The second robustness check is about the censoring thresholds in the Tobit model. When estimating the Tobit model, we intentionally avoided making strict assumptions regarding the precise level at which expenditure outcomes were censored. To demonstrate that this choice does not fundamentally influence our results, we explored models using both minimum and maximum censoring thresholds. In this con- text, minimum (left) censoring implies that all observed values of the dependent variable below a specific lower bound are truncated, so any value less than this threshold is effectively set at the limit. Conversely, maximum (right) censoring means that values above a certain upper bound are similarly truncated, with any values exceeding this point recorded at the maximum threshold. This approach allows us to assess the