Syndicate content

Add new comment

Five things we learnt from a loan and grain storage intervention in Tanzania. Guest post by Hira Channa

Development Impact Guest Blogger's picture
Two of the main post-harvest concerns that many small holders in Sub Saharan Africa (SSA) face are:
  1. Maintaining the quantity and quality of staple grains throughout the year. 
  2. Managing the persistent price seasonality in grain commodity markets. In Mbeya, Tanzania, where our experiment was conducted, for the last two years maize prices in the lean season were 80% higher than prices at harvest.
The important question for researchers and practitioners is then to determine which intervention will enable farmers to manage these challenges. This matters for overall welfare because it has the potential to increase mean incomes and reduce consumption seasonality which is known to be particularly detrimental for children (Christian and Dillion, 2018).

The role of credit constraints (Burke et al., 2018) and storage constraints (Omotilewa et al., 2018; Aggarwal et al., 2018) have been studied separately. Basu and Wong (2015) combined a storage and credit intervention, but their credit product, which is not given at harvest, is not designed to measure how liquidity constraints drive output storage and sale decisions.

As such, the main contribution of this paper is that it is the first to use a randomized control trial to empirically estimate the impact of both physical storage constraints, and credit constraints for smallholders at harvest. We focus on maize which is the main crop in much of the Mbeya region and shows the sharpest price seasonality.

We have two main outcomes of interest that we hypothesize would be affected by the intervention. Maize inventory in the January following the intervention, which is six months following the harvest and when maize prices tend to peak. The other is net maize sales throughout the year, we estimate the impact on net maize sales (maize sales-(maize purchases +interest rate for the credit group)) aggregated for the entire year, and disaggregated by quarters across the intervention year We would expect that the intervention that addresses the more pressing constraint would create a larger increase in maize inventory in January and increase the value of sales in the lean season.

Our tools and the intervention
Purdue Improved Crop Storage (PICS) bag
The PICS bag is a hermetic (airtight) storage bag that creates an airtight seal and has been shown in multiple experimental studies to reduce physical losses and to slow down fungus growth without using storage pesticides.
This tool is designed to address the storage constraint that maybe preventing farmers from storing grain into the lean season. As part of this intervention some of the farmers received two PICS bags.

Loan
The loan product was designed in collaboration with our lending partners in Tanzania, Pheretajo. They are a local NGO responsible for the registration and training of the credit clubs, called Village Savings Cooperative Bank (VICOBA) in the Mbeya region of Tanzania. The term VICOBA (kikundi in Swahili) refers to a group of individuals (15-30) who come together so that they can access credit (for most formal credit is inaccessible) and save and invest money. The group meets every week (or every other week), and each member buys “shares” in the VICOBA. This is a form of saving for the group members, who then lend this money out to other group members.

To address the liquidity crunch that many farmers face at harvest, the loan product was distributed at harvest and was due back in the lean season when prices were expected to be higher. Farmers were expected to store grain at harvest, sell the grain at the higher price in the lean season and payback the loan with the 12 % interest, and keep the remaining profit. The 12% interest rate is higher than the 10% internal lending rate of the credit group, but is much lower than the 20-25% interest rate which would be the cheapest outside option for farmers (a group loan from CRDB Bank, however only 2% of our sample indicated that they had access to any formal sources of credit).

A requirement of our partners, who were lending the loan was that the loan would be collateralized with maize stored in the PICS bag and be stored in groups.
For this intervention the loan product was worth USD 40, which is approximately equivalent to two bags of maize valued at harvest time in June 2017. This amount is significant, considering that half of our sample borrowed less than USD 11 from the credit group which was their other main source of credit.

Intervention
Our experiment covered seven districts in the Mbeya region of the southern highlands of Tanzania, with a sample size of 1583 respondent. The randomization occurred at the level of individual credit clubs (the VICOBAS), where 131 clubs participated in total.

In each treated group, 10 individuals were randomly selected to be treated, and another 10 or fewer (depending on turnout) were trained on the use of the bag and surveyed but did not receive the bags or the loans. While the randomization for the groups was done beforehand, at the individual level it was in the form of a public lottery.
  • 44 clubs were randomly allocated to the storage intervention (involved training and receiving 2 hermetic storage bags).
    • Treated (received training and storage bags) -424 individuals
    • Training only-173 individuals
  • 43 clubs were randomly allocated to the credit intervention (offered access to the loan product described above).
    • Treated (loan product, storage bags and training) -402 individuals
    • Training only-65 individuals
  • 44 were randomly allocated to the control group.
    • 420 individuals
What did we learn?
  1. The loan intervention was popular, 80% of those who were offered the loan accepted the offer. This is higher than the 60-65% that Burke et al. (2018) find for a comparable product in Kenya, which as they point out is already much higher than adoption of micro-credit products, found in the previous literature that generally ranges from 2-55%.
  2. We find evidence that combining the loan product with PICS bags increases both maize inventory and net maize sales later in the year (increase by 20% and 11% respectively at the mean). These farmers decreased net maize sales earlier in the year and increased sales 6 months after the harvest, primarily by adjusting quantities stored and sold. While these estimates are noisy (significant at the 10% level) they suggest a return of 40% on the loan amount.
  3. Macro-level factors can affect interventions and create unexpected outcomes. In our case this came in the form of an export ban on maize by the Tanzanian government, which considerably reduced the maize price rise in the lean season. This very likely created an attenuation effect on our outcomes, because the gain from holding on to the grain was now much lower.
  4. For our partners, who lent the money, the main take-away was that the PICS storage bags can help secure collateral for loans. This lower risk is passed on as lower interest rate for the farmers. Even though the lack of price rise meant that repayment was delayed, and lower than expected at 85%, our partners scaled-up the loan to 200 groups in the year following the intervention.
  5. Our study also provides some insight to the emerging literature on how the impact of microcredit interventions are heterogeneous with regards to individual characteristics. For example, as we might expect intuitively, those who borrowed less from the credit group (at baseline, possibly indicating credit constraints), sold more maize later in the year (had a higher treatment effect).
Hira Channa is a PhD candidate in Agricultural Economics at Purdue University.