Survey of Businesses Receiving The People's Business Credit 2021

An administrative database (SIKP), which contains basic characteristics of all KUR borrowers since 2016, served as the sampling frame for the quantitative data collection. Weighted stratified random sampling was used to select the sample. Strata were based on four characteristics that may influence beneficiaries’ experiences with KUR and how KUR may change their business: gender of KUR recipient, size of KUR loan, financial institution that issued the KUR loan, and geographic region. Strata including less than 1% of KUR beneficiaries were oversampled in order to ensure that each subgroup of interest would have sufficient representation in the sample in order to draw precise estimates at the subgroup level.

Stratified sampling methodology was chosen because the team wanted to ensure that subgroup analysis was feasible across certain dimensions. Some of the subgroups of interest represent only a small portion of KUR borrowers, so a random sampling approach without using strata may not have provided a sufficient number of observations to draw any conclusions about some of these subgroups. Gender was included as a stratification variable to ensure that a gender-sensitive analysis was feasible. Female entrepreneurs in Indonesia face greater financing constraints than male entrepreneurs (World Bank 2023), so KUR may have particularly strong impacts for female entrepreneurs. Nevertheless, the market-based implementation of KUR may also limit the ability of KUR to reach female entrepreneurs, if it does not alleviate gendered constraints to accessing financing. Micro KUR loans and small KUR loans have different requirements and offer different sizes of subsidies to the KUR distributors. As such, it is critical to be able to analyze them separately. Because less than 10 percent of KUR loans are small KUR loans, stratification on this variable ensures that there is enough statistical power to draw conclusions about small KUR loans. One financial institution, BRI, issues the majority of KUR loans. Because KUR is implemented by different distributors and some aspects of implementation are left to the distributor’s discretion, it is important to understand whether the implementation of KUR looks different when issued by the dominant bank or when issued by other distributors. Finally, financing conditions and alternatives vary across geography. Because the environment may shift how important KUR is to MSMEs, it is important to be able to understand how trends vary across different regions. Some regions have less than 10 percent of KUR borrowers in them, so a simple random selection may not have produced enough observations in some regions to allow for analysis disaggregated by region.

Generally, strata including firms with KUR loans of more than 25 million and those outside of Jawa were over-sampled, while firms receiving loans of less than 25 million in Jawa were under-sampled to ensure the total sample size rested within budget and logistical constraints. Finally, an even number of firms were selected for the sample from each strata so that they can be split into halves, where one half would answer the modules in questionnaire A and the other half would answer modules in questionnaire B. This allows the design weights to remain constant for all variables in the survey and facilitates data analysis. The modules to be asked were randomly assigned and balanced across sampling strata to ensure all modules included nationally representative information. Due to the weighted sampling design, design weights are used in all descriptive analysis in this report, and once incorporating the design weights the analysis is representative of all KUR recipients since 2016.

The survey firm received a randomized order list of firms within each strata and were instructed to call respondents until reaching the quota per strata.