Problems in Forecasting Water Requirements1

ABSTRACT.Methodological problems associated with forecasting water requirements by use of regression analysis are examined. Problems occurring when long‐range forecasts are based on linear and nonlinear extrapolation of time series models include possible changes in socioeconomic conditions, water allocation system structure, and limits to growth. Problems arising in forecasting based on multiple regression models are likely to involve serially dependent errors, multicollinear explanatory variables, and difficulties inherent to the presence of explanatory variables that must themselves be predicted.