Time Use and the Macroeconomy. April 2015

Transcription

1 Tme Use and the Macroeconomy Mark Aguar Prnceton Unversty Erk Hurst Unversty of Chcago Aprl 2015

2 Broad Overvew Why s the allocaton of tme mportant to macroeconomsts? o Market work central to most busness cycle and growth models. o Shocks to productvty to the market sector a focal pont n busness cycle and growth models. o Havng a home sector can potentally make market labor supply more elastc. o Shocks to the home sector can also generate tme seres (and busness cycle?) varaton n market work. o Alternate uses of tme are essental to understandng both lfecycle and busness cycle patterns n expendtures. o Many macro lteratures focus on other uses of household tme: shoppng behavor, job search, human captal nvestment, lesure, etc.

3 Handbook Chapter Goals Express a smple model to llustrate many of the key features of the nteracton between tme use and expendtures. Show some stylzed facts about the tme seres, lfe cycle and busness cycle patterns n key tme use categores: market work, chld care, home producton, and lesure. Throughout, hghlght the mportance of tme use n recent macro lterature.

4 Data - Part 1: Tme Seres Patterns n Market Work

5 Key Outstandng Queston What explans the tme seres patterns n market work for men and women of dfferent skll? Changes n partcpaton rates a key puzzle n recent years. We stll do not have a good grasp on changes n hours worked among subgroups over the past 50 years. Part of our goal s to set the stage for future research wth tme seres trends n market work.

6 Settng the Stage: Trends n Market Work Sample: CPS Data, Men, Age 21-75

7 Settng the Stage: Trends n Market Work Sample: CPS Data, Women, Age 21-75

8 Settng the Stage: Trends n Market Work Sample: CPS Data, Men by Skll, Age 21-75

9 Settng the Stage: Trends n Market Work Sample: CPS Data, Women by Skll, Age 21-75

10 Settng the Stage: Trends n Market Work (Hours Per Week) Sample: CPS Data, Hgher Sklled Men by Age

11 Settng the Stage: Trends n Market Work (Hours Per Week) Sample: CPS Data, Lower Sklled Men by Age

12 Settng the Stage: Trends n Market Work (Hours Per Week) Sample: CPS Data, Hgher Sklled Women by Age

13 Settng the Stage: Trends n Market Work (Hours Per Week) Sample: CPS Data, Lower Sklled Women by Age

14 Summary of Trends n Market Work Key Take-Aways o Trend down n market hours for men of all skll. o Increase n employment nequalty across men by skll. o The declne n employment for hgher sklled men much larger for ages 55 (untl recently) o Younger and older lower sklled men have consstently experenced declnes n market hours. o Trend up n market hours for women untl md-1990 for all skll. o Women hours are declnng over the last 15 years. o Patterns smlar by skll levels for women. o Men and women market hours are convergng

15 Summary of Trends n Market Work Some questons. o What s drvng the downward trend n male labor supply? It s not just wages as hgh sklled men experenced wage ncreases and labor supply declnes. o Why relatvely large drop for less sklled men n 1970s to early 1980s, then relatve stablty untl Great Recesson? o Why does upward trend n women stop/reverse n md-1990s? o Why has partcpaton fallen so much durng the decade for all groups (partcularly the young)? o Can tme use allocated away from the market sector help us to understand the tme seres and lfecycle patterns?

16 A Beckeran Model of Expendture: Theory and Implcatons

17 Ghez and Becker (1975); t V( a,, t) max U( C,..., C ) EV( a', ', t 1) I subject to: C F( H, X ), 1,...,I H L 1 a' (1 r) a (1 ) wl T p X L 0, a' a. (assume C.E.S., CRS) Let μ, λ, θ, and κ be the respectve multplers on the tme budget constrant, the money budget constrant, the postve hours constrant and the postve assets constrant. Assume U(.) s addtvely separable across tme and across goods. ψ= s vector of wages, commodty prces (p), taxes and transfers

18 Frst Order Condtons U F X : p, C X H U F :, C H L : w If θ = 0 (L > 0), prce of tme (n permanent ncome unts) (μ/λ = w) More generally (gven L often = 0), μ/λ = ω

19 Frst Order Condtons Intra-perod tradeoff between tme and goods: F F w H X p p (f L > 0) (1) Margnal rate of transformaton between tme and goods n producton of s equated to the relatve prce of tme.

20 A few addtonal assumptons: Frst Order Condtons o F s constant elastcty of substtuton (σ ) o p s are constant over tme Some algebra d X F F ln dln H H X (2) d ln X H d ln (3) Note: To get (3), sub (2) nto (1)

21 Statc Frst Order Condton The statc F.O.C. pns down expendture relatve to tme nputs. If we know σ and the change n the opportunty cost of tme, we should be able to pn down the relatve movement n expendtures relatve to tme. %ΔX -%ΔH =σ %Δω Notce, ths equaton does not requre us to make any assumptons about borrowng or lendng, perfect foresght, etc.

22 Home Producton vs. Lesure Work of Aguar and Hurst make dstncton based on commodty producton functons. Home Produced Goods : o Goods where σ > 1 o Examples meals ; cleanng servces ; etc. Lesure Goods : o Goods where σ < 1 o Examples entertanment goods.

23 Estmatng σ s %ΔX -%ΔH =σ %Δω Insght from ths equaton has been crtcal to lterature estmatng σ. Rupert, Rogerson, Wrght (1995) Use PSID data to estmate σ for home producton. Do ths for dfferent groups (sex*martal status). Estmates of σ for all groups (hghest for marred women). Aguar and Hurst (2007) Use data from ATUS, CEX and shoppng data (to nfer opportunty cost of tme) to estmate σ for home producton. Estmates of σ ~ 1.8. Can use ths nsght to estmate σ for other consumpton commodtes.

24 Some Further Assumptons/Defntons Constant Intertemporal Elastcty of Substtuton (good specfc): U C C U C 2 2 ( ) Share of Tme n Commodty Producton s H F H H C Separablty Across Consumpton Goods

25 Expendture Elastcty Wth Cost of Tme dln X H s d ln d 0 The elastcty of expendture on category wth respect to changes n the opportunty cost of tme can be postve, negatve or zero. If home producton tme s not mportant (s H = 0), the elastcty s zero. The elastcty ncreases as σ ncreases. If tme and goods are substtutes n the commodty producton (home produced goods), more wllng to substtute towards expendtures as tme gets expensve. The elastcty falls as γ ncreases. As the prce of tme ncreases rases the cost of consumng today (relatve to tomorrow) reducng both X and H. The extent of the substtuton s governed by γ.

26 Expendture Elastcty Wth Margnal Utlty of Wealth Dfferentate FOC for x n wth respect to ω holdng λ constant. Get: n dln c d ln d 0 o value of (hgh ncome o rather than o (retrement). Spendng should fall the most (wth declnes n the margnal wealth) for goods that have hgh elastctes of substtuton elastctes). Spendng should fall more for luxures (e.g. entertanment) necesstes (e.g. food). Some common explanatons for spendng declnes wth age

27 Predctons: Lfecycle Movements Standard Lfecycle Consumpton Models (Gournchas and Parker, 2002; Blundell et al 2001) o o Luxures (entertanment) should declne more late n lfe relatve to necesstes (food) No mportance of changng opportunty cost of tme over lfecycle Beckeran Model o o Goods for whch home producton s mportant can move over the lfecycle n ways that are dfferent than goods for whch expendture and tme are complements. If opportunty cost of tme declnes after mddle age, food may declne more than entertanment later n lfe.

28 Aguar and Hurst (2005, 2007b, and 2013) Show that home producton s an mportant reason that expendture falls after mddle age. As households leave labor force, they allocate more tme to home producton reducng expendtures on hgh σ goods and ncreasng expendtures on low σ goods. Show that food expendtures fall sharply after mddle age whle entertanment expendtures contnue ncreasng after mddle age. Use Beckeran model to explan lfecycle profles of expendture, Deaton- Paxon facts, and the retrement consumpton puzzle. Show that nsghts from the Beckeran model are mportant for estmates of the amount of rsk household face and preference parameters.

29 Non-Market Tme Elastcty Wth Cost of Tme d lnh 1 s s d ln d 0 h h The elastcty of home producton tme n category wth respect to changes n the opportunty cost of tme s unambguously negatve. As prce of tme ncreases, wll shft away from home producton to market expendtures n the producton of the good. As prce of tme ncreases, wll shft away from consumpton today vs. tomorrow. Implcaton: As opportunty cost of tme ncreases, home producton tme should fall (holdng lfetme resources fxed).

30 Market Tme Elastcty Wth Cost of Tme dlnl dln d 0 I h (1 s ) s 1 L h h Weghted average of the non-market tme elastctes. Labor supply elastcty depends on what you do wth your tme outsde of the market sector. If outsde tme s spent on hgh σ actvtes (home producton), labor supply s more elastc. Women spend more tme n home producton then men so they should have more elastc labor supply (Mncer 1962). As home producton tmes of men and women have been convergng (a fact we hghlght below), labor supply elastctes should be convergng.

31 Addtonal Model Elements Can set the model n general equlbrum Can endogenze wage movements (market productvty shocks). Can allow for shocks to the home producton and lesure sectors. Explore the busness cycle mplcatons (Benhabb, Rogerson, and Wrght (1991) ; Greenwood and Hercowtz (1991). o Generate hgher labor supply elastctes at busness cycle frequences. Explore mplcatons of nnovatons n the home producton sector (Greenwood and Seshadr (2005), Greenwood et al. (2005). o Generate ncreasng market hours and declnng home hours for women.

32 Shoppng Tme Detaled tme use data combned wth prce data has allowed researchers to estmate shoppng functons. Aguar and Hurst (2007) frst to use scanner data to estmate shoppng functon parameters (the trade off between tme nvestments and prces). Lfecycle Implcatons Aguar and Hurst (2007) Busness Cycle Implcatons Kaplan and Menzo (2014) Cobon et al (2015) Ros-Rull and Huo (2013) Nevo and Wong (2014) Consensus: Shoppng tme s an mportant margn of substtuton n explanng prces, employment, consumpton, mark-ups, etc.

33 More Data: Tme Seres, Busness Cycle and Lfe Cycle Patterns n Tme Use

34 Data : Amercan Tme Use Survey; 1985 : Amercan s Use of Tme, Amerca s Use of Tme All based on 24-hour tme dares. Market Work All tme workng on all pad jobs, tme spent commutng to work, any tme spent on work related meals and actvtes. Non-Market Work All tme n (1) core home producton (cookng, cleanng, ronng, dustng, vehcle mantenance, etc.), (2) home ownershp actvtes (pantng rooms, reparng roof, mowng lawn, gardenng), (3) shoppng and obtanng goods and servces, and (4) adult care. Chld Care All tme spent carng for, educatng, and playng wth chldren (not for pay). Lesure All tme spent on lesure actvtes (TV watchng, socalzng, gong to moves, exercsng, computer tme, eatng meals, sleepng, etc.) Other Everythng else (own health care, educaton, cvc oblgatons, etc.)

35 Lfecycle Profle of Market Work Sample: ATUS Data, Pooled , Ftted 4 th Order Polynomal

36 Lfecycle Profle of Non-Market Work Sample: ATUS Data, Pooled , Ftted 4 th Order Polynomal

37 Lfecycle Profle of Chld Care Sample: ATUS Data, Pooled , Ftted 4 th Order Polynomal

38 Lfecycle Profle of Lesure Sample: ATUS Data, Pooled , Ftted 4 th Order Polynomal

39

40

41

42

43 Can We Identfy Busness Cycle Effects n Another Way Explot cross-state varaton durng the recesson (Aguar, Hurst and Karabarbouns 2013). For each state pool data over two or three years: o 2007 and 2008 o 2009 and 2010 o 2011, 2012, and 2013 Take changes n tme use for all ndvduals wthn the state between successve pooled year pars. Regress change n tme use n state k durng perod t aganst change n market work hours n state k durng perod t. 102 state*tme pars.

44 Sample: ATUS, Ages 21-75, Change n Non Market Work, β = (0.04)

45 Sample: ATUS, Ages 21-75, Change n Lesure, β = (0.06)

46 Sample: ATUS, Ages 21-75, Change n Job Search, β = (0.014)

47 Tme Use of the Unemployed: Men More Educated Actvty Employed Unemployed NILF (age<63) NILF (Age 63) Lesure Market Work Job Search Home Producton Chld Care Other Observatons 13, ,054 Less Educated Actvty Employed Unemployed NILF (age<63) NILF (Age 63) Lesure Market Work Job Search Home Producton Chld Care Other Observatons 22,319 1,625 3,603 3,399

48 Tme Use of the Unemployed: Men More Educated Actvty Employed Unemployed NILF (age<63) NILF (Age 63) Lesure Market Work Job Search Home Producton Chld Care Other Observatons 13, ,054 Less Educated Actvty Employed Unemployed NILF (age<63) NILF (Age 63) Lesure Market Work Job Search Home Producton Chld Care Other Observatons 22,319 1,625 3,603 3,399

49 Tme Use of the Unemployed: Men More Educated Actvty Employed Unemployed NILF (age<63) NILF (Age 63) Lesure Market Work Job Search Home Producton Chld Care Other Observatons 13, ,054 Less Educated Actvty Employed Unemployed NILF (age<63) NILF (Age 63) Lesure Market Work Job Search Home Producton Chld Care Other Observatons 22,319 1,625 3,603 3,399

50 Tme Use of the Unemployed: Men More Educated Actvty Employed Unemployed NILF (age<63) NILF (Age 63) Lesure Market Work Job Search Home Producton Chld Care Other Observatons 13, ,054 Less Educated Actvty Employed Unemployed NILF (age<63) NILF (Age 63) Lesure Market Work Job Search Home Producton Chld Care Other Observatons 22,319 1,625 3,603 3,399

51 Tme Use By Unemployment Duraton We explore whether the allocaton of tme changes wth the duraton of unemployment. Hard to do wth ATUS data (need to merge to the CPS but there s a gap n measurement). Found no evdence of changes n tme use by unemployment duraton. However, power s an ssue. Would be nterestng to see f people do change the allocaton of ther tme (job search, home producton, chld care, etc.) the longer they remaned unemployed. Selecton? Duraton?

52 Conclusons/Next Steps A Beckeran model of tme use s qute powerful n explanng n lfe cycle, busness cycle, and tme seres trends n both labor supply and expendture. Wth the advent of new data, new opportuntes are openng up to explore even more questons. o o o o o o Job search Shoppng Home producton Chld care Human Captal Travel tmes It s an exctng tme to thnkng about the relatonshp between tme use and macroeconomcs.