New User Special Price Expires in

Let's log you in.

Sign in with Facebook


Don't have a StudySoup account? Create one here!


Create a StudySoup account

Be part of our community, it's free to join!

Sign up with Facebook


Create your account
By creating an account you agree to StudySoup's terms and conditions and privacy policy

Already have a StudySoup account? Login here

Public Economics

by: Madie Schinner

Public Economics ECN 230C

Madie Schinner
GPA 3.57

Hilary Hoynes

Almost Ready


These notes were just uploaded, and will be ready to view shortly.

Purchase these notes here, or revisit this page.

Either way, we'll remind you when they're ready :)

Preview These Notes for FREE

Get a free preview of these Notes, just enter your email below.

Unlock Preview
Unlock Preview

Preview these materials now for free

Why put in your email? Get access to more of this material and other relevant free materials for your school

View Preview

About this Document

Hilary Hoynes
Class Notes
25 ?




Popular in Course

Popular in Economcs

This 167 page Class Notes was uploaded by Madie Schinner on Tuesday September 8, 2015. The Class Notes belongs to ECN 230C at University of California - Davis taught by Hilary Hoynes in Fall. Since its upload, it has received 86 views. For similar materials see /class/191851/ecn-230c-university-of-california-davis in Economcs at University of California - Davis.


Reviews for Public Economics


Report this Material


What is Karma?


Karma is the currency of StudySoup.

You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!

Date Created: 09/08/15
LECTURE SOCIAL SECURITY HILARY HOYNES UC DAVIS EC23O OUTLINE OF LECTURE Introduction and definitions Institutional Details in Social Security Social Security and Redistribution Justification for Government Involvement bP N SOCIAL SECURITY Social Insurance Programs Features of social insurance programs in contrast to public assistance Compulsory yes Contributory payroll taxes Provides benefits when condition is met old age Benefits limited to those who have paid in work history Mechanism for society to pool risk for events that have catastrophic consequences loss of a job or events people do not plan adequately for retirement Benefits tied to previous labor market experience Biggest of the social insurance programs Different from other income security programs we have studied so far in that it is less designed to insure against unanticipated shocks instead getting old Social Security Same issues of protection and distortion are relevant here in the optimal design of SS Protection Insurance to protect against poverty and reductions in standard of living that can occur when earnings stop Distortion Benefits are conditional on employment and are funded by taxes while working Deadweight loss from the program s alteration of people s behavior Important questions in SS literature How does Social Security affect private savings How does Social Security affect retirement What are the distributional implications for SS Privatization of SS Important concepts for understanding pensions Defined Benefit DB Benefits in retirement defined by rules age earnings etc and benefit formulas Rate of return on contributions certain at time of contribution Defined Contribution DC Benefits in retirement determined by contributions Rules dictate contributions by employer and employee Funds placed in account worker gets gross return Rate of return on contributions uncertain at time of contribution Funded Assets are in the bank to pay obligations outstanding benefits Unfunded No assets are accumulated current workers contributions fund current retiree s benefits Pure unfunded programs are often referred to as Pay as you go PAYG Actuarily Fair Defined for an individual worker PDV taxes or contributions PDV benefits Examples DB DC Funded Private pensions IRA 4O 1 K older larger firms programs traditional private pension vehicle Private pensions newer smaller firms SampL Govt pensions most common form Univ of CA Newly adopted SS programs outside US Chili Mexico Argentina Australia Unfunded US Social Security Less common Sweden US Social Security Program OASDHI OASI Old age Survivors DI Disability Insurance HI Hospital Insurance SMI Supplemental Medical Insurance Timeline 1935 Social Security Act initiated Social Security 1960 DI added 1961 Retirement allowed at age 62 1965 HI SMI Medicare added Funding Historically SS is PAYG system now some accumulation Recall definition of pay as you go Current taxes Current benefits 19501985 No surplus pure pay as you go Tax rates low since the ratio of workers to retirees is high 1985 Huge surplus Tax rates increased 1977 1983 to build up the fund so that as the ratio of workers to retirees decreases baby boom generation the system does not go bankrupt Trust fund now 1 trillion liabilities of 9 trillion Financing Paid out of payroll tax on earnings Tax rates Social Security 62 employee 62 employer at tax up to maximum earnings 153 for self employed Medicare 145 no maximum Benefit Formula Benefits are a function of age at retirement average earnings growth in real wages and family structure Defintion AIME Average Indexed Monthly Earnings Average monthly earnings in covered employment Indexed to changes in real wages Based on 35 years of highest earnings Defintion PIA Primary Insurance Amount Set at retirement Used as base for all future benefits In 2003 PIA 090mst 606 of AIME 032of 606 3653 ofAIME O15AIMEgt3653 Redistribution in formula higher replacement rate for lower AIME workers Benefit in year t BI PIA t CPI retirement Retirement Date 62 year old early retirement benefits based on 80 PIA 65 year old normal retirement benefits based on 100 PIA Existing legislation gradually increasing normal retirement age to 67 Adjustment from 80100 from 6265 is actuarially fair adjustment for average recipient Earnings Test In 1992 Earnings lt 7440 per year penalty free Earnings gt7440 per year taxed at 50 rate Adjustments to PIA Married choose better of either a 150 of higher PIA or b sum of PIAs Surviving spouse earns 100 of deceased spouse s PIA Death benefits to children Redistribution in Social Security Program Several papers have used representative samples of retirees to construct the pdv of lifetime SS taxes and projected SS benefits This can be used to construct an internal rate of return on the investment MaxAge Definition 1 such that Z M 0 520 DAG This can be used to compare the internal rate of return IRR within and across cohorts The research shows that there is redistribution gtAcross generations Part of early justification of the program was to transfer from younger richer cohorts to older cohorts IRR clearly falling over time See Table 1 in Feldstein and Liebman gtWithin generations by income Higher IRR for lower income earnings PIA formula Offsets progressive tax structure gtWithin generations by family structure Higher IRR for single earning married couples gtWithin generations from unhealthy to healthy Lower life expectancy means lower PDV of benefits How does Social Security differ from a simple annuity Definition of simple annuity Take tax contributions and purchase bond at age 65 that yields certain return forever 1 Redistribution within cohort see PIA formula 2 Not actuarially fair across cohorts 3 Surviving spousechild 4 Earnings test 5 Indexed to wages prices 6 Cannot be sold Rationale for Social Security 1Market failure adverse selection Social security is a real annuity e g COLA adjustment in US SS Until recently no such securities available US Treasury now offers these Annuity companies face adverse selection only health sign up 2 Paternalism People are myopic and undersave Why Lack of information high discount rate What is enough How to measure 3 Samaritan s dilemma Without SS forced savings then some people will end up poor and old At that point you want to help them So you might as well force them to save to begin with 4Redistribution based on lifetime earnings Other sources of redistribution well documented Basic Economics of Pav as Vou 20 Social Securitv Outlined in Samulson s seminal paper JPE 1958 paper Generaliztyion of Samyelson s model in 2 period Overlapping Generations Model Feldstein and Liebman Handbook of Public Economics Assumptions Identical Individuals Each agent lives for 2 periods Work a fixed amount in period 1 retire in period 2 Population growth rate n No capital no money Can not save for old age good is perfectly perishable Notation L of workers in period t W wage rate 8 SS benefit T t SS taxes 6 SS tax rate Implications Lt1 2 1 nLt labor growth populationworkers Z QLtw aggregate tax revenue BI 2T BLtw PAYG benefitstaxes in yeart Rate of Return for person in cohort t Taxes paid as worker Z QLtw Benefits received as retired Br1 Qme 61 nLtw Taking ratios generates the internal rate of return IRR i W 2 1 n Internal rate of return T QLIW t Samuelson called 72 the biological rate of interest Even an unfunded SS program pays interest Also note the Windfall received by the 1st generation Bl 00 T O 0 Connecting back to rationale for SS SS is desirable since it permits individuals to retire and consume despite all perishable goods Key result of Samuelson In a world Without a capital good SS leads to pareto improvement by making inter generational trade feasible Adding technological progress wage growth wt1 1 gwt Internal Rate of Return B 1 61nL 1gw Lz rz 1 17 T M lt gx gt t Despite simplicities the model demonstrates realities of PAYG SS system High g andor high n will yield a high rate of return pareto improvement In the United States currently both g and n falling over time yielding falling returns 19601999 Real wage bill grown by 28 real wage growth 1 labor force growth 18 Wagesbeneficiary 327 in 1995 204 in 2030 Reason for increasing wages and building up trust fund is to not have to raise tax rates by so much later Adding Capital Stock With a nonperishable good then consumers can save from one period to the other Evaluating impacts of SS are different in that model F eldstem and Liebman present this in their chapter TaxBenefit Linkage Note There is a deadweight loss on preretirement earnings due to payroll taX AS LONG AS taX is viewed as a taX and not a benefit Back to balancing distortion and protection Optimal SS system has to balance these two parts 1 Protection Providing basic consumption for those who are too myopic to save for themselves 2 Distortion Providing a lower rate of return for those who ARE NOT myopic and therefore would have saved adequately for their retirement Crowding out other saving Distorts retirement and labor supply F eldstem and Liebman present SS Optimal tax in chapter Lecture Taxes and Labor Supply hwhoynesucdavisedu EC230 Outline of Lecture 1Basic labor supply model with linear budget set 2Adding taxes to budget set 3 Empirical literature N Eissa Taxation and Labor Supply of Married Women The Tax Reform Act of 1986 as a Natural Experiment E Saez Do Taxpayers Bunch at Kink Points Motivation and empirical regularities 0 Main lesson from optimal taX literature optimal taX rate depends inversely on compensated wage elasticity of labor supply 0 In this lecture we will discuss issues around the estimation of this important parameter 0 Enormous literature in public finance and labor 0 Different measures of labor supply 0 Extensive margin work or not 0 Intensive margin hours of work 0 Retirement transition into retirement 0 We will also discuss the elasticity of taxable income as a more general characterization of labor supply taX avoidance old Feldstein point that there are other margins then hours of work 0 Our plan 0 Lay out basic static model one earner and two earner 0 Note role of taxes 0 Discuss alternative approaches for identification 0 Discuss Eissa Saez 0 Labor supply and low income population 0 Labor supply and high income population Static Labor Supply with Linear Budget Constraint Individual faces exogenous deterministic wage W and price of other goods p They also may receive nonlabor income N Utility is a function of leisure E and other goods x The choice problem is Maximize U E x subject to wh N px h T E Where W hourly wage h hours worked Eleisure T time endowment x composite commodity p Hicksian price index px can be replaced by total income y or consumption c N nonlabor income Maximize U E c subject to wh N y h T E Usual Assumptions Increasing in Z and y decreasing in h Leisure and consumption are normal goods 3 wTiN Deriving budget constraint yWhN hT E yWT N N yWTN WE intercept WT N full income at full hours slope W loss in income of one more hour of leisure Therefore indifference curves have usual shape We are typically interested in studying the determinants of hours worked but we model the determinants of leisure and then translate back to hours First Order Conditions for interior solution wUCUhO WEE U Maximizing U with respect to h yields the labor supply function h h w N Corner Solution Define w reservation wage evaluated at h0 C wgt w thenwork hgt0 w lt w then no work h O equivalent to hlt0 Comparative Statics u 6h Uncompensated elast1c1ty of labor supply 8 W ma w substitution effect lt0 income effect gt0 if 1 normal Can be positive or negative backward bending labor supply 6h Income effect parameter 77 w 6N If leisure is a normal good then negative Imbens Rubin Sacerdote AER 2001 Compensated elasticity of labor supply EC 2 w h ahc 6w Always positive Slutsky equation 6 0h 6w 6w 6N Implies EC 2 8 77 EC gt 077 S 08 lt 808quot ltgt 0 EC is the most important parameter because it measures the cost of distorting wages using taxes Elasticity of participation Positive increase in wages leads to an increase in participation no income effect when considering extensive margin Adding taxes to labor supply model Example 1 a uniform proportional tax denoted as t y Max U E y st wl thNyhT E FOC 6U 0h a U wl t 5y LS function h hwl t N Observations 1 net of tax wages belong in the labor supply equation not gross wage 2 Policy question How do taxes affect hours worked Elasticity can tell us theory does not even tell us the sign 3 How do taxes affect labor force participation taxes gt reduction in net of tax wages no change in reservation wage gt probability of work decreases The empirical labor supply literature has evolved from looking only at hours worked to focusing on labor force participation Clear predictions of theory Hours harder to model perhaps not free to choose Participation margin more elastic than hours margin Increasing focus on more elastic labor supply groups eg women Where participation is important Example 2 Progressive marginal tax rates Consider three increasing marginal tax rates Budget constraint becomes Y wh N TaxeswhN y t t1 if E lt E1 tt2ifE1ltEltE2 f Y3 t t3 if E gt E2 w1t3 sz t3gtt2gtt1 w1t2 Given that utility function is concave and budget set is convex then we know there is a unique tangency or corner solution on one of the segments Possibilities not working hlt0 tangency on 1st 2nd or 3rd segment on a kink expect people to be bunched on convex kinks Consider someone on the highest segment First Order Condition aU ah aU ay Which implies the labor supply function wl t3 hhwl t3YV3 Or more generally h h wn YV Where Wn net of taX wage YV Virtual income Determining participation h lt 0 then no work hwltlNlt0 Suppose you want to empirically implement this model with some data set You use a linear hours of work equation and relate hours worked to net wages and net nonlabor income Virtual income h0tBwn5YVyZe KEY OBSERVATION Labor supply theory tells us that the labor supply equation is a function of net wages and net nonlabor income Yet these are themselves a function of hours worked Endogeneity We will see LOTS of examples this quarter of other government programs that change the budget set rendering net wages endogenous Empirical issues with identification of labor supply elasticity Data Current Population Survey annual starting in 1960s Panel Study of Income Dynamics panel 1968 Starting with basic OLS framework hi a wi gNl 0Xl 81 Pencayel 1986 survey for men 8 O 77 Ol EC 2 01 Killingsworth and Heckman 1986 survey for women Much larger elasticities with larger range 0 to over 1 Identification issues lcross sectional identification of w high wage guys have more taste for work independent of wage 2Measurement error wage usually computed as earnings divided by hours Spurious negative correlation 3Measurement of hours itself is quotlabor supply just hours worked 4Functional form sensitivity linear model means elasticity is w hm Varies across persons How to compute average elasticity How sensitive is elasticity to functional form 5 Taxes Early literature ignored taxes Theory tells us that equation should depend on net of tax wages and nonlabor income Taxes generate endogeneity of net wages 6Nonparticipation OLS regression can only be run for those working because there is no wage for those not working But participation is obviously correlated with tastes for work and is an important issue for women especially married women Result is that basic estimate will miss extensive margin which may be the largest margin of response Empirical Estimates Negative income tax experiments NIT conducted in late l960s early 197 Os in SeattleDenver SIMEDIME and other rural sites in the US Basic design of the program was a lump sum transfer with a 50 7 0 phaseout rate Basic result was signi cant labor supply response but small Men elas small 01 Women elas 05 concentrated on extensive margin This early attempt at experimentation in the US was not ultimately successful Experiments were poorly designed nonrandom selection into experiment selected on income nonrandom assignment to T and C groups self reported earnings with incentives for T to underreport so that they got NIT payment Lesson need to match to administrative records UI SS rm tax records sample attrition Nonrandomness undoes the simple TC comparison that is so powerful in randomized studies So much statistical modeling was used here Hausman nonlinear budget set model There was a lot of attention to taxes and nonparticipation in the late 1970s early 1980s Structural nonlinear budget set models Hausman Sample Selection methods Heckman Idea of Hausman model 0 Wanted to identify deep parameters 0 Assume functional form for preferences usually linear hours equation specify Where observables and unobservables enter the model Specify budget set 0 Solve algorithm for maximizing utility ranges of unobservable that imply location on kinks and budget segments 0 Find estimates with maximum likelihood methods Hausman 1981 results large compensated elasticities due primarily to large income effects Larger elasticities for women 0 Very in uential and method applied a number of times 16 Criticism of Hausrnan model 0 Sensitivity to functional forrn Does not address issue that wages are correlated With tastes for work Static labor supply model predicts that indiViduals should bunch at the kinks in the taX schedule Little eVidence that they do Saiz Do Taxpayers Bunch at Kink Points Basic prediction of kinked budget constraint model is that we should see people bunched at the convex kinks And we should see a gap in the distribution at nonconvex kinks Some papers have examined particular applications social security earnings test welfare recipients around notch WFTC and hours restriction but no study has examined this among taxpayers in US Simple clever paper using the best data tax data Modeling insights Less curvature in indifference cures higher substitution elasticity 9 more bunching dzz e dtIt esub elastiticty tMTR ztaxable income 0 Therefore if there is little evidence of bunching and model is valid 9 small elasticity of taxable income 0 Later he considers changes to model to explain lack of bunching uncertainty in income constrained hours choice Data 0 IRS annual crosssection of taXfilers 1960 1997 NlO0000year He does not use all of the years high in ation years when taX parameters were not indexed Methods 0 Simple descriptive unconditional exercise 0 Uses histograms and kernel density local smoother of histogram within a band observations further from the central point are weighted less in average Results 0 Some evidence of bunching around lSt kink MTR goes from O 9 15 Figures 23 More evidence for single and HH returns First kink probably the most Visible to taxpayer But could the finding be an artifact that those left of 1st kink do not have to file and may not be in data 0 No eVidence at 2r101 or later kinks Some eVidence of bunching around EITC first kink Results concentrated for those with self employment income no effect for those with only wage income Implication Small elasticities Simulations using extended model again shows no clustering So these models are not right or elasticities are small or agents do not know where kinks are Problematic for research using kinked budget constraint methods 20 39I39Id 11125 39d an quot15 arc I C J39I IZH quot 39 l1arrel rm i39 Damn cummgr E 51quotka 311 I 2quot mm FIEIE 2 I I KJJIMC 521 3423le Tumble IZCI39Iquot IECEJ 1 am Tana Invr e QUE dcl Fig 5 5iribut39z39n5 around the rst kink Danni 1 E E g9 21 Ensnary Emmim HIM 03155quot EanEl Z JZH Z C E 31 11 EIZI 515100 SEZJIJEI S45E10 SEEJZZEI quot313th IF BDTE EDD dollarst 22 A Marre TBIJSEEP Fl I39lljl39 f 125517 Dbl E EI39IJ E TEIDEIEEIquot 539 25 39 935 cm 1 Ca 131 butt s 53 E E E 5 3 m mlnl 54cm mum IL I 5le LilZIZM mmu EDGE Tmm rm 22E 3 am Tanae Inzcer39e EDI ucl an Fig 7quot Density di51r bump amund kink 5 IE 23 WEBB12E 23 Early Instrumental Variables Mroz in uential study that reviews literature on married women39s labor supply Identifies instruments that are credible using Hausman specification and overid tests Estimate IV with small and large instrument set and test for equality of estimates can be low power Credible instruments unemp rate parent39s ed wife39s age and ed Not credible labor market experience age hourly earnings preVious reported wages This study contributed to emerging View that policy variation taxes was necessary to identify parameters Blundell et al Econometrica Use demographics taX reform 24 Tax Reforms and Labor Supply New Public Finance 0 Using tax reforms as a natural experiment to evaluate the effect of taxes on labor supply and other outcomes Can get around problem of endogenous net of tax wages and wages more generally 0 Advantage of tax reform policies can affect some groups and not others creating natural treatment and control groups 0 We have seen lots of changes in tax laws to provide experiments to examine TRA86 Tax Reform Act of 1986 o The program where the most work has been done Why See Auerbach and Slemrod JEL 0 Most significant policy change in postwar period 0 Goals of TRA86 Horiz Equity Efficiency eliminate tax preferences Simplicity Result Broaden base reduce rates MTR o 1986 14 brackets 11 50 o 1990 5 brackets O 15 28 33 28 0 Increase standard deduction and personal exemptions We will see later papers using this variation to look at impact of taxes on low end EITC and high end 25 Eissa TRA86 and Married Women s Labor Supply NBER WP Never published not sure why but great teaching paper and also very in uential paper Established convincingly that married women are sensitive to taxes have higher elasticity of labor supply Added to our knowledge that participation margin is more sensitive than hours margin Good example of difference in difference methodology most commonly cited DD approach to taxes and labor supply Eissa focuses on high income women because they had the highest reductions in MTR see figure from paper 26 Hind Finn39s I nun 713 mm mam Tm twain Elma 27 Economics Secondary Earner Labor supply model Most common approach is to model labor supply of husband and wife sequentially l Husband or primary earner maximizes utility ignoring wife just like single agent model MaXlh SI Whl lh l N Z Y 9 l l1 2 Wife or secondary earner maximizes utility conditioning on husband s optimal labor supply decision Therefore she takes N Whhh as given MaXlw C st wwhW whhh N C Graph this 28 Comparative statics of secondary earner model 0 Earnings of husband increase T through increase in h or W 9 nonlabor income of wife T 9 income effect 9 hours and employment of wife i 0 Taxes Decrease in taxes leads to T net nonlabor income 9 hours and employment of wife fall T wW 9 hours employment T KEY with progressive taxes she gets the change in MTR which is exogenous to her own labor supply but comes through her husband Her first hour MTR is his last hour MTR 29 Empirical AppraochDifference in difference hours and part Before After Treated Hto Htl Control Hco H01 Treatment women in gt99th percentile of distribution Control 75th percentile E 90th percentile Tradeoff 90th better control but gets some treatment Htl Hto not good if secular trends or other contemporaneous shocks to labor market secular trends T over period Htl 39 Hto Hcl H00 D in D Nets out A in control group change that treated would have if they weren t treated Key identifying assumption controls are good comparison No contemporaneous shock to treated No assortative mating on unobservables 30 Data CPS 1984 1986 before 83 85 1990 1992 after 89 91 TRA86 phased in by 88 Predictions Employment of women in 99th p will rise relative to women in 90th p 9 Her MTR i 9 net wage T 9 LFP T But we have to believe that her net of taX nonlabor income did not change much Why 9 husband s MTR i 9 T earnings or no change if elas small 9 But TRA86 broadened base 9 overall effect on her after taX nonlabor income is small To the extent which his 1 net earnings are not captured then this estimate is an underestimate of total effect 31 Hisii Flamquot Inamg l 2 mm 539 Ii r 3quot 75111 r r I 124nm Tpdlh but new 32 Results Unconditional difference in difference Ave Y AMTR Tab Ha ALFP D in D 99th p gt 90K 139 pp 90 pp 90th p 67K 69 pp 45 pp 45 pp 28 13 75th p 47K 41 pp 53 pp 37 pp 28 12 Ave Y AMTR Tab Ha Ahours D in D 99thp gt9OK 139 pp 163 90th p 67K 69 pp 96 67 648 6 75th p 47K 41 pp 55 108 651 9 33 Conditional DD PrW0rk a0 0512 a2 highi a3 P05t86t a4 High P 05t86 Zit age educ kids young kids race CC year amp state fixed effects Expectations oc2lt 0 baseline inc effect oc3gt 0 secular trend oc4gt 0 Main test of TRA86 34 Results Significant increase in LFP less for hours Consistent W lit showing greater responsiveness on participation margin than hours margin Mroz Hausrnan 35 Tam H mm Mamquot Rnulh Labmr Farce Paniczpnium Variath mum 1395quot Barrail Euutrul 9039 Pertmil Higll mu IZITU IIZIJ l D 12E 1 137139 4 AWEIV 071 I 1175 WEIR DM 4205 a39l 3035 Mm 152 115 IE H51 Huh 57E T1 an all 1 D35 1j1395i fl 1 ED I ME Ail IJil 05 I ll i H Iiil Agc mum ms Hm mm Eihnmim m5 2315 WW I 39I I m Edmuimr m a 3141 v 4101 09 ChilBrmm r uni1E JET Ij LOW Blart 373 u 441 I 094 121 Human SewEmple 131 03 i M23 LUBE LEI ILIk lh quot131 4935 4830 5359 Uh rvmlm 3556 555 BENT HM Regimens in mlurruu l mm 133 mm arm almmm Winn in udumns 12 and all mmlud 3121 min EH dumi RISE lni mum tilt mmrniu Dun an Mm CPS WSWH and HEDGE Samdl 4mm are in premium TatIv Vin Praline1i Pmicipilinm Ilene l39rubil Esljmmes Gum e are Amer hang m srmcinlnw TMHE 1m infMm I Churn TF5 Pernn le Hislt 4450 m 1137 l EJ EJ 1395quot Percentile 741 73 2031 J33quot l I J39lei39al gum Cunlm Bilquot FacEIu lle High 511 063 Ill Ill 9039 Percentile ll 54 Jill1 ci 2 25 laym Pndklcd Pmixmiun es e uulilh u FEW milled WEE 1 currentcud u ll39l39l evenly atmrinin 39in Ihe39 WIT 39the Villa Elasmaw var F icipalb I HuhImam Gmup Cmml Gruup Elm quot15 1395 Pcrmmilc 1M 9239 Percentile I15 Criticism of Eissa s research design Heckman Comment General criticisms Heckman does not support structural estimation but feels that a theoretical DD models are misguided 0 do not identify any parameters of interest e g elasticities 0 require strong assumptions no differential trends across groups 0 throws away information in this case using after tax wages in model Specific criticisms Assortative mating on unobservables Trend toward quotpower couples Used to be that prof men had nonworking spouses now more common to have working prof spouse Yet in middle class more stable situation with working middle class spouse 0 Demand or supply shock to 99th p eg work in different sectors 9 different trends for T and C re ecting inequality literature 0 Other A taxes affect 99th p more than 75th p 0 Does taX reform affect selection into group T C groups 0 Things to examine in DD model that were not known then 0 Placebo treatment use data for pre periods redo DD using placebo treatment say comparing yar O and year 1 0 Useful necessary to plot outcome variables in T and C year on year for whole period examine whether the trends are similar in pre period Look for change 38 Summary What do we know about the size of elasticities across groups lPrime Aged Males Wage elasticity O mostly gt O I T 9 w i 9 h i Inc elasticity lt 0 but small DWL fairly small maX 20 taX reV 2Married women Much more sensitive Wage elasticity 5 10 Inc elasticity 25 Larger DWL but lower LFPR so this holds it down At 9 no A in h for lots of non workers 3FHH In the middle 39 Econometrics of Kinked Budget Constraints Convex Budget Set Hausman s model After tax and transfer income Preferences hquotlt h W y 8 h observed hours a taste shifter functional form for labor supply equation 40 Comments Virtual income y3 y5 are a function of observed nonlabor income and tax system Need one preference assumption either labor supply equation IUF or DUF Assume gross wage is exogenous Assume gross nonlabor income is exogenous Ex Functional form Hausman used for labor supply equation was hiawi yizyg which implied the following for the IUF vwz39yz39 exp wiyi wi AZ 41 4 Steps in constructing the likelihood function 1 What do you observe 2 Identify possible states 3 Determine economic decision rule that justifies each choice 4 Derive probabilities associated with each choice Step 1 What do you observe Hours 0 or continuous hours worked Hourly wage rate for workers Nonlabor income Covariates Step 2 States of the World 0 h0 l 0lthlth2 2 hh2 3 h2lthlth4 4 hh4 5 h4lthlth6 6 hh6 42 De ne the labor supply function for each segment hwiyie linear labor supply curve for net wage wi and net nonlabor income yi Ex i123 wlw no taxes w3w 1 t1 lSt marginal tax rate w5wlt2 2nd marginal tax rate y 1N observed nonlabor income y3 Virtual income y3 y5 Virtual income y5 43 Step 3 1 Economic Decision Rules State 0 State 1 State 2 State 3 State 4 State 5 State 6 h 0 hw1y1 0 Desired hours given wl yl are lt0 0 lthlth2 hhw1y18 Desired hours given w1y1 are between 0 and h2 h h2 hW1y1e 2 h2 AND hW3y3e 5 h2 Note that being on kink has higher probability than any given point on segment h2lthlth4 hhw3y38 hh4 hw3y3e 2 h4 AND hw5y5e h4 h4lthlth6 hhw5y58 hh6 hW5y58 2 h6 44 Then translate desired hours rule into rule about unobservable e Derive probability that choice was made Step 4 Create Likelihood Function Lh n Pr5i0 5i n fhwiyi 8i i 0246 1 135 where 8i 1 if state i is observed and 0 otherwise 45 LECTURE MEDICARE HILARY HOYNES UC DAVIS EC23O OUTLINE OF LECTURE 1 Overview of Medicare 2 Dimensions of variation used in the literature 3 Card Dobkin and Maestas The Impact of Nearly Universal Insurance Coverage on Health Care Utilization and Health Evidence from Medicare Medicare Background and Program Details Basic F acts 0 Nearly universal coverage for elderly 65 0 Costs 260 billionyear 12 federal spending 2 of GDP 0 Benefits Part A Hospital Part B Physician s services both with deductible and copayments History 0 Medicare enacted in July 1965 nearly universal implementation in July 1966 0 Single largest increase in health insurance coverage in the US increase in insurance coverage by 75 percentage points among elderly Eligibility Can get at age 65 if you or spouse has worked for 40 quarters Part A is free Part B is optional and has low monthly premium Part D is new drug coverage Who is likely to be more affected by Medicare 0 Those With low insurance coverage disadvantaged This is important to understand the reason for the program redistribution as well as to keep in mind in empirical analyses How to examine impact of program when it is universal and With no interstate variation 9 Look for discontinuities in year using introduction of program and age using eligibility at age 65 But it also addresses one of the most important questions in health economics how health insurance affects health care utilization and health status Alternative RAND health insurance experiment Similar motivation for interest in literature on Medicaid expansions Card Dobkin amp Maestas The Impact of Nearly Universal Insurance Coverage on Health Care Utilization and Health Evidence om Medicare AER forthcoming Regression discontinuity approach using variation around age 65 They examine impacts across groups with an interest in evaluating impacts on inequality in utilization outcomes Identification No confounding factors that are also changing near age 65 All other factors are smoothly changing near age 65 in contrast to discontinuous change due to Medicare Outcomes examined Health insurance coverage 1st stage CPS NHIS Health care utilization BRFSS NHIS Health behaviors BRFSS NHIS Hospital stays Discharge data subsequent project Death Vital statistics Confounders Retirement Income Family Structure Changes in medical guidelines at 65 RD captures short term shift from lt65 to gt65 RD Methodology 3quotquot 7 r 1 239 2 32 L quot1 39 4 J myquot TeraJ P jal i law 51 quot him l Dani 39 V where lljtal Lia ui39Jgljtal T L g l F HJ HJDI TJ and 1 1 139 41quot 1 Tip llija R 1139an 139 aga j Estimate outcome variable as a function of polynomial in age hJa with an intercept shift at age 65 nyj Derivative potentially discontinuous at 65 In principle the shift at age 65 can re ect an increase in coverage e g for someone wo coverage and generosity for someone With an existing plan F1gln elC 7D 39 R Mm 7 an n 100 n D B gi eggaoe o oegggwe awgagnagaaggagaag 3 ELL D a quotrug 0 o 0 Ba Du D Dan E u D nun Duo 0 o 000 O 000 00 so 0 o o o o 000 00 0 09000 0090000 00 co 0 0000000 000 o I 030 7 I I I I I ml o N N u 0 3 90 o o 3 0 39 j 0507 F 0 0 u o 3 Any Cm erage HighEd Wintes I Any Cove 0 All Groups 9 2 Any Coverage LowEd Mjuodues rwn Pohcies39 HighEd W hites Two Pohcies All Groups 010 i O TWOP011Cies LOWEd Minorities 1 000 VViViiVVVVVVVVVVVVViViiViiViViiViViiVVVVVVVVVVVVViiViViiVVVVVVVVVVVVVVVVVVVVVV 55 36 57 58 63 64 6 66 67 68 6 70 Age In Quarters First stage sharp increase in coverage more for disadvantaged From NHIS 73 First stage impacts on insurance coverage using 19992003 NHIS Tame t msurarrce Chamdertsncs Just Eerore Age 55 and Esttmared msmrrttnunres atAge 55 Arty msumnce Prwate Coverage was rum rte m 5 3 t4 s t DVEIEH sampte a7 9 9 5 7t 9 r2 9 a a t ctassmea by Etrrrrrcterarm Educatrerr Wmle New spah e Hrgrr Schcu Dropout 2tt 555 341 30 535 752 t5u 445 4st 7259 4 6 t2 7 a 3 4 a 4 5 3 men smemcraauee 114 547 52D 75 895 519 wt 5t 5 539 ram 50 77 t e 35 25 4 4r Least Same image a t 63 4 94 e 4 4 35 a V2 3 a a 55 t 59 t 49 t 4 7 a 5 t a 4 a 2 a Mummy 5 Hrgrrsmeutomumrt 195 445 559 2t 5 332 712 1t 4 194 39t 93 3t 2t 25 9 at s ttrgrr smear creature a 7 44 a a5 2 a 9 an 9 75 a 13 e 23 4 54 2 45 4 4 7 2 a 51 4 s a 5 7 4 Least Some CDHEge tn 3 52 r set 5 s 73 3 754 tt t as 4 as 2 722 3 4 S t2 0 4 3 3 5 7 2 ctasst ed by Etrmrmr OHM s wrrtteNerspanrc an 552 at a 73 797 723 04 5t 9 at 9 7335 4 s to 5 t 4 35 2 3 9 Eiack NcnrHtsuzmc Att 7 9 4s 5 s4 5 9 57 t 74 2 13 4 27 a 4st 2 43 5 3 6 t2 0 2 E 3 7 3 7 ta Hrspantctmt tea 444 mo 73 425 729 as 2t 7 529 42t 37 30 t 7 2t 37 Huge increase in coverage and dual coverage u HighrEd W lmesrAanal 7777 quotHiglirEd VVlutesrPremcxed Ox emll SamplerAmml Oyemll SamplerPredicted LowrdMmmmemcmai Lcerd Mmoxmesinedxcled 55 S6 37 58 59 60 61 51 63 64 65 56 67 68 69 7o 71 7 73 74 A22 Testing for confounders estimate same model on employment No evidence of an employment discontinuity at age 65 Stronger reduction at age 62 they also look at marriage poverty mobility no discontinuity Table 3 Measures or Access in Care 151 Before 55 and 5511M a 01m 39mumeszi manna NHIS 1992723113 NHlS aw Doctur Delayed Care UK 1101 cm Lure Hosp11a1 s1ay Last Veer Last Veer Last Veer mam RD 15 mm 1 12gt a 10vemllsanmle 2 45 11a 12 11141 114 Classmed 1m Emmuty and ndlmamn wme anrMSuamc HighSchooleDom 115 45 7g 702 317 31 44 1a 1141 11c 13gt 13gt 3 Hensmawemauam 71 in 55 713 351 7m 123 13 1281 2 a 115 117 4 AtLeastSumeCuHege so 45 37 714 315 no 92 21 1141 E13 113 117 Mummy 5 nghSchauleDom 135 753 117 742 s02 54 45 an 11m 19 122 114gt 5 High chuoleraduate 43 733 12 15 343 19 114 1a 1321 37 27gt 14gt 7 AtLeastSumeCuHege 54 416 43 702 350 37 95 17 1111 mm 1393 12m ClassmedlmEmn1c1ty om a WmteNunrHlSDamt 69 46 44 712 553 as 115 13 1a 41 n 3 1n 3 1n 5 a ElankNumHEDaniclXH 73 4g 54 703 342 35 144 15 1111 11gt 119 111 1a H1sgamcwl 111 4g 33 735 794 32 113 11 mm run 1m 11m Health care utilization Modest increase in access to care and utilization larger for more disadvantaged groups Hospital discharge data CA FL NY 19922002 Figure 3 Hospnal Aduu39ssion Rates by RaceEthnicity u magma as am n wm Np w x mamm Mme 39Ipand ee Ragazimnl ma m am n p and me mam sm Hi and Ma mam amaze 5 E n m u i v 1000 nu WJA 3vr a f u 5 vvrquot 7 E Jar anquot a c A A E V A 35quot n u DualADJ Jeanna a a n 39n a so an e as 64 m Increase is driven by discretionary medical care diagnostic heart treatments Figure 4 HospnalAdmlsslon 1n Calemma by mum5111p Type 199 0 16000 7000 3 14000 rr r A 5111 5000 s g 000 1 39 E Z Pmmcbmmho x 02 y g vx 39 r 5000 5 5 10000 frfceha wAf39n mquot hurchrRun lt Puma for Pro t j a 4000 a 3000 Huspmlmsmc 300quot M 1 7 2000 C omn mmmm 1000 Mix of hospital usage changes Conclusions Medicare causes sharp increase in coverage especially for disadvantaged Medicare causes increase in health care utilization Low cost services increase more for previously underinsured Higher cost services increase more for dual insured In a follow up paper Does Medicare Save Lives the authors nd that this change in coverage leads to a nearly 1 percentage point drop in 7day mortality for patients at age 65 20 percent reduction The mortality gap persists for at least two years following the initial hospital admission Figure 1 Ageadjusted Elderly Morlallty Rate m 1959 1955 1950 1955 197a 1975 1950 1995 1990 1995 2000 year Finkelstein amp McKnight What did Medicare do And was it worth it Examine introduction of Medicare on mortality and out of pocket health care expenditures They use a differenceindifference framework difference 1 beforeafter introduction difference 2 age young elderly 6574 vs near elderly 5564 As an alternative second differencing they compare changes with the introduction among groups with higher and lower prior insurance coverage DDl Changes across young elderly and near elderly before and after 1965 Figure 2 Mortahty Rate Trends by Age Group mg m k 7 8 a w E E F w o 5239 8 E E lt2 w e o a 22 N E 1550 1555 WEB 1965 19m 1575 year 47 Elderly 6575 Nonrel Trending down for both groups elderly turns down pre1965 Conditional differenceindifference model 111cleat119w 81 h1popnw selderlg 61 391S IIfeg r quot 1Yem39 mm 2 elderly 10790 1139Bs F2951 Figure 3 Estimates or equation 1 m age memmmion strategy ror ages 55774 a Q l 1950 1955 1960 l l 1965 I970 1975 how is this a DD model Estimated X Same as unconditional graphidecline in relative mortality prior to 1965 DD2 Geographic variation in of elderly wo insurance Why do this as a DD instead of DDD They drop the age interaction which seems strange in this case They mention later that DDD is similar to DD but I would show DDD Note problem is that vital statistics data e g death certi cates is short on demographic variablesino education for example So a common alternative is to use geographic variation to proxy for these missing demographic variables No break in trend at figure 4 Estimates of equmiou 4 me gengmpiiic variation su aleg 1966 Panel A Ages 55 Also no impact on speci c diseases quot 39 39 not shown in paper Their interpretation of the results when people were in need with possible exception of segregated south people wo insurance sought and received care in county hospitals This is consistent with little change in mortality Out of pocket expenditures Looking for some impact of Medicare they use two crosssectional data sets on health care utilization and expenditures 1963 1970 They want to capture the impacts across the distribution as out of pocket expenditures are highly skewed see Fig 5 QTE in DD setting Again comparing dq spendq 1970 elderly 1 spendl 963 elderly 1 young elderly to spe11dc1 1970 elderly U spend q 1 963 elderly 0 near elderly beforeafter Results for QTE with covariates Figure 71 Ages 55 774 Covanate7Adjusied 8 5 r o E B u o c 3 E a t 8 m r i i i i u 20 40 so 50 mo percenme cenmeIreatmentestvmate 77777 951Howerbound 77777 95 Ciupperbound Large reduction concentrated at the to of t e distribution Is this result just pickingaup an undquyiglg trend Figure 8a Falsi cation Exercise 8 Ages 55764 CovanateiAdjusled C ompare c o I 5 559 to o o y g3 6064 E C N o o o o 7 6 2 0 4 0 6390 8390 I60 percenule 7 cenme treatmem eshmate quot 95 Cl lower bound 95 c1 upper bound Figure 8b Falsi cation Exercise 0 Ages 6574 emanateAdjusted Compare 8 6569 to m o N 7074 11 g o S N y o o o v w w w w w a 20 40 an mo 60 percenme cenuletrea memesumate quot 95 CHowerbound 95 or upper bound Looks like n0 a1th0ugh there is a downward trend in 8b Comments not sure Why they do not present everything by race This is in the Vital statistics data 21 LECTURE MEDICAID HILARY HOYNES UC DAVIS EC23O OUTLINE OF LECTURE 1 Overview of Medicaid 2 Medicaid expansions 3 Economic outcomes with Medicaid expansions 4 Crowdout Cutler and Gruber QJE 1996 Medicaid Background and Program Details Basic F acts I Health insurance to low income population I Really 4 programs in one 0 Low income women and children families 14 SS 3 of people 0 Gap coverage for Medicare low income elderly 0 Low income disabled 13 people 0 Nursing home care for low income elderly 34 S I Third largest entitlement program after Social Security Medicare I Currently is the fastest growing entitlement program History I Social Security amendments started Medicare and Medicaid I States joined slowly from 19661982 I States had exibility in setting up programs Medicaid Eligibility Cash aid welfare population Eligibility for cash transfers AFDC 881 9 automatically eligible for Medicaid Nonwelfare population Medically needy Income not low enough but close Large medical expenses relative to low income Can spend down Mostly used by elderlydisabled Medicaid services They have less discretion on the covered services Fed guidelines States may impose limits e g days in hospital Provider reimbursement Medicaid is the insurer but individual maintains the choice of care Provider and hospital reimbursement rates are low but vary across states relative to Medicare and private insurance Medicaid Costs Typically costs are an uncapped entitlement Cost sharing between the state and federal government Fed is inversely related to the state per capita income 5083 Same cost sharing as old AFDC program Costs have increased greatly particularly in the late 19605 early 19905 Gruber Medicaid Table 51Jnduplicaled Nninbei of Medicaid Recipients By Eligibility Category Fiscal Years 95 1972 Fiscal Spending 1997 Total 30926 3742 33 432 3863 3 4053 3 41 1 361 I 8 4285 34872 3955 36700 37500 4700 124429756 37721431 303 Numbers in thousands lin dues Permanent Dependent 108 101 18 135 2222 109 2355 97 2572 92 2710 8 2636 79 2674 92 2817 86 2993 84 2806 77 28 39 79 2834 80 2937 87 3100 8 3296 86 3401 95 3496 3 3635 RR 1 on 84 43 X4 4932 5372 92 5707 13 6126 1129 6800 7000 54130401 435 17500 11658 5 126 Adults in 7600 12307239 99 Otlmr39 Caseload has also increased driven by disabled and more recently kids Gruber Medicaid The focus in the literature concerns Medicaid for women and children Why Better policy variation Economic agents have more scope for moral hazard than the elderly This is where the research is Medicaid Expansions Phase 1 198487 incremental expansions for populations with similar financial circumstances as AFDC recipients pregnant women children in two parent families Phase 2 1987 0 Applied to children and pregnant women goal to decouple AFDCMedicaid through increasing cutoffs 0 Increased the income cutoff for all kids regardless of family structure 0 Federal mandates tied to age of child of poverty line by certain date 0 Example ofmandate By 1992 cover all pregnant women and kids lt 6 up to 133 of the poverty line All kids born after 93083 eligible up to 100 of poverty line 0 States in many cases went beyond requirements andor met requirements at different times O erall changes in eligibility Gruber Medicaid and TPA 1997 o z m A mm39 Fugmo t m fruLUquoten Emmi4 47 3 1 quotA 5 195d 3965 IQSE 15149 s39uu Fag Isa 1339 15627 M Fianna 31 mm m m 3mm am mlzlren 0 Va Eligible 390 Mmluiki A quotuCavaredbyMedicald 31 M But takeup IS not eliglblhty 23 23 IB J3 vvlwww 8 383945 55 65 78 88 99 0919293949596 V83 Figure 1 Eligibility and Coverage of Children 015 Example of how changes vary across states Gruber Medicaid 39J39abla 3 SLau Mudiaaid Age and 1111me Eligibility quotThraahulda fur Childmn Januargir 1 9113 I December 1 98 9 December 1911 1 I December 1 519 3 Static Age Medicaid A g Medicaid Medicaid Age Medicaid Alabama 1 1115 11 133 111 133 A 1 aska 3 11111 8 13 3 111 13 3 A rimna 1 11111 2 1 1111 8 1411 1 2 1 411 Arkansas 3 7395 139 11111 11 185 111 133 Ca1 i lm i a 5 1 1391 5 S 1 8 5 111 21111 Cd11 3rad1391 1 75 8 133 111 133 Cannaatiaut 115 11111 2 5 1115 8 185 111 1115 Dciawara 115 11111 2 5 1 1111 8 1 611 1 8 1 E 5 11C 1 11111 2 11111 S 1 S 5 111 1 B39 5 1 11 1rida 1 5 11111 5 1 1111 1 511 111 1 11 5 Gdurgia 115 11111 3 1 1111 1 3 3 1 3 1 1391 5 11awaii 4 1 1111 8 1 8 5 111 1 11 5 Idaha 1 1395 8 133 111 133 111i mi 1 11111 3 13 3 111 1 3 3 Indiana 3 1 111 S 1511 111 1 511 luwa 115 11111 55 185 8 185 111 13985 Kansas 5 1 511 1 511 111 1 511 Kamucky 15 11111 2 125 8 185 111 135 Luuisiana 6 11111 S 133 111 133 Maine 5 135 I 185 13 1115 hiaq39laiid 115 11111 61 135 185 111 1115 Mass achuacua 115 11111 5 185 8 185 111 21111 Another example of how policv varies across states Gruber TPA 1997 Table 3 Eligibility by Sun Children 1544 Women State 1934 1992 Diff Slate 1984 1992 Diff AL 0111 0252 0141 AL 0150 0528 0373 AK 0179 0208 0029 AK 0193 0306 0114 AR 0163 0292 0129 AR 0129 0453 0324 CA 0294 0406 0112 CA 0262 0510 0243 CC 0073 0211 0138 CO 0116 0379 0264 CT 0122 0273 0151 CT 0140 0341 0171 DE 0133 0198 0065 DE 0057 0373 0316 DC 0427 0474 0047 DC 0333 0494 0162 FL 0116 0337 0221 FL 0127 0492 0364 GA 0120 0327 0207 GA 0121 0393 0272 H1 0126 0278 0152 H1 0205 0372 0167 1D 0065 0292 0227 ID 0115 0456 0341 1L 0204 0287 0083 IL 0193 0358 0164 1N 0039 0272 0183 IN 0050 0423 0373 A 0206 0266 0060 A 0233 0494 0261 Concern Federal mandates mean that states that started out at low coverage experienced larger increases in eligibility Are the policy changes exogenous Madam Eig39n minim I mkymkmm m wwlmnmrmmu umwhmg m Dmnamiu My mum at IMWIMH minimum uni mum m pww wi 11 cumuamm m p WWW Wm me llhm aim initiation m mm mm Helm Olll39mlu Economic Issues in Medicaid Expansions Sourcez Gruber T ox Policy and the Economy 1997 9 Ultimately we care about outcomes but many steps involved in getting there 9 Research is available at nearly every step Medicaid expansions Issues TakeUp I Trade off gain from Medicaid against stigma costs Moffitt 1983 I Why might overall takeup of Medicaid be low 0 low quality low reimbursement rates 0 Stigma I Descriptive evidence on takeup Note that eligibility is increasing faster than participation leading to a decreasing take up rate I Why would take up rate decline with expansions o Newly eligible have less to gain from new coverage I Less disadvantaged higher up income distribution I Less information not on welfare I More private insurance 23 of newly eligible have private HI 0 Descriptive evidence below Gruber Medicaid Note the difference in takeup for kids 23 vs women 34 Economic implications of Medicaid Expansions Crowd Out Does Medicaid crowd out private insurance Important for knowing expected impacts on outcomes and for distributional implications Outcomes Does Medicaid improve health outcomes 1Use of preventative care 2Mortality 3 Health status birthweight Efficiency gains Labor Supply Medicaid expansions loosen up the welfare lock staying on welfare to keep Medicaid coverage 9 reduction in welfare participation and increase in labor supply Medicaid Expansions Empirical Methods Consider K a Xl yMl 81 Where Y1 outcome of interest Mi 1 if on Medicaid Naive CrossSectional Estimator Participation Suppose you simply regress outcome on dummy for Medicaid participation I Takeup ll1 is correlated with unobservables such as taste and demand for healthcare Naive CrossSectional Estimator Eligibility Suppose you replace participation M with eligibility E for Medicaid K a Xi 7El 81 I Eligibility is related to other factors leading to bias 0 nonlinear function of income family structure 0 may be endogenous EX Having a sick child leads to lower family income constrains work options and high use of services 0 May be correlated with stateyear trends 6 g recession Simulated Eligibility I Take a national sample of kids women I In each state in each year for pregnant women and by age of children calculate the eligible I Use national sample to avoid possibility that state demographics re ect policy somehow I Eligibility varies by year state age of child highly nonlinear I This instrument parameterizes the state Medicaid generosity using the observed density of distribution of eligibility yariation income age of kids I This can be done BY age group I With this variation you can control in the regression for 0 fixed year effects 0 fixed state effects 0 fixed effects for child s age 0 stateyear effects 0 statechild age yearchild age I Note not all of these controls have been used in all of the studies Application 1 Medicaid and Crowd Out I Large of newly eligible for Medicaid already have private insurance I Analyze decision making of family choosing between Medicaid and some form of private insurance I Assume that private plans are more generous meaning that they feature more providers more services better services Case 1 No government program PH Q People with a preference for health insurance HI will select into D more generous rather than B other goods Case 2 Government program with generosity M Take it or leave it public program value M Predictions 0 Introduce M Those with a low valuation of insurance take M others stay with private insurance D or E other 0 As value of M rises More goods people will move out of private insurance and into public Crowd Out Mechanism Matters 0 Most insurance is through employer 0 If cost of insurance is passed on to worker Gruber AER then question arises as to whether group insurance or individual insurance matters 0 If pass through is at the group level then if you switch to Medicaid there is no compensating increase in wage 9 less attractive to switch out of private 0 But if you already pay out of pocket then expansions in M are still attractive Employers may respond to the expansion by decreasing generosity of their private insurance plans for employees to keep costs down Cutler and Gruber Does Public Insurance Crowd Out Private Insurance QJE 1996 First to examine crowdout issue Data March CPS 19881993 lmputed Medicaid eligibility for women amp children using income state laws and child s age Women ages 1544 childbearing age pregnant women eligible Descriptive data I Table 2 demonstrates the large scope for crowd out in expansions many have private insurance 20 mm m Unuwld pmquot mm and uninsured mm m In Percent ulClmdmn wilh Diman mm oflnnumncx Pmmw insured was was man my we Year mem Hb Percent anmnan 1544 wh Different Tyves of Inaumnce oz 075 mm mm m 0 m saw was was ago 199 we mm mm Pnccnl 7mm Mwnnu pm Decrease in private health insurance and an increase in Medicaid Couldbe other contaminating factors so must look closer Model I Use simulated instrumental variable method described above 0 Addresses endogeneity of eligibility o SIMELIGist instrument is mean eligibility using a national sample in state s at time t 0 estimate model separately for women and for children I Controls 0 race sex and age single year of age dummies for the child 0 marital status of woman household type number of workers one vs two parent family 0 state and year fixed effects 0 text says adding stateyear and ageyear fixed effects do not change results not included in paper Should also have stateage I Dependent variables 0 Medicaid coverage expect T 0 Private insurance expect i o Uninsured expect i 22 Results Table 4 IV Estimates KIDS Takeup 0f 24 if ELIG 1 for kids Crowd out For every 10 percentage point increase in eligibility 07 percentage point decrease in private insurance 12 percentage point decrease uninsured crowdout 074 2 31 Crowdout estimate takeup 024 WOMEN 0 increase in Medicaid Troubling implies very large crowdout gt100 since private insurance declined 23 TABLE N Emmamm ExpMINING CILWEMGE ml Win1w um Gulwuun h dmm Women Independent variable Medicaid Private Ude Marniraid Private Uninmmed Eli zlr Eur Medicid 0235 44174 41119 INS 0IIHE 0047 mam 1W1 IHIIEJ 101313 003 MISS Demg m hm Mal1 11W UM 3 il LMl manna 0001 Mite 01mg 01151 IDIIIQ quot 53 0036 quot 0 m 111102 mma El1102 MW 003 0 WEI Married 118111 Uu 0016 comm 0014 MAME Hg th Inuit h umber ul penny 01325 LIME 0010 0051 39D3E DW3 0110 DAME KIWI 11002 0002 E g 39 l MEMTEszlE II 151 01116 i056 Ilt135 136 392139 01106 I1 Ell 01M mom l 39l 0003 MW 24 Table 56 Explore employer provided coverage since that is likely source of decline in insurance Possible channels I employer reduces generosity of coverage as Medicaid expands I workers decline coverage from employer I workers keep own coverage but drop children from plan Use data from special CPS on fringe benefits Fewer observations Table 5 Employer offered insurance no change lndividual takeup of insurance conditional on offer i Table 6 Shows that dependants coverage falls with expansion lncrease not significant for adult reduction in cost of family coverage Response is all on the individual side 25 HIU dollars of household s expected health expenditures that are covered by Medicaid eligibility not actual coverage Instrumented with SIMELIG analog TABLE v March annual Emylvyee bene ts supplements demographic supplement Coveng fmm Employer Tukeup 1 Coverage rmm Independent van nblc emplnycx offers o md employer in nr HIU dullnrs 41133 002 was a115 0107 0039 10127 10033 Summug statistics m 03112 0301 1189 0355 Dependent vnn39uble menu 0676 0808 0837 0598 38460 38528 31201 391351 an ample u wrin 16rd 395 Mn N M a nhud39 Inf 391 quot M J la n A v I n n n TABLE VI THE Dun 1 w MEN wmr DEPENDENTS POTENTIALLY Eusxsu FOR MEDICAID Whn covered in fam Any employer y Independent variable coverage Individual Dependents of HIU dollars 0066 0237 0065 0043 0069 Summary statistics R2 0299 0074 0273 Dependent variable mean 0689 0081 0608 N 119000 119000 119000 39 26 Y m nf annual s as in Tnlsle v Dam are weighed 11 nahonal cams Smndnrd camps are corrected fur heuroskedunieicy Issues CPS insurance information changes over time Analyses using SIPP show much smaller effects Show results using stateyear effects include other FE No placebo effects the estimates are applied to a sample of ALL educationincome levels Maybe should estimate model on a group not expected to respond to the policy change 27 Currie and Gruber Health Insurance Eligibility2 Utilization of Medical Care and Child Health QJE 1996 The justification for expanding Medicaid and subsequently SCHIP is to increase child health outcomes Yet there is little evidence on how Medicaid affects outcomes Why might it not matter Takeup low Poor live in underserved areas Increased utilization does not impact health outcomes 28 Data NHIS National Health Interview Survey 19841992 I 30000 children in the sample each year I Utilization of medical care in past 2 weeks and past year I Weaker demographic characteristics compared to CPS 0 income is in brackets and sometimes missing 9 imputing ELIG is not clean 0 solution use CPS to impute missing values randomly chose income value in range better to use CPS to inform this choice Outcomes I child did not see doctor in past year pure preventative care measure I doctor s visit in past 2 weeks physician s office ER outpatient other clinic I hospitalization in past year can be confounded by morbidity 29 Model I Use simulated instrumental variable method described above 0 SIMELIGgst instrument is mean eligibility using a national sample in state s at time t for age group g I Controls 0 race sex and age single year of age dummies for the child 0 income female head number of children education siblings other relatives central city 0 state and year fixed effects age 5 groups year fixed effects age state fixed effects 30 Results Table 4 Water Coe icients are multiplied by 100 I Control variables show that utilization is higher for Whites higher education first children smaller families female heads outside rural areas higher income for doctors Visits and lower income for hospital Visits I Medicaid expansion 0 Significantly reduces probability of going Without a PHY Visit in the past 12 months by 50 o Smaller not significant increase in 2week PHY utilization 0 Large increase in hospital Visits Table 5 Examining utilization by site of care DR office ERClinic other Increase in 2week utilization is concentrated in DR office Impacts on child health I Child health only measure in NHIS is self reported health 0 No results reported but footnote says no sig effect 31 0 They claim this measure is not good because it captures true health and reported health 0 Weak argument they should present results I Vital statistics is used to examine child mortality 0 data is at stateyearage 2 agegroups 0 IV as above CPS data used to create ELIG and SIMELIG 0 Results show decline in mortality concentrated With internal as opposed to external Violence etc causes This is consistent With expectations Related Study Currie and Gruber Saving Babies JPE 1996 Examine impacts on utilization of prenatal care NLSY and birth outcomes Vital statistics 32 LECTURE SOCIAL INSURANCE HILARY HOYNES UC DAVIS EC23O OUTLINE OF LECTURE 1 Introduction and definitions 2 Justification for Government Involvement 3 Optimal benefits U1 Social Insurance Features of social insurance programs Compulsory yes Contributory payroll taxes Provides benefits triggered by an event old age disability unemployment death workplace injury Eligibility is tied to prior labor market experience yes Benefits tied to previous labor market experience yes For reference this is in contrast to public assistance programs Where Compulsory no contributory no eligibility limited by prior labor market choices no benefits related to prior labor market choices no means tested yes Motivation for social insurance program to provide for insurance against shocks to income Facts about social insurance 2005 Federal outlays 20 GDP Total federal outlays 2472 billion Total federal outlays distribution Medicare 12 Social Security 21 National defense 20 Interest 7 nme 1 Social lnsunncr swam ass new 0 Percent a Percem cam mm mm Gum M cur a2 yes swam ms in am am lt25 2 a Ca v w 4231 in UP 5 may a 52 2 a 32 7 Um 2 59 31 mm 5 19M Cz squot mm H a as at a9 is a f f e 3 f fquot at 99 f a a 9 6 3 Krueger and Meyer 2002 Questions in social insurance 1 Justification for social insurance Why insurance Why social 2 Optimal social insurance how much to have given that we have it Trade off between protection and distortion Protection benefits of program are to reduce uctuations in consumption protect against poverty and reductions in standard of living Distortion changes in incentives for workers and firms that lead to inefficient outcomes and deadweight loss Moral hazard Empirical literature examines both protection and distortion Justification for Governmental Intervention in Social Insurance Why insurance Individuals are risk averse prefer to smooth consumption They benefit from actuarially fair insurance Social insurance program Insures against Social Security Old age earnings capacity Disability Insurance Disability Unemployment Insurance Involuntary unemployment Worker s Compensation Work based injury Medicare Health shock Why government intervention and provision of these programs What is the market failure Reasons for government intervention in social insurance 1 Adverse selection informational problems incomplete insurance markets Rothschild and Stiglitz 1976 21ndividual optimization failures paternalism Myopia self control improper planning Role of information EX Consumption falls too much at retirement Banks Blundell and Tanner 1998 Bernheim Skinner and Weinberg 2001 EX People do not save enough Scholz et al JPE 2006 3Macro shocks 4Redistribution based on lifetime earnings rather than current year income Adverse Selection Basic result If we have asymmetric information insurer does not know your type then there is a market failure incomplete insurance Government can create pareto improvement because of mandatory participation We can think of this model to justify government intervention for any social insurance program But here let s consider it with unemployment insurance Model static one period model no savings can not tell who is which type 2 types of people Fraction f are high risk of unemployment H Fraction f are low risk of unemployment L Income w for both types if employed 0 if unemployed w0 pH gt pL probability of becoming unemployed Note no distortion labor supply is fixed Insurance contract is a 051052 Pay a1 if employed Receive a2 if unemployed Consumption With insurance contract then becomes w 051052 Expected utility Via l pluw a1 piu02 Equilibrium set of insurance contracts such that both types cannot find a better contract than the one they chose and all insurance providers are earnings zero profits Consider pooling equilibrium one policy for both and separating equilibrium H guys choose aH and L guys choose aL Full information every one equal w2 mum I Full insurance Earn expected income I pw in each state Pooling equilibrium No pooling equilibrium exists If high and low types are offered the same contract then low types Will be charged an unfairly high premium to be insured 17 Zero profit condition requires that 062 Tpal Where p is average risk ofbeing employed But since 5 gt pL then L is paying too much W2 39 1 is the proposed pooling equilibrium With zero profit But firm can offer Bwhich L prefers It follows that a firm can make a positive profit and equilibrium breaks down qunrz u Separating Eguilibrium Here we can get full insurance for H but underinsurance for L H can get full insurance because the L guys will never try to get it and if they do it only raises profits for firm L is underinsured because if fullinsurance then H will want this insurance and this will create negative profits Have to set L premium low enough so that H will not want it wI H fully insured at 05H L would prefer full insurance at 3 But H would like that too 9 neg profit Best pooling eq is DZLJZH Incomplete insurance for L But someone could come in and offer y which can draw L and H away from pooling mmm Then this fails Government intervention Key is that they mandate participation No one can opt out Government can find a program that is a pareto improvement over the separating equilibrium One possibility is y from above Both L and H better off then separating eq L are still not fully insured If there are relatively few H guys then low risk guys benefit from pooling with them But as above this will not work in private market because someone will try to enter and steal the L guys away leading to a death spiral Individual Optimization Failures People should self insure against these shocks But they do not seem to be doing so at an adequate level Clearly if people misperceive the risk of unemployment or other SI shock then mandatory government program can help Still undeveloped area but growing behavioral public finance Macro or aggregate risk Insurance firms want to diversify risk so that in any given period L can payoff H Unemployment is clear macro risk hard to diversity old age risk can be thought of as demographic swings baby boom less relevant for disability and worker s comp Where the risks are more idiosyncratic Optimal social insurance In Rothschild amp Stiglitz we have that perfect insurance is optimal But that is all about protection and ignores distortion Distortion if you are perfectly insured then you will never work So to get optimal social insurance we have to balance protection with moral hazard Growing research area We will focus on optimal insurance in case of unemployment insurance First a bit of background on unemployment insurance Unemployment Insurance Overview Started 1935 social security act State provided Benefits Replacement rate about 50 of preunemployment wages but varies a lot across states is subject to min and maX benefit Duration of benefits 26 weeks most states can be extended by federal government by another 13 weeks median spell duration is 2 months Eligibility covered employment about 97 of workforce in eligibly jobs must satisfy work history requirements weeks at job high enough wageearnings not eligible if voluntary separation quit misconduct fired not able to work disabled typical eligibility work 2 out of last 4 quarters have to be looking for work Takeup not automatic only about 12 to 23 of eligibly unemployed takeup benefits Financing I financed by employed based taxes I Experience rated tax system 0 tax rate applied to each indiVidual firm is related to preVious benefits collected by preVious employees of the firm 0 Implies that there is a w to laying off a worker increases future taxes 0 US is only country in the world that experience rates unemployment taxes 0 lessens moral hazard problems on firm side I Firm tax rate is a facct balancetotal payroll subject to min and max rates 0 Acct balance taxes paid by firm benefits paid to laid off workers System is imperfect experience rated pdV future taxes ltpdV U1 benefits due to min max tax rate no interest charged etc Optimal U1 Benefits Possible distortions in U1 Longer unemployment durations Greater use of j obs with high unemployment risk landscaper in Wisconsin Greater shirking risk of job loss less severe Less savings Protection in U1 Smoother path of consumption Baily JPUBE 1978 did seminal work on this Chetty JPUBE 2006 updates Baily Baily s model all workers are identical firm perfect experience rating no distortions to firm behavior two periods work in period 1 some unemployment risk in period 2 utility is function only of consumption so when unemp still active labor market policy so little leisure fixed probability of job loss Chetty s extension of Baily still maintains the following partial equilibrium no GE effects on wages no externalities in search no crowdout no distortions to firm behaVior layoffs Basic Baily model Representative agent comes to period 0 With assets A0 Faces risk of job loss of probability p earnings are w or 0 If unemployed must search for job Can control unemployment duration D by varying search effort L11D concave increasing function captures leisure value of unemp benefits of improving job match With search Budget constraint has to hold Within each state UI taX rate While working of I benefit While unemployed is b cg cu consumption in employed and unemployed states Individual s maximization problem Max puce Pu0u L1 D sf A0 w r ce 20 A0 bDwl D cu 20 Gives the value function as a function of b Vb Social Planner s problem is to choose b that maximizes agent s utility While satisfying balanced budget constraint 11 1le Vb st l pr pr same structure as optimal taX problem 20 Solution Optimal benefit b is implicitly defined by Ac 7 b s 80 0 Ac 0 c e 2 change in consumption With unemployment protection 0 ce 2 uTcece coefficient of relative risk aversion value of greater smoothness u cg d 10 D 8D b 2 d1 g b elast1c1ty of duration wrt benefits distortion 0g marginal social benefit of search equal to MSC of search Empirical literature focuses on estimating these parameters Least amount of work on risk aversion 21 Many people have extended Baily s model and further explored the implications for optimal benefits Chetty JPUBE 2006 derives a model that yields a reduced form expression for optimal benefits as a function of the estimable elasticities 1 shows that Baily expression for optimal benefit level apply more generally then previously thought depends on three critical parameters 1 results hold under borrowing constraints endogenous spousal labor supply leisure benefits of unemployment 1 makes the point that the optimal benefit formula has parameters given current U1 system capturing the direct and indirect channels gtEX 8m captures full effect of b on D So if a higher b means less private savings which feeds back into D 2 further this means that the formula is the same for a richer model eg bringing in leisure gtGreater leisure value raises b through Acc greater leisure value means willing to tolerate a greater fall in c 9 higher b But conditional on knowing Acc and 8 then leisure has no additional value in optimal tax formula because they are already taken into account in agent formula 22 LECTURE TAX SALIENCE AND BEHAVIORAL PUBLIC FDIANCE HILARY HOYNES UC DAVIS EC230 Papers Chetty Looney and Kroft Salience and Taxation Theory and Evidence Amy Finkelstein EZTax Tax Salience and TaX Rates Motivation and Context Tax Salience Tax policy a is more salient than tax policy I if calculating the grossof tax price under policya requires less computation than calculating grossof tax price under policyb Basic tax results relevant for this analysis 0 Incidence of a tax does not depend on Whether tax is levied on consumer e g added at point of sale or levied on firm eg posted price is inclusive of the tax 0 Behavior should respond to prices after tax Behavioral response should be identical to prices and taxes 0 Optimal tax resultRamsey Ruleinverse elasticity rule Tax more heavily goods that have a lower elasticity Bounded rationality Agents face a cost of processing information They therefore rationally use heuristics to solve complex problems These papers show that if there are costs of processing information then the salience of the taX can lead to the following results Salience of the taX can affect measured elasticities o If you are not aware that the taX changes does your behaVior respond 0 If the costs of processing information are large then you do not adjust in the same way 0 Optimal taX result Higher taxes being leVied on the less Visible or salient taxes lower elasticity 0 Political economy result preference to raise less salient taxes 0 Tax incidence neutrality no longer holds Deadweight loss taxes with small utility losses if ignored by indiViduals can still create large DWL overall Ultimately they still pay the taX and the DWL effects depend on HOW they adjust other spending and what the other spending is Connections to other literatures 0 IO There is empirical evidence on the differential responsiveness to different components of prices costs etc Examples include cost of appliance and energy costs car purchases and manufacturer rebates Taxes and the size of the government the less visible the tax 9 tendency to have a larger government 0 To my surprise this was a point made by the 2005 President s Advisory Panel on Federal Tax Reform in a concern to recommend a VAT Which was perceived as being less salient than an income tax 0 Liebman and Zeckhauser Schmeduling labor supply responds more to average tax rates than marginal tax rates These papers illustrate the kind of work being done by the best young people in empirical public finance a move away from the pure policy evaluation of the identification emphasized reduced form differenceindifference regression discontinuity literature instead use those methods to reveal something about behavior theory Much more connected to economics and economic theory Also part of emerging area of behavioral public finance Individual faces cognitive constraints in achieving true optimum when faces with a complex taX system Chetty et a1 Salience and Taxation Theory and Evidence What they test whether a commodity tax has a larger effect on demand if it is included in the posted price rather than added at point of sale Idea tax included posted price is more salient Examples of taxes that are included in price or not Included in price Not included in price Excise tax gas cigarette alcohol income tax Airline tax sales tax Their empirical evidence 1 Experiment in grocery store augmenting posted price to advertise tax inclusive price 2 Analysis of behavioral response of alcohol consumption to variation in excise and sales taxes Organizing framework for paper 1 x Ux 2a y 1b 3 2 goods x andy pprice of x y is numeraire x is subject to ad valorem tax ts so inclusive of tax price is pt 2 p1ts Economy consists of 6 consumers who maximize subject to the gross of tax price pt and l 6 who maximize subject to the pretax price of p This results in aggregate demand for x after approximation and logs of log pt6 a logp Q log ts Goal is to estimate 6 the fraction who take the sales tax into account Under canonical neoclassical model 61 and the elasticity of response to p and t is identical Em irical evidence 1 Groce store ex eriment Orig Tag Exp Tag mm mm x 194mm x107 Alter posted prices for 3 full categories oftaxable items in grocery store in No California over 3 week period Sales tax is 7375 0 Note they chose categories that were relative high price so sales taxes were nontrivial high elasticity so demand response would be detectable o Cosmetics hair care accessories deodorants o All items in category were treated 0 Data comes from scanner data sets which is used at lot in 10 and records all transactions in the store 0 Data 1 spans period before and after intervention 2 covers two stores chosen as control stores and 3 covers treated categories and control categories toothpaste skin care and shaving products Triple difference estimation Treatment store Control Stores Before During Diff Before During Diff Treated Mi M A yfo y A products Control ydo Mi A3 yoco Mi A3 products Diffdiff for T store AIT Ag Diffdiff for C store AIC AOC Differenceindifferenceindifference AIT Ag AIC Ag Why triple difference What does it capture trends in treated category trends in store over time trends Identifying assumption no specific shock to treated category in treated store Back to framework model what does this experiment capture In control world we have the following aggregate demand log pt6 a logp 6 logl ts In the treated world everyone knows posttaX price costlessly so 61 log ptl a logp logl ts So the change in log of X with the treatment is 1 6 logl ts where B is the price elasticity of demand If everyone responds to posttaX prices then the effect of the experiment should be zero If everyone responds to pretaX prices posted prices then the response should be equivalent to the behavioral response of an increase in price of t TREATMENT STORE Penod Control Cate ONES Trea ed Cate ones vaference thtle change for the Easehne 20 48 2517 4 31 control store good 200517 0 22 0 37 0 43 N 2006 5 5510 754 6264 SO DDD DD Expenmem 27 32 23 07 73 45 2006 B 0 87 102 0764 Us1ng an elastlclty of 2006 10 285 39 324 159 estlmated 1n Dwfference 0 84 71 30 DD 214 over me 0 75 U 92 El 64 separate regreSSIOH 57951 7931 6588 you get an estlmate of CONTROL STORES 0 03 5 Penod Control Cate ones Tremed Cate ones vaference Easehne 30 57 2794 2 63 200517 0 24 0 30 032 2005 5 I 1020 1500 12528 Experiment 30 76 2819 2 57 2000 87 0 72 106 109 2000 10 570 78 648 Dwfference 0 IS 0 25 DDCS 006 over 0 e 0 54 0 92 090 11590 1586 13176 DDD Estimate 220 053 19764 12 Conditional DDD Note to calculate a triple difference you need to control for Main effects store category time 2way interactions storeXcategory storeXtime categorthime 3way interaction store X category X time treatment effect Controls price time Effect of Poslmg Taxelncluswe Pnces Regressmn Estima es Quanmy per Quanmy per Revenue per Log quanmy Log revenue Dependenlvanable Categow Categow Category 3 per category per Categow 7 Z 3 4 5 Treatment 220 220 1 12 43101 41123 0 5mm 0 59Vquot 4 88Vquot o myquot 0 04 Average Pnce 73 15 73 24 0 25Vquot 1 74 Average Pnce Squared u 05 u 06 0 ourquot 0 03quot Log Average Prrce 59 VD 39 0 mm 0 m Category smre Week FES x x x x Sample srze 19764 19764 19754 15327 15327 Estimated elasticity from this model well identi ed Empirical Evidence 2 State excise taxes on alcohol Alcohol is subject to two state taxes Excise tax imposed on Wholesale price fixed volume tax Sales tax added at register Idea is to use variation across states and over time in typical state panel identification model to see if consumers respond differentially to the more salient excise and less salient sales taxes Back to basic model log22tEtS6 a log tE 6 logl ts B elasticity on excise tax related price changes 6B elasticity on sales tax related price changes Model is estimated in changes since autocorrelated taxes state j period t Alog a0 Alogl ti 6 Alogl tit X p g Data 19702003 Excise taxes are higher 64 than sales taxes 43 They are also more variable but also change in nominal terms less frequently Not a super great identification strategy lot s of trending in both RHS and LHS variables Secular year fixed effects included but no state specific time trends Little attention to exogeneity of tax changes Effect 0 EXCISE and Sales Taxes on Beer Consumpt on Dependent Variable Change m Logper caplta beer consumption Easehne Bus Cycle Bus Cyc e Lag ch Regulations 2 3 4 ALog1Excise Tax Rate 037 031 4336 039 0 mm 017 0 mm D mm ALog1Sales Tax Rate 020 000 005 mm 0 3m 0 30 u 30 u 30 ALogPopulauun u 03 in D7 0 05 yo D7 0 06 0 07 u 19 u 07 ALoglncome per Caplia u 22 u 18 u 22 0 05 0 05 D 05quot ALogLInempxoymemRate in D1 in m r m 0 0U U 0 0 0 1 Lag Bus Cycle Controls x A cohol Regulation Controls gtlt Year Fixed Effects x x X X FrTest for Equahty of Tax 1 Vanables PwobgtF 0 05 D m U 039 D 039 Samp e Swze 1607 MB 1440 1487 Much iarger response to more salient tax 6006 Bottom line Both approaches show that there is a larger response under salience They conducted a small survey at same grocery store showing that this is not due to people simply not knowing about the sales taX either what it applies to or how large it is Finkelstein EZTax Tax Salience and TaX Rates 0 The setting is the introduction of electronic toll collections on US roads tunnels and bridges She collects data on tolls traffic and the timing of introduction of toll collections in 123 of the 183 sites with tolls in place in 1985 0 She examines how the introduction of ETC affects o Tolls analogy to size of government or taX rates 0 Elasticity of road use to toll price 0 Evidence is compelling that taX rates rise when less salient taX is created tolls rise With ETC introduction Idea once you use electronic payment you no longer pay attention to the toll amount Government Objective Function maX SWF of indi utils j by choosing taxes mfaX Zujwj 1 1 09M Leads to the inverse elasticity rule 1 EU 3 the taX rate on group j is decreasing in the Tj l j j j j demand elasticity the MU of income 9 and the social welfare Weight p 239 Ej j 1 zu j J J Prediction of the model for electronic toll introduction With electronic toll the taX is less salient therefore a change in the taX toll Will lead to smaller changes in behavior driving elasticity falls this Will lead to an increase in the toll This analysis is predicated on the assumption that income effects of this change are small which makes sense since tolls are small share of individual spending and government revenue 20 Survey Evidence What is the role of this Her own survey of Mass Pike drivers also a commuter survey in NYNJ Shows that those using ETC are 0 Less likely to know amount of toll they pay could this just be higher income folks are more likely to have ETC and are not paying attention Figure 1 Distribution of ETC Start Dates Adoption of ETC Across states Frequancy 1O 5 Basic Model 2 y lETCAdopt zETCit 81 where ETC1 if electronic in place in year t ETCAdoptl if adopted this year Year controls y y Alogminimum toll minimum toll when ETC implemented sometimes there is a discount offered in the early years To get around endogeneity of this her preferred sample excludes the areas where the discount is used Since the model is in differences then the gammas capture average growth rates by year and the betas capture deViations from those growth rates Why identi an impact offhe year of adoption why include ETCAdopt Source of identification in the model 22 ETCAdopt is used to capture the initial impact of the introduction maybe with a discount The identification is a simple DD model comparing changes across areas with and without ETC Identifying assumption ETC are not endogenous places with ETC implemented are on same trend line and placed wo ETC ETC implementation is not correlated with changes in toll setting relative to its norm Details in estimation weighting she weights 49 operating authority equally Why does that make sense Why not weight by usage she clusters on state why 23 Tuhle 2 Toll Rates A Log Mix A 0g A Log A Log A log Min A log Mm Tall Manual Toll 1011 1011 Toll Insulng ETC 3 leads to 750 I I 3 4 5 61 0 ETC 0 015 0 00 0 024 more increase 0 006 0 005 o 011 0 m 0004 o 051 in tolls 1520 AETC 0 623 7 compared to Fenetmnon 0 35 o 252 0 161 0 044 o 045 0 057 average 39 0 ETCAdopl 051 716 70 033 o 051 70105 70 097 Increase Of 2 Aquot 0 035 0 032 0 019 0 035 0109 0105 0155 0 521 0 097 0166 0 348 0 320 IV t t Menu nap my 0 010 0 02 0 017 o 017 0 020 o 020 39 ms rumen of slate 2 4 17 17 4 14 Wlth ETC op aurhm 49 49 31 31 49 49 famine 123 123 70 70 123 13 penetration Wlth 079 5079 2275 2751 4x15 4815 Esmuanon LS OLS 0L8 LS 1v 1V dummy for Sam le No ETC Na ETC ETC resmmiou dISCOnnt hitounl introduction Overall diffusion of ETC to steady state 60 leads to an percent increase in rates increase of 2040 Flgure 3A Full Sample Figure 35 Balanced Panal ETC Penetration a R 9 Ram Axis Frc Yaur Examine timing of ETC using event study Look for pretrend to be at note text around balanced panel Examine how treatment effects change increase with time since treatment Impacts on traffic Alogtm z c yt lA logmin telly 2A10gmintoll Never ETC AA logmin toll ETC penetration NeverETC ETCpenetratz390n 8 I think we know that the price elasticity of demand for gas and driving has decreased over time I think that means we want to control for mintolltime in the regression C 26 Table 5 The elaslirih nl traf c will I espem 10 I011 3 4 5 6 A log mm lollu 70 049 70 058 70 051 70 057 70 052 70 160 0 015 0 018 O 019 0 017 039 0 037 Base 0 004 0 008 0 009 0 006 0145 0135 e1ast101ty A log mm 011 4 O 154 0141 0 ETCiPeneli alwnl 0 035 0 076 0 0051 0 091 L er if A log mm 1011 0 o 00 erg ETCiYear 0 001 0 00 never ETC 0 0011 00611 00610071 A log mm c111n 1 70 009 70 005 Nee1 ETC 0136 0131 0 2 9 0 05 0 611 0 533 0 966 0 97s Smaller 1f ETC Mean dell Va 0 049 0 042 L 043 L 040 0 039 21 12 12 12 1 makes sense 0 urhors 32 1e 1e 15 s1nce less Lr offncllmes 76 33 33 33 33 39 N 1100 727 571 293 305 S le Sample No ETC No ETC N0 ETC No ETC N0 ETC resmmou5 dismums disconms discoums disconms dismums Overall though really small elasticities Political Economy baseline assumption is that legislators do not want to increase taxes in election years if less salient taxes do not change behavior people are less aware of the taX changes then there should be less of an election year effect With the less salient tax 2 2 yr lETCAdopt ZETC 3lEZecYearst 4lEZecYearst ETCAdoptit 51EZBCYBClI st ETC 8 Expect B3negative fewer toll increases in election years BSpositive less negative due to less salient 28 Table 6 Mm Toll 0g Mm r611 Lug Log 16 To A Log Mm run A 1 g 21 T1511 Mm T611 Raisem Min r011 R 159d Mm r011 Rmsed M111 7611 Rmse Mm To Raisedquot l 139 3 J 5 5 7 S 9 0 ETC 0 015 0 073 0 016 0 074 0 016 0 074 0 044 0006 0 044 0 006 0 024 0 006 0 024 0 006 0 024 0 022 0 009 0 022 0 010 0 006 0 017 0 006 0 016 0 005 0 04 0 494 0 042 AnyElec 70 015 r 02 r 029 Year 0 006 0 011 0 010 0 017 0 023 0 003 GovElec 70 017 V0 036 70 016 70 036 yearn 0 006 0 013 0 005 0 012 0 011 0 010 0 001 0 002 LegOnly r 013 70 014 70 015 70 021 ElecYEMS 0 007 0 013 0 005 0 012 0 064 0 263 0 005 0 005 AnyElec 39 Bang 0 027 ETQL 0 041 GovElec 0 004 0 016 Bang 0 014 0 033 4H5 0 791 0 617 LEgOuly 0 030 0 094 ElecYem J 0 014 0 033 2FTFH 0 012 0 005 Indeed under ETC there is less of an election year effect Less sensitive to electoral business cycle


Buy Material

Are you sure you want to buy this material for

25 Karma

Buy Material

BOOM! Enjoy Your Free Notes!

We've added these Notes to your profile, click here to view them now.


You're already Subscribed!

Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'

Why people love StudySoup

Steve Martinelli UC Los Angeles

"There's no way I would have passed my Organic Chemistry class this semester without the notes and study guides I got from StudySoup."

Janice Dongeun University of Washington

"I used the money I made selling my notes & study guides to pay for spring break in Olympia, Washington...which was Sweet!"

Bentley McCaw University of Florida

"I was shooting for a perfect 4.0 GPA this semester. Having StudySoup as a study aid was critical to helping me achieve my goal...and I nailed it!"

Parker Thompson 500 Startups

"It's a great way for students to improve their educational experience and it seemed like a product that everybody wants, so all the people participating are winning."

Become an Elite Notetaker and start selling your notes online!

Refund Policy


All subscriptions to StudySoup are paid in full at the time of subscribing. To change your credit card information or to cancel your subscription, go to "Edit Settings". All credit card information will be available there. If you should decide to cancel your subscription, it will continue to be valid until the next payment period, as all payments for the current period were made in advance. For special circumstances, please email


StudySoup has more than 1 million course-specific study resources to help students study smarter. If you’re having trouble finding what you’re looking for, our customer support team can help you find what you need! Feel free to contact them here:

Recurring Subscriptions: If you have canceled your recurring subscription on the day of renewal and have not downloaded any documents, you may request a refund by submitting an email to

Satisfaction Guarantee: If you’re not satisfied with your subscription, you can contact us for further help. Contact must be made within 3 business days of your subscription purchase and your refund request will be subject for review.

Please Note: Refunds can never be provided more than 30 days after the initial purchase date regardless of your activity on the site.