Geographic Neighbourhood and Cluster Formation: Evidence from Indian Agriculture
研究发现收入集群中既有地理邻居也有非邻居，且并非所有邻居都同属一个集群；基于印度各邦农业数据的逻辑回归表明，集群成员身份主要由非地理因素（如天气冲击、资源约束、技术投入等）驱动。
We study an empirical occurrence largely overlooked in studies on income clusters: (i) most clusters include geographic neighbours and non-neighbours; and (ii) not all geographic neighbours are cluster-co-members. Using agricultural income across Indian states, we find a similar pattern in income-clusters over the last 45 years. Logistic regressions that consider state-pairs as the unit of analysis show that cluster membership is not driven by geographic variables but rather by non-geographic factors like weather shock, resource constraints, technology/input usage, extent of crop diversification, infrastructure, policy and institutions