12 Preemptive and Preventive Measures: DP Approach

Transcription

1 12 Preemptive and Preventive Measures: DP Approach Abbreviations CT: counter-terror DP: direct and preventive MB: marginal benefit Concepts and Learning Points to Terrorism MC: marginal cost Org: terrorist organization Preventive measures reduce the terrorists demand for engaging in terror as a means to achieve their end, whereas (as we ve learned in previous models) preemptive measures tend to reduce or limit the capability of terror organizations to produce terror. As long as the marginal costs of preemption are increasing, in the face of an increase in militancy or threat of terrorism, the State will not enhance its preemptive measures to the extent which will fully counter the increase in militancy. As result an increase in preemption will be associated with an increase in terror. In this sense, winning the war on terror (WoT) by preemptive measures only is impossible. Combining direct counter-terror measures like preemption with preventive measures is more effective in combating terrorism and can win the WoT INTRODUCTION 1 In the preceding chapters we ve assumed that a terrorist organization derives direct utility from the expected damage it causes to the target nations through producing and unleashing terror. 1 The material of this chapter is based on Das and Lahiri (2018). 1

2 12.1. Introduction While the models based on this assumption generated reasonable predictions on the behavior of terror organizations and reasonable conclusions on how security and preemptive measures work, they do not capture terrorism as a means to a long-term end. Remember that terrorism as a form of violence is defined as the use of terror to instill public fear in order to achieve some long-term political, religious or socio-economic goal. While any ideological goals cannot be accepted at any cost, some other could be arguably legitimate. For instance, IRA in Northern Ireland wanted more political participation, which is very different from ISIS s goal of Caliphate. Even Hamas s goal of a separate Palestinian state has some legitimacy. In general, withdrawal of outside armed troops from the ground or removal of an oppressive regime supported by a major power can be thought as a legitimate goal. In this chapter we incorporate such a long-term objective of a terror organization into our analysis. Doing so adds an important and extra dimension to the set of counter-terror (CT) measures that can be used to address the problem of terror. Meeting the long-term legitimate demands of a terror organization or of the population it hails from amounts to preventive counterterror measures in terms of redressing their legitimate grievance. CT measures such security, preemption and intelligence gathering studied in previous chapters are direct means to fight terrorism. We can compare these measures to medicines or procedures to combat a disease in the field of medical science, whereas redressal of grievances of communities from which terrorist organizations emerge is akin to preventive measures that aim to counter the origin or the root cause of a disease. Right after 9/11, the U.S. President George W. Bush declared the war on terror (WoT) or global war on terror (GWoT). WoT or GWoT meant cracking down of terrorists, terrorist leaders and their bases along with increased measures of security. More than a decade later, there is a growing realization that WoT has not been won. While the likelihood of spectacular attacks such as 9/11 has receded thanks to the deployment of comprehensive security measures across many countries, strikes on terrorist havens, scrutiny of movement of funds, larger volume of intelligence gathering and the like the number of terror attacks and the number of human lives lost by these attacks globally have increased dramatically. According to Institute for Economics and Peace (2015), there were less than 2,000 terrorist incidences globally in the year 2000, killing less than 4,000, and in 2014 nearly 13,000 terrorist incidents were reported which killed over 30,000 people. While most terrorist activities are confined to SANIP Syria, Afghanistan, Nigeria, Iraq and Pakistan or SANIPY that includes Yemen as well, the rest of the world witnessed 54% increase in terror incidents in 2013, compared to The threat of terrorism is global and persistent. And, in this sense victory in WoT remains a far cry. There seems to a growing awareness that WoT is not just about raising security, killing or capturing terrorists, squeezing the flow of funds or destroying their bases, infiltrating their organizations, etc. Furthermore, poverty and lack of education may be contributing factors, but are not the primary causes behind the rise of terrorism. The key is to understand what the terrorist Page 2 April 4, 2018

3 12.1. Introduction organizations really want or what their ultimate goal may be and there is a recognition of the need to redress some of terror organizations grievances as a part of a holistic approach to deal with terrorism. Painstaking collection of quality data by Pape and Feldman (2010) suggests that direct and indirect control or occupation by foreign armed forces is the primary cause behind suicide terrorism. The lack of political and religious freedom and failed states provide the breeding ground of terrorists and terrorism (Krueger (2007), Berman and Laitin (2008) and Berman (2011)). Krueger (2007) capsules these factors as geopolitical grievances. 2 In her speech delivered at the OSCE (Organization for Security and Cooperation in Europe) Expert Conference in Astana, Kazakhstan, the U.S. Deputy Coordinator for Homeland Security and Multilateral Affairs, Anne Witkowsky discussed the need for international cooperation at various levels to counter terrorism and at the same time said the following: To make progress on this front, we must resolve legitimate grievances peacefully and strive to foster good governance, reduce poverty and corruption, and improve education, health and basic services. In particular, we need to focus our efforts on locations where the risk for radicalization is exceptionally high. Once we understand the unique combination of grievances and needs that are driving recruitment in these hot spots, we can set about creating tailor-made programs and strategies in response. (Witkowsky (2010)) David Sedney, the Deputy Assistant Secretary of Defense for Afghanistan,Pakistan and Central Asia from and the Deputy Assistant Secretary of Defense for East Asia from , writes Our singular focus on killing, without any serious attempt to ameliorate basic societal problems and the absence of a moral core for our actions have led huge swathes of the world to see us as the evil doers. Extremists today seek revenge for those we have killed, to punish us for abuses they suffer, and to end our support for abusive, corrupt rulers. (Sedney (2015)) In the White House Summit on Countering Violent Extremism in 2015, President Obama had voiced that we have to address grievances terrorists exploit. One of the notable success stories in more or less eliminating terrorism and conflict is the Northern Ireland case. The main grievance of the nationalists (Sinn Fein and the Irish Republican Army) was the lack of representation of the Catholics in actual policymaking, and the main tenet of the peace agreement has been power-sharing by all the groups in the Northern Ireland administration. Recent negotiations between the Colombian government and FARC at the behest of the United Nations that are likely to end more than fifty years of violence and terrorism include 2 A broader list of geopolitical grievances or fears of the population from which terrorist organizations emerge is provided by Hegghammer (2009). April 4, 2018 Page 3

4 12.1. Introduction redressing longstanding grievances of the rebels in terms of government s promise of large-scale investments in rural areas, which, historically, have been neglected at the expense of big cities in Colombia. 3 This chapter argues that while direct measures to curb terrorism are essential, they may not be sufficient. We define the War on Terror (WoT) to be lost if, in response to an exogenous increase in militancy or terrorism, the State steps up its CT measures but not sufficiently enough, with the outcome that the new level of terror is higher despite more intervention by the State. WoT is said to be won if CT measures are enhanced such that terror is reduced. We ll derive a key result that as long as the marginal cost of direct measures is increasing, it is impossible to win WoT by direct measures only. Hence it is also necessary to actively pursue preventive measures in terms of redressing grievances that are legitimate and reasonable, although they carry their own costs for the target countries. The current chapter integrates the direct and preventive measures, and, together, they may be called the DP approach to terrorism. As said earlier, a key feature our analysis will be that the expected damage from terror is not directly utility-enhancing for the Org. Terror is used by the Org against the State in order to induce the State to grant a redressal of grievances. Because preventive measures reduce the incentive to engage in terror activities, commitment to some redressal of terrorists grievances may be critical. A potential concern with measures that meet the demands of terrorist organizations is that such measures may promote a sense that terror works and thus encourage the terror organization to increase terror. This issue will be addressed in section However, it will be argued that these consideration may somewhat reduce but they do not eliminate the scope of preventive measures. The other side of the coin is that not addressing the legitimate concerns may not increase the chances of victory on WoT either. In order to keep our analysis in perspective, it must be noted however that most recent, major instances of terrorism perpetrated by ISIS and modern Al Qaeda seem to be driven more by a global agenda of spreading extreme ideology, and, such objectives are not aligned with eventuating redressal of geopolitical grievances. They are justifiably illegitimate, and definitely unacceptable. Authors such as Kagan et al. (2016) strongly warn that such extreme ideological goals pose an existential threat effectively a WoT of a very complex kind to liberal democracies and requires non-conventional, out-of-the-box, innovative counter-terror strategies. Iannacone and Berman (2006) recommend instituting religious competition in order to deter the emergence and growth of such religious extremism This, however, would a very different kind of preventive CT measures, the analysis of which is outside the purview of this chapter. 3 In Sri Lanka, while the military force dealt a defeat for LTTE in 2009, the civil war bordered on terrorism in that the main target of LTTE was the government; killing of civilians by LTTE occurred toward the end of the conflict. Page 4 April 4, 2018

5 12.2. Analytical Environment 12.2 ANALYTICAL ENVIRONMENT Let there be one terror organization, the Org, and, one target state, the State. They are the two players in a game. The population which the Org represents lives in an environment that is partly dictated by the State, towards whom the population and the Org feel resentful. It could be occupation by armed forces of the State, a regime abhorred by the population but supported by the State or a failed state due to prior policy of the State. Improving such an environment is tantamount to redressing the population s grievances. FIGURE 12.1: Utility Function, Grievance Redressal Deficit and Utility Deficit facing the Org 0 U(g 1 ) g = g 1 Utility Deficit grievance redressal deficit Let grievance redressal be measured in a single (composite) dimension, namely, the unit interval g [0, 1]. An increase in g means more grievance redressal. It is a unilateral concession, not an outcome of any bargaining process. The upper limit on g, equal to ḡ, is normalized to one, and achieving it is the primary goal of the Org. Both the State and the Org inherit the problem of terrorism, which is captured by the initial g being less than ḡ. For simplicity, the initial g is normalized to zero. Any g < ḡ = 1 signifies grievance redressal deficit. 4 The Org obtains utility from grievance redressal, say U(g). We call U(ḡ) U(g) the utility deficit. Figure 12.1 illustrates U(ḡ) the utility function, the grievance redressal deficit and the utility deficit. Redressing grievances of the Org, i.e., granting g, is costly to the State. Let C(g) denote the utility to the State, e.g., from keeping troops, supporting an unpopular regime for security and global military strategic purposes etc. This is exhibited in Figure The negative slope of C(g) captures that granting concessions is costly to the State in terms of utility. Let C(1) = C(ḡ) = 0, where 0 signifies (lowest) level of utility of the State from granting full redressal. ḡ g 4 It is worth stressing again that g is intended to capture terrorist demands that have some justification or legitimacy not goals like political, cultural or religious subjugation of the target countries. April 4, 2018 Page 5

6 12.2. Analytical Environment FIGURE 12.2: Utility Function of the State from Granting Grievance Redressal C(g) 0 Apart from grievance redressal, the State undertakes preemptive measures, the scale of which is denoted by m as in previous chapters. An increase in m tends to increase the Org s MC of producing terror. For simplicity, we abstract from the choice of security measures by the State and assume that the level of such measures is exogenous. The State and the Org choose their strategies or actions sequentially. Consider a three-period scenario where 0 is end of period 0, same as beginning of period 1. The initial level of grievance redressal is g(0) = 0. During period 1 the State chooses its strategies: the scale of preemptive measures (m) and the extent of preventive measures g 1 between 0 and 1. If the State chooses g 1 = 1, the terrorists demands are met and (by assumption) the terror problem is solved. There is no need to engage in direct CT measures and m = 0. If the State chooses g 1 < 1, possibly 0, the problem of terrorism continues to the future. Note that the choices of m and g by the State respectively predetermine the Org s utility deficit and marginal cost of producing terror. In the first part of period 2, say period 2a, the Org produces planned attacks or terror input 2x, that translate into a random flow of potential damages to the State. Half of it, x, is executed in terms of attacks in period 2a, and the remainder is reserved for period 3. (More generally, a fixed proportion of total terror input produced in period 2a is executed in period 2a and the remaining is kept for period 3.) Let the expected damage from terror to the State in period 2a, say D, be proportional to the terror input x, i.e., D = β x, where β > 0. The magnitude of β would depend, among other factors, on the security level chosen by the State, which we treat as exogenous. Hence β is constant. After the unleashing of terror input equal to x against the State in period 2a, in the second half of period 2, say period 2b, the State reconsiders whether to grant further redressal keeping in mind the terror threat in period 3 equal to x. Thus, terror attempts by the Org in period 2a are ḡ g Page 6 April 4, 2018

7 12.3. Solving the Model used as a credible signal of the magnitude of terror to come in the future, unless all grievances are met. The above scenario is constructed, so as to capture the notion that terror is used by the Org as a threat or a means to an end: it does not provide direct utility to the Org. FIGURE 12.3: Time Line of Decision Making Back to the State s decision on grievance redressal in period 2b, let us assume for simplicity that the State faces a binary choice situation: either don t increase g further or grant full redressal, i.e., g 2 = g 1 < 1 or g 2 = ḡ = 1. Further, it cannot reverse its choice of grievance redressal, i.e., cannot choose g 2 < g 1, reflecting commitment on the part of the State, or, that rolling back on concessions is too costly for the State in terms of reputation and credibility. If g relates to military occupation with supporting military hardware and logistics, the resource cost in reversing a pull back may be too high. If g 2 = ḡ = 1, the game ends in period 2. Otherwise, if g 2 = g 1 < 1, in period 3, the Org unleashes terror input x, produced earlier. The timeline of strategies and actions is illustrated in Figure The State and the Org choose their strategies based on their payoffs. Since there are multiple periods of time, they need to consider the discounted value of payoffs. For simplicity, we assume that the discount rate is zero. Nothing qualitative hinges on this simplifying assumption SOLVING THE MODEL Our analysis will involve the understanding of MBs and MCs facing the Org and the State, with respect to their respective strategies. Further, since strategy choices are sequential, the MBs of the State in period 1 from preemptive and preventive measures depend on the anticipated behavior of the Org in period 2. In turn, in first part of period 2 the MBs to the Org from choosing the terror input depends on the anticipation of State behavior with regard to grievance redressal in the latter half of period 2. Such interdependence makes the model fairly intricate. Being a sequential model or game, the game model is solved backward. Beginning first with the last period 3, we note that there is no fresh decision making in this period by either player. The Org either simply disposes x off freely if grievances are fully met, or, uses terror equal to x as planned earlier, otherwise. April 4, 2018 Page 7

8 12.3. Solving the Model THE STATE S DECISION MAKING IN PERIOD 2B In period 2b, the State inherits its own choice g 1 from period 1 and the Org s level of terror production x from period 2a. By assumption, it faces two choices: grant full redressal, i.e., g 2 = ḡ = 1, or, keep the status quo, i.e., g 2 = g 1. The general expression of expected surplus of the State in period 2 is S 2 C(g 2 ) D = C(g 2 ) β x. If the State grants full redressal, C(g 2 ) = C(1) = 0, β x = 0 since the Org would not use terror in period 3 and thus its (expected) surplus in period 2 equals is 0. If the State maintains status quo, i.e., chooses g 2 = g 1, its surplus is C(g 1 ) β x. What will be the State s choice? Keeping the zero surplus associated full redressal, if, given g 1, the level of terror unleashed in period 2a is low enough such that C(g 1 ) β x > 0, then the State s preferred choice will be status quo. Instead, if x is high enough so that C(g 1 ) β x < 0, the State would prefer full redressal. These choices make intuitive sense: if the threat of terror is high, the State would accede to what the Org wants (that is full redressal) and otherwise it would not. Precisely put, the State s choice rule is: g 1 (< 1) if C(g 1 ) β x 0 g 2 = (12.1) 1 if C(g 1 ) β x < 0. This underscores the role of terror by the Org as a means to an end, rather than an end by itself. We next consider the choice of terror production by the Org in period 2a THE ORG S DECISION MAKING IN PERIOD 2A After observing g = g 1 granted by the State and the State s preemptive actions m which determines its cost function, the Org decides its terror production. Assume that the Org doesn t have full information on the State s preference towards g, that is, C(g). For instance, the Org may be unsure about how unwilling or willing the State is toward granting redressal, which may depend on political pressure groups preferences in the State. Thus, the Org does not know for sure whether C(g 1 ) β x 0 for any given terror output it chooses, whereas the State knows its own preference towards g, i.e. there is asymmetric information. 5 Given the anticipated choice rule by the State, the choice of terror output by the Org generates a probability of full redressal from the Org s perspective. A higher x implies a better chance that C(g 1 ) β x < 0 and hence a higher probability of full redressal. Letting p denote the probability 5 Likewise, the State may not have full knowledge of the Org s utility function U(g). But this is not critical to our analysis. Das and Lahiri (2018) analyze a model where U(g) partly unknown to the State, just as C(g) is partly unknown to the Org. Page 8 April 4, 2018

9 12.3. Solving the Model that the State would grant full redress, we have p = p(x), p > 0. We further assume p (x) < 0, i.e., the marginal effect of terror on the probability of full redressal is positive but diminishing. 6 Given the function p(x) and noting that the Org derives utility U(ḡ) in case of full redressal and U(g 1 ) in case of partial or no redressal, the total expected utility or benefit from using terror is: p(x)u(ḡ) + (1 p(x))u(g 1 ). We can rewrite this as: U(g 1 ) + [U(ḡ) U(g 1 )] p(x), (12.2) }{{} utility deficit which is easy to interpret: the higher the utility deficit, the greater is the benefit from increasing terror. If the utility deficit is zero, there is no gain to the Org from producing terror. FIGURE 12.4: Org s Marginal Benefit and Marginal Cost Functions (a) Org s Marginal Benefit from Terror (b) Org s Marginal Cost of Producing Terror The expected benefit expression (12.2) implies that the marginal benefit to the Org from using terror being equal to: MB terror = [U(ḡ) U(g 1 )] p (x) 0. 0 Note that the MB to the Org from terror depends on the change in p or how the odd changes, not on p or the odd itself. Our assumption of p (x) < 0 implies that the MB from producing terror diminishes with the amount of terror input produced. This is exhibited in panel (a) of Figure We now introduce the cost side of producing terror. Let s assume that the total cost function is 6 Here are a couple of examples of the p function that satisfy p > 0 > p : (a) p(x) = 1 e x and (b) p(x) = 1 (1 + x) a. April 4, 2018 Page 9

10 12.3. Solving the Model linear in the terror input produced: b (a + m) (2x), b > 0, where b, a positive constant, is a technology parameter. As in previous chapters, a decrease in a signifies an exogenous increase in militancy or terrorism and z denotes preemptive measures. An increase in preemptive measures tends to raise the MC of producing terror. Remember that the Org produces 2x amount of terror for period 2 and 3 together. The above total cost function implies the MC of terror production (extra cost of producing one unit for two periods together, i.e., half unit for each period) is b (a + m). Hence, the MC of producing per-period terror is 2 b (a + m). For convenience of notation, let s choose b = 1/2, which does not change any qualitative results. Thus MC terror per period = a + m. (12.3) Henceforth, for simplicity, we will refer the MC of per-period terror a + m as marginal cost of terror. Panel (b) of Figure 12.4 shows the MC curve facing the Org. Because they appear additively in the MC function, the marginal effects of a and m on the MC of terror production are constant (equal to unity). FIGURE 12.5: Org s Optimal Production of Terror Realize that the State s choice of preemptive and preventive measures in period 1 determine the positions of the MB and MC curves of the Org in period 2a. Given these schedules, the rational production level of terror is governed by the first-order condition that the MB from terror = the MC of producing terror. That is, [U(ḡ) U(g 1 )] p (x) = a + m. (12.4) Page 10 April 4, 2018

11 12.3. Solving the Model FIGURE 12.6: An Increase in Terrorism and Preemptive Measures Only Figure 12.5 displays this. It is noteworthy that an increase in grievance redressal reduces the MB (shifts MB curve in) from using terror and hence leads to a decline in the quantity produced of terror. Preemptive measures work differently: they shift up the MC function of producing terror and tend to reduce terror. Combining the two is a two-pronged counter-terror intervention to reduce terrorism THE STATE S CHOICE OF CT MEASURES IN PERIOD 1 AND THE DEFINITION OF WINNING OR LOSING THE WAR ON TERROR The central question we ask in this chapter is: how would the State respond in terms of its CT measures in period 1 to an exogenous increase in militancy or terrorism? In our model it boils down to asking: how would the state adjust its counter-terror measures m and g in response to a decline in a? We are now in a position to state what we mean by winning or losing the war on terror. That is, if the State s counter-terror response to a decline in a is such that the equilibrium level of terror production by the Org is not higher than what it was before the decline in a, we say that the WoT is won. Otherwise, we say that WoT is lost. In other words, if an increase in militancy or incentive to engage in terrorism is sufficiently countered by the State in terms of preemptive and preventive measures such that the actual (equilibrium) production of terror by the terror organization is less than before, then it is a win situation. Otherwise, if the State s response is not sufficient so that the actual terror production is higher and thus the expected damage from terror is greater we label it as losing the WoT. April 4, 2018 Page 11

12 12.3. Solving the Model FIGURE 12.7: Marginal Benefit from Preemption against Marginal Cost facing the Org THE STATE S CHOICE OF PREEMPTIVE MEASURES ONLY The conditions under which the State would win or lose the WoT can be understood clearly if we focus on preemptive measures only. Suppose (for now) that the State s continuing policy from past has been to not budge from g = g 0. But it can vary its preemptive measures m. Let the original level of exogenous militancy or terrorism be a = a 0, and suppose that the corresponding choice of preemptive measures be m = m 0, as shown in Figure (How the State chooses m rationally will be considered later.) The MC facing the org is then a 0 + m 0. It intersects the MB curve for the Org at E 0.The equilibrium level of terror is x 0. Suppose the Org and the State inherit a 0, m 0 and x 0 from the past at the end of period 0, and, the State plans to choose m 0 again as it anticipates how the game is going to play out in the next three periods. At this point in the beginning of period 1, imagine that there is an unanticipated increase in militancy in terms of a decline in a from a 0 to a 1 that will last for the next three periods. Notice that if the State keeps the level of preemptive measures unchanged at m 0 despite the increase in militancy (which is unrealistic of course), the marginal cost of the Org in producing terror falls, shown by the dotted line a 1 + m 0. The new equilibrium point for the Org will be E 1. In Figure 12.6, the equilibrium level of terror increases from x 0 to x 1. This is quite intuitive. But, of course, the State wouldn t sit idle and not adjust its preemptive measures. It is expected that it would intensify its preemptive measures. If the State responds by increasing preemption from m 0 to m 1 or higher, then the increase in militancy is fully countered in the sense that the equilibrium terror produced is unchanged at or less than x 0. By our definition, WoT is won. However, if the State increases preemption from m 0 but the new level is less than m 1, then the Page 12 April 4, 2018

13 12.3. Solving the Model MC of the Org (= a 1 + m 1 ) will be less than before (= a 0 + m 0 ) and as a result there will be more production of terror and higher expected damage from terror despite more intensive preemptive measures. The State will be chasing the terrorism problem without being able to overcome it. By our definition, WoT is being lost. You may ask, why doesn t the State simply choose m m 1? The problem is that increasing preemptive measures is costly in terms of resources, lost lives of soldiers as well as collateral damage. Whether the (rational) State would increase m up to m 1 or beyond depends on the MB from preemptive measures and its tradeoff with the MC of preemption MARGINAL BENEFIT TO THE STATE FROM PREEMPTIVE MEASURES Realize that the total and marginal benefits from preemptive measures to the State works through the marginal cost function facing the Org: more intensive preemptive measures raise the MC of producing terror, force the Org to produce and unleash less terror and ensure less expected damage from terror. Assume further that the higher the initial level of MC facing the Org, the less will be the marginal effectiveness of preemption in reducing expected damage from terror. This implies positive but diminishing MB from preemption. As shown in Figure 12.7, the MB to the State from preemption is plotted against the MC facing the Org (a + m) in the horizontal axis, rather than against the level of preemptive measures (m) per se. We read this curve as follows. If initial level of exogenous MC facing the Org is a 0, then the MB facing the State starts from A 0 and to the right; if a = a 1, it starts from A 1 and so on. We mustn t confuse between the marginal benefit to the Org from using terror in Period 2, shown in in Figures 12.4, 12.5 or 12.6, and the marginal benefit to the State from preemption in Period 1, shown in Figure Furthermore, notice that in Figure 12.5 or 12.6, the MC of producing terror is shown on the vertical axis, where it is measured along the horizontal axis in Figure We next introduce the MC of preemption MARGINAL COST OF PREEMPTIVE MEASURES Preemptive measures are costly. There are direct cost of military operations to destroy terrorist bases and hideouts, kill terrorist leaders etc. Indirect costs include the value of lives lost of military personnel, the individual-family and national grief and trauma, collateral civilian damage associated with drones and military strikes, frictions in international and diplomatic relations etc. We consider two scenarios: constant MC of preemption (where MC of preemption is independent of the level of preemption) and increasing MC of preemption. In order to compare with the MB from preemption, we also graph the MC of preemption along the vertical axis vis-à-vis the MC of terror production by the Org, a + m along the horizontal axis. See Figure 12.8, where panel (a) and panel (b) exhibit respectively constant and increasing marginal cost of preemption. Like the April 4, 2018 Page 13

14 12.3. Solving the Model FIGURE 12.8: Marginal Cost of Preemption (a) Constant MC of Preemption (b) Increasing MC of Preemption MB, we measure the MC of preemption (m) from the value of a and toward the right. Interestingly, the graphs measure one player s MC against the the other player s MC EQUILIBRIUM AND RESPONSE TO TERROR: CONSTANT MARGINAL COST OF PREEMP- TION We next put together MB and MC curves facing the State and characterize its decision making. Recall our assumption that the State is unwilling to grant any redressal, i.e., g 1 = g 0 = 0 and it uses preemption only. Figure 12.9 (a) illustrates the case of constant MC of preemption. At a = a 0 initially, the equilibrium MC of the Org is at the point M = a 0 + m 0. The implied rational choice of preemption by the State is m 0. We can read the equilibrium per-period terror production in panel (b): it is x 0. Consider now an increase in militancy: a falling from a 0 to a 1. Since the marginal effect of m on the Org s MC function is constant, any change in a doesn t affect the MB curve from preemption facing the State. In panel (a) of Figure 12.9, there is no shift of the MC curve facing the State, because MC of preemption is constant. The optimal point is M as before, where MB = MC. But there is a difference: the implied new level of preemption is m 1, which is greater than m 0. The upshot is that stepping up preemption to the level m 1 fully offsets the decline in a, so that there is no change in the MC facing the Org and hence no increase in the equilibrium production of terror. It stays unchanged at x 0 in panel (b) of Figure 12.9; the dotted lines do not apply. Page 14 April 4, 2018

15 12.3. Solving the Model FIGURE 12.9: Constant Marginal Cost of Preemption and Response to Increase in Militancy or Terrorism (a)state (b) Org Result 12.1 If MC of preemption is constant, in response to an exogenous increase in militancy or terrorism, the State steps up preemptive measures sufficiently enough such that the Org s marginal cost of producing terror is unchanged, the Org produces the same amount of the terror input and thus the expected damage from terror does not increase, despite an increase in militancy. Terror is contained to its old level and in this sense the WoT is won EQUILIBRIUM AND RESPONSE TO TERROR: INCREASING MARGINAL COST OF PREEMP- TION Different implications follow if preemptive measures entail increasing marginal cost to the State. Refer to Figure (a). The upward sloped lines are the MC functions of preemption which increase with the level of preemption, whose horizontal intercepts are determined by the value of a. Initially, a = a 0 and the MC of preemption (m) starts at point a 0 on the horizontal axis. See that it intersects with the State s MB curve at C 0. The equilibrium MC of the Org is M = a 0 + m 0, where m 0 is the rational choice of preemption by the State. In panel (b), the corresponding equilibrium production of terror by the Org is x 0. As militancy increases, so that a falls to a 1, the upward sloping marginal cost curve of preemption moves to the left. The new equilibrium point is C 1 and new MC of the Org is at M 1 = a 1 + m 1. However, M 1 < M means that MC facing the Org is less than before, after enhanced preemptive measures undertaken by the State. This is the main point and illustrates the limitation of direct April 4, 2018 Page 15

16 12.4. Combining Preemptive and Preventive Measures FIGURE 12.10: Add Caption (a) State (b) Org CT measures like preemption. If the State does fully offset the increase in militancy by increasing m from m 0 to m 2, MC of preemption would be C 2, which is higher than MB = C 0. Therefore, the rational State would not step up m to m 2. This is due to increasing MC of preemption, reflected by the upward sloping MC curve for the State. A lower MC facing the Org, equal to a 1 + m 1, is shown as the dotted in line in panel (b). The equilibrium production of terror is higher at x 1. As a consequence, the State s expected damage from terror is greater. In this sense, the State does not win WoT. Result 12.2 In response to an exogenous increase in militancy or terrorism, the State steps up preemptive measures but not enough to fully offset the decline in marginal cost facing the Org, if the marginal cost of preemption is increasing. The Org produces more terror input and the expected damage from terror is greater. WoT is not won. Result 12.2 highlights the limitation of direct CT measures COMBINING PREEMPTIVE AND PREVENTIVE MEASURES Suppose the State is open toward offering grievance redressal. It is beyond the scope here to fully characterize the rational choice of grievance redressal. But it is straightforward to see how combining preemptive measures with preventive measures (grievance redressal) can lead to a better outcome. The key point (noted earlier) is that grievance redressal reduces the Org s MB from using terror by reducing the utility deficit and hence shifts it to the left. Hence it reduces terrorism through its utility deficit reducing effect. Consider panel (a) of Figure The original situation is described by (a 0, m 0, g 0, x 0 ) and Page 16 April 4, 2018

17 12.5. Partial Redressal May Encourage the Terror Organization to Produce More Terror FIGURE 12.11: Preventive Measures and Org s Choice of Terror (a) WoT is lost (b) WoT is Won the original The MB curve facing the Org is B 0. Under increasing marginal cost of preemption, as an increase in militancy lowers a to a 1, the State steps up its preemptive measures to m 1. As we saw earlier, this is not enough to deter the Org from producing more terror. If the State didn t grant any redressal so that the MB curve facing the Org is described by B 0, x 2 is the new level of terror, which is higher than x 0. But if the State provides a relatively modest redressal so that the new MB curve of the Org is B a (which is quite close to the original curve B 0 ), the equilibrium terror production is x a. WoT is still lost. But the outcome in terms of terror production and expected damage from terror is better than if there were no grievance redressal. Panel (b) of Figure illustrates a situation where the State grants more grievance redressal so that the inward shift of the Org s MB curve is greater. The new MB curve is B b. The equilibrium terror production is now x b, which is less than x 0. WoT is won. The upshot is that combining preemptive and preventive measures is likely to yield a better solution than just using preemptive measures only and has the potential of winning the WoT. It is important to note that preemptive and preventive measures are not substitutes. Rather, they complement each other: while preemptive measures enhance the Org s MC of producing terror, preventive measure curtail the Org s MB from producing terror PARTIAL REDRESSAL MAY ENCOURAGE THE TERROR ORGANIZA- TION TO PRODUCE MORE TERROR It may be argued that granting partial redressal may lead the Org to think that the State is soft liner and terror works on the State. Thus partial redressal would encourage terror instead. We can call this an expectations revision effect in the sense that partial redressal lead the Org to revise its expectation on the State s type, although the true preference of the State remains unchanged. April 4, 2018 Page 17

18 12.5. Partial Redressal May Encourage the Terror Organization to Produce More Terror There are two counter arguments however. First, let the Org s expectations revision because of partial redressal be reflected in the p( ) facing the Org, where recall that p is the probability that the State would grant full redressal in period 2b from the perspective of the Org. Instead of p = p(x + ), let p = p(x +, g 1 + ). The partial p/ g 1 > 0 means that a partial redressal leads to the Org to believe that the State is more like to offer full redressal at any given level of terror x. rule: Does it necessarily imply that the Org will produce more terror? Recall the terror output choice [U(ḡ) U(g 1 )] p }{{} x = MC of producing terror. MB to the Org Note that terror output choice does NOT depend on p itself. It depends on p x : how p changes (i.e. the odds change) with respect to an increase in terror. It is possible that even if p = p(x +, g 1 + ), p x may be independent of g 1. For example, suppose p = x 1 + x + γg 1. γ > 0. It is easily checked that p x is independent of g 1 in this case. Hence if the p function is one like the above, terror production is not affected by any revision of belief by the Org. Thus, if the marginal effect of terror on the probability of full redressal is unaffected by partial redressal, the expectation revision effect doesn t necessarily lead to more terror production. Second, even if the p(x, g 1 ) function is such that the expectations revision effect implies more terror production, it does not wash away the utility deficit reducing effect so long as the terror organization eventually cares only about achieving its goal. The latter effect is likely to be stronger if the grievance redressal is not relatively minor. To see this, consider the following example: p = γg 1 x 1 + γg x. In this case, there is a potential problem with partial redressal, because p x = γg γg 1 (1 + x), 2 which increases with g 1. This is the expectations revision effect. However, the choice rule for terror production is: γg 1 1 [U(ḡ) U(g 1 )] p x = [U(ḡ) U(g 1 )] = MC of producing terror. }{{} 1 + γg 1 (1 + x) 2 utility deficit reducing effect }{{} expectations revision effect Page 18 April 4, 2018

19 12.5. Partial Redressal May Encourage the Terror Organization to Produce More Terror Notice that an increase in grievance redressal (g 1 ) has two effects on the MB of the Org from terror: a utility deficit effect and an expectations revision effect. It is not necessary that the latter effect outweighs the former. Indeed, it is shown by Das and Lahiri (2018) that as long as g 1 is above a threshold, the utility deficit effect will outweigh the expectations revision effect and hence partial redressal results in less terror despite the expectations revision effect. April 4, 2018 Page 19

20 Exercises 12.1 Assume that the State is unwilling to provide any grievance redressal, i.e., g 1 = 0. Let the State s marginal benefit function with respect to preemptive measures be MB from preemption = 110 (a + m), which is decreasing in m. The total cost of terror production by the Org is (a + m)x, so that the marginal cost of terror production is a + m. Notice that an increase in preemption m increases the marginal cost of producing terror. Let the initial value of a is: a 0 = 10. A decrease in the value of a would reflect an increase in militancy or terrorism. Suppose the total cost function of preemption is C(z) = 90m. The the marginal cost of preemption is constant, equal to 90. a. What preemeption level will be chosen by the State? What is the Org s equilibrium level of marginal cost producing terror? b. Suppose there is an increase in militancy or terrorism in terms of a decrease in the value of a from a 0 = 10 to a 1 = 0 (an extreme case). What will now be the optimal level of preemption for the State and what would be the Org s equilibrium marginal cost of producing terror? c. Compare your answers to parts (a) and (b) and provide an intuitive reason Consider the decision making by the Org in period 2 now. The probability of success in securing full grievance redressal is given by the function p(x) = x 1 + x. Check that as long as x is zero or positive, p lies between 0 and 1; further, it increases with x at a decreasing rate, i.e., p 1 (x) = (1 + x) > p (x) = (1 + x) < 0. 3 Let the Org s utility function from grievance redressal is: U(g) = v g, v = 980. a. If the State is unwilling to grant any redressal, so that g 1 = 0, what is the utility deficit facing the Org? 20

21 b. What will be the Org s choice of terror x when a = a 0 = 10? c. What will be the Org s choice of terror x when a falls to a 1 = 0? d. Compare your answers to parts (b) and (c) and very briefly describe why the answers may be different or the same Suppose instead that the cost of engaging in preemptive measures has the function C(m) = 4.5 m 2, implying that the marginal cost of preemption is 9m, which increases linearly with preemption. a. Answer part (a) and part (b) of Question #12.2. b. Briefly discuss in words why you get different results for part (b). c. Computer the levels of terror x that will be chosen by the Org when a = a 0 = 10 and when a = a 1 = 0. d. Is the State winning the war on terror or militancy when there is such an increase in terrorism or militancy? Why or why not? 12.4 Continue to assume that the marginal cost of preemption is increasing and equal to 9m. Suppose that in response to an increase in militancy or terrorism, the State is now willing to grant redressal. What is the level of grievance redressal needed so that the equilibrium terror at a = 0 is same as that when a = 10? 21

22 References Berman, Eli (2011), Radical, Religious and Violent: The New Economics of Terrorism: The MIT Press. 3 Berman, Eli and David D. Laitin (2008), Religion, Terrorism and Public Goods, Journal of Public Economics, Vol. 92, pp Das, Satya P. and Sajal Lahiri (2018), Why Direct Counter-Terror (CT) Measures Only May Fail: An Analysis of Direct and Preventive CT Measures. Mimeo. 1, 8, 19 Hegghammer, Thomas (2009), Jihadi-Sala s or Revolutionaries? On Religion and Politics in the Study of Militant Islamism, in Roel Meijer ed. Global Salafism: Islam s New Religious Movement: Columbia University Press. 3 Iannacone, Lawrence R. and Eli Berman (2006), Religious Extremism: The Good, the Bad, and the Deadly, Public Choice, Vol. 128, pp Institute for Economics and Peace (2015), Global Terrorism Index 2015 Measuring and Understanding the Impact of Terrorism. 2 Kagan, Frederick W., Kimberly Kagan, Jennifer Cafarella, Harleen Gambhir, and Katherine Zimmerman (2016), U.S. Grand Strategy: Destroying ISIS and al Qaeda, Report One Al Qaeda and ISIS: Existential Threat to the U.S. and Europe, Technical report, Institute for the Study of War. 4 Krueger, Alan B. (2007), What Makes a Terrorist? Economics and Roots of Terrorism, New Jersey: Princeton University Press. 3 Pape, Robert A. and James K. Feldman (2010), The Explosion of Global Suicide Terrorism: Cutting the Fuse: The University of Chicago Press. 3 Sedney, David (2015), America s Counterterrorism Policy Is Failing, January Witkowsky, Anne (2010), Preventing Terrorism: Strategies and Policies to Prevent, OSEC Expert Conference, Astana, Kazakhstan, October