Application of Price-Elastistic Supply Input-Output Model on the Economic Impacts of Hurricanes Katrina and Rita: Disruption in U.S.

Transcription

1 Alication of Price-Elastistic Suly Inut-Outut Model on the Economic Imacts of Hurricanes Katrina and Rita: Disrution in U.S. Oil-Industry JiYoung Park Von Kleinsmid Center 382 School of Policy, Planning, and Develoment University of Southern California Los Angeles, CA Phone: (213) Final version of draft

2 Alication of Price-Elastistic Suly Inut-Outut Model on the Economic Imacts of Hurricanes Katrina and Rita: Disrution in U.S. Oil-Industry Abstract Recent Hurricane Katrina and Rita caused severe economic cessations without exectations in Louisiana State. The natural disasters temorarily roduced quantity losses in -industry and lead to further economic losses via interindustrial relations. In a sense that industry lays a key role to suort national economy, estimation of the economic imacts of the disasters should be based on the sulydriven inut-outut (IO) model. However, common limitation in IO models is over-estimate total imacts in a relatively long-term duration due to the linear characteristics of IO coefficients. Emirically, we observe consumers change their behaviors as rice changes. The rice change would follow rice elasticity as well as quantity changes in the market. Therefore, by constructing a new suly-driven model using rice elasticity of demand for refinery, that is, rice-elastistic suly-driven model, this study will analyze the national economic imacts of disruted -industry due to Hurricane Katrina and Rita. This rice-elastistic suly-driven model is a temoral exansion of suly-driven IO study. This study, also, has an imortant imlication in the asect that it is the first emirical alication of rice-tye suly-driven model for the U.S. 1

3 I. Introduction Economic imact analyses on various natural disasters have been studied actively. Esecially, we exerienced a severe tragedy from Hurricanes on New Orleans in August 2003, which reminds us the imortance of rearation once again; what should have been reared to reduce the damages from similar revious exeriences? As a art of answers on the question, certainly we can reduce the chained damages with our efforts, although it is not easy to block the nature s attacks on our community erfectly. The endeavors include tied-communications between all articiants in our community, hysical reairs of frail laces through checking regularly, exress-oeration activity from federal and local government, and so on. However, with what can we reare those necessary activities? The estimation of economic imacts on the disasters has been layed a key role in imlementing federal tax and distributing necessary resources to the ublic. Because the economic models estimating the economic losses from historical disasters of certain regions serve to distribute aroriate resources of federal and local governments to the targeted region, many social and regional scientists have interested in and contributed to develoing an aroriate economic model. Certainly, the economic models rovide base information on reventing and mitigating a disaster occurred from the disrution of communities and residents of the targeted region. To estimate aroriate imacts, hence, the economic model should be well designed to estimate the actual damages from a disaster onto the region. In the sense, this study involves two issues. One is to introduce a new national inutoutut model combined with rice elasticity of demand. Another investigates the U.S. economic 2

4 imacts from the recent huge natural disasters of hurricanes Katrina and Rita, based on the rice elasticitic suly-driven IO model. The rest of this aer includes the introduction of the issues and results in the revious study of economic imacts analyses from the Hurricanes Katrina and Rita. In the third section, the rice-elastistic suly-driven IO model for the U.S. and the data for the elastistic model are exlained. Based on the model, results of this study are addressed. Also, the conclusions and remarks are followed. II. Issues Recent hurricanes Katrina and Rita caused severe economic cessations without exectations in Louisiana state. Holtz-Eakin (2005) stated the direct caital losses from the hurricanes Katrina and Rita ranged from $70 billion to $130 billion. According to Louisiana federal reimbursement reort, the losses are estimated u to $115 billion (Kent, 2006). Another estimate is conducted by National Hurricane Center (2007) and reorting $92.3 billion damages from the two hurricanes. However, those economic imacts do not count for the economic interrelations, although many direct economic losses lead to further economic losses via interindustrial and interstate relations. Using satially disaggregate Inut-Outut model, Park et al. (2006) reorted total economic losses, including direct and indirect losses, resulted from the shut-down of New Orleans ort during seven months as $62.1 billion. Further, Kim et al. (2007) addressed the economic imacts from the hurricane disasters in -refinery industry as $4.8 billion for 13 months disrutions, based on temorarily and satially extended Inut-Outut (IO) model. 3

5 Certainly, to estimate the economic imact more accurately, it is imortant to construct better model able to examine the interindustry connections. Before introducing a new IO model addressing the interindustry connections reflecting behaviors of consumers, it would be necessary to address following issues that hel to understand the limited conditions of the IO analysis. Can regional secific direct losses resulted from a disaster extend to the national direct losses? Can the suggested economic model address regional and/or temoral (industrial) substitution effects? Do the disturbances from a disaster have forward or backward industrial linkage effects? That is, which model is more aroriate between suly- or demand- driven model? To observe the first issue, it is necessary to overview the economic effects of two hurricanes on the U.S. economy. Based on the GDP data, the economic situations of the U.S. are overviewed historically. As seen in Table 1, the real GDPs (constrained to 2000 dollars) of the U.S. are consistently increasing from 2002 to Similarly, all regions during the identical eriods show the same atterns as the U.S. However, while the real GDPs of Louisiana increased until 2004, the GDP of Louisiana between 2004 and 2005 decreased by 1.49 ercent. Also, we can observe that the GDP roortions which Louisiana takes out of the U.S. are small and decreasing from 1.3 ercent (on 2002) to 1.23 ercent (on 2005). However, the small GDP roortion of Louisiana is not alicable to the Mining sector. Second and third column in Table 2 show To 10 states which take high real GDP in the U.S. 4

6 for the Mining sector, where Mining industry including Oil industries. Three states, e.g. Texas, Louisiana, and New Mexico among the states in Gulf of Mexico (that is, Petroleum Administration for Defense District III) are ranked at the first, second, and seventh resectively, taking over 50 ercent of the U.S. Mining industry. Although Texas is the largest state of GDP for Mining industry, the GDP of Mining industry of Louisiana is decreased most severely in the U.S., showing 13.7 ercent decrease. Table 1. Changes of Real GDP (millions of chained 2000 dollars) and the changes of roortions to the U.S. of each region, Louisiana state, and the U.S.: Region and State GDP ($M) Percent GDP ($M) Percent GDP ($M) Percent GDP ($M) Percent NEW ENGLAND 568, % 581, % 605, % 619, % MIDEAST 1,851, % 1,887, % 1,961, % 2,017, % GREAT LAKES 1,553, % 1,592, % 1,624, % 1,645, % PLAINS 650, % 669, % 691, % 707, % SOUTHEAST 2,183, % 2,249, % 2,355, % 2,464, % SOUTHWEST 1,072, % 1,094, % 1,147, % 1,207, % ROCKY MOUNTAIN 320, % 327, % 344, % 362, % FAR WEST 1,781, % 1,834, % 1,931, % 2,016, % Louisiana 129, % 131, % 137, % 135, % U.S. 9,981, % 10,237, % 10,662, % 11,041, % Notes 1. Source: U.S. Bureau of Economic Analysis 2. Definition of Regions: a. NEW ENGLAND: Connecticut, Maine, Massachusetts, New Hamshire, Rhode Island, and Vermont b. MIDEAST: Delaware, District of Columbia, Maryland, New Jersey, New York, and Pennsylvania c. GREAT LAKES: Illinois, Indiana, Michigan, Ohio, and Wisconsin d. PLAINS: Iowa, Kansas, Minnesota, Missouri, Nebraska, North Dakota, and South Dakota e. SOUTHEAS: Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, Mississii, North Carolina, South Carolina, Tennessee, Virginia, and West Virginia f. SOUTHWEST: Arizona, New Mexico, Oklahoma, and Texas g. ROCKY MOUNTAIN: Colorado, Idaho, Montana, Utah, and Wyoming h. FAR WEST: Alaska, California, Hawaii, Nevada, Oregon, Washington 5

7 Total decreased value of GDP of the U.S. for the Mining sector is $2.8 billion. The Gulf of Mexico region takes over 74 ercent of the U.S. decrease of Mining sector. The decreases of the U.S., therefore, can be exlained by the natural disasters of the Hurricanes Katrina and Rita occurred sequentially in In the sense, the effect of the two Hurricanes on Oil industries are not constrained to the Gulf of Mexico, but extended to the national event (BEA, 2005). That is, the temorary losses of roduction and refinery caacity resulted from the Hurricanes Katrina and Rita made affection to the losses of U.S. caacity. Because Oil industries are critically connected to other industries, the roduction losses of the U.S. requests to examine the economic imacts extended to other relative industries via inter-industry relationshis (Park et al., 2006; Kim et al., 2007). Table 2. To 10 states taking high roortions out of the U.S. for the Mining sector: ) States Real GDP constrained to 2001 ($M.) 1) Proortions of Real GDP to the U.S. 2) Change in Real GDP by State: ) 2004) Difference Value ($M.) Percentage Texas % 38.11% 0.61% % Louisiana % 9.06% -1.17% % Oklahoma % 6.50% 0.63% % Alaska % 4.65% -0.39% % California % 4.50% -0.31% % Wyoming % 3.95% 0.07% % New Mexico % 3.94% -0.03% % Colorado % 3.92% 0.12% % West Virginia % 2.72% 0.08% % Kentucky % 2.18% 0.11% % U.S % % 0.00% % Notes 1) U.S. Bureau of Economic Analysis. 2) The author calculated. Although the Oil sector, which is a sub-sector of the Mining sector, reflects the changes better, the data for 2005 are not available until Jun., Also, as seen in Table 2, the real GDP of other states outside Gulf coasts are changed differently; Alaska and California are negatively changed, while Oklahoma is ositively changed 6

8 in real GDP between 2004 and To include all regional substitutions after the disasters to the analysis of economic imact, I used the national inut-outut (IO) model, which called USIO consisting of 47 USC industry sectors shown in Table A1 of Aendix (29 commodity sectors and 18 service sectors; see Park et al. (2007) for more detail on the sector descrition). Further, in a sense that industry lays a key role to suort national economy by roviding energy, the interindustrial economic imacts from the -refinery disrutions due to the disasters should be estimated with forward linked industrial connections, instead of backward connections. Most revious efforts have focused on constructing various demand-driven IO models because of their widely acceted usefulness in regional science. After Ghosh s suggestion of the suly-driven IO model, a debate over its lausibility ensued (Oosterhaven, 1988; 1989). Much of this was resolved with Dietzenbacher s (1997) suggestion of the interretation of a rice model, equivalent to Leontief s rice model. Although there has been a debate on the inoerability of suly-driven model, however Park (2007) shows that it is still aroriate to use the suly-driven IO model when analyzing indirect imacts from the direct losses of monetary value. In static market equilibrium, roducers will not change the current technical relationshis that are based on historical sales during the immediate eriod after an exogenous event. This reflects the fact that Ghosh s suly-driven model is in terms of monetarily exressed quantities and hence alicable when using the suly-side IO in the circumstance of static market equilibrium with abnormal economic cessations. Based on the static market equilibrium condition, Park (2006) and Park et al. (2006) tested the economic imacts on closures of major orts due to hyothetical dirty-bomb attacks or Hurricane Katrina. 7

9 Although the alications are still useful, even in the case that normal market equilibrium is not maintained, instead of the direct use of suly-side quantity model, Ghosh s (1958) case can be switched to a rice-tye suly-driven model, and lay a role in estimating economic imacts. To address this switching rocess, exogenous rice elasticities of demand are combined with the suly-driven model, adjusting quantity resonses to rice imacts. This study adoted the Park s (2007) logic underlying the theoretical background necessary to utilize the suly-side model. Common limitation in traditional IO models is over-estimate total imacts in a relatively long-term duration due to the linear characteristics of IO coefficients. Recently, Park (2007) addressed a new aroach to exanding the suly-driven IO model by combining rice elasticity of demand. The suggestion is useful to overcome the IO limitation, because the exanded model can reflect resilient effects conducted by consumers resulted from a disaster. Hence, this study adot a new suly-driven model using rice elasticity of demand only for refinery, that is, rice-elastistic suly-driven model, in order to analyze the national economic imacts of disruted -industry due to Hurricane Katrina and Rita. This rice-elastistic sulydriven model is not only a temoral exansion of suly-driven IO study, but also the first emirical alication of rice-tye suly-driven model for the U.S. Based on these conditions and limitations, a new model is suggested in the next section. 8

10 III. Model and Data III-1. Model The most rigorous limitation in IO models is a fixed coefficients characteristic which rovides over-estimate total imacts in a long-term. The fixed coefficients assumtion ignores the reaction of market system via flexible behaviors of suliers and consumers. Of course, market ower can be maintained during some short-term if behaviors in market exect a disaster would exist shortly, and hence, the fixed coefficients assumtion is still useful. However, if a disaster were exected to exist in long-run eriod, market equilibrium would adjust following the market signal, that is, rice. In the case, the direct use of Inut-Outut model might overestimate the interindustrial effects (Park, 2007). Figure 1. Oil rice changes due to Oil-refinery roduction losses, following fixed (average) rice elasticity. Oil rice Demand Curve q 0 q Oil refinery roduction 9

11 As shown in Figure 1, the losses of refinery roduction Δq( = q 1 q 0 ) due to Hurricane Katrina and Rita would increase rice, following the demand curve. Then, new market equilibrium is temorarily decided at the new rice 1, where consumers demands are reflected. Introducing exogenous rice elasticity of demand ( ε ) will make demand curve linear, and hence, the changes of monetary quantity and rice on demand curve will convert the traditional quantity-tye demand-driven models in equation (1.) and (2.) to a rice-tye suly-driven model as shown in equations (3.) and (4.). X = AX + Y (1.1) s T s T 1 = (I A) Y (1.2) where, s X denotes the monetary value row vector of total inuts (outlays) for each sector and its elements are exressed as s x j, which is the row sum of intermediate flows and value added factors of sector j. In the market s d equilibrium, the transose of X ( X ) equals to total oututs vector, X. s T Y includes various kinds of final demands for industry sector i. The element y ik denotes the deliveries in dollar values from industry sector i to final users k. Generally, k contains rivate consumers, governments, investments, and exorts. A is the technical coefficient matrix of intermediate interindustry flows. The element aij denotes the roortional deliveries in dollar s worth from industry sector i to j. 10

12 Suerscrit T denotes the transose of the matrix. Hence, the total imacts according to the direct losses of final demands for the industry would be obtained via, ΔX = (I A) 1 ΔY (2.) s T From the equation (2.), rice-tye inut-outut model are obtained as, s ΔX T s s 1 1 = T T [I Xˆ B(Xˆ ) ] ΔY (3.1) s 1 s 1 = Xˆ T (I B) (X T ˆ ) ΔY (3.2) s where, = T s T 1 T Xˆ B(Xˆ ), because X s d A = X. Hence, ˆ s T 1 s ) ΔX T 1 s 1 = T (I B) (Xˆ ) ΔY, (4.1) (X and ΔP = 1 ΔP Y (I B) (4.2) s T 11

13 δq / q Because the rice elasticity of demand for industry sector, ε, is defined as, δ / s where q = x, the rice change δ is obtained based on the exogenous rice elasticity of demand ε as, δq / q δ = (5.1) ε / δq = ε q (5.2) = δq π (5.3) where = π is exogenous for industry sector. It is relatively easier to find ε q the fixed and q right before the event than to find the exogenous rice elasticity of demand. As Dietzenbacher (1997) noticed, in the case of rice increases, there is no reason to slit where the rice increases of refinery roduct, e.g. from final demand sectors or intermediate roducts, because final users should ay for all increased rices in any cases. Therefore, the rice increases of refinery roducts would be assumed as rice increases of final users. Therefore, based on the equations (4.) and (5.), the vector of derived total (relative) rice 12

14 changes Δ P ~ s T are obtained as, Δ ~ 1 = (I B) ΔQ Π (6.) s P T where, ΔQ is a vector only including industry quantity losses, δq, for the USC sectors, and Π is a exogenous vector involving valid π. III-2. Data Figure 2 shows total -refinery roducts attern for the U.S. by month for 29 months between Aril 2004 and Setember The simle comarison for the identical months between the Window 04 and Window 05 shows that the former window doesn t include the serious disrutions due to a hurricane while the latter window involves serious quantity losses due to Hurricanes Katrina and Rita. Since 2006, the U.S. refinery roduction certainly seems to be recovered to the status right before the disasters occurred. Based on the figures, I assume that the U.S. -refinery roducts had disrutions only for last four months of Table 3 shows the summary of inut data necessary to estimate the total outut vector s Δ P ~ T. First, I used searate the -refinery roducts into five tyes: Finished Motor Gasoline, Kerosene-Tye Jet Fuel, Distillate Fuel Oil, Residual Fuel Oil, and Proane (Commercial use), in order to calculate the amount of q. To estimate rice decreases of industry (USC sector 10) later for consumtion tyes, I classified those -refinery tyes combining with consumtion sector, based on the etroleum consumtion note available at the htt:// From the classification, I 13

15 distribute Distillate and Residual Fuel Oil to four-consumtion tyes; residential, commercial, industrial and transortation, based on 2003 roortions from Adjusted Sales of Fuel Oil by End Use available at the htt://tonto.eia.doe.gov/dnav/et/et_cons_to.as. Therefore, the vector q by consumtion sector and -refinery tye is suggested in the third column of Table 3. Second, the fourth column vector is calculated as average rices using the monthly rices between January through July of 2005, which exclude the Hurricanes effects. Those rices only deend on -refinery tye. Third, I obtained rice elasticities of demand, ε, by consumtion tye from the htt:// Then, using the definition of π suggested in equations (5.), I calculated the column vector of π. Fourth, I forecasted the δq with Holter-Winters method for last four months of 2005 for each -refinery roduction tye. The statistical results are shown in Table A2 of Aendix. The results are statistically accetable, because Theil s U is close to 0 (Maddala, 1977). After then, I distribute the direct quantity losses with the same roortions as used in calculating the. q Finally, based on the equation (5.3), the vector δ is estimated. Also, the weighted δ s, which are multilied by the weight obtained from the roortions of the δq to the sum of δq, are suggested in the last column. The sum of the element in column vector of δ would be total rice increase due to -refinery disrutions, reflecting economic behaviors in the U.S. market. Therefore, the total rice-tye direct losses on the U.S. market due to the two Hurricanes would be $ er barrel, that is, $1.23 er gallon. 14

16 Figure 2. Total Oil Refinery Production for the U.S. by Month: Aril 2004 through Setember 2006 Total 1000B total Window 04 Window Ar- 04 May- 04 Jun- 04 Jul- 04 Aug- 04 Se- 04 Oct- 04 Nov- 04 Dec- 04 Jan- 05 Feb- 05 Mar- 05 Ar- 05 May- 05 Jun- 05 Jul- 05 Aug- 05 Se- 05 Oct- 05 Nov- 05 Dec- 05 Jan- 06 Feb- 06 Mar- 06 Ar- 06 May- 06 Jun- 06 Jul- 06 Aug- Se Date Source: Author calculations of raw data from EIA official website (htt://tonto.eia.doe.gov/dnav/et/et_n_wiu_dcu_r30_w.htm) 15

17 Table 3. Data Summary Necessary to Estimate Total Outut Vector Consumtion Oil-Refinery q ε Sector δq Product Tye π W 2) δ W* δ Finished Motor Gasoline Kerosene-Tye Jet Fuel Transortation Distillate Fuel Oil Residual Fuel Oil Residential Distillate Fuel Oil Distillate Fuel Oil Commercial Proane Residual Fuel Oil Distillate Industrial Fuel Oil Residual Fuel Oil TOTAL Unit Barrel(B) 1) $/B 1000B $/B $/B Notes 1) The classifications between consumtion sector and -refinery roduct tye are available at htt:// and 2003 ercentages of Adjusted Sales of Fuel Oil by End Use from htt://tonto.eia.doe.gov/dnav/et/et_cons_to.as were used to distribute Distillate and Residual Fuel Oil to the consumtion sectors. 2) W c = δq c / δq c, where subscrit c denotes Consumtion Sector. c IV. Results Based on 2001 U.S. IMPLAN data set, I created suly-side version of the U.S. national Inut Outut model, which consists of 47 industry sectors. Table 4 shows the total rice increases of each industry sector. The disrution in refinery affects to other relative non-service industries, e.g. Nonmetallic minerals (USC sector 8), Fertilizers (USC sector 13), and Basic chemicals (USC sector 11), increasing additional rices for those sectors. Also, the disrution increases the rices of service industries, such as Utility (USC sector 30), Transortation (USC sector 33), Real estate and rental and Leasing (USC sector 38), and Management of comanies and enterrises industries (USC sector 40) more than other service industries. 16

18 Table 4. Total Imacts via Price-tye Suly-driven USIO Model USC Sector Total Price Increase Proortion Total Imact ($M.) USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % USC % TOTAL %

19 Although the rice-tye model shows the rice increases for each sector, our concern might be more related to understanding how much the economic imacts would be from the increased rices. Because total quantity losses during the last four months of 2005 are 140 Million Barrels, total value losses including the increased rice are easily obtained, only by multilying the losses of quantity and the increased rice. The amounts are $721.5 million for four months. Based on the direct losses, total economic losses via the USIO are $1.537 billion. For the Oil industry (USC sector 10) lose $1 billion totally and other industries in the U.S. occurred $537 million losses due to the two Hurricanes. To check the reasonability of this result, I comared the total losses simly with revious research. It is very interesting; this study shows that the simle total economic losses er month would be $384 million, while Kim et al. (2007) shows that the total losses er month would be $373 million. Although the two studies used the similar data source, this study is conducted based on the national level, while they used only PADD III data. Further, the rice-tye suly driven USIO is comletely different from their temorally and satially extended IO model. Therefore, the result in this study can be accetable, although this study is conducted at the different basis from the revious study. V. Conclusions Inut-Outut models are attractive because they can be made oerational and accessible at low cost. However, the widely used inut-outut aroaches routinely include the caveat about the fixed technologies assumtion and it has been imortant issues to overcome the overstated results (Park et al, 2007). I alied a new model reflecting the economic behaviors in the market, 18

20 which is recently suggested by Park (2007), to the refinery disrutions due to the Hurricanes Katrina and Rita. Recent Hurricane Katrina and Rita caused severe economic cessations, and led to further economic losses via interindustrial relations. This study is conducted based that industry lays a key role to suort national economy, and hence the estimation of the economic imacts of the disasters should be based on the suly-driven inut-outut (IO) model. The new rice-tye suly-driven model using rice elasticity of demand for refinery showed that total rice increased u to 11 dollars nationally based on 5.15 dollars direct losses. The total economic losses would be $1.54 billion during four months. This rice-elastistic suly-driven model is a temoral exansion of suly-driven IO study, and showed a reasonable result comared to the revious research. However, this aroach didn t touch the fixed coefficient assumtions; only reflected the economic behaviors to adjust the direct losses for the IO inuts. Therefore, further researches include the satial and temoral extensions of the rice-tye suly-driven IO model. 19

21 References BEA (2005) available at the htt:// Dietzenbacher, E. (1997) In Vindication of the Ghosh Model: A Reinterretation as a Price Model, Journal of Regional Science, Vol.37: Ghosh, A. (1958) Inut-Outut Aroach to an Allocative System, Economica, Vol. XXV: Holtz-Eakin, H.D.J. (2005) Statement in AFTER THE HURRICANES: IMPACT ON THE FISCAL YEAR 2007 BUDGET, available at htt:// Kent, Joshua D. (2006) 2005 Louisiana Hurricane Imact Atlas Volume 1, Louisiana Geograhic Information Center. Available at htt://lagic.lsu.edu/lgisc/ublications/2005/lgisc-pub _2005_HURRICANE_ATLAS.df Maddala, G. S. (1977) Econometrics, New York: McGraw-Hill National Hurricane Center of National Weather Service (2007) Troical Weather Summary 2005 Web Final, Available at htt:// Oosterhaven, J. (1988) On the Plausibility of the Suly-driven Inut-Outut Model, Journal of Regional Science, Vol.28: Oosterhaven, J. (1989) The Suly-driven Inut-Outut Model: A New Interretation but Still Imlausible, Journal of Regional Science, Vol.29: Park, J.Y. (2006) The Economic Imacts of a Dirty-Bomb Attack on the Los Angeles and Long Beach Port: Alying Suly-driven NIEMO, Paer resented at 17 th Annual Meeting of the Association of Collegiate Schools of Planning, Fort Worth, TX, USA, November

22 Park, J.Y. (2007) The Suly-Driven Inut-Outut Model: A Reinterretation and Extension, resented at 46 th Annual Meeting of the Western Regional Science Association, Newort Beach, CA, USA, February Park, J.Y., P. Gordon, S.J. Kim, Y.K. Kim, J.E. Moore II, and H.W. Richardson, (2006) Estimating the State-by-State Economic Imacts of Hurricane Katrina, Paer resented at CREATE symosium: Economic and Risk Assessment of Hurricane Katrina, University of Southern California, California, USA, August Park, J.Y., P. Gordon, J. E. Moore II, and H. W. Richardson (2007) Simulating The State-by- State Effects of Terrorist Attacks on Three Major U.S. Ports: Alying NIEMO (National Interstate Economic Model), in H.W. Richardson, P. Gordon and J.E. Moore II, eds., The Economic Costs and Consequences of Terrorism. Cheltenham: Edward Elgar. 21

23 Aendix Table A1. Definitions of USC sectors Classification USC Descrition USC01 Live animals and live fish & Meat, fish, seafood, and their rearations USC02 Cereal grains & Other agricultural roducts excet for Animal Feed USC03 Animal feed and roducts of animal origin, n.e.c. USC04 Milled grain roducts and rearations, and bakery roducts USC05 Other reared foodstuffs and fats and s USC06 Alcoholic beverages USC07 Tobacco roducts USC08 Nonmetallic minerals (Monumental or building stone, Natural sands, Gravel and crushed stone, n.e.c.) USC09 Metallic ores and concentrates USC10 Coal and etroleum roducts (Coal and Fuel s, n.e.c.) USC11 Basic chemicals USC12 Pharmaceutical roducts USC13 Fertilizers USC14 Chemical roducts and rearations, n.e.c. Commodity Sectors USC15 Plastics and rubber USC16 Logs and other wood in the rough & Wood roducts USC17 Pul, newsrint, aer, and aerboard & Paer or aerboard articles USC18 Printed roducts USC19 Textiles, leather, and articles of textiles or leather USC20 Nonmetallic mineral roducts USC21 Base metal in rimary or semi-finished forms and in finished basic shaes USC22 Articles of base metal USC23 Machinery USC24 Electronic and other electrical equiment and comonents, and office equiment USC25 Motorized and other vehicles (including arts) USC26 Transortation equiment, n.e.c. USC27 Precision instruments and aaratus USC28 Furniture, mattresses and mattress suorts, lams, lighting fittings, and illuminated signs USC29 Miscellaneous manufactured roducts, Scra, Mixed freight, and Commodity unknown USC30 Utility USC31 Construction USC32 Wholesale Trade USC33 Transortation USC34 Postal and Warehousing USC35 Retail Trade USC36 Broadcasting and information services USC37 Finance and Insurance Non-Commodity USC38 Real estate and rental and leasing (Service) Sectors USC39 Professional, Scientific, and Technical services USC40 Management of comanies and enterrises USC41 Administrative suort and waste management USC42 Education Services USC43 Health Care and Social Assistances USC44 Arts, Entertainment, and Recreation USC45 Accommodation and Food services USC46 Public administration USC47 Other services excet ublic administration 22

24 Table A2. Statistical Results for Holt-Winters Aroach to Forecasting Normal Status of Petroleum Products. Variables Finished Motor Gasoline Kerosene- Tye Jet Fuel Distillate Fuel Oil Residual Fuel Oil Proane WEIGHT Weights WEIGHT Seasonal Parameters WEIGHT Jan Feb Mar Ar May Jun Jul Aug Se Oct Nov Dec Total sum of squares (SST) Sum of the squared residuals (SSE) R-Squares (= 1-SSE/SST) THEILU