Labor economics B Hokkaido University Fall Yukiko Abe Hokkaido University, Japan

Transcription

1 Labor economics B Hokkaido University Fall 2016 Yukiko Abe Hokkaido University, Japan 1

2 [1] On research, writing, & publishing (available in the internet) Dixit, A. My system of work (not!) Passion and Craft: Economists at Work, ed. Michael Szenberg, University of Michigan Press, Fullerton, D. General & Specific guidelines for writing research papers Moffitt, R. Getting published in economics journals, CSWEP newsletter, Spring tr_sprsum_2011.pdf 2

3 [2] On research, writing, & publishing (available in journals) Hamermesh, D. (1992) "The Young Economist's Guide to Professional Etiquette," Journal of Economic Perspectives, v. 6, iss. 1, pp Thomson W. (1999) The Young Person s Guide to Writing Economic Theory, Journal of Economic Literature, Vol. 37, pp , sections 1 & 2 Hamermesh, D. (2000) The Craft of Labormetrics, Industrial and Labor Relations Review, Vol. 53, No. 3., pp

4 [3] On research, writing, & publishing (available in books) Goldin, C. The Economist as Detective Passion and Craft: Economists at Work, ed. Michael Szenberg, University of Michigan Press, 1998 McCloskey, D. Economical Writing, Second Edition, Waveland Press,

5 Things to note Paragraphs and topic sentences Introduction Detailed rules of writing Rewriting! Read aloud 5

6 Those who will be attending the class Write an to the stating that you will be attending the class, by October 7 th, 13 pm. Obtain Blundell & MaCurdy (1999). Also, learn how to using Econlit & downloading journal papers from the library s website. Obtain materials in [1]-[3] above, to the extent you can. Yukiko Abe 2014 All rights reserved 6

7 Plan of the lecture Labor force participation: facts & the longterm trends Labor supply: static model Labor supply: dynamic model If there is extra time, we discuss other topics such as wage determination & labor supply 7

8 Useful links Raj Chetty s lecture notes & video (can be seen in YouTube) The part closely related to this lecture is Topic 5, Part 3 Emmanuel Saez s lecture notes Labor Supply Responses to Taxes and Transfers They are from public economics courses, but contain very useful discussions on labor supply and taxes. 8

9 Preparation 1: Elasticities Growth rates are convenient measures since they are unit-free: for instance, if price changes are measured in growth rates, they are comparable across goods, across countries, or across time (the CPI, energy prices, commodity prices). GDP growth rates are comparable across countries. Elasticity is defined as the ratio of the growth rates. In understanding growth rates, it is useful to know the natural log function. 9

10 The natural log function (1) y = ln( x) y x ln(x) line with slope=1 10

11 The natural log function (2) The difference in ln(x) is approximately equal to the growth rate, as follows: ln( x ) ln( x ) = ln[1 + ( x x ) / x ] x/ x where = Differentiating x x1 x0 ln( x) ln( x) 1 = x x This part is proved here by x, we have The natural log function is the inverse function of the exponential function. 11

12 What is the exponential function? Definition: the exponential function e z = z n= 0 n! n Addition formula: Noting that Derivative: 0! = 1 e z z = z+ w z w e = ee z z, we have ee = e z 1 12

13 The natural log function (3) The derivative of the natural log equals 1 at x =1. In the figure, the slope of the ln(x) at x=1 is 1. Noting that ln(1)=0, the straight line that is tangent to ln(x) at (1,0) is y = x 1. Yukiko Abe 2014 All rights reserved

14 The natural log function (4) The straight line that is tangent to ln(x) at (1,0) is y = x The linear line that is tangent to ln(x) approximates ln(x), i.e., 1 ln(1 + x) x. Thus we have ln[1 + ( x x ) / x ] x/ x Yukiko Abe 2014 All rights reserved

15 Labor supply: Usual assumptions W is exogenous and H is endogenous; wages are market wages people take as given, and hours are the choice variable. Labor supply model predicts H from (exogenous) wages & other variables: H = f( wx, ) This can happen at both participation (extensive) and hours (intensive) margins. 15

16 Labor supply decision Labor force participants = workers +unemployed workers: who work for pay unemployed: who does not work but looks for employment Non-participants (Out of Labor Force) who do not work for pay and who do not search jobs (students, retirees, housewives) 16

17 Static labor supply equation We try to estimate a static labor supply as a function of economic factors (wage, income, demographics, etc.) ln( H ) = β + β ln( W ) + β ln( Income) + β X + ε H: hours; W: wage rate; Income: household income or own income; X: other factors; ε: error term 17

18 (a) Division bias Many workers are NOT paid by hour. So what we observe in data are usually earnings & hours, instead of hourly wages. Then, we impute hourly wage by W=E/H. When H is measured with error, W is going to carry over that error. Then we have a problem in the regression ln( H) = β + β ln( W) + ε

19 (a) Division bias (cont.) This part follows Chetty s slides (albeit different notations). ln( H) = ln( H *) + µ where H * is the true hours. This means multiplicative measurement error. * * lnw = ln E ln H = ln E ln H µ = lnw µ W * : true wage 19

20 (a) Division bias (cont.) Then, using the standard regression argument (for the case of univariate regressor), we have * * cov(ln H + µ,ln W µ ) plim β2 =. * var(ln W ) + var( µ ) This means that the estimated coefficient of β 2 has a downward bias, because of the negative correlation in the numerator, and the positive variance term of μ in the denominator. 20

21 Attenuation bias Attenuation bias in measurement error (see, e.g., Greene pp , Wooldridge pp ) This refers to the term in denominator, i.e., var(ln W * ) + var( µ ) This is the iron law of econometrics (Deaton, The Analysis of Household Surveys, p.99) Note that the problem here is even worse, with the term in the numerator. 21

22 (b) Selection (unobserved wages) Non-workers have earnings zero, because H=0. But non-workers have latent wages (i.e., the wages they would have received had they worked), but we do not observe them in the data. Therefore, we somehow need to make assumptions to get them. This idea was pioneered by Heckman. Specifically: (1) Make distributional assumptions (Heckman); (2) do some kind of matching (use observed wages of similar people). 22

23 Examples of matching Juhn (1992, QJE), Juhn & Murphy (1997, Journal of Labor Econ), and Devereux (2004, JHR) In these papers, non-workers wages are set to the wages of workers who are likely to have weak attachment to the labor force (i.e., those who work less than 13 weeks per year). 23

24 (c) Unobserved heterogeneity In a cross section, we have different people who has different propensity to work. The differences in wages (which allow us to identify the β 2 parameter) may be the reflection of heterogeneity, not the exogenous changes in wages. The problem is, the differences in wages that come from heterogeneity in taste is quite likely to be correlated with the error term, ε. 24