Panel filling

[goldma99]

I have two unbalanced panels: beaver presence and land use. I need to fill in the missing years.

Beaver survey

• Variable type: dummy of beaver presence in river grid cell in year
• Years available: 2012, 2017, 2020
• Approach: assume first recorded positive was the first year of presence (e.g., if 2012 has 0 and 2017 has 1, assume 2013-2016 have 0).
• Robustness: assume first non-recorded negative was the first year of presence (e.g., if 2012 survey has 0 and 2017 has 1, assume 2013-2016 have 1)
  • A further check could be to randomly select a first year between 0s and 1s (e.g., if 2012 has 0 and 2017 has 1, randomly select a year in 2013-2016 to be the first year).

Land use

• Variable type: continuous proportion of river grid cell with ag land use
• Years available: 1990, 2000, 2007, 2015, 2017, 2018, 2019, 2020, 2021, 2022
• Approach: linearly interpolate between data years.
• Robustness: ?

[goldma99]

• [ ] Start merging as much as possible. For each cell-year, we want beaver presence, ag share, river levels, ag parish.
• [ ] Start thinking about controls: annual precip/temp averages, nat'l or global prices of major cereal commodities