## wpiea2021277-print-pdf

## Source details

**Canonical URL:** [wpiea2021277-print-pdf](https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021277-print-pdf.pdf)

## Other formats

- [Markdown version](/-/media/files/publications/wp/2021/english/wpiea2021277-print-pdf.pdf.md)
- [Structured JSON version](/-/media/files/publications/wp/2021/english/wpiea2021277-print-pdf.pdf.json)

---

### Background and literature
- Prior costing studies (Duvvury et al. (2013) survey) classify costs as: direct tangible, indirect tangible, direct intangible, indirect intangible (classification after Day et al. 2005).
- Relative to GDP, those studies often find costs in the range between 1.2 and 2 percent of GDP.
- The paper emphasizes endogeneity concerns between economic activity and violence against women and girls (VAWG); education and household wealth are negatively associated with intimate partner violence (Wilson (2019)).

### Data sources and definitions
- Sample: 29 integrated Demographic and Health Survey from USAID’s DHS Program from 18 sub-Saharan African countries.
- Sample coverage: more than 440,000 women, representative of around 75 percent of sub-Saharan Africa's female population; analysis at the district level covering 224 districts.
- Nighttime lights: cleaned luminosity satellite data since 1992; pixel-year roughly 0.86 square kilometers at the equator; light digital number (DN) ranges from 0 to 63.
- Nightlights per capita computed as sum of pixel values within a district divided by district population (population from Gridded Population of the World, Version 4).
- VAWG indicators: multiple DHS questions on punching, slapping, pushing/throwing, threats, emotional violence, sexual violence, and visible marks; constructed a composite index as the simple average of these variables.
- Controls: share of cropland, urbanization rate (share living in urban areas), time to water source, total road lengths, average years of schooling, conflict-related deaths per population, religious fractionalization (Montalvo and Reynal-Querol (2005) methodology).

### Descriptive statistics and stylized facts
- On average, more than one quarter of women experienced intimate partner violence in recent years.
- Prevalence by income/region:
  - In low-income countries, nearly one-third of women experienced domestic violence.
  - Around one fifth in upper middle-income countries.
  - Sub-Saharan Africa has the highest rate with more than 30 percent of women experiencing intimate partner violence.
- Illustrative cross-country heterogeneity:
  - Punching with fist or something harmful ranges from 1.1 percent in Comoros to 17.8 percent in Gabon.
  - Sexual violence ranges from 1.5 percent in Burkina Faso to 27.1 percent in Democratic Republic of Congo.
  - Emotional violence ranges from 5.5 percent in Comoros to 47 percent in Gabon.
- Change over time ("Before 2010" vs "After 2010" averages):
  - Any physical results of husband's actions: 11.8 (Before 2010) vs 14.6 (After 2010); Difference 2.8.
  - Ever slapped: 26.0 (Before 2010) vs 28.7 (After 2010); Difference 2.7.
  - Ever experienced any emotional violence: 24.9 (Before 2010) vs 30.5 (After 2010); Difference 5.6.
  - Ever experienced any sexual violence from partner: 12.3 (Before 2010) vs 14.4 (After 2010); Difference 2.1.

### Empirical approach
- Main specification: nightlights per capita = α + β Violence + γ X′ + country fixed effects + time fixed effects + error.
- Endogeneity concerns: reverse causality, omitted variables, measurement error (underreporting in surveys).
- Instrumental variable: average level of domestic violence in neighboring districts (districts sharing borders); controls include average nightlights per capita in bordering districts to address spillovers.
- Identification strategy: two-stage least squares (2SLS) with district, year, and country fixed effects in various specifications.

### Main quantitative results — OLS and IV
- OLS baseline (Table 4): coefficient on the composite violence index negative and strongly significant across specifications.
  - Example OLS estimates range from -1.8488*** to -3.1402*** across columns (standard errors reported in table).
- IV baseline (Table 5): instrument performs well; first-stage and overidentification/weak instrument tests support identification.
  - IV main result (Table 5, column (1), composite index): Violence against women = -8.6954*** (1.285) — an increase of domestic violence against women by 1 percentage point could reduce nightlights per capita by 8.7 percent.
  - Type-specific IV estimates (examples):
    - Spouse ever punched with fist or something harmful: -15.2377*** (0.944).
    - Spouse ever experienced any emotional violence: -18.0609*** (4.224).
- Controls with consistent signs: urban area, roads, education positive; share of agriculture, time to water source, conflict, religious fractionalization negative/significant in many specifications.

### Transmission channel: female employment
- IV estimates of VAWG on female employment (Table 6):
  - Composite index: Violence against women = -2.4739*** (0.403) — interpreted as a 1 percentage point increase in VAWG → 2.5 percentage point reduction in female employment.
- Mediation analysis (Table 7):
  - Including female employment in nightlights regressions reduces magnitude of violence coefficient but it remains negative and significant:
    - Example: Violence against women = -6.4237*** (1.298); Female employment coefficient = 0.7026** (0.354).
  - Interpretation: adverse effect of VAWG on economic activity is partially transmitted through reductions in female employment; other likely channels include productivity, health, human capital, and resource allocation.

### Nonlinearities and heterogeneities
- Law protecting against domestic violence:
  - Violence * law = 5.2341** (2.424); Violence against women = -7.3210*** (2.604) — protective laws dampen the negative effect.
- Natural resources (mining sites):
  - Violence * natural resources = -10.3803*** (3.071); Violence against women = -14.9031*** (3.395) — costs higher in resource-rich districts.
- Economic conditions (commodity price index):
  - Violence * economic conditions = 0.7044** (0.224); Violence against women = -9.4469*** (1.547) — economic costs are higher during bad times (declining commodity prices).
- Decision-making power and education gap:
  - Violence * decision power = 44.9079*** (8.812); Decision power = 7.0456*** (1.643) — higher decision-making power for women reduces economic costs of VAWG.
  - Violence * education gap = -21.3012* (12.495); Education gap = -4.7622** (1.195) — wider gender gap in education exacerbates economic impact of VAWG.

### Robustness checks
- Alternative attitudes-based measures of violence (Table 9): attitudes that justify beating instrumented by neighbors remain negatively and strongly associated with nightlights per capita (examples: Attitudes towards violence coefficients -3.8205***, -4.8672***, -13.1854*** across columns).
- Alternative IV construction (median of neighbors' violence with inverse population weights) — results still hold (Table 10): Violence against women = -8.9645*** (1.324) for composite index.
- Excluding outliers (top and bottom 5 percent of violence occurrence) — results similar or larger in magnitude (Table 11): composite index estimate -9.0503*** (1.389).
- Additional covariates (population density, precipitations, wealth index, income gap, oil field, access to banks, access to electricity) — composite-index violence coefficient remains negative and highly significant in all columns (Table 12).
  - Access to banks coefficient positive and strongly significant: 1.7419*** (0.364).
  - Income gap coefficient negative and mildly significant: -0.8601* (0.515).

### Key quantitative summaries (preserved exactly as reported)
- Cost estimates from prior studies often find costs between 1.2 and 2 percent of GDP.
- Sample: 29 DHS surveys from 18 countries; more than 440,000 women; around 75 percent of sub-Saharan Africa's female population; 224 districts.
- Nightlight pixel DN range: 0 to 63.
- Correlation between nightlight and economic activity up to 70 percent (cite Hu and Yao, 2019; Pinkovskiy and Sala-i Martin, 2016 as reported).
- IV main result (composite index): Violence against women = -8.6954*** (1.285) — 1 percentage point increase in VAWG → up to 8.7 percent reduction in nightlights per capita (Table 5, column (1)).
- Female employment effect (IV): Violence against women = -2.4739*** (0.403) — 1 percentage point increase in VAWG → 2.5 percentage point reduction in female employment (Table 6, column (1)).
- Nonlinear interaction examples:
  - Violence * natural resources = -10.3803*** (3.071).
  - Violence * law = 5.2341** (2.424).

### Conclusions and policy implications
- Higher levels of VAWG are causally associated with lower economic development in sub-Saharan Africa.
- An increase in violence against women by 1 percentage point can reduce nightlights-based economic activities by up to 8.7 percent on average.
- Evidence indicates the adverse effect is driven in part by negative impacts on female labor supply.
- Heterogeneities: negative economic effects are larger in countries without protective laws against domestic violence, in natural resource-rich districts, where women lack decision-making power, during economic downturns, and where there is a significant gender gap in education.
- Policy implications emphasized:
  - Adopt and reinforce laws against domestic violence and strengthen women’s decision-making power to lower violence and its economic cost.
  - Encourage girls’ education to reduce the gender gap and attenuate male dominance/control.
  - Pay attention to increased levels of domestic violence during the COVID-19 pandemic as economic costs are higher during downturns, potentially lengthening recovery.
  - Implement targeted policies to ensure windfalls from natural resource exploitation are not used to dominate and control women.

*Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021277-print-pdf.pdf*

### 0.94 percentage points in 2017. Other studies estimate annual monetary costs of VAWG.  Duvvury et al. (2013)

### wpiea2021277-print-pdf - 0.94 percentage points in 2017. Other studies estimate annual monetary costs of VAWG.  Duvvury et al. (2013)

### Background and literature
- Prior costing studies (Duvvury et al. (2013) survey) classify costs as: direct tangible, indirect tangible, direct intangible, indirect intangible (classification after Day et al. 2005).
- Relative to GDP, those studies often find costs in the range between 1.2 and 2 percent of GDP.
- The paper emphasizes endogeneity concerns between economic activity and violence against women and girls (VAWG); education and household wealth are negatively associated with intimate partner violence (Wilson (2019)).

### Data sources and definitions
- Sample: 29 integrated Demographic and Health Survey from USAID’s DHS Program from 18 sub-Saharan African countries.
- Sample coverage: more than 440,000 women, representative of around 75 percent of sub-Saharan Africa's female population; analysis at the district level covering 224 districts.
- Nighttime lights: cleaned luminosity satellite data since 1992; pixel-year roughly 0.86 square kilometers at the equator; light digital number (DN) ranges from 0 to 63.
- Nightlights per capita computed as sum of pixel values within a district divided by district population (population from Gridded Population of the World, Version 4).
- VAWG indicators: multiple DHS questions on punching, slapping, pushing/throwing, threats, emotional violence, sexual violence, and visible marks; constructed a composite index as the simple average of these variables.
- Controls: share of cropland, urbanization rate (share living in urban areas), time to water source, total road lengths, average years of schooling, conflict-related deaths per population, religious fractionalization (Montalvo and Reynal-Querol (2005) methodology).

### Descriptive statistics and stylized facts
- On average, more than one quarter of women experienced intimate partner violence in recent years.
- Prevalence by income/region:
  - In low-income countries, nearly one-third of women experienced domestic violence.
  - Around one fifth in upper middle-income countries.
  - Sub-Saharan Africa has the highest rate with more than 30 percent of women experiencing intimate partner violence.
- Table 2 (types of violence) reports wide cross-country heterogeneity; illustrative figures:
  - Punching with fist or something harmful ranges from 1.1 percent in Comoros to 17.8 percent in Gabon (selected stat preserved as reported).
  - Sexual violence ranges from 1.5 percent in Burkina Faso to 27.1 percent in Democratic Republic of Congo.
  - Emotional violence ranges from 5.5 percent in Comoros to 47 percent in Gabon.
- Change over time (Table 3): overall increase in most types of domestic violence after 2010 compared to before 2010; aggregate "Before 2010" vs "After 2010" averages:
  - Any physical results of husband's actions: 11.8 (Before 2010) vs 14.6 (After 2010); Difference 2.8.
  - Ever slapped: 26.0 (Before 2010) vs 28.7 (After 2010); Difference 2.7.
  - Ever experienced any emotional violence: 24.9 (Before 2010) vs 30.5 (After 2010); Difference 5.6.
  - Ever experienced any sexual violence from partner: 12.3 (Before 2010) vs 14.4 (After 2010); Difference 2.1.

### Empirical approach
- Main specification: nightlights per capita = α + β Violence + γ X′ + country fixed effects + time fixed effects + error.
- Endogeneity concerns: reverse causality, omitted variables, measurement error (underreporting in surveys).
- Instrumental variable: average level of domestic violence in neighboring districts (districts sharing borders); controls include average nightlights per capita in bordering districts to address spillovers.
- Identification strategy: two-stage least squares (2SLS) with district, year, and country fixed effects in various specifications.

### Main quantitative results — OLS and IV
- OLS baseline (Table 4): the coefficient on the composite violence index is negative and strongly significant across specifications (example coefficients shown in table as reported).
  - Example OLS coefficients preserved as in Table 4: Violence against women estimates range from -1.8488*** to -3.1402*** across columns (standard errors reported in table).
- IV baseline (Table 5): instrument performs well in first stage; tests (F-test of exogeneity, Cragg-Donald, Kleibergen-Paap, Anderson-Rubin) support identification.
  - Key IV quantitative finding (column (1), composite index): an increase of domestic violence against women by 1 percentage point could reduce nightlights per capita by 8.7 percent (Violence against women = -8.6954***; robust standard error (1.285) as reported).
  - Type-specific IV estimates (examples from Table 5):
    - Spouse ever punched with fist or something harmful: -15.2377*** (0.944).
    - Spouse ever experienced any emotional violence: -18.0609*** (4.224).
  - Controls with consistent signs: urban area, roads, education positive; share of agriculture, time to water source, conflict, religious fractionalization negative/significant in many specifications.

### Transmission channel: female employment
- Causal effect of VAWG on female employment using IV (Table 6):
  - Composite index: Violence against women coefficient = -2.4739*** (0.403) — interpreted as: an increase in the share of women who experienced domestic violence by 1 percentage point would reduce female employment rate by 2.5 percentage points (coefficient preserved as reported).
- Mediation analysis (Table 7): including female employment in the nightlights regressions reduces the magnitude of the violence coefficient but it remains negative and significant:
  - Example coefficient (Table 7, composite index with female employment included): Violence against women = -6.4237*** (1.298); Female employment coefficient = 0.7026** (0.354).
  - Interpretation: the adverse effect of violence against women on economic activity is partially transmitted through reductions in female employment; other channels likely include productivity, health, human capital, and resource allocation.

### Nonlinearities and heterogeneities (Table 8)
- Law protecting against domestic violence:
  - Violence * law interaction is positive and significant (Violence * law = 5.2341** (2.424)); Violence alone remains negative and significant (Violence against women = -7.3210*** (2.604)); protective laws dampen the negative effect.
- Natural resources (mining sites):
  - Violence * natural resources = -10.3803*** (3.071); Violence against women = -14.9031*** (3.395); economic costs of domestic violence are higher in resource-rich districts.
- Economic conditions (commodity price index):
  - Violence * economic conditions = 0.7044** (0.224); Violence effect remains negative and significant (Violence against women = -9.4469*** (1.547)); economic costs are higher during bad times (declining commodity prices).
- Decision-making power and education gap:
  - Violence * decision power = 44.9079*** (8.812); Decision power = 7.0456*** (1.643) — higher decision-making power for women reduces economic costs of VAWG.
  - Violence * education gap = -21.3012* (12.495); Education gap = -4.7622** (1.195) — wider gender gap in education exacerbates the economic impact of VAWG.

### Robustness checks
- Alternative attitudes-based measures of violence (Table 9): attitudes that justify beating (multiple scenarios) instrumented by neighbors remain negatively and strongly associated with nightlights per capita (examples: Attitudes towards violence coefficients -3.8205***, -4.8672***, -13.1854*** across columns with standard errors reported).
- Alternative IV construction (median of neighbors' violence with inverse population weights) — results still hold (Table 10); example: Violence against women = -8.9645*** (1.324) for composite index.
- Excluding outliers (top and bottom 5 percent of violence occurrence) — results similar or larger in magnitude (Table 11); composite index estimate -9.0503*** (1.389).
- Additional covariates (population density, precipitations, wealth index, income gap, oil field, access to banks, access to electricity) — composite-index violence coefficient remains negative and highly significant in all columns (Table 12).
  - Access to banks coefficient positive and strongly significant: 1.7419*** (0.364).
  - Income gap coefficient negative and mildly significant: -0.8601* (0.515).

### Key quantitative summaries (preserved exactly as reported)
- Cost estimates from prior studies often find costs between 1.2 and 2 percent of GDP.
- Sample: 29 DHS surveys from 18 countries; more than 440,000 women; around 75 percent of sub-Saharan Africa's female population; 224 districts.
- Nightlight pixel DN range: 0 to 63.
- Correlation between nightlight and economic activity up to 70 percent (cite Hu and Yao, 2019; Pinkovskiy and Sala-i Martin, 2016 as reported).
- IV main result (composite index): Violence against women = -8.6954*** (1.285) — 1 percentage point increase in VAWG → up to 8.7 percent reduction in nightlights per capita (Table 5, column (1)).
- Female employment effect (IV): Violence against women = -2.4739*** (0.403) — 1 percentage point increase in VAWG → 2.5 percentage point reduction in female employment (Table 6, column (1)).
- Nonlinear interaction examples:
  - Violence * natural resources = -10.3803*** (3.071).
  - Violence * law = 5.2341** (2.424).

### Conclusions and policy implications (as presented)
- Higher levels of VAWG are causally associated with lower economic development in sub-Saharan Africa.
- An increase in violence against women by 1 percentage point can reduce nightlights-based economic activities by up to 8.7 percent on average.
- Evidence that adverse effect is driven in part by negative impacts on female labor supply.
- Heterogeneities: negative economic effects are larger in countries without protective laws against domestic violence, in natural resource-rich districts, where women lack decision-making power, during economic downturns, and where there is a significant gender gap in education.
- Policy implications emphasized in the paper:
  - Adopt and reinforce laws against domestic violence and strengthen women’s decision-making power to lower violence and its economic cost.
  - Encourage girls’ education to reduce the gender gap and attenuate male dominance/control.
  - Pay attention to increased levels of domestic violence during the COVID-19 pandemic as economic costs are higher during downturns, potentially lengthening recovery.
  - Implement targeted policies to ensure windfalls from natural resource exploitation are not used to dominate and control women.

*Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021277-print-pdf.pdf*

### References

### References

### Bibliographic references
- Acemoglu, D., Naidu, S., Restrepo, P. and Robinson, J. (2019). Democracy Does Cause Growth. Journal of Political Economy, 127(1): 47-100.
- Alesina, A., Devleeschauwer, A., Easterly, W., Kurlat, S., and Wacziarg, R. (2003). Fractionalization. Journal of Economic Growth, 8: 155-194
- Alesina, A., Özler, S., Roubini, N. and Swagel, P. (1996). Political Instability and Economic Growth. Journal of Economic Growth, 1 (2): 189–211.
- Amegbor, P.M. and Rosenberg, M.W. (2019). What geography can tell us? Effect of higher education on intimate partner violence against women in Uganda. Applied Geography, 106: 71-81.
- Barro, R. J. (2001). Human Capital and Growth. The American Economic Review, Vol. 91 (2): 12-17.
- Barro, R. J. and Sala-i-Martin, X. (1995). Economic Growth. New York: McGraw-Hill.
- Bell, H. (2003). Cycles within cycles: Domestic violence, welfare, and low-wage work. Violence Against Women, 9: 1245-1262.
- Bobonis, G., Gonzalez-Brenes, M., and Castro, R. (2013). Public transfers and domestic violence: The roles of private information and spousal control. American Economic Journal: Economic Policy, 5(1): 179–205.
- Böök, B., van Hoof, F., Senden, L., and Timmer, A. (2020). Gendering the COVID-19 crisis: a mapping of its impact and call for action in light of EU gender equality law and policy. European Equality Law Review 2020/2: 22-44
- Brueckner, M. and Lederman, D. (2015). Effects of Income Inequality on Aggregate Output. World Bank Policy Discussion Paper 7317.
- Brush, L.D. (2002). Work-related abuse: A replication, new items, and persistent questions. Violence and Victims, 17: 743-757.
- Campbell, J. (2002). Health Consequences of intimate partner violence. The Lancet, 359 (9314): 1331–1336.
- Campbell, J., Torres, S., Ryan, J., et al. (1999). Physical and nonphysical partner abuse and other risk factors for low birth weight among full term and preterm babies: a multiethnic case-control study. American Journal of Epidemiology, 150(7):714–726
- Caselli, F. and Reynaud, J. (2019). Do fiscal rules cause better fiscal balances? A new instrumental variable strategy. IMF Working Papers Series 19/49, Washington DC.
- Center for International Earth Science Information Network (CIESIN). (2013). Global Roads Open Access Data Set, Version 1 (gROADSv1). Columbia University, and Information Technology Outreach Services—ITOS—University of Georgia. Palisades, NY: NASA Socioeconomic Data and Applications Center (SEDAC).
- Chen, X. and Nordhaus, W. D. (2011). Using Luminosity Data as a Proxy for Economic Statistics. Proceedings of the National Academy of Sciences, 108(21):8589-8594.
- Cherif, R., Hasanov, F., and Wang, L. (2018). Sharp instrument: a stab at identifying the causes of economic growth. IMF Working Papers Series 18/117, Washington DC.
- Conger, R.D., Elder, G.H. Jr., Lorenz, et al. (1990). Linking economic hardship to marital quality and instability. Journal of Marriage and the Family, 52:643–656.
- Cools, S. and Kotsadam, A. (2017). Resources and Intimate Partner Violence in sub-Saharan Africa. World Development, 95: 211-230.
- Damonti, P. (2014). Can Gender-based Violence result in a process of social exclusion? A quantitative-qualitative analysis. Procedia-Social and Behavioral Sciences, 161: 41-47
- Day, T., McKenna, K., Bowlus, A., (2005). The Economic Costs of Violence Against Women: An Evaluation of the Literature. United Nations Expert Brief.
- Devries, K.M., Kishor, S., Johnson, H. et al. (2010). Intimate partner violence during pregnancy: analysis of prevalence data from 19 countries. Reproductive Health Matters, 18(36):158–70.
- Dhakal, L., Berg-Beckhoff, G., and Aro, A. R. (2014). Intimate partner violence (physical and sexual) and sexually transmitted infection: Results from Nepal Demographic Health Survey 2011. International Journal of Women’s Health, 6: 75–82.
- Dreher, A., and Lohmann, S. (2015). Aid and Growth at the Regional Level. Oxford Review of Economic Policy, 31(3-4): 420-446
- Dugan, L. (2002). Domestic violence legislation: exploring its impact on domestic violence and the likelihood that police are informed and arrest. University of Maryland, Maryland.
- Duvvury, N., Callan, A., Carney, P., and Raghavendra S. (2013). Intimate partner violence: economic costs and implications for growth and development. World Bank Women’s Voice, Agency, & Participation Research Series No. 3.
- Edleson, J. (1999). Children’s witnessing of adult domestic violence. Journal of Interpersonal Violence, 14(8): 839-870.
- Egert, B., Kozluk, T. and Sutherland, D. (2009). Infrastructure and Growth: Empirical Evidence. CESifo Working Paper Series 2700, CESifo.
- Felson, R. B., Messner, S.F., Hoskin, A.W. and Deane, G. (2006). Reasons for reporting and not reporting domestic violence to the police. Criminology, 40(3): 617-648.
- García-Moreno, C., Ellsberg, M., Heise, L. and Watts, C. (2005). WHO multicountry study on women’s health and domestic violence against women: Initial results on prevalence, health outcomes and women’s responses, Technical report. Geneva: World Health Organization.
- Grau, J., Fagan, J. and Wexler, S. (1984). Restraining Orders for Battered Women: Issues of Access and Efficacy. Women & Politics, 4(3): 21-25.
- Henderson, J. V., Storeygard, A., and Weil, D. N. (2012). Measuring Economic Growth from Outer Space. American Economic Review, 102(2):994-1028.
- Holt, S. (2015). An exploration of the impacts that experiencing domestic violence can have on a child’s primary school education: view of educational staff. British journal of community justice, 13(2): 7-26.
- Hu, T. and Yao, J. (2019). Illuminating Economic Growth. IMF Working Paper WP/19/77. Washington Dc: International Monetary Fund
- International Labour Organization, (2007). Girls in mining: Research findings from Ghana, Niger, Peru, and United Republic of Tanzania. ILO Working paper, Geneva, Switzerland.
- International Labour Organization and UN Women, (2019). Handbook: addressing violence and harassment against women in the world of work. New York, USA
- International Monetary Fund. (2019). Sub-Saharan Africa Regional Economic Outlook: Recovery Amid Elevated Uncertainty. Accessed June, 2020. https://www.imf.org/en/Publications/REO/SSA/Issues/2019/04/01/sreo0419.
- Jewkes, R. (2002). Intimate partner violence: Causes and prevention. The Lancet, 359, 1423–1429.
- Kinda, T., Mlachila, M. and Ouedraogo, R. (2018). Do commodity price shocks weaken the financial sector? The World Economy, 41(11): 3001-3044, November.
- Koenig, M. A., Ahmed, S., Hossain, M. B., and Mozumder, A. B. (2003). Women’s status and domestic violence in rural Bangladesh: individual- and community-level effects. Demography, 40(2): 269–288.
- Kotsadam, A., Ostby, G. and Rustad, S.A. (2017). Structural Change and Wife Abuse: A Disaggregated Study of Mineral Mining and Domestic Violence in Sub-Saharan Africa. Political Geography, 56: 53-65
- Krantz, G. (2002). Violence against women: a global public health issue. Journal of Epidemiology and Community Health, 56(4):242-243.
- Kumar, S., Jeyaseelan, L., Suresh, S., Ahuja, R.C. (2005). Domestic violence and its mental health correlates in Indian women. British Journal of Psychiatry, 187(1):62–67
- La Mattina, G. (2017). Civil conflict, domestic violence and intra-household bargaining in post-genocide Rwanda. Journal of Development Economics, 124: 168-198
- Levine, R. (1997). Financial development and economic growth: views and agenda. Journal of Economic Literature, 35: 688-726.
- Liker, J.K., Elder, G.H. Jr. (1983). Economic hardship and marital relations in the 1930s. American Sociological Review, 48(3):343-359.
- Mamo, N., Bhattacharyya, S. and Moradi, A. 2019. Intensive and extensive margins of mining and development: Evidence from sub-Saharan Africa. Journal of Development Economics, 139: 28-49.
- Mankiw, N. G., Romer, D. and Weil, D.N. (1992). A Contribution to the Empirics of Economic Growth. Quarterly Journal of Economics, 107: 407-437.
- Mocan, N. H., and Cannonier, C. (2012). Empowering women through education: Evidence from sierra leone, Working Paper 18016. National Bureau of Economic Research.
- Montalvo, J. G. and Reynal-Querol, M. (2005). Ethnic polarization, potential conflict and civil war. The American Economic Review 95(3): 796-816.
- Neamah H.H., Sudfeld C., McCoy D.C., et al. (2018) Intimate Partner Violence, Depression, and Child Growth and Development. Pediatrics, 142(1):e20173457
- Ostry, J. D., Berg, A. and Tsangarides, G. D. (2014). Redistribution, Inequality, and Growth. IMF Staff Discussion Note No. SDN/14/02, February.
- Pierotti, R. (2013). Increasing rejection of intimate partner violence: Evidence of global cultural diffusion. American Sociological Review, 78(2): 240–265.
- Pinkovskiy, M., and Sala-i Martin, X. (2016). Lights, Camera . . . Income! Illuminating the National Accounts-Household Surveys Debate. The Quarterly Journal of Economics, 131(2): 579–631.
- Raghavendra, S., Kijong, K., Sinead, A., Mrinal, C., Asante, F., Petri, T P., and Duvvury, N. (2019). The macroeconomic loss due to violence against women and girls: the case of Ghana. Levy Economic Institute of Bard College, Working Paper No. 939.
- Rahman, M., Nakamura, K., Seino, K., and Kizuki, M. (2013). Does gender inequity increase the risk of intimate partner violence among women? Evidence from anational Bangladeshi sample. PLoS One, 8(12), e82423.
- Reeves, C., and O'Leary-Kelly, A. (2007). The effects and costs of intimate partner violence for work organizations. Journal of Interpersonal Violence 22(3): 327-44.
- Ross, M. (2008). Oil, islam, and women. American Political Science Review, 102 (1): 107-123.
- Ross, M. (2012). The oil curse: How petroleum wealth shapes the development of nations? Princeton University Press, Princeton/Oxford (2012)
- Schneider, D., Harknett, K, and McLanahan, S. (2016). Intimate Partner Violence in the Great Recession. Demography, 53(2): 471-505.
- Shamu, S., Abrahams, N., Temmerman, M., Musekiwa, A. and Zarowsky, C. (2011). A systematic review of African studies on intimate partner violence against pregnant women: prevalence and risk factors. PLoS One, 6(3):e17591.
- Swanberg, J. and Logan, T.K. (2005). The effects of intimate partner violence on women’s labor force attachment: Experiences of women living in rural and urban Kentucky. Journal of Occupational Health Psychology 10(1): 3-17.
- Swanberg, J. E., Logan, T., and Macke, C. (2005). Intimate partner violence, employment, and the workplace: Consequences and future directions. Trauma, Violence & Abuse 6(4):286-312.
- UN Women, (2020). COVID-19 and Ending Violence Against Women and Girls. New York: UN Women.
- United States Geological Survey (USGS), 2014. Mineral Facilities of Africa and the Middle East. National Minerals Information Center.
- Wilson, Nicholas (2019). Socio‐economic Status, Demographic Characteristics and Intimate Partner Violence. Journal of International Development, 31(7):632-657.
- World Bank. (2020). Women, Business and the Law 2020. The World Bank Group, Washington DC. https://openknowledge.worldbank.org/bitstream/handle/10986/32639/9781464815324.pdf
- Yount, K., and Li, L. (2009). Women’s ‘justification’ of domestic violence in Egypt. Journal of Marriage and Family, 71: 1125–1140

### Appendices — Tables and empirical results

- Table A 1: List of country surveys (columns: Country, Survey year, Number of districts, Number of individuals)
  - Angola 2015 14 14,379
  - Burundi 2016 17 17,269
  - Cameroon 2004 10 10,656
  - Cameroon 2011 11 15,426
  - Democratic Republic of Congo 2007 99 9,995
  - Democratic Republic of Congo 2013 9 18,827
  - Ethiopia 2016 11 15,683
  - Ghana 2008 12 4,916
  - Kenya 2003 11 8,195
  - Kenya 2008 11 8,444
  - Kenya 2014 11 31,079
  - Malawi 2010 28 23,020
  - Malawi 2016 27 24,562
  - Mali 2006 8 14,583
  - Mali 2012 6 10,424
  - Mozambique 2011 10 13,745
  - Namibia 2013 12 10,018
  - Nigeria 2008 5 33,385
  - Nigeria 2013 5 38,948
  - Senegal 2017 14 16,787
  - South Africa 2016 78 8,514
  - Zimbabwe 2005 10 8,907
  - Zimbabwe 2010 10 9,171
  - Zimbabwe 2015 10 9,955
  - Tanzania 2010 24 10,139
  - Tanzania 2015 27 13,266
  - Burkina Faso 2010 13 17,087
  - Zambia 2007 8 7,146
  - Zambia 2013 8 16,411
  - Total 224 440,937

- Table A 2: Descriptive statistics (columns: Variable Obs Mean Std. Dev. Min Max)
  - Nightlights per capita, log 358 1.10 1.22 0.00 5.77
  - Non-agriculture employment 358 0.51 0.27 0.04 1.00
  - Violence against women, composite index 358 0.17 0.08 0.02 0.46
  - Any physical results of husband's actions 339 0.13 0.10 0.01 0.70
  - Spouse ever punched with fist or something harmful 351 0.12 0.06 0.01 0.42
  - Spouse ever slapped 351 0.26 0.12 0.01 0.63
  - Spouse ever pushed, shook, or threw something 358 0.13 0.08 0.00 0.46
  - Spouse ever threatened with harm 358 0.11 0.07 0.01 0.54
  - Spouse ever experienced any emotional violence 358 0.27 0.12 0.03 0.80
  - Spouse ever experienced any sexual violence from partner 348 0.13 0.09 0.00 0.47
  - Beating is justified for any reasons 358 0.44 0.23 0.03 0.88
  - Beating is justified if woman argues with him 358 0.28 0.18 0.01 0.77
  - Beating is justified if woman burns food 358 0.15 0.12 0.01 0.56
  - Beating is justified if woman goes out without telling him 358 0.29 0.18 0.01 0.73
  - Beating is justified if woman refused to have sex 358 0.24 0.17 0.00 0.72
  - Beating is justified if woman neglects the children 358 0.32 0.19 0.02 0.80
  - Agriculture 337 0.34 0.22 0.00 0.73
  - Urban area 358 0.34 0.23 0.04 1.00
  - Time to water source 358 23.60 12.82 0.16 78.18
  - Roads, log 358 2.36 5.40 0.00 18.64
  - Education, log 358 1.88 0.44 0.47 2.56
  - Conflict 358 0.05 0.51 0.00 9.26
  - Religious fractionalization 358 0.61 0.25 0.01 1.00
  - Violence against men 349 0.04 0.03 0.00 0.23
  - Neighbors' domestic violence 358 0.16 0.05 0.06 0.33

- Table A 3: Baseline results using OLS, disaggregating violence against women
  - Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1. Dependent variable: nighttime light per capita.
  - Column (1) Dependent variable: Any physical results of husband's actions
    - Violence against women: -1.1905** (0.493)
    - Agriculture: 0.1529 (0.204)
    - Urban area: 1.1785*** (0.229)
    - Time to water source: -0.0079** (0.003)
    - Roads: 0.0636*** (0.015)
    - Education: 0.2873* (0.163)
    - Conflict: -0.1085*** (0.028)
    - Religious fractionalization: -0.0834 (0.346)
    - Constant: -0.4550 (0.471)
    - Observations: 318
    - R-squared: 0.789
    - District fixed effects: Yes; Year fixed effects: Yes; Country fixed effects: Yes
  - Column (2) Dependent variable: Spouse ever punched with fist or something harmful
    - Violence against women: -1.1746* (0.673)
    - Agriculture: 0.2271 (0.204)
    - Urban area: 1.1282*** (0.228)
    - Time to water source: -0.0108*** (0.004)
    - Roads: 0.0608*** (0.015)
    - Education: 0.2930* (0.152)
    - Conflict: -0.1153*** (0.025)
    - Religious fractionalization: -0.3708 (0.328)
    - Constant: -1.6239*** (0.476)
    - Observations: 337
    - R-squared: 0.774
    - District fixed effects: Yes; Year fixed effects: Yes; Country fixed effects: Yes
  - Column (3) Dependent variable: Spouse ever slapped
    - Violence against women: -1.1980*** (0.414)
    - Agriculture: 0.2344 (0.198)
    - Urban area: 1.0939*** (0.223)
    - Time to water source: -0.0097*** (0.004)
    - Roads: 0.0603*** (0.015)
    - Education: 0.3659** (0.148)
    - Conflict: -0.1067*** (0.029)
    - Religious fractionalization: -0.3495 (0.315)
    - Constant: -1.5570*** (0.480)
    - Observations: 337
    - R-squared: 0.781
    - District fixed effects: Yes; Year fixed effects: Yes; Country fixed effects: Yes
  - Column (4) Dependent variable: Spouse ever pushed, shook, or threw something
    - Violence against women: -2.7919*** (0.606)
    - Agriculture: 0.1453 (0.191)
    - Urban area: 1.1374*** (0.218)
    - Time to water source: -0.0077** (0.003)
    - Roads: 0.0611*** (0.014)
    - Education: 0.4121*** (0.151)
    - Conflict: -0.1157*** (0.026)
    - Religious fractionalization: -0.2299 (0.298)
    - Constant: -1.5601*** (0.473)
    - Observations: 337
    - R-squared: 0.791
    - District fixed effects: Yes; Year fixed effects: Yes; Country fixed effects: Yes
  - Column (5) Dependent variable: Spouse ever threatened with harm
    - Violence against women: -2.4185*** (0.604)
    - Agriculture: 0.1523 (0.202)
    - Urban area: 1.1009*** (0.225)
    - Time to water source: -0.0102*** (0.003)
    - Roads: 0.0611*** (0.015)
    - Education: 0.4344*** (0.167)
    - Conflict: -0.1363*** (0.024)
    - Religious fractionalization: -0.4290 (0.303)
    - Constant: -1.6982*** (0.486)
    - Observations: 337
    - R-squared: 0.783
    - District fixed effects: Yes; Year fixed effects: Yes; Country fixed effects: Yes
  - Column (6) Dependent variable: Spouse ever experienced any emotional violence
    - Violence against women: -0.6882** (0.340)
    - Agriculture: 0.2133 (0.207)
    - Urban area: 1.1415*** (0.231)
    - Time to water source: -0.0106*** (0.004)
    - Roads: 0.0592*** (0.015)
    - Education: 0.3053* (0.159)
    - Conflict: -0.1183*** (0.026)
    - Religious fractionalization: -0.3635 (0.324)
    - Constant: -1.5826*** (0.494)
    - Observations: 337
    - R-squared: 0.774
    - District fixed effects: Yes; Year fixed effects: Yes; Country fixed effects: Yes
  - Column (7) Dependent variable: Spouse ever experienced any sexual violence from partner
    - Violence against women: -2.1538*** (0.627)
    - Agriculture: 0.2136 (0.202)
    - Urban area: 0.9274*** (0.229)
    - Time to water source: -0.0109*** (0.003)
    - Roads: 0.0588*** (0.014)
    - Education: 0.4511*** (0.163)
    - Conflict: -0.1113*** (0.029)
    - Religious fractionalization: -0.3687 (0.304)
    - Constant: -1.7114*** (0.473)
    - Observations: 327
    - R-squared: 0.789
    - District fixed effects: Yes; Year fixed effects: Yes; Country fixed effects: Yes

*IMF WORKING PAPERS The Heavy Economic Toll of Gender-based Violence: Evidence from Sub-Saharan Africa*

---


_Source: https://www.imf.org/-/media/files/publications/wp/2021/english/wpiea2021277-print-pdf.pdf_