SEMINAR IN LABOR ECONOMICS



                                                                     Spring 1995




Prof. Arthur Diamond


Office:  CBA 512E


Office Phone:   554-3657


Office Hours:   11:15-noon Tues. & Thurs.; 4:30-6:00pm Thurs.; and by appointment.


CWIS userid (also CBA LAN userid):  adiamond


Internet address:





Course Objectives:

            Economics consists of tools of analysis that have proven useful in explaining a wide variety of human behavior.  Labor economics in particular makes use of supply and demand in the factor market for labor in order to explain the wage rate and the quantity of labor employed.  The basic objective of the course is to give the student a working familiarity with the relevant tools of analysis.  A secondary objective is to learn how institutions, such as unions and the government, affect the wage rate and the quantity of labor employed.  Thirdly, the student should learn how individual differences in human capital influence the supply and demand for a person's labor and thereby in part explain individual differences in income and likelihood of unemployment.  An additional important objective of the course is to give the student significant experience in completing all aspects of an empirical research project in labor economics.  Included in the skills used for satisfying this objective will be:  topic selection, literature search, data collection and transformation, econometric estimation, and report writing.




            Economic Theory: Micro (ECON 3200) and Economic Theory: Macro (ECON 3220) or permission of the instructor.



            Ronald G. Ehrenberg and Robert S. Smith, Modern Labor Economics:  Theory and Public Policy (5th ed.) (required)


            George Kosicki, Study Guide:  Labor Market Problems and Applications (to accompany Modern Labor Economics, 5th ed.) (required)




Other Useful Books:


            Peter Kennedy.  A Guide to Econometrics. (2nd or later edition), Cambridge, Mass.:  MIT Press, 1985.  


            Thomas L. Wyrick.  The Economist's Handbook:  A Research and Writing Guide.  St. Paul, Minn.:  West Publishing Co., 1994.  


            SAS Institute Inc.  SAS Language and Procedures:  Introduction, Version 6, First Edition.  Cary, NC:  SAS Institute Inc., 1990.  


            SAS Institute Inc.  SAS System for Regression, Second Edition.  Cary, NC:  SAS Institute Inc., 1991.  




Readings Schedule:

            The following tentative readings schedule divides the semester into 15 one-week blocks.  The schedule is very optimistic--the odds are high that we will have to drop a couple of chapters from the list.  If necessary, chapter 16 will be the first to go and chapter 14 will be the second to go.  (The material in chapter 16 is fully covered in macroeconomics courses, and the material in 14, while important, is less crucial than other material in the text.)


             1.  Intro., Ch. 1, Ch. 2


             2.  Ch. 3 - The Demand for Labor (including discussion of minimum wage laws) (read appendix 3A)


             3.  Review of econometrics on PC-SAS; accessing the NLS data set; research methods and clear writing.


            READ:  McCloskey, Donald.  "Economical Writing."  Economic Inquiry 23 (April 1985):  187-222.



             4.  Ch. 4 - Labor Demand Elasticities, Technological Change, and Foreign Trade


             5.  Ch. 5 - Quasi-Fixed Labor Costs and Their Effects on Demand


             6.  Ch. 6 - Supply of Labor to the Economy:  The Decision to Work (read                               appendix 6A)


             7.  Ch. 7 - Labor Supply:  Household Production, the Family, and the Life Cycle

                 Ch. 8 - Compensating Wage Differentials and Labor Markets (read                                     appendix 8A)


                        Midterm exam



             8.  Ch. 9 - Investments in Human Capital:  Education and Training (read                                   appendices 9A and 9B)


             9.  Ch. 10 - Worker Mobility:  Turnover and Migration (read appendix 10A)


            10.  Ch. 11 - Pay and Productivity


            11.  Ch. 12 - Gender, Race, and Ethnicity in the Labor Market


            12.  Ch. 13 - Unions and Collective Bargaining in the Private Sector (skip appendix 13A)

                 Ch. 14 - Public Sector Labor Markets (skip appendix 14A)


            13.  Ch. 15 - Inequality in Earnings (read appendix 15a)

                 Ch. 16 - Unemployment (skip appendix 16A)


            14.  Student presentations


            15.  Review for final




            Students are strongly encouraged to work on the problems and multiple choice questions in the Study Guide, but these will not be turned in or graded.  When the lectures for a chapter have been completed, time will be allowed for students to ask about questions that they are having trouble with.  A few of the questions from the Study Guide will appear on the tests (sometimes in modified form to fit the multiple choice format).



            Grades will be based on:  a 15 minute presentation of preliminary results on the NLS project (20 points), a midterm (60 points), a comprehensive final (100 points), a 13 page paper using the NLS to answer a question in labor economics (100 points).  Both the midterm and the final will consist in part of questions similar to those assigned from the book and the study guide.  Questions frequently involve the analysis of graphs and algebra and may occasionally require knowledge of basic differential calculus.  Programmable calculators, including the Hewlett-Packard Business Consultant II model, may be used during exams, but their memory will be cleared by the instructor prior to the start of the exam.


Point Deductions for Late Paper:

            For the paper assignment, 5 points will be deducted for the first day late and 1 point will be deducted for each additional day late.



            Exams will be attentively monitored.  The result of academic dishonesty will be a grade of F for the seminar and a recommendation for expulsion from the university.


Course Grade Reporting:


In a memo dated January 21, 1991, Vice Chancellor Otto Bauer has advised faculty that posting grades may be a violation of the "Privacy Rights of Parents and Students Act" of 1974.  Grades will not be posted.  A student who needs early reporting of the course grade at the end of the semester should provide the instructor with a stamped, self-addressed envelope.



Important Dates:


            March 2:  tentative date for midterm

            March 19-26:  Spring vacation; no classes.

            March 31:  Last day until 4:00 p.m. to drop course with a grade of "W"; or change course to "audit".

            April 27:  Paper due date.

            May 4:  Final exam 6:00-8:40 pm.






(This section will be filled completed in a future edition of the syllabus.)



                          13 Page Paper Using the National Longitudinal Surveys (NLS)




The paper should be written to be understandable by a first year economics graduate student.  That means you should assume that you are writing for someone who is intelligent, interested, short on time, and does not have a deep knowledge of econometric techniques beyond basic regression analysis.  In practice, this implies that you should very briefly explain the meaning of the statistical techniques and tests that you use beyond the most basic ones. 


The "core" sections of the paper should sum to approximately 13-16 pages.  Papers in which the core sections exceed 16 pages will be penalized, unless the extra pages are truly needed (i.e., not "padding").  On the other end, it is nearly impossible to do a good job on the paper in less than 11 pages. 


The paper should have a title page including the title, your name, the name of the course, the course number and the date on which the paper is turned in.  Following the title page, on a separate page should be a 100 word abstract.  Neither the title page nor the abstract page counts toward the 13-16 page limit of the paper.  Pages in the body of the text (beginning with the Introduction) should be consecutively numbered.  The text of the paper should be double-spaced and should have inch and a half margins on the top, bottom, right and left.  The font size should be 10 point (traditionally called "pica," which is the larger of the two typefaces traditionally used on typewriters).  Please do not put your paper in a special binder that would limit the space for my comments.  Instead, staple it in the upper left-hand corner, or clip it together. 


Endnotes should be used only if really necessary.  When a source is used in the text, internal references should be given in the text that consist of a parenthetical mention of the author's last name, the date of the publication and a specific page number (if appropriate).  For each brief parenthetical reference, a complete bibliographic reference should be provided in the Bibliography (that is not counted in the 13-16 page limit).  The ultimate arbiter for reference format is the latest edition of The Chicago Manual of Style.


Before you turn in your final draft, be sure to submit it to a spell-checking program (available with all major word-processing programs).  Do not forget that "data" is the plural of "datum," and so should appear, e.g., as "data are" rather than "data is."  Spell out acronyms the first time they are used.


As you work on the text of your paper, be sure to periodically backup your latest draft onto a floppy disk.  Also be sure to keep a clean copy of your paper for your files---the copy you turn in to me will be marked-up in red ink.


You should append to the paper (not included in the 13-16 page limit) print outs of your SAS program, the SAS output that provides you the descriptive statistics of your data set, and the output from the regressions that you estimate.  If you estimated a large number of regressions, you do not need to attach print-outs of all of them.  But at a bare minimum, you should attach the print-outs for your "best" regression and for the regression used to obtain your elasticities (if they are not the same regression).


The following is a rough outline of the "core" sections of the paper (those that count in the 13-16 page "limit") and the (very approximate) number of pages that should be devoted to each.


1.         Introduction (1-3 pages).  Describe the issue that you are investigating.  What, if any, are the policy implications?  What are the findings of previous researchers?  This would be a good place to mention the contents of any broadly relevant articles that you turned up in your Econlit, UNCOVER and ABI-Inform searches on your topic.  If there is no relevant material on UNCOVER or ABI-Inform for your topic, mention the absence of relevant material (and include in the appendix, a list of the keywords that you used to search under and the articles that resulted).  (That is, it would be very unusual to have no relevant articles appear, so if you claim that there are none, the burden of proof is on you.)  Briefly (in a few sentences) summarize what you will be doing in the rest of the paper. 


2.         Data (2-3 pages).  Describe the variables that you think should be included in your regression functions.  Describe those for which you were able to obtain information.   Describe and reference the sources for each of your variable series, if they do NOT come from the NLS.  Describe any special techniques that you had to use to make variables comparable (such as interpolating values into annual data to make the variables comparable to quarterly data).  Describe the technique used to transform variables reported in nominal dollars into variables reported in real dollars.  State the base year for comparison of the real dollars and the price index used for the transformation (e.g., most commonly the Consumer Price Index is used).  This could be stated in terms such as:  "all money variables are reported in constant 19?? dollars" (where you fill in the ?? as appropriate).  Be sure to provide a reference for your source for the CPI (or other index used).


3.         Table of Descriptive Statistics (1 page).  For each basic variable used in the analysis (i.e., not the log transformations), include information on:  the number of observations for which data was available, the mean value and the standard deviation.  If you have space you may also include other relevant statistics, such as the minimum and maximum values.  For most variables three significant digits to the right of the decimal point is sufficient.  More digits usually add clutter without adding useful information.  Descriptive statistics should be estimated only for observations included in the regressions.


4.         Regressions (2-3 pages).  Describe the regressions you estimated.  Discuss which ones have the best "goodness of fit" as measured by the adjusted R-square.  For robustness, you should normally estimate at least a couple of functional forms, most commonly the linear and the log-linear forms.  The semi-log form is also often worth trying.  


5.         Table of Regressions (1-2 pages).  Somewhere in the title or the body of the table or a footnote, indicate what the dependent variable was.  For each independent variable used in the regressions report the coefficient value and the t-statistic for the regression.  (Please note:  the "t" in t-statistic is not capitalized.)  It is redundant and wastes space to include both standard errors and t-statistics.  Usually t-statistics are preferred.  Put an asterisk or star (*) next to every coefficient value that is significant at the .05 level (unless they are all significant, in which case the asterisks would not be useful).  The convention is to put the star to the right of the coefficient number (not next to the coefficient name or the t-statistic.)  For each regression report the number of observations, the R square and the adjusted R square.  For most coefficient estimates, t-statistics, and R-square statistics, three significant digits to the right of the decimal point is sufficient.  More digits usually add clutter without adding useful information.


6.         Policy conclusions and future work (1-2 pages).  Discuss what different data or techniques you would apply to the problem if you had additional time.  Discuss what you have learned (and your level of confidence in what you have learned).  Discuss any relevant policy conclusions from your paper.    


                                       SOME CRITERIA FOR GRADING PAPERS


Some Criteria


Follow Directions

Spacing, fonts, bibliography format, inclusion of raw regressions in appendix, etc.



Follows McCloskey



Evidence of literature search, . . .


Data Section

Reasonable variables, use of CPI, . . .


Data Table

Clear and informative


Data Analysis; Regressions

Uses appropriate techniques; reports clearly


Regression Table

Reports those requested; clear presentation


Conclusions and Future Work

Clear, appropriate.



Difficulty and importance






The following sample bibliography is intended to illustrate the preferred reference format for several different kinds of publications. 




Ackermann, Robert John.  Data, Instruments and Theory.  Princeton, N.J.:  Princeton University Press, 1985.


Acs, Zoltan J., David B. Audretsch and Maryann P. Feldman.  "Real Effects of Academic Research:  Comment."  The American Economic Review 82, no. 1 (March 1992):  363-367.


Adams, James D.  "Fundamental Stocks of Knowledge and Productivity Growth."  Journal of Political Economy 98, no. 4 (August 1990):  673-702.


Alchian, Armen A.  "Private Property and the Relative Cost of Tenure."  In Economic Forces at Work.  Indianapolis:  Liberty Press, 1977 (essay originally published in 1958).


American Chemical Society.  Directory of Graduate Research 1989.  Washington, D.C.:  American Chemical Society, 1989.


Becker, William E., Jr.  "Maintaining Faculty Vitality through Collective Bargaining."    In Clark, Shirley M. and Lewis, Darrell R., eds. Faculty Vitality and Institutional Productivity:  Critical Perspectives for Higher Education.  New York:  Columbia University, Teachers College Press, 1985, pp. 198-223.


Davis, Elizabeth.  "Compensation Gains of Faculty Unions and the Unemployment Effect of Extending Unemployment Insurance Coverage."  (PhD Dissertation,  University of Michigan, 1988).


Diamond, Arthur M., Jr.  "Age and the Acceptance of Cliometrics." The Journal of Economic History 40, no. 4 (December 1980):  838-841.


       .  "The Career Consequences of Accepting a Scientific Mistake."  working paper, 1992.


Friedman, Milton.  "An Open Letter for Grants."  Newsweek  (May 18, 1981):  99.


Garfield, Eugene, (chairman).  Science Citation Index.  Philadelphia:  Institute for Scientific Information, Inc., 1961-present.


Lucas, Robert E., Jr.  "Incentives for Ideas."  New York Times (April 13, 1981):  23.

                                         SUPPPLEMENTARY SOURCES OF DATA




Macroeconomic Data and Analysis


Each of the Federal Reserve banks publishes a (usually quarterly) Review which they mail free of charge to anyone who requests it.  They also have free statistical reports on various subjects that they are willing to send out free upon request.



Three of the best* of the Reviews are:



Federal Reserve Bank of Minneapolis Quarterly Review


            Research Department

            Federal Reserve Bank

            Minneapolis, MN  55480


            phone:  (612) 340-2341



Federal Reserve Bank of New York Quarterly Review


            Federal Reserve Bank of New York

            Public Information Department

            33 Liberty St.

            New York, NY  10045


            phone:  (212) 720-6134



Federal Reserve Bank of St. Louis Review


            Research and Public Information Department

            Federal Reserve Bank of St. Louis

            P.O. Box 442

            St. Louis, MO  63166





Since Nebraska is in the Tenth Federal Reserve District overseen by the Federal Reserve Bank of Kansas City, the Review with the most information on our area is the bimonthly:


Federal Reserve Bank of Kansas City Economic Review


            Public Affairs Department

            Federal Reserve Bank of Kansas City

            925 Grand Avenue

            Kansas City, MO  64198-0001


            phone:  (816) 881-2000


A quarterly publication put out by the Kansas City Fed that contains a wealth of statistics on our region is:


Regional Economic Digest


            (address and phone number same as for Economic Review)




Complete addresses and phone numbers for all of the Federal Reserve sources of information follow:






KANSAS CITY, MO 64198-0001


Or you may call.  The local number is 816-881-2683 and the toll-free number, which can be used anywhere in the continental United States, is 800-FED-1010, Ext. 2683.


Addresses/Telephone Numbers


Board of

Governors                  Publications Services, MS-138

                                    Washington, D.C. 20551



FRB of:


Boston                        Public and Community Affairs, T-6

                                    P.O. Box 2076

                                    Boston, MA 02106-2076



New York                   Public Information

                                    33 Liberty Street

                                    New York, NY 10045



Philadelphia                Public Information

                                    P.O. Box 66

                                    Philadelphia, PA 19105-0066



Cleveland                   Public Affairs and Bank Relations

                                    P.O. Box 6387

                                    Cleveland, OH 44101-1387

                                    216-579-2157 or 3079


Richmond                   Public Affairs

                                    P.O. Box 27622

                                    Richmond, VA 23261



Atlanta                        Public Affairs

                                    104 Marietta Street NW

                                    Atlanta, GA 30303



Chicago                       Public Information Center

                                    P.O. Box 834

                                    Chicago, IL 60690-0834



St. Louis                     Research and Public Information

                                    P.O. Box 442

                                    St. Louis, MO 63166

                                    314-444-8444, Ext. 8808 or 8809


Minneapolis               Public Affairs

                                    P.O. Box 291

                                    Minneapolis, MN 55480-0291



Dallas             Public Affairs

                                    Station K

                                    Dallas, TX 75222



San Francisco Public Information

                                    P.O. Box 7702

                                    San Francisco, CA 94120-7702





Another useful (and free) source for economic data and analysis for the state of Nebraska is the monthly or bi-monthly:


Business in Nebraska


            Bureau of Business Research

            200 CBA

            University of Nebraska at Lincoln

            Lincoln, NE  68588-0406






Labor Data


U.S. Department of Labor, Bureau of Labor Statistics.  Handbook of Labor Statistics.




General Data


U.S. Bureau of the Census.  Statistical Abstract of the United States:  1989.  (109th edition.)  Washington, DC, 1989.





The phone number for the nearest branch of the U.S. Government Printing Office bookstore (in Kansas City) is:  (816) 765-2256.




* The top three bank reviews (ranked by number of citations in 1987-88 from the 3 major macro journals) are:  Minnesota, St. Louis and Kansas City.  Tied for 4th and 5th are Richmond and San Francisco.  (See:  Jansen, Dennis W.  "Ranking Federal Reserve System Research Departments by Publications in Professional Journals."  Journal of Macroeconomics 13, no. 4 (Fall 1991):  733-742.)



                                                        RANDOM TIPS ON SAS

                                                        (last revised:  March 9, 1995)






Some potentially useful PROCs:



Descriptive Statistics:


            PROC MEANS;


            Provides descriptive statistics (mean, standard deviation, minimum, maximum) for all your data.  This is probably the procedure that you should always run first, in order to see if you have read your variables correctly off of the dataset.  It is also useful to see if your independent variables are varying sufficiently to permit precise estimation of their effect on the dependent variable.  It is also useful to see if your dataset is a represtentative sample of your intended population (e.g., if the mean for "years schooling" is 18, then your sample is probably not representative of the US population as a whole).




            PROC PRINT;


            Provides a complete print of all your data.  If your data set is large, this would use up a lot of paper.  Sometimes, as a diagnostic, you might want to print and look at just the first 100 observations.  You could do this by putting the following statement after your input statement. 


                        if _n_ < 101;



            PROC CORR;


            This procedure estimates pair-wise correlations for all variables in your data set. population).  This could be useful, for instance, as a first pass at seeing whether you might have a multi-colinearity problem with any of your proposed independent variables.



            PROC FREQ;


            This procedure provides you a frequency table for all your variables.  (Which might be useful, for instance, in performing a chi-square test on the hypothesis that two samples come from the same population).





Econometric Modelling:



            PROC REG SIMPLE;


            The "simple" option is useful in obtaining the descriptive statistics associated with the data used in a particular regression.  The objective is to make sure that missing data for some variables does not result in a sample that is unrepresentative of the population in ways that bias the resulting estimates.




            PROC AUTOREG;


            This procedure corrects for autocorrelation in time-series data.



            PROC LIFEREG;


            This procedure provides a method for estimating "hazard functions" where the dependent variable is the number of periods before some event occurs (e.g., death, unemployment, retirement, marriage, divorce). 


                        example for running a lifereg:


                        proc lifereg;

                                    model r=x1 x2 x3;




            PROC LOGISTIC;


            This procedure provides a variety of methods for handling categorical dependent variables usine a logit model.  If OLS regression is used when the dependent variable takes on categorical (like:  zero or one) values, then the coefficient estimates will be biased.


                        example for running a logit:


                        proc logistic;

                                    model r=x1 x2 x3;





If you include discrete categorical independent variables (apart from a zero-one dummy variable) you need to remember to exclude one of the categories.  Otherwise you would have a problem of exact multicolinearity.  To illustrate, consider the following example where C is the constant and where all observations are categorized by a three-way geographical division: N for north, S for south and W for west.




























So, any variable can be expressed as an exact function of the constant and the other variables (e.g., W=C-S-N) which would be perfect multicolinearity.



If you estimat log-linear functional forms, you have to be careful that none of the values of your variables is equal to zero (since you can't take the log of zero).  The convention among many applied econometricians when they run into this problem is to transform the zero values into very small positive numbers.  For the variable x, this could be accomplished with the following statement following the input statement:


                        if x=0 then x=.000001;




In the NLS CD-ROM data sets, missing values are indicated by values such as -99 or -999 (depending on the number of columns that the variable is allocated).  Unless told otherwise, SAS assumes that a missing value is indicated by a period.  Unless told otherwise SAS will interpret any number literally.  In order for SAS to read these values as missing, you will need to include statements such as the following after your input statement:


                        if x=-999 then x=.;

                        if y=-99 then y=.;



Sometimes you may want to create a dummy variable out of a discrete variable.  For example, you may have data on hours worked (H) and want to create a zero-one dummy variable (WORK) to incidate whether the person worked or not.  Then you could create such a variable with the following statements:


                        if H>0 then WORK=1;

                        if H=0 then WORK=0;




Some of you will need to combine datasets (as, e.g., when your final data set will combine observations from more than one of the four cohorts).  The following example sketchs one way to do this.  In the example the four original data sets are YM, YW, OM, OW (for Young Men, etc.)  [I am typing the following from memory, so if it doesn't work, bring it to me and we will check if it needs minor modification.]


                        data a;

                                    infile 'a:ym' delimiter=',';

                                    input x y z;


                        data b;

                                    infile 'a:yw' delimiter=',';

                                    input x y z;


                        data c;

                                    infile 'a:om' delimiter=',';

                                    input x y z;


                        data d;

                                    infile 'a:ow' delimiter=',';

                                    input x y z;


                        data e;

                                    set a b c d;


                        data f;

                                    set e;

                                    file 'a:\all';