The concept of microfinance has existed for centuries and can be traced back to the time of indigenous Rotating Savings and Credit Associations (ROSCAs), money- lenders, and local co-operatives (Armendariz de Aghion & Morduch, 2005). However, the emergence of the modern version of microfinance as it is known today is usually credited to Bangladesh’s Mohammed Yunus, who in the 1970s started offering small loans to basket-weavers and bamboo furniture makers in rural Bangladesh to help them with their businesses. Yunus continued to give out these loans for nearly a decade, before forming the well-known Grameen Bank in 1983, with the aim of reaching out to a wider group of people.

With the emergence of the so-called microfinance revolution, Bangladesh is seen as a pioneer in the industry and is the home of several microfinance institutions (MFIs)

and extensive microfinance operations throughout the world (Khandker, 2005). While the industry in Bangladesh has received much praise in terms of poverty reduction, others remain sceptical about the positive impact of microfinance. Thus, donors (both prospective and current), government agencies, policymakers, and stakeholders are all very interested in understanding what works and what does not in microfinance, hence the need for greater evidence to ascertain what the impacts of microfinance interventions are. Recent systematic reviews such as Duvendack et al. (2011), Stewart et al. (2010) and van Rooyen et al. (2012) conducted non-empirical syntheses of the existing literature on the impact of microfinance, and conclusions from these studies suggest no meaningful impact of microfinance.

In this study, we focus mainly on Bangladesh and examine if findings from an empirical synthesis would corroborate the findings from the existing systematic reviews. We focus on Bangladesh because it is the home of modern microfinance, and has benefited from several microfinance-related endowments. While several reports suggest a positive effect of microcredit, with the likelihood of publication bias it is important to know whether or not the impact of microcredit on poverty in Bangladesh is truly positive. Publication selection bias occurs when researchers, editors, and reviewers are predisposed to select only statistically significant results that are consistent with their existing theory. It has been established that publication selection bias is a threat to empirical economics (Stanley, 2008).

Subsequent studies such as those conducted by Card and Krueger (1995), Abreu et al. (2005), Doucouliagos (2005), and Stanley (2008) all raise concerns about publication bias. With regards to microfinance, this bias can actually extend to the predisposition to reject studies that report negatively on the impacts of microfinance interventions.

In addition, the problem of heterogeneity across studies makes it difficult to come to a general conclusion about the effects of microcredit on poverty. In this study, we consider three proxies for poverty in the existing literature – income, assets, and consumption/expenditure – and present a statistical approach to dealing with the issues of heterogeneity. We also address issues of publication bias and examine whether a genuine effect exists in the case of microcredit and poverty in Bangladesh. First, we provide fixed effect estimates of the weighted mean effects of the impact of microcredit on poverty, which provide an understanding into the overall evidence base that exists. Second, we conduct precision effect tests (PETs) and funnel asymmetry tests (FATs) to help determine a verifiable and reliable summary measure of the genuine effects of microcredit on poverty beyond publication bias, and towards a more informed evidence-based policymaking. Lastly, with a meta-regression analysis (MRA) that includes various moderating variables, we examine the potential sources of variations in the literature.