Slide1: Discussion on Extended Validity of an Alternative Framework to Estimate Short-Term Negative Impacts of an Unexpected (Unprecedented) Event Tad Hara: (Presenter) Regional Science Program & School of Hotel Administration, Cornell University The 5th Business Enterprises for Sustainable Travel Education Network Think Tank and Conference Session 2B: Economics of Tourism Crises and Disasters 11:45~ June 17th 2005


Prologue: Economic Impact of Terrorism: Prologue: Economic Impact of Terrorism Negative Impact of Terrorism over Economy Widely assumed, but hardly quantified Vulnerability of Certain Sectors E.g. Tourism industry One of the largest & fastest growing export sectors for many nations, its potential looks bright, but its vulnerability (risk of revenue volatility) seldom quantified Benefit of quantification for policy makers Proper response, better plan for prevention E.g. Which industry will suffer most, jobs, income, tax revenue, etc. How much budget can be justified for preventive measures


Tourism Industry: Tourism Industry “Tourism can be one of the few development opportunities for the poor.” World Tourism Organization 2002


Prologue: Economic Impact of Terrorism: Prologue: Economic Impact of Terrorism Need to study this topic?


Economic Impact of Terrorism: Introduction: Economic Impact of Terrorism: Introduction Economic Impact of Terrorism is hard to quantify – Best estimates of “experts” often used How do we estimate? 1 Quote whatever the other sources said 2 Econometrics / Time Series Model 3 Input Output/Social Accounts framework


Literature Review (1/5): Literature Review (1/5) Literature on the Negative Economic Impact of Natural Disasters Horwich (2000): “it is capital stock, not output, that is directly reduced by the disaster” (Kobe Earthquake) Murdoch, Singh & Thayer: S.F Earthquake caused 2% reduction in housing values.


Literature Review (2/5): Literature Review (2/5) Negative Economic Impact of a Man-Made Disaster Cohen (1995) Impact of Exxon Valdez Oil Spill Concede difficulty of Ex-post analysis Globar (1993) Econometric model (ex-post analysis) Identified Inverse relation between defense expenditure and investment in Sri Lanka. Coshall (2003) Intervention analysis (impacts of events on UK Air travel using ARIMA) Negative Surprise tend to increase volatility than positive one (asymmetry of volatility: Engle et al 1990)


Literature Review (3/5): Literature Review (3/5) Input Output Models Isard & Kuenne (1953) Estimating Impact of Steel industry in NY-Philadelphia region Demonstrated employer effect is much higher than prior assumption of 1:1. Ahlert (2001) Impact of Soccer World Cup 2006 to German Economy English (2002) Impact of Converting Corn to Ethanol Production Nakajima (1994) International I-O model (Japan & Asia) Lee (1994) I-O with fuzzy final demands


Literature Review (4/5): Literature Review (4/5) I-O Model Application to Tourism Industry Fletcher (1981) Economic Impacts on Gibraltar While final demand is dominated by UK defense expenditure, tourism generated the highest marginal increase in income and employment Archer (1982) Usage of I-O for tourism policy Heng & Low (1990) Tourism industry’s impact on Singapore Economy


Literature Review (5/5): Literature Review (5/5) Negative Impacts with the I-O model Caskie, Davis & Moss (1999) Simulation of Negative Impacts of BSE on N. Irish Economy Zhou, Yanagida, Chakravorty & Leung (1997) Simulation of 10% decrease in tourism revenue in HI (partially using CGE by GAMS) Okuyama, Hewings & Sonis (1997) Kobe Earthquake’s shockImpacts from unscheduled events are not only the negative effects but also positive effects of reconstruction.


Estimation--Assumptions: Estimation--Assumptions Initial Shock – fairly simple one Assume that the reductions in number of employed between Sept. and Oct. 2001 are all attributed to the huge exogenous shock Allocate the initial shocks to SIC 1-digit industry categories


Simulations—Initial Shock (Services adjusted: Adjusted Version): Simulations—Initial Shock (Services adjusted: Adjusted Version)


Simulations--Results (Services Adjusted: NYS Results: Latest Version): Simulations--Results (Services Adjusted: NYS Results: Latest Version)


Simulations--Results (Services Adjusted: NYC results: Adjusted version): Simulations--Results (Services Adjusted: NYC results: Adjusted version)


Review of Other Studies: Review of Other Studies NY City Economic Impact “Total Loss $83 billion” (NYC partnership & Chamber of Commerce: Nov 2001) “Total Cost $54 billion” (NY Governor: Oct 2001) “WTC Replacement Cost & Cleanup $25~29 billion” (FEB NY: April 2002) “Total Cost $83 billion (quoting NYCP-COC) but $67 billion covered by Insurance (US GAO: May 2002) NY City JOBS Lost 108,500, 115,300, 105,200, 125,000, 84,000, 78,200, 129,000…. NY State Jobs Lost “99,000 in 2001, 78,000 in 2002, 77,000 in 2003” (NYS Senate Finance Committee: DRI-WEFA: January 2002) “Resulted at peak loss of 78,200” (DRI-WEFA: March 2002) “50,000 immediately, 70,000 in 4th Quarter” “Much of this loss is likely linked to WTC attack” (FEB NY: April 2002)


Ch 4: Verdict--Forecast VS Reality: NYC: Ch 4: Verdict--Forecast VS Reality: NYC


Ch 4: Verdict--Forecast VS Reality: NYS: Ch 4: Verdict--Forecast VS Reality: NYS


Ch 4: Verdict--All Studies In Perspective Compare with Actual Data: Ch 4: Verdict--All Studies In Perspective Compare with Actual Data


Ch 4: Verdict—Historical Data: Ch 4: Verdict—Historical Data


Concluding Remarks & Future Research: Concluding Remarks & Future Research Concerns Important assumptions Varieties of interesting problems in “ex post” analysis Time Frame Issues (in what time span) Rooms for Refinement Multivariate Econometric Model Time Series Analysis Let’s have a quick look


Review of Other Types of Models: Review of Other Types of Models 1. Multivariate Econometrics Model Future depends on past associations of dependent variable with independent data Convenient model, widely used, convincing Works well after all the data are on the table (when dust settles down) 2. Time Series Model Future depends on past behaviors of the own data Powerful, good at finding patterns But…“Because many models make predictions based on market’s past behavior, they can be easily wrong-footed by unusual market moves” “We look at the worst probable risk, not the worst possible risk.” WSJ9/27/02 “Rocky Markets Foil Firms’ Bets Based on ‘Risk Models’


Review of Econometrics model : Example: Review of Econometrics model : Example Can you find these data at t=0? Or How long you wait?


Review: I-O/SAM model: Review: I-O/SAM model Where, X=Total Output (an nx1 vector), I=Identity Matrix an nxn matrix), A=“A” matrix (standardized inter-industry coefficient matrix: nxn), Y=Final demand (an nx1 vector) You may rewrite the equation…. Increase [loss] in final demand in some sector will result in Increase [loss] in Total Output (overall economic activities)


Data Availability?: Data Availability? Source: Time managize: Suzanne Plunkett/AP http://www.time.com/time/photoessays/wtc/7.html


Review of Time Series model: Example Time Series Analysis: Review of Time Series model: Example Time Series Analysis Source: the authors by SAS based on data from Labor Department of New York State.


Concluding Remarks & Future Research: Concluding Remarks & Future Research Advantages of I-O/SAM framework for Estimating Negative Impacts of Terrorism General Criticism of I-O/SAM: “No Capacity Constraints” does not hold Structurally good at capturing “a huge shock” Superior to other models which may estimate direct impacts only in that I-O/SAM will capture total negative impacts to a national/ regional economy by describing indirect and induced effects of terrorism. External Validity We can quantify impacts of large negative shock for a nation/region GIVEN I-O/SAM table and solid labor data.


Concluding Remarks & Future Research: Concluding Remarks & Future Research Rooms for Improvements General EquilibriumCGE model (you can make variables endogenous; interest rate, price, exchange rate etc: WARNING: CGE’s accuracy highly depends on the very accuracy of I-O/SAM data) Hybrid with econometrics for final demand (∆FD), some stochastic model (Once data become available), ARCH, GARCH, E-GARCH, or Structural Equation with Regime Shifts? Can We Predict whether structural shift occurred given current surge/plunge of data (t=0)? Frontier topic of Time Series Based on constraints of yt = αyt-1 + εt . it is challenging to endogenize the shock. (Bayesian, Kalman filter?) I-O/SAM seem to have good validity to deal with huge exogenous shocks amid the chaos immediately after the huge unprecedented shock.


Concluding Remarks & Future Research: Concluding Remarks & Future Research Research on Negative Shocks: Economic Impacts and Beyond Quantification of potential threatsEasier to justify expenditures for proactive measures Final demandinfluenced by psychological factors? Roles of mass media--intermediary of information The whole system of Terrorism Confrontational eradication by “War on Terrorism”, or Research on why some resort to terrorism?—multidisciplinary subjectits understanding may provide cost-effective prevention Systematic Prevention is important to Realize Potential of Stable Growth of Tourism Industry


Slide29: Case: WTC Attacks on 9/11/2001 NY City


Geographical Distribution of Terrorism Events: Geographical Distribution of Terrorism Events Figure 1-1 "Patterns of Global Terrorism: 2000." Source: U.S.State Departmenthttp://www.state.gov/s/ct/rls/pgtrpt/2000/ Will you believe…that history repeats itself?