ABSTRACT

In many parts of the world, women face barriers to accessing improved agricultural inputs. In Tanzania, a program called the National Agricultural Input Voucher Scheme (NAIVS) was implemented from 2008 to 2014 to provide smallholder farmers with vouchers to purchase inputs at a subsidized price, with the goal of improving productivity and increasing income and food security. Using data from the Tanzania Living Standards Measurement Study panel, we analyzed the effects of NAIVS on female-headed households’ market participation, sales, and profitability. The analysis showed that NAIVS increased market participation of female-headed households by 11%, particularly in downstream positions in domestic value chains. However, there were no discernible effects on sales and profitability, indicating that although the program reduced input costs, production costs remained high enough to offset the efficiency gains from the subsidized inputs.

INTRODUCTION

Modern farm inputs, specifically improved seeds and fertilizers, are increasingly adopted in Africa and other parts of the world to improve productivity and welfare (Johnson et al. 2003; Sheahan and Barrett 2017). To assist such endeavors, governments have implemented input subsidy programs (ISPs) to reduce the burden on farmers, increase food stocks, and regulate prices. Examples of large ISPs in Africa include those of Ethiopia, Kenya, Uganda, Malawi, Zambia, Ghana, and Nigeria (Jayne et al. 2018). In 2008, the government of Tanzania implemented the National Agricultural Input Voucher Scheme (NAIVS) in response to the 2007–2008 international food crisis. The program aimed to ensure food security at both the household and national levels by providing improved seeds and chemical fertilizers to smallholder maize and/or rice farmers (World Bank 2014).

Despite these efforts, many smallholders, particularly women, continue to lack access to a range of inputs necessary to enhance their productivity and profitability. For example, data from the Tanzania National Panel Survey (TZNPS) between 2008 and 2012 show that female-headed households use only about a quarter of the modern inputs used by male-headed households. Moreover, a study by Sheahan and Barrett (2017) shows that male-headed households in Nigeria, Niger, Ethiopia, Malawi, Uganda, and Tanzania are more likely to adopt modern inputs. While there is limited empirical evidence on ways to improve access to these inputs, particularly for women, little is known about the impact of NAIVS on market participation and profitability of farm business in Tanzania.

In this article, we study the impact of NAIVS on market participation and farm-related enterprises in Tanzania with a gender lens. Even if there is substantial support from the government or any other stakeholders, we must not ignore how gender imbalances in society can deter the implementation of the development agendas. In sub-Saharan Africa (SSA), the gender gap can be observed at household, community, and organizational levels (Woldu et al. 2015). According to Woldu et al. (2015), men in Ethiopia are favored because of educational attainment and their roles as household heads, which make them the owners of family lands. This is also the case in Tanzania; data from 2008, 2010, and 2012 from Tanzania-NPS shows that about 75% of households are led by men, hence decision-making in the production and agricultural market participation. As women face constraints in agricultural production and market participation, input subsidies may change the situation provided that they have access to such programs. For example, Fisher and Kandiwa (2014) prove that female-headed households receiving input subsidies are likely to modernize maize production by 222%.

Failure to address gender issues in the implementation of important development programs reduces the chances to attain Sustainable Development Goal 5 pertaining to gender equality. There is evidence that women in rural areas are improving household economies, ensuring food security, and acquiring new skills as they participate in agriculture and entrepreneurship (Ferguson and Kepe 2011; King, Sintes, and Alemu 2012). Another key example of women’s contribution can be drawn from the share of women in the labor force in crop production—about 50% in Tanzania, and 40% on average in SSA (Palacios-Lopez, Christiaensen, and Kilic 2017). As women are increasingly involved in agribusiness, considerable attention is taken by stakeholders aiming to facilitate the efforts to tackle gender matters. To date, there is still a debate about the definition of entrepreneurship. Previous literature has mixed results when it comes to the definition; some studies define it as the establishment of a new venture, which involves creating ideas and adapting to market opportunities. In this study, we use the definition by Venkataraman (2019) in which entrepreneurship involves the opportunities for the creation of future goods and services in the market. According to this definition and in the context of agricultural sector, entrepreneurship includes the involvement of smallholder farmers in agribusiness activities. As farmers experience surplus production, say due to subsidies, they are likely to enter the market for their extra output or create new goods for sale by processing their crops.

A few studies pinpoint the roles of ISPs on gender profitability. This is to say, little attention has been paid to understanding if the women beneficiaries can improve their participation in the market as well as their profitability. Sibande, Bailey, and Davidova (2017)—one of the close studies that investigated the effects of subsidized fertilizer on maize marketing in Malawi—found that participating in these ISPs increases the maize market participation and quantity sold. But the study did not explore the impact of ISPs on positioning in the value chain with a gender lens. In this article, we extend this knowledge by examining the impact of the input subsidies on women entrepreneurs and their farm-related enterprises by disintegrating the market structure into which farmers participate. The gender-view analysis of NAIVS sheds light on the gender-based impacts of input subsidies. In general, input subsidies would increase productivity, likelihood to participate in the market chain, and profits derived thereafter. Our study discusses the impacts on women-headed households and sees if participating in the agricultural market as a result of the program improves entrepreneurship. The key research question, therefore, is: do NAIVS receipts increase women’s participation in agricultural value chains?

This article makes two specific contributions. First, it expands the small but growing literature on the causal impacts of input subsidies in SSA with a gender lens. Second, it provides reliable estimates that will be useful when formulating women’s empowerment strategies and initiatives. For our empirical strategy, we exploit panel data from three waves (2008–2012) of Tanzania’s Living Standards Measurement Study-Integrated Surveys on Agriculture (LSMS-ISA). An advantage of using panel data compared with cross-sectional data is that we can take into account unobserved individual heterogeneity and observe the trajectories in the access to input subsidies and labor market participation and how labor market participation changes following eligibility to input subsidies. The key finding shows that NAIVS significantly increases female and male farmers’ participation in the market by 11.2% and 4.8%, respectively for the period included in the analysis, mostly downstream, that is, closer to the final market; this likelihood is larger than that of male farmers. This suggests that female inclusion in NAIVS increases their chances to participate in the domestic value chain. The fact that the results show a higher likelihood among women than men proves that the ISP can not only contribute to reducing the gender divide in the agricultural value chain, but also improve agribusiness. However, no significant impact can be found on profitability.

The rest of the article proceeds as follows: Section 2 provides a review of related literature on ISPs and agricultural development, followed by a brief introduction about the NAIVS program in Tanzania. Section 3 provides a detailed description of Tanzania’s LSMS-ISA data and the empirical strategy. Section 4 presents the results and Section 5 concludes.

BACKGROUND

Literature Review

Women’s participation in agricultural markets is an important aspect in reducing gender disparity and achieving sustainable development. This offers a means to create female entrepreneurs who in turn may earn extra household income useful in fighting poverty and improving welfare (Osei and Zhuang 2020). Ackah and Aryeetey (2012) show that women’s participation in cocoa farming is positively associated with household welfare, but they often have lower returns than men due to their limited access to resources. To address these constraints, policies must be targeted toward women and designed to address specific challenges they face, such as access to inputs, credit, and extension services. Fisher and Kandiwa (2014) found that agricultural input subsidies targeted toward women in Malawi can reduce the gender gap in maize adoption, which can lead to increased yields and improved market participation.

To improve household welfare in SSA, ISPs should be fairly accessible between male and female-headed households. In Malawi, Karamba and Winters (2015) show that subsidy program improved women’s access to inputs and increased their agricultural productivity. The same is true in Tanzania’s NAIVS, which intentionally targeted female-headed households. As more women have unconstrained access to vital agricultural inputs, household productivity, income, and welfare are set to be improved. This may create a society that values the contribution of women to household, rural, and national development.

Well-implemented agricultural subsidy programs can yield outcomes, which may be witnessed in the quality of their produce (Fisher and Kandiwa 2014). According to Fisher and Kandiwa (2014), ISPs reduce the gender gap because as women participate in the program, they increase their crop cultivation. This can also impact their entrepreneurship activities induced by surplus production and market participation. It is important to analyze if women are benefiting from these programs through participating in agricultural markets while learning entrepreneurship and improving their skills. Women’s participation in both production and market is vastly determined by their decision-making ability. Women’s participation in cash cropping increases their bargaining power within the household, which can lead to improved decision-making and well-being (Ackah and Aryeetey 2012). In cases where women have met fewer obstacles in agricultural production, studies show that household incomes were improved (Roy et al. 2017).

Other studies consider the introduction of ISPs as a milestone to address gender issues in agricultural production by focusing on ownership of means of production, mainly land (Fisher and Kandiwa 2014; Karamba and Winters 2015; Slavchevska 2015). Even though most land in developing countries are owned by men, women still find ways to participate in decision-making, which may help to determine what kind of crops can be planted and which markets they should be sell in. However, little attention has been given to understanding if participating in ISPs can improve women’s participation in the market as well as their profitability. The understanding of women’s participation and the benefits thereafter will shed light on how these programs have addressed gender issues.

NAIVS Program

Agriculture is a significant contributor to Tanzania’s economy, accounting for approximately 29% of the country’s GDP and providing employment for 65% of the population, mostly living in rural areas (World Bank 2020). Despite agriculture being the primary occupation of most people in rural Tanzania, the sector faces various challenges, including those related to climate shocks. Tanzania experiences both unimodal and bimodal rainy seasons, which adversely affect agricultural productivity and farmers’ food insecurity level. Consequently, households in areas that experience unimodal rainy seasons may face food insecurity for several months before the next harvest season.

To improve food insecurity in rural households, the ISPs were implemented. In addition, since the majority of rural farmers engage in maize and rice production, improving such crop cultivation has the potential to improve food security and reduce poverty. For example, about 5.6 million tons of maize were produced in 2019, harvested from nearly 3.5 million hectares of land (Food and Agriculture Organization [FAO] 2021). On the other hand, about 3.5 million tons of rice were produced in the same year from the land size that was about a third of the maize cultivated area. Growth in agricultural production is related to improved farming practices. Such practices may include the application of irrigation, the use of machinery, and the use of improved seeds and chemical fertilizers. However, prices of important inputs, specifically improved seeds, and fertilizers may not be stable mainly due to factors, such as disruption in supply chains. To respond to the increase in grain and fertilizer prices in 2007 and 2008, the government provided subsidies to smallholder farmers (World Bank 2014). The introduction of NAIVS in 2008 was not only set as a rescue to the troubled food production, but also marked a turning point in agricultural production in Tanzania.

During the early years of operations, NAIVS targeted only 12 provinces in the country that were considered to have high productivity. Most of these provinces are in the southern and northern highlands, including Mbeya, Iringa, Ruvuma, Kilimanjaro, and Manyara. From the 2011/2012 agricultural season, the government extended the program to the national level covering a total of 21 provinces, which impacted about 2.5 million smallholder farmers. The program was voucher-based, where beneficiaries received three vouchers to be redeemed at the local dealers within their respective villages. The three vouchers included benefits of 10 kg of either hybrid seeds or improved open-pollinated maize variety, 15 kg of paddy seeds, 50 kg bag of basal fertilizer, and 50 kg bag of top-dressing fertilizer. The program subsidized 50% of the payment for the inputs.

In principle, the NAIVS program was designed to target “full-time farmers” who cultivate maize or rice on plots of 1 hectare or less and are willing to follow the advice of extension workers. Furthermore, farmers who were unable to afford the cofinance were excluded. Priority was given to households who had purchased few or no inputs in the last 5 years and female-headed households because the program aimed to increase input utilization. Regarding choosing the eligible farmers, the village voucher committee (VVC) was responsible to select the beneficiaries. VVC was believed to know more about the households in their community and hence, reduce the chance of choosing elites or unqualified households. Once the VVC identified the eligible farmers, they had to make sure that the beneficiaries had the co-payment so that they could sign and receive the vouchers. The agro-dealers received the vouchers, submitted them to the local National Microfinance Bank (NMB) for reimbursement of the other half of the voucher value.

DATA AND METHODOLOGY

Data

This article examines the impact of the NAIVS on women entrepreneurs in the agricultural sector. Data on households, community, and crop production and marketing are necessary to understand the relationship between the provision of input subsidies and women’s entrepreneurial activities. Therefore, we use the first, second, and third waves of Tanzania’s LSMS-ISA data of 2008–2012, which is well-suited to answer the research questions. The first wave had 3,265 households, while the second and the third had 3,924 and 5,010 households, respectively. The data is collected by Tanzania’s National Bureau of Statistics in collaboration with the World Bank.

The datasets have detailed information on households’ agricultural activities, giving us opportunity to evaluate the effects of the NAIVS participation. A question like “Did you receive [and redeem] a voucher/certificate for this [FERTILIZER/SEED]?” was used to identify farmers who participated in the program. Furthermore, information on the quantity of fertilizers and improved seeds received, cash paid by farmers as co-pay, the quantity of maize and rice harvested, sales of the harvested crop, and voucher redeemed were included in the analysis. Moreover, the household data provided information on household demographics where our main focus was on female-headed households. For community data, the survey provides information on the availability of services at the community level. Such services include Savings and Credit Cooperatives (SACCO), which will be used to determine women’s participation in microfinancial activities that are expected to improve their entrepreneurship. As it is important in other sectors, finance is also an important improvement of farm productivity and business operations.

We focus on agricultural profitability by women-headed households. We calculate the profitability in maize and rice cultivation as the difference between the total receipts from participating in the market and the costs of operations. We measure the land size by considering only the size of plots from which crops were harvested. In this regard, we ignore the land size estimates provided in the survey by farmers. The main reason is that the farmers’ estimates are bigger than the measurements and some plots were not measured at all. Even though the land size is important in determining accessibility to input subsidies, the dataset does not show that. We find some farmers with more than 1 hectare, but received subsides only for 1 hectare input. Consequently, we cannot assume that the inputs received will only be used in that 1 hectare out of the total land cultivated. This can also be used to measure the impact of NAIVS with the expectation that productivity will somehow improve with additional inputs.

In Figure 1, we map the incidence of NAIVS programs in the different years of our sample. The incidence is 2.4% in 2008, which seems to be low as it reflects the beginning of the program. In 2010, the incidence of NAIVS reached 7.2% but decreased to 4.6% in 2012. The incidence in 2012 is lower than in 2010 because of the graduation of many farmers from the program and because the number of beneficiaries did not increase despite the increase in the area covered by the program (World Bank 2014). We can see that there are large variations across provinces and over time in the incidence of NAIVS. We will exploit these variations using fixed effects regressions to identify the effect of the NAIVS program on household market participation.

FIGURE 1

NAIVS Incidence Rate. Note: This map reports NAIVS incidence rate in our sample. Provinces not included in our sample are reported as “Dropped.”

FIGURE 1

NAIVS Incidence Rate. Note: This map reports NAIVS incidence rate in our sample. Provinces not included in our sample are reported as “Dropped.”

Close modal

Next, we present the basic relationship between household expenditure as well as sales as binned scatterplots, controlling only for the region- and year-fixed effects with a bivariate regression line fit to the plotted bins. Figure 2 Panel A is for household expenditure. We see a similar downward trend for both female and male: poor households have been paying higher costs to redeem input vouchers than richer households. Panel B shows the association with sales. We see a strong association with sales, with clear gender differences in the slopes. This could imply that the availability of other sources of finance could have played a role in increasing access to NAIVS for the poor smallholders.

FIGURE 2

Correlation among NAIVS Cost, Total Expenditure, and Total Sales. Panel A: NAIVS and household expenditure. Panel B: NAIVS and crop sales.

FIGURE 2

Correlation among NAIVS Cost, Total Expenditure, and Total Sales. Panel A: NAIVS and household expenditure. Panel B: NAIVS and crop sales.

Close modal

Our sample for the rest of the variables is summarized in Table 1. Panel A gives the market participation characteristics of the sample while Panel B and C show demographics and community-level characteristics, respectively. As expected, NAIVS households have a higher mean in all the market participation outcomes, except for sales and profit. From the table, NAIVS participants have higher output; they harvest about 76% more crops than nonparticipants. Their presence in the market is significantly more than nonparticipants’ by 18.8%. Consequently, they sell more within the possible value chain by approximately 20%. Demographic characteristics are almost identical, except for household size and rural location. Community-level characteristics show that NAIVS households live in communities that are closer to town centers, markets, and roads by about 16%, 11%, and 4%, respectively. These variables constitute the most common determinant of market access.

TABLE 1

Descriptive statistics

PooledFemaleMale
NAIVSNAIVS
minus
Non-NAVIS
NAIVSNAIVS
minus
Non-NAVIS
NAIVSNAIVS
minus
Non-NAVIS
 Panel A 
Seller 0.506 0.188*** 0.389 0.117* 0.536 0.205*** 
Inside the market chain 0.457 0.201*** 0.352 0.144** 0.483 0.213*** 
Downstream 0.283 0.135*** 0.278 0.164*** 0.284 0.126*** 
Upstream 0.174 0.066*** 0.074 −0.020 0.199 0.087*** 
Quantity harvested 6.720 0.757*** 6.270 0.751*** 6.834 0.743*** 
Sales (share of harvest) 0.265 0.023 0.160 −0.053 0.292 0.041 
Profit 0.595 −0.182*** 0.550 −0.204* 0.604 −0.178*** 
Fertilizer voucher 0.834 0.834*** 0.852 0.852*** 0.829 0.829*** 
Seeds voucher 0.362 0.362*** 0.333 0.333*** 0.370 0.370*** 
       
 Panel B 
Household size 5.193 −0.520*** 3.870 −0.703*** 5.533 −0.515*** 
1 if married 0.602 0.038 0.500 0.086 0.629 0.020 
School 0.087 0.021 0.056 0.030 0.095 0.017 
Age 48.211 −1.079 56.148 0.841 46.180 −1.352 
1 if rural area 0.754 −0.113*** 0.630 −0.201*** 0.786 −0.092*** 
       
 Panel C 
1 if SACCO is available 0.310 −0.008 0.396 0.080 0.288 −0.030 
Distance to town 36.978 −16.197*** 38.341 −10.829*** 36.628 −17.717*** 
Distance to market 73.620 −11.789*** 76.643 −3.891 72.843 −13.991*** 
Plot distance to road 17.728 −4.191*** 17.783 −2.268 17.713 −4.752*** 
Rainfall 826.936 21.417 835.222 40.393 824.805 16.163 
       
Observations 265 5,047 54 1,135 211 3,912 
PooledFemaleMale
NAIVSNAIVS
minus
Non-NAVIS
NAIVSNAIVS
minus
Non-NAVIS
NAIVSNAIVS
minus
Non-NAVIS
 Panel A 
Seller 0.506 0.188*** 0.389 0.117* 0.536 0.205*** 
Inside the market chain 0.457 0.201*** 0.352 0.144** 0.483 0.213*** 
Downstream 0.283 0.135*** 0.278 0.164*** 0.284 0.126*** 
Upstream 0.174 0.066*** 0.074 −0.020 0.199 0.087*** 
Quantity harvested 6.720 0.757*** 6.270 0.751*** 6.834 0.743*** 
Sales (share of harvest) 0.265 0.023 0.160 −0.053 0.292 0.041 
Profit 0.595 −0.182*** 0.550 −0.204* 0.604 −0.178*** 
Fertilizer voucher 0.834 0.834*** 0.852 0.852*** 0.829 0.829*** 
Seeds voucher 0.362 0.362*** 0.333 0.333*** 0.370 0.370*** 
       
 Panel B 
Household size 5.193 −0.520*** 3.870 −0.703*** 5.533 −0.515*** 
1 if married 0.602 0.038 0.500 0.086 0.629 0.020 
School 0.087 0.021 0.056 0.030 0.095 0.017 
Age 48.211 −1.079 56.148 0.841 46.180 −1.352 
1 if rural area 0.754 −0.113*** 0.630 −0.201*** 0.786 −0.092*** 
       
 Panel C 
1 if SACCO is available 0.310 −0.008 0.396 0.080 0.288 −0.030 
Distance to town 36.978 −16.197*** 38.341 −10.829*** 36.628 −17.717*** 
Distance to market 73.620 −11.789*** 76.643 −3.891 72.843 −13.991*** 
Plot distance to road 17.728 −4.191*** 17.783 −2.268 17.713 −4.752*** 
Rainfall 826.936 21.417 835.222 40.393 824.805 16.163 
       
Observations 265 5,047 54 1,135 211 3,912 

Note: Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Empirical Strategy

To estimate the effect of NAIVS on household market participation in a given time period, we exploit the panel nature of our data and the fact that access to NAIVS was introduced and changed in different years across several provinces between 2008 and 2012. Following Wooldridge (2021), we model the relationship using the following equation:

Yicrt=α+βNAIVSicrt+Xitδ+Krtρ+γr+δdt+μi+εicrt
(1)

where, yicrt is a set of outcomes of household i producing crop c in region r at time t. This includes the likelihood to be a seller, to participate in the value chains, and to be in the upstream, downstream, harvested quantity, sales, and profit. NAIVSirt is the measure of NAIVS exposure for household i producing crop c in region r at time t.1 Given that NAIVS is mainly targeting women, we run our analysis separately by reported gender. Xit is a vector of household characteristics, both time-variant and specific to the assignment rule of the NAIVS, which selects households as beneficiaries based on: highest degree completed, age, real annual income per capita household, consumption, the number of adult equivalents, and the size of landholdings. Krt are time-varying region characteristics: land rental price index, distance to the nearest road, distance to the nearest population center with more than 20,000 inhabitants, distance to the nearest market, average annual rainfall, and temperature, and share of individuals working for political parties. The variables are introduced to capture other regional policies that may interact with market participation and regional macroeconomic trends. μi are individual fixed effect, δdt are district-year fixed effect, and γr are region-fixed effect. εicrt is the error term and standard errors are clustered at the household level.

We are interested in β1, which is the effect of the NAIVS program. Our specification relies on household-level variation in NAIVS exposure. Specifically, the variation in identification comes from comparing the average change over time among observably similar households, producing the same crop, but who happened to be exposed to NAIVS relative to that same change among unexposed households in the same region. The main challenge to identification is that participation in the NAIVS is not random and is shaped by political and economic factors. For instance, 60% of the available vouchers were allocated to village leaders and VVC members, who were also in charge of the execution of the program in the village (Pan and Christiaensen 2012). Second, and related to this, the NAIVS program targeted smallholder farmers with 1 hectacre but with a condition that they cofinance 50% of the voucher costs. This could lead to cases where farmers could sell their vouchers to wealthier households who could redeem them. These factors are also more likely to affect market participation. Thus, controlling for region-fixed effects and district-year fixed effects could still lead to biased estimates because of unobserved characteristics. In addition, bias may remain if households’ characteristics, such as farming ability, risk aversion, and motivation to participate in NAIVS, or other unobserved factors predict selection into NAIVS, even when household fixed effects are included in the panel analyses.

Another potential threat is the fact that participation to NAIVS is underreported in the household survey. Our data suggest that NAIVS household represents 5% in our sample, while official data shows that the program beneficiary reached 10% of eligible farmers by 2011. One of the reasons is that many smallholder farmers who received the vouchers could not afford to redeem the inputs. While the guideline for the implementation and targeting of NAIVS may cause some households in some areas to be selected in the program, the observed participation is not likely to be exogenous. Table A1 in the  Appendix shows that participation to NAIVS is highly predicted by the log of expenditure, seed usage, and inorganic fertilizer use. The significance of other determinants, such as improved seeds, show that the targeting might not have been perfect. It shows that richer farmers could also have benefited from NAIVS. In this setting, the practice of controlling for the NAIVS selection criteria as covariates in Equation (1) does not allow for the assessment of covariate balance.

To deal with these challenges, we employ matching methods, namely the entropy balance. This technique creates a perfect balance by reducing the imbalance between the characteristics of the treatment and control groups. It reweights the sample so that the mean and the variance of the observations in the control group become similar to those of the NAIVS group for all variables considered. Entropy balancing is useful for its nonparametric properties and does not rely on assumptions about the functional form that is necessary for propensity score comparisons. We use entropy balance as our robustness test.

RESULTS

Main Findings

Table 2 presents our main results.2 All specifications include district- and year-fixed effects and a vector of controls for both individual and location-specific characteristics. We only report the coefficient on NAIVS from estimating Equation (1) by gender. Note that participation in NAIVS is much better observed for male in our data, which makes the gender comparison more interesting. While the sample for female recipients of NAIVS is rather small, analyzing within gender subsamples rather than adding a gender interaction term ensures that the regression specification accounts for how gender relates not only to our main variables of interest, but also to the full vector of control covariates. Perhaps, most importantly, stratification by gender ensures that the coefficient on the NAIVS indicator is determined solely on the basis of policy variation within a region, rather than on differences in outcomes for women or men across provinces, which may reflect many gender-related determinants of farming.

TABLE 2

Impact of NAIVS on market participation by gender

SellerInsideDownstreamUpstreamHarvested
quantity
SalesProfit
(1)(2)(3)(4)(5)(6)(7)
 Panel A 
NAIVS 0.112* 0.166** 0.122* 0.044 0.241 0.130 −0.311 
 (0.067) (0.077) (0.071) (0.062) (0.185) (0.141) (0.308) 
        
Observations 1,115 1,115 1,115 1,115 1,045 1,045 101 
R-squared 0.606 0.568 0.527 0.487 0.782 0.362 0.720 
        
 Panel B 
NAIVS 0.048 0.081** 0.063* 0.019 −0.005 0.154** 0.010 
 (0.042) (0.041) (0.037) (0.037) (0.081) (0.077) (0.084) 
        
Observations 3,857 3,857 3,857 3,857 3,648 3,648 514 
R-squared 0.604 0.577 0.524 0.473 0.757 0.447 0.590 
        
Controls Yes Yes Yes Yes Yes Yes Yes 
Household FE Yes Yes Yes Yes Yes Yes Yes 
District-Year FE Yes Yes Yes Yes Yes Yes Yes 
SellerInsideDownstreamUpstreamHarvested
quantity
SalesProfit
(1)(2)(3)(4)(5)(6)(7)
 Panel A 
NAIVS 0.112* 0.166** 0.122* 0.044 0.241 0.130 −0.311 
 (0.067) (0.077) (0.071) (0.062) (0.185) (0.141) (0.308) 
        
Observations 1,115 1,115 1,115 1,115 1,045 1,045 101 
R-squared 0.606 0.568 0.527 0.487 0.782 0.362 0.720 
        
 Panel B 
NAIVS 0.048 0.081** 0.063* 0.019 −0.005 0.154** 0.010 
 (0.042) (0.041) (0.037) (0.037) (0.081) (0.077) (0.084) 
        
Observations 3,857 3,857 3,857 3,857 3,648 3,648 514 
R-squared 0.604 0.577 0.524 0.473 0.757 0.447 0.590 
        
Controls Yes Yes Yes Yes Yes Yes Yes 
Household FE Yes Yes Yes Yes Yes Yes Yes 
District-Year FE Yes Yes Yes Yes Yes Yes Yes 

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Note: Standard errors are clustered at the household level

We report the estimates for female-headed households in Panel A and for male-headed households in Panel B. We see in Column 1 that NAIVS has a positive but discernible effect on the probability of selling for female. The results show that women participating in the NAIVS are 11.2% points more likely to sell to the market. The impact is large and statistically significant if compared to the model for male, which is 4.8% points and not statistically significant. This is consistent with the design of the program, which was set to give more advantage to female-headed households who also tend to be smallholders without market access. Column 2 shows the impact of NAIVS on the probability to participate in the value chain. We also see a distinctly positive effect: NAIVS appears to be associated with a high probability to participate in the value chains with clear differences between female and male. The coefficient of female is 0.166 and that of male is 0.081. As women have entered the market, they have also increased their position in the value chain, leading us to explore, in Columns 3 and 4, the effect of NAIVS on the type of value chains. For female-headed households, we find that NAIVS has a positive and significant effect on the participation in downstream value chains but did not affect their position in upstream chains. We also find similar results for male-headed households, but the magnitudes are smaller compared to female-headed households. Columns 5 through 7 show results for harvested quantity, sales, and profit. Focusing on women, Panel A shows that the ISP had a positive but not statistically significant on harvested quantity and sales, but a negative and not statistically significant effect on profit. Results for male in Panel B shows only positive and statistically significant effect on sales.

One issue with OLS is that it gives similar weight to all households. This raises the question of whether households with small NAIVS participation are excessively driving the results. Table 3 maintains the same specifications as Table 2 but reweight the estimator to balance the observable characteristics between NAIVS and non-NAIVS households. The weights are obtained from the entropy balance, controlling for a comprehensive set of observable variables that capture NAIVS exposure and market participation preferences as accurately as possible.3 The matching yields a smaller sample size but the findings are qualitatively similar. The weighted estimates are higher in this specification, and the corresponding effects for male-headed households increased slightly higher. Specifically, our coefficients for male-headed households are now 0.137, 0.224, and 0.108 for selling, participation, and downstream.4 We also see that the coefficient upstream for male-headed households is now large and highly significant. The coefficient for harvested quantity and profit become also large although insignificant, whereas the effect on sales becomes small and not significant. The fact that the reweighted estimates are higher than the baseline estimates is consistent with the general trend that the estimated NAIVS effects become more positive when more sources of selection are accounted for.

TABLE 3

Robustness test

SellerInsideDownstreamUpstreamHarvested
quantity
SalesProfit
(1)(2)(3)(4)(5)(6)(7)
 Panel A. Female 
NAIVS 0.145 0.247** 0.274** −0.027 0.401* 0.121 −0.619 
 (0.113) (0.114) (0.118) (0.146) (0.214) (0.588) (0.426) 
        
Observations 761 761 761 761 713 713 87 
R-squared 0.802 0.756 0.783 0.749 0.911 0.347 0.828 
        
 Panel B. Male 
NAIVS 0.137* 0.224*** 0.108* 0.116* 0.163 0.074 0.079 
 (0.074) (0.074) (0.064) (0.068) (0.132) (0.064) (0.136) 
        
Observations 2,542 2,542 2,542 2,542 2,405 2,405 449 
R-squared 0.778 0.786 0.699 0.682 0.871 0.393 0.717 
        
Controls Yes Yes Yes Yes Yes Yes Yes 
Household FE Yes Yes Yes Yes Yes Yes Yes 
District-Year FE Yes Yes Yes Yes Yes Yes Yes 
SellerInsideDownstreamUpstreamHarvested
quantity
SalesProfit
(1)(2)(3)(4)(5)(6)(7)
 Panel A. Female 
NAIVS 0.145 0.247** 0.274** −0.027 0.401* 0.121 −0.619 
 (0.113) (0.114) (0.118) (0.146) (0.214) (0.588) (0.426) 
        
Observations 761 761 761 761 713 713 87 
R-squared 0.802 0.756 0.783 0.749 0.911 0.347 0.828 
        
 Panel B. Male 
NAIVS 0.137* 0.224*** 0.108* 0.116* 0.163 0.074 0.079 
 (0.074) (0.074) (0.064) (0.068) (0.132) (0.064) (0.136) 
        
Observations 2,542 2,542 2,542 2,542 2,405 2,405 449 
R-squared 0.778 0.786 0.699 0.682 0.871 0.393 0.717 
        
Controls Yes Yes Yes Yes Yes Yes Yes 
Household FE Yes Yes Yes Yes Yes Yes Yes 
District-Year FE Yes Yes Yes Yes Yes Yes Yes 

Standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1

Note: Controls variables are explained in the text. Standard errors are clustered at the household level.

Heterogeneity

To better understand the mechanisms behind these effects, we leverage rich heterogeneity in the data. Given large differences between the NAIVS type and value chain position, as well as significant variation among NAIVS beneficiaries, we explore heterogeneous effects by NAIVS type and value chain positions. We also directly compare the effects of NAIVS to those of SACCO.

In Table 4, we first explore the results by NAIVS type. We reestimate Equation (1) but this time, we replace NAIVS with fertilizer vouchers and seeds vouchers. As shown, we find stark differences in effects by NAIVS type. For female-headed households, a fertilizer voucher increases the probability to be a seller by 13.4% points and the probability to participate in the value chains by 18.6% points. These results, to some extent, resemble Sibande, Bailey, and Davidova (2017) in which subsidized fertilizers in Malawi increase the likelihood to sell by 4% points although no gender distinction is made. Access to NAIVS for seeds appears to have no impact on our outcomes. We interpret this direct impact to be a function of farming characteristics where women rely more on manual work and traditional farming systems. For male-headed households, the results indicate that access to fertilizer vouchers increases the probability of participating in the value chains, downstream, and sales by 8.6%, 7%, and 17.2% points, respectively. The effect of the seed voucher is positive and statistically significant for the likelihood of being a seller.

TABLE 4

Effect by NAIVS type

SellerInsideDownstreamUpstreamHarvested
quantity
SalesProfit
(1)(2)(3)(4)(5)(6)(7)
 Panel A. Female 
Fertilizer voucher 0.134* 0.186** 0.110 0.076 0.198 0.076 −0.435 
 (0.073) (0.085) (0.084) (0.068) (0.203) (0.137) (0.394) 
Seeds voucher 0.006 0.039 0.033 0.005 0.050 0.099 0.029 
 (0.127) (0.124) (0.103) (0.079) (0.335) (0.162) (0.162) 
        
Observations 1,131 1,131 1,131 1,131 1,061 1,061 102 
R-squared 0.605 0.568 0.531 0.479 0.778 0.362 0.676 
        
 Panel B. Male 
Fertilizer voucher 0.036 0.086* 0.070* 0.015 −0.019 0.172** 0.001 
 (0.045) (0.045) (0.041) (0.042) (0.086) (0.084) (0.097) 
Seeds voucher 0.112* 0.080 0.051 0.029 0.206 −0.045 −0.020 
 (0.066) (0.059) (0.053) (0.054) (0.146) (0.064) (0.146) 
        
Observations 3,905 3,905 3,905 3,905 3,694 3,694 517 
R-squared 0.604 0.577 0.523 0.475 0.756 0.447 0.582 
        
Controls Yes Yes Yes Yes Yes Yes Yes 
Household FE Yes Yes Yes Yes Yes Yes Yes 
District-Year FE Yes Yes Yes Yes Yes Yes Yes 
SellerInsideDownstreamUpstreamHarvested
quantity
SalesProfit
(1)(2)(3)(4)(5)(6)(7)
 Panel A. Female 
Fertilizer voucher 0.134* 0.186** 0.110 0.076 0.198 0.076 −0.435 
 (0.073) (0.085) (0.084) (0.068) (0.203) (0.137) (0.394) 
Seeds voucher 0.006 0.039 0.033 0.005 0.050 0.099 0.029 
 (0.127) (0.124) (0.103) (0.079) (0.335) (0.162) (0.162) 
        
Observations 1,131 1,131 1,131 1,131 1,061 1,061 102 
R-squared 0.605 0.568 0.531 0.479 0.778 0.362 0.676 
        
 Panel B. Male 
Fertilizer voucher 0.036 0.086* 0.070* 0.015 −0.019 0.172** 0.001 
 (0.045) (0.045) (0.041) (0.042) (0.086) (0.084) (0.097) 
Seeds voucher 0.112* 0.080 0.051 0.029 0.206 −0.045 −0.020 
 (0.066) (0.059) (0.053) (0.054) (0.146) (0.064) (0.146) 
        
Observations 3,905 3,905 3,905 3,905 3,694 3,694 517 
R-squared 0.604 0.577 0.523 0.475 0.756 0.447 0.582 
        
Controls Yes Yes Yes Yes Yes Yes Yes 
Household FE Yes Yes Yes Yes Yes Yes Yes 
District-Year FE Yes Yes Yes Yes Yes Yes Yes 

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Note: Standard errors are clustered at the household level

So far, our results suggest that NAIVS did not increase profit despite increasing market participation and participation in value chains. Figure 2 shows that poor households paid a higher cost per hectare to redeem the inputs than wealthier households. In particular, the costs were lower for commercial farmers than for less-market-oriented farmers. While NAIVS increased the supply of inputs, it did not create a mechanism to enable farmers to purchase inputs. To further investigate, we directly compare the impacts of the NAIVS to those of other sources of business finance. For instance, we find that 33.4% of female-headed agricultural households rely on SACCO against 30.9% of male. Over time, access to SACCO has largely increased for women, growing from 31.1% in 2008 to 38.9% in 2012, whereas access for male increased from 29.8% to only 33.8%. Given the nonrandomness of SACCO, we employ the same specification as in Equation (1), adding SACCO as an additional covariate. This dummy variable equals 1 for households with both NAIVS and SACCO. Since exposure to NAIVS may be correlated with access to SACCO, we control for time-varying variables, such as the total expenditure for improved inputs and network location. Results are displayed in Table 5. We find that the estimated NAIVS impacts are quantitatively similar and consistent with our previous results. More importantly, we find that the joint use of SACCO and NAIVS increased sales for male-headed households but reduced profit for female-headed households.

TABLE 5

Effects of NAIVS and SACCO

SellerInsideDownstreamUpstreamHarvested
quantity
SalesProfit
(1)(2)(3)(4)(5)(6)(7)
 Panel A. Female 
NAIVS 0.151** 0.226*** 0.177* 0.049 0.139 0.158 −0.217 
 (0.075) (0.086) (0.091) (0.062) (0.176) (0.156) (0.251) 
NAIVS and SACCO −0.082 −0.152 −0.163 0.012 0.186 −0.092 −1.127*** 
 (0.114) (0.138) (0.126) (0.081) (0.282) (0.118) (0.190) 
        
Observations 1,106 1,106 1,106 1,106 1,044 1,044 100 
R-squared 0.604 0.567 0.531 0.477 0.779 0.362 0.760 
        
 Panel B. Male 
NAIVS 0.010 0.058 0.063 −0.005 −0.044 0.058 0.039 
 (0.048) (0.045) (0.044) (0.043) (0.095) (0.066) (0.101) 
NAVIS and SACCO 0.105 0.057 0.010 0.046 0.130 0.314** −0.117 
 (0.079) (0.081) (0.068) (0.067) (0.140) (0.146) (0.139) 
        
Observations 3,773 3,773 3,773 3,773 3,575 3,575 507 
R-squared 0.600 0.573 0.519 0.474 0.756 0.447 0.582 
        
Controls Yes Yes Yes Yes Yes Yes Yes 
Household FE Yes Yes Yes Yes Yes Yes Yes 
District-Year FE Yes Yes Yes Yes Yes Yes Yes 
SellerInsideDownstreamUpstreamHarvested
quantity
SalesProfit
(1)(2)(3)(4)(5)(6)(7)
 Panel A. Female 
NAIVS 0.151** 0.226*** 0.177* 0.049 0.139 0.158 −0.217 
 (0.075) (0.086) (0.091) (0.062) (0.176) (0.156) (0.251) 
NAIVS and SACCO −0.082 −0.152 −0.163 0.012 0.186 −0.092 −1.127*** 
 (0.114) (0.138) (0.126) (0.081) (0.282) (0.118) (0.190) 
        
Observations 1,106 1,106 1,106 1,106 1,044 1,044 100 
R-squared 0.604 0.567 0.531 0.477 0.779 0.362 0.760 
        
 Panel B. Male 
NAIVS 0.010 0.058 0.063 −0.005 −0.044 0.058 0.039 
 (0.048) (0.045) (0.044) (0.043) (0.095) (0.066) (0.101) 
NAVIS and SACCO 0.105 0.057 0.010 0.046 0.130 0.314** −0.117 
 (0.079) (0.081) (0.068) (0.067) (0.140) (0.146) (0.139) 
        
Observations 3,773 3,773 3,773 3,773 3,575 3,575 507 
R-squared 0.600 0.573 0.519 0.474 0.756 0.447 0.582 
        
Controls Yes Yes Yes Yes Yes Yes Yes 
Household FE Yes Yes Yes Yes Yes Yes Yes 
District-Year FE Yes Yes Yes Yes Yes Yes Yes 

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Note: Standard errors are clustered at the household level

CONCLUSION

The ISPs in Tanzania is one of a few SSA cases that have attracted great scholarly and policymakers’ attention. The program design stimulated the efforts to reduce poverty in rural areas while improving productivity and ensuring food security. Using nationally representative LSMS-ISA panel data (2008–2012) from Tanzania, this study has examined the impact of NAIVS on market participation and farm-related enterprises with a gender view. Specifically, this article seeks to determine whether NAIVS receipts increase women’s participation in agricultural value chains, hence adopting and improving their entrepreneurial skills.

Overall, our empirical results suggest that NAIVS increases the likelihood of farmers participating in the agricultural markets. However, NAIVS had a larger effect on female-headed households than male-headed households; involvement in the NAIVS increased the probability of participating in the market by 16.6% and 8.1% points for female-headed households and male-headed households, respectively. Moreover, NAIVS significantly increases the female-headed households’ probability of selling in the market and participating in the downstream value chain. This suggests that the program has a positive impact on reducing the gender gap. Since women were given priority in the program’s implementation, we can consider this as a successful operation. In addition, the program increased female participation in the value chain and the effect is larger in the downstream value chain. On the other hand, the results are partial because NAIVS increases the likelihood of women participating in the market, but no impact can be found on sales or profit.

Based on our empirical results, further research is required to explore the underlying causes of the poor uptake of modern inputs by female farmers. We may think that it is because men are the decision-makers at the household level, but some existing studies find that women who are plot managers are also less likely to adopt modern maize varieties unless they get free seeds (Karamba and Winters 2015). The study concludes that participating in the NAIVS program increases the utilization of modern inputs by women leading to expanding agricultural entrepreneurship. Hence, there is a great need to implement subsidies as a tool to empower women in agriculture. As women are likely to be affected more by the subsidies, it is important to reduce bureaucratic and political hurdles to ensure fair supply promptly. Moreover, proportionate reduction of input costs to reduce the burden to poor farmers has the potential to stimulate the adoption of modern inputs.

APPENDIX

TABLE A1

Determinants of NAIVS participation

(1)(2)(3)(4)(5)(6)
NAIVSNAIVSNAIVS
(fertilizer)
NAIVS
(fertilizer)
NAIVS
(seed)
NAIVS
(seed)
Female 0.289 0.237 0.076 0.037 0.227 0.190 
 (0.180) (0.179) (0.080) (0.077) (0.188) (0.181) 
Land size in log 0.005 0.006 0.002 0.003 −0.003 −0.003 
 (0.008) (0.008) (0.007) (0.007) (0.005) (0.005) 
Expenditure in log −0.028*** −0.026** −0.016* −0.014 −0.018*** −0.017** 
 (0.011) (0.011) (0.010) (0.010) (0.007) (0.007) 
Agriculture is the main occupation 0.006 0.006 0.008 0.008 −0.005 −0.004 
 (0.018) (0.018) (0.016) (0.017) (0.013) (0.013) 
1 if purchased seeds 0.028** 0.029** 0.010 0.011 0.044*** 0.044*** 
 (0.012) (0.012) (0.011) (0.011) (0.010) (0.010) 
1 if purchased improved seeds 0.032 0.032 0.024 0.023 0.045*** 0.043*** 
 (0.022) (0.022) (0.020) (0.020) (0.016) (0.016) 
1 if purchased certified improved seeds 0.038 0.037 0.000 −0.001 0.047* 0.047* 
 (0.027) (0.027) (0.024) (0.024) (0.025) (0.025) 
1 if used pesticide fertilizer, 0 if not 0.009 0.010 0.006 0.008 0.037** 0.037** 
 (0.023) (0.023) (0.022) (0.021) (0.016) (0.016) 
1 if used inorganic fertilizer, 0 if not 0.330*** 0.328*** 0.352*** 0.349*** 0.025 0.025 
 (0.032) (0.032) (0.032) (0.032) (0.018) (0.018) 
1 if used organic fertilizer, 0 if not −0.021 −0.022 −0.016 −0.016 −0.009 −0.009 
 (0.020) (0.019) (0.019) (0.018) (0.014) (0.014) 
Constant 0.292** 0.273* 0.184 0.163 0.183* 0.173* 
 (0.145) (0.147) (0.128) (0.129) (0.100) (0.102) 
       
Observations 3,135 3,135 3,135 3,135 3,135 3,135 
R-squared 0.621 0.625 0.637 0.642 0.478 0.485 
Controls Yes Yes Yes Yes Yes Yes 
Household FE Yes Yes Yes Yes Yes Yes 
District-Year FE No Yes No Yes No Yes 
(1)(2)(3)(4)(5)(6)
NAIVSNAIVSNAIVS
(fertilizer)
NAIVS
(fertilizer)
NAIVS
(seed)
NAIVS
(seed)
Female 0.289 0.237 0.076 0.037 0.227 0.190 
 (0.180) (0.179) (0.080) (0.077) (0.188) (0.181) 
Land size in log 0.005 0.006 0.002 0.003 −0.003 −0.003 
 (0.008) (0.008) (0.007) (0.007) (0.005) (0.005) 
Expenditure in log −0.028*** −0.026** −0.016* −0.014 −0.018*** −0.017** 
 (0.011) (0.011) (0.010) (0.010) (0.007) (0.007) 
Agriculture is the main occupation 0.006 0.006 0.008 0.008 −0.005 −0.004 
 (0.018) (0.018) (0.016) (0.017) (0.013) (0.013) 
1 if purchased seeds 0.028** 0.029** 0.010 0.011 0.044*** 0.044*** 
 (0.012) (0.012) (0.011) (0.011) (0.010) (0.010) 
1 if purchased improved seeds 0.032 0.032 0.024 0.023 0.045*** 0.043*** 
 (0.022) (0.022) (0.020) (0.020) (0.016) (0.016) 
1 if purchased certified improved seeds 0.038 0.037 0.000 −0.001 0.047* 0.047* 
 (0.027) (0.027) (0.024) (0.024) (0.025) (0.025) 
1 if used pesticide fertilizer, 0 if not 0.009 0.010 0.006 0.008 0.037** 0.037** 
 (0.023) (0.023) (0.022) (0.021) (0.016) (0.016) 
1 if used inorganic fertilizer, 0 if not 0.330*** 0.328*** 0.352*** 0.349*** 0.025 0.025 
 (0.032) (0.032) (0.032) (0.032) (0.018) (0.018) 
1 if used organic fertilizer, 0 if not −0.021 −0.022 −0.016 −0.016 −0.009 −0.009 
 (0.020) (0.019) (0.019) (0.018) (0.014) (0.014) 
Constant 0.292** 0.273* 0.184 0.163 0.183* 0.173* 
 (0.145) (0.147) (0.128) (0.129) (0.100) (0.102) 
       
Observations 3,135 3,135 3,135 3,135 3,135 3,135 
R-squared 0.621 0.625 0.637 0.642 0.478 0.485 
Controls Yes Yes Yes Yes Yes Yes 
Household FE Yes Yes Yes Yes Yes Yes 
District-Year FE No Yes No Yes No Yes 

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Note: Standard errors are clustered at the household level

NOTES

1.

For simplicity of interpretation, we use OLS/linear probability models for all dependent variables in our empirical analysis. Using a probit model for the binary dependent variables produces results of the same sign and significance level.

2.

We replicate the main results when we use survey weights.

3.

See the methodology section for details.

4.

While larger and statistically significant, the impact on seller and participation for male do not seem the be statistically different from that of female.

REFERENCES

Ackah, C, and E, Aryeetey.
2012
. “Cash Cropping, Gender and Household Welfare: Evidence from Ghana.” In
Globalization, Trade and Poverty in Ghana
, edited by Charles Ackah and Ernest Aryeetey,
202
20
.
Sub-Saharan Publishers
,
Ghana
.
Ferguson, H., and T. Kepe.
2011
.
“Agricultural Cooperatives and Social Empowerment of Women: A Ugandan Case Study.”
Development in Practice
21
(
3
):
421
29
.
Fisher, M., and V. Kandiwa.
2014
.
“Can Agricultural Input Subsidies Reduce the Gender Gap in Modern Maize Adoption? Evidence from Malawi.”
Food Policy
45
:
101
11
.
Food and Agriculture Organization
.
2021
.
FAOSTAT: Cereal Production in Tanzania: 1960–2020
, https://www.fao.org/faostat/en/#country/215.
Jayne, T. S., N. M. Mason, W. J. Burke, and J. Ariga.
2018
.
“Review: Taking Stock of Africa’s Second-Generation Agricultural Input Subsidy Programs.”
Food Policy
75
:
1
14
.
Johnson, M., P. Hazell, and A. Gulati.
2003
.
“The Role of Intermediate Factor Markets in Asia’s Green Revolution: Lessons for Africa?”
American Journal of Agricultural Economics
85
(
5
):
1211
16
.
Karamba, R. W., and P. C. Winters.
2015
.
“Gender and Agricultural Productivity: Implications of the Farm Input Subsidy Program in Malawi.”
Agricultural Economics
46
(
3
):
357
74
.
King, S., H. Sintes, and M. Alemu.
2012
.
“Beyond Participation: Making Enterprise Development Really Work for Women.”
Gender & Development
20
(
1
):
129
44
.
Osei, C. D., and J. Zhuang.
2020
.
“Rural Poverty Alleviation Strategies and Social Capital Link: The Mediation Role of Women Entrepreneurship and Social Innovation.”
Sage Open
10
(
2
):
1
11
.
Palacios-Lopez, A., L. Christiaensen, and T. Kilic.
2017
.
“How Much of the Labor in African Agriculture is Provided by Women?”
Food Policy
67
:
52
63
.
Pan, L., and L. Christiaensen.
2012
.
“Who is Vouching for the Input Voucher? Decentralized Targeting and Elite Capture in Tanzania.”
World Development
40
(
8
):
1619
1633
.
Roy, P., S. Haque, A. Jannat, M. Ali, and M. Khan.
2017
.
“Contribution of Women to Household Income and Decision Making in Some Selected Areas of Mymensingh in Bangladesh.”
Progressive Agriculture
28
(
2
):
120
29
.
Sheahan, M., and C. B. Barrett.
2017
.
“Ten Striking Facts About Agricultural Input Use in Sub-Saharan Africa.”
Food Policy
67
:
12
25
.
Sibande, L., A. Bailey, and S. Davidova.
2017
.
“The Impact of Farm Input Subsidies on Maize Marketing in Malawi.”
Food Policy
69
:
190
206
.
Slavchevska, V.
2015
.
“Gender Differences in Agricultural Productivity: The Case of Tanzania.”
Agricultural Economics
46
(
3
):
335
55
.
Venkataraman, S.
2019
.
The Distinctive Domain of Entrepreneurship Research
.
Emerald Publishing
.
Woldu, T., F. Tadesse, and M.-K. Waller.
2015
. “Women’s Participation in Agricultural Cooperatives in Ethiopia.”
Paper presented at the International Conference of Agricultural Economists
,
Milan, Italy
,
Aug.
9
14
,
2015
.
Wooldridge, J. M.
2021
. “Two-Way Fixed Effects, the Two-Way Mundlak Regression, and Difference-in-Differences Estimators.” Working Paper,
Michigan State University
.
World Bank
.
2014
.
Tanzania Public Expenditure Review: National Agricultural Input Voucher Scheme (NAIVS)
.
Washington, DC
:
The World Bank
.
________.
2020
.
World Development Indicator
.
Washington, DC
:
The World Bank
.