Poverty and Witch Killing

1
Poverty and Witch Killing

• Edward Miguel
• 2003

2
Introduction

• Strong Negative Relationship between Economic
  Growth and Crime
• Miguel uses local rainfall to identify income
  shocks in Tanzania
• Drought and Flood lead to poor crop harvests and
  famine conditions

3
Introduction

• Findings consistent with theoretical model of
  household resource allocation
• Kill unproductive household members to protect
  other members
• Most Witch Killings occurred in low
  socio-economic status villages during the
  hungry season

4
Introduction

• Alternate Theory Scapegoats
• Families need someone to blame for their
  suffering
• Both rainfall and disease should lead to more
  witch killings
• Only poverty shocks (rainfall) lead to more
  witch killings

5
Background

• Witchcraft beliefs are widely held in
  Sub-Saharan Africa
• Witches inflict harm on other community members
  in their immediate circle whom they envy

6
Tanzania

• Total GDP 21.58 billion (U.S. 11 trillion)
• GDP growth 5.2
• Total population 36,588,225
• Average life expectancy 44.39 years
• Age structure
• 0-14 years 44.2
• 15-64 years 53.2
• 65 years and over 2.6

7
Tanzania

• Witchcraft beliefs strong no adoption of
  Christianity or Islam
• Relatives, Kin, Neighbors typically behind the
  murders
• Government efforts to stop killings unsuccessful

8
Other Factors

• Sungu-Sungu political Institution
• All male
• Consider combating witches central to village
  security
• Witches considered criminals just as dangerous
  as thieves or murderers killers pursuing
  justice

9
Other Examples

• Ghana, Kenya, Zimbabwe, South Africa
• Young Children
• South America
• Infants
• Europe 16th 18th Centuries

10
Allocation of Resources

• Four Assumptions
• Allocation Choices made entirely within the
  household
• One household member determines allocation
• There is a minimum level of food consumption
  necessary to sustain life
• Household members are identical except in terms
  of their future economic production

11
Allocation

• Patriarch divides income to maximize utility
• Two Cases
• Normal enough income to sustain all members
  above subsistence
• Extreme not enough income to meet nutritional
  needs of all members

12
Options

• Elderly lowest future income of all household
  members
• Infants remain unproductive for years and are
  more susceptible to mortality
• Seen more in poorer areas where households are
  closer to subsistence

13
Options

• Women
• Local Political leaders are almost all male
• Women move into their husbands village
• Men have greater average physical strength

14
Anthropology

• Poor pre-industrial societies frequently respond
  to environmental stress by killing those seen as
  a burden on the community
• Believing a murder victim truly is a witch plays
  an important role in alleviating psychological
  trauma and social stigma associated with murder

15
Data and Measurement
16
How was the data obtained?

• 2 surveys conducted in 2001-2002
• Village Council Survey
• Household Survey
• Satellite Vegetation Data

17
Village Council Survey

• Sample of 67 villages
• 2 questions to council members in each village
• Has this village faced any natural disasters or
  calamities in the past ten years?
• Had there been any murders in the village during
  the previous ten years? If so, the number and
  years of the murders.
• Data more reliable
• Memorable events
• Openness in discussing killings due to lack of
  consequences

18
Household Survey

• Total of 1293 households surveyed in 2001.
• 15-20 households in each village surveyed.
• Information
• Demographic Age and Sex composition of each
  village
• Consumption expenditure
• Results
• District extremely poor
• High level of food insecurity
• - 2 households use irrigation rather than
  rain-fed agriculture
• - 75 of income goes to food consumption
• - 73 of villages received some sort of food
  relief

19
Satellite Vegetation Data

• NDVI used to measure weather variability.
• Deviation from average vegetation during the
  rainy season.

20
Descriptive stats of survey data
21
(No Transcript)
22
(No Transcript)
23
Empirical Results
24
Empirical Results

• Key assumption Rainfall exogenous.
• Regressions run on equations that include
• Village socioeconomic conditions.
• Demographics.
• Diseases and epidemics.
• Indicator variable for extreme rainfall.

25
Table 3

• Shows factor effects on annual per capita
  consumption.
• Proves extreme rainfall statistically significant
  with respect to consumption.
• Shows disease epidemic not associated with per
  capita income.
• Shows other variables not conclusive.

26
(No Transcript)
27
Table 4

• Shows extreme rainfall as statistically
  significant to witch murders.
• Shows famine insignificant factor in witch
  killing.
• Bolsters poverty theory over scapegoat.

28
(No Transcript)
29
Table 6

• Shows rainfalls effect on violent crimes.
• Impacts witch killing.
• No change in non-witch murders.
• Indicates little loss of general order in extreme
  weather and poverty.

30
(No Transcript)
31
Table 7

• Less robust findings with respect to village
  idiosyncrasies.
• Average per capita income only significant
  numbers related to murders.
• The rest only weakly support hypothesis.

32
(No Transcript)
33
Infant Statistics

• Boys mortality rates generally higher.
• Negative income shocks increase female mortality
  rates.
• Most data shows economic based favoritism toward
  boys in leaner years.
• Infant based regressions not robust, poor
  collection accuracy factor in questionability of
  data.

34
Conclusion

• Strong link between poverty and witch killings in
  Tanzania.
• Extreme rainfall ? income drops ? doubling of
  witch murders
• Focus on microeconomic issue to show link between
  poverty and violence in developing countries.
• Miguel the situation will not improve until
  living standards improve.

35
What we think

• Survey data
• Very unreliable data with no formal system of
  record keeping
• Risk of inaccuracy in exact number of witch
  killings. Data relying on honesty of council
  members.
• Extreme rainfall ingenious exogenous variable
  for income
• Convincing empirical findings
• Very specific result can it be generalized?

36
Discussion Questions

• Do you believe rainfall truly exogenous, or do
  witches control the weather?
• Can you think of any other reasons to kill a
  witch due to rainfall, other than impending
  economic shortfalls?
• Is witch killing the solution to impending US
  social security woes?
• Why are witch killers lax in their duties when
  facing sickness and plague? (perhaps theyre too
  ill to swing the machetes effectively?)

37
Discussion Questions

• Did this convince you that witch killings are
  attributed to poverty?
• Proposed solutions
• Aggressive prosecution of witch killers
• Provide elderly women with regular pension
• Improve system of formal insurance for extreme
  rainfall shocks
• Do you think these solutions are realistic?
• Any other suggestions?