# Regression and Simultaneous Equations

Regression
Linear regression is a common tool in econometrics and can be used for inflation forecasting. Economists create forecasts with regression techniques by estimating an equation or model, for the effect, or dependent variable, to be estimated and substituting values for each independent variable in the regression model. For example, a regression model predicting the future rate of inflation might include predictors like future price expectations, rate of interest and the economic conditions prevailing in oil-producing countries.   When using basic regression analysis as a forecasting tool, one must make use of a model with a good fit (as measured by the value of R2) to produce an accurate prediction.
Simultaneous Equation Models
Interdependence is a reality in the economic world, with many variables having two-way causal relationships. For example, GDP and investment levels affect each other simultaneously. A single regression equation with only one dependent variable cannot capture such relationships. Simultaneous equation models are a helpful forecasting tool when variables such as GDP, investment and others can be both a cause and effect. The simultaneous approach uses two regression models in which jointly determined, or endogenous, variables appear in both equations. An analyst could use a simultaneous equation approach to forecast the price and quantity of beef. One equation would express the quantity of beef as a function of consumers' incomes, beef prices, the costs of beef production and the prices of competing meats, while the second expresses beef prices as a function of beef quantity, prices of other meats and beef production costs. A drawback of the simultaneous equation approach, however, is that it violates the classical econometric assumption that each explanatory variable and the error term (the amount of variance unexplained by a regression model) are independent of each other.