join.gdp.ts: Join past GDP time series to future.

View source: R/zchunk_L102.GDP.R

join.gdp.tsR Documentation

Join past GDP time series to future.

Description

When we have to join two GDP time series, we usually find that they don't match up at year of overlap (the "base year"). What we do in these cases is we compute, for the later time series, ratios of GDPs in the future years to those in the base year. We then multiply the future ratios by the past base year value. That future time series can then be grafted onto the past without leaving a seam.

Usage

join.gdp.ts(past, future, grouping)

Arguments

past

Tibble with the past time series (year, gdp, and grouping).

future

Tibble with the future data (year, gdp, scenario, and grouping).

grouping

Name of the grouping column (generally either 'iso' or 'GCAM_region_ID', but could be anything

Details

In practice, the past is often a single time series, while the future is often a collection of scenarios. Therefore, we assume that the past time series has no scenario column. If the future does not have a scenario column, it is given a dummy one, which is dropped before the new table is returned. Note that we look for lower-case 'scenario' for this.

The base year is calculated automatically. It is the maximum of the years that overlap between the two data sets.

We also have to know how to group the data for calculating the gdp ratios. Normally this will be either by country ('iso') or by GCAM region ('GCAM_region_ID'). The choice of which is passed in as the 'grouping' argument.

Finally, although we have discussed this function in terms of joining two GDP time series, in the future time series we use only the ratios of GDP to base year GDP. Therefore, any time series with the correct ratios will work. For example, if we have a time series of growth rates, we can convert those to ratios using cumprod and pass those ratios as the future time series. For similar reasons, even if the two time series have different units (e.g., different dollar-years or PPP vs. MER), they can still be joined. The units of the output time series will be the same as the units of past.

Value

Time series with the past and future joined as described in details.


JGCRI/gcamdata documentation built on March 21, 2023, 2:19 a.m.