An open source based approach to predict nuclear proliferation decisions

This paper presents an attempt to predict a country’s nuclear proliferation decisions based on open source information. The approach is based on the combined use of data on a country’s financial status, technological capability, and political motivation. Nuclear fuel cycle capabilities were included as part of technological capability. The database was compiled from the information from the IAEA data center and the collected information of political and economic status data by the Correlates of War Project. To identify determinants of nuclear proliferation and to reveal the relationship between the proliferation decision of a nation and the basis of the decision, correlations among the input variables were analyzed. Based on the use of selected sets of input variables, predictive models were developed by using Weibull and Cox event history modeling. The developed models were used to predict proliferation decisions against the historical records of nuclear proliferation from 1945 through 2000. Results of predictions using these models were compared with historical proliferation events observed in various countries with respect to “explore”, “pursue”, and “acquire” decisions. For the “acquire” decision, accounting for the nuclear fuel cycle capabilities of a nation was found to be important in improving the predictions of nuclear proliferation. Overall, this study indicates that predictive models based on open source information would be useful in providing warnings for potential nuclear proliferation attempts.