we need two time horizons: one for amortization and recurring costs, another for pop projection

[jea98]

currently NP has only one time horizon for the whole scenario. and we've been told to always run scenarios with a time horizon of 30 years, otherwise something like the investment in a hydro power plant or MV line will note likely seem cost-effective. but SE4All and other development time-frames are typically from "now" to 2030, so ~15 years. growth rates for population are some of the best growth rates in terms of data quality (much better than econ growth estimates) and they tend to be available from census sources. So people, including us, often want to input them for pop projections. but if these are going to give an accurate result for a 15 SE4All time horizon, they need to be adjusted down and then run in a 30 year model. Like some other issues with NP, this isn't enormously difficult to work around, but it does present a few real problems: 1) it requires a single operator to run and analyze the same scenario; otherwise the analyst will never understand why the pop growth rates are strange; 2) it breaks the intuitive experience of the software a bit, and if you add this to other fudges, you soon get scenarios that are really loaded with caveats, and using the software quickly becomes a kind of black art.

[chrisnatali]

Thanks for clarifying. Will try to work this into the metric-model decomposition effort.