Table of contents

Table of contents

?Introduction 1
Literature review 2
2.1 Angelucci et al. (2015) 2
2.2 Field et al. (2012) 4
2.3 Field et al. (2013) 6
2.4 Beaman et al. (2015) 8
2.5 Attanasio et al. (2015) 10
Research Design 11
3.1 The microcredit provider’s background 12
3.2 Experimental design 12
3.3 Data collection 13
3.4 Final outcomes 14
3.5 Estimation method 15
3.6 External validity 15
Conclusion 16?

Introduction
According to Mohammed Yunus, the founder of the first microfinance institution – the Grameen Bank, “the poor are bankable only if the right lending mechanism is used (Yunus, 2007).” Despite being hailed by the proponents of the microfinance concept, microfinance as a poverty alleviation policy has been subjected to intense public scrutiny and criticism. Many experimental studies have been conducted by researchers worldwide in order to evaluate the impact of microfinance.
Since microfinance is a wide concept that includes not only lending but also saving and insurance services, this paper would like to narrow down the scope on microcredit which is an important aspect of financial services. Low take-up rates, problems with measurement and heterogeneity blunt transformative effects of microcredit. Nevertheless, some believe that better loan contract designs can produce greater impacts on the poor’s living standard (Karlan and Morduch, 2010; de Janvry and Sadoulet, 2016). This leads us to the research question: “Will better loan contract designs have greater impact on the poor?”
In this paper, we have further examined five major experimental studies with different contract designs. These major studies have helped to provide an insight into the variance of microcredit effectiveness in different contexts. They are the works by Angelucci et al. (2015), Field et al. (2012), Field et al. (2013), Beaman et al. (2015), and Attanasio et al. (2015).
Our paper is divided into three parts. The first part is a literature review of the five works mentioned above. Each review ends with a possible gap which requires further research. Part two of our paper is our proposed research design which highlights the experimental design, data to be collected and the description of loan products. The proposed research to fill in the gap of interest will be implemented in the northern region of Ghana. The third part covers the conclusion.

Literature review
2.1 Angelucci et al. (2015)
The first literature we review is a study by Angelucci et al. (2015) which represents a typical microlending contract with features such as group liability, weekly fixed payment, and female entrepreneur target. The study sought to estimate the effect of micro loans on the poor in Mexico.
The authors cooperated with the largest lending institution in Mexico-Compartamos, to roll out new credit services from 2009 to 2012 in North-central Sonora, Mexico where banks had not yet penetrated. Randomization was conducted in a geographic cluster or municipality unit. Specifically, 125 clusters were randomly assigned to receive loan products, while the other 125 clusters served as a comparison group. Targeted clients were mostly business or intent-to-be-business women from the ages of 18 to 60, accounting for 51% of studied low-income borrowers.
Initial loan amounts were approximately 125-500 USD which had to be repaid weekly with an equal amount in the group meeting over the course of 16 weeks. Interest rates were at 3.89 percent per month. Loan takers with good repayment records had lower interest rates as an incentive at 2.9 percent per month and loan amount increased up to 2,250 USD.
Despite being the first lending institution opening in the area, the compliance rate of treated clusters was moderately low at 17.3 percent as compared to 26.7 percent and 13 percent in the studies of Banerjee et al. (2015a) and Crépon et al. (2015) respectively. However, the default rate was as low as 1 percent. Due to low take-up rate, the authors estimated the intention-to-treat (ITT) effect in 37 multiple outcomes formed as 6 main brackets: microentrepreneurship, income, labor supply, expenditures (consumption) and subjective well-being.
There was a plausible increase in business growths with revenues rising by 27 percent and expenditures rising by 36 percent. Regarding income, results were not statistically significant, meaning that microcredit did not help to substantially improve earnings by business, labor and remittance. As for consumption, access to microcredit led to a decrease of 9 percent in the assets bought and 6 percent in temptation goods. Another favorable result found was a slight increase (0.8 percentage points) in women’s contribution in family matters. More importantly, results found in this study are considered to be akin to other microcredit experiments (Gíne and Karlan, 2014; Banerjee et al., 2015a; Crépon et al., 2015) in terms of small but not transformative effectiveness of microfinance on poverty reduction.
Regarding the internal validity, the high attrition rate in the sample of approximately 37 percent, together with the use of endline survey instead of baseline survey pose a threat to the unbalance between treatment and control groups at the baseline. The low take-up rate at 17.3 percent has also made it difficult to assess the program impact.
The interest rates in this study had two inferences. On the one hand, its incentive reduction increased the probability of Compartamos’ loan taking by 11.5 percentage points, while Compartamos did not lose its profits. On the other hand, the annual percentage rate was extremely high at 110 percent. Despite being normal in Mexico market, it is uncertain whether this study can be replicated in other developing countries’ markets where interest rates were far lower ranging from 12 percent to 27 percent (Banerjee et al., 2015b, p. 4). Both have a further implication for policy makers as well as financial providers to take into account in designing microcredit products in the future.
Moreover, studies by Angelucci et al. (2015), Banerjee et al. (2015b), and Woodruff et al. (2007) have shown the heterogeneous treatment effect of microcredit, that is its benefits for some loan recipients than others. Therefore, an evident gap from this literature reviewed is the inability to examine the degree of gain or loss associated with the changes in outcomes on the beneficiaries due to the introduction of loan products in North-central Sonora, Mexico. It is necessary for the future studies to identify the factors such as the previous existence of financial services, level of education, business skills and experience, business cycles, or level of education which lead to different impacts of microcredit programs on borrowers who have signed onto the same program and are being treated under similar conditions.
2.2 Field et al. (2012)
Field et al. (2012) conducted an experiment on how contracts with flexible terms of loan repayment helps to boost business outcomes of assigned borrowers, and in turn lead to fiscal stress reduction. Traditional microfinancing by design, requires that recipients start repayment merely a week after the loan has been disbursed. Credit applicants then pay in instalments week after week until the full amount is settled. This arrangement makes it difficult for clients to deal with shocks in the short term.
Furthermore, the authors opined that microfinance projects which required borrowers to repay in weekly instalments had tendencies of having negative psychological effects on borrowers who might be anxious about impending payments. Thus, the conventional weekly repayments have tendencies of escalating the stress levels of borrowers thereby driving some to even commit suicide. The authors further argued that this could be a likely case since most poor borrowers lacked the financial literacy to proficiently manage their inflows and outflows.
To examine this further, Field et al. (2012) ran an RCT to put in experimental perspective, the effect of repayment frequency on borrowers’ economic status and psyche. Hulme (2007) agrees with Field et al. (2012) that most borrowers tend to be very frightened at the possibility of owing these MFIs due to the treatment that one is subjected to in the event of any default.
Borrowers who got randomised into the treatment group were offered a more flexible contract. This entailed a five-weekly (monthly) repayment. In contrast, the comparison group had the conventional weekly frequency. The study was done in conjunction with a microfinance institution known as Village Financial Society (VFS) in Kolkata, India. A sample size of 213 clients out of a population of 740 clients was used for the study. In the period January to September 2008, VFS formed 148 groups of 5 members each. A total of 112 potential borrowers were randomised into the treatment group, and the other 111 borrowers ended up in the comparison group. Borrowers in the treatment group were on the average required to pay 27.10 USD every month while their comparison counterparts were due to pay 5.40 USD every week. The loan contract had a full term of 45 weeks.
The study was based on four primary hypotheses. The first hypothesis was on income. That a more flexible repayment method would afford clients the opportunity to plough back profits over the flexible period, into more profitable assets for their business and increase their income gains. The second hypothesis was that flexible repayment might encourage fiscal indiscipline and defaults. Thus borrowers will be tempted to spend on personal goods which are not crucial to their business. In addition, they advanced on the hypothesis that a less frequent repayment model might lessen the cost of consumption smoothing. This in effect could enable borrowers to deal with unexpected family or business expenditures. The last hypothesis was that monthly repayment frequency could reduce the time burden on borrowers and ultimately, the anxiety level.
Along various dimensions, the contract with the flexible terms of repayment was seen to reduce borrowers’ stress levels along various dimensions and increase their income since they could conveniently invest in more lucrative ventures. Stress levels for the treatment group decreased significantly by 9.3 percentage points. In addition, there was a significant increase in business income by 561.33 Rupees and overall income by 614.9 Rupees as compared to those of the comparison group. Similarly, investment in business assets rose significantly by 523.7 Rupees. Interestingly, no increases in the short run default nor spending on temptation goods could be found.
On the basis of the findings made in this study, there is the impression that the debt contract design of a microfinance helps the poor to smoothen consumption in the short run. Yet, it is difficult to clearly estimate this in the literature under the review. Therefore, there is the need for further experiments to study the impact of more flexible microcredit contract designs on consumption smoothing as well as consumption behaviours.

2.3 Field et al. (2013)
Another study by Field et al. (2013) also experimented with other types of contract designs. Those in the treatment group, however, were granted a 2-month grace period, while those in the comparison group were granted 2-week grace period. Besides, low-income borrowers had to fortnightly repay their loans. Differences in results between them would help to shed light on whether debt contract designs based on repayments would solve illiquidity of microentrepreneurs and then achieve greater impacts of microfinance.
Researchers questioned whether a relief from the initial repayment obligation after loan disbursement would encourage microentrepreneurs to invest in riskier yet more profitable assets which bring about higher returns. They also enquired whether lenders would face higher risks due to the grace period. As such, they conducted an experiment in West Bengal, India introducing “grace period” in cooperation with VFS between March and December 2007. The study concentrated on women entrepreneurs, individual liability, and fixed interest rates, in spite of the difference in grace periods between treatment and control groups.
With reference to study design, 845 potential female borrowers were formed into 169 groups with 5 members each. Half of the 169 groups (the treatment group) were randomly offered individual-liability terms and a two-month grace period, whereas the other 84 groups (the control group) were assigned individual-liability contract with regular repayment. Both groups after the first repayment would have to repay every two weeks within a duration of 12 months. Loan recipients received various loan sizes, ranging from around 90 USD to 225 USD.
Postponing the initial repayment until two months in the treatment group had a crucial implication since it allowed them to use the full loan amount in the expansion of their enterprises. In the short run, researchers found that 91 percent of treated borrowers invested in their business and on average 83 percent of the loan amount was used for business expenses. They invested particularly in less-liquid assets more than their regular grace counterparts. Furthermore, new start-up businesses in the treatment group were three times as likely as in the the control group. This suggests that without longer grace periods introduced, borrowers might have to reserve a part of the loan amount for the purpose of meeting the first repayments instead of investing in less-liquid assets for business growths.
At the end of the study, the authors reported that the two-month grace period group increased weekly returns on business as well as family earnings, by 41 percent and 19.5 percent respectively, more than outcomes of those with conventional repayment significantly. Also, different rates observed were profoundly different in comparison with other studies such as approximately 1% in Angelucci et al. (2015) and Beaman et al. (2015). Specifically, the authors found an increase of 6 – 9 percent in default rates of the two-month-grace-period group. This implies the likelihood of shortage in the treatment group’s incomes during loan period due to opting riskier investments in the short run. Besides, three years after the intervention had taken place, is not long enough to entirely affirm that the grace period design induced low business inflows making clients cash trapped to meet repayment obligations. The case could be different in the long run when investments start paying off.
Heterogeneous treatment effect in which some may gain but some lose is shown in this literature. Different results emerged from various borrowers’ qualities in the short run. Those who are low-risk takers, patient, and possess high business acumen enjoyed higher benefits from the 2-month-grace-period assignment. In contrast, grace period contracts did not favor high-risk takers as much as low-risk takers because there was a significant decrease in monthly profits (-1,557.9 Rupees) of risk-loving borrowers in the interaction term between the characteristics of the borrowers and grace period.
It is evident from the study that default rates increased in the treatment group relative to the comparison group. However, it will be expedient to conduct further research into how the grace period could be exploited to suit both high risk and low risk borrowers. One area that requires further empirical evidence is the repayment with arrears which was observed with the extended grace period. What makes borrowers default on repayment; a moral hazard such as fiscal indiscipline or investment in riskier but profitable assets?
The contracts we have considered so far are all of a general nature and do not pay close attention to the peculiar business needs of the individual borrowers. What will be the impact of a contract design which is sensitive to the business cycle of various classes of borrowers? For instance, in the agricultural sector, due to seasonal dynamics financing needs to differ from one enterprise to the other. What would be the impact of a microfinance program which is particularly tailored to meet the needs of borrowers depending on their industries?
2.4 Beaman et al. (2015)
The above inquiries lead us to the fourth literature which was written by Beaman et al. (2015). This study examined if financial constraints hinder Malian farmers from agricultural investments, and the tendency of more productive farmers to self-select into a microcredit program. When it comes to agriculture, most farmers’ incomes are subject to fluctuations. That is, farmers mainly obtain their incomes after the harvest season, once or twice each year. Fluctuations in cash flows make rigid repayments challenging for farmer borrowers.
In order to help Malian female farmers in Mali ease financial constraints, the authors collaborated with a microfinance institution in Mali called Soro Yiriwaso (SY) to provide the farmers with microcredit. Especially, during the time of the study, SY was one of the few financial institutions having products that were particularly tailored to meet financial needs of the farmers with repayment based on their cash flows.
Regarding the credit offering experiment, loans were given out by SY during the cultivation seasons of 2010 and 2011 under group liability. The interest rate charged was 25 percent with loan size being equal to 113 USD on average. 88 among 198 eligible villages were randomly assigned to the treatment group in which they were offered microcredit products at the start of their cultivation period and were obliged to repay full amount. However, untreated villages (the comparison group) were not offered any loans.
Over the course of the study, defaults in repayment observed was 1%. The loan take-up rate was at a low of 21%, so ITT estimate was used to measure average impact of microcredit. Study results revealed that borrowers used loans to invest in fertilizer (an increase of 10.35 USD) and pesticide (an increase of 5.08 US), and that in turn led to an increase in the magnitude of harvest by 32 USD, but not in profits. However, the authors highlighted the fact that the prospect of less capital constraints encourages some farmers to self-select into the lending program. Loan takers may have more advantages and gain market power against those who did not take up loans. This could be a factor of the heterogeneous treatment effect of microfinance which may be “good for some but bad for others” (Banerjee et al., 2015b, p. 14).
In spite of not being offered loans, sub-group of farmers in the comparison group, as well as non-compliers in the treatment group received cash transfers supported by the Innovations for Poverty Action organization. This is a particularly interesting intervention since, in addition to loans offered, those who were not provided loans also had an equal chance to reduce capital constraints by obtaining a one-time cash transfer amount of 40,000 FCFA (140 USD). As a result of the cash incentives, those who received in the comparison group increased expenses for agricultural investments (by 14 percent) leading to more outputs (13 percent) along with profits (12 percent). This indicates that what hinders investments in agriculture is attributable to low liquidity. Thus, it is important to ensure that microcredit contracts do not leave borrowers with liquidity problems. When borrowers are illiquid, they are unable to make the kind of investments which will facilitate the transformative impact of the credit-offering programs.
The flexible term used in this study not only led to a relief of liquidity constraints on farmers but also contributed to the increased investment and increased revenues in farming. This suggests that microcredit with its promise is likely to help alleviate poverty through relieving financial constraints that impede business growth and cultivation expansion in particular (Banerjee et al., 2015b). The likely capital constraint due to the lump-sum-payment obligation used in this experiment might have discouraged loan take up rates. Following from this, an identified gap which needs to be filled with further research is how a lump repayment after harvest deters farmers from signing on to a microcredit program.
2.5 Attanasio et al. (2015)
A feature of the microfinance design is where the liability for the loan lies, either group liability or individual liability (Giné and Karlan, 2014). Until now, we have reviewed studies which experimented with one kind of liability exclusively. In tackling the challenges or drawbacks of microfinance, (thus microfinance not yielding the expected results for which the designers of the scheme and service providers tout it to be) most MFIs now resorted to group-liability lending as a means of In the agricultural sector, due to seasonal dynamics financing needs to differ from one enterprise to the other to curb the incidence of default (J-PAL, 2015). Attanasio et al. (2015) conducted an RCT in Mongolia to investigate the differences in impact of individual and group liability contracts on underprivileged females.
The study involved randomising 1,148 perceived poor women in 40 villages into three groups. 15 villages out of 40 got access to individual-liability loans, and another 15 were given group-liability loans. Both served as the treatment groups. Participants in the last 10 villages, served as the control group, but received no loans during the period of the study. The highest loan size given was about 435 USD with annual interest rate of 27 percent and subsequent reduction by 0.1 percent per month upon good repayment records. In this experiment, monthly payments were required with the maturity between three and 12 months. Eligible borrowers were granted loans by the microfinance institution, XacBank.
The researchers did an ITT analysis to ascertain the effect the intervention had on the beneficiaries. The probability of taking up a loan was higher in group lending areas than in individual lending areas, 57 percent versus 50 percent respectively. Overall, business ownership increased significantly for group-liability borrowers with the plausibility of running new businesses being 9 percentage points higher than non-loan group. It was observed that female entrepreneurship increased considerably, led by less educated women in group treatment areas. Results revealed such women being 31 percent more likely to run their own enterprises.
This changes led to an increase in the working hours of treated female borrowers by 35 percent. Business assets index in the group-liability treatment areas also rose by 13.7 units, but there were no significant impacts found in income. Additionally, the researchers observed an increase in human well being measured in terms of consumption. Monthly budget for consumption was 18.46 USD higher for low-income borrowers in group-lending areas as compared to the control group. Individual-liability borrowers, on the other hand, showed minor change in entrepreneurship and consumption despite the increase in credit.
In terms of loan repayment, no empirical evidence was found to confirm that group-liability borrowers were less likely to default than individual-liability borrowers. Thus, no difference in default likelihood was found to suggest that the nature of loan liability influenced default rates. This result can be corroborated by Giné and Karlan (2014). What runs through the results of the study, however was the strong results measured in favor of group liability as against individual liability.
In this experiment, outcomes suggest a causal relationship between group liability and female-business-ownership expansion. However, this needs to be examined further in the long run to confirm the existence of such causality since this study only focused on the short-term results (18 months after the intervention).
Research Design
Our prospective research will be focused on examining critically the impact of flexible agricultural-specific microcredit products on female farmers. The experiment would enable us to establish whether loan products which are sensitive to the agricultural seasons will help farmers to alleviate their liquidity constraints. And that, they can make profitable agricultural investments in the short run, which in turn lead to increased impacts on business and welfare outcomes in the long run.
We would partner with The Abdul Latif Jameel Poverty Action Lab (J-PAL) and the Hopeline Institute, a non-profit-organization in Ghana, to run an RCT which offers micro loans to remote villages in the Northern Region of Ghana. It is the largest savannah area in Ghana and has extreme temperatures in both the dry and rainy seasons. Inhabitants of the proposed regions are engaged predominantly in farming but only on subsistence basis. Thus, they have no extra sources of income to expend on health, education and other basic human needs. This unfortunately leaves inhabitants cash trapped and places them in the category of the impoverished.
3.1 The microcredit provider’s background
The Hopeline Institute, founded in 2007, is well known for not only providing business training skills but also offering microloans to the Ghanaian microentrepreneurs. Since its establishment, Hopeline has successfully supported a large number of start-up businesses, particularly its contribution to shift microenterprises into small-and-medium-sized ones.
3.2 Experimental design
Out of the 26 districts in the northern region of Ghana, we intend to carry out an experiment in Bunkpurugu, Savelugu, Saboba Chereponi, Bunkpurugu- Yunyoo, Sawla-Tuna-Kalba. These are areas with lack of access to official credit products. Since villages in the selected districts are far apart, it is a means to ensure internal validity in that spill over or externalities would not happen during the period of the intervention.
The main aim for our intended loans is to enable the women finance their farming activities and expand beyond subsistence level. Our population of interest is 200 underprivileged villages drawn from the aforementioned five districts. Within these villages, we will randomly assign 100 villages into the treatment group, and the other 100 villages would serve as the comparison group. Both groups would be offered loans under group liability but in different terms of repayment frequency.
Eligible clients must be women ranging from the ages of 18 to 60. We are interested only in female farmers, or females who are planning to go into farming. For group liability, each group would consist of five members. Upon application for loans, they would be required to provide proof of residence and oral business plans.
We project the intervention to span from 2019 to 2020 with two loan cycles. The first loan cycle will be from January to December 2019, and the second will be from January to December 2020. Loan size given to both groups of clients will be 250 USD at an annual interest rate of 10 percent. This rate is below the average market interest rate in Ghana (approximately 15 percent). However, since our program is focused on poverty alleviation, it is reasonable. Besides, the interest rate will be slashed down to 8 percent when clients pay back their loans dutifully. This is to incentivise high patronage of the program and possibly discourage payment with arrears.
The microcredit program will grant borrowers in the treatment group a grace period which covers the period of the farming season before harvesting, expected to be between January and June. For an intervention which is to purposively ensure that these borrowers commercialise their farms, they will be entitled to a grace period of six months to cushion them against seasonal changes. This will help to minimize capital constraints among poor farmers in the short run; thereby, relieving them from financial stress so that they can concentrate on farm work and investments. This grace period is long enough to cover the cultivation period until harvesting between January and June. Treated clients will be expected to repay 50 percent of the loan amount only after harvesting. Thereafter, the rest will be spread over six months, from July until December, in equal instalment to group treasurers who will be accountable to the loan officers for monies collected on behalf of the group. Conversely, borrowers in the comparison group will be expected to repay their loans by instalment every month.
3.3 Data collection
Before the intervention, we are going to organize the baseline-data collection from June 2018 to December 2018. We would exploit the baseline data to make sure treatment-control balance in a wide range of household and socioeconomic characteristics. The next follow-up survey would be conducted 12 months after loan disbursements, and it will be used to estimate the causal effect of the program in the short run. Because one of our main purposes in the study is to measure the impact in the long term, we would then interview study borrowers in the endline survey five years after the microcredit program.
At baseline, we will collate data on household characteristics and demographics (age; marital status; educational level; number of children; children’s educational attainments; household expenses for food; family assets; education and health; number of dependants; proof of residence; history of credit and livestock ownership). We are also interested in socio economic factors (whether they are engaged in farming; the produce they cultivate; the number of labor; sales and expenses; whether they have any intention of going into farming should they be granted loans; any land title to their name; land size; nature of crops; years of experience in farming; source of capital (if any); availability of storage facilities; their knowledge about the current market; the existence of a ready market for their produce; expected returns after every planting season and community participations). All these will be collected and well documented to aid in endline analysis.
As part of the experiment, every borrower will be issued with a mobile phone to enable surveyors contact them annually to ascertain whether there have not been any incidence of relocation. This will enable us overcome any likely effect of attrition on our final results. Should there be attrition which can pose a threat to the study’s internal validity, through the mobile phone, they can still be contacted for interviewing. The use of mobile phones for tracking and monitoring in our experiment is inspired by Field et al. (2012).
3.4 Final outcomes
There are five outcome categories we would like to measure in the short run (12 months after each loan cycle) and the long run (five years after the loan program). We will measure for micro entrepreneurship with proxies being agricultural investments. This will give an indication of whether they were liquid enough to spend on crucials for their farms. Secondly, information on investment returns will be required from respondents. Such data will give an idea about any expansion beyond subsistence level. In addition, we will consider data on household income and labour supply. Further, general data on family expenditure will be gathered. The responses for these outcomes will help estimate whether the intervention led to consumption of essential goods, temptation goods or the patronage of durables. Finally, we will collect data from our respondents on women empowerment using female community participations as a proxy. In addition, we will estimate effects on basic well being such as expenses for health or education.
3.5 Estimation method
We expect to analyse ITT effect since despite our efforts to encourage everyone to register for the program, compliance rates for microfinance programs are often below 100 percent. This estimate allows us to examine the effectiveness of the intervention on average, regardless of whether clients actually comply or not. Our basic regression equation is:
Yivt = 0 +1 . Tv+ X . Xi0 + 1i (1)
Y is the outcome variable of interest for individual i in village v at time t (t = 0
(1, 2) at baseline survey (follow-up surveys));
Tvis a dummy variable equal to 1 for those who were assigned to the treatment group (0 for the comparison group);
Xi0 represents a set of baseline characteristics of participants;
1i is the error term clustered in village (at the randomization) level.
3.6 External validity
Having put in measures to ensure the internal validity of our design, there is also the need to ensure its external validity or its potential of being generalized in other contexts. The grace period per this design is sensitive to the gestation period of the produce cultivated by our targeted borrowers in northern Ghana. For any potential replication to be successful, there is the need for the designers to pay critical attention to the seasonal dynamics and design a contract with a befitting grace period.

Conclusion
Microfinance as a poverty alleviation tool has succeeded to deliver on its mandate in some areas. There are however some grey areas which need to be addressed with studies and policy changes. Putting asides saving and insurance products, we pay more close attention at microcredit services which are a crucial part of microfinance. From previous studies reviewed, different contract designs which have flexible repayment terms in the longer term have shown some promising results.
For this reason, we would like to experiment further with a contract design which is more focused on borrower characteristics, business conditions and its impact in the long run. The design outlined features a grace period to shield farmers from having to make payment when they have not even harvested their produce. By doing so, capital constraints and any tendencies of repayment causing financial stress will be minimised. We believe that both short run and long run evaluations need to be conducted after treatment. The timing of impact evaluations after an intervention is considered in our design. Doing this will help avoid any premature judgement of an intervention as having achieved desired results or being a failure.
Summarily, the interaction between microcredit contract designs, seasonal dynamics of agriculture and timing of assessments might be of important implications in the transformative outcomes that can be observed with farmers. We believe this matter should drive further research by academia, MFIs and policy makers on the concept of microfinance in general and microcredit in particular.

References
Abdul Latif Jameel Poverty Action Lab (J-PAL). “Where Credit Is Due.” Policy Bulletin, February 2015, 1–16. https://www.povertyactionlab.org/.
Angelucci, Manuela, Dean Karlan, and Jonathan Zinman. 2015. “Microcredit Impacts: Evidence from a Randomized Microcredit Program Placement Experiment by Compartamos Banco.” American Economic Journal: Applied Economics, 7(1): 151–182.
Attanasio Orazio, Britta Augsburg, Ralph De Haas, Emla Fitzsimons, and Heike Harmgart. 2015. “The Impacts of Microfinance: Evidence from Joint-Liability Lending in Mongolia.” American Economic Journal: Applied Economics, 7(1): 90–122
Banerjee, Abhijit, Rachel Glennerster, and Cynthia Kinnan. 2015a. “The Miracle of Microfinance? Evidence from a Randomized Evaluation.” American Economic Journal: Applied Economics, 7(1): 22–53.
Banerjee, Abhijit, Dean Karlan, and Jonathan Zinmann. 2015b. “Six Randomized Evaluations of Microcredit: Introduction and Further Steps.” American Economics Journal: Applied Economics, 7(1): 1-21.
Beaman, Lori, Dean Karlan, Bram Thuysbaert, and Christopher Udry. 2015. “Self Selection Into Credit Markets: Evidence From Agriculture in Mali.” National Bureau of Economic Research (NEBR) Working Paper 20387.
Crépon, Bruno, Florencia Devoto, Esther Duflo, and William Parienté. 2015. “Estimating the Impact of Microcredit on Those Who Take It Up: Evidence from a Randomized Experiment in Morocco.” American Economic Journal: Applied Economics, 7(1): 123–150.
de Janvry, Alain, and Elisabeth Sadoulet. “Financial Services for the Poor.” In Development Economics: Theory and Practice, 463–494. New York: Routledge, 2016.
Field, Erica, Rohini Pande, John Papp, and Y. Jeanette Park. 2012. “Repayment Flexibility Can Reduce Financial Stress: A Randomized Control Trial with Microfinance Clients in India (Repayment Flexibility Can Reduce Financial Stress).” Public Library of Science, 7(9): e45679.
Field, Erica, Rohini Pande, John Papp, and Natalia Rigol. 2013. “Does the Classic Microfinance Model Discourage Entrepreneurship Among the Poor? Experimental Evidence from India.” American Economic Review, 103 (6): 2196–2226.
Gine, Xavier, and Dean Karlan. 2014. “Group versus Individual Liability: Long Term Evidence from Philippine Microcredit Lending Groups.” Journal of Development Economics, 107 (2014): 65–83.
Harper, Malcom. “Some final thoughts.” In What’s Wrong With Microfinance?, eds. Thomas Ditcher and Michael Harper, 257–259. Rugby: Practical Action, 2007.
Hulme, David. “Is Microdebt Good for Poor People? A Note on The Dark Side Of Microfinance.” In What’s Wrong With Microfinance?, eds. Thomas Ditcher and Michael Harper, 19–22. Rugby: Practical Action, 2007.
Karlan, Dean, and Jonathan Morduch. “Access to Finance.” In Handbook of Development Economics, eds. Dani Rodrik and Mark Rosenzweig, Vol. 5: 4703–4784. Amsterdam: Elsevier, 2010.
Woodruff, Christopher, David McKenzie, and Suresh de Mel. 2007. “Returns To Capital In Microenterprises: Evidence From A Field Experiment.” World Bank Research Working Paper 4230.
Yunus, Muhammad. Banker to the Poor: Micro-lending and the Battle against World Poverty. New York: PublicAffairs, 2007.
Hopeline Training Center. “http://hopelineinstitutegh.org” (accessed May 11. 2018)
Government of Ghana Official Portal
“http://www.ghana.gov.gh/index.php/about-ghana/regions/northern” (accessed May 11. 2018)