdoor means of fnancng t at a deceptvely low rate. 23 Ths argument gnores the fact that state and local government, lke the federal government, s already payng prces that nclude federal taxes that would n turn be replaced by the FarTax. Second, t gnores the fact that state and local government would be able to mantan current servces at no greater burden to state and local taxpayers by some smple adjustments n ther tax codes. I explan the need for those adjustments n the appendx and show how by makng them, state and local government would mantan ts purchasng power, along wth that of ndvdual consumers. So t turns out that Bartlett s sxth mstake s really a combnaton of mstakes: falsely argung that the FarTax would necessarly cause government spendng to rse; falsely clamng that government does not currently tax tself; gnorng the fact that whether t taxes tself or not s rrelevant; falng, agan, to understand the rrelevance of prce changes for measurng real effects; confusng the prce change under full accommodaton (t e ) wth the nclusve tax rate (t ); and, most egregously, falsely clamng that the FarTax represents a burden on government. Concluson Ths artcle has dentfed sx mstakes (and more) made by Bartlett n hs crtque of the FarTax and, collaterally, hs attack on BHI/Kotlkoff. Bartlett accuses BHI/Kotlkoff of presentng deceptve mathematcs to make the FarTax more palatable to poltcans. That he takes ths tack s ronc, consderng hs own nclnaton to put poltcal realty ahead of prncple. Whch, ndeed, he does. Bartlett engages n the dsngenuous game of argung aganst an dea by clamng t to be poltcally nfeasble. When he says that Congress would never mplement the rebate as planned and that t would be subject to poltcal manpulaton, he s, n effect, sayng that scholars should avod confrontng poltcans wth deas that would lmt ther ablty to make bad polcy. In that fashon, Bartlett s lke the doctor who won t tell her overweght patent that he mght de of heart dsease f he doesn t det ths on the assumpton that the patent wouldn t stck to a det anyway. 23 Id. And t s worse than that. Bartlett, n gvng gratutous advce on practcal poltcs, gets the practcal poltcs backwards. The FarTax does not attempt just to encourage savng and smplfy tax admnstraton and complance. The FarTax would make t harder not easer for poltcans to manpulate the tax code to the end of satsfyng specal pleaders. To extend the earler metaphor, t would put elected offcals on a det, nsofar as every addtonal dollar put toward government spendng and toward the rebate would mean a hgher tax rate vsble at every retal transacton and a poltcal lablty to anyone who suggested t. The fact that the FarTax would restran, not expand government, s the btter pll that polcymakers would have to swallow. Bartlett tres to put down the FarTax as an nventon of the Church of Scentology. But t was not Tom Cruse, but Thomas Hobbes, who wrote, some 350 years ago, that when the mpostons are lad upon those thngs whch men consume, every man payeth equally for what he useth. I submt that Hobbes s nsght offers the better foundaton for applyng economc scence to our understandng of the FarTax. Appendx Implementaton of the FarTax would cause state and local government to lose purchasng power unless t made some adjustments n ts tax law. At the same tme, those adjustments would permt state and local government to mantan the real value of ts purchasng power wthout reducng that of consumers or of the federal government tself. Consder what happens when we add state government spendng to the tax base. Let (A1) t( C+G+SG)(1 + ) = ( G+tB )(1 + ) where G s federal government spendng and SG s state government spendng. Solvng for t, (A2) t = G C+G+SG B Assume that, under the current system, the ncome receved by consumers s $1,250, that G = $200 (that s, the federal ncome tax rate s 16 percent), and that state governments mpose a sales tax at 5 percent, measured on a tax-exclusve bass. Ths mples a tax-nclusve state sales tax rate of 4.76 percent. Further assume, as under current law, that nether the federal nor the state government pays the state sales tax on ts purchases. Then after-tax ncome s $1,050. Assumng that all of ths ncome s consumed, the state collects $50 n sales tax revenue. Assume that there s a sngle good, wdgets, and that the producer prce s $1. Then the market prce s $1.05. But, because they do not pay the state sales tax, both federal and state governments pay $1. Ths means that ndvduals would be able to consume 1,000 wdgets, the federal government would be able to consume 200 wdgets, and the state government would be able to consume 50 wdgets. TAX NOTES, February 4,

8 Now apply Equaton (A2) to compute the FarTax nclusve rate, assumng that there s no rebate: $200 (A3) t = = 0.16 $1,000 + $200 + $50 Because we assume that all ncome s spent, the FarTax nclusve rate s equal to the ncome tax rate. Assume nonaccommodaton. 24 The producer prce decreases to $0.84 (= $ $0.16), and ncome falls from $1,250 to $1,050 (= $1, ). However, there are now no ncome taxes and the ndvdual can consume the entre $1,050. Because the state sales tax s mposed on the producer prce, state sales tax revenue falls unless the state adjusts ts tax system. Consder what happens f the state makes no change n ts tax system. Because the sales tax falls on the producer prce, the state collects only $0.042 (= 0.05 $0.84) for every unt sold to the consumer. Ths also means that the prce the consumer faces s no longer $1.05, but $1.042 per unt. Therefore, the consumer can purchase 1, (= $1,050/$1.042) wdgets. However, the consumer s gan s the state s loss. The state collects $42.32 (= 1, $0.042). Because the prce pad by the state for a wdget, $1, ncludes the FarTax, the state can buy wdgets, not 50 wdgets, as prevously. The consumer can buy 7.68 more wdgets, and the state can buy 7.68 fewer wdgets. However, f the state adjusts ts legslaton to gan back the purchasng power lost to the consumer, both the state and the consumer would be able to buy the same number of wdgets as under the current system. The smplest correctve s to apply the state sales tax on the FarTaxnclusve prce, that s, $1. The state would once agan collect $0.05 (= $ percent) on each unt sold, and the prce faced by the consumer would once agan be $1.05. The consumer would consequently be able to 24 As above, the level of monetary accommodaton s rrelevant to our results. Assumng nonaccommodaton makes the llustraton smpler. purchase 1,000 wdgets (= $1,050/$1.05), just as under current law. The state would collect $50 (= 1,000 $0.05), whch would allow t to purchase 50 wdgets, as before. Now suppose the state mposes an ncome tax nstead of a sales tax. Suppose also that the taxpayer stll earns $1,250 n ncome, that the federal government collects $200 n ncome tax revenue (mposng a rate of 16 percent), and that the state collects ncome tax revenue of $50 (at a rate of 4 percent). Snce all ncome s spent, the consumer buys $1,000 worth of goods, the federal government buys $200 worth of goods, and the state government buys $50 worth of goods. In ths case the market prce faced by the consumer s the same as that faced by the state and federal governments, $1, so the ndvdual consumes 1,000 wdgets, the federal government 200, and the state government 50. Now we mpose the FarTax, and because total consumpton and federal tax revenue s the same as n the prevous example, the FarTax tax-nclusve rate s, agan, 16 percent. Assumng nonaccommodaton, the producer prce falls to $0.84, and ncome to $1,050. Once more we consder two scenaros: one n whch the state government does not adjust the ncome tax rate and one n whch t does. If the state government does not adjust ts ncome tax rate, t wll collect $42 n tax revenue (= $1, ). The consumer s dsposable ncome wll therefore be $1,008 (= $1,050 - $42). Market prce remans constant at $1, snce t ncludes the FarTax. Therefore, the consumer buys 1,008 wdgets and the state buys 42 wdgets. Once more, the state loses purchasng power matched by a gan n purchasng power enjoyed by the consumer. The state has to ncrease ts revenue collectons by the FarTax tax-exclusve rate, percent (= 0.16/ (1-0.16)). Ths s not a concdence; t s because ncome has decreased by the FarTax (tax-nclusve) rate. So suppose then that the state ncreases the ncome rate to 4.76 percent (= 4 percent ( percent)). The ndvdual would pay $50 as state ncome tax (= $1, percent), whch would make hs dsposable ncome $1,000 (= $1,050 - $50). As before, the ndvdual would buy 1,000 wdgets and the state 50, thus keepng the purchasng power of both constant. 646 TAX NOTES, February 4, 2008