Professional Documents
Culture Documents
lemaire@wharton.upenn.edu
Abstract. The 2000 US life expectancy only ranks 28th among the worlds 34 richest countries. At 76.9 years, it stands 2.29 years below the average of these affluent countries. We investigate one of the possible causes of this life expectancy gap: the widespread availability of firearms and the resulting high number of US firearm fatalities: 10,801 homicides in 2000. The European Union, with a total population of 376 million, experienced 1,260 homicides. Japan, with a population of 127 million, only reported 22 homicides. Using multiple decrement techniques, we show that firearm violence shortens the life of an average American by 104 days (167 days for white males, 362 days for black males). Among all accidental deaths, only motor vehicle accidents have a stronger effect. The elimination of all firearm deaths in the US would increase the overall life expectancy more than the total eradication of all breast and prostate cancers. We also show that the insurance premium increases paid by Americans as a result of firearm violence are probably of the same order of magnitude as total medical costs due to gunshots or the increased cost of administering the criminal justice system due to the availability of firearms.
1. Violent Deaths due to Firearms: A US Phenomenon? Life expectancies have steadily increased throughout the 20th century in the United States, reaching 74.1 years for males and 79.5 years for females in 2000. The gap between the life expectancies of Caucasians and African-Americans remains significant, and is not decreasing: 74.8 years vs. 68.2 years for males, 80.0 vs. 74.9 for females. Levine et al (2001), noting that there has been no sustained decrease in black/white inequalities in mortality rates since 1945, expect the disparity in life expectancy to either remain stable or increase over the next decade. While improving, life expectancies in the US lag compared to other industrialized countries. The US only ranks 27th for males and 28th for females among the 34 countries that had a GDP per capita, expressed in purchasing power, in excess of $20,000 in 2000. Interestingly, the rank correlation coefficient between life expectancy and GDP per capita among these 34 countries is a non-significant 0.062: once a prosperity threshold has been reached, income does not influence life expectancy any more1. Worldwide, the US ranks 48th in life expectancy. Andorrans live the longest even though their GDP per capita is below $20,000: 80.58 years for males, 86.58 years for females. The worldwide rank correlation coefficient between life expectancy and GDP per capita stands at 0.814. This research focuses on a potential cause of the relatively low life expectancy in the US: the widespread availability of guns and the resulting huge number of firearms fatalities. The US is a country that has chosen to live with guns, and now must face the consequences of this choice. In 2000, 28,663 people in the US died from firearm injury:
Spearmans rank correlation coefficient was deemed more appropriate than Pearsons correlation coefficient as (i) it is less sensitive to outliers in a small sample: and (ii) a graph of life expectancies vs. GDP per capita showed a non-linear relationship.
1
16,586 from suicide, 10,801 from homicide, 776 unintentional, 270 from legal intervention, 230 undetermined. 2000 was only the second year since 1971 that firearm deaths numbered below 30,000. The rate of firearm fatalities in the US far exceeds that of other industrialized countries: gun violence is a huge tax on US quality of life. Figure 1 compares the crude firearm death rate per 100,000 inhabitants in countries that have a population in excess of 3.8 million and a GDP per capita, adjusted for purchasing power, in excess of $20,0002. The death rate in the US is about double the next highest rate (Finland) and 2.5 to 37 times the rate in other European affluent countries. The rate of firearms deaths in Asian countries is negligible: only 6 deaths were reported in Singapore and in Hong Kong. A comparison of age-adjusted death rates (such as Fingerhut et al, 1998) would provide essentially the same results. Figure 2 compares homicide firearm death rates. The US rate dwarfs the rate of any other industrialized country. It is 5.5 times the next highest rate (Italy). Several European countries have a firearm homicide rate that is insignificant: only 45 firearm homicides were reported in the UK, 15 in Denmark, 10 in Norway, 7 in Ireland. A total of 1,260 homicides occurred in the 15 countries of the European Union (vs. 10,801 in the US), a region with a total population of over 376 million inhabitants. Only twenty-two firearm homicides were reported in Japan (population: 127 million), a country where it has never been possible for private individuals to acquire a handgun except under extraordinary circumstances.
These countries all have reliable mortality surveillance systems. The data from a few countries do no enable to separate deaths caused by explosives from firearm deaths, resulting in a slight overestimation of firearm mortality rates. Missing data: Switzerland.
These figures suggest that the widespread availability of guns in the US increases the number of homicides and suicides due to firearms. Cook (1981) estimates that every 10,000 guns sold are involved in about 3,000 robberies and 100 homicides. Firearms deaths do not affect all classes of the population uniformly. In almost every region of the world, but especially in the US, adolescents and young males are the primary victims and perpetrators of violence. In the US, violent deaths hit the AfricanAmerican population disproportionately. Table 1 compares the 2000 total number of firearms deaths, the population, and the crude rate of firearm deaths per 100,000 inhabitants for all major racial groups. Table 1: Firearm deaths and crude rates per race and gender RACE AND GENDER White Males White Females Black Males Black Females Native Males Native Females Asian Males Asian Females Total # DEATHS 17,750 3,195 6,284 770 196 44 352 72 28,663 POPULATION 111,196,305 115,055,528 16,776,358 18,527,393 1,206,143 1,230,010 5,446,867 5,826,395 275,264,999 RATE 15.96 2.78 37.46 4.16 16.25 3.58 6.46 1.24 10.41
Table 2: Firearm homicides and crude rates per race and gender RACE AND GENDER White Males White Females Black Males Black Females Native Males Native Females # DEATHS 3,686 1,120 5,084 615 70 16 POPULATION 111,196,305 115,055,528 16,776,358 18,527,393 1,206,143 1,230,010 RATE 3.31 0.97 30.30 3.32 5.80 1.30
166 44 10,801
Table 3: Firearm suicides and crude rates per race and gender RACE AND GENDER White Males White Females Black Males Black Females Native Males Native Females Asian Males Asian Females Total # DEATHS 13,214 1,960 965 125 109 24 166 23 16,586 POPULATION 111,196,305 115,055,528 16,776,358 18,527,393 1,206,143 1,230,010 5,446,867 5,826,395 275,264,999 RATE 11.88 1.70 5.75 0.67 9.03 1.95 3.05 0.39 6.03
Firearms not only kill many people in the US, they kill young people. Table 4 presents the average age at death, when death is due to homicide or suicide, by race and gender. Table 4: Average age at death due to homicide and suicide per race and gender RACE AND GENDER White Males White Females Black Males Black Females HOMICIDE 32.45 39.29 28.23 31.87 SUICIDE 48.93 46.04 36.89 38.44
From tables 2 to 4, it can be concluded that firearm homicide is a problem that disproportionately targets young black adult males. The age-adjusted, gender specific, black/white mortality ratio for homicides, that was 7.08 for males in 1979, has barely reduced to 6.73 in 1998 ( Levine et al, 2001.) Firearm suicides mostly affect middleclass, middle-aged, white males, and more and more so: the black/white mortality ratio for suicides decreased from 0.67 to 0.57 from 1979 to 1998. 5
Of course, deaths at early ages have a profound effect on life expectancies. While the discovery of a new drug or procedure slowing down the effect of prostate cancer would only have an insignificant effect on life expectancies, a drastic reduction in violent firearm deaths would add many years of life to potential victims and possibly lead to a significant improvement in US life expectancies. The purpose of this article is to
evaluate the cost of firearm violence in terms of life expectancies. We use actuarial multiple-decrement techniques to estimate the reduction of life expectancy, by race and gender, due to firearm deaths. In other words, we calculate the potential gain in life expectancy that would result from a total elimination of firearms deaths. We then estimate the fraction of the total premiums for term and whole life insurance due to firearm deaths. Many studies attempt to estimate the direct cost of firearm violence in terms of medical treatment (Cook et al, 19993, Cook and Ludwig, 2000). Other costs are more difficult to quantify. They include the cost of public resources devoted to law
enforcement, private investment by individuals in protection and avoidance4, lost productivity of victims, and changes in the quality of life: limits on freedoms to live or work in certain places, restrictions on residential and commercial location decisions, limitations in hours of operations of retail establishments, emotional costs to the forced adaptation to increased risk, and the cost of pain and fear. Cook and Ludwig (2000) estimate the aggregate cost of gun violence in the US at about $100 billion annually, or about $360 for every American. Other costs yet to be evaluated included the loss of
Cook et al estimate the mean cost of a firearm injury at about $17,000, in 1994 dollars. With about 135,000 fatal and non-fatal injuries per year, the total annual medical cost of gunshot injuries is around $2.3 billion, with about half this cost borne by US taxpayers. 4 $1,800 annually per American household, according to one estimate (Anderson, 1999).
prestige of the US in the international scene or the rejection of the US as a model society to emulate. This article emphasizes costs that the scientific literature has yet to consider: the increased cost of essential insurance policies, and the cost in terms of life expectancy reduction. Section 2 summarizes the calculation of life expectancies under different scenarios. Section 3 describes the sources of the data used in this study. Results concerning changes in life expectancies are derived and discussed in section 4. Section 5 presents changes in life insurance premiums. 2. Estimation of Life Expectancies
( ( Consider a life table {l x ) , x = 0,1,2,..., w}, where l x ) is the number of individuals
( alive at exact age x. l 0 ) , the radix of the table, is chosen arbitrarily, usually 100,000. w
is the last possible age at death. The probability that an individual age x is still alive at age x+n is
( ) n px = ( l x+)n . The complementary probability that an individual alive at age x ( l x )
ex =
( p x ) dt. It is the expected number of years lived from age x. The life expectancy
0
at birth e0 is the most commonly used measure to compare levels of public health among different populations, as it summarizes mortality at all ages. It is suitable to make comparisons among populations with different age structures, as it is not affected by the age distribution. It is the most accurate measure of the quality of life in a society, summarizing in a single number all the natural and man-made damages that can affect an individual, ranging from poor health care system and civil war to unhealthy nutrition and sexual behavior. The potential gain in life expectancy is considered to be the best measure to evaluate the impact of the total eradication of a disease or condition. Another
widely used measure, the number of years of potential life lost, suffers from a failure to correctly account for the effect of competing risks; it is also heavily influenced by the age structure and the total population size (Lai and Hardy, 1999.) The variance of age at death can be calculated as 2 al, 1977).
( The annual probability of death q x ) can be decomposed in two terms: ( ( ( ( q x ) = q x1) + q x2) , where q x1) denotes the probability that an individual age x dies from a 2
0 t t p dt e0 (Bowers et
( ) 0
( firearm injury within a year, while q x2) is the probability to die from any other cause.
( ( q x1) and q x2) are called dependent probabilities, reflecting the crucial feature of multiple
decrement theory that all causes of death interact to produce a societys mortality pattern.
( q x1) depends not only on firearm injuries but also on all other causes of death. In order to
die from a firearm injury, an individual first has to survive all other causes of death. Paradoxically, an improvement of HIV-related mortality a disease often prevalent in the same subgroups of society as firearm violence- would actually increase the number of firearm deaths. It is possible to calculate independent probabilities or net probabilities of decrement with appropriate assumptions concerning the distribution of deaths within each year of age, but in this case they are not required given the available data.
( ( ( As the data enable the estimation of q x ) and q x1) , q x2) can be obtained by simple
( difference, and the life table l x 2 ) describing mortality patterns after exclusion of firearm ( deaths derived. Integration of t p x2 ) then provides life expectancies when firearm deaths
are removed.
3. Data
Excellent data are available to calculate the impact of firearms on life expectancies. The 2000 US Life Tables, published by the National Center for Health Statistics, Center for Disease Control and Prevention, following the decennial census, are
( used to calculate annual death probabilities q x ) for the entire US population, as
well as gender- and race-specific probabilities. 2000 population figures, by age, gender, and races, are available from the 2000 decennial census. The National Center for Injury Prevention and Control, Center for Disease Control and Prevention, publishes, through its Web-based Injury Statistics Query and Reporting System, an annual Injury Mortality Report that provides the number of deaths by cause, age, gender, and race. Along with census population figures, this report provides the required data to calculate annual firearm death
( probabilities q x1) . Three sets of probabilities are calculated, related to homicides,
suicides, and all causes. The National Cancer Institute, through its SEER program (Surveillance, Epidemiology, and End Results,) provides probabilities of dying of cancer, by sex, age, race, and site of cancer. The World Report on Violence and Health, World Health Organization, provides international firearm-related mortality, by manner of death and country, for the most recent year available between 1990 and 2000.
The World Factbook from the Central Intelligence Agency provides the GDP per capita, adjusted for purchasing power, for all countries, as well as international population figures and life expectancies.
For the US population and its main sub-groups, table 5 provides the current life expectancy, the life expectancy if all firearm homicides were eliminated, if all firearm suicides were eliminated, if all firearm deaths were eliminated. Note that this table assumes that there is no substitution effect: firearm deaths are not replaced by violent deaths by another mean. This important assumption is discussed at the end of this section. Table 5. Life expectancies under different scenarios POPULATION US US Males US Females White Males White Females Black Males Black Females CURRENT HOMICIDES 76.9 77.03 74.1 74.30 79.5 79.54 74.8 74.91 80.0 80.03 68.2 69.01 74.9 75.00 SUICIDES 77.04 74.33 79.54 75.08 80.05 68.34 74.92 ALL 77.18 74.56 79.58 75.21 80.09 69.19 75.02
Table 6 presents the reduction in life expectancy, in days, due to firearm deaths, for the different groups and causes of death. The main result is that the average American loses 103.8 days of life due to firearm deaths. The average white male loses five months, the average black male nearly one full year. Table 6. Reduction in life expectancies under different scenarios, in days
10
POPULATION US US Males US Females White Males White Females Black Males Black Females
In addition to increasing the life expectancy, the elimination of firearm deaths would also reduce the variance of the age at death, a desirable outcome since this variance summarizes inequality in the presence of death. Currently, this variance stands at 275.90. Eliminating all homicide firearm deaths would reduce it to 270.46; eliminating all suicides, to 271.71; eliminating all firearm deaths, to 265.69. For comparison purposes, table 7 presents the 2000 number of deaths and reductions in life expectancy in days due to various injuries, for an average American. It shows that only motor vehicle accidents are more costly than firearm violence in terms of reduced life expectancy.
Table 7. Number of deaths and reduction in life expectancy, in days, due to various injuries INJURY Firearms Motor vehicle accidents Other transportation Adverse effects of medical care and drugs Drownings Falls Fires Poisoning Suffocation Contacts with object # DEATHS 28,663 43,354 1,413 3,059 4,073 14,002 3,907 20,230 12,098 1,292 REDUCTION 103.8 160.7 4.9 6.4 17.2 25.4 13.1 66.5 39.1 4.5
11
Machinery 676 2.2 Table 8 presents, for the major sites of cancer, the increase in life expectancy, in days, that would result from a total eradication of the cancer, as well as the proportion of Americans who die from each cancer. For breast and ovarian cancers, the proportion of female deaths is indicated. For prostate cancer, the table provides the proportion of male deaths. For all other cancers, male and female deaths are combined. The table
demonstrates that the effect of firearm violence, in terms of reduction in life expectancy, is higher than the combined effect of all breast and prostate cancers. The elimination of all firearm deaths would increase the overall life expectancy more than the total eradication of all breast and prostate cancers. Table 8. Increase of life expectancy, in days, from total eradication of cancer, and percentage of Americans who die from cancer CANCER SITE Lung Breast Colon Prostate Ovaries All cancers INCREASE 205.7 73.3 70.1 25.7 22.6 856.5 % OF DEATHS 5.42 (M+F) 3.09 (F) 2.32 (M+F) 3.20 (M) 1.04 (F) 21.26 (M+F)
Discussion: The previous tables have been established under the crucial assumption of
total elimination of firearm deaths. This raises the question of the existence of a substitution effect. In countries where guns are not easily available, is there a substitution process, where individuals commit suicide and kill each other with other means and weapons? In the case of homicides, many studies show that there is no substitution effect. Clarke and Mayhew (1988) find that the rate of homicides committed with a handgun in
12
the US is 174.6 times the rate observed in England and Wales. The US rate of all gun homicides is 63.4 times the UK rate. Despite that, the rate of homicides committed by other means is 3.7 times higher in the US. Sloan et al (1988) compare the crime rates in Seattle and Vancouver, two very similar cities in terms of population, climate, income per household, poverty and unemployment rates. The only significant demographic difference lies in the racial composition of minorities, with more Asians in Vancouver and more Hispanics and Blacks in Seattle. Gun regulations are much stricter in Vancouver. In Seattle, handguns may be purchased legally for self-defense in the street or at home. After a 30-day waiting period, a permit can be obtained to carry a gun as a concealed weapon. Handguns need not be registered. In Vancouver, self-defense is not considered a valid or legal reason to carry a gun. Concealed weapons are not permitted. The purchase of a gun requires registration and a restricted-weapons permit. Handguns can be transported in a car, but only if stored in a locked box in the trunk. As a result, an estimated 41% of Seattle inhabitants own a gun compared to only 12% of Vancouver inhabitants. The authors find that the two cities essentially experience the same rates of burglary, robbery, homicides and assaults without a gun. However, in Seattle the rate of assault with a firearm is 7 times higher than in Vancouver, and the rate of homicide with a handgun is 4.8 times higher. The authors conclude that the availability of handguns in Seattle increases the assault and homicide rates with a gun, but does not decrease the crime rates without guns, and that restrictive handgun laws reduce the homicide rate in a community.
13
Killias (1993) uses gun ownership figures obtained through 28,000 telephone interviews in 14 countries: Australia, Canada, the US, and 11 European countries. The proportion of homes owning guns varies widely, from 1.9% in the Netherlands to 48% in the US (with large variations within the US, from 29% in the east to 54% in the south). The author then computes Spearman rank correlation coefficients and obtains highly significant correlations between the percentage of households owning guns and the rate of homicide with a gun (correlation = 0.746), the rate of suicide with a gun (0.900), the overall homicide rate (0.658) and the overall suicide rate (0.515). If a substitution effect exists, the correlations between gun ownership percentages and rates of homicides and suicides by other means than a gun would be significantly negative. They are not (resp. 0.441 and 0.015, both non-significant). The correlation of 0.441, although not
significant with this small sample, suggests that the number of homicides by means other than a gun increases with raising levels of gun ownership. In other words, the data do not support at all the existence of a compensation effect for homicides. A substitution effect probably exists for suicides. Some despondent individuals contemplating suicide who have no access to a firearm may attempt to take their life by another mean. A recent study by Shenassa et al (2003) allows an estimation of this substitution effect. The authors study all 10,287 completed suicides and all 37,352 hospital admissions due to attempted suicide recorded in Illinois between 1990 and 1997. Table 9 provides, for all males, the percentage of all episodes coded with a particular suicide method5, as well as the percentage of fatal cases for each method. Table 10
Poisons = all episodes involving prescription drugs, over the counter drugs and toxic substances such as gasoline and household cleaning substances. Suffocation = episodes involving hanging, strangulation, and suffocation. Cuts = all episodes involving cutting or piercing instruments. Crash/jump = episodes
14
provides the corresponding figures for females. Noteworthy is the extreme lethality of firearm suicide attempts (a fatality rate of 96.5% for males, 96% for females.) Other techniques are relatively inefficient, due to a significant part on the success of public policy measures such as reductions in the carbon monoxide content of domestic gas, the development of less toxic sleeping pills and antidepressants, and restrictions in the prescription of potentially lethal drugs.
Table 9. Percentage of cases and fatality of suicide methods, males METHOD Firearms Poisons Suffocations Cuts Crash/jumps Exposure Other % OF TOTAL 20.2 58.8 9.6 6.1 2.2 0.5 2.5 % FATAL 96.5 11.1 90.7 9.9 76.1 64.0 2.4
Table 10. Percentage of cases and fatality of suicide methods, females METHOD Firearms Poisons Suffocations Cuts Crash/jumps Exposure Other % OF TOTAL 2.3 87.8 1.8 4.1 0.8 0.3 2.9 % FATAL 96.0 3.9 89.0 2.6 68.8 43.5 1.4
Excluding firearms, the average fatality rate of all other suicides techniques is 22.42% for males, 5.99% for females. Assume that, in the absence of a firearm, suicide candidates select an alternative technique in such a way that the relative proportion of
involving crash into a moving object or jump from a high place. electrocution or exposure to heat or cold. Exposure = episodes involving
15
each technique remains stable. The unavailability of a firearm then would reduce the fatality rate of firearm suicides from 96.5% to 22.42% for males, and from 96% to 5.99% for females. These figures probably overestimate the substitution effect in a significant way. Indeed, they assume that every single firearm suicide attempt is replaced by an attempt using another mean. Also, Shenassa et al only consider suicide attempts that result in death or hospitalization. This overestimates the lethality of non-firearm techniques: while the vast majority of firearm attempts result in a completed suicide or hospitalization, it is likely that a significant percentage of unsuccessful attempts involving poisons or cuts do not result in hospital treatment. Suicide attempts by other means than firearms are thus undercounted, and their lethality is consequently overstated. Table 11 revises table 6, by introducing a substitution effect for suicides. It presents the reduction in life expectancy for the various subgroups of the US population, assuming (i) no substitution effect for homicides; and (b) a substitution effect for suicides that reduce the lethality of firearm suicides to 22.42% for males and 6.99% for females.
Table 11. Reduction in life expectancies under different scenarios, in days, assuming no substitution effect for homicides, and a substitution effect for suicides.
POPULATION US US Males US Females White Males White Females Black Males Black Females
16
Conclusion: The average life expectancy in the US is 76.9 years. The average life
expectancy, weighted by population, of the other 33 richest countries in the world is 79.19 years. The US thus suffers from a life expectancy gap of 2.29 years. Table 5 indicates that 0.28 years of this gap is due to firearm deaths, if no substitution effect is introduced. It is concluded that 12.2% of the US lag can be explained by the availability of firearms and the resulting number of fatalities. Breaking down calculations by gender, 2.8% of the female life expectancy gap and 23.5% of the male gap is explained by firearm deaths. With the substitution effect described in this section, 11.5% of the US lag is explained by firearm fatalities (2.7% for females, 20.7% for males.) Within the US, 10.66% of the life expectancy differential between white and black males is due to firearm homicides, a result that confirms Potter (2001). The corresponding figure for females is only 1.29%. Excess firearm suicides among whites reduce the male life expectancy difference by 2.15% and the female difference by 0.58%.
5. Changes in Term Life Insurance Premiums
1 Table 12 compares the net single premiums 1,000 A25:20 of a $1,000, 20-year, fully
discrete term insurance policy on a life age (25) at issue, under the different scenarios: under the current situation, after elimination of all firearm homicides, all firearm suicides, and all firearm deaths. An annual rate of interest of 5% is used. Table 12. Cost of $1,000 20-year term insurance, age at issue 25, under different scenarios POPULATION US US Males US Females White Males CURRENT $17.16 $22.87 $11.48 $20.67 HOMICIDES $16.35 $21.53 $11.22 $19.95 SUICIDES $16.37 $21.54 $11.24 $19.07 ALL $15.48 $20.07 $10.97 $18.21
17
Table 13 presents the percentage discounts that could be awarded if firearm deaths could be eliminated. Table 13. Discounts, 20-year term insurance, age at issue 25, under different scenarios POPULATION US US Males US Females White Males White Females Black Males Black Females HOMICIDES 4.70% 5.88% 2.20% 3.48% 2.03% 15.84% 3.26% SUICIDES 4.61% 5.85% 2.03% 7.77% 2.88% 2.56% 0.59% ALL 9.78% 12.28% 4.42% 11.91% 5.13% 19.02% 3.98%
Table 14 compares the net single premiums 1,000 A25 of a $1,000, 20-year, fully
discrete whole life insurance policy on a life age (25) at issue, under the different scenarios. Table 14. Cost of $1,000 whole life insurance, age at issue 25, under different scenarios POPULATION US US Males US Females White Males White Females Black Males Black Females CURRENT $ 98.11 $111.36 $ 85.32 $107.57 $ 82.59 $146.69 $108.37 HOMICIDES $ 97.33 $110.08 $ 85.06 $106.87 $ 82.37 $140.93 $107.68 SUICIDES $ 97.14 $109.74 $ 85.03 $105.65 $ 82.24 $145.65 $108.23 ALL $ 96.28 $108.34 $ 84.75 $104.81 $ 82.00 $139.66 $107.52
Table 15 presents the percentage discounts that could be awarded if firearm deaths could be eliminated.
18
Table 15. Discounts, whole life insurance, age at issue 25, under different scenarios POPULATION US US Males US Females White Males White Females Black Males Black Females HOMICIDES 0.80% 1.14% 0.31% 0.65% 0.26% 3.92% 0.63% SUICIDES 0.99% 1.45% 0.33% 1.79% 0.42% 0.71% 0.13% ALL 1.87% 2.71% 0.67% 2.57% 0.71% 4.79% 0.78%
Conclusion: According to the American Council for Life Insurance (2001,) there were
148 million group and 35 million individual term life insurance policies in force at the end of 2000. There were 125 million group and 8 million individual whole life policies in force. The total annual premium income was $31,589 million in term and $98,289 million in whole life. It is tempting to apply the US population discounts of 9.78% for term (table 9) and 1.87% for whole life (table 11) to these total premium volumes to get an estimate of $4.9 billion of the insurance cost of firearm violence. This calculation overstates costs, as these discounts were obtained using US population life tables, and the mortality of insured lives may markedly differ from population mortality6. Also, an age at issue of 25 results in high discounts, as homicides mostly affects young adults. Still, this extrapolation suggests that increased insurance costs due to firearms are probably of the same order of magnitude as total medical costs ($2 to $2.3 billion, Cook and Ludwig, 2000, Cook et al, 1999) or the increased cost of administering the criminal justice system, including incarceration costs ($2.4 billion, Cook and Ludwig, 2000.)
Note, however that, while the teenage victim of a drug-related shooting incident is unlikely to be insured, the middle-age, middle-class, white male who commits suicide most probably has employer-provided term insurance coverage, and possibly self-purchased term or whole life insurance.
6
19
References
American Council for Life Insurance, 2001, Life Insurance Fact Book 2000 (Washington, D.C.:ACLI).
Andersen, David A., 1999, The Aggregate Burden of Crime, Journal of Law and
Economics, 42: 611-42.
Bowers, Nelson, Hans Gerber, James Hickman, Donald Jones, and Cecil Nesbitt, 1997,
Actuarial Mathematics (Schaumburg, IL: Society of Actuaries).
Clarke, Ronald and Pat Mayhew, 1988, The British gas suicide story and its criminological implications, Crime Justice, 10: 79-116.
Cook, Philip J., 1981, Guns and Crime: The Power of Long Division, Journal of Policy
Analysis and Management, 1: 120-125.
Cook, Philip J., Bruce A. Lawrence, Jens Ludwig, and Ted R. Miller, 1999, The Medical Costs of Gunshot Injuries in the United States, Journal of the American Medical
Association, 282: 447-454.
Cook, Philip J. and J. Ludwig, 2000, Gun Violence. The Real Costs (Oxford, UK: Oxford University Press)
20
Fingerhut, Lois A., Christine S. Cox, and Margaret Warner, 1998, International
Comparative Analysis of Injury Mortality, Advance Data, Vital and Health Statistics,
Killias, Martin, 1993, International correlations between gun ownership and rates of homicide and suicide, Canadian Medical Association Journal, 148: 1721-1725.
Lai, Dejian and Robert J. Hardy, 1999, Potential gains in life expectancy or years of potential life lost: impact of competing risks of death, International Journal of
Epidemiology, 28:894-898.
Levine, Robert S, James E. Foster, Robert E. Fullilove, Mindy T. Fullilove, Nathaniel C. Briggs, Pamela C. Hull, Baqar A. Husaini, and Charles H. Hennekens, 2001, BlackWhite Inequalities in Mortality and Life Expectancy, 1933-1999: Implications for Healthy People 2010, Public Health Reports, 116: 474-483.
Louise Potter, 2001, Influence of Honicide on Racial Disparity in Life Expectancy United States, 1998, Morbidity and Mortality Weekly report, Center for Diseases Control, 50 (36), 780-783.
Sloan, John, Arthur Kellermann, Donald Reay, James Ferris, Thomas Koepsell, Frederick Rivara, Charles Rice, Laurel Gray, and James LoGerfo, 1988, Handgun regulation, crime,
21
assaults, and homicide. A tale of two cities, New England Journal of Medicine, 319: 1256-1262.
Shenassa, E.D, S.N. Catlin, and S.L. Buka, 2003, Lethality of firearms relative to other suicide methods: a population based study, Journal of Epidemiology and Community
Health, 57: 120-124.
22