Coursehero >> Ohio >> Ohio State >> ECON 642

Course Hero has millions of student submitted documents similar to the one below including study guides, homework solutions, papers, exam answer keys and textbook solutions.

home Take exercise 1: EViews guide with exercise; Econ 6421 April 2008 R. de Jong Being able to use a statistical package such as EViews and being able to interpret the output correctly is one of the most important things that Econ 642 should enrich you with. EViews is a reasonably user-friendly package (although you may not think so when you use it for the first time). In principle, however, most statistical packages produce similar output, and once you have mastered EViews, learning another package should be relatively straightforward. If you should have any questions, please ask Jungick Lee, the TA for this class. His email address is lee.2570@osu.edu. Please start the exercise early. The 318 Arps computer lab is closed on weekends, and therefore you should be finished with your exercise on Friday. Start on Thursday at the very latest! A. Basics 1. How to start EViews a. If you have an Eviews CD: Put your EViews 6 Student Version2 CD into the CD-ROM drive of the computer. A window will appear that gives you five options. Choose "run EViews from CD". EViews will now start. If EViews does not start automatically, click "Start-Run" and type "d:autorun" to start the CD. b. If you do not have an EViews CD and you are in Econ lab (AP305 or AP318): Simply double-click the EViews icon on the desktop (username: labuser, password: economics). EViews should also be available on PCs in the following computer labs: 145 Brown Hall, 590 Baker Systems, 5 Agricultural Administration. 2. Creating a workfile We will now create a workfile, which contains variable names, number of observations, whether the data is monthly, quarterly etc. We shall read the data into this workfile as well. Click: File New Workfile. guide is based on EViews 5.x or 6.0. can be purchased online (http://www.eviews.com/eviews6/eviews6s/evstud6.html) at the price of $39.95 if you want to own your own copy. 2 This 1 This 1 Choose "Unstructured / Undated" (used for cross-sectional data) in Workfile structure type. Observations (i.e., sample size): 15 in Data range. Click OK or press Enter 3. Loading data 1. Open NotePad (Start - Programs - Accessories - NotePad) from Start-Programs-Accessories. Enter the data shown in Table 1. Enter the data, starting with a first line containing "x y" (i.e. the variable names; use lowercase letters for your variable names), and put "1 0.58" on the second line (make no empty line spacing), and so forth. To deliminate x and y (or 1 and 0.58) use Tab or Spacebar on the keyboard. Save the file as "chicken.txt" and remember where you saved the file. 2. Close the data file you just made and return to the EViews screen again. Click (from the workfile window, not the program window) Procs Import Read Text/Lotus/Excel. Find the file chicken.txt from the folder where you saved it and open it. A dialogue box appears where you should type "2" in the box labeled Name for series or Number if named in file, which enables the package to understand that your data has two columns. In the field named Delimiters, check Tab if you use Tab key or check Space if you use Spacebar to deliminate your data before. Click OK or press Enter. 4. Viewing your data To view the x series, simply double-click on x and you will see a window in which the values of x are displayed. Do the same for y. Then click on x and cntrl-click y (hold the cntrol key on the keyboard while clicking on y). You should now have highlighted both x and y. Right-click the highlighted variables and choose "Open as Group". You will now see a window containing the values of both x and y. 2 5. Running regressions Click Quick-Estimate Equation from the main menu or Proc-Make Equation from the menu in Group window. A dialogue box appears where you have to specify your regression. Enter "y c x" in the main window. This means the dependent variable is y, which is regressed on a constant and x. Click OK or press enter. A window with the regression results will appear. Answer the following questions: 1. Do we get an upward-sloping regression line? ^ ^ 2. What are the values of 0 and 1 from this regression ? 6. Viewing graphs Next, we will draw a graph of y against x and the regression line. Close the Equation window. Click variable x first to highlight it. Then press control, hold it, and click on y. Note: the first variable that you click will be plotted on the horizontal axis. Right-click the highlighted variables and click "Open as Group". In the new window, click "View-Graph-Scatter-Scatter with Regression". Click OK. Close this window after inspecting the graph. 7. Saving the file Go to File Save As and save the file as chicken.wf1. The wf1 extension (EViews Workfile) is provided automatically. Close the Workfile window. You can reload the workfile by clicking "File Open EViews Workfile" and finding the chicken.wf1 file from the folder where you saved. You should be back to exactly the point where you left off. Workfiles are easy to reload, and allow you to resume work exactly where you left off. 3 8. Generating new data series On the current workfile, we will create variables containing the log of x and the log of y. Click Quick Generate Series from the main menu or click Genr from the workfile window. A dialogue box appears. Enter "logx=log(x)". click OK or press enter. In the same way, create the log of y in the variable "logy". To run a regression with log(y) as the dependent variable and the constant and log(x) as regressors, follow Step 5, only instead of "x" write "log(x)" and instead of "y" write "log(y)". Answer the following questions: ^ ^ 3. What are the values of 0 and 1 from this regression ? 4. Do we get an upward-sloping regression line? Next, graph the regression line for this regression. 5. Now based only on both graphs, what regression do you like best, the one where y is regressed on x, or the one where log(y) is regressed on log(x) ? 9. Using data files in text format from the web The file "consinc.txt" contains national consumption and national income of The Netherlands for a number of years. Using web-browser it an be found at http://www.econ.ohio-state.edu/dejong/consinc.txt You can see the two series of data. Answer the following question: 6. What does it look like? Is the first or the second series national income? Save the file "consinc.txt" somewhere on your computer using Internet Explorer's File-Save Page 4 As function. Open the "consinc.txt" file in NotePad and add a new first line to the file that contains the variable names "cons" and "inc" in the correct order, separated by a space. Then start a new workfile, and load this data set into the workfile. This datafile has 42 observations. 7. From the regression of national consumption on national income, what is your estimate of the marginal rate of consumption? 10. Alternative program: WebRegress Instead of using Eviews, you can also use my program WebRegress in this class. program This is web-based and can be used from any internet-enabled computer. Open the Mozilla Firefox browser and go to http://web.econ.ohio-state.edu/dejong/. Click "Start WebRegress". Click "Browse" and choose the file "consinc.txt" that you saved on your computer. Click "Start". The computer will now give you some information about the data. Click "Continue". Now choose "cons" as your dependent variable and "intercept" and "inc" as your regressors. Click "Ordinary Least Squares". 8. Did you get the same results? 9. Click "draw a graph". You will see some clustered points below the regression line. What did you think was happening at the time of these observations? Click "Do another regression" and "Transformations". We will now create the logarithms of x and y in WebRegress. For the name of the new variable, specify "logcons" and as transformation, specify "log(cons)". In the same fashion, create a variable named "loginc" as the logarithm of "inc". 10. Run a regression of "logcons" on "loginc". What slope coefficient do you get? 11. Using Eviews workfiles from the web Close the old workfile. 5 Load the file "grades.wf1". Do this by starting Mozilla Firefox and typing the address http://www.econ.ohio-state.edu/dejong/grades.wf1 on the address line. Check to make sure the file has 24 observations. When the computer asks what to do next, save the file as "grades.wf1". This file contains midterm and homework grades of a course that I taught long ago in The Netherlands. Next, load this workfile into Eviews. Run a regression of the variable "midterm" (the grade a student got for the midterm) on "home1" (the grade for the first homework). 11. How do you interpret the results of the last regression? Next, run a regression of the variable "midterm" on "home2" and draw a graph. Answer the following questions: 12. What is the value of R-squared in this regression? 13. Which homework seems to have more influence on the midterm grade, the first or the second? Why? 14. For an individual with a homework 2 grade of 90, what is the predicted midterm grade? B. Estimation with panel data 1. Finding data file We use the data used in Example 13.8 of your textbook, which can be available from http://web.econ.ohio-state.edu/jungick/e642data.htm Download and save the ASCII data file, ezunem.raw. 2. Making workfile You can make an EViews Workfile directly from the data file as follows: 6 Open ezunem.raw by clicking File Open Foreign Data as Workfile in EViews main menu. You will encounter ASCII Read wizard windows (3 steps): Take the default, that is, just click Next Next Finish. You should be able to see now Workfile window and Group window (that shows the data in spreadsheet format). Close the Group window; you can save it or delete it whatever you want. 3. Pooled OLS estimation You can replicate the result of Example 13.8 in your textbook as follows: Click Quick Estimate Equation in the main menu; Specify your equation to be estimated as in Equation (13.32) of your textbook: guclms c d82 d83 d84 d85 d86 d87 d88 cez Note: (i) guclms = log(uclms); the growth rate of unemployment claims. (ii) cez = (ez). 15. What is the estimate of 1 , the coefficient on cez, and its standard error? Are they the same as in your textbook? How do you interpret this estimation result? (You may want to save the Equation object for future comparison. Close the Equation window after providing it with a name, say "pooled", to save it.) 4. Panel estimation Note that the city fixed effect, ai , disappeared in the above specification due to the first differencing (Example 13.8). If you specify your equation to be estimated as in Computer Exercise C14.4 of your textbook, i.e., log(uclmsit ) = ai + cit + 1 ezit + uit , where ai and ci are both unobserved effects, you can allow for more heterogeneity across cities. (Note: The term cit allows each city to have its own time trend.) By first differencing this equation, you can obtain log(uclmsit ) = ci + 1 ezit + uit () as in part (i) of Computer Exercise C14.4. Because the differenced equation contains a fixed effect, ci , it is appropriate to estimate it by fixed effects. Here is the procedure for FE panel estimation: 7 To use panel estimation in EViews, first you need to structure your workfile so that EViews can recognize it as panel data: Click Proc Stucture/Resize Current Page in Workfile window menu. Select "Dated Panel" in Workfile structure type field. Then you can some new dialog boxes in the window. In Panel identifier series field, type city for Cross-section ID series and year for Date series. For the other parts, you can take the default. Click OK or press Enter. Now you will notice that new series called dateid is created in Workfile window. You can also notice that Range and Sample in Workfile window (right below the menu) is changed from Range: 1 198 198 obs Sample: 1 198 198 obs to Range: 1980 1988 22 198 obs Sample: 1980 1988 198 obs This change means that the workfile structure is changed to dated panel data. Now, for FE estimation Click Quick Estimate Equation in the main menu; Specify the above equation () to be estimated: guclms cez [Note: Even though you did not include a constant in panel data regressions, EViews always include it whether you specify it or not. (You will see this in a second.) That is, specifying regression equation as guclms c cez gives you the same estimation result.] Next, click "Panel Options" tab and select "Fixed" only for Cross-section in the Effects specification field. Click OK or press Enter. 8 16. What is the estimate of 1 , the coefficient on cez, and its standard error? Are they different from those obtained in Question 15 above? Which one is the larger in magnitude? Is the effect of enterprise zones still statistically significant? If we add all the year dummies to equation (), the estimated enterprise zone effect will be reduced. (Why?) To see this, Click Estimate in Equation window and modify equation specification as guclms d82 d83 d84 d85 d86 d87 d88 cez [Note: Since EViews always include a constant whether you specify it or not, you need to drop one dummy, here d81.] You don't need to change "Panel Options." Click OK or press Enter. 17. What is the estimate of 1 , the coefficient on cez and its standard error? Is the result consistent with your intuition? What happens to the estimate of 1 if we specify time-period fixed effects as well (in addition to cross-section fixed effects), instead of adding the year dummies to the model? Hint for the last question: Click Estimate in Equation window and change equation specification back to guclms cez Click "Panel Options" tab and select "Fixed" for Period as well in the Effects specification field. Click OK or press Enter. 9 Answer sheet for homework 1 (Eviews exercise) de Jong Your name: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 10

Find millions of documents here - Study Guides, Homework Solutions, Papers, Exam Answer Keys and more. Course Hero has millions of course related materials that will enable you to learn better, faster and get an A in all your courses.
Below is a small sample set of documents: