Ex-ante economic impact analyses are required to demonstrate the development impact and economic viability of loans and grants extended by multi-lateral development banks. These assessments are performed under tight time constraints and often in data poor environments. This paper develops a framework for assessing development interventions in data poor environments and applies it to the analysis of the Sustainable Tourism Program, a US$15 million loan from the Inter-American Development Bank to the Government of Belize to foster tourism development in emerging destinations and enhance participation of low income people in the tourism value chain. This paper contributes to the literature in two critical ways: (i) the paper develops a generalizable approach to building dynamic computable general equilibrium models for development policy analysis in data poor environments; (ii) realistic expectations of agent behavioral responses to development interventions are required to calibrate model simulations. To estimate business as usual tourism arrivals and expenditure, auto-regressive integrated moving average methods were used. To estimate agent response to the development intervention, a survey-based quasi-contingent valuation approach was employed. These projections and information on investment structuring and costs were used to calibrate the model shocks. Results of this analysis show that the proposed investment will have positive impacts on Belize's economy by hastening economic growth. Gross domestic product increases 3% by 2040 and unemployment falls from 12% to 10%. Cross validating with a break-even scenario confirms that the Government of Belize would recover all investment costs even if actual agent demand response were considerably less than forecast. The model developed here may be applied to the ex-ante economic analysis of other sectoral development interventions ranging from agricultural policy to fiscal policy, from integration and trade, to health and education.
ex-ante economic impact analysis tourism development economy-wide model auto-regressive integrated moving average contingent valuation data poor environment