劳动力供给的长期弹性研究

1、劳动力供给的长期弹性研究Orley C. AshenfelterKirk B. DoranBruce Schaller2Orley C. Ashenfelter劳动经济学手册第1、2卷主编AER主编，美国劳动经济学协会创始人，美国法律与经济评论创始人之一。早期研究关注于工会的经济分析。工会中的种族歧视。七十年代，关于劳动供给的研究。与James Heckman一起将新古典理论拓展到家庭决策，“单一家庭模型”（unitary household model)。1972年任美国劳工部教育局局长。开始关注于培训项目有效性的衡量问题。1978年发表的一篇论文，提出”difference in diff

2、erence”方法。这些方法被广泛用来评估“自然实验”的有效性。经济学的“实验研究”，收集双胞台数据研究教育回报问题。Orley C. Ashenfelter, Kirk B. Doran, Bruce Schaller, 2009. A Shred of Credible Evidence on the Long Run Elasticity of Labor Supply. WORKING PAPER #551 INDUSTRIAL RELATIONS SECTION PRINCETON UNIVERSITY September 2009 Working Paper 15746. 2010

3、 Orley C. Ashenfelter, Kirk B. Doran, Bruce Schaller, 2010. A Shred of Credible Evidence on the Long Run Elasticity of Labor Supply.Economica Volume 77, Issue 308, pages 637650, October 2010ECON 650 Labor Economics I Purdue University Fall 2013 “Labor Supply Elasticity Estimation”34问题的提出纽约出租车市场基本描述数

4、据分析结论及局限性5 U (y, ) ，U1 0; U2 0 h + = T ， y = wh + G U(wh + G, T h) 678Effects of nonlabor e on labor supply Effects of the hourly wage rate on labor supply substitution effect e effect 9税收和收入分配等公共政策都与有关劳动力供给的工资率长期效应假说有着密切关系。Prescott (2004) ，Rogerson and Sargent等宏观经济学家们认为欧洲较高的收入税使得欧洲人普遍比美国人工作时间短。其中重要

5、的假设在于税收收入以按人补贴的方式返还给家庭，从而起到工资削减补偿的作用。多数研究表明，持久工资上升对于男性劳动力供给的影响是有限的。工资对劳动力供给的影响与工作类型或者工资计算类型相关。工资是外生的，且变化是持久的。10劳动力供给的收入效应与闲暇效应。现实中，究竟何种效应更重要。研究对象的限制。1）大多数人无法改变自己的工作时间，除非换工作；2）难以衡量收入的变化是否为外生。出租车司机可以根据收入状况自主决定工作时间 (Farber, 2005)。研究时期内正好经历了两次持久外生的收入增加。根据对纽约出租车数据的研究，长期来看，收入效应更加显著。11数据表明，劳动对于的费结构外生增加的反应是

6、有限的。而且是负面的。但是司机的收入却显著地受到的费结构的影响。研究劳动供给长期弹性的一个限制条件：外生，持久的工资变化。纽约出租车运营价格受TLC掌控。自1952年7月以来，已经调整了13次。最近的一次，调整的结果是使每小时的总收入增加。12The London VersionBut the modern one, not the original1213The Yellow TaxiThe Modern One, not the Clasic814151617Taxi Fares Are Regulated by the TLCA Meter Drop FeeFixed Per RideA

7、 Mileage ChargeOther Charges1819数据来源说明。TLC一年三次的汽车检验数据。94.905.12上次检验以来的时间、检验的时间、行驶里程以及收入。劳动供给由行驶里程来衡量。（未用工作时间衡量，两者的差异在于等待时间，这时候时间与里程并不一致，绝大多数情形下，工作时间和行驶里程密切相关）计算每英里距离挣得的实际收入。一辆出租车视同一个劳动者（剔除日夜轮班倒车辆）。只有3%的Medallion易手。20New York City Taxi Medallion21The Standard Model of Labor Supplyu=u(h,y),where h is

8、hours worked and y is e from driving, and uh 0. Work and e are related byy=g(h;), represents the parametric part of the fare structure. The driver optimizes by working at a point where- uh / uy = gh ,the rate of substitution of leisure for goods equals the marginal effect of hours on e.22A Structura

9、l Interpretation of Labor SupplyA convenient parameterization is - uh / uy = h g(h;)= h is a first order approximation for the earnings function, is measured as revenue per mile driven. 1Labor supply function is:ln(h)=(1/)ln ln(/)1 This functional form was first apparently used by Burtless and Hausm

10、an (1978) and the implied utility function and other aspects of it are discussed extensively by Stern (1986).23Market and Individual SupplyA fare structure, driver behavior, and a demand function determine equilibrium in the aggregate market including:Cab availabilityCab useageDriver revenue per mil

11、eWe assume each driver faces a perfectly elastic demand for hours at the equilibrium revenue per milea standard assumptionwe measure individual level labor supply, ie, work effort24Pre-Post Differences in Miles Driven and Revenue/MIleChange in Revenue per mileChange in Miles DrivenLabor Supply Elast

12、icity1996 Fare Increase +$0.140 (17%) -477 (-3.%)-3/17=-.172004 Fare Increase +$0.15 (19%) - 824 (-5.6%) 5.6/19=-.2925简单计算劳动供给弹性：-5%/20%=-.25 26Figure 1: How March 1996 fare change affected real revenue/mile and miles driven27Figure 2: How May 2004 fare change affected real revenue/mile and miles dr

13、iven28Change in Revenue per MileChange in Miles Driven 1996 Fare Increase+ $0.14 (+ 17 %)- 477 miles (- 3.2 %)2004 Fare Increase+ $0.15 (+ 19 %)- 824 miles (- 5.6 %)Table 2: Simple Difference Table:(Medallion Fixed Effects; no other controls)All changes are computed as the coefficient of a dummy var

14、iable indicating the year noted and are significant at the 0.1% level. Revenue is in December 2005 Dollars. Miles driven measures the number of miles driven since the last inspection. The average number of days between inspections is 122 (4 months) with a standard deviation of 4 days. Since the pane

15、l is not fully balanced, these results are computed from a regression that includes medallion fixed effects in order to use all the data.Change in Revenue per MileChange in Miles Driven1996 Fare Increase+ $0.14 (+ 17 %)- 399 miles (- 2.7 %)2004 Fare Increase+ $0.15 (+ 19 %)- 818 miles (- 5.6 %)Diffe

16、rence Table:(Medallion Fixed Effects; controls for month and days since last inspection)All changes are computed as the coefficient of a dummy variable indicating the year noted and are significant at the 0.1% level. Revenue is in December 2005 Dollars. Miles driven measures the number of miles driv

17、en since the last inspection. The average number of days between inspections is 122 (4 months) with a standard deviation of 4 days. Since the panel is not fully balanced, these results are computed from a regression that includes medallion fixed effects in order to use all the data. The regressions

18、in this table also contain a variable measuring the number of days since the taxi was last inspected.29VariableMeanStd. Dev.MinMaxObsowner-driver?(1 = yes, 0 = no)0.490.5001102275Days since the last inspection122 days4 days40 days237 days102275Miles driven since the last inspection15989 miles6163 mi

19、les4000 miles41997 miles102275Miles driver per day131 miles per day50 miles per day24 miles per day408 miles per day102275Revenue earned since the last inspection$21,597$8,465$,3007$68,53667317Revenue earned per day$177 per day$69 per day$25 per day$553 per day67317Revenue earned per mile (a measure

20、 of the wage)$0.68 per mile$0.14 per mile$0.33 per mile$1.41 per mile65888Real revenue earned per mile (in Decemebr 2005 Dollars)$0.81 per mile$0.14 per mile$0.44 per mile$1.56 per mile49112Table 1: Simple StatisticsSimple Statistics: by inspectionVariableMeanStd. Dev.MinMaxObsowner-driver?(1 = yes,

21、 0 = no)0.400.38014658Number of Inspections22121464658Simple Statistics: by medallion30(1)(2)OLSFixed EffectsPost fare increase*0.190.19(0.00)(0.00)ln(days since inspection)-0.02-0.01(0.05)(0.03)February-0.010.01(0.01)(0.04)March-0.00-0.03(0.01)(0.04)April-0.02-0.06(0.01)(0.04)May0.000.00(0.01)(0.01

22、)June-0.010.00(0.01)(0.04)July-0.01-0.05(0.01)(0.04)August-0.03-0.07(0.01)(0.04)September-0.02-0.03(0.01)(0.00)October-0.04-0.02(0.01)(0.04)November-0.02-0.06(0.01)(0.04)December-0.03-0.07(0.01)(0.04)Constant-0.18-0.21(0.24)(0.15)Observations1228112281R-squared0.240.53# of Medallions2514 Table 3: Re

23、venue per Mile as a Function of the Fare Changes(First stages of specifications (2) and (4) in Table 3)Standard errors in parentheses Unit of Observation: One Driver during a 4 month periodFixed Effects: Medallion LevelPost fare increase = 0 for inspections that take place during the 365 days before

24、 each fare change was implemented. Post fare increase = 1 for inspections that take place during the 365 days beginning four months after each fare change was implemented31(1)(2)OLSFixed EffectsPost fare increase*-0.0239*-0.0423*(0.00618)(0.00374)ln(days since inspection)0.633*0.792*(0.102)(0.0643)F

25、ebruary-0.0007460.0318(0.0168)(0.0906)March-0.0293*-0.00251(0.0154)(0.0888)April-0.00813-0.0492(0.0156)(0.0888)May0.01510.0219*(0.0180)(0.0106)June0.0426*0.0608(0.0167)(0.0904)July-0.01540.0177(0.0156)(0.0888)August-0.0107-0.0393(0.0154)(0.0888)September-0.0231-0.0135(0.0173)(0.0100)October-0.01250.

26、0363(0.0171)(0.0908)November-0.0300*-0.00338(0.0154)(0.0888)December-0.00118-0.0432(0.0154)(0.0888)Constant6.489*5.726*(0.492)(0.318)Observations1228112281R-squared0.0070.033# of Medallions2514 Table 4: Miles Driven as a Function of the Fare ChangesStandard errors in parenthesesUnit of Observation:O

27、ne Driver during a 4 month periodFixed Effects:Medallion LevelPost fare increase = 0 for inspections that take place during the 365 days before each fare change was implemented. Post fare increase = 1 for inspections that take place during the 365 days beginning four months after each fare change wa

28、s implemented32Ordinary Least Squares and IV Estimates of the Labor Supply ElasticityOrdinay Least SquaresIV/No Fixed EffectsOLS/WFixed EffectsIV/WithFixed Effects-0.42-0.13-0.40-0.2333(1)(2)(3)(4)OLSOLS IVFixed EffectsFixed Effects IVln(real revenue/mile)-0.42-0.40-0.12-0.23(0.01)(0.01)(0.03)(0.02)ln(days since inspection)0.630.720.630.79(0.06)(0.04)(0.10)(0.06)February0.000.02-0.000.04(0.01)(0.02)(0.02)(0.09)March-0.03-0.00-0.03-0.01(0.01)(0.02