Phnom Penh is the only large city in Cambodia and a major entrepot port attracting a substantial proportion of the country’s markets economic growth and investment. The city has grown rapidly as investors (mostly North Asian and ASEAN based) were attracted by low costs in the wake of the 1989 market reforms that ended the Soviet style command economy. The industrial sector has provided, until recently, the most growth in the economy and much of this occurred in and around the city. The city is a natural draw for the land-less poor who provide labour for large garment, service and light manufactures. Through 1997-1998, there was a sharp decline in investment and growth and a return to high inflation. These conditions severely affected the livelihood of the wage classes and no doubt increased conflicts and risk-taking.
Table 2: Type of Crime as a Proportion of All Recorded Crime 1996
Criminal Offences Cambodia Phnom Penh
Murder 10.3 3.0
Rape and Molestation 2.5 0.1
Assault 19.4 2.9
Armed Injuries 3.7 6.3
Other Against Person 3.4 2.8
Armed Robbery 23.8 38.3
Stealing 26.3 35.9
Other Theftab 7.4 4.2
Otherc 3.1 4.0
All Offences 100.0 100.0
Source: MOI Judicial Police Centre (1998). Notes: (a) grenade attacks, kidnap/abduction and, threats; (b) other offences against the state, cultural thefts, breaches of trust, pick-pocketing and fraud; (c) damage property, illegal weapons, drug and immigration offences.
The pattern of recorded crime shown in Table 2 differ strikingly from that found in most advanced jurisdictions. Property crime and theft account for only 36.8% of recorded crime while offences against the person make up the remainder. The usual ratio of violent to property crime is 1 in 10 or less. In Australia and Hong Kong violence against the person accounts for less than 10% of all recorded crime. Thus, non-lethal crime, especially property crime, is either vastly under-reported or relatively rare in the villages and communes of the Kingdom. Because of the situational determinates of theft risk in villages it is likely that property crime is low in rural areas but this will also be magnified by poor communications with national police. Property crime was more frequent in the city and, violent crime, except armed robbery, occurred more frequently in rural areas as would be predicted by a pre-modern version of the developing state.
In 1996, Phnom Penh recorded nearly a quarter [23.7%] of all recorded crime in Cambodia at over two and half times the rate, at 156.1 per 100,000, of the whole country. Nearly a third (31.6%) of the recorded property theft, 36.6% of the armed robberies and 22% of the pick-pocket, and fraud offences occurred in a city that comprises only 7.6% of the total population. However, only 7.2% of the recorded homicides and 3.5% of assaults occurred in the city suggesting that violence is better regulated but relative wealth and anonymity creates opportunities for theft.
Although independent sources are limited, the data suggests that crime declined in the immediate post-war. However, the withdrawal of UNTAC civil police and investigative support reduced the capability of the RGC police and this presumably impacted on reporting practice25. Even if crime fell in the 2 years after the1993, UN supervised election and the creation of the RGC this decline was quickly reversed. The uneven impact of re-construction, demobilisation, and economic development delivered scant relief to the most vulnerable and displaced. Added to these difficulties was increasing tension within the political coalition which culminated in the coup d'état of July 1997. For a few days Phnom Penh was turned into a battlefield and the Second Prime Minister Hun Sen of the Cambodian People’s Party ousted First Prime Minister Prince Ranridth leader of the royalist Funcinpec Party26. Equally, adverse economic conditions, especially the financial crisis engendered by the July 1997 coup, was severely aggravated by the collapse of the Thai economy and the October 1997 Asian financial crisis. This was followed by intense and frequently violent pre and post 1998 election tension. Such multiple pressures would predict an increase in crime in 1998. It is possible that both decreased police efficiency and post-war fatigue temporarily drove down recorded crime while extreme economic adversity from mid-1997 increased conflicts and stress and drove crime up. We have little means to formally test such explanations and only tentative conclusions can be drawn. Police data for 1998 show a relatively sharp increase from 1996 in all recorded crime including murder and armed robbery although abating in 1999. The very low 1997 figures are discounted because of the acute disruption to policing and recording in the wake of the coup.
The Prevalence of Lethal Violence
The official crime statistics are likely to significantly under-estimate the rate of homicide and other crime. The figures supplied by the MOI are not audited and nor are counting protocols clearly defined. Problematic assumptions are made about the completeness of police data and that recording practice is uniform throughout Cambodia. It is also assumed that the definition of murder and homicide is comparable to that elsewhere and distinctions between murder and other forms of (potentially) criminal death in the MOI sources are retained. Because precise situational and personal details are required to determine if a killing constitutes murder, it is assumed recording practices reflect these important distinctions in a cultural meaningful way. Thus, given these caveats and adjustments for known sources of error an estimate of the prevalence of homicide may be made.
The first source of error is that reported murders record the number of events and not the number of victims. Thus, multiple victim events are recorded as a single incident. To estimate the number of homicide victims we must refer to Khmer newspapers and official references to “losses” to estimate the number of victims. Second, it is unlikely that all homicides are recorded and the sources do not report accidental death, suicide or separately describe the usual homicide categories of attempted murder, infanticide and manslaughter27. Apart from the 542 murders for 1996 other forms, of homicides are recorded. Noted are 19 “intentional killings”, 54 grenade attacks that led to 201 injuries or deaths, 108 offenders shot dead and 36 injured. In addition, a further 53 presumably non-fatal poison and 44 kidnaps were recorded (Judicial Police Centre, July 1998). Notionally the deaths may be classified as homicides although based on cases a proportion of the grenade events are accidental and deaths arising from police action may be justified. Third, the remote provinces of Kratie, Kompong Speu and Mondulkiri, with a combined population of approximately 836,000, returned no reports of criminal offences for 1996 to National Judicial Police and accordingly rates are adjusted.
Finally, it is unclear but likely that the deaths of suspects due to police/military or mob/militia actions are excluded from the murder count. Deaths allegedly motivated by political conflict reported by the UNCOHCHR are usually included. Other acts of intimidation, harassment and non-lethal violence of a political nature may not overlap with official sources and thus may be in addition to the crime rate28. Judicial Police records for 1997 are incomplete and exclude the estimated 40 plus battlefield deaths arising from the July coup conflicts. Many of the approximately 100 post-coup and national elections “political” deaths reported by UNCOHCHR are recorded in 1998 by MOI. A considerable degree of harassment and intimidation related to post-coup conflict and the 1998 election was noted but relatively few of the confirmed “killings” reported by the UNCOHCHR arose from exclusively political conflicts29. Politics may also serve to mask motives that are more mundane and often political and personal rivalries are intertwined. However, the circumstances produce special modes of violence including extreme but diminishing political violence. Raiding activities of the remnants of the Khmer Rouge and other soldier-bandits, for example, cannot be characterised as political or battlefield actions because coherent ideological and command structures no longer motivate these acts of violence.
Until a credible system of death registration and investigation based on international standards30 is established the data on murder and homicide will remain uncertain and the subject of continual controversy. The problem is aggravated by low investigative and forensic scientific capability. The main criminal investigative department has few trained police, no laboratory/mortuary or forensic pathologist and insufficient resources to staff all provinces. Thus available data is not comprehensive and estimates are conservative.
Details on the type or motive of murders and age and sex of victims and offenders are not available from MOI sources. However, analysis of newspapers suggests robbery murder, revenge, quarrels and disputes are the most common reasons for murder. Khmer press coverage is extremely pictorial and has a significant influence on the perception of the risks of homicide. Approximately 61% of victims died because of gunfire, a further 14.6% from knifes and other sharp instruments and 12.5% from unknown causes of death. Most homicide victims, reported by the press, are males usually shot by offenders armed with military weapons. Women make up 11.6% of victims and those 21-29 years old were the most frequently cited age group. Suicides, accidental deaths and traffic accidents are also frequently noted suggesting substantial levels of these forms of mortality are also experienced. Although newspaper reports are an unreliable means of measuring crime, they provide a ratio of deaths to homicide events.
Table 3 describes the victims and fatalities identified in Khmer newspapers from January to early July 1998. This shows 26.7% of the 240 reported deaths involved the death of a suspect or offender at the hands of the police, local “mobs” or militia. One hundred or 41.7% were prima facie murder or manslaughter victims sometimes killed in multiple-victim episodes involving multiple offenders. About 24.2% of the deaths recorded were accidental of which the majority were traffic related and 7.5% were successful suicides. Most of the victims of the category “attempted murder” were survivors of an armed robbery. The164 notional homicides arose from 130 incidents averaging 1.26 deaths per event.
Table 3: Crime & Fatalities Khmer Press Reports January 1-July 7, 1998
Type of Event n %
Police Action 48 15.4
Vigilante Action 16 5.1
Murder/Manslaughter 100 32.0
Attempted Murder* 46 14.7
Suicide 18 5.8
Attempted suicide* 2 0.6
Kidnap/Robbery* 20 6.4
Rape* 4 1.2
Road fatality 30 9.6
Misadventures 28 9.0
All victims 312 100.00
Source: Phnom Penh Post Vol. 7 (1-15) from crimes reported in Khmer newspapers Rasmei Kampuchea and Koh Santeheap January 1 to July 7 1998. Note: (*) indicates non-fatal event.
The 1998 MOI report also refers to the number of deaths, injuries, and property losses recorded for 1992-1995. From this source, the total number of deaths recorded by national police was 1,222 or 12.86 per 100,000 in 1993. These deaths arose from 599 murders and 157 grenade attacks that resulted in 1,008 deaths at an average of 1.34 fatalities per event, and 214 “offenders” reported shot dead during police action. Using this ratio of 1:1.34 deaths to events and the ratio derived from the Khmer press [1:1.26] we can estimate the approximate number of victims for 1996 by applying the average ratio [1:1.3]. This weighting produced 775 homicide deaths from the 596 recorded events plus the 108 offenders killed by police yields a total of 883 deaths at a rate of 9.32 per 100,000. Table 4 shows the 1992-1998 homicide rates including [rate B] or excluding [rate A] suspect deaths and conservatively represents, the estimated range for the prevalence of homicide.
As noted, about one in five newspaper crime reports refer to police or mob killings of suspects. These forms of extra-judicial homicide make up over a quarter of all the deaths and nearly two fifths of all homicides reported by the press. Table 4 indicates the proportion of extra-judicial deaths of all homicides has varied from 17.5% in 1993, 9.0% in 1994, 5.8% in 1995, 12.2% in 1996 and 10.2% in 1998. The proportion of “offender” deaths in Table 4 for 1998 are much lower than estimated from news reports. This appears to be a function of news selection and the possible failure of police to diligently record deaths arising from vigilante action. Consequently, such a measure serves only as a crude and conservative indicator of fidelity to rule by law. The high number of extra-judicial homicides is attributed to the armed nature of most offending, a weak rule of law culture and the poor discipline and training of police. The risk of spontaneous acts of violence and mayhem are also relatively high given the number of grenade incidents and accidents reported in the press and official sources. However, these behaviours do not appear to equate with notions of “amok” although often ill directed and non-utilitarian (Winzeler 1990).
Table 4: Estimates of the Prevalence of Homicide 1992-98
1992 1993 1994 1995 1996 1998
Murder events 429 599 303 397 542 793
rate 4.71 6.3 3.09 3.97 5.70 7.33
Grenade events 40 157 79 27 54 68
rate 0.44 1.65 0.81 0.27 0.57 0.63
Estimated victims 487 1008 427 551* 775* 1119*
rate A 5.35 10.60 4.35 5.50 8.18 10.34
“Offender” deaths n/a 214 42 34 108 139ab
rate -- 2.25 0.43 0.34 1.14 1.28
All deaths 487 1222 469 585* 883* 1258*
rate B 5.35 12.86 4.78 5.84 9.32 11.62
Source: RGC MOI Reports of Judicial Police Centre 1998,1999 and 2000. Notes: (a) rates adjusted for missing provincial data; (b) estimated number of offender fatalities from newspaper and other sources; (n/a) data not available; (*) multiplier of 1.3 fatalities per event; estimates for 1997 not attempted and; offender fatalities unavailable for 1992.
The Comparative Prevalence of Homicide
Returning to our opening question about the relative level of lethal violence in Cambodia compared with other countries an attempt, through comparative analysis, is made in Table 5 to address the question of the severity of violence of Cambodia. In a comparative approach, we hope to identify cultural, economic, political, situational and social factors that explain macro-differences in homicide. The aim is to understand the variables that appear to weaken or strengthen the risks of crime at the broadest level. Comparative studies enable key aspects of culture, especially the definition and social reaction to crime, to be observed and also help identify factors that shape the prevalence and nature of crime (Arthur and Marenin 1995).
There are many macro-level factors thought to influence cross-national differences in crime including amongst others religious and cultural diversity, relative economic strength, and the effectiveness of state governance. The concept of modernisation conventionally proposes that the essential relationship between crime and development is that modernity reduces violence31. Table 5 locates the homicide rates and socio-economic conditions for Cambodia in comparison with other countries32. Factors indicative of socio-economic development and modernity included: person per telephone [PPT] and person per television [PPTV] as measures of modern culture; average life expectancy in years [life expect.] measures mortality and stress; the percent of the total population living in urban areas [% urban] and the percent of the total population who can read and write [% literate] measures of modernity; and gross net product per person [GNP $US] measures socio-economic strength33. Analysis of the relationship between these factors and homicide using rank correlation methods was inconsistent and no single or derived factor accounted for more than 12% of the variation observed. Thus, few of the socio-economic factors listed in Table 5 appear to relate to the homicide rate although increased wealth appears to have some protective value. Measures of state strength or capacities and autonomy provide better correlations with stronger states associated with lower levels of homicide. Differences in criminal justice policy and practice may be more crucial with high homicide nations such as Cambodia, Philippines, Thailand enduring relatively unfettered firearms; low enforcement capability and regime influenced policing institutions. Perhaps because of a fortuitous comparison Cambodia had a rate of homicide similar to other Buddhist states Thailand and Mongolia, but less than Sri Lanka did in 1996. The latter nation disrupted by chronic civil war and Mongolia a newly independent state. Amongst the developed nations, only Taiwan and the USA approached the level of Cambodia34. Regardless of which measure was utilised the homicide rate for the Philippines is informative and is further examined.
Table 5: Homicide Rates in Asian and Selected Countries
Country Homicide GNP Religion Life % % PPT PPTV
rate per capita expect. urban literate
Nepal 2.3 225 Buddhism 55 14 28 174 355
Cambodia* 9.3 270 Buddhism 53 21 38 1212 119.5
Bangladesh 2.0 283 Islam 57 18 32 380 143
Mongolia 9.6 335 Buddhism 64 61 95 26.7 17.3
India 4.6 387 Hindu 61 27 52 65.3 16.6
China 1.3 738 Secular 71 30 82 22.4 4.1
Sri Lanka 11.1 760 Buddhism 72 22 89 65.1 15.2
Myanmar 1.4 765 Buddhism 59 26 82 265 22
Indonesia 0.9 998 Islam 63 34 84 47.7 6.8
PNG 2.9 1005 Christian 57 17 72 4.9 11.2
Philippinesab 33.5 1203 Catholic 67 46 94 38.6 7.9
Thailand 9.6 2450 Buddhism 69 36 94 13.5 4.4
Fiji 5.9 2500 Christian 72 40 90 8.9 11.1
Malaysia 2.1 4287 Islam 72 47 89 5.5 4.7
South Korea 1.4 9511 Buddhism 72 81 97 2.1 3.4
Taiwan* 8.1 13303 Polytheism 75 58 93 2.1 3.1
N. Zealand 1.8 16970 Christian 77 86 100 1.4 2.0
Macau* 3.6 17475 Polytheism 73 75 94 2.5 3.6
Canada* 2.7 19740 Christian 78 78 99 1.3 1.5
Australia 1.9 20020 Christian 78 86 100 1.5 2.1
UK* 1.4 22268 Christian 76 90 100 1.9 2.2
FRG* 1.5 25720 Christian 76 87 100 1.8 1.8
Hong Kong 1.6 26400 Polytheism 79 95 92 1.5 3.0
USA* 9.1 29950 Christian 77 76 96 1.3 1.2
Singapore 1.8 31900 Polytheism 77 100 92 2.0 2.6
Japan* 1.0 33800 Shinto 80 78 100 1.5 1.6
Sources: The average per capita INTERPOL homicide rates between 1988-1994 cited in Neopolitan (1997) unless otherwise indicated. Notes: (*) national rates for a single year [1995/1996] cited from national sources; PPT = persons per telephone and, PPTV = persons per television; (a) principal or state religion; (b) reported by Philippines National Statistic Office as 18.0 incidents in 1994 for murder and other homicide incidents.
Crime and lethal violence in Cambodia is now compared briefly with Hong Kong and the Philippines. These states have useful contrasting colonial histories, socio-religious values and, economic development that may help disentangle the relationship between modernity, extreme conflicts and crime. The emerging economy of the Philippines with its Spanish-American colonial legacy and dominant Catholic values can be contrasted with the Taoism and Confucianism values of the former British colony of Hong Kong now an advanced market and service economy.