# ExpForecast

Syntax

ExpForecast(Year1, Value1, Year2, Value2,... YearN, ValueN) or
ExpForecast(ExcelFile, ExcelRange)
ExpForecast(Branch:Variable)
ExpForecast(FirstYear, LastYear)

Summary

Forecasts future values based on an exponential regression of historical data. Regression is not forced through base year value.

Description

Exponential forecasting is used to estimate future values based on a time series of historical data. The new values are predicted using linear regression to an exponential growth model (Y = m + X^c) where the Y terms corresponds to the variable to be forecast and the X term is years. Exponential forecasting is most useful in cases where certain values can be expected to grow at constant growth rates over the period in question (e.g. population levels).

Use this function with caution. You may need to first use a spreadsheet or some other package to test the statistical validity of the forecast (i.e. test how well the regression "fits" the historical data). You can also use the LEAP's built-in statistical functions such as RSquared, StandardError, and Observations to help with this task.  Moreover, bear in mind that future values may be markedly different from historical ones, particular if structural or policy shifts in the economy such as changing energy policies are likely to have an impact on future trends.

Typically, you will enter this function in your Current Accounts data and use it to forecast values for the base year and forward. If your historical data includes a value for the base year, it will be included in the regression, but the forecast value for the base year will not necessarily equal the base year value (depending on how well the regression fits the actual data). If you wish to force the regression through the base year value use the similar ExpForecastBY function.

The syntax ExpForecast(FirstYear, LastYear) is a special syntax that can only be used in scenarios (not in Current Accounts).  This syntax generates a forecast in a scenario based on the data in Current Accounts.  Only the values between and including FirstYear and LastYear are used as input to the regression that generates the forecast.  FirstYear must be on or after the base year. LastYear must be before the first scenario year. LastYear must be at least three years after FirstYear.