The latest research in economics on Africa: The CSAE round-up

This post is a joint production with contributions from Niklas Buehren, Aletheia Donald, Michael O’Sullivan, Sreelakshmi Papineni, and Julia Vaillant.

This week’s CSAE conference saw an epic 109 sessions with 351 papers presented.  The whole program can be found here.   Videos of some of the paper sessions can be found here, in addition to a great keynote by Macartan Humphreys on evidence in the age of fake news and a thought provoking panel discussion of refugees and migrants.      

What follows is a snapshot of some of the work, organized by themes.  

Agriculture

• The decision to adopt new agricultural technologies is positively influenced by the degree to which farmers feel they are in control. #Ethiopia (Abay)
• Video messaging can be an effective way to teach farmers about seed selection and storage practices. #RCT #Uganda (Vandevelde)
• Farmers have a good sense of the prevalence of counterfeit herbicides and accordingly use it less than at the desired level. #Uganda (Karachiwalla)
• Is there an inverse relationship between plot size and yield:  no, using national survey data from #Ghana (Tuffour). 
• Explaining the inverse plot size-productivity using plot-level panel data in #Uganda:   It's because productivity is higher on the edges.   Differential allocation of labor may play a role in why edges are more productive (Bevis). 
• The inverse plot size-productivity relationship is due to systematic measurement error (aka bias) in self-reported production, based on #Ethiopia data.  When you use crop cuts as an estimate of yields there is a positive relationship between plot size and productivity (Desiere).  
• Using data from #Uganda, with heaps of measures, another argument that the inverse productivity-plot size relationship is due to biases in reporting.  Potentially due to heaping of farmer's responses.  Bonus: satellite estimates of yields line up more with crop cut data than farmer reported yields (Kilic).  
• Extension and subsidies in #DRC increase knowledge, yields and food security, but no long term effects on adoption.   Subsidies boost input use.  Evidence from a diff in diff (Nillesen). 
• Expanding extension in #Ethiopia increases switch to commercial crops.   Women benefitted equally, but the intervention didn't close gender gaps.  Evidence from a diff in diff (Vaillant)
• Increased tractor use doesn't appear to be associated with increased yield and area planted using an IV in #Ghana (Cossar).
• Land formalization yields sustained investment effects in Benin #RCT (O’Sullivan et al.)
• The inverse relationship between size and farm productivity withstands the inclusion improved measures for land area and inter-cropping in #Tanzania (Boulay)
• Contract farming is associated with lower income variability in #Madagascar (Bellemare et al.) 

Conflict

• Exposure to sexual violence during the civil war increases post-war domestic violence among veterans in #Angola. Violence perpetrated is physical and arises from psychological factors. (Brück & Stojetz)
• Boycott of “conflict minerals” from the #DRC resulting from the Dodd Frank Act increased infant mortality in villages near the mines. Effect could be linked to lower use of health care services. #Diff-in-diff (Parker et al.)
• Rising diamond prices can increase conflict locally if the diamonds are lootable. If they are not, conflict increases nationally. (Rigterink)
• The public health impacts of civil conflict go far beyond combat deaths and can still be felt decades later. (Dunne)

Education (note: this section is based primarily on papers submitted rather than presentations)

• Can aspirations be too high? The relationship between aspirations (relative to current status) and human capital investment has an inverse-U shape (Ross). #India

• Among wealthy families, there is a positive effect of family size on schooling: i.e., no quantity-quality tradeoff (Alidou & Verpoorten). [Data from 35 African countries]
• Different household assets have different effects on child education: Housing quality = positive; agriculture assets = negative (Kafle). #Tanzania #IV
•  “On average, men and women do not have different preferences for investments in children, but more than 80% of the couples have different time preferences” (Ringdal & Sjursen) #Tanzania #LabInField
• Providing solar lamps to 7^th graders increased math grades, probably by increasing study at school after sunset (Hassan & Lucchino). #Kenya #RCT
•  “Cross-section data…suggest no gain in the completed years of education and matriculation completion for both black African men and women relative to whites, after Apartheid” (Patel) #SouthAfrica
• Technology-aided instruction that “precisely targets instruction to students’ preparation level” has big learning impacts (Muralidharan et al.) #India #RCT [Dave Evans’ blog post]
• Better school management = more student learning. But private schools are not better managed than public schools despite more autonomy (Crawfurd). #Uganda
• Teachers matter: “A 1-SD increase in teacher quality leads to at least a 0.27 SD improvement in student performance on a reading test” (Buhl-Wiggers et al.). #Uganda #RCT
• Universal primary education increased education (wait for it!) and also consumption. Effect in urban areas is twice that in rural areas (Delesalle). #Tanzania #IV
• Making secondary school free increased students taking the secondary exit exam by 16% AND may have even raised test scores (Masuda & Yamauchi) #Uganda
• Eliminating primary school fees increased schooling by 1.5 years for women, reduced births, and delayed sexual activity and marriage (Chicoine). #Ethiopia #IV
• Improving the quality of community-based childcare centers plus providing group-based parenting support had short-term child development impacts, but these had dissipated 18 months later (Ozler et al.) #Malawi #RCT
• Want kids to learn English? Teaching them in their mother tongue first result in better math and literacy scores after they transition to English (Seid). #Ethiopia #TripleDifferences #YoungLives
• How much does the quality of your preschool matter? It may vary depending on parent education (Morabito et al.). #Mauritius #RCT
• How much does preschool matter in low-income countries? 25 percent more likely to complete secondary education (Woldehanna & Araya). #Ethiopia #YoungLives

Firms

• Non-cognitive skills based business training yields a 31% increase in profits for firms in #Togo based on #RCT evidence and in comparison to a more standard business training (Goldstein - coming soon).
• Export shocks draw workers out of self-employment (enterprises), and we can better understand underinvestment in capital.  A nice model and IV estimates from #Mexico (Koelle).
• Information plus some convincing gets some but not many firms to formalize in an #RCT in #Benin.  Enhancing the benefits of formalization helps some (Pouliquen).

Institutions

• When playing the dictator game, coal miners in #SouthAfrica were more generous with members of their ethnic group. Union membership works as a club, generating favoritism among insiders. #LITF (Ravetti et al.)
• Colonial history can explain today's land institutions and inequalities in access to land between ethnic groups in the #DRC #IV
  (Vinez - Coming soon!)
• High-frequency panel data of intertemporal choice among dairy farmers in #Kenya shows that cash constraints affect preferences over the timing of income (Geng et al.) 

Labor

• Incentivizing job applications can have positive impacts on the number of applications and the quality of job applicants. #RCT #Ethiopia (Caria)
• Using reference letters in applications increase call back rates and make it more likely for women to secure jobs. #RCT #SouthAfrica (Burger)
• Decent work increases consumption just as much as having a household member migrating. #Senegal (Gueye)
• Working next to higher ability workers increases your output in an #RCT on tea estates in #Malawi.  It looks like social preferences may play a role: people want to work next to people who work hard because they think it will motivate them (Kerwin).
• Increased food prices associated with higher child labor in #Uganda using an #IV (Frempong).
• Videos of inspiring business women/farmers plus discussion increases income, health, child absence from school, and informal savings for women living with HIV.   Evidence from an #RCT in #Uganda (Narciso).
• Decoupling wages from outcome leads to effort that is increasing in wages.   Effort is lower if the wage offered is unfair.  Lab in the field results from #Uganda (Pamuk)

Poverty and Inequality

• Estimating the semi-elasticity of the poverty rate to initial poverty shows that there is in fact evidence of poverty convergence. However the convergence rate is too slow to reach the target of ending extreme poverty by 2030. (Wacker et al.)
• Using data from four large towns in #Ethiopia:  rural households reclassified as urban as cities expand are poorer than surrounding rural households (Mezgebo).  
• Using LSMS-ISA data to examine the circular relationship between land degradation and poverty in #Malawi and #Tanzania:   50.6% of households in Malawi and 25.9% in Tanzania experienced land degradation between 2009-2013 (Kirui).  
• Using continent-wide data: good soil and no infrastructure may be a curse -- the resource curse in the dirt (Wantchekon).    
• Using IV and RD (respectively):   Child grants and old age grants in #South Africa are associated with lower multidimensional poverty.   Stronger effects from the (larger) old age grant (Pasha).
• Land redistribution in #Zimbabwe was associated with a drop in cultivated cropland and lower vegetation quality #DiffInDiff (von Fintel & Chingozha)
• Mandating the registration of mobile SIM cards in #Niger resulted in a 33 p.p. drop in non-agricultural household enterprises #IV (Annan & Sanoh)
• Unconditional cash transfer program in Malawi increased resilience, consumption, & food security #RCT (Handa & Otchere)
• High-quality off-grid solar may still be beyond the reach of the rural poor in #Rwanda, suggesting the need for subsidies #RCT (Grimm et al.)

Rural households

• DHS data show that when spouses disagree on decision-making over large purchases, men generally presume power over their wives, while women concede power to their husbands. Decision-making is one of several dimensions of women’s agency whose measurement need to be improved in empirical studies. (Donald et al.)  
• What's the willingness of farmers in #Ethiopia to adopt conservation agriculture?   Offering a hypothetical climate smart agriculture indicates that only 53% would take minimum tillage, and 44% would take zero tillage.   Those who would adopt what compensation of $117 and $183 respectively (Mistru).  
• Higher commodity prices increase demand for land titles in #Uganda.   Channels appear to be both higher income and greater insecurity (Perego).  
• Evidence shows that land titling improves perceived tenure and leads to increased access to credit, off-farm earnings and activity on land markets. #RDD #Ghana (Buehren)
• Mobile phone services may provide people with access to mobile financial services in the absence of physical infrastructure. (Mothobi)
• Increasing the accountability in farming groups can having positive impacts on production but may cause farmers to be wary of cooperation and other undesirable outcomes. #RCT #Liberia (Beekman)
• Unconditional cash transfers paid to mothers can increase women’s savings and entrepreneurial activities. #RCT #Zambia (Natali)
• Cash transfer do provide a basic mechanism to smooth income shocks. #RCT #Zambia (Bonilla)
• Accelerometry devices – implemented together with diaries – can shed new light on gendered patterns of time use: women’s greater burden of domestic activities comes at the expense of their social interactions #Ghana (Nkegbe et al.)
• Households in #Malawi strategically pool income: members partially insure one another for expenditures on essential goods #IV (Josephson)
• Women in Sub-Saharan Africa who speak languages with stronger distinctions between masculine and feminine are less likely to participate in the labor force (Rao)
• The introduction of universal primary education in #Uganda significantly increased girls’ education, which reduced the probability of a brideprice being paid at their marriage #RD (Nagashima and Yamauchi)
• Black women in #SouthAfrica earn more from an additional year of tenure with the same firm and less from labor market experience compared to other groups, but the observed gap is largely driven by how labor market experience is measured (Malindi)
• Adoption of improved agricultural seeds is associated with higher household welfare in Uganda and Tanzania (Kinuthia & Mabaya)
• Large extended households in #BurkinaFaso reduced their farm labor inputs – possibly due to coordination challenges – after a labor-intensive technology was introduced #RCT (Ouedraogo et al.)

Social Networks and Households

• Positive capital shocks at the village level in the #Gambia reduce economic interactions between individuals (using network mapping).   Mostly around goods and gifts (Jaimovich).
• People give the most when they can’t hide income and they are subjected to claims by others.   Income increases result in more giving only when people can't hide the income.  And people share more with people who are presented as closer to them.   Evidence from lab-in-the-field in #Tanzania (di Falco).  
• Using DHS data across SSA:  electrification is correlated with lower acceptance of domestic violence and unobservables matter a lot (Sievert).
• Turning the lights on: solar lamps (including a phone charger) in an #RCT in #Tanzania lead to spending less on kerosine, charging your phones and on lighting.   And then more people working outside the home and (when a book was given too) more time studying (Bold).  
• Women’s involvement in decision-making decreases with the importance of the decision. They have internalized gender norms and have less demand for agency than men. #Pakistan #LITF (Afzal et al.)
• Monetary incentives increase knowledge sharing in a cost-effective way. #RCT #Uganda (Sseruyane & Bulte)

• In #Senegal, the death of a female adult increases girls’ weekly time spent on household chores by 9 hours. Children directly under the responsibility of a deceased adult have a significant decrease in schooling and increase in work. #Diff-in-diff (De Vreyer & Nilsson)
• Women’s empowerment should be measured through attitudes and behaviors, in the economic, social, political and psychological domains. (Fox & Romero)
• Large within country, within year variations of attitudes towards intimate-partner violence are found using DHS data. Negative climate shocks increase acceptance of violence among women. (Anttila-Hughes et al.)
• Using an #IV, #Cameroonian women who are more educated get married to wealthier men who, in turn, are more likely to be polygamous (Dupraz).  
• Early marriage reduces schooling, leads to more agreement with traditional gender norms and leads to less progressive networks.   Results from #Bangladesh using an #IV. (Wahhaj)
• An experiment in the field to measure female empowerment has a low correlation with survey based measures.   A graduation intervention in #Kenya lead to changes in the experimental measure, but not on survey measures (Santos).  
• An increase in wives’ estimated consumption gains from remarriage leads to a higher probability of divorce #Malawi (Walther)
• Extended family households are more inefficient than nuclear families, and in-laws behave less cooperatively with each other compared to spouses #Lab #India (Ksoll et al.)
• Marriage exacerbates the individual’s present bias problem by providing more resources for them to consume: present-biased individuals turn over less money to their spouses while receiving more, and are more likely to keep cash in the house #Lab #Vietnam (Kono and Tanaka) 
• Women in urban Mali imitate their peers – especially those whom they respect – when deciding whether to purchase improved cookstoves #RCT (Bonan et al.)
• Giving a transport subsidy to job-seekers reduces their peers’ job search efforts and reduces job information sharing #RCT (Caria et al.)