The Effectiveness of Rocket Attacks and Defenses in Israel

Abstract

This empirical article studies rocket attacks and defenses in Israel during operations Protective Edge, Pillar of Defense, and Cast Lead, and also during the Second Lebanon War. It analyzes publicly available counts of rockets fired, fatalities, casualties, and property damage. The estimates suggest that interceptor deployment and civil defense improvements both reduced Israel's losses slightly during Pillar of Defense and substantially during Protective Edge. They also imply that interceptor performance during Pillar of Defense may have been overstated. Ground offensives were the most expensive way to prevent rocket casualties. Interceptors were at least as cost-effective as military offensives, and their advantage improved over time. Without its countermeasures, Israel's rocket casualties could have been more than fifty times higher during Operation Protective Edge. These results imply that Israel's rocket concerns were more justified than critics admit, but its military operations were less worthwhile than intended.

Introduction

From 2005 to 2014, militants such as Hamas in the Gaza Strip and Hezbollah in Lebanon fired more than 16,500 rockets carrying roughly 260 metric tons of warheads into Israel, causing more than 2,600 wounded or dead and $160 million in property damage. In defense, Israel spent billions of dollars (including more than $1 billion of American aid) on interceptors, shelters, and sirens. It also ran three military operations against Gaza that cost the Israeli Defense Forces (IDF) about $3.6 billion and 530 casualties, and inflicted substantial collateral damage on Palestinian civilians.¹

There has been much controversy as to whether these operations were justified. Some observers consider the rocket attacks a “strategic threat” (Rubin 2011), while others discount them (Perry 2014). Controversy also surrounds Israel's Iron Dome interception system. While some call it a “game changer” (Hamilton 2012), others question its effectiveness (Broad 2013) or its effects on peace efforts and defense priorities (Kober 2013; Shapir 2013a, 2013b).

Given these debates, this article analyzes rocket attacks and defenses in Israel during Operation Protective Edge in 2014, Operation Pillar of Defense in 2012, Operation Cast Lead in 2008–2009, and the Second Lebanon War in 2006. It leaves aside their political, diplomatic, legal, and ethical dimensions, as well as their effects on Gaza's civilians. It focuses instead on the effectiveness and efficiency of Israeli countermeasures as tools for reducing losses inflicted by rocket attacks. Are offensives against Gaza worthwhile? Are ground assaults needed, or are air strikes sufficient? Is rocket effectiveness declining (Hamilton 2012; Rubin 2015)? If so, is that due to interceptor batteries or to civil defenses? If these measures work, are they economical?

The article proceeds as follows. First, I describe the existing civil-defense-or-interception debate and the related controversy over Iron Dome's performance. I next summarize the available data and use it to calculate fatality, casualty, and property damage rates—or loss rates—per rocket for each operation. Preliminary analysis shows that ground assaults reduced daily rocket fire within each Gaza operation, but air strikes alone did not. It also confirms that per-rocket effectiveness has been decreasing, despite increases in volume, range, consistency, and warhead size.

I then compare these loss rates to infer whether decreasing rocket effectiveness should be attributed to interceptor deployment and/or to civil defense improvements. The empirical analysis suggests that Iron Dome batteries intercepted less than 32 percent of all hazardous rockets during Pillar of Defense, but between 59 and 75 percent during Protective Edge. Civil defense improvements further reduced casualties by up to 36 percent for the former operation and between 57 and 75 percent for the latter. The declining casualty rates highlight the growing relative importance of the conflict's economic costs due to, for example, property damage, reduced business productivity, and defense spending. The calculations further suggest the number of rockets hitting populated areas during Pillar of Defense may have been understated. The number of threats to populated areas, on the other hand, may have been overstated. This implies that Iron Dome's effective interception rate may have been significantly lower than reported.

The research also estimates the cost-effectiveness of Israel's countermeasures for the stated goal of preventing rocket losses. Preemptively destroying rockets in Gaza via military operations, especially ground assaults, incurred high costs while preventing few casualties. By comparison, interception grew increasingly cost-competitive over time. Civil defenses prevented the most casualties with the lowest marginal costs. The article concludes by discussing some implications of the analysis.

Rockets, Interceptors, and Civil Defenses

Apparent Trends and Competing Explanations

While the first rockets from Gaza landed in Israeli in 2001, the attacks “transformed from nuisance to strategic threat” (Rubin 2011, 1) after the IDF's withdrawal in 2005. The attacks included a variety of unguided artillery rockets (for specifications, see Federation of American Scientists 1999 or Dullum 2009); of these, the locally produced short-range Qassams were the most common. They increased in size during the 2005–2014 period, with warhead weights growing from 5 to 15 kg and ranges extending up to about 17 km. Medium range rockets like Grads were also common and carried 18–21 kg warheads up to 45 km. Long range M75 and Fajr-5 rockets were rare. All are relatively inaccurate and best suited to large area targets.

Several analysts have suggested that the rockets’ lethality has been declining. Rubin (2011, 21–22) estimated that the average number of rockets fired per fatality climbed from about 100 in 2002 up to 2,100 by 2008. Hamilton (2012) reported a similar trend, from between fifty and seventy-five rockets per fatality in the early 2000s, up to five hundred in 2012, despite larger warheads.

Two competing arguments have been made to explain these apparent declines in per-rocket lethality. Some analysts primarily credit Israel's deployment of Iron Dome interceptor batteries and the Tamir interceptors they fire (Hamilton 2012; Rubin 2015, 27). The first battery arrived in March 2011. Three more were deployed before Pillar of Defense began, while a fifth was added during that operation (Katz 2011; Shapir 2013b). Israel added two more batteries before Protective Edge and three during the operation. The technology reportedly cost more than $1 billion to develop, followed by costs of battery construction (some $50 million each), interceptor resupply, and ongoing upkeep (Reuters 2014).

Improved civil defenses provide the alternative explanation for decreasing losses (Postol 2014). Israel's spending on shelters and reinforced buildings totaled $384 million from 2005 to 2012 (State of Israel 2015, 110), and by 2014 more than 70 percent of homes had shelters (Lappin 2016a). In 44 communities near Gaza, 10,140 home shelters have been constructed (Kamisher 2016). Israel's $140 million investment in warning systems has yielded increasingly precise alerts (State of Israel 2015, 110); the country was divided into only 25 warning zones in 2006, but 248 in 2015 (Isby and Richardson 2014; Lappin 2015). Three thousand zones are expected by 2018 (Cohen 2016).

Thus, while Israel's interceptor deployments and civil defense improvements complement each other as rocket-fire countermeasures, they compete as explanations of declining rocket effectiveness. This competition is linked in turn to controversy over the performance of the Iron Dome batteries.

Interceptor Performance Controversies

There are several ways to quantify the performance of interceptor systems like Iron Dome, depending on which rockets are considered. Publicly available data allow the calculation of aggregate measures like the percentage of arriving rockets intercepted. These measures consider all rockets, whether headed toward areas defended by batteries or toward undefended areas. They represent how well the batteries in aggregate protected the country overall, making them suitable for strategy and policy analysis. This article focusses on such metrics.

By contrast, the interception rate is the percentage of rockets destroyed before they hit defended areas; it ignores rockets over undefended areas. This assesses a battery's tactical performance against the rockets it faced directly, but is generally impossible to verify from public data. Very high interception rates are claimed for Iron Dome. Ginsburg (2015) reports that batteries intercepted 735, or 91.99 percent, of the 799 rockets and mortar shells engaged during Protective Edge. Harel and Cohen (2015) claim that batteries intercepted 85 percent of their targets during Pillar of Defense and 89.6 percent during Protective Edge.

However, several writers have challenged these claims (Broad 2013; Pedatzuer 2013; Lloyd 2014; Postol 2014). Intercepting a ballistic rocket, as opposed to a more vulnerable cruise missile (Armstrong 2014b), requires breaking apart the warhead. Merely damaging the metal casing is inadequate, as the rocket could continue along its original trajectory or another equally dangerous one.

The controversy arises in part because public photographs lack enough detail to confirm the warheads’ destruction. It is also partly due to tendencies for missile defense advocates to overstate systems’ capabilities. For example, during the 1991 Gulf War, American Patriot batteries initially claimed interception of 95 percent of incoming Scud missiles, but later reduced that to 59 percent. Some analysts suggest that the true rate was below 10 percent (Postol 1991; Sullivan et al. 1999). More recently, Israel's Defense Ministry was accused of exaggerating test results for its Arrow interceptors (Melman 2014; Cohen 2014).

Iron Dome critics have advanced two main arguments. The first notes that interception claims seem unbelievably high for such a challenging task, and most claims cannot be verified independently. It is consequently possible that actual performance was worse than claimed or that some reportedly intercepted rockets never really existed (Globes 2014). The claims also are often vague or mistaken regarding the rockets they include. (Even Rubin [2015, 11] has complained about this lack of clarity.) For example, one military interviewee said the IDF managed to “successfully intercept 85 percent of 1,500 rockets” during Pillar of Defense and “up to 90 percent out of 4,700 rockets” during Protective Edge (Lapidot 2015). Most of those rockets, however, were never even engaged.

A better-substantiated criticism comes from assessments of the Tamir interceptors’ behavior. Postol (2014) and Lloyd (2014) separately studied videos and photos of interception attempts. They used the images to infer interception rates. Based on the Tamirs’ approach trajectories and design, they argued that most attempts probably did not destroy rockets. Lloyd (2014) estimated effective interception rates of 30 to 40 percent, while Postol (2014) put them below 5 percent. The authors also noted that, while the IDF reported 58 rocket hits on urban areas during Pillar of Defense, Israeli police reported 109 (Pedatzur 2013).²

Proponents of Iron Dome have responded with two main rebuttals. The first notes that critics rely on amateur photos and videos of lower quality than those of the IDF, and that IDF claims are thus more credible (Richardson 2014a; Shapir 2013a). The second counterargument compares loss rates from different conflicts. Rubin (2015) notes that fatalities and damage claims per rocket fired decreased between Second Lebanon, Pillar of Defense, and Protective Edge. He concludes the batteries therefore must have been effective, or else loss rates would have been much higher. Rubin (2013) similarly compares the Second Lebanon War and Pillar of Defense, arguing that without Iron Dome the latter conflict would have had 14,400 damage claims, instead of 3,165. However, his analyses omit Cast Lead and discount improvements in civil defense.

Analysis Approach

I address these controversies by estimating the influence of Iron Dome and civil defenses at reducing Israel's losses, using a loss rate approach that builds on Rubin's (2013, 2015). I begin by summarizing the rocket-related data available for the four Israeli conflicts and use it to calculate a variety of loss rates on a per-rocket basis. I then adjust the rates to control for the effect of changes in rocket warhead sizes from one conflict to another.

Next, I use the following logic to distinguish between interceptor and civil defense influences. Interception destroys rocket warheads in midair to prevent them from hitting populated areas. This protects both people and property. By contrast, civil defenses shelter civilians from rocket attacks, but do not stop the rockets from hitting buildings or other property. Therefore, they primarily protect people.

This implies that, if only civil defenses are improving in effectiveness, then fatality and casualty rates should fall, but with little to no change in property damage claim rates. Conversely, if only interception improves, there should be roughly equal declines in claim, casualty, and fatality rates. Finally, if both countermeasure improvements are effective, there should be small decreases in claim rates and large decreases in casualty and fatality rates.

This indirect approach can only show correlation, not causation. In principle, loss rates could also be influenced by many other secondary factors, such as the tactical choices made by individual militant and IDF units. However, I am not aware of any arguments for such factors as alternative (i.e., more influential) explanations for declining rocket effectiveness; the debate is between civil defense and interception. Furthermore, while my approach is imperfect, it is nonetheless much more detailed than previous efforts. For example, Hamilton (2012) and Rubin (2013, 2015) attributed decreasing loss rates entirely to interception while ignoring other factors. Similarly, Postol (2014) discounted interception and credited the low rates largely to civil defense. By contrast, I consider both interception and civil defense as possible influences, while also controlling for warhead size.

Rocket Data and Preliminary Calculations

The data below on rocket attacks mostly come from the Israeli Security Agency (ISA) online archive of annual (e.g., ISA 2014) and monthly (e.g., ISA 2014 August) reports. Where reports combined casualties from rockets and mortars, I estimated rocket casualties via prorating. To attain a more comprehensive measure of the attacks’ human impact (Fazal 2014), the study takes into account casualties (dead plus wounded) as well as fatalities (dead).

Attacks Outside of Operations, 2005–2014

Table 1 shows the number of rockets fired and casualties inflicted each year from 2005 to 2014, excluding those during IDF operations.³ (The rockets also damaged property, reduced economic output, and increased psychological stress.)

The two sets of numbers can be combined to yield the average number of casualties per rocket by year. Figure 1 shows that this casualty rate per rocket had a clear downward trend. The decade's overall average was 0.085 casualties per rocket (or about 12 rockets per casualty), but this fell from 0.1049 for 2005–2009 to just 0.0223 for 2010–2014. Note that the decline was already apparent before Iron Dome's introduction in 2011.

Operation Protective Edge, 2014

Israel launched Operation Protective Edge in 2014 to “stop Hamas’ incessant rocket attacks against Israel's civilians” (IDF 2014b) and “restore sustained calm and security to the Israeli civilian population” (State of Israel 2015, 22). It began with air strikes from July 8 to 17, followed by ground combat from July 18 to August 4. Alternating skirmishes and ceasefires ensued from August 5 to 26, including eight days without rocket fire. The IDF claimed to have destroyed 3,000 rockets on the ground, plus other targets (Yaakov 2014).

Each day the IDF (2014a, 2014b, 2014c) announced tentative rocket counts (see Figure 2). The heaviest fire came on July 10, when militants launched 192 rockets and 185 arrived in Israel. Fire was lowest on July 26 due to a brief cease-fire. An IDF summary reported 3,356 rockets fired as of August 4: 475 misfired or landed within Gaza, 578 were intercepted, 116 hit “populated areas,” and 2,187 landed elsewhere (Yaakov 2014); it did not define “populated areas.” The ISA (2014) later reported 2,968 rockets and 1,724 mortar shells launched during the operation. Some 1,963 rockets (presumably Qassams) flew under 20 km, 760 (presumably Grads) flew 20 to 50 km, and 255 traveled more than 50 km into Israel (ISA 2014 July, 2014 August).

To counter this barrage, Israel deployed seven Iron Dome batteries before the operation and three more during the first week (Lappin 2014), though one remained out of action due to crew shortages (David 2014). Reports variously said the batteries intercepted 735 of the 799 rockets and shells they engaged (Ginsburg 2015), 89.6 percent of their targets (Harel and Cohen 2015), or 83 percent of all rockets headed into populated areas (Richardson 2014b).

July and August 2014 each saw one rocket death (Hartman and Udasin 2014; ISA 2014 August). The ISA (2014) reported 110 civilians wounded by mortar and rocket fire, while the Ministry of Foreign Affairs (2014) reported 126. For August, the ISA (2014 August) reported 30 people wounded by rockets and 10 by mortars; it did not give a breakdown for July. The Israeli Tax Authority received 4,600 insurance claims worth $38.9 million for property damaged by rockets or mortars (Azulai 2014), and the Bank of Israel estimated the economy suffered $1 billion in indirect losses (Globes 2015).⁴ The government budgeted $225 million of American aid to replenish its supply of Tamir interceptors, nominally priced at $100,000 each (Dagoni 2014). Estimates for the military operation's cost ranged from $1.9 to $2.6 billion (Barkat 2014), plus 67 IDF soldiers killed and 469 wounded (ISA 2014; Hartman 2014).

The IDF summary indicates that 14.15 percent of fired rockets remained in Gaza and 85.85 percent arrived over Israel (Yaakov 2014). Of those arrivals, 20.06 percent were intercepted, 4.03 percent hit populated areas, and 75.91 percent landed elsewhere. This means that 24.09 percent of the arrivals were “threats” that either hit populated areas or were intercepted and that batteries claimed interception of 83.29 percent of those threats. Applying the IDF's arrival breakdown percentages to the ISA's 2,968 arriving rockets implies 715 threats, 595 interceptions, 120 populated area hits, and 2,253 landings elsewhere.

The first step before using this data is to remove any parts not related to rocket fire. The ISA indicated that rockets caused 30/(30 + 10) = 75 percent of the total civilians wounded in August 2014 (ISA 2014 August); the other 25 percent were due to mortars. Assuming July was similar, the civilians wounded by rockets totaled 110 × 0.75 = 83, giving 83 + 2 = 85 rocket casualties in total. Applying the 75 percent share to property damage gives 3,450 claims worth $29.2 million and averaging $8,457 each. If rockets consumed 595/735 = 81 percent of the expended interceptors, then their share of the resupply cost was $182 million. Table 2 summarizes this refined data.

The next step is to compare basic statistics for the operation's air and ground phases. Linear regression of the daily number of rockets fired against calendar days during the air phase showed no statistically significant trend (see online appendix for regression table), indicating that air strikes did not change the rocket fire rate from day-to-day during the operation (regression slope p = 0.836). The ground phase likewise showed no statistically significant trend (p = 0.348). The air and ground phases had statistically significant differences in mean fire rates (t-test p = 0.000), percentages of fired rockets that arrived (z-test p = 0.000), and percentages of arriving rockets that were threats (z-test p = 0.048); see Table 3. This indicates the fire's quantity and quality were lower during the land phase than during the air phase.

The last step is to convert some of the data from counts into ratios; this facilitates comparisons between operations. The IDF had seven times as many wounded as dead. There were 1,725 damage claims per civilian fatality and 40.59 claims per casualty. Casualties averaged 0.0286 per arriving rocket, 0.1189 per threatening rocket, and 0.7083 per hitting rocket. Table 4 shows these rates, plus those for fatalities and claims.

Operation Pillar of Defense, 2012

Operation Pillar of Defense consisted of air strikes from November 14 to 21, 2012, and claimed 980 rocket launchers. It was “launched in response to incessant rocket attacks . . . to cripple terror organizations in the Gaza Strip and defend Israelis living under fire” (IDF 2012b). The IDF (2012a) counted 1,506 rockets fired: 152 stayed in Gaza, 875 landed in open areas, 421 were intercepted, and 58 hit “urban areas.” Figure 3 displays the daily counts. Only 10 long-range rockets arrived (ISA 2012b). Israel started with four Iron Dome batteries near Gaza and later added a fifth farther north (Agence France Press 2012).

The ISA (2013) reported five killed and 232 wounded by rockets, while the IDF (2012a) reported 240 wounded. There were 3,921 damage claims worth $14.81 million (Rubin 2015, 2013). The indirect cost to the Israeli economy was $240 million. The government spent $285 million on the military operation and another $200 million of US aid (Lev 2012) to replenish its nearly depleted interceptor supply (Bergman 2014).

The calculations follow the same logic as before; see Tables 3 and 4. About 10.09 percent of fired rockets stayed in Gaza and 89.91 percent arrived over Israel. Of the 1,354 arrivals, 31.09 percent were intercepted, 4.28 percent hit urban areas, and 64.62 percent landed elsewhere. Thus 35.37 percent (479) of arrivals were threats, of which 87.89 percent were intercepted. The daily fire rate had no statistically significant trend (p = 0.700).

Operation Cast Lead, 2008–2009

Operation Cast Lead began with air strikes from December 27, 2008 to January 2, 2009, followed by land battles during January 3–18 (see, e.g., Farquhar 2009; Rubin 2009). The goal was to “stop Hamas’ almost incessant rocket and mortar attacks upon thousands of Israeli civilians and its other acts of terrorism” (State of Israel 2009, 1). The IDF destroyed 1,200 rockets (Sharnoff 2009), and the government's final report estimated 617 rockets and 178 mortar shells landed in Israel, killing four and wounding 182. Furthermore, 6.5 percent of all rockets fired remained within Gaza (State of Israel 2009, 23). The ISA (2008, 2009) reported 664 rockets and 257 mortar shells, while another source counted 640 rockets and 162 shells (Journal of Palestine Studies 2009, 201–6); see Figure 4. About 230 were Grads, while the rest were Qassams (Rubin 2015, 25). The number hitting populated areas is unknown. The 1,900 property-damage claims worth $10.4 million resulted from the combined fire (Sderot 2010; Kana and Avital 2012). IDF cost estimates ranged from $0.90 to $1.3 billion (Lehav 2014; Filut 2009), and 10 soldiers died in the operation (Reuben 2014).

Given 617 arrivals and 6.5 percent of fired rockets falling short, there were 43 short and 660 fired. Since 77.61 percent of the arriving rounds were rockets, about 141 wounded, three dead, and 1,475 damage claims worth $8.1 million can be attributed to them. Assuming the same seven-to-one IDF wounded-to-dead ratio of Protective Edge, there may have been 70 wounded soldiers.

During the air phase, daily rocket fire showed no statistically significant trend (p = 0.773). Fire during the land phase did show a statistically significant downward trend of 0.79 rockets per day (p = 0.030 and R² = 29.2 percent). The difference in mean fire rates also was statistically significant (t-test p = 0.013), suggesting that while the rate of fire was not reduced by air strikes, it decreased after the army advanced.

Second Lebanon War, 2006

The Second Lebanon (or Israeli-Hezbollah) War consisted of air strikes and ground combat against Hezbollah militants in Lebanon from July 13 to August 14, 2006 (Cordesman 2007; Harel and Issacharoff 2008). Air strikes claimed more than 300 long range rockets (Cordesman 2007, 123), but had little impact on daily fire (Harel and Issacharoff 2008, 96). Police reported 972 rockets hitting “built-up areas,” 3,256 in open areas, and 53 deaths. Between 33 and 250 rockets arrived each day. Almost all were Grads, but 250 landed more than 50 km inside Israel (Rubin 2006). The IDF reported 901 rockets hitting “population centers” and 3,069 in rural areas, leaving 42 dead and 1,489 wounded (Cordesman 2007, 103). There were 26,653 property claims worth $108.1 million (Rubin 2015, 29). Indirect economic losses totaled $1.6 billion and the operation cost $3 billion (Elis 2012). One-hundred-twenty soldiers died (Cordesman 2007, 17). The IDF numbers indicate 22.70 percent of arriving rockets hit population centers.

Rocket Fire Trends

More than 9,492 rockets were fired at Israel during the four aforementioned operations, causing 2,005 casualties (including 52 fatalities) and 35,499 damage claims worth $160 million. Adding the peacetime counts from Table 1 brings the totals to 16,510 rockets and 2,605 casualties. The preceding calculations indicate two general trends: increasing scale of the rocket fire and decreasing effectiveness in terms of losses inflicted per rocket.

Increasing Scale

Between 2008 and 2014, rocket fire from Gaza grew more comparable to that from Lebanon in 2006. The volume of fire and the proportion of long-range rockets both increased, thereby threatening more of Israel's territory, population, and economy. The amount of explosives also increased. During the Second Lebanon War, about 3,720 Grads and 250 long-range rockets struck Israel. Since Grad warheads weigh about 20 kg and the long-range warheads likely weighed 175 kg (Cordesman 2007), the warhead weight totaled about 118.2 metric tons and averaged 29.76 kg per rocket. Cast Lead involved 230 Grads and 387 Qassams. If the Qassam warheads averaged 7.5 kg (Rubin 2011, 21) each, then the barrage totaled 7.5 tons at 12.16 kg per rocket.

Protective Edge included 255 long range, 760 medium range (presumably Grads), and 1,953 short-range rockets (presumably advanced Qassams). Assuming 20 kg warheads for the two former types (Rubin 2015, 13) and 12.5 kg for the Qassams gives 44.7 tons total and 15.07 kg on average. If the Grad-Qassam mix for Pillar of Defense was halfway between Cast Lead and Protective Edge, with 20 kg Grads and 10 kg Qassams, then it totaled 18.4 tons at 13.57 kg per rocket. Similar estimates indicate about 73 tons between operations, putting the decade's warhead total at 262 tons, or about 8.3 B-52H bomber payloads (USAF 2005).⁵

Table 5 shows these warhead averages and totals. It also converts the losses per arriving rocket from Table 4 into losses per arriving metric ton of warheads. For example, Cast Lead's claims rate becomes 2.391 × 1000/12.16 = 196.5 claims per ton. From this “throw weight” perspective, Protective Edge losses seem much lower than those of earlier operations.

In terms of rocket numbers, Pillar of Defense saw the heaviest daily rate, largely due to its lack of a ground assault. Its average was 139 percent higher than that of Protective Edge overall, but only 19 percent higher than that operation's air phase. The remaining decrease may have been due to militants firing from tunnels during Protective Edge to reduce their air strike exposure (Rubin 2015, 21). This slower fire was also more consistent day-to-day: the coefficient of variation, which measures variability relative to the average, was just 0.166 during the Protective Edge air phase, less than half that of Cast Lead's air phase (0.404) and of air-only Pillar of Defense (0.405). The militants apparently had improved their ability to operate during air strikes, perhaps due to their tunnels. Although air strikes helped reduce the average rate of fire from what it could have been, and destroyed many unfired rockets, they did not decrease the number fired day-to-day within any of the operations. In fact, militants were apparently able to protect or replace their launch crews and equipment. The daily rate decreased only after ground units advanced, indicating that control of the ground was a prerequisite for effective control of airspace.

Decreasing Effectiveness

While the scale of fire increased, Table 4 shows that effectiveness per rocket decreased. That is, the number of fatalities, casualties, and claims per arriving rocket generally declined over time. This is similar to the decreases in casualty rates shown in Figure 1 and in fatality rates noted by other observers. The largest drops occurred between Second Lebanon and Cast Lead, where, for example, the claims rate fell 64 percent, from 6.714 to 2.391. Table 4 also shows fatality and casualty rates falling faster than claim rates, making claims relatively more numerous. This highlights the increasing economic costs of the conflicts.

In the next section, I explore some potential explanations for this observed decrease in effectiveness.

Evaluating Rockets, Interceptors, and Civil Defenses

Tables 3 and 4 hint that rocket changes, interceptor deployment, and civil defense improvements all influenced loss rates to some extent. The large rate drops between Second Lebanon and Cast Lead correspond to greatly decreased warhead weights. The further decreases for Pillar of Defense and Protective Edge suggest the interceptors made a difference. The fact that casualties and fatalities fell faster than claims suggests that civil defense improvements also helped. The following subsections investigate these influences in more detail.

Warhead Sizes

Warhead sizes differed materially between rockets, but a warhead that is twice as large does not necessarily cause twice the loss. One way to account for this is to compare their relative blast areas. Blast pressure at a given distance from an explosion is proportional to explosive mass taken to the third power (FEMA 2003), while the affected circular area is proportional to that radius squared. Thus, the lethal blast area of 20 kg of explosives is ((20/10)^1/3)² = 2^2/3 = 1.587 times as large as that of 10 kg.

To compare operations, each row of Table 6 shows a warhead mass, that mass taken to the two-thirds power (i.e., to represent the relative blast area) and the numbers of those warheads per operation. The bottom row shows the weighted average area by operation. For example, Second Lebanon had 250 warheads of 175 kg and 3,720 of 20 kg. The war's weighted average area was (250 × 175^2/3) + (3720 × 20^2/3)/(250 + 3720) = 8.874. I use these nominal averages in the next set of calculations.

Comparing Losses to Arrivals: Interception

This subsection analyzes losses per arriving rocket to see what they suggest. Since neither Cast Lead or Second Lebanon had batteries, they can serve as baselines for estimating interceptor effectiveness during the later operations. Cast Lead is the more obvious choice because it occurred in the same part of Israel as Pillar of Defense and Protective Edge. The analysis includes Second Lebanon as a second baseline to give a sense of the estimates’ uncertainty and allow for comparisons with earlier studies (e.g., Rubin 2013, 2015).

The first step is to take the loss rates from Table 4 and convert them into proportions of the Second Lebanon rates after adjusting for the warheads’ average blast areas in Table 5. For example, expressing the claims rate for Protective Edge as a proportion of Second Lebanon's gives (1.162/6.714)(8.874/6.048) = 0.2540. That is, Protective Edge saw only 25 percent as many claims per rocket as Second Lebanon, or 75 percent less, after considering warhead differences. Table 7 displays these loss rate proportions.

Table 8 similarly shows the adjusted loss rates as proportions of Cast Lead's. Here, the Protective Edge rate relative to Cast Lead was 0.4117, or 59 percent lower. Thus Tables 5 and 6 suggest that during Protective Edge, Iron Dome intercepted between 59 and 75 percent of all hazardous rockets, whether headed toward defended areas or elsewhere.

For Pillar of Defense, the adjusted claims rate was 32 percent lower than that of Second Lebanon but 10 percent higher than that of Cast Lead, suggesting the batteries intercepted at most 32 percent of the hazardous rockets, and perhaps much less. The fact that one calculation shows increased claims indicates the uncertainty in the estimates and suggests Iron Dome's influence may have been small enough to be overshadowed by changes in other factors. Figure 5 illustrates the estimates for each operation versus the two baselines, along with their means. It highlights how interception was far less influential for Pillar of Defense than for Protective Edge.

The interception calculations above, including Tables 6–8, try to account for the complexities of rocket hits by assuming that losses are roughly proportional to warhead mass taken to the two-thirds power. A simpler approach could assume instead that losses are proportional to the warhead mass itself (e.g., that a 20-kg warhead causes twice the loss of a 10 kg one). The calculations for this alternative are provided in the online appendix, using the loss per ton values from Table 5. Those results suggest that during Protective Edge, claims rates decreased between 61 and 66 percent relative to the two baselines. During Pillar of Defense, the claims rate estimates range from a 5 percent decrease to a 9 percent increase. That is, the interception estimates from this simpler approach are slightly lower and much narrower than the ones calculated previously. However, they still indicate substantially lower loss rates during Protective Edge and marginally lower ones during Pillar of Defense.

Comparing Losses to Arrivals: Civil Defense

The next step is to estimate the effects of civil defense improvements by comparing fatality and casualty rates to the corresponding claims rates from Table 4. For example, Protective Edge saw 1,725 claims per fatality, whereas Second Lebanon had 635. This means Protective Edge had 1,725/635 = 2.718 times fewer fatalities per claim, or that fatalities decreased by proportion 1 – (635/1725) = 1 – 0.3679 = 0.6319, or 63 percent.

Comparing scaled proportions from Table 7 is another way to reach the same result. The Protective Edge claims rate after warhead adjustments is 0.2540, or 25.4 percent of Second Lebanon's. Interception presumably also reduced casualties to a similar extent. However, the Protective Edge relative fatality rate was only 0.0935, or 9.35 percent. The ratio 0.0935/0.2540 = 0.3681 indicates fatalities were only 37 percent of the expected rate, so civil defense improvements could be credited with 63 percent savings. Table 9 shows the remaining ratios.

These estimates suggest that during Protective Edge, civil defense improvements reduced fatalities by 63 to 72 percent and casualties by 57 to 75 percent. These savings were in addition to fatalities and casualties prevented by interceptors or by preexisting civil defenses (i.e., those built in 2008 or earlier). The Pillar of Defense benefits were more modest, with fatalities between 19 and 37 percent lower, and casualty estimates ranging from 9 percent higher to 36 percent lower. Figures 6 and 7 illustrate these results.

Although the two baselines lead to similar qualitative conclusions about interceptor and civil defense effectiveness, their numerical details differ. Part of that difference may be due to imperfect adjustments for rocket warhead weights. When adjusting by estimated weights taken to the two-thirds power (Table 7), Cast Lead shows 38 percent lower claims, 20 percent lower fatalities, but 5 percent higher casualties per rocket, relative to Second Lebanon. When using unadjusted simple weights (online appendix), Cast Lead instead shows 13 percent lower claims, but 12 percent higher fatalities and 48 percent higher casualties per rocket. The “best” adjustment may lie somewhere in between.

Another part of the difference is likely due to differences between the two conflicts’ contexts: Second Lebanon took place in northern Israel in 2006, while Cast Lead occurred in southern Israel in 2008. On the Israeli side, the two regions had differences in e.g., population densities and civil defense development. On the militant side, Hamas and Hezbollah could have differed in, for example, weapons skills and firing tactics. The combined effect of such factors likely led to Cast Lead having fewer claims and more casualties, relative to Second Lebanon, even after adjusting for warhead sizes in various ways.

Finally, many smaller factors could add “noise” to the estimated loss rates for all four conflicts. For example, the “fog of war” means that neither side was entirely certain how many rockets arrived over Israel or what their warheads weighed; the estimated numbers should be good but not perfect. Similarly, luck always plays at least a small role; for example, a few feet of drift in one direction or another could decide whether a given rocket hits an occupied house, an unoccupied shed, or an empty ditch.

In summary, the analysis regarding arriving rockets suggests that both interceptor deployment and civil defense improvement reduced Israel's losses modestly for Pillar of Defense and substantially for Protective Edge. Table 10 summarizes the estimates.

Comparing Losses to Hits

The number of damage claims per rocket hitting populated or urban areas, as defined by the IDF, provides another way to evaluate interceptor performance. The ratio of claims to hits in Table 4 for Pillar of Defense was more than double those of Protective Edge and Second Lebanon. While Pillar of Defense's rockets may have been aimed differently or simply benefitted from more luck, an alternative explanation is that its number of hits was understated.

Plausible estimates can be found via the other operations’ loss rates. For example, Pillar of Defense involved 3,921 damage claims. To generate that many claims using the Protective Edge claims-to-hits ratio of 28.75 would require (3921/28.75) x (6.048/5.616) = 147 hits after adjusting for warhead sizes. The Second Lebanon numbers imply 209 hits.

These comparisons assume that the ratio of losses in populated areas to those in rural areas was similar in all three operations. However, interceptors mostly defended populated areas, so those should have seen fewer losses as batteries were added. This likely means the 147 hits calculated via Protective Edge (which had twice as many batteries) is an underestimate, while the 209 calculated via Second Lebanon (which had no batteries) is an overestimate.

It is difficult to say why these estimates are much higher than the reported 58 (IDF) or 109 (police) hits. Perhaps the definition of “populated area” varied over time. Some reports may have counted only rockets within city boundaries, while others included outlying suburbs. Each report then would be correct on its own, but distorted relative to the other.

Comparing Threats and Interceptions to Arrivals

While the number of hits during Pillar of Defense seems understated, the number of threats seems overstated. The IDF data for that operation indicate that 35.38 percent of arrivals either hit urban areas or were intercepted. By contrast, only 24.09 and 22.70 percent hit or were intercepted during Protective Edge and Second Lebanon, respectively. This means that the threat percentage was about one-half larger during Pillar of Defense. Similarly, 31.09 percent of arrivals were reportedly intercepted during Pillar of Defense, versus only 20.05 percent during Protective Edge, despite the latter having twice as many batteries.

Pillar of Defense saw 1,354 arriving rockets, of which 479 (35.38 percent) were reportedly threats. If the true proportion was 24.09 percent, as in Protective Edge, that would imply 1,354 × 0.2409 = 326 threats, rather than 479. If it was 22.70 percent, as in Second Lebanon, then the estimate becomes 307. If the proportion intercepted was 20.06 percent, as in Protective Edge, then the interceptions become 1,354 × 0.2005 = 272 instead of 421. The threats would then become 272 + 58 = 330, or 272 + 209 = 481.

It is unknown why these estimates are mostly much lower than the reported counts. Perhaps “defended areas” differed significantly from “populated areas” during Pillar of Defense. For example, if the Iron Dome software and/or crews added safety margins around populated sites, then rockets heading into the added area would have been deemed a threat without actually threatening the population.⁶ The apparent threats then would have increased due to batteries engaging “extra” rockets. Regarding interceptions, perhaps the IDF counted both “confirmed” and “probable” ones during Pillar of Defense, but only “confirmed” ones for Protective Edge.

Alternatively, imperfections in the battery hardware or software may have caused it to misclassify some incoming rockets as threats—“false positives,” in terms of Armstrong (2014a). For example, Patriot batteries in the 1991 Gulf War had software that caused tracking errors after extended operation (GAO 1992). More recently, an Iron Dome battery fired at mortar shells falling outside of Israel (Shoval and Brenner 2016). Operating procedures and operator decisions may also have affected performance. For example, during Protective Edge, “[i]n about 65 instances, operators decided to utilize the defense system contrary to Iron Dome launching policy” (Harel and Cohen 2015).

Combining the Estimates

The loss-per-arrival analysis suggests that during Protective Edge, Iron Dome intercepted 59 to 75 percent of all hazardous rockets, while civil defense improvements reduced human losses a further 57 to 75 percent. These substantial savings support both civil defense advocates and interception advocates.

Interception was much less influential during Pillar of Defense, with the loss-per-arrival analysis suggesting that batteries intercepted less than 32 percent of hazardous rockets. As to why, the loss-per-hit analysis implies there were more than 58 urban area hits, anywhere between 147 and 209. The threats-per-arrival analysis implies that there were 307–326, not 479 threats, and 272, as opposed to 421 interceptions.

Consider the implications of combining these estimates. The IDF reported 58 hits and 421 interceptions during the operation, meaning the batteries intercepted 88 percent of 479 threats. Suppose there really were 58 hits, but only 326 threats. This would mean the batteries only intercepted 82 percent of the threats. On the other hand, if there were actually 479 threats and 147 hits, then the batteries intercepted 69 percent of the total. Assuming 147 hits and 272 interceptions, the score becomes 64 percent. Table 11 summarizes these possibilities, along with several others.

Since this analysis aggregates all incoming rockets, whether headed toward defended areas or elsewhere, it is not directly comparable to the tactical interception rate. However, if the percentage of threats intercepted during Pillar of Defense really was lower than the reported 88 percent, then the effective interception rate may also have been lower than the reported 85 percent. This analysis therefore partly supports earlier technical estimates based on imagery analysis. Postol's (2014) 5 percent interception estimate looks too low, but Lloyd's (2014) 30 to 40 percent suggestion seems plausible.

By contrast, these results differ from Rubin's (2013, 2015) for several reasons. First, by comparing rockets one-for-one, Rubin's work implicitly assumed the warheads had equal payloads. Second, by explicitly assuming civil defenses were constant, it credited any improvements to interception. Third, by using only Second Lebanon and not Cast Lead as the baseline, it calculated the largest possible improvements.

Limitations

The indirect analysis herein cannot prove that Israel's interception claims were incorrect, any more so than interception proponents can prove they were correct. Thus, those proponents might dismiss these results as being merely circumstantial. Conversely, critics might argue that the analysis relies on limited data from Israeli government reports that could be intentionally misleading for security reasons. Those critics might therefore dismiss the results as illusionary. It consequently seems unlikely that “the decades-long debate in Israel between proponents and critics of missile defense has now been laid to rest” (Rubin 2015, 32). However, whereas past debates previously centered on interception's potential feasibility and desirability, it now focuses on the actual effectiveness and efficiency.

The analysis also cannot reveal why interception performance differed between Pillar of Defense and Protective Edge, though several possibilities are apparent. The number of batteries doubled between operations, allowing them to defend more areas and engage more rockets. The quality of the batteries also improved; Israel was using the fourth version of Iron Dome technology in 2014 and continues to upgrade it (Reuters 2014; JPost 2015). In that regard, it is easy to imagine programmers refining Iron Dome's software between operations. For example, the 1991 Patriot software bug was fixed within two weeks of its discovery (GAO 1992), and more recently the IDF rebuilt its battle management software within three months (Lappin 2016b). Similarly, the IDF could have used its experience from the earlier conflict to improve its future procedures. For example, after facing difficulties during Second Lebanon, IDF ground forces improved their doctrine in time for Cast Lead (Farquhar 2009). Regarding system hardware, Lloyd's (2014, 1) report had suggested several enhancements, though implementing those could have taken longer.

One other limitation is worth noting. This analysis takes an indirect approach to evaluating loss rate changes. It treats rocket battles as “black boxes,” analyzing inputs and outputs outside each box to infer the events inside. In effect, it verifies the data's internal consistency: do the numbers make sense relative to each other? It is less able to validate the data for accuracy: do the numbers represent reality? An ideal study would physically observe rocket trajectories, interception attempts, and impact sites, but such hard evidence is rarely made public, and even the IDF likely has only a subset.

The analysis compensates for these challenges by calculating three loss rates (fatalities, casualties, and property damage) that have different strengths and weaknesses. For example, fatality counts are very reliable, as the press prominently reports each death, but they are also very small, meaning that random chance could significantly influence their number. By contrast, counts of wounded are larger and so have better statistical properties, but the definition of “wounded” might vary from source to source, and their numbers are difficult to verify. The calculations also compare the relationships of the various loss rates with rocket fire in several different ways, helping to corroborate the results.

Were Israel's Countermeasures Worthwhile?

Countermeasures’ Benefits

Both Israel's defensive strengths and the rockets’ inherent weaknesses have prevented higher losses. For example, during Protective Edge, only 24.09 percent of all arriving rockets threatened targets. If all the rockets had been that accurate, casualties could have been 4.151 times as high, or 315 percent higher. The IDF's preemptive destruction of 3,000 rockets in Gaza kept casualties from being 3,000/2968 = 101 percent higher, or 2.011 times as high. If batteries intercepted 66.71 percent (the mean estimate) of hazardous rockets, they prevented casualties from being 3.004 times or 200 percent higher. Zucker and Kaplan (2014) estimated that civil defenses kept rocket casualties from at least tripling during the 2000–2010 period. If improvements prior to Protective Edge further reduced casualties by the mean estimate of 65.94 percent, then that kept them from being another 2.936 times as high, or 194 percent higher. This implies that civil defenses overall kept casualties from being 3.000 × 2.936 = 8.808 times as high.⁷

For Pillar of Defense, similar calculations imply that preemption prevented 72 percent more casualties. If civil defense improvements reduced casualties by 13.61 percent, then they prevented 16 percent more. If batteries intercepted 10.70 percent of hazardous rockets, then they prevented 12 percent more casualties. If rocket inaccuracy was about halfway in between that of Second Lebanon and Protective Edge, then it prevented 327 percent more casualties. For Cast Lead, preemption prevented 194 percent more casualties.

Figure 8 displays these percentage casualties avoided. The figure illustrates the importance of rocket inaccuracy and civil defenses. If attackers ever manage to improve rocket accuracy or negate civil defenses, then Israel's losses would sharply increase. By contrast, preemptively destroying rockets on the ground had surprisingly little effect on casualties, except during Cast Lead. This is because each rocket had little chance of success. For example, during Protective Edge only 0.2409 x (1 – 0.6671) ≈ 8 percent would have flown accurately and evaded interception, and even those would have harmed relatively few people due to civil defenses.

Higher Potential Losses

These multipliers also reveal how much greater the rocket danger would have been without Israel's countermeasures. For example, during Protective Edge the country suffered 85 rocket casualties. That number could have been 8.808 times larger if Israel had not built civil defenses, 3.004 times larger had it not developed Iron Dome, and 2.011 times larger had it not destroyed rockets in Gaza. Rocket casualties therefore could have been 8.808 × 3.004 × 2.011 = 53.21 times as high, or around 4,523 people, a figure triple that of Second Lebanon.⁸

Critics should keep these multipliers in mind when judging the proportionality of Israel's actions. While such judgements should consider the relatively low losses that the country actually suffered, they should also consider the much greater ones that could have occurred absent its extensive countermeasures.

Expensive Military Operations

IDF operations publicly aimed to protect Israeli civilians from rocket attacks. Each operation destroyed rockets and related facilities, but at considerable cost. For example, Pillar of Defense destroyed 980 rockets and cost $285 million, or $291,000 per rocket. That saved Israel from rocket losses 72 percent higher, that is, 177 casualties and $10.7 million in damage, for a marginal cost of (285 – 10.7)/177 = $1.55 million per casualty avoided.

Cast Lead rocket destructions avoided 194 percent higher losses: 280 civilian casualties and $15.8 million in damage. The IDF spent (910 + 1,280)/2 = $1,095 million and suffered 80 military casualties. Prorating of rockets and mortars assigns the rockets 77.6 percent: $850 million and 62 military casualties. The cost to prevent each civilian casualty was $2.98 million plus 0.221 military casualties, or $3.83 million after netting civilian and military casualties.

Protective Edge prevented 101 percent higher losses: 86 casualties and $29.5 million of damage. It cost (1900 + 2600)/2 = $2,250 million and 536 military casualties; prorating assigns the rockets 75 percent, or $1,688 million and 402 military casualties. Each prevented civilian casualty cost $19.3 million plus 4.67 military casualties, indicating the operation incurred many more casualties than it prevented.

These rough calculations include only the operations’ direct marginal costs and benefits and assume the only goal was to prevent rocket losses. They omit many costs that are not meaningful to allocate per-rocket or per-casualty, such as the IDF's sunk costs of purchasing weapons and training soldiers, and its fixed costs of maintenance and salaries. They also exclude Israel's large indirect economic losses. The calculations similarly neglect the operations’ other outcomes, such as damage to militant tunnels and civilian buildings.

The calculations ignore deterrence of Hamas because it seemed negligible in this period. Each operation was followed by another a few years later, with hundreds of rockets fired in between, a routine called “mowing the grass” (Inbar and Shamir 2014). If partial deterrence were applicable, one could add the estimated losses prevented between operations, but also the costs of supporting the IDF deterrent force.⁹

Competitively Priced Interceptions

Israel budgeted $225 million of American aid to replenish its interceptor supply after Protective Edge (Dagoni 2014). Prorating implies $182.1 million for the 595 rocket interceptions, or $306,000 each. By keeping rocket losses from being 143 to 294 percent higher, interceptions prevented 122 to 250 casualties, and $41.7 to $85.7 million in damage. This puts the marginal cost per prevented casualty between $386,000 and $1.151 million.

The country similarly spent $200 million on resupply after Pillar of Defense (Lev 2012). This equates to $475 thousand for each of the 421 claimed interceptions, or $1.18 million each if there were only 326 – 147 = 179 effective interceptions. If interceptions prevented at most 47 percent higher losses (i.e. $6.9 million in damage and 115 casualties) then the net marginal cost exceeded $1.68 million per casualty avoided. If they only prevented 12 percent higher losses, then that cost was $6.84 million apiece.

These estimates assume the funding only replaced previously fired Tamirs. If, contrary to the news reports, it also increased stocks beyond their original levels, then the marginal cost would be lower. Conversely, the estimates exclude the capital costs of developing the technology and building the batteries, as well as their ongoing support costs.

An alternative “bottom-up” costing approach could multiply the cost per interceptor by the number fired, but this would require knowing the number of Tamirs fired, the price paid, and any marginal expenses for delivery, launcher refurbishment, etc. For example, if the 799 engagements during Protective Edge each used 1.5 Tamirs on average, and if each Tamir cost $100,000, then the total would have been at least $120 million.¹⁰

Durable Civil Defenses

Unlike infantry and interceptors, civil defense usage involves negligible marginal costs, making it very cost-effective by that measure. Israel's civil defense improvements prevented up to 138 casualties during Pillar of Defense and an additional 113 to 252 during Protective Edge. Civil defenses also protect against mortars and other attacks.

Discussion

One implication of this study is that the controversy over Israel's military operations is partly misplaced. On the one hand, its justification for responding militarily is better than some critics admit, as the rocket attacks pose substantial potential danger. Israel's losses have been relatively low precisely because of its multilayered defenses. On the other hand, the benefits from the military operations seem not to have been worth their costs (Perry 2014). As such, analysts should devote more attention toward evaluating the productivity of military operations, as opposed to focusing singularly on their proportionality.

Another implication is that the controversy over Iron Dome has been overly polarized. While the system was perhaps “little more than a bluff” (Globes 2014) originally, it has become a valuable shield (Elis 2012). However, it is not a “game-changer that heralds the end of rockets” (Hamilton 2012). Instead, Iron Dome represents another step within an ongoing arms race (Shapir 2013b). Hamas has rebuilt its rocket stockpile since Protective Edge (JPost 2017), while Hezbollah has amassed more than 120,000 rockets (Ahronheim 2017a) and additional guided missiles and drones. Given sufficient quantity and accuracy, rockets could overwhelm a battery and greatly increase Israeli losses (Armstrong 2014a).

This relates to the increasing importance of the Israeli-Palestinian conflict's economic dimension. Israel's defenses have greatly reduced fatalities and casualties and somewhat reduced direct property damage, but done little to curb indirect economic losses. Furthermore, casualties and fatalities are light only because the country has spent billions on defenses. All these costs detract from Israeli prosperity and complicate defense policies and spending priorities (Kober 2013; Shapir 2013b).

This article has analyzed the actual performance of rocket countermeasures, but their perceived performance also matters. Despite the effectiveness and efficiency of civil defense, some Israelis may see it as overly passive or even defeatist. By contrast, military operations are assertive responses that might satisfy domestic political pressures or signal determination to foreign enemies, regardless of their actual impacts (Shapir 2013b).

Public perception is similarly complicated for Iron Dome. Even if interception became influential only during Protective Edge, it was perceived as a game changer during Pillar of Defense and boosted civilian morale during both operations. Consequently, civilians now demand interceptor protection, even if military facilities sometimes ought to have priority. That perception might reduce political pressure for offensives against Gaza, but also for negotiating peace. It also may have fostered a complacent belief that interception will always work, even against Hezbollah, whose attacks could include 1,000 rockets per day (Ahronheim 2017a; Cohen 2016; Limor 2017).

On the whole, Israel seems to have a complicated menu of countermeasures at its disposal. Civil defense seems to provide the greatest actual benefits, by preventing many casualties and fatalities while also being cost-effective. However, it does not prevent property damage, and its passivity may be unattractive domestically.

Interception occupies a middle position. Iron Dome now seems able to provide substantial rocket protection at a lower cost than airstrikes. As an active defense, it offers a positive public image at home and abroad. However, future attacks might overwhelm or bypass the batteries, and political pressures could complicate their future deployment.

Military operations incurred the highest costs and prevented the fewest losses.¹¹ Airstrikes were more cost-effective than ground assaults, while the latter were largely inefficient or even counterproductive. However, ground operations carry an assertive image, and my analysis indicates they were the only option that at least temporarily reduced rocket fire rates.

These rocket challenges exist in the context of intensifying missile proliferation around the Middle East. Israel faces missile threats from Syria and especially Iran, and the latter's missiles have increased in size and quantity, though not necessarily in accuracy (Siryoti 2017; Harel 2017). Israel therefore has deployed two more interceptor systems: David's Sling for medium range missiles, and Arrow for long range ones (Ahronheim 2017b). Iran also poses a threat to Saudi Arabia, which has already suffered missile and drone attacks from Houthi militants in Yemen (Frantzman 2017).

Other countries are also interested in Israel's interceptors. The United States so far has supplied $1.3 billion for Iron Dome and $1.7 billion for Israel's other missile defense efforts and has committed to providing $500 million annually for the next decade (White House 2016). It might eventually adapt some of the Israeli technology for its own antimissile programs. Buyers of Israel's interceptors will likewise care how well they perform. South Korea and Singapore have been mentioned as potential candidates, though so far only Azerbaijan has purchased any (JPost 2016; Armstrong 2017).

Supplemental Information

Supplementary information is available at the Journal of Global Security Studies data archive.

Footnotes

References