Table 4 Time-variant variables expected to differentially affect deforestation within CFM and MNP, included as covariates in the event study analysis
Covariate (annual) | Units | Spatial resolution | Source |
|---|---|---|---|
Distance to forest edge | m | 90 m | Vielledent et al. (2018)31 updated to 2020; annual distance to forest edge calculated using Google Earth Engine80 |
Population density | People/km2 | 1 km | WorldPop73 |
Maximum accumulated precipitation | mm | 5 km | TerraClimate81 |
Maximum temperature | °C | 5 km | TerraClimate81 |
Drought severity | Palmer drought severity index | 5 km | TerraClimate81 |
Maximum wind speed | m/s | 5 km | TerraClimate81 |
Average rice price | Madagascar Ariary | NA | World Bank82 |
Standard deviation of rice price | Madagascar Ariary | NA | World Bank82 |
Time-invariant covariates used as interaction terms | |||
Distance to a road | m | 90 m | Rasolofoson et al. (2015)39 |
Distance to a village | m | 90 m | Rasolofoson et al. (2015)39 |
Distance to an urban center | m | 90 m | Rasolofoson et al. (2015)39 |
Population density in 2005 | People/km2 | 1 km | WorldPop73 |
Development level (index of material assets) in 2007 | 0 (below median) or 1 (above median) | Fonkontany (administrative unit) | Wu Yang, Conservation International, based on Communes Database83 |
Security conditions and risk of theft of property in 2007 | 0 (below median) or 1 (above median) | Fonkontany | Wu Yang, Conservation International, based on Communes Database83 |
