Autoregressive and distributed lag models:

There are  many  situations  in  which  economic  agents,  such  as  consumers  or producers,  respond  to changes  in  the  economic  environment with  a time  lag.  The main  causes  of  the  delayed  responses are  the  time  lags  in  the  transmission and reception  of information,  although the high cost of speedy adjustment can also be a factor, as can institutional constraints.

In  the sections above we mentioned several examples of economic activities where agents would typically have delayed responses, such as, the decision to invest  in new capital  equipment  by  a firm  as  sales  go  up,  the household's decision to increase consumption expenditure following an increase  in income, and the effect on a firm's profit subsequent to a large investment expenditure.

To  accurately  model  these delays we  need  to  incorporate  lagged  values  of  the independent and dependent variables  in our model structure. Some of these models, besides being empirically accurate, can be  justified  theoretically as well. Simple distributed lag models, with no  lagged,dependent variables may be estimated using, ordinary least  squares  techniques. However,  dynamic models  require more complicated. estimation techniques, such  as instrumelital variables or  iterative,  non- linear least squares.

Inference in  these models is straight forward and proceeds as in any standard multiple regression model.