Tourism Crisis Management: adjusting to a temporary downturn

Transcription

1 Toursm Crss Management: adustng to a temporary downturn ABSTRACT Adam Blake and M. Thea Snclar Unversty of Nottngham, UK There have been many nstances where the toursm ndustry n one or more countres has suffered an unexpected and sudden downturn n demand. Polcy makers are faced wth the predcament of f and how they should respond to such crses n the absence of research nto the relatve effectveness of dfferent responses. The downturn n toursm demand n the Unted States followng September 11 s a partcularly vvd example of toursm crss. If such a downturn s of a temporary nature, government polcy tends to be amed towards reducng the level of adustment that takes place; f the downturn s udged to be permanent, government polces tend to be amed towards makng adustment to a new equlbrum less costly. Ths paper examnes the means of measurng the costs of adustment wthn a CGE model and assesses how government polcy can be most effcently targeted towards reducng those costs. We construct a computable general equlbrum model of the US to analyze the effects of the September 11-nduced toursm crss and potental and actual polcy responses to the crss. A CGE model s a useful tool n evaluatng toursm crses because t allows toursts to purchase a bundle of commodtes, accounts for general equlbrum effects and allows the modelng of government responses to crss. The model s a statc 98-sector model of the US economy, wth the specfcaton of 23 types of labor to enable the modelng of employment responses to September 11. Toursm demands are specfed as CES functons over a varety of goods and servces, wth dfferent demands from foregn, domestc, and ntermedate busness toursts, each of whch undertake non-ar travel and ar travel toursm trps. Shocks are appled to ar travel toursm to smulate the effects that September 11 has had on the demand for travel, and the resultng mpact on welfare, GDP, employment s calculated. The levels of factor adustment that take place n the economy are measured. Further to ths, the effcency of federal government measures to counteract the fall n demand for ar transport are modeled. The effcency of varous other possble federal government actons s also modeled. The model s bult around a set of data taken from US natonal accounts sources. The nputoutput table s derved from the 1997 benchmark nput-output table, wth addtonal data on GDP by ndustry and natonal ncome and product account data used to update ths table to construct a SAM that represents the US n 2001 wthout the effects of September 11. Other US natonal accounts sources are used to construct a matrx of tourst demands, by product and type of tourst. We fnd that sector-specfc targeted subsdes and tax reductons are the most effcent means of toursm crss management. The ntal federal government responses to the crss were effcent n counteractng the fall n toursm demand, but were not large enough to fully nsulate producers from the effects of that fall. Calls for more general tax relef to toursm and toursm busnesses are far less effectve at savng ob losses n toursm related ndustres. 07March2003 TOURISM CRISIS MANAGEMENT The Case of the U. S. Response to September 11 TOURISM CRISIS MANAGEMENT M. THEA SINCLAIR 1 UNIVERSITY OF NOTTINGHAM, UK BLAKE AND SINCLAIR ARTICLE ADAM BLAKE CHRISTEL DEHAAN TOURISM AND TRAVEL RESEARCH INSTITUTE NOTTINGHAM UNIVERSITY BUSINESS SCHOOL, JUBILEE CAMPUS NOTTINGHAM NG8 1BB, UNITED KINGDOM TEL FAX <A.BLAKE@NOTTINGHAM.AC.UK> ABSTRACT CONTENTS 1

2 1. INTRODUCTION POLICY RESPONSES TO SEPTEMBER FEDERAL GOVERNMENT RESPONSES TO THE CRISIS...6 THE TRAVEL INDUSTRY RECOVERY COALITION PLAN...7 BUDGET MEASURES MODELING TRAVEL AND TOURISM IN THE WIDER ECONOMY...8 PRODUCTION SECTORS...9 MARKETS...10 TOURISM DEMAND...10 RESULTS WITHOUT OFFSETTING POLICIES...ERROR! BOOKMARK NOT DEFINED. THE RESULTS OF CRISIS MANAGEMENT POLICIES...ERROR! BOOKMARK NOT DEFINED. 4. CONCLUSIONS...15 REFERENCES...17 FIGURES AND TABLES...19 : There have been many nstances where the toursm ndustry n one or more countres has suffered an unexpected and sudden downturn n demand. Polcy makers are faced wth the predcament of f and how they should respond to such crses n the absence of research nto the relatve effectveness of dfferent responses. The downturn n toursm demand n the Unted States followng September 11 s a partcularly vvd example of toursm crss. We Per our Style, the entre paper should be wrtten n the thrd person construct a computable general equlbrum model of the US to analyze the effects of the September 11-nduced toursm crss and potental and actual polcy responses to the crss. We fnd that sector-specfc targeted subsdes and tax reductons are the most effcent means of toursm crss management. Keywords: toursm crss management, September 11, computable general equlbrum. Please shorten t by a few words

3 1. Introducton The contrbuton of toursm and travel to both ndustralzed and developng countres s now so great that any downturns n the level of actvty n the ndustry are a cause for concern. The repercussons extend beyond actvtes drectly assocated wth toursm, notably arlnes, hotels and caterng, to sectors that supply ntermedate or fnal goods that are purchased by frms and employees n the ndustry, so that all sectors of the economy are affected to a greater or lesser extent. The general response by toursm and travel-related organzatons to sgnfcant downturns n toursm demand s to lobby governments to mplement a range of polces to offset the downturn. Polcy makers are then faced wth the dffcult decson of what, f any, measures to take, partcularly snce the mplementaton of many polcy measures s costly and the relatve effectveness of alternatve measures dffers. Polcy makers are confronted, n partcular, by three ssues that are relevant to any downturn n toursm actvty. The frst s the ssue of whether the downturn s suffcently large to mert offsettng measures. Informaton that s relevant to ths ssue ncludes the magntude of the decreases n ncome n travel and toursm actvtes, the dsperson of nduced effects across other sectors of the economy and the effects on welfare. The second s the ssue of the duraton of the downturn. If an economc shock s beleved to be long lastng or even permanent, polcy makers should concentrate ther attenton on mnmzng adustment costs as the economy moves towards a new equlbrum. If a shock s beleved to be of temporary nature, the concern should be to nsulate the economy from the adverse effects of the shock. The thrd ssue concerns the choce of polces for mplementaton. In ths context, polcy makers need to evaluate the relatve merts of alternatve polcy responses and consder the combnaton of polces and rates of relef that s most cost-effectve n rasng ncome and lmtng adustment costs. Toursm demand s partcularly senstve to securty and health concerns. The lst of cases where toursm demand has taken a sudden and unexpected downturn s long, and ncludes the effects of the 1991 Gulf War on toursm n the Mddle East, the Medterranean and other European countres, terrorst attacks at Luxor n Egypt n 1997, the Kosovo conflct n 1999, the effects of the foot and mouth dsease outbreak n the UK n 2001 and, more recently, the effects of the September 11 terrorst attacks n the US and of terrorst attacks n Bal and Kenya. In most of these cases, where governments have actually recognzed the mportance of the toursm downturn and acknowledged the need for a polcy response, they have operated n an envronment where very lttle research nto the merts of offsettng polcy responses has been conducted. The case of the September 11 terrorst attacks n the US s a notable example of the dffcult decsons that polcy-makers confront. Followng those events, falls n ncome and employment spread beyond travel and toursm throughout the entre US economy. Federal government response was swft, wth the Ar Transportaton Safety and System Stablzaton Act (ATSSSA) beng passed on September 22. Ths act provded federal government loans and compensaton to arlnes, as well as other fnancal and safety measures. The Travel Industry Recovery Coalton, comprsng 25 toursm and travel ndustry organzatons, proposed a Sx- Pont Plan amed at restorng the levels of actvty n the travel and toursm ndustry n the US to those exstng pror to the September 11 events. The Plan conssted of a seres of polcy measures coverng traveler and workforce tax credts, busness loan programs, the extenson of allowances for net operatng losses, and government and prvate sector fundng for advertsng to stmulate travel to and n the US. Government polcy makers were faced wth the dffcult decson of how to respond, gven the dfferng costs and relatve effectveness of the dfferent measures. The choce of polcy responses should, deally, be based on pror estmaton of the magntude and dsperson of the mpact of the downturn and of the mpact of alternatve polcy measures 3 <#>Adam Blake s Research Fellow and Thea Snclar s Professor at the Chrstel DeHaan Toursm and Travel Research Insttute, Nottngham Unversty Busness School (Jublee Campus, Nottngham NG8 1BB, UK. Emal: <adam.blake@nottngham.ac.uk>). Ther research nterests cover all aspects of the economcs of toursm, n partcular CGE modelng, toursm demand models, compettveness, destnaton mpacts, and economc polces n ths feld. If you wsh, you can add another sentence here (about 10 words). <#>Keywords: toursm crss management, September 11, computable general equlbrum. <#>Runnng Head: Toursm Crss Management Page Break TOURISM CRISIS MANAGEMENT <#>U.S. Response dng to September 11 long-lastng most

4 across dfferent sectors of the economy. Few other studes have quantfed the effects of alternatve polcy responses to toursm crses from an economc perspectve, although there have been a number of non-economc studes of toursm, terrorsm and poltcal nstablty. These nclude the revew by Sonmez (1998) and the earler volume by Pzam and Mansfeld (1996). More recently Sonmez, Apostolopoulos and Tarlow (1999) dscussed the management of the mpact of terrorsm from the perspectve of organzng crss management task forces and preparng nformaton. Cross-cultural dfferences n travel agents perceptons of poltcal nstablty have been examned by Seddgh, Nuttall and Teocharous (2001). Estmaton of the economc mpact of changes n the demand for travel and toursm has, typcally, been undertaken usng nput-output (IO) analyss (for example, Fletcher and Archer 1991; Archer, 1995; Archer and Fletcher, 1996; Henry and Deane, 1997). IO modelng s a well-establshed means of takng account of the nter-sectoral mpacts of dfferent types of exogenous shocks to an economy. However, IO modelng reles on a number of assumptons, such as fxed prces and wages, whch may not be approprate for the crcumstances of specfc economes. Some of these assumptons have been relaxed n a number of studes, for example Andrew (1997), Wagner (1997), Jensen and Wanhll (2002). An alternatve methodology that has been ntroduced to allow for flexblty n the prces of factors of producton and fnal goods, as well as consstency of ncome and expendture, s computable general equlbrum (CGE) analyss. To date, ths methodology has been used to examne the economc mpacts of toursm and travel n only a small number of countres, for example Hawa (Zhou, Yanagda, Chakravorty and Leung 1997), Australa (Adams and Parmenter 1995; Dwyer, Forsyth, Madden and Spurr 2000), Indonesa (Sugyarto, Blake and Snclar 2002), Span (Blake 2000), the UK (Blake, Snclar and Sugyarto forthcomng), and n more general cases (Alavalapat and Adamowcz, 2000). It has a strong track record n the feld of nternatonal trade and development (revewed by Deveraan, Lews and Robnson 1982; de Melo 1988; Shoven and Whalley 1992; Franços and Renert 1997). IO modelng typcally takes a change n fnal demand such as an ncrease (or decrease) n expendture by toursts and traces the effects of ths change through an economy, wth each expandng (contractng) sector s demands for ntermedate products leadng to further expanson (contracton) n other sectors. The technque s often extended by allowng nduced effects, followng ncreases (reductons) n household and government ncome, to lead to ncreases (reductons) n prvate consumpton and government expendture. Crucally there s no constrant on the employment of factor servces such as labor and captal, whch are avalable n any quantty at unchanged wage rates. Prces are assumed to be constant because wage rates do not change, and also because of the smple producton and compettve structures nherent n IO models. In realty, however, wages and prces do change when an economy adusts to a substantal shock. After September 11 there was notceable dscountng of prces by toursm frms not only n the US but also around the world. Some workers who become unemployed as a result of a shock take up lower-pad employment n other sectors of the economy. Workers who retan ther obs may experence falls n average wages through mmedate reductons of overtme workng and, n the longer-term, through smaller annual wage ncreases or (more rarely) through actual wage decreases. Wage and prce flexblty s an nherent part of the mechansms through whch an economy adusts to a shock, and therefore t s mportant to nclude such flexblty when modelng the effects that a shock wll have. IO analyss, by gnorng wage and prce flexblty, gnores the ablty of workers to fnd employment n other ndustres and therefore overestmates the effect that a demand reducton wll have on GDP and unemployment. IO models also have dffcultes n modelng the effects of polces. The man focus of these models s to trace fnal demand shocks through to GDP, so they are ncapable of modelng changes n tax rates or subsdes. CGE models, by allowng a dfference between the prce that consumers pay for products and the prce that producers receve for producng them, can useful m For references wth more than two co-authors: when cted frst n the paper, name all the co-authors, from that pont on use the format you have, but do not type et al n talcs. wages, that et al. et al. et al. et al. et al. Franços et al. 1986; B, prces do not change W, however, such wll 4

5 ncorporate taxes and can allow these tax rates to be changed. Therefore, whle an IO model could (under the restrctve assumptons of such models) examne the effects of a change n government expendture, t could not examne the effects of a change n tax rates or subsdes. The ablty of governments to gve targeted subsdes or tax reductons to specfc ndustres s a sgnfcant form of response to toursm crses, and therefore needs to be ncorporated nto a model that seeks to examne government responses. CGE modelng s therefore well suted to examnng the economc mpacts of exogenous changes n the demand for travel and toursm, such as those assocated wth acts of terrorsm, as well as the mpacts of a range of related polcy measures. As n the case of the Sx-Pont Legslatve Plan n the USA, these nclude alternatve types of government expendture, taxaton and credts. A CGE model can take account of dfferng contexts; for example, dfferences n the degree of flexblty of money or real wages, changng returns to captal, changes n productvty, alternatve levels of capacty n dfferent economc sectors, and alternatve constrants on the publc sector borrowng requrement. The man lmtatons of the CGE model are the dependence of the results on the estmated values of key parameters and the lack of provson of statstcal confdence ntervals for the results that are obtaned. However, these can be overcome by ensurng that the parameter values are accurate estmates and/or by undertakng senstvty analyss to evaluate the responsveness of the results to changes n the parameter values. The model reles on the use of data from an IO table estmated for a perod other than that of the ncdent n queston; for example, the CGE model for the US uses data from an IO table based on 1992, updated wth data for Ths s not a serous lmtaton as, n practce, the structural relatonshps between the dfferent sectors of the economy rarely change sgnfcantly over short tme perods. Those cases n whch structural relatonshps do change sgnfcantly, such as the case of the ol prce shocks, are easy to dentfy, so that the model can be modfed to take them nto account. The CGE model s, therefore, approprate for examnng the mpact of alternatve combnatons of mcroeconomc measures wthn dfferent macroeconomc settngs. Ths paper wll use a CGE model to estmate the mpact of the downturn n travel and toursm, caused by the September 11 events n the US, on dfferent sectors of the economy. Ths wll be the frst tme that toursm crss management n the US has been examned usng such an approach and wll buld on the record of the mmedate mpacts of the September 11 attack provded by Goodrch (2002). The model wll be used to estmate the mpact of dfferent polcy responses ncluded wthn the ATSSSA and the Sx-Pont Plan. Ths wll provde nformaton that s useful for evaluatng the relatve effectveness of dfferent polcy responses n terms of ther mpacts on dfferent sectors of the economy, as well as n relaton to alternatve mpact crtera, notably ncome generaton, labor and captal adustment costs and employment generaton. The approach used n the paper s appled to the case of the US, so that the results are specfc to the effects of September 11 on the US economy. However, the approach s of general applcablty and the case study llustrates the ways n whch the model can be used to quantfy the effects of alternatve types of crses and of the range of possble polcy responses, n order to dentfy those that are most effectve n dfferent crcumstances. Ths s straghtforward wthn a CGE modelng context, n whch model parameters can be changed to sut dfferent real world crcumstances. The next secton of the paper wll brefly document the polcy responses that occurred after the September 11 crss and some of the other polces that were proposed. The model of the US economy and the relatve merts of dfferent polcy nterventons are then dscussed. The results from the model are provded and the relatve magntudes of the effects are compared across economc sectors. The effects of the dfferent measures n contrbutng to ncome and employment generaton are also gven. The fnal part of the paper provdes some conclusons and mplcatons for toursm management polces. M 7 model It specfc but. that we develop to estmate the effects of the crss 5

6 2. Polcy Responses to September 11 The magntude of the toursm downturn followng September 11 was large, as s evdent n Fgure 1, whch shows the annual growth rates of domestc and nternatonal passenger enplanements at US arports from January 2000 to June A clear downturn s observed n September 2001 n both fgures. The percentage changes n enplanements n September 2001 compared wth the prevous twelve months are 34% for domestc travel and 23% for nternatonal travel. Whle the September fgures nclude a perod n whch all commercal ar traffc n the Unted States was grounded, the October fgures do not; they show falls of 22% and 33% compared wth the prevous 12 months nternatonal travel beng even lower n October than t was n September - demonstratng the ongong effects of the crss. By June 2002 some recovery had taken place, wth enplanements down by 11% and 7% over the precedng 12 months. Ths s stll less than the growth that the arlne ndustry would have expected f the September 11 attacks had not occurred. [Insert Fgure 1 around here] Federal Government Responses to the Crss The mmedate government responses to the September 11 attacks ncluded law enforcement actvty, dsaster relef and securty measures. Eleven days after the attacks, Congress passed the Ar Transportaton Safety and System Stablzaton Act (ATSSSA), to preserve the contnued vablty of the Unted States ar transportaton system (Lbrary of Congress, 2001a). Ths Act provded () federal credt to arlnes totalng $10bn; () compensaton to arlnes totalng $5bn; () compensaton for arlnes facng ncreased nsurance premums; (v) lmtaton of the extent of lablty by arlnes for the results of terrorst acts; (v) allowances for arlnes to make late payment of excse taxes; (v) compensaton for ndvduals klled or nured n the September 11 attacks; and (v) spendng of $3bn on arlne safety. The ATSSSA measures can be categorzed nto two types: frst, measures desgned to keep the arlne ndustry n operaton as a result of the September 11 attacks and, second, measures desgned to compensate those harmed, physcally and economcally, as a result of the attacks. The offer of federal credt to arlnes, lmtaton of lablty aganst future terrorst actons, allowances for late tax payments and measures to ncrease arlne safety were all ntended to ncrease confdence n ar travel and prevent the arlne ndustry from grndng to a halt. Wthout these measures, arlnes would have been unable to borrow captal and would have been refused nsurance by brokers. Compensaton payments to arlnes, compensaton of arlnes facng ncreased nsurance premums and compensaton of ndvduals were all measures amed at provdng compensaton for harm suffered as a result of the September 11 attacks. The compensaton payments gven to arlnes under the ATSSSA are only applcable where an arlne can show that t has suffered economc harm as a result of the September 11 attacks. An arlne can clam compensaton equal to the lower amount of (a) the economc harm t has suffered, and (b) ts August 2001 market share, multpled by $4.5bn, for passenger traffc, and ts market share n August 2001 of cargo traffc multpled by $500m. As of the end of 2002, a total of $4.6bn had been pad out to 410 ar carrers (Department of Transportaton 2002), and examnaton of pendng clams ponts to all of the $5bn allocaton for compensaton beng pad to arlnes. The ssue of whether the compensaton pad to the ndustry fully offsets the cost ncurred s open to queston. Three cases are possble: () the amount of economc harm caused to arlnes was between $4.3bn and $5bn, and all the compensaton pad to arlnes has exactly offset these losses; () arlnes ncurred less than $4.3bn n losses, but the ATSSSA rules enable them to clam for more damages than were ncurred; or () arlnes ncurred more than $5bn n losses but the upper lmt on compensaton prevents them from clamng more. The thrd case seems to be the more lkely, gven that ntal estmates of the cost of September 11 to the arlne ndustry exceeded ths fgure. Jones (2001), for nstance, n a news report on 6 POLICY RESPONSES TO SEPTEMBER 11 s and s and 2 The Ar Transportaton Safety and System Stablzaton Act and Avaton and Transportaton Securty Act Measures Please shorten ths secton headng md July, 3 382

7 September 13 estmated that the groundng of arcraft mmedately after September 11 would cost arlnes $10bn. The ATSSSA also places some restrctons on any arlne that receves compensaton or a loan under the act. Most sgnfcantly, any such arlne must ensure that communtes that receved ar travel servces pror to September 11 contnue to do so. In other words, the arlne cannot wthdraw all ts routes from arports where there are no other arlnes operatng, although the ATSSSA s not specfc about mantanng each specfc route or the frequency at whch flghts must contnue. The Act also places restrctons on the levels of compensaton that such arlnes can pay to ther own employees. The ATSSSA was extended on November wth the Avaton and Transportaton Securty Act (ATSA) (Lbrary of Congress, 2001b) whch () provded addtonal securty measures, such as the ntroducton of ar marshals, mprovements to arport securty and permsson for flght-deck crews to be armed wth non-lethal weapons; () provded $1.5bn to arports to pay for these addtonal securty measures; () provded $70m for research and development of arport securty equpment; (v) extended the ATSSSA s compensaton payments to cover ar tour operators and some other ar operatng companes not prevously ncluded n the compensaton scheme; and (v) lmted the lablty of New York Cty for the events of September 11. Further acts were passed by the federal government n the course of 2001 and 2002, relatng to arport securty and compensaton of vctms of terrorsm (Lbrary of Congress 2001c, 2001d). The Travel Industry Recovery Coalton Plan The Travel Industry Recovery Coalton (TIRC), a group of ndustry members formed to lobby government for more acton n response to September 11, proposed an addtonal Sx-Pont Plan for publc polcy desgned to restore the economc health of the U.S. travel and toursm ndustry to pre-september 11 levels. The sx ponts were that the federal government should: () provde a $500 tax credt for travel orgnatng and occurrng wthn North Amerca on ar, cruse, tran, bus, rental car, hotel and motel expendtures; () expand loan programs to small busnesses; () provde a workforce tax credt for employment n the travel and toursm ndustry; (v) provde substantal federal fundng for marketng campagns; (v) expand tax allowances to enable busnesses to offset losses sustaned because of September 11 aganst future earnngs; and (v) restore full tax deductblty for busness entertanment expenses that are currently only 50% deductble. The TIRC plan addresses ssues not tackled n the ATSSSA. Although the ATSSSA provded compensaton to arlnes, t dd not provde compensaton to other travel and toursm busnesses. The TIRC plan s not focused on the arlne ndustry but ams to provde support for the whole travel and toursm ndustry notably ncludng busnesses n other transportaton sectors. The measures ncluded n the TIRC plan can be classfed nto two categores: frst, measures desgned to stmulate economc actvty across the whole ndustry n order to offset the downturn caused by the September 11 events and, second, measures amng to ncrease lqudty and to prevent frms from ncurrng maor costs and possbly gong out of busness. Loan programs and tax allowances fall nto the second category. The frst category of measures provdes drect subsdes for producton by the toursm ndustry, such as the compensaton provded for arlnes, and for consumer expendture on t, for example the proposed tax credts for expendture on toursm. The frst category of measures has effects on the economy that are amenable to quantfcaton. They are also measures that ncur sgnfcant costs for government, so that t s partcularly useful for the government to have some knowledge of ther relatve effectveness. The CGE model wll be developed to estmate the relatve magntudes of ths category of polcy measures. 7

8 Budget Measures On 12 December 2001 the Presdent of the USA ntroduced an economc securty plan to Congress that was stll under consderaton by the Senate n md August The Budget for 2003 (OMB, 2002:406,409) ncludes provsons for ths economc securty plan, budgetng for tax cuts of $62bn n 2002 ($65bn n 2003) and $27bn ncreased expendtures n 2002 ($8bn n 2003). The economc securty plan s focussed on the Economc Securty and Worker Assstance Act (ESWAA), whch would, f passed by the Senate, () reduce some busness tax rates; () provde temporary assstance for needy famles; and () provde tax benefts for busness and prvate property n New York Cty that was damaged as a result of the September 11 attacks (lmted to a total of $15bn). In addton, the ESWAA ncludes other measures to extend a number of tax provsons that would otherwse expre, provde tax provsons for dsaster vctms, and other mscellaneous and techncal provsons. It should be noted that the largest economc effects of the economc securty plan ental tax breaks on drect personal and busness taxaton, whch have dstrbutonal mplcatons but are otherwse ntended as beng a fscal measure to stmulate general economc actvty. The 2003 budget also contans extra September 11 related expendture on defense and homeland securty. The defense budget s set to rse from $306bn n 2001 to $331bn n 2002 and $369bn n 2003 (OMB, 2002:101). $48bn of the 2003 expendture s an ncrease over prevous plans (Tobn, 2002). Homeland securty expendture s a category created n response to September 11, and wll rse to $37.7bn n 2003 (OMB, 2002:399). Tobn (2002) sees the ncreased expendtures on defense and homeland securty as havng a fortunate by-product of beng a drect fscal stmulus to the economy, and dscusses the merts of tax reductons covered n the economc stmulus package relatve to these expendtures. He concludes that the defense and homeland securty measures should be brought forward from 2003 and ntroduced as early as possble as a means of stmulatng the US economy. It should be noted that the combned ncrease n expendtures on these categores s $86bn, whch s sgnfcantly larger than the tax reductons n the economc securty plan, and also larger than the ATSSSA/ATSA measures and the TIRC s Sx-Pont plan. Therefore they ental greater costs to the federal government n terms of ts economc polcy response to September Modelng Travel and Toursm n the Wder Economy The effects of the September 11 crss and the relatve mpacts of the dfferent polcy responses on toursm and travel can be examned usng a CGE model of the US economy. The model ncludes 98 sectors and commodtes wth 53 manufacturng sectors, 35 servce sectors and 10 prmary sectors. Each sector uses toursm ntermedate nputs and other ntermedate nputs of each of the 98 commodtes, together wth captal servces and nputs of 23 types of labor. Each sector produces one or more of the 98 products, although most sectors manly produce ther own product. Each sector s taxed at a rate whch encompasses all forms of ndrect taxaton n the US economy. A representatve prvate household receves all factor ncome, pays ncome taxes to the government and spends ts ncome on resdent toursm, prvate (nontourst) consumpton and savngs. The government receves all tax ncome and spends t on tourst and non-tourst government consumpton. Trade occurs under the assumpton (Armngton, 1969) that goods produced n the USA dffer from goods n the same classfcaton mported from abroad. Exports are also qualtatvely dfferent from goods produced for the domestc market. The model allows for unemployment n each of the 23 labor markets and for under-utlzaton of captal. Toursm flows exst for prvate households domestc resdent toursm and toursm mports government, non-resdents and for each of the 98 producton sectors that purchase goods that are used for busness travel and toursm. For each of these toursm flows, separate flows are modeled for travel and toursm trps ncludng ar travel and travel and toursm trps not ncludng ar travel. Substtuton can occur between ar travel trps and non-ar travel trps for 8.Tobn. M trps,

9 example n response to cheaper flghts due to ATSSSA compensaton payments. The parameters n the demand functons for toursm are changed to smulate the effects of September 11 on the demand for toursm, wth ar travel trps becomng less attractve. The CGE model ncludes producton, consumpton, trade and market equlbrum relatonshps for 98 commodtes and producton sectors. Model equatons consst of two general types: accountng relatonshps such as market equlbrum equatons for each commodty and factor of producton and ncome-expendture equatons for each producton sector, prvate consumers and the government; and behavoral equatons such as producton functons and utlty functons. These behavoral functons are drawn from the constant elastcty of substtuton (CES) famly of equatons that allow a parameter to control the degree of substtuton possbltes between nputs. Second order equatons are derved from these relatonshps that ensure that producers are proft maxmzng and that consumers are utlty maxmzng. Producton Sectors Producton functons are defned as: ( 1 λ )( σ ( σ 1) ) λ (( σ 1) σ) Q =ΘK al, Ll, (1) l The model s formulated so that government expendture on goods and servces s exogenous, and government expendture on travel and toursm trps s changed exogenously to smulate the effects of September 11. The government adusts ts budget through lump-sum transfers to prvate households. Net foregn savngs are fxed, so that the changes that are modeled do not affect nternatonal captal flows. World market prces are assumed to be determned exogenously. These condtons are known as the government, macroeconomc and external closures. whch defnes a Cobb-Douglas relatonshp between ndustry output 9 Q, captal nputs K and a CES functon of labor nputs L l, where s the set of 98 ndustres and l s the set of 23 types of labor. σ s the elastcty of substtuton between labor types (set equal to 2 to allow greater substtuton between labor types than between labor and captal) and calbrated parameters. Industry output Θ, λ and δ l, are Q s spread across the output of more than one commodty (although the make matrx ndcates that the maorty of each ndustry s output s of the correspondng commodty) through a constant elastcty of transformaton (CET) functon wth elastcty of unty: Q = B Π S β (2),, where S, s the supply of commodty by ndustry ; B and β are calbrated parameters. Income-expendture consstency s ensured through an equaton that sets ndustry revenues equal to producton costs: ( ) PP 1 t S = r. K + w L, l l, l N AT NT + PQ ( α, + α, ( 1+ tbat ) + α. ( 1+ tbnt )) (3) + P α Q M M where PP s the producer prce of commodty pror to the deducton of an ndrect producton tax t. r and w l are the wage rates of captal and each type of labor. Inputs of each commodty (purchased at prce P ) are a lnear functon of output ( Q ). These nputs appear three tmes n equaton 3: nputs of non toursm products, nputs of products purchased n ar toursm trps and nputs of products purchased n non-ar toursm trps. The total pad for N ntermedate products that are not toursm related s PQα,. Wth the ncluson of a tax on

10 the two toursm components, whch s zero n the base and used only when modelng the margnal mpact of addtonal polces, the total pad for ar-trp toursm related nputs s AT PQ α, ( 1+ tbat) and the total pad for non-ar toursm related nputs s PQ ( t ) NT α, 1+ BNT M lnear functon α Q of output.. Inputs of non-compettve mports (purchased at prce P M ) are also a Equatons 1, 2 and 3 are combned under the proft maxmzng assumpton to create nput demand equatons for K and L, whch are used n the model. Markets Domestc producton (the sum of supply of good by each ndustry ) s allocated to export and domestc markets accordng to a CET functon: where ( ( ) ) (( ) ) ( τ ( τ 1) ) τ 1 τ τ 1 τ S, = T( χx + ( 1 χ) D ) (4) X s exports of good and D s domestc supply of domestcally produced goods. τ s the elastcty of transformaton between exports and domestc goods; T and χ are calbrated parameters. The accountng dentty that ensures that payments receved by producers equals the value of exports plus the value of domestc products s: S, PP = DDP + XXP (5) Total uses n the domestc market mports where M. Ths functon s CES: ( ( ) ) (( ) ) ( ( 1) ) 1 ( ( 1 μ μ μ μ μ 1 μ θ θ) ) U s a product of domestcally produced products and U =Ω D + M (6) μ s the elastcty of substtuton between mports and domestc goods, commonly known as the Armngton elastcty. Values for these elastctes are taken from estmates for the US by Franços and Hall (1997). The correspondng accountng equaton s: where UP = DDP + MMP (7) P s the prce pad by producers for ntermedate nputs n equaton 3 and also by consumers, and MP s the mport prce. Supply and demand equlbrum ensures that total uses are equal to ntermedate demands for products plus consumer demands, government demands, nvestment demands and tourst demands. Toursm Demand Foregn toursm demand F s equal to a functon of toursm prces: ( ) F =Ψ PFT FE ε (8) 10

11 where PFT s the aggregate prce of the commodtes purchased by foregn toursts, FE s the prce of foregn exchange (the exchange rate), ε s the prce elastcty of foregn tourst demand (set equal to 2) and Ψ s a shft parameter that allows foregn toursm demand to be exogenously changed. PFT s a result of the CES nestng of toursm demand: where ( ρ ( ρ 1) ) (( ρ 1) ρ) π TD (9) F =ϒ TD s toursm demand for each ndvdual commodty, ρ s the elastcty of substtuton of toursts between commodtes (set equal to 2 to allow substtuton between commodtes), and ϒ and π are calbrated coeffcents. The correspondng accountng equaton for foregn toursm s: ( ) F. PFT = TDP 1+ tft (10) where t FT s a tax rate that can be appled to foregn toursm, although t s zero n the base case and s only used to examne margnal tax effcency. Domestc toursm demand s a part of prvate consumpton: γsave γnt γ DT U =Γ. SAVE. NT. DT (11) where U s the utlty level of the representatve household, SAVE s the level of real savngs, NT s the level of real non-toursm consumpton and DT s the real level of toursm consumpton. Γ, γ SAVE, γ NT and γ DT are calbrated parameters. Domestc toursm s tself composed of ar-trp toursm AT and non-ar toursm NAT components: ( ( ) ) (( ) ) ( ( 1) ) 1 ( ( 1 ω ω ω ω ω 1 ω ϖ ϖ) ) DT =Ε AT + NAT (12) where ω s the elastcty of substtuton between ar- and non ar- trps. Ths elastcty s set to a value of 2 to allow substtuton between the types of toursm. Both types of toursm are themselves a Cobb-Douglas functon of purchases of ndvdual commodtes: AT =Α (13) C φ AT, NAT =Ν (14) where C AT, and C, C ϕ NAT, NAT are the levels of real consumpton of commodtes used by domestc toursts on ar trps and on trps that do not nvolve ar travel. Α, Ν, φ and ϕ are calbrated parameters. The budget constrant for the prvate household s:, N, AT ( RAT ), NAT ( RNAT ) (15) Exp = SAVE. PSAVE + C P + C P 1+ t + C P 1+ t where PSAVE s the prce of savngs (the prce of purchasng one unt of new captal), C N, s consumpton of goods for non-toursm purposes, t RAT and t RNAT are tax rates on resdent ar toursm trps and resdent non-ar toursm trps. These tax rates, n common wth t BAT, t BNT and 11

12 t FT are zero n the base case and are only used when assessng the margnal mpact of subsdes to toursm actvtes. The equatons shown above do not completely defne the CGE model. Other equatons that are not shown here account for household and government ncomes, prvate consumpton behavor, as well as the operaton of nvestment, mports and exports and the balance of payments. Further assumptons that are often grouped under the term model closure nclude: () the government balances ts budget through the savngs and nvestment market, n other words the changes that are beng appled to the model wll lead to changes n the government defct but wll not be substantal enough to lead to changes n levels of taxaton or government expendtures; () the trade balance (ncludng toursm flows) s fxed, whch follows from an assumpton that the events beng modeled wll not lead to changes n US net captal flows; () the US faces exogenous world prces for ts exports and purchases mports at an exogenous world prces; (v) unemployed labor moves nto employment (and employed labor moves out of employment) when real wages change. Ths fnal pont allows ob savngs to be modeled, as the effects of September 11 lead to a fall n real wages n occupatons that are related through patterns of employment to tourst expendtures, and therefore employment of labor n these occupatons falls. Targeted subsdes can, partally at least, prevent ths fall n employment. Only a few of the many varables n the model are reported here. Constant dollar GDP s a standard measure of economc actvty. The net effect on the government budget s the net change n (federal, state and local) tax revenues mnus extra expendtures ncurred as a result of September 11. Constant dollar factor adustment s a measure of captal and labor adustment, gven n dollar terms. More specfcally, t s the constant dollar value of all factors that move from ther orgnal sector of employment, ether nto another sector or to become unemployed or, n the case of captal, underutlzed. It does not measure the costs of adustment, of retranng, retoolng and payments of unemployment beneft, but t s an ndcator of the relatve sze of these adustment costs. Relatve factor adustment s a smlar measure, but s gven as a percentage of all factor employment rather than n constant dollars. Constant dollar employment shows the real value of labor employment whch, unlke the prevous two measures, ncludes the net effect of labor beng re-employed n other sectors. FTE employment shows the net employment pattern n terms of full-tme equvalent workers, whle FTE obs lost shows the number of obs lost as workers are dsplaced from ther orgnal sector of employment, ether to be unemployed or re-employed n another sector. Two further ndcators show the number of obs lost n () arlnes and () hotels and other accommodaton establshments. Data are obtaned from varous Bureau of Economc Analyss (BEA) sources. The benchmark nput-output table for 1992 (Lawson 1997) s used n conuncton wth data for GDP by orgnatng ndustry (Lum and Moyer 2001) and natonal ncome and product account (NIPA) data (Bureau of Economc Analyss 2002) to construct an nput-output table for the second quarter of Data from the prototype toursm satellte account (Okubo and Plantng 1998; Kass and Okubo 2000) are combned wth estmates of toursts use of motor vehcles and vacaton homes (Okubo, Fraumen and Fahm-Nader 2001) and updated usng NIPA data to provde estmates of toursm consumpton n the second quarter of These are smlar to the BEA s own release of updated quarterly toursm satellte account results except for the ncluson of toursts use of motor vehcles and vacaton homes, as recommended by Okubo et al. (2001). Employment data are obtaned from the Bureau of Labor Statstcs (BLS 2002) and updated usng GDP by orgnatng ndustry and NIPA data to provde data on employment by labor type n the second quarter of 2001.,,,,, et al. second-quarter, 12

13 4. The Effects of September 11 on Toursm The frst set of results from the model ndcate the effects of the September 11 events n the absence of any offsettng polcy responses. It s clear that there are severe effects n terms of a varety of crtera, ncludng ncome, government revenue, labor and captal adustment costs, employment and ob losses, as s shown n Table 1. As the frst column of the table shows, the fall n toursm expendtures reduces Gross Domestc Product (GDP) to almost $30 bllon less than t would otherwse have been, and worsens the government budget by over $7 bllon. Factor adustment, the amount of captal and labor that leave ther orgnal sector of employment, s $31 bllon. The loss of employment s hgh, at 383 thousand full tme equvalents (FTE). A total of 559 thousand obs are lost, of whch 203 thousand are n arlnes and 174 thousand are n accommodaton establshments. [Insert Table 1 around here] The toursm effects of September 11 can be decomposed nto the effects of reduced demand for ar travel and toursm trps by non-resdents and by US resdents, as s shown n the second and thrd columns n Table 1. It s apparent that the reductons n demand by US resdents are domnant for all types of effects, but partcularly for factor adustment costs and for FTE obs lost n the arlne ndustry. Ths demonstrates the mportance of domestc toursm and travel n the US economy and the sgnfcant effects that decreases n domestc demand can brng about. The effects of the reductons n demand by non-resdent toursts are lower but are spread more wdely across the economy. 5. Alternatve Polcy Responses Gven the fndng that the effects of the decreases n toursm demand brought about by September 11 are severe, the related queston s whether the mplementaton of crss management polcy responses succeeded n offsettng the downturns. The CGE model of toursm and travel was used to estmate the effects of all of the quantfable polcy measures that were mplemented by the ATSSSA and ATSA;.e. compensaton to arlnes and ndvduals, spendng on arlne safety and expendture on marketng and the effects of September 11 wth and wthout these crss management responses are ncluded n Table 2. [Insert Table 2 around here] Perhaps the most mportant fndng n the table s that the polcy responses are very effectve n offsettng the adverse effects of the crss. The fall n GDP s much lower than t would otherwse have been, at under $10 bllon compared wth nearly $30 bllon n the wthoutpolcy case. The measures worsen net government revenues by the relatvely large amount of around $11 bllon, whch ncludes $5 bllon n arlne compensaton and $4.5 bllon n securty spendng outlned n the ATSSSA and ATSA acts. The polces are less effectve n decreasng the factor adustment costs relatng to the crss, almost halvng the value of the labor and captal that leave ther orgnal sector of employment. The loss of employment s less than half of what t would otherwse have been, so that the number of people affected s below 150,000 compared wth almost 400,000 n the non-response case. The number of FTE obs lost s around 60% of the total that would otherwse have occurred, at 335,000 compared wth over half a mllon wthout a polcy response. The number lost n the arlne ndustry s 93,000 obs (28% of all obs lost) after the polcy response, compared wth over 200,000 (36% of all obs lost) wthout a polcy response. The model was also used to estmate the relatve and absolute effectveness of the ATSSSA and ATSA measures on the worst ht sectors of the economy, wth respect to the dfferent measurement crtera (changes n ncome, labor and captal adustment and ob loss). The results are provded n Tables 3 and 4. [Insert Table 3 around here] 13. T It s nterestng to note that the estmated fgure of..., prevously provded n Blake and Snclar (2002), s vrtually the same as the Travel Industry of Amerca s (2002) subsequent estmate of... dfferent types of polcy responses on

14 Table 3 shows the fall n constant-dollar employment n the ten sectors that are affected the most by the crss. Ar transport s the worst ht sector pror to the polcy responses, but when polcy responses are ncluded, hotels are the worst ht sector. A total of 35 of the 98 sectors lose out n terms of factor employment from September 11, wth 31 of these sectors stll losng after the polcy response. [Insert Table 4 around here] Table 4 shows the results of modelng several dfferent possble polcy responses to the crss. Each of these responses nvolves a specfc type of subsdy scheme, whch can be nterpreted as an equvalent tax reducton. The dfferent types of crss management responses are dvded nto fve man types n Table 4: subsdes to producton, subsdes to consumers expendture, subsdes to labor employment, subsdes to captal profts, and more general fscal stmulus measures. The former provdes a drect subsdy to sectors that are partcularly affected. Subsdes to consumers expendture subsdze (or provde tax relef to) toursm and travel expendture. Subsdes to labor and captal gve specfc support to these factors of producton (n the case of labor, to all of the 23 types of labor). Thus, for example, compensaton to arlnes s ncluded n the former, whle tax credts for personal and busness expendture on toursm and travel are ncluded n the second category and a workforce tax credt s ncluded n the thrd category. Ths enables the effects that these polces have after September 11 and the ntal polcy responses of the ATSSSA and ATSA to be compared. Margnal changes are made to the approprate subsdy nstrument to facltate comparsons of the effects of the polces. The change n constant dollar GDP and the other ndcators shown are dvded by the change n net government revenue as a result of the margnal subsdy change to gve the margnal effects per mllon dollars of subsdy expendture. A producton subsdy to the arlne sector (whch can also be nterpreted as a tax reducton) ncreases GDP by $3.1 mllon for every mllon dollars n subsdy. It has a mnmal effect on the net government budget, as t s so effectve at haltng the declne of the arlne sector and reducng unemployment that the stmulated economc actvty generates almost as much revenue through drect and ndrect taxaton as s spent on the arlne subsdy. Every $1 mllon of arlne sector subsdy reduces factor adustment by $1.6 mllon and reduces the total number of obs lost by 27.8, of whch 20.7 obs are saved n the arlne sector tself, but only 0.1 obs are saved n the accommodaton sectors. The results n the table show that the provson of subsdes targeted to the arlne ndustry are the most effectve polcy response accordng to most crtera. An arlne producton subsdy outperforms all other types of producton subsdy n each of the crtera wth the excepton of the number of accommodaton obs lost. Whle more effcent n these terms, however, an arlne subsdy fals to take account of ob losses outsde the arlne ndustry that are caused by the fall n demand from toursts who would have traveled by ar. The provson of subsdes to hotels and other accommodaton establshments, though less effectve than a subsdy to arlnes, s reasonably effectve at boostng GDP and savng obs n accommodaton. These subsdes are sgnfcantly more effectve than subsdes to caterng and entertanment. In fact, subsdes to the latter can have the effect of worsenng GDP and labor and captal adustment, as they encourage workers to move out of the arlne and accommodaton ndustres, thereby ncreasng the ob losses n these sectors. Ths demonstrates that poorly targeted subsdes can not only fal to allevate adustment costs, but can also add to them. Consderaton of the effects of subsdzng toursm and travel consumers shows that the provson of subsdes for ar travel expendture s more effectve than the provson of a more generalzed subsdy for all types of toursm and travel expendture - as shown by the tem all domestc toursm. The results for the latter are ndcatve of the effects of provdng consumers wth a tax credt for expendtures on ar, cruse, tran, bus and rental car travel, hotels and motels, as requested n the frst pont of the TIRC Sx-Pont Plan. Such a subsdy performs badly because t s not focussed on the types of expendtures that have fallen snce 14 As the results n Table 3 show, a arlnes are only part of the story; We model a set of toursm management The are In order to make comparsons of the effects of the polces, m travelers

15 September 11 tourst expendtures on trps nvolvng ar travel. Smlarly, subsdes for expendture on all resdent trps are less effectve than those drected solely towards ar travel. Subsdes for busness ar trps are the most benefcal n terms of GDP (though not n terms of adustment costs) because busness travel, as an nput nto producton n other sectors, ncreases the effcency of those other sectors. Subsdzng non-resdent trps has the greatest effect n reducng adustment costs and ob losses and has the second largest effect on preservng obs n the arlne ndustry. Subsdzng resdent ar trps preserves more obs n the arlne ndustry per $ mllon of subsdy than any of the other toursm and travel expendture subsdes. Subsdzng all ar trps spreads the effects of the ob savngs more evenly across the arlne and accommodaton sectors. Subsdes to employment n arlnes are ust as effcent n terms of GDP and net budgetary effect as arlne producton subsdes, but have a much greater mpact on savng obs n the arlne ndustry, as would be expected. Proft subsdes perform ust as well n terms of GDP and factor adustment, but perform badly n terms of savng obs. However, whle there s nevtably some poltcal pressure to mnmze ob losses, there s also a need to mnmze the adustment costs to captal. The employment and captal subsdes to all toursm sectors (arlnes, hotels and other accommodaton) perform smlarly to the producton subsdy on these sectors, but wth dfferent employment mplcatons. Cuts n drect taxaton and mltary expendture both perform relatvely poorly n terms of stmulatng GDP, reducng factor adustment and savng obs n sectors affected by September 11. They are also expensve n budgetary terms, as very lttle of the tax cut and mltary expendture s returned to the government through ncreases n other forms of taxaton, as the net budget effect of every $1 mllon spent s close to $-1 mllon. Whle these polces may perform better n the context of a government response to general economc recesson, they are not a partcularly effcent means of counteractng the fall n toursm demand followng September Conclusons Three ssues of central mportance to toursm crss management were posed at the begnnng of ths paper: frst, that of whether the downturn n actvty assocated wth the crss s suffcently large to mert offsettng measures; second, whether the shock s suffcently longlastng to mert measures to mnmze adustment costs; and thrd, what types of polces should be mplemented. The analyss of travel and toursm n the US and the applcaton of the CGE model to the case of the September 11 events has provded substantve answers to these questons. The results provded by the model ndcate that, n the absence of offsettng polces, the terrorst attacks would have had the effect of decreasng GDP by almost $30 bllon, wth the arlne ndustry, followed by hotels and other accommodaton, beng partcularly badly affected. The total number of obs lost would have exceeded half a mllon. Hence, t s clear that the downturn n actvty was, ndeed, a serous cause for concern. The magntude of the adverse effects stemmng from the September 11 events was so great that any possblty that they would be of short duraton was precluded. Moreover, examnaton of statstcs for nternatonal enplanements showed that the decreases n the month of October were even greater than those n September and that the decreases below the averages for the same months n the prevous year were ongong. The assocated declnes n the use of labor and captal n affected sectors, notably arlnes and accommodaton, are not matched by mmedate ncreases n ther use elsewhere, owng to mperfectons n nformaton, costs of nter-sectoral factor moblty and other frctons wthn the market. For these reasons, along wth the socal costs of unemployment and dslocaton, there s a clear case for polces to decrease the adustment costs assocated wth the September 11 events. 15