Hello, I am investigating the effects of climate finance on climate vulnerability scores in Sub-Saharan countries from 2003 to 2018 by using a fixed effects regression and lagging finance by 2 years. Attached are images of a univariate model and a multivariate model. The other independent variables are climate risk index (cri), agriculture contribution to GDP (agric), institutional ability to effectively use adaptation investment (readiness), GDP per capita (gdppc) and population (pop). My choice of variables is based on consensus in the literature that vulnerability is a factor of exposure, sensitivity and adaptive capacity. From my results, finance is negatively associated with vulnerability, it is also significant in both models. However, my other explanatory variables are proving to be poor predictors despite showing expected signs in the coefficients except for readiness . The f-test in the univariate model is <0.5 which is okay but in the multivariate model, there is no statistic. My main question is, does that mean it's a poor model? I'm not a big fan of transforming variables but is that a recommended option? Also, is the intraclass correlation (rho) of 0.993 further evidence of a poor model ? I'm not sure of the validity of my multivariate model and would highly appreciate your advice. Thank you

Array Array