CREATE Research Archve Publshed Artcles & Papers 10-31-007 Defensve Counterterrorsm Measures and Domestc Poltcs Todd Sandler Unversty of Texas at Dallas, tsandler@utdallas.edu Kevn Squera Unversty of Texas at Dallas, squera@utdallas.edu Follow ths and addtonal works at: http://research.create.usc.edu/publshed_papers Recommended Ctaton Sandler, Todd and Squera, Kevn, "Defensve Counterterrorsm Measures and Domestc Poltcs" (007). Publshed Artcles & Papers. Paper 64. http://research.create.usc.edu/publshed_papers/64 Ths Artcle s brought to you for free and open access by CREATE Research Archve. It has been accepted for ncluson n Publshed Artcles & Papers by an authorzed admnstrator of CREATE Research Archve. For more nformaton, please contact grbben@usc.edu.
Ths artcle was downloaded by: [Unversty of Southern Calforna] On: 9 Aprl 009 Access detals: Access Detals: [subscrpton number 906869076] Publsher Routledge Informa Ltd Regstered n England and Wales Regstered Number: 107954 Regstered offce: Mortmer House, 37-41 Mortmer Street, London W1T 3JH, UK Defence and Peace Economcs Publcaton detals, ncludng nstructons for authors and subscrpton nformaton: http://www.nformaworld.com/smpp/ttle~content=t713640174 DEFENSIVE COUNTERTERRORISM MEASURES AND DOMESTIC POLITICS Kevn Squera a ; Todd Sandler a a School of Economc, Poltcal and Polcy Scences, Unversty of Texas at Dallas, Rchardson, TX 75080-301, USA Onlne Publcaton Date: 01 December 008 To cte ths Artcle Squera, Kevn and Sandler, Todd(008)'DEFENSIVE COUNTERTERRORISM MEASURES AND DOMESTIC POLITICS',Defence and Peace Economcs,19:6,405 413 To lnk to ths Artcle: DOI: 10.1080/104690701775483 URL: http://dx.do.org/10.1080/104690701775483 PLEASE SCROLL DOWN FOR ARTICLE Full terms and condtons of use: http://www.nformaworld.com/terms-and-condtons-of-access.pdf Ths artcle may be used for research, teachng and prvate study purposes. Any substantal or systematc reproducton, re-dstrbuton, re-sellng, loan or sub-lcensng, systematc supply or dstrbuton n any form to anyone s expressly forbdden. The publsher does not gve any warranty express or mpled or make any representaton that the contents wll be complete or accurate or up to date. The accuracy of any nstructons, formulae and drug doses should be ndependently verfed wth prmary sources. The publsher shall not be lable for any loss, actons, clams, proceedngs, demand or costs or damages whatsoever or howsoever caused arsng drectly or ndrectly n connecton wth or arsng out of the use of ths materal.
Defence and Peace Economcs, 008 Vol. 19(6), December, pp. 405 413 DEFENSIVE COUNTERTERRORISM MEASURES AND DOMESTIC POLITICS KEVIN SIQUEIRA AND TODD SANDLER Downloaded By: [Unversty of Southern Calforna] At: 0:43 9 Aprl 009 Taylor GDPE_A_77547.sgm and Francs Ltd School of Economc, Poltcal and Polcy Scences, Unversty of Texas at Dallas, GR31, 800 W Campbell Road, Rchardson, TX 75080-301, USA (Receved 31 July 007; n fnal form 31 October 007) 10.1080/104690701775483 Defence 104-694 Orgnal Taylor 000 008 ToddSandler tsandler@utdallas.edu & Artcle Francs (prnt)/1476-867 Peace Economcs(onlne) Unlke most of the lterature, ths paper ncludes domestc poltcal consderatons n whch two countres must decde defensve countermeasures aganst a common terrorst threat. A delegaton problem arses as voters strategcally choose a polcymaker whose preferences dffer from ther own. As a consequence, countres lmt the presumed oversupply of defensve countermeasures. Thus, the ncluson of domestc poltcs gves a new perspectve on counterterrorsm. The tmng of electons s also shown to make a dfference. Keywords: Terrorsm; Externaltes; Counterterrorsm; Delegaton problem; Domestc poltcs JEL Codes: H41, D7 INTRODUCTION Snce the unprecedented terrorst attacks on September 11, 001, economsts and poltcal scentsts have ncreasngly turned ther attenton to theoretcal and emprcal analyses of transnatonal terrorsm (see, for example, Drakos and Kutan, 003; Enders, 007; Fara and Arce, 005; Heal and Kunreuther, 005; Kunreuther and Heal, 003). To date, lttle attenton has been pad to how domestc poltcs (e.g. voters electon of polcymakers) can mpact polcy decsons, such as defensve efforts to harden targets. When two or more countres ndependently confront a common terrorst threat or network, countres are portrayed as oversupplyng defensve measures compared to an optmum level (e.g. Arce and Sandler, 005; Sandler and Squera, 006). These measures not only reduce the probablty that a country wll be attacked at home, but they also curb the amount of damages n the event of an attack. Defensve actons are costly to the provder and to other potental target countres. The latter follows because hardenng targets at home ncreases the probablty of an attack on relatvely softer targets abroad. It s ths transference that may nduce targeted countres nto a compettve and wasteful fortfcaton race, especally f such defensve actons do not greatly reduce the terrorsts proclvty for attackng. Such compettve defensve efforts may be attenuated somewhat as the assocated negatve (transference) externalty s partly nternalzed f a naton s nterest abroad s at rsk. Correspondng author. E-mal: tsandler@utdallas.edu ISSN 104-694 prnt: ISSN 1476-867 onlne 008 Taylor & Francs DOI: 10.1080/104690701775483
406 K. SIQUEIRA AND T. SANDLER The purpose of ths paper s to show that voters strategc actons to choose a polcymaker may lessen excessve defensve spendng as voters partly nternalze the transference externalty through delegaton. In partcular, the medan voter elects a polcymaker who places less weght than the voter on expected damages from a terrorst attack. Wth strategc delegaton (see, for example, Persson and Tabelln, 199), countres have ncentves to commt to lower defensve countermeasures than n the absence of such delegaton. Such delegaton-nduced strategzng by mutually targeted countres can mprove ther well-beng as defense s allocated more effcently. THE MODEL Downloaded By: [Unversty of Southern Calforna] At: 0:43 9 Aprl 009 We assume that ust two countres are n the terrorsts cross hars and we represent these countres defensve decson wth a two-stage game. In the frst stage, voters n each targeted country elect a polcymaker, whle accountng for the mpact that ths electon wll have durng the second stage, when defensve measures are determned. Based on ther preferences, polcymakers n the two countres smultaneously and ndependently choose protecton levels n the second stage. Terrorst behavor s captured n reduced form by a probablty of attack functon that reflects attack transference between targeted countres. Let θ denote country s defensve efforts to reduce ts vulnerablty to terrorsm, and let π (θ,θ ) represent the probablty of a terrorst attack n country. Ths probablty functon ndcates that s defensve measures lmt the lkelhood of a home attack wth dmnshng returns to effort: π / < 0 and π/ θ > 0. Addtonally, t also assumes that greater defense n country rases the lkelhood of an attack n, but at a decreasng rate: π / > 0 and π/ θ < 0. Greater defensve acton by one country s assumed to ncrease the margnal effectveness of the other country s defense, so that π / < 0. Ths s also consstent wth reducng the margnal transference of an attack ( π / ) that stems from the other target s defensve efforts. Fnally, π (θ,θ )=1 π (θ,θ ), so that the probablty of an attack n equals the probablty of no attack n, as the terrorsts go for the softer target or flp a con when both countres are equally fortfed. The terrorsts are, however, determned to attack one of the countres, whch s consstent wth al-qada afflated groups that are not deterred by target hardenng (US News & World Reports, 007). The Polcymaker The elected polcymaker, g, n country, s assumed to mnmze hs or her antcpated taste-weghted damages and defense costs as denoted by: g g Z = α { π( θ, θ) D( θ) +[ 1 π( θ, θ) ] d}+ C( θ) ( 1) In equaton (1), D(θ ) and d denote the assocated damages to country and ts nterests when ether attacked at home or abroad, respectvely. D(θ ) s assumed to be decreasng and lnear n θ, whle d s constant wth respect to s defense actons at home. For all θ, home attacks are more damagng to s nterests than attacks abroad so that D > d. C(θ ) represents country s strctly ncreasng and convex cost of defensve efforts. Also, let α g depct the weght that s polcymaker places on the expected damages from a terrorst attack. Ths weght vares contnuously over the unt nterval, so that voters can be characterzed as havng a contnuous
COUNTERTERRORISM MEASURES 407 choce over competng polcymakers wth dfferent abhorrence for terrorst-nduced damages. Analogous obectve functons characterze the agents n country. The polcymaker chooses defensve efforts to mnmze equaton (1) subect to the constancy of defensve efforts n country, θ. Ths gves the followng frst-order condton (FOC): α g π g ( D d) + α π θ θ θ D'( ) + C'( ) = 0 ( ) Downloaded By: [Unversty of Southern Calforna] At: 0:43 9 Aprl 009 In equaton (), the two left-hand terms are negatve and ndcate the margnal benefts assocated wth defensve efforts, stemmng from transferrng the attack abroad and lmtng damages from attacks at home, respectvely. These margnal benefts are set equal to margnal defense costs. Gven our assumptons, the second-order condton, Z g / θ s satsfed. 1 > 0, Our assumptons are also suffcent for each country s best-response functon (BR) to be upward slopng (see equaton (3)), wth θ and θ on the vertcal and horzontal axes, respectvely: θ θ = α g π g π ( D d) α D'( θ ) θ g > 0 () 3 Z θ Ths postve slope ndcates that defensve measures are strategc complements. We also have that country s defensve efforts ncrease wth greater weght gven to terrorst damages: Utlzng the second-stage FOCs for both countres, we can characterze the comparatve-statc change to Nash equlbrum levels (denoted by *) of the countres defensve measures n response to a greater weght gven by to damages from terrorsm attacks. Because defensve actons are strategc complements, country s enhanced concern for damages augments defensve measures n both countres through a rghtward shft n s best-response curve and a movement along s best-response curve. The results are gven by the followng system of equatons, summarzed n matrx form: π ( D d) πd'( θ) g = g > 0 ( 4) α Z where: é ë A A * é dq ù ù da d û q g ë a û A A d g éb ù * = ( 5) 0 ë û 1 The second-order condton s g Z g π = g π α ( D d) + α D'( θ + C > 0 ) "( θ).
408 K. SIQUEIRA AND T. SANDLER and g π g π A α ( D d) + α D'( θ) + C"( θ) > 0 g π g π A α ( D d) + α D'( θ ) < 0 π g π g A α ( D d) + α D'( θ ) < 0 π g π g A α D d + D + C ( ) α '( θ ) ( θ) > 0 Downloaded By: [Unversty of Southern Calforna] At: 0:43 9 Aprl 009 Usng Cramer s rule, we obtan: dθ dα g BA = AA AA B = π ( D d) πd ( θ) > 0 * * dθ BA > 0, and g = > 0 ( 6) dα AA AA where the denomnators are assumed to be postve to ensure the statc stablty of the equlbrum (see Appendx). Graphcal Representaton Fgure 1 dsplays the polcymakers equlbrum n the second stage. Polcymaker s choce of defensve measure, θ, s on the vertcal axs, whle polcymaker s choce of defensve measure, θ, s on the horzontal axs. The two thck best-response paths, BR and BR, represent the respectve polcymaker s choce of hs or her θ for alternatve levels of the other polcymaker s defensve measure. The second-stage Nash equlbrum s at N 1, where the two paths (assumed lnear for smplcty) ntersect. The relatve slopes of the best-response paths ensure a stable equlbrum. If, say, polcymaker had more nterests n country (.e. foregn nvestments) such that d ncreases, then BR would shft left and both polcymakers would reduce ther defensve measures wth polcymaker engagng n the greater reducton. FIGURE 1 Polcymakers second-stage equlbrum Next, we focus on the thnner BR path that follows from an ncrease n polcymaker s concern for expected damages from a terrorst attack at home see equaton (4). As α g ncreases, BR shfts to the rght as shown and stage-two equlbrum levels of defensve actons ncrease n both countres see N wth polcymaker expendng the greater ncrease n defensve measures. These changes n the equlbrum levels of the θs correspond to equaton (6). The Medan Voter We now turn to the frst stage where each country s voters elect a polcymaker to whom they delegate authorty for defensve counterterrorsm measures n stage two. We thus assume that the voters n each country effectvely solve ther collectve acton problem wth respect to ther
COUNTERTERRORISM MEASURES 409 BR ' BR ** N BR * N 1 Downloaded By: [Unversty of Southern Calforna] At: 0:43 9 Aprl 009 0 FIGURE 1 * ** Polcymakers second-stage equlbrum country s provson of natonal publc goods through votng. Under maorty rule and other standard assumptons, the medan voter proves to be decsve n electng the polcymaker who, voters antcpate, wll act n hs or her self-nterest at the second stage of the game. These voters, n effect, behave strategcally by takng nto account the defensve race that may ensue between the two countres polcymakers. As such, the voters act as a leader wth ther respectve polcymaker actng as a follower. Snce voters preferences dffer by the sngle-dmensonal weghts that each voter places on expected damages, the medan voter theorem apples when preferences are strctly convex n α g. Assumng such preferences, we represent the choce of the electorate as that of the medan voter, who chooses α g to mnmze: The assocated FOC s: = [ ( ) ( )+ ( ( )) ]+ ( ) m m * * * * * * Z α π θ, θ D θ 1 π θ, θ d C θ ( 7) é m p a ë q * * ù q p q m d é m ù d - + a p q + q g a D d g û d + - a q da ë û ( D d) D'( ) C'( ) ( ) = 0 ( 8)
410 K. SIQUEIRA AND T. SANDLER Utlzng the FOC for s polcymaker from equaton () n equaton (8) and rearrangng slghtly, we obtan the followng expresson: é m g p a a ë q ù dq * é p dq * m ù g + a D - d g û da q da ë û ( ) ( D - d) + pd'( q) ( ) = 0 ( 9) Downloaded By: [Unversty of Southern Calforna] At: 0:43 9 Aprl 009 for whch t follows that α m > α g, based on the sgns of the terms n equaton (9). Ths last nequalty means that the medan voter for each country strategcally elects a polcymaker who puts less weght than the voter on expected damages. Ths then lmts defensve spendng. Consequently, delegaton and strategc votng may mtgate a tendency to overspend on hardenng targets and, thus, make both countres better off. Ths result can be best understood by descrbng the above equaton not only as the trade-off that the medan voter faces n terms of benefts and costs, but also n terms of the non-strategc and strategc effects of delegaton respectvely, the frst and second terms of equaton (9). The non-strategc benefts are host-country based, reflectng the reduced damages and smaller probablty of attack derved from shorng up home defenses. In contrast, the strategc effect represents costs for s voters, because ther selecton of a polcymaker who places greater weght on expected damages results n ncreased defensve spendng by the other country and, thus, an augmented lkelhood that s attacked, π / > 0. When, therefore, the medan voter delegates defensve decsons to someone who puts less emphass on such actons, the voter uses foresght and delegaton to lessen defensve spendng both at home and abroad. The effects of delegaton are dsplayed n Fgure, where E ndcates the equlbrum n the absence of delegaton when α g = α m. In the presence of delegaton wth α m > α g, BR shfts leftward to BR and BR shfts downward to BR, so that the delegaton equlbrum at E mples much reduced levels of defensve measures n both countres, compared wth E 1. The mpled leader-follower behavor of delegaton causes voters partly to nternalze the transference externalty when multple countres are targeted by the same transnatonal terrorst group. In so dong, they vote n polcymakers who downplay defensve measures to lmt compettve defensve races. Such actons by voters n both targeted countres may mprove welfare and resource allocaton. FIGURE Effects of delegaton n frst stage EXTENSIONS Next, we ask what would happen f electons n the two countres are not smultaneous. Wll the sequental tmng of the electons undermne a country s ncentves to delegate, thereby alterng ts defensve measures? Recall that the terrorsts are equally lkely to ht ether target country f θ = θ. Suppose then that both countres are alke n all aspects except for the tmng of electons: country holds electons n the frst stage, whle country holds electons n the second stage. Startng n the second stage, t can be easly shown that voters n country wll elect a polcymaker wth the same weght on damages as that of the medan voter. Insofar as country holds electons after those of country, there s no strategc advantage to lmtng ts defensve measures. In terms of our earler dscusson, the strategc effect from delegaton for country vanshes. However, n the frst stage, the electorate n country stll has an ncentve to elect a polcymaker who lmts spendng. Unfortunately, ths can work aganst country as t draws a hgher probablty of attack, gven the asymmetry of the resultng equlbrum. Thus, electon tmng n targeted countres can determne whch country faces the greater lkelhood of sufferng from a terrorst attack.
COUNTERTERRORISM MEASURES 411 G BR ' BR BR E 1 BR ' Downloaded By: [Unversty of Southern Calforna] At: 0:43 9 Aprl 009 0 E FIGURE Effects of delegaton n frst stage The above dscusson may also provde clues to the possble mplcatons of what mght happen f voters choose polcymakers usng consderatons other than, or n addton to, strategc ones. Suppose, for example, that the voters choce of polcymaker s also partly expressve, meanng that voters also gan from expressng support toward a canddate who supports a partcular poston (see, for example, Brennan and Hamln, 1999). To allow for ths possblty n the present context, assume that t s more costly for voters to express support for polcymakers who are dfferent from them. In other words, voters fnd t less costly to elect polcymakers who are more lke them n terms of the weght that they place on expected damages from terrorsm. In ths scenaro, voters concerns may tend to offset the strategc effects of votng n the frst stage of the game and, thus, dmnsh the moderatng effects of delegaton on defensve spendng durng the second stage of the game. That s, when confronted wth a common terrorst threat, the source of voters motvaton can also make a dfference. G CONCLUDING REMARKS In the context of a transnatonal terrorsm threat, we have shown that domestc poltcal consderatons can lessen over-deterrence when voters strategcally elect a polcymaker to account for attack transference. If, however, a country confronts ust domestc terrorsm, then
41 K. SIQUEIRA AND T. SANDLER over-deterrence s not a concern (Sandler, 005) and there s no reason for the electorate to act strategcally, snce hardenng targets at home are not transferrng attacks abroad. ACKNOWLEDGMENTS Ths research was partally supported by the US Department of Homeland Securty (DHS) through the Center for Rsk and Economc Analyss of Terrorsm Events (CREATE) at the Unversty of Southern Calforna, grant number 007-ST-061-000001. However, any opnons, fndngs, and conclusons or recommendatons are solely those of the authors and do not necessarly reflect the vews of DHS. References Downloaded By: [Unversty of Southern Calforna] At: 0:43 9 Aprl 009 Arce, D.G. and Sandler, T. (005) Counterterrorsm: a game-theoretc analyss. Journal of Conflct Resoluton 49() 183 00. Brennan, G. and Hamln, A. (1999) On poltcal representaton. Brtsh Journal of Poltcal Scence 9(1) 109 17. Cornes, R. and Sandler, T. (1984) Easy rders, ont producton, and publc goods. Economc Journal 94(3) 580 598. Drakos, K. and Kutan, A.M. (003) Regonal effects of terrorsm on toursm n three Medterranean Countres. Journal of Conflct Resoluton 47(5) 61 641. Enders, W. (007) Terrorsm: an emprcal analyss. In Handbook of Defense Economcs, vol.., edted by T. Sandler and K. Hartley. Amsterdam: North-Holland, 816 866. Fara, J.R. and Arce, D.G. (005) Terror support and recrutment. Defence and Peace Economcs 16(4) 63 73. Heal, G. and Kunreuther, H. (005) IDS models of arlne securty. Journal of Conflct Resoluton 49() 01 17. Kunreuther, H. and Heal, G. (003) Interdependent securty. Journal of Rsk and Uncertanty 6(-3) 31 49. Persson, T. and Tabelln, G. (199) The poltcs of 199: fscal polcy and European ntegraton. Revew of Economc Studes 59(4) 689 701. Sandler, T. (005) Collectve versus unlateral responses to terrorsm. Publc Choce 14(1-) 75 93. Sandler, T. and Squera, K. (006) Global terrorsm: Deterrence versus pre-empton. Canadan Journal of Economcs 39(4) 1370 1387. US News & World Reports (007) Al Qaeda regroups, News desk, July 16, avalable at http://www.usnews.com/ blogs/news-desk/007/7/1/al-qaeda-regroups.html, downloaded 18 July 007. APPENDIX Followng Cornes and Sandler (1984) and others, our noton of statc stablty s based on the smple adustment process: dθ dt [ ] = = s θ ( θ ) θ ( t) for, 1, and ( A1) where the constant parameters s 1 and s are the postve speeds of adustment, θ (θ ) s country s best-response functon, and θ (t) denotes the actual value of defensve measures at tme t. Takng a lnear (Taylor seres) approxmaton of the system (A1) around the neghborhood of the equlbrum, we get: é ë dq dt dq dt 1 ù é = û ë s 1 1 q1 ù -s s * q éq q 1 - ù 1 q q q -s ë - û q1 û * (A) where / = A /A, for, =1, and, s the slope of each country s best-response functon. Statc stablty then requres that the determnant of the matrx be negatve defnte. Gven the postve speeds of adustment, ths requrement reduces to:
COUNTERTERRORISM MEASURES 413 1 1 ss 1 θ > 0 1 (A3) or 1 1 > θ 1 from whch we get our assumed condton, requred for stablty n the text, namely that A A A A > 0. Ths requrement on the relatve slopes of the country s best-response paths s ncorporated n Fgures 1 and where the equlbrums dsplay statc stablty, snce: Downloaded By: [Unversty of Southern Calforna] At: 0:43 9 Aprl 009 < θ In equaton (A4), the left-hand expresson s the slope of country s best-response path and the rght-hand expresson s the nverse of the slope of country s best-response path. 1 (A4)