The survey was conducted among the low-income, urban micro-entrepreneurs (MEs) in three countries, Bangladesh, India and Sri Lanka. The study defined micro-entrepreneurs as those who employed less than ten hired workers, i.e 0-9. The hired workers are paid employees or full-time equivalent, excluding the owner. This is an adaptation of international definition followed by World Bank and European Commission1. Low-income or the poor is identified as those who occupy C, D & E levels in the socioeconomic category (SEC) classification in Bangladesh and Sri Lanka. It was SEC D & E levels in India. Please see Annexure 1 for the SEC classification in three countries. In all the three countries, only cities or urban localities as defined by the respective national governments were selected for the study. The owners of the microenterprise who are older than 14 were included in the study.
There were small changes made in the study during the middle of fieldwork, as we were not able to find the desired number of MEs in the certain worker categories. In Bangladesh, SEC B1 and B2 were included to cover the MEs in 1-3 and 4-9 worker categories. In India, SEC B2 was added to achieve the quotas for 1-3 and 4-9 workers categories in Patna city. Whereas in New Delhi, SEC C was added to cover MEs with 4-9 employees. In Sri Lanka, we were not able to find the adequate number of MEs in the 1-3 and 4-9 workers category and extended the framework to include SEC B1 and B2.
In each country, two cities were selected. The capital city and another city which is weaker in terms of governance. The governance was inferred by the proxy indicators related to the human development of the province/state the city is located. The study assumed that low- income MEs would experience differential efficiency levels in customer services in these two locations due to the overall difference in governance levels. The performance of strong and weak provinces/states in which cities are located are presented in Annexure 2. In Bangladesh, Dhaka, the national capital city and Gaibandha and Kurigram cities from the Rajshahi Division were selected. In India, New Delhi, the National Capital city, and Patna, the capital city of Bihar State were selected on the basis of data available on cities. In Sri Lanka, Colombo, the national capital city from the Western province and four urban centers from Northwestern province, Kurunegala and Kuliyapitiya from Kurunegala district and Puttalam and Chilaw from the Puttalam district were selected. More than one urban centers or cities were selected in the weaker cities in Bangladesh and Sri Lanka due to an insufficient number of MEs in one location.
The respondents for the study were selected through a probability sampling method. In Sri Lanka, the A-Z map book published by The Survey Department of Sri Lanka was used. The entire province is divided into square blocks. The blocks were selected using a random number table. In each block, respondents were found first at the northwest corner and then subsequent interviews were conducted in the same block by continuous walking. In a few instances, for the 6-9 workers category, we used snowball sampling to find the respondents, but within the same block. In India, the city is divided into wards by the Election Commission of India. The list of wards was taken and sixteen wards were selected through a systematic sampling process. Within each ward, all the streets were listed alphabetically. Every third street was selected and respondents were found for the study. In Bangladesh, cities were divided into wards, then mahallas and then streets. A similar method as in India was followed. As some of the wards were inaccessible due to incessant hartals (public protests) and governmental restrictions, additional wards were selected using the similar process.
The sample size decided for each country was as follows: Sri Lanka – 900, India – 1200 and Bangladesh – 900. In each country, the respective sample was divided equally between the two locations differed by levels of governance, strong and weak. The numbers were decided after considering the required number of cases for statistical analysis and amount of funds available for the study.
As the countries did not have specific data on MEs, in other words, there was lack of data on low-income MEs, closest data extant national statistics on microenterprises was used to determine the quotas of the MEs in the sample. Also, the composition of MEs on parameters like location, gender, industrial domain and number of workers were not uniformly available across the countries, in each country different parameters, determined by the availability, were used in deciding the composition of the sample. The composition of MEs in the population and the sample is presented in Annexure 3.
A single questionnaire was used across the three countries. A colloquium of fifteen experts from all the countries was assembled in Colombo city and possible items for the questionnaire were deliberated. The experts were from electricity, telecom, governance and MEs domains. The questionnaire was constructed using the items gathered in the colloquium. The draft was reviewed by the experts over email and a skype conference. After modifications, the questionnaire was pilot tested in the three countries with a total of 30 respondents. On the basis of feedback, the questionnaire was modified and finalized. The final questionnaire was translated into four languages, Sinhala, Tamil, Hindi and Bangla. The translation was cross-checked by the native speakers who are also fluent in English.
An international market research agency was commissioned to conduct the fieldwork. It organized a three-day training program for the interviewers separately in the three countries. The training program involved discussion of the study objectives, understanding of the questionnaire and the sampling process. Few mock interviews were also conducted to enhance the familiarity with the questionnaire.
The fieldwork was conducted during March-May 2013. After identifying the valid respondents, the response rate for Bangladesh, India and Sri Lanka were 80 %, 99% and 86% respectively. The final sample size is 3180 (Bangladesh – 986, India – 1279 and Sri Lanka – 915). The final report presents the currency values in US dollars. The exchange rates used are 78 Taka (Bangladesh), 54 Indian Rupees (India) and 124 Lankan Rupees.