Source: LIRNEasia national baseline survey, 2015 7 Respondents from higher income groups were more frequent teleusers , with just 15% having never used a phone before, compared with 38% of those from lower income groups. Eighty percent of this use was from a mobile phone. When the purpose of calling was considered, owners tended to have more livelihood-related calls, although this kind of usage was low compared to social-purpose-related calls. Mobile ownership and factors influencing mobile adoption 8 At an individual level, mobile phone ownership (having an active SIM, with or without a handset ) among 15–65 year olds in Myanmar was 39%. A relatively large gender gap of 29% was seen, with women’s ownership being just 33% compared to 47% among men. Mobile penetration was greater in higher spending households (i.e., those with higher income or more wealth) 42% of them had a mobile phone and SIM. But even in poor households the number was 35%. This figure is remarkable, given that in 2006, nearly 9 years after market liberalization, only 9% of the Indian poor had a phone, 9 23% of Pakistan and 22% in Sri Lanka. The urban-rural ownership gap was 58%, with urban mobile ownership at 65% (driven largely by the three big cities: Yangon, Mandalay and Nay Pyi Taw) and rural at 27%. However, these levels were still higher than the 2006 levels seen in India, Pakistan and Sri Lanka (who were each several years post-liberalization compared to Myanmar at 1 year from liberalization). Affordability and a lack of a need for a phone were given as the main reasons for non-ownership. A logistic modeling of the factors determining the ‘odds’ of adopting was conducted. The ‘Odds’ is directly related 10 to (but not the same as) the probability of something happening. A binary logistic regression is a way of modeling the probability of an event when the event is a binary outcome, so mobile adoption = 1 (yes) or 0 (no). The coefficient of a logistic regression (Column 1 in Table 1) cannot be directly interpreted, therefore the Odds Ratio (Column 2) is calculated from which the, the change in the odds of 11 adoption associated with a 1-unit increase in the explanatory variable (e.g., gender, education, etc) may be calculated (given in Column 3). This analysis shows that the largest impact on the odds (and therefore probability) of mobile adoption is from the completion of tertiary education in Myanmar, although the actual number of respondents who have completed tertiary education is low. A person who has completed tertiary education in Myanmar has 378% higher odds of mobile adoption than someone who has not completed tertiary education but has all other measured characteristics (e.g., gender, household characteristics, etc) the same. Having completed secondary education is associated with increased odds of 55%. Being a female is associated with a reduced odds of adoption of 42%. Age has no significant association. Household characteristics appear to have a positive association with mobile adoption, with having a TV, electricity and household spending being positively associated with the odds of adoption. Network variables are significant and positive, with each additional 10% of a respondent’s family owning a mobile associated with an increased odds of adoption of 36%. Positive perceptions of the benefits of 7 That is having used any) phone before. Although responses show that a negligible number (0.7%) own an active SIM but not a handset of their own, we consider these SIM-only owners as mobile ‘owners’. 8 9 As seen in LIRNEasia’s Teleuse@BOP2 data. 10 The ‘odds’ of and event happening is equal to the probability of the event happening divided by the probability of the event not happening; in this case, Odds of mobile adoption = probability of adoption / probability of not adopting 11 This type of modeling and data can not be used to ascertain causality, only association.