Which forecasting model is best used when relevant data is scarce?

When relevant data is scarce, the best approach is often to rely on a subjective model. Subjective models incorporate expert opinions, insights, and judgment to make forecasts. This method is particularly valuable in situations where quantitative data is limited or unreliable, allowing for the integration of knowledge and experience that quantitative methods may not capture.

In contrast, moving average and exponential smoothing models rely heavily on existing data trends to produce forecasts. These methods may not perform well when data is sparse, as they depend on historical patterns to make predictions. The causal model also necessitates having sufficient data to identify and understand the relationships between different variables, which is challenging when relevant data is lacking.

By utilizing a subjective model, forecasters can still generate useful estimates and make informed decisions, drawing on the collective insights and intuition of those familiar with the context or subject matter. This makes it a suitable choice in scenarios where data scarcity hinders the effectiveness of more data-driven approaches.