Multilateral Climate-Finance Approvals and Climate Vulnerability in Latin America and the Caribbean: Dynamic Panel Evidence, 2006–2024

Dynamic two-way fixed-effects models are estimated on a common sample of 333 observations from 18 countries with Driscoll–Kraay inference. The preferred specification indicates strong vulnerability persistence (coefficient = 0.784; p < 0.001), while total approvals (0.245; p = 0.505), adaptation approvals (−0.266; p = 0.877), institutional quality (−0.054; p = 0.532), and the finance–institution interaction (0.162; p = 0.800) are imprecisely estimated.

Disaster incidence is positively associated with vulnerability under Driscoll–Kraay inference (0.0716; p = 0.032) but not with country-clustered inference (p = 0.122). Results remain substantively unchanged across alternative covariance bandwidths, lagged and cumulative approvals, leave-one-country-out estimates, and a half-panel jackknife correction. The evidence does not establish that climate finance is ineffective; rather, it shows that sparse annual approval flows are an inadequate proxy for executed resilience investment.

Credible impact evaluation requires linked information on approvals, disbursements, implementation, completion, and outcomes.</p>