Pooled data - Time-Series & Cross-Sectional Data

Some bodies of data have both cross-sectional and time-series dimensions. For example, you may have data organized by country and by year. EViews will help keep track of the data, do operations automatically in the cross-sectional dimension, and apply econometric methods that extract information from both dimensions. This part shows some basic skills to handle it.


Create A Pool Object

Step 1 : Chosse the Objects on the tool bar of the EViews.
Step 2 : Then a select menu will pop up. Choose New Object....
Step 3 : After that a dialog will show. Choose Pool in the Type of Object box and type a name in the Name for Object box. The name GDP is an example in this case.
Step 4 : Then a window will pop up. Type the cross-section member names in the window.
Step 5 : The cross-section member names is the country names in this case ( CAN, FRA, GER, ITA, JPN, UK, US ).
Step 6 : After that, a pool object will be created. It is GDP as named before in this case.
Back to the top


Manipulating Pooled Data

Step 1 : Open the pool object ( GDP in this example ) and choose the PoolGenr on the tool bar of the window.
Step 2 : Then a dialog box will pop up. Type the equation in the box to generate new series. It is "ratio? = gdp_? / gdp_us" in this example. In this case "?" can be used as a general expression for a pool series name. For example, "gdp_?" represents gdp_can, gdp_fra, gdp_ger, gdp_ita etc.
Step 3 : Since "ratio?" is used, the ratios for different countries will be created such as ratiocan, ratiocan, ratiofra, ratioger etc.
Back to the top


Pooled Estimation

Step 1 : Open the pool object ( GDP in this example ) and choose the Estimate on the tool bar of the window.
Step 2 : After that, a dialog box will show. Type the name of depentant variable in the top left box. Then type the names of common coefficients and cross-section specific coeffients in the box below. "?" can also be used in this case as the general expression. The area labeled "Common coeffients" should contain variables that are to receive the same coefficient across all members of the pool. The area labeled "Cross-section specific coeffients" should contain variables that are to receive different coefficients for each member of the pool.
Step 3 : At the bottom of the window, there are two kinds of check boxs, Intercept and Weighting. Under the type of Intercept, there are four options: "None" means no intercepts, "Common" means same intercept for all pool members, "Fixed effects" means that separate intercepts estimated for each pool member (fixed effects), "Random effects" means that treats intercepts as random variables across pool members (random effects). Under the type of Weighting, there are three options: "No weighting" means all observations are given equal weight, "Cross section weights" means weights estimated in preliminary regression with equal weights and then applied in weighted least squares in second round, "SUR" means seemingly unrelated regression--the covariance matrix across pool members is estimated in a preliminary regression and then applied in generalized least squares in the second round. Choose the suitable one among them and click OK to run the regression.
Back to the top


All right reserved by Dr. Bill Hung.