1 Journal of Health Economcs The demand for prvate health care n the UK Carol Propper ) Department of Economcs, CASE and CEPR, UnÕersty of Brstol, Brstol BS8 1TN, UK Receved 9 November 1998; receved n revsed form 2 March 2000; accepted 3 March 2000 Abstract Polcy change has eroded the enttlement of UK resdents to free state-provded health care, wth a resultng rse n the use of the prvate sector. Ths paper examnes the choce between publc and prvate health care. It models the use of prvate health care as a functon of ts costs and benefts relatve to state care and no care. The results ndcate a dfference between users of prvate care and other care, and the mportance of past use as a predctor of current use. But they also show consderable movement between the publc and prvate sectors, ndcatng a complex relatonshp n publc and prvate sector use. q 2000 Elsever Scence B.V. All rghts reserved. JEL classfcaton: I1; H4 Keywords: Prvate health care; Publc health care 1. Introducton Health care n the UK s predomnately state-fnanced. However, whle approxmately 85% of fundng comes from the publc purse, the use of prvate health care servces s rsng. Whle poltcans have stressed ther commtment to tax-fnanced free hosptal care, polcy change has reduced elgblty for publcly provded treatment, ncreased copayments for dental, ophthalmc servces and pharmaceutcals, and reduced the payments made to ndependent contractors who provde state ) Tel.: q ; fax: q E-mal address: C. Propper r00r$ - see front matter q 2000 Elsever Scence B.V. All rghts reserved. PII: S X
2 856 ( ) C. PropperrJournal of Health Economcs fnanced dental care. These changes have been accompaned by a growth n the mportance of the prvate sector n the provson of health care n the UK. Ths growth could affect publc provson Ž the NHS. n a number of ways. Frst, reductons n the avalablty of free care, partcularly reductons whch may not have been the ntended consequence of polcy, may affect whether NHS equty goals are beng met. Second, the demand for prvate sector servces affects the publc sector. As n many health care systems n whch the prvate sector operates alongsde a larger publc sector, labour n the NHS s also employed n the prvate sector, often smultaneously. In the short run, a sgnfcant expanson n prvate demand would reduce the avalablty of staff to the publc sector and so reduce the qualty of publc sector servces. More subtly, an ncrease n the use of prvate servces may be accompaned by a decrease n the support for, and wllngness to pay, taxes for the publc sector. Hgh prvate usage leadng to lack of voce and taxpayer dscontent could lead to the evoluton of the NHS nto a poor servce for the poor. Ths last effect s lkely to be less mportant f ndvduals who use prvate servces contnue to use the NHS at the same tme and f the use of the prvate sector for one type of servce s not lnked to use of another. Gven the pecemeal nature of polcy change on the use of prvate fnance n UK health care, ths s perfectly possble: ndvduals may use both publc and prvate servces and may retan an overall strong commtment to the state fnancng of health care even though they are prvate servce users. On the other hand, the changes n polcy may have led to the development of a group of users who demand almost all ther health care from the prvate sector and who have lttle commtment to publc fundng of health care. A small body of research has dentfed the mportance of ncome and poltcal atttudes n the use of prvate health care servces ŽPapadaks and Taylor Gooby, 1987; Taylor-Gooby, 1989; Calnan et al., Propper Ž 1989, found ncome and poltcal belefs affected the decson to buy prvate medcal nsurance. Besley et al. Ž show a lnk between medcal nsurance purchase and qualty of NHS servces. Ths research has not been able to establsh whether the dstnct economc and socal profle of prvate sector users s smply due to ndvdual fxed effects or whether changes n ncome or atttudes would ncrease prvate sector use. Nor has any research examned the dynamcs of prvate sector use: whether use s related to past demand, or whether current demand for one prvate servce s accompaned by prvate demand for another. Ths paper focuses on these ssues. It estmates a model of the use of health care servces whch takes nto account the choce a potental user of care n the UK has among the prvate sector, the NHS, and no care. It examnes not only hosptal and physcan use but also the use of dental and other servces based n the communty. The estmated model allows past use to affect the costs and benefts of these alternatves. The data are from the Brtsh Household Panel Survey Ž BHPS., a natonally representatve survey of around 5000 households
3 ( ) C. PropperrJournal of Health Economcs ntervewed each year snce In terms of health care, and partcularly prvate health care use, t s a rch data set. It contans nformaton on medcal need, on measures of poltcal belefs about the role of the publc sector n the provson of health care, and on a large set of ndvdual and household characterstcs. In addton, t has data on health care utlsaton that dstngushes between use of NHS and prvate servces and, wthn prvate servces, between those that are publcly funded and those that are pad out of pocket or by nsurance. 1 The results ndcate that use of prvate care s strongly related to ncome, a set of dentfable ndvdual demographc characterstcs and poltcal atttudes. Prvate users are healther than ther NHS counterparts. Prvate use n the past s sgnfcantly assocated wth current use. Further, all these factors contnue to be sgnfcantly assocated wth prvate use after controllng for unobserved ndvdual effects. So, n ths sense there s perhaps a defnable prvate welfare class. On the other hand, prvate sector users do not lve by prvate use alone. The results ndcate consderable movement of ndvduals over tme between the publc and prvate sector. Current use of the prvate sector s postvely assocated wth both past and future use of the NHS. Current use of the NHS ncreases the chances of future use of the NHS, and for some servces, also ncreases the probablty of use of prvate servces n the future. So, n addton to an understandng of who buys prvate health care, the paper also contrbutes to an understandng of the lnks between the prvate and publc sectors n the UK. Whle there s consderable research on the lnk between publc and prvate health care sectors n the Amercan context Žfor example, Cutler and Gruber, 1996a,b., the amount of UK research s small. MacAvnchey and Yannopolous Ž estmate a cost shares model usng aggregate data n whch they fnd sgnfcant cross-prce elastctes between publc and prvate care, Martn and Smth Ž use ward level data and fnd an effect of watng lsts on demand for NHS electve surgery, and Besley et al. Ž use mcrodata and fnd a postve relatonshp between watng lsts and prvate nsurance. The results presented here show the patterns of assocaton between publc and prvate sector use to be qute complex. Prvate use n the past s more lkely to lead to current prvate use than NHS use s n the past, but prvate servce use n the past s also assocated wth NHS current use. The paper s organsed as follows. Secton 2 presents an overvew of prvate fnanced health care n the UK, and a descrpton of the users of ths care from the BHPS. Secton 3 presents a model of the choce among use of prvate care, publc care and no care, whch motvates the demand for prvate care where a free publc alternatve exsts. The econometrc model derved from ths model s then pre- 1 Ths dstncton s mportant as a sgnfcant component of UK government reform of the welfare sector has taken the form of contractng out of provson of servces to the prvate sector, whlst mantang publc fundng of these servces.
4 858 ( ) C. PropperrJournal of Health Economcs sented. Secton 4 presents estmates of the choce between the three alternatves. Secton 5 focuses on the effect of past use and controls for ndvdual effects. The fnal secton dscusses the results. 2. Prvate health care n the UK Prvate expendture on health care n the UK has grown from 9% of total health care expendture n 1979 to 15% n Ths fgure s made up of prvate medcal nsurance premums Ž pad for by employers or ndvduals. and out of pocket expendture on prvate medcal servces and goods. Out of pocket expendture ncludes payment for npatent prvate care, other hosptal servces, outpatent servces, dental care, eye care, copayments for prescrbed medcnes and over the counter medcnes. Prvate provson of hosptal servces has always exsted alongsde the NHS. The 1991 NHS reforms attempted to gve an ncentve for the expanson of such care through tax relef on prvate medcal nsurance for the over-60s, but despte ths, growth n prvate medcal nsurance has remaned slow durng the 1990s. On the other hand, polcy change n dental and eye care appears to have had a bgger mpact on prvate use. In 1985, NHS provson of glasses was restrcted to certan groups Žchldren, students under 19, ndvduals on low ncome, and users of certan complex lenses., and publc provson was replaced the followng year by vouchers. In dentstry, free dental check ups were restrcted to the same groups n Although NHS treatment s supposed to be avalable to all, copayments Ž user fees. have rsen. Changes to the level of fees pad to dentsts who provde NHS care has led to wdespread dssatsfacton amongst dental care provders, and to anecdotal evdence of a reducton n the dental servces avalable under the NHS. Ths paper examnes the use of prvate servces where an NHS alternatve exsts. In some cases ths NHS alternatve s free at pont of demand. In other cases, elgblty rules mean certan groups are not enttled to free NHS care but NHS provson, wth copayments for those not enttled to free care, s supposed to be avalable to all. The paper examnes any use of prvate care and then focuses on two specfc types of prvate health care servce: npatent hosptal care and dental care. Prvate npatent use s of nterest because NHS qualty measures, partcularly watng lsts, are thought to be mportant n determnng use, and f there s a lnk between these measures and prvate health care nsurance t s through the 2 expected use of prvate npatent servces. Dental servces are of nterest because, although all ndvduals are enttled to NHS dental care, t s argued the 2 Prvate medcal nsurance prmarly covers n- and out-patent hosptal servces.
5 ( ) C. PropperrJournal of Health Economcs Table 1 Proporton of sample usng prvate health servces n Brtsh Household Panel Proporton 1990r 1991r 1992r 1993r 1994r Prvate dental care Prvate eye care Prvate physotherapy, chropody or health vstor Prvate npatent stay Any prvate health servce use Base weghted avalablty of such care s beng reduced by supplers wthout any explct change n government polcy. Table 1 shows the patterns n prvate demand for the dfferent servces for each of the frst 5 years of the survey. Approxmately 16% of ndvduals n the BHPS sample use ether prvate dental, eye, hosptal or outpatent servces. The table ndcates a general rse n prvate servce usage across the 5 years of the survey, wth only prvate npatent use showng no growth. 3. The demand for prvate health care 3.1. The economc model Goddard and Smth Ž outlne a smple model of demand for medcal care where publc and prvate care exsts. Ths can be used to explore the mpact of ncome, prce and qualty of the publc alternatve, atttudes to the role of state n the provson of health care, and past use on the current demand for prvate care. The model follows Goddard et al. Ž n recognsng that an ndvdual has three dscrete choces: to seek no medcal care, to use prvate care, and to use publc care. These choces wll be affected by the severty of llness, the costs and the qualty of NHS care, and the costs and qualty of prvate care. For any ndvdual, ndexed by, let V be the beneft of prvate health care and p be the cost Ž relatve to ncome.. The exstence of the publc sector constrans the supplers of prvate care to provdng a servce whch s of at least as hgh a qualty as the publc alternatve: no one wll buy the servce f t s of lower qualty than the publc sector. Let ths qualty dfference be represented by a sngle
6 860 ( ) C. PropperrJournal of Health Economcs parameter t. At ts smplest, ths can be watng tme but can be thought of more generally as encompassng other qualty dmensons Žsuch as the provson of nformaton, or the ablty to choose the exact date and locaton of treatment.. Indvduals vary n ther valuaton of ths qualty by the parameter g.ift s thought of as watng tme, g can be thought of as the rate at whch the value of treatment decays because t s receved later rather than sooner ŽLndsay and Fegenbaum, The value of NHS treatment for ndvdual s V exp Ž yg t.. NHS care has no drect user prce but the ndvdual has to access NHS facltes. Let ths cost be c. Faced wth these costs and benefts, the potental user s ndfferent between prvate and NHS care when VyVexpŽ ygt. spyc Ž 1. s ndfferent between prvate care and no care when V sp Ž 2. and s ndfferent between NHS care and no care when VexpŽ ygt. sc. Ž 3. Ž.Ž. Eqs. 1 3 can be used to examne the mpact of changng parameters on the choce of the three alternatves. Decreasng p wll ncrease the use of the prvate alternatve and decrease the use of the NHS and no care alternatves. Decreasng c wll ncrease the use of NHS treatment and reduce the use of prvate and no care. An ncrease n g wll ncrease the use of prvate care, reduce the use of NHS care and ncrease the use of no care. The effect of an ncrease n t s the same. An ncrease n V wll ncrease the use of prvate care and NHS care. These parameters wll vary across ndvduals. The value of the benefts of medcal treatment V wll obvously be related to the severty of llness. In addton, as V s the perceved beneft to the user of care, t may also be a functon of the mportance of good health to the ndvdual. Ths s often argued to be postvely assocated wth educaton. Where t s taken to be the watng tme for NHS treatment, g s the ndvdual s valuaton of tme. Ths can be expected to be a postve functon of ncome and of type of employment. For example, those who are self-employed are less lkely to get pad whlst watng for medcal treatment. 3 Where t s the qualty of the NHS, g s the valuaton of NHS qualty. Ths mght be expected to be a functon of the ndvdual s general atttudes to the 3 For an emprcal estmate of the value of tme n a medcal context see Propper 1995.
7 ( ) C. PropperrJournal of Health Economcs NHS. For example, those who strongly value state provson of health care may be more tolerant of poorer qualty. Then those who hold such atttudes wll use more NHS servces for a gven value of V and t. The value of p wll be lower for those ndvduals who have medcal nsurance whch covers the cost of prvate care and lower for those wth hgher ncome Ž snce p s defned relatve to all other goods.. Snce c s the access cost to NHS facltes, t wll be lower, leadng to greater avalablty of NHS facltes. The nature of medcal care means these prces wll also be a functon of past use of servces. Frst, the prce of care n each sector ncludes the costs of search. In health care, consumers cannot always tell good from poor qualty. As a result, regulaton lmts advertsng and consumers are relant on ther own knowledge and that of frends, relatves etc. As a consequence, search costs for a suppler of care may be hgh. Second, nformaton asymmetres mean trust s an mportant component of the relatonshp between supplers and demander of care. Thrd, an ndvdual nvests effort n communcatng ther medcal hstory to a suppler of health care. So once an ndvdual has found one suppler, they may be less lkely to change to another. A consumer who has used the prvate sector n the past wll have lower search costs for prvate sector use and wll have made an nvestment wth a prvate suppler. Smlarly, a consumer who has recently used the NHS wll have lower NHS care search costs and wll have made an nvestment wth an NHS suppler. So the costs of care n each sector are lkely to be a functon of past use, and current use s lkely to depend upon past use The econometrc model At any tme t, the ndvdual chooses between prvate care, publc care and no care. She thus has J alternatves. Let js1 denote no care, js2 denote NHS care, and js3 prvate care. From the economc model of Eqs. Ž. 1 Ž. 3, the latent net valuatons assocated wth each alternatve wll depend on the characterstcs of the ndvdual Ž her valuaton of health states, of tme, of prce etc... The weght attached to each characterstc wll vary by alternatve. Allowng for random error, the latent net valuatons of the jth choce can be wrtten as: m ) sb X j j zqej Ž 4. where z s a vector of attrbutes of the ndvdual Ž whch may be allowed to vary. 4 by alternatve as well and e s a random error term. j 4 Ths model s of the form of the model of occupatonal choce of Schmdt and Strauss 1975.
8 862 ( ) C. PropperrJournal of Health Economcs If the ndvdual s observed makng choce j, t s assumed that m ) j has the hghest net valuaton. The statstcal model s drven by the probablty that choce j s made, whch s Pr m ) )m ) j k for all k/j. If the ej are assumed to be..d. wth Webull dstrbuton, then the statstcal model that results for the choce of alternatve s the multnomal logt model: J Ý X X b j z b k z Ž. Pr m sj se r e 5 ks1 where m s the observed choce of ndvdual. To estmate the parameters of ths model t s necessary to normalse and mpose the restrcton that one of the bjs0. In estmaton of ths model, I mpose the restrcton that b1 s0. From the dscusson above, the specfc characterstcs that wll nfluence the choce the ndvdual makes Ž.e. the varables n z. nclude ncome and health status, the costs of accessng each servce Žwhch wll be a functon of the costs of prvate care, past use, and the avalablty of servces. and the perceved qualty of the publc servce Žwhch wll be a functon of watng lsts and possbly atttudes to the publc sector role n fnancng health care.. In addton, there may be regonal effects. So the latent net valuaton of alternatve j wll depend on: m ) sf X,Q,m, R;e, js1,2,3 Ž 6. j r jty1 j where X s a set of personal and household demographcs, soco-economc varables ncludng ncome, and measures of the belefs ndvduals hold about the approprate role for the prvate sector n the provson of health care, Qr a set of ndcators of the qualty of NHS provson n the regon n whch the ndvdual lves, R a vector of regonal dummes, m jty1 s use of alternatve j last perod, and ej s whte nose error. The data set records use at fve tme ponts and of several health servces, denoted s Ž defned below.. So the data permt examnaton of the mpact of past use of one servce on another and of past use of one alternatve on another. To estmate these effects I pool the panel data and expand the set of varables n Eq. Ž. 6 to nclude these cross-servce and cross-alternatve effects and also to allow for tme effects. The latent utlty for alternatve j of servce s at tme t s thus: m s) sb X qb Q qb m qb m X qb Rqb Tqe, jt 0 t 1 rt 2 jty1 3 jty1 4 5 jt js1,2,3 Ž 7. where m and m X are now vectors of past servce use, j X jty1 j ty1 ndexes the other alternatves, T s a vector of tme dummes, and e s whte nose error. From Eq. jt
9 ( ) C. PropperrJournal of Health Economcs Ž. 5 the parameter estmates wll dffer across the two alternatves but for smplcty of notaton we have dropped the subscrpts on these. As the restrcton that the parameter estmates are 0 for alternatve 1 Ž no care. s mposed, parameters are estmated only for alternatves 2 and 3. In poolng the data and condtonng on nformaton at tme t, the assumpton made s that the errors are not correlated over tme for an ndvdual. In addton, the MNL model mposes the assumpton of ndependence of the errors across optons The data From the BHPS data I defne the use of three servces or sets of servces. The frst set of servces s defned as use of any of the servces gven n Table 1 Ždental care, eye care, npatent stays, certan communty servces 6.. If the respondent used any of these servces n year t and any of her use was prvate, she s recorded as havng postve prvate use n year t. If she used any of these servces and none of her use was prvate, she s recorded as havng postve NHS use n year t. The second servce s an npatent hosptal stay. If ths stay s prvately pad for, the respondent s recorded as havng a prvate stay. If the stay s not prvately pad for, then she s recorded as an NHS user. The thrd servce s dental care, where f the respondent has dental care and any of that care s prvate, she s recorded as havng prvate dental care. If she has some dental care and none of t s prvate, she s recorded as havng NHS dental care. 7 In addton to measures of health servce use and standard soco-economc and demographc data for each ndvdual n the household, the BHPS ncludes measures of current self-assessed health status, and detaled measures of longer term health status. Indvduals are asked whether they are lmted n ther daly actvtes ncludng work and lesure Ž ADLs.. If they have any lmtatons they are asked whether these lmt ther ablty to work. They are also asked what specfc lmtatonsrcondtons they have. The data also record whether the ndvdual s a smoker, and f the answer s postve the number of cgarettes consumed per day. Earler research on the demand for prvate health care n the UK has ether used no measures of health status Ž Besley et al., 1999., or rather more lmted measures Ž Propper, Here the BHPS data are used to control for a large number of measures of health status. In addton, condtonng on health status enables a 5 Whle ths s an unattractve feature of the MNL model, the rch set of regressor varables should reduce correlaton between the errors. 6 These communty servces are all servces for whch NHS provson exsts. The BHPS records use of other prvate servces for whch there s no NHS alternatve. As we are nterested n the choce between no use, NHS use and prvate use, we do not examne these servces. 7 Ths codng gves the broadest defnton of prvate use, whch fts our nterest n any use of the prvate sector.
10 864 ( ) C. PropperrJournal of Health Economcs reducton n the potental contamnaton of the ncome coeffcent, whch arses from the correlaton between health and ncome. To these data are matched regonal ndcators of the qualty of the NHS. These are the length of NHS watng lsts over- and under-12 months deflated by the regonal populaton Žrelevant for npatent care. and an ndex of dental servce avalablty Žrelevant for dental care.. 8 The prces pad by ndvduals are not observed so the mpact of prce s not examned. 9 As prce at pont of use for hosptal npatent servces wll depend on nsurance coverage but the BHPS does not record prvate nsurance cover, occupatonal dummes are ncluded as nstruments for corporate cover The determnants of prvate, NHS and no demand A multnomal logt MNL model of the use of publc and prvate care s estmated. Table 2 presents the estmates of the model where the omtted alternatve s no care. The table presents separate results for the three sets of servces defned above. Interpretaton of the parameters n an MNL model s not straghtforward, so nstead of coeffcent estmates, the table gves the estmated margnal effect of varable x on choce of alternatve j. 11 The nterpretaton of the margnal effect s the mpact of a unt change of the varable x on the choce of alternatve j. Table 2 presents the margnal effects for the choce of the prvate and the NHS alternatves. To show whch varables have coeffcents that are sgnfcantly dfferent from 0, the z-statstcs for the coeffcent estmates are presented as well. In general, the coeffcents are well defned and ndcate sgnfcant dfferences between users of the dfferent alternatves. The results for the use of any care show that users of prvate care and users of NHS care share some demographc characterstcs. Women are more lkely to demand care n ether sector than men, though the log-odds of prvate use compared to NHS use s lower for women. There s some ndcaton that users of health care servces n ether sector rate ther health as better health than the non-users of servces. Indvduals who smoke more, condtonal on beng smokers, are less lkely to use ether NHS or prvate servces. These health status results perhaps reflect the fact that the any servce set ncludes preventve servces. Educatonal attanment s postvely assocated wth use of both sectors, though the magntude of the estmated margnal effects s 8 These measures and the ssues of possble endogenety are presented more fully n Propper Ž A measure of the prce of npatent medcal care s the average annual prvate nsurance prce. Ths vares by year but not by regon. Ths measure was used n ntal estmates, but had a small and nsgnfcant coeffcent estmate and therefore was dropped from subsequent analyss. 10 Besley et al. Ž found that corporate health nsurance cover vares by occupaton. 11 The estmated margnal effects are d pjr d x. The relatonshp to the estmated bj s d pjr d xs p Ž b ys p b. whch depends on the estmated parameters for all the optons. j j j j
11 ( ) C. PropperrJournal of Health Economcs Table 2 Multnomal logt estmates of margnal effects of NHS and prvate use Any servce use Inpatent stay Dental vst NHS only Prvate NHS only Prvate NHS only Prvate dprdx Ž z-stat. dprdx Ž z-stat. dprdx Ž z-stat. dprdx Ž z-stat. dprdx Ž z-stat. dprdx Ž z-stat. Gender 0.05 Ž Ž Ž Ž Ž 4.9. y0.005 Ž 0.9. Socal renter Ž 5.5. y0.09 Ž Ž 0.1. y Ž 2.8. y0.021 Ž 7.2. y0.060 Ž 8.7. Hgher educaton Ž Ž 4.6. y0.005 Ž 1.0. y Ž Ž Ž Age, Age, Famly composton Yes Yes Yes Yes Yes Yes Self employed y0.12 Ž Ž 1.1. y0.06 Ž Ž 0.1. y0.063 Ž Ž 1.6. Employed y0.09 Ž Ž 1.5. y0.06 Ž 9.5. y0.003 Ž 1.5. y0.05 Ž Ž 1.7. Publc sector Ž 0.5. y0.01 Ž Ž Ž Ž 0.4. y Ž 0.7. Household ncome y Ž Ž Ž Ž 6.4. y Ž Ž 7.4. H hold ncome squared 2.66ey07 Ž 0.3. y4.18ey06 Ž 4.5. y1.56ey07 Ž 2.1. y5.75ey08 Ž ey08 Ž 0.7. y1.85ey07 Ž 3.6. Industry dummes Yes Yes Yes Yes Yes Yes Health lmts daly actvty y0.002 Ž 1.5. y0.02 Ž 2.4. y0.03 Ž 6.0. y0.003 Ž Ž 0.2. y0.023 Ž 2.3. Health lmts work y0.02 Ž Ž 0.8. y0.03 Ž Ž Ž Ž 2.0. Smoker 0.01 Ž Ž 1.8. y0.009 Ž 1.3. y Ž Ž Ž 0.8. Number of cgarettes y0.002 Ž 3.1. y Ž 2.2. y Ž 2.0. y Ž 1.0. y0.003 Ž 3.1. y Ž 1.6. Other health controls Yes Yes Yes Yes Yes Yes Conservatve party supporter y y Dsagree wr All health care should be free y0.001 Ž Ž 4.7. y Ž Ž Ž Ž 5.2. Dsagree wr Unfar money buys prorty y0.004 Ž Ž Ž Ž 2.8. y0.002 Ž Ž 0.4. Publc use t y Ž y0.015 Ž Ž y0.002 Ž Ž Ž Prvate use t y Ž Ž Ž Ž Ž Ž % RHA resd watng )1 year Yes Ž ns. Yes Ž ns. Yes Ž ns. Yes Ž ns. % RHA resd watng -1 year Yes Ž ns. Yes Ž ns. Yes Ž ns. Yes Ž ns. Dental avalablty Yes Ž ns. Yes Ž ns. Regonal dummes Yes Yes Yes Yes Yes Yes Log lkelhood y17,960 y6630 y14,844 Sample sze 21,002 20,962 20,598 Ž. 1 z-stats are for coeffcent estmates and are calculated usng robust standard errors. Ž. 2 Health controls are dummy varables for whether ndvdualhas health problems wth or suffers from the followng condtons: lmbs, sght, hearng, skn, chest, heartrblood pressure, stomachrdgeston, dabetes, anxetyrdepresson, alcoholrdrugs, eplepsy, and other condtons not specfed n ths lst. Ž. 3 Publc and prvate use t y 1 are the last year s use Ž of the publc and prvate sector, respectvely. of the dependent varable of the column for whch the estmates are gven.
12 866 ( ) C. PropperrJournal of Health Economcs larger for the prvate alternatve. Prevous analyses of NHS use have also found a postve assocaton between NHS use and educaton Žsometmes argued to be evdence of mddle class capture of the welfare state.. On the other hand, there are clear dfferences, n the expected drecton, n the soco-economc characterstcs of the users of the two sectors. Prvate users are less lkely to lve n socal rented housng, and NHS health care users are more lkely to do so. Beng employed s negatvely assocated wth use of publc care, but postvely assocated wth use of prvate care. Income s more heavly assocated wth use of the prvate sector than use of the NHS. Indvduals who report that they are lmted n ther daly actvtes are less lkely to use any prvate health servces than to use the NHS. There also appears to be a clear assocaton between prvate use and poltcal atttudes. Beng a Conservatve supporter s assocated postvely wth prvate sector use and negatvely wth NHS use. Users of NHS servces are slghtly more pro-free care than non-users, but prvate users are sgnfcantly less supportve of the prncples of the NHS. It s possble that these atttude measures are endogenous: ndvduals who use prvate servces justfy ther use by holdng relatvely ant NHS vews. The endogenety of these varables s nvestgated elsewhere Ž Burchardt and Propper, where t was found that, whle prvate users hold less postve atttudes to the NHS, there s no clear evdence that prvate sector use leads to less support for the NHS than does NHS use. It s possble also that these atttudnal varables are correlated wth the regressors and wth the error terms but play no casual role n health care decsons. To nvestgate ths, the model was re-estmated constranng the coeffcents on the party support and atttudnal varables to be 0. The coeffcent estmates on the other varables changed very lttle and none of the changes were statstcally sgnfcant. 12 I therefore conclude that the results are robust to ncluson of the atttudnal varables, and that these varables play an ndependent role n the choce of alternatve. There appears to be no clear assocaton between the length of ether watng lsts under a year or over a year and use of ether alternatve. The coeffcent estmates of these varables were not sgnfcantly dfferent from zero at conventonal levels. Therefore, the margnal effects of these varables are not presented. The estmates for npatent and dental servce use ndcate demographc dfferences between users and non-users of these servces. These, n part, reflect the type of servce. For example, the demanders of dental care n ether sector are younger, whch may reflect the fact that dental care s n part preventatve and those who are younger perceve more beneft from such care. There are also clear soco-economc dfferences between the users of prvate and NHS care for these two servces. NHS npatent users are less lkely to be employed and are more 12 These results are avalable from the author.
13 ( ) C. PropperrJournal of Health Economcs lkely to lve n socal rented housng than ether non-users or prvate users. NHS dental servce users are poorer and less lkely to be employed, reflectng the elgblty requrements for free dental check-ups. In terms of poltcal atttudes, prvate npatent users are less supportve of the equty prncples of the NHS than ether NHS users or non-users. For dental servces, Conservatve party support s postvely assocated wth prvate use and negatvely assocated wth NHS use. However, both NHS and prvate users of dental servces are less supportve of the equty prncples of the NHS than are non-users. Ths perhaps reflects the fact that users of dental servces, whether publc or prvate, make some payment and so are more used to payng for care than non-users of dental servces. The results also ndcate a strong assocaton between lagged and current use. Health status s controlled for, so ths s not due to observed dfferences n health status. For any servce use, NHS-only use s assocated postvely wth use last year of ether sector. Prvate use s negatvely assocated wth use last year of NHS servces but postvely assocated wth lagged use of prvate servces. Further, the estmated effect of past use of servces n the same sector Žthe same-sector effect. s substantal larger than the estmated effect of past use of servces n the other sector Ž the cross-sector effect.. For example, the margnal effect of lagged use of NHS-only servces on current use of NHS servces s 0.3 whle the effect of lagged use of NHS-only servces on current use of the prvate sector s y The same- and cross-sector effects for lagged prvate care are 0.24 and 0.02, respectvely. The npatent and dental estmates show rather dfferent same- and cross-sector effects. For npatent servces, only lagged same-sector use s assocated wth current use. The small sze of the estmated margnal effect reflects the nfrequency of use of npatent servces. In contrast, for dental servces, the margnal effects of lagged use of both NHS and prvate care are postvely assocated wth current NHS and current prvate sector use. Ths perhaps ndcates that ndvduals move between publc and prvate care when the cost dfference between publc and prvate treatment s not large and when dental nsurance s not wdely avalable Ž as durng the perod covered by these data.. But even for dental care the mpact of lagged own-sector use s consderably hgher than that of lagged cross-sector use. The estmated margnal mpact of past NHS dental use on current NHS dental use s 0.36, whle the margnal effect of past prvate dental use on current NHS use s only Smlar dfferences n the estmated margnal effects of lagged use can be seen for current prvate sector use. These results ndcate that there s a tendency for ndvduals to re-use the sector they used last tme Suppler-nduced demand, where NHS dentsts encourage patents to go prvate would suggest that the effect of lagged publc use on current prvate use should be larger than the mpact of lagged prvate use on current NHS use. The estmates here do not provde support for ths dea.
14 868 ( ) C. PropperrJournal of Health Economcs These MNL estmates provde broad support for the model of Secton 3. Those who value ther health care more hghly, as measured by educaton, use more servces. Those wth hgher costs of watng tme Ž.e. a hgher g. the employed and those wth hgher ncomes use more prvate servces. If the parameter t s nterpreted more broadly as NHS qualty, ndvduals wth a hgher g as measured by poltcal atttudes tend to be hgher users of prvate servces. There appears to be no drect mpact of the measures of NHS qualty used here. If past use s taken as a measure of Ž lower. search costs, those for whom the relatve prce of prvate servces s lower tend to use more prvate servces and those for whom the prce of NHS servces s lower tend to use more NHS servces. The mportance of the lagged effects merts further exploraton. The data permt the estmaton of not only the effect of use of servce s at ty1 on servce s at t but also of other servces, denoted s X,atty1 on use of servce s at t. Table 3 presents estmates of these margnal effects. The estmated model s the MNL model of Table 2 where lagged past servce use s dsaggregated nto use of four specfc servces. These are dental care, eye care, npatent stays, and a group of communty-based servces. Only the estmates of these lagged use varables are presented. The other coeffcent estmates and margnal effects are very smlar to those n Table 2. The results show that use of a servce n one sector s n almost all cases sgnfcantly assocated wth use of the same servce a year later n the same sector. It s also often assocated wth use of the same servce a year later n the other sector. But these latter cross-sector effects are always smaller than the own-sector effects. In other words, whle ndvduals do change sector between years, they are more lkely to use the same sector agan next year rather than swtch between sectors. For npatent servces, there s a postve mpact of lagged use n one sector on current use n the same sector. These own-sector effects are generally sgnfcant but are small. There s lttle assocaton between an npatent stay and other servce use whch, gven the low probablty of an npatent event, s not surprsng. The cross-sector effects are much smaller than for the other servces and n most cases the coeffcents from whch the margnal effects are derved are not well defned. There s an nterestng excepton to ths: the postve assocaton between lagged prvate use of communty servces and current NHS npatent use. Ths perhaps ndcates a lack of avalablty of NHS communty based servces. The patterns of assocaton wth current dental servce use also provde evdence of lnked use wthn one sector over tme. NHS dental use s assocated wth lagged use of both NHS dental and NHS eye servces. Prvate dental servce use s assocated wth lagged use of three of the four prvate servces. The estmates also show the cross-sector assocaton of use n dental servces seen n Table 2: lagged prvate use s assocated wth current NHS use and lagged NHS use s assocated wth current prvate use.
15 Table 3 The mpact of past servce use: multnomal logt estmates Any servce use Inpatent stay Dental vst NHS only Prvate NHS only Prvate NHS only Prvate dprdx Ž z-stat. dprdx Ž z-stat. dprdx Ž z-stat. dprdx Ž z-stat. dprdx Ž z-stat. dprdx Ž z-stat. Publc dental ty Ž y0.002 Ž y0.002 Ž Ž Ž Ž Prvate dental ty1 y0.03 Ž Ž y0.009 Ž Ž Ž Ž Publc eye ty Ž Ž Ž Ž Ž 3.7. y0.01 Ž 0.4. Prvate eye ty1 y0.04 Ž Ž y0.007 Ž Ž Ž Ž 4.5. Publc npatent stay ty Ž 2.9. y0.03 Ž Ž y0.001 Ž Ž 0.5. y0.006 Ž 0.5. Prvate npatent stay ty1 y0.10 Ž Ž Ž Ž Ž Ž 1.2. Publc other servces ty Ž Ž Ž Ž Ž 0.2. y0.001 Ž Prvate other servces ty1 y0.16 Ž Ž Ž Ž Ž Ž 3.3. Log lkelhood y17,091 y6692 y14,754 Sample sze 20,586 20,569 20,519 Ž. Ž. 1 z-stats are for coeffcent estmates and are calculated usng robust standard errors. 2 Models estmated ncludng all other parameters n Table 2. C. PropperrJournal of Health Economcs 19 ( 2000 )
16 870 ( ) C. PropperrJournal of Health Economcs Further nvestgaton of dynamcs The results ndcate a strong assocaton between past and current use. If past use does determne present use, condtonal on ncome and health status, ths has mportant mplcatons for the evoluton of use of the prvate sector over tme. So as the data s a panel, t s used to examne whether the mpact of past use s robust to ncluson of ndvdual effects. To do ths I focus on prvate use only. NHS and no use are combned nto one category. For each of the 5 years of data and for each servce, a bnary ndcator of whether the ndvdual has used prvate health care s observed: n s si n s ) )0 t t where I Ž.. s the ndcator functon takng a value 1 f the expresson n parentheses s true and 0 otherwse, n s t ) s the latent demand for prvate health care, ndexes the ndvdual, t tme, and s the servce defned as above. Ths s a model wth one error, where that error can be nterpreted as the propensty to choose prvate care. The separate parameter estmates of the three alternatve model cannot be recovered 14 but, whle the parameter estmates wll not be the same as those of the multnomal logt model, the latent utlty of prvate care wll agan be a functon of all the parameters of the economc model. Usng the data as a panel means ndvdual effects can be allowed for, so the latent utlty of prvate care s modelled as: n s t ) sg 0 Xtqg1Qrtqg 2 n s ty1qg 3Rqg4Tqyq t Ž 8. where n s ty1 s past prvate use of servce s, y s an ndvdual effect, t s whte nose, and all other varables are defned as above. Eq. Ž. 8 s estmated as a random effects probt model. Ths allows for an ndvdual specfc effect, so s a rcher error structure than the MNL model estmated above. Eq. Ž. 8 also ncludes the lagged dependent varable as a regressor. In a dynamc panel model where the number of tme perods s short, correlaton between the random effects term and the ntal observaton of the dependent varable Ž the so-called ntal condtons problem. renders random effects ML estmaton of the parameters of nterest nconsstent f the ntal condtons problem s gnored Ž Hsao, To overcome ths problem, a method 14 If the determnants of the ndrect utlty of the three optons dffered, t would be possble to recover the parameters of each of the three optons from a bnary outcome by estmatng a model wth partal observablty. However, the economc model ndcates that all varables affect all choces, so ths s not possble wthout mposton of arbtrary excluson restrctons.
17 ( ) C. PropperrJournal of Health Economcs suggested by Orme Ž s used, an outlne of whch s gven n the Appendx. The approach s to augment the model of nterest wth an estmated term that corrects for the correlaton between the ntal observaton of the dependent varable Ž n. and the random error Ž y. 0. Orme suggests that ths wll allow estmaton of ths correlaton and, where ths s not too hgh, the technque wll yeld adequate nferences for the parameters of nterest. To estmate the correcton term requres that the varables that determne the ntal observaton of prvate demand are exogenous. To estmate the ntal observaton, I use parental soco-economc status, age and gender, usng the argument that parental soco-economc status affects the ntal level of prvate demand, but not subsequent changes. The estmates ndcate that ntal demand Ž the 1990r1 BHPS observaton. s sgnfcantly assocated wth these varables. Table 4 presents the random effects probt estmates wth the correcton term ncluded. Estmates of the margnal effects are presented together wth the z-statstcs for the underlyng coeffcents from whch the margnal effects are derved. The table ndcates that the estmates of lagged use reman sgnfcant after correcton for the ntal condton problem. The coeffcent on the correcton ) term e ndcates a postve correlaton between n and y Ž 0 0 the estmate varyng from 0.25 for all prvate use to only 0.08 for npatent use.. The results wthout the Orme correcton are not presented here but ndcate a larger coeffcent on the lagged term, so gnorng the ntal condtons problem results n overestmaton of the lagged effects. However, even wth the correcton term ncluded, the estmated effects of lagged past prvate demand reman large n magntude and hghly sgnfcant for all three servces. 15 The results n Table 4 show that the patterns of assocaton dentfed n the MNL estmates are not smply due to ndvdual heterogenety. Condtonal on random ndvdual effects, prvate sector use remans sgnfcantly assocated wth employment status, educaton, ncome, health status, poltcal atttudes and past use of the prvate sector. The lack of mportance of watng lst and NHS avalablty measures remans. Prvate use at tme t y 1 s a sgnfcant determnant of prvate use at t. Although from a dfferent stochastc structure, the results can be compared to the prvate use columns of the MNL model n Table 2. Ths comparson ndcates close smlarty of most of the estmates, though the random 15 Two assessment of the goodness of ft of ths model were made. The frst was a comparson of the assumed parametrc estmator Ž the random effects probt estmator. aganst a sem-parametrc estmator Ž the lowess estmator.. Ths showed a close assocaton between the two, suggestng robustness to the parametrc assumptons. The second was a graphcal test based on Klen Ž The observatons were grouped nto 200 quantles of the ndex functon of the random effects probt estmator. The test statstc for the quantle, Q, suggested by Klen, was plotted aganst the quantles. The results showed lttle evdence of msspecfcaton.
18 872 ( ) C. PropperrJournal of Health Economcs Table 4 The use of prvate health care servces: random effects probt estmates wth correcton for ntal condtons Any servce use Inpatent stay Dental vst dprdx Ž z-stat. dprdx Ž z-stat. dprdx Ž z-stat. Prvate use t y Ž Ž Ž e 0.25 Ž Ž Ž Socal renter y0.08 Ž 8.3. y Ž 2.6. y0.048 Ž 7.0. Demographcs Ž age, household composton. Included Included Included Self employed 0.07 Ž 4.3. y Ž Ž 4.5. Employed y Household ncome H hold ncome sq. y4.7ey 07 Ž 4.4. y3.8ey 08 Ž 1.9. y3.0ey 07 Ž 3.3. Industry dummes, publc sector employee Included Included Included Health lmts daly actvty, health lmts work, Included Included Included smoker, number of cgarettes smoked % RHA resdents watng ) 1 year y32.0 Ž 1.2. y2.26 Ž 0.9. % RHA resdents watng - 1 year 3.4 Ž Ž 1.3. Dental avalablty y Conservatve party supporter 0.04 Ž Ž Ž 5.0. Dsagree wr All health care should be free Ž Ž Ž 4.1. Dsagree wr Unfar money buys prorty Ž Ž Ž 1.0. Regonal effects Included Included Included Tme effects Included Ž ns. Included Ž ns. Included Ž ns. Indvdual effects Included Included Included Sample sze Mean probablty of prvate use Predcted probablty Ž at means Ž. 1 dprdx s the probablty of usng prvate care. Derved for contnuous varable by estmaton of the partal dervatve at means of all varables. For dummy varables calculated as dfference n evaluated probabltes when dummys 1 and dummys 0 at means of all other varables. Ž. 2 z-stats for coeffcents are calculated usng robust standard errors.
19 ( ) C. PropperrJournal of Health Economcs effects probt model gves hgher estmates of the margnal effect of past prvate use on current prvate use for both dental and npatent servces Dscusson and conclusons Lack of avalable data has lmted the statstcal analyss of the determnants of the choce between publc and prvate health care n the UK. Prevous studes have looked only at prvate health nsurance or used aggregate data. Ths study estmates a competng rsks model of the health care alternatves open to a user of care n the UK usng ndvdual data. It s also the frst study to examne the mpact of past use usng mcrodata. Use of prvate health care s found to be strongly determned by ncome, demographcs, atttudes to the equty goals of the NHS, poltcal allegance, and past use. Users of prvate servces are rcher, more lkely to support the poltcal rght, less supportve of the equty goals of the NHS and more lkely to have used prvate care n the past than the rest of the populaton. A separate and sgnfcant contrbuton of each of these factors s dentfed. There s strong evdence of assocaton between past and present use of prvate health care, and between past and present use of NHS care. The past prvate use of one health servce s postvely assocated wth current prvate demand for another and the past use of an NHS servce s postvely assocated wth current use of another NHS servce. There s also a cross-sectoral flow. Past use of the NHS s postvely assocated wth present use of prvate servces. The flow s not just one way out of the NHS. The past use of servces n the prvate sector s postvely assocated wth current publc sector use. The estmated postve mpact of past prvate use on current NHS use of dental servces s larger than the estmated mpact of past NHS use on current prvate use. Whle these fndngs support the dea that the users of prvate servces are an dentfable socal group and so Ž f examned cross-sectonally. could be argued to consttute a prvate welfare class, when a longtudnal analyss s undertaken, the support for a prvate welfare class s less strong. The analyss does not support the dea that there s a group of users n the UK who move nto the prvate sector and stay there. Instead, there s consderable movement between the publc and the prvate sectors. Takng all the prvate servces for whch there s an NHS alternatve together, publc and prvate use appear complementary. For dental care, prvate and publc care also appear to be complements. Only for npatent stays, whch ndvduals requre nfrequently and for whch they may buy nsurance, does there appear to be a group that predomnantly uses the prvate sector. Even ths 16 The smlarty wth the MNL results perhaps ndcates that the mposed assumpton of uncorrelated errors n the MNL model does not serously dstort the results.
20 874 ( ) C. PropperrJournal of Health Economcs group s unlkely to use only prvate care, as several npatent servces Že.g. accdent and emergency servces. are not provded prvately. Nor does support for the equty goal of the NHS appear to be lnked most strongly to sector n whch health care has been taken. Support for the equty goals of the NHS s strongest amongst non-users, and publc and prvate users are closer to each other than to non-users. 17 The results suggest that despte the recent growth n the use of prvate fnanced health care, there s perhaps not a dstnct group of prvate servce users who have completely opted out of the NHS. Instead, the use of prvate servces appears to be complementary to the use of the publc sector. Those who use one sector use the other, and those who use the prvate sector retan consderable support for the NHS. Such patterns have mplcatons for the growth of prvate and publc care n the UK. Acknowledgements I am grateful to the Data Archve for access to the Brtsh Household Panel Survey and to the Joseph Rowntree Foundaton for fundng. Tana Burchardt provded excellent data assstance and comments. Two anonymous referees helped to produce a much mproved paper. My thanks are also due to Smon Burgess and semnar partcpants at the Unverstes of Brstol, Manchester, Essex and the LSE. All remanng errors are mne. Appendx A The model of nterest takes the followng form: n ) sg n qb X t ty1 x tqyq t, s1,...,n,ts1,...,t Ž A.1. where n ) t s the latent use of prvate health care, x t s a vector of ndvdual characterstcs, and y an unobserved ndvdual effect. The ntal condtons problem occurs where the ntal observaton, n 0, s correlated wth the random error term y Ž because the start of the observaton perod does not concde wth the start of the stochastc process generatng n ). t. To allow for ths problem, Heckman Ž can be followed and a reduced form equaton s specfed for the ntal observaton: n ) sl X 0 zqh Ž A.2. 2 Ž where z s a vector of strctly exogenous nstruments, var h ss and corr y, h 17 See also Burchardt et al
An Alternatve Way to Measure Prvate Equty Performance Peter Todd Parlux Investment Technology LLC Summary Internal Rate of Return (IRR) s probably the most common way to measure the performance of prvate
REVIEW OF RISK MANAGEMENT CONCEPTS LOSS DISTRIBUTIONS AND INSURANCE Loss and nsurance: When someone s subject to the rsk of ncurrng a fnancal loss, the loss s generally modeled usng a random varable or
Internatonal Journal of Busness and Economcs, 2011, Vol. 10, No. 2, 159-164 Can Auto Lablty Insurance Purchases Sgnal Rsk Atttude? Chu-Shu L Department of Internatonal Busness, Asa Unversty, Tawan Sheng-Chang
PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIIOUS AFFILIATION AND PARTICIPATION Danny Cohen-Zada Department of Economcs, Ben-uron Unversty, Beer-Sheva 84105, Israel Wllam Sander Department of Economcs, DePaul
Searchng and Swtchng: Emprcal estmates of consumer behavour n regulated markets Catherne Waddams Prce Centre for Competton Polcy, Unversty of East Angla Catherne Webster Centre for Competton Polcy, Unversty
4.02 Quz Solutons Fall 2004 Multple-Choce Questons (30/00 ponts) Please, crcle the correct answer for each of the followng 0 multple-choce questons. For each queston, only one of the answers s correct.
Inequalty and The Accountng Perod Quentn Wodon and Shlomo Ytzha World Ban and Hebrew Unversty September Abstract Income nequalty typcally declnes wth the length of tme taen nto account for measurement.
SP 2005-02 August 2005 Staff Paper Department of Appled Economcs and Management Cornell Unversty, Ithaca, New York 14853-7801 USA Farm Savngs Accounts: Examnng Income Varablty, Elgblty, and Benefts Brent
Economcs Letters 76 (2002) 151 158 www.elsever.com/ locate/ econbase Gender dfferences n revealed rsk takng: evdence from mutual fund nvestors a b c, * Peggy D. Dwyer, James H. Glkeson, John A. Lst a Unversty
Fnancal Mathemetcs 15 Mathematcs Grade 12 Teacher Gude Fnancal Maths Seres Overvew In ths seres we am to show how Mathematcs can be used to support personal fnancal decsons. In ths seres we jon Tebogo,
Evaluatng the Effects of FUNDEF on Wages and Test Scores n Brazl * Naérco Menezes-Flho Elane Pazello Unversty of São Paulo Abstract In ths paper we nvestgate the effects of the 1998 reform n the fundng
The OC Curve of Attrbute Acceptance Plans The Operatng Characterstc (OC) curve descrbes the probablty of acceptng a lot as a functon of the lot s qualty. Fgure 1 shows a typcal OC Curve. 10 8 6 4 1 3 4
Summary of Analyss of Premum Labltes for Australan Lnes of Busness Emly Tao Honours Research Paper, The Unversty of Melbourne Emly Tao Acknowledgements I am grateful to the Australan Prudental Regulaton
Workng Paper n 07-02 The nfluence of supplementary health nsurance on swtchng behavour: evdence on Swss data Brgtte Dormont, Perre- Yves Geoffard, Karne Lamraud The nfluence of supplementary health nsurance
Heterogeneous Paths Through College: Detaled Patterns and Relatonshps wth Graduaton and Earnngs Rodney J. Andrews The Unversty of Texas at Dallas and the Texas Schools Project Jng L The Unversty of Tulsa
HIGHER DOCTORATE DEGREES SUMMARY OF PRINCIPAL CHANGES General changes None Secton 3.2 Refer to text (Amendments to verson 03.0, UPR AS02 are shown n talcs.) 1 INTRODUCTION 1.1 The Unversty may award Hgher
LIFETIME INCOME OPTIONS May 2011 by: Marca S. Wagner, Esq. The Wagner Law Group A Professonal Corporaton 99 Summer Street, 13 th Floor Boston, MA 02110 Tel: (617) 357-5200 Fax: (617) 357-5250 www.ersa-lawyers.com
A dynamc model of demand for prvate health nsurance n Ireland Clare Fnn a *, Colm Harmon a b Workng Paper 17 Research Programme on Health Servces, Health Inequales and Health and Socal Gan Ths programme
Lagrange Multplers as Quanttatve Indcators n Economcs Ivan Mezník Insttute of Informatcs, Faculty of Busness and Management, Brno Unversty of TechnologCzech Republc Abstract The quanttatve role of Lagrange
For customers Small pots lump sum payment nstructon Please read these notes before completng ths nstructon About ths nstructon Use ths nstructon f you re an ndvdual wth Aegon Retrement Choces Self Invested
Returns to Experence n Mozambque: A Nonparametrc Regresson Approach Joel Muzma Conference Paper nº 27 Conferênca Inaugural do IESE Desafos para a nvestgação socal e económca em Moçambque 19 de Setembro
CelersSystems DEFINING %COMPLETE IN MICROSOFT PROJECT PREPARED BY James E Aksel, PMP, PMI-SP, MVP For Addtonal Informaton about Earned Value Management Systems and reportng, please contact: CelersSystems,
Supplemental health nsurance and equalty of access n Belgum by Erk SCHOKKAERT Tom VAN OURTI Dane DE GRAEVE Ann LECLUYSE Carne VAN DE VOORDE Publc Economcs Center for Economc Studes Dscussons Paper Seres
The Effects of Tax Rate Changes on Tax Bases and the Margnal Cost of Publc Funds for Canadan Provncal Governments Bev Dahlby a and Ergete Ferede b a Department of Economcs, Unversty of Alberta, Edmonton,
Cahers de la Chare Santé The nfluence of supplementary health nsurance on swtchng behavour: evdence from Swss data Auteurs : Brgtte Dormont, Perre-Yves Geoffard, Karne Lamraud N 4 - Janver 2010 1 The nfluence
How to measure assocaton I: Correlaton. 1. Measurng assocaton usng correlaton and regresson We often would lke to know how one varable, such as a mother's weght, s related to another varable, such as a
Nordc Journal of Health Economcs Onlne ISSN: 1892-9710 Determnants of employment-based prvate health nsurance coverage n Denmark ASTRID KIIL* firstname.lastname@example.org Unversty of Southern Denmark Abstract: Ths study
Busness Brths and Deaths Impact of busness brths and deaths n the payroll survey The CES probablty-based sample redesgn accounts for most busness brth employment through the mputaton of busness deaths,
14 The Ch-squared dstrbuton PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 1 If a normal varable X, havng mean µ and varance σ, s standardsed, the new varable Z has a mean 0 and varance 1. When ths standardsed
The Development of Web Log Mnng Based on Improve-K-Means Clusterng Analyss TngZhong Wang * College of Informaton Technology, Luoyang Normal Unversty, Luoyang, 471022, Chna email@example.com Abstract.
Socal Excluson and the Two-Tered Healthcare System of Brazl 1 Densard Alves Unversty of São Paulo Chrstopher Tmmns Yale Unversty Resumo No Brasl exste um sstema de saúde com dos acessos. Aqueles, com recursos,
CHAPTER 14 MORE ABOUT REGRESSION We learned n Chapter 5 that often a straght lne descrbes the pattern of a relatonshp between two quanttatve varables. For nstance, n Example 5.1 we explored the relatonshp
Perre Foy Statstcs Canada 4 Calculaton of Samplng Weghts 4.1 OVERVIEW The basc sample desgn used n TIMSS Populatons 1 and 2 was a two-stage stratfed cluster desgn. 1 The frst stage conssted of a sample
THE EFFECT OF PREPAYMENT PENALTIES ON THE PRICING OF SUBPRIME MORTGAGES Gregory Ellehausen, Fnancal Servces Research Program George Washngton Unversty Mchael E. Staten, Fnancal Servces Research Program
Module LOSSLESS IMAGE COMPRESSION SYSTEMS Lesson 3 Lossless Compresson: Huffman Codng Instructonal Objectves At the end of ths lesson, the students should be able to:. Defne and measure source entropy..
Factors Affectng Outsourcng for Informaton Technology Servces n Rural Hosptals: Theory and Evdence Bran E. Whtacre Department of Agrcultural Economcs Oklahoma State Unversty firstname.lastname@example.org J.
ISSN 1084-1695 Agng Studes Program Paper No. 23 Pre-Retrement Lump-Sum Penson Dstrbutons and Retrement Income Securty:Evdence from the Health and Retrement Study 1 Gary V. Engelhardt Center for Polcy Research
14.74 Lecture 5: Health (2) Esther Duflo February 17, 2004 1 Possble Interventons Last tme we dscussed possble nterventons. Let s take one: provdng ron supplements to people, for example. From the data,
The Onlne Journal of New Horzons n Educaton Volume 4, Issue 3 MargnalReturnstoEducatonForTeachers RamleeIsmal,MarnahAwang ABSTRACT FacultyofManagementand Economcs UnverstPenddkanSultan Idrs email@example.com
The Economc Impacts of Cgarette Tax Reductons on Youth Smokng n Canada Dane P. Dupont and Anthony J. Ward Economcs, Brock Unversty December 2002 Abstract Cgarettes are the most commonly consumed recreatonal
The Complementartes of Competton n Chartable Fundrasng Andreas Lange Unversty of Hamburg Department of Economcs Von-Melle-Park 5 D-20146 Hamburg Germany firstname.lastname@example.org Andrew Stockng Congressonal
ADVERSE SELECTION IN INSURANCE MARKETS: POLICYHOLDER EVIDENCE FROM THE U.K. ANNUITY MARKET * Amy Fnkelsten Harvard Unversty and NBER James Poterba MIT and NBER * We are grateful to Jeffrey Brown, Perre-Andre
ILRRevew Volume 65 Number 2 Artcle 10 2012 The Wllngness to Pay for Job Amentes: Evdence from Mothers' Return to Chrstna Felfe Unversty of St. Gallen, email@example.com The Wllngness to Pay for Job
5 Fxed ncome rsk attrbuton Chthra Krshnamurth RskMetrcs Group firstname.lastname@example.org We compare the rsk of the actve portfolo wth that of the benchmark and segment the dfference between the two
Survve Then Thrve: Determnants of Success n the Economcs Ph.D. Program Wayne A. Grove Le Moyne College, Economcs Department Donald H. Dutkowsky Syracuse Unversty, Economcs Department Andrew Grodner East
Inter-Amercan Development Bank Banco Interamercano de Desarrollo Latn Amercan Research Network Red de Centros de Investgacón Research Network Workng paper #R-436 Socal Excluson and the Two-Tered Healthcare
Secton 5.4 Annutes, Present Value, and Amortzaton Present Value In Secton 5.2, we saw that the present value of A dollars at nterest rate per perod for n perods s the amount that must be deposted today
STAMP UTY ON SHARES AN ITS EFFECT ON SHARE PRICES Steve Bond Mke Hawkns Alexander Klemm THE INSTITUTE FOR FISCAL STUIES WP04/11 STAMP UTY ON SHARES AN ITS EFFECT ON SHARE PRICES Steve Bond (IFS and Unversty
Analyss of Subscrpton Demand for Pay-TV Manabu Shshkura * Norhro Kasuga ** Ako Tor *** Abstract In ths paper, we wll conduct an analyss from an emprcal perspectve concernng broadcastng demand behavor and
HE DISRIBUION OF LOAN PORFOLIO VALUE * Oldrch Alfons Vascek he amount of captal necessary to support a portfolo of debt securtes depends on the probablty dstrbuton of the portfolo loss. Consder a portfolo
A research and educaton ntatve at the MT Sloan School of Management Understandng the mpact of Marketng Actons n Tradtonal Channels on the nternet: Evdence from a Large Scale Feld Experment Paper 216 Erc
Transton Matrx Models of Consumer Credt Ratngs Abstract Although the corporate credt rsk lterature has many studes modellng the change n the credt rsk of corporate bonds over tme, there s far less analyss
For customers Uncrystallsed funds penson lump sum payment nstructon Don t complete ths form f your wrapper s derved from a penson credt receved followng a dvorce where your ex spouse or cvl partner had
ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING Matthew J. Lberatore, Department of Management and Operatons, Vllanova Unversty, Vllanova, PA 19085, 610-519-4390,
10-170 Research Group: Econometrcs and Statstcs 2010 UK Letter Mal Demand: a Content Based Tme Seres nalyss usng Overlappng Market Survey Statstcal Technques CTHERINE CZLS, JEN-PIERRE FLORENS, LETICI VERUETE-MCKY,
Fnancal Instablty and Lfe Insurance Demand + Mahto Okura * Norhro Kasuga ** Abstract Ths paper estmates prvate lfe nsurance and Kampo demand functons usng household-level data provded by the Postal Servces
SIMPLE LINEAR CORRELATION Smple lnear correlaton s a measure of the degree to whch two varables vary together, or a measure of the ntensty of the assocaton between two varables. Correlaton often s abused.
BER WORKIG PAPER SERIES CROWDIG OUT AD CROWDIG I OF PRIVATE DOATIOS AD GOVERMET GRATS Garth Heutel Workng Paper 15004 http://www.nber.org/papers/w15004 ATIOAL BUREAU OF ECOOMIC RESEARCH 1050 Massachusetts
An Emprcal Study of Search Engne Advertsng Effectveness Sanjog Msra, Smon School of Busness Unversty of Rochester Edeal Pnker, Smon School of Busness Unversty of Rochester Alan Rmm-Kaufman, Rmm-Kaufman
ADVERSE SELECTION IN INSURANCE MARKETS: POLICYHOLDER EVIDENCE FROM THE U.K. ANNUITY MARKET Amy Fnkelsten Harvard Unversty and NBER James Poterba MIT and NBER Revsed May 2003 ABSTRACT In ths paper, we nvestgate
Insurance Markets and Companes: Analyses and Actuaral Computatons, Volume 1, Issue 2, 2010 José Antono Ordaz (Span), María del Carmen Melgar (Span) Covarate-based prcng of automoble nsurance Abstract Ths
Lecture 3: Force of Interest, Real Interest Rate, Annuty Goals: Study contnuous compoundng and force of nterest Dscuss real nterest rate Learn annuty-mmedate, and ts present value Study annuty-due, and
Is There A Tradeoff between Employer-Provded Health Insurance and Wages? Lye Zhu, Southern Methodst Unversty October 2005 Abstract Though most of the lterature n health nsurance and the labor market assumes
HARVARD John M. Oln Center for Law, Economcs, and Busness ISSN 1045-6333 ASYMMETRIC INFORMATION AND LEARNING IN THE AUTOMOBILE INSURANCE MARKET Alma Cohen Dscusson Paper No. 371 6/2002 Harvard Law School
Regresson Lectures So far we have talked only about statstcs that descrbe one varable. What we are gong to be dscussng for much of the remander of the course s relatonshps between two or more varables.
MARCH 211 C.D. Howe Insttute WORKING PAPER FISCAL AND TAX COMPETITIVENESS The Effects of Tax Rate Changes on Tax Bases and the Margnal Cost of Publc Funds for Provncal Governments Bev Dahlby Ergete Ferede
Margnal Beneft Incdence Analyss Usng a Sngle Cross-secton of Data Mohamed Ihsan Ajwad and uentn Wodon World Bank August 200 Abstract In a recent paper, Lanjouw and Ravallon proposed an attractve and smple
Wllngness to Pay for Health Insurance: An Analyss of the Potental Market for New Low Cost Health Insurance Products n Namba Abay Asfaw Center for Dsease Control and Preventon\Natonal Insttute for Occupatonal
Management Qualty, Fnancal and Investment Polces, and Asymmetrc Informaton Thomas J. Chemmanur * Imants Paegls ** and Karen Smonyan *** Current verson: December 2007 * Professor of Fnance, Carroll School
The Racal and Gender Interest Rate Gap n Small Busness Lendng: Improved Estmates Usng Matchng Methods* Yue Hu and Long Lu Department of Economcs Unversty of Texas at San Antono Jan Ondrch and John Ynger
TESTING FOR EVIDENCE OF ADVERSE SELECTION IN DEVELOPING AUTOMOBILE INSURANCE MARKET by Oksana Lyashuk A thess submtted n partal fulfllment of the requrements for the degree of Master of Arts n Economcs
Hgh Correlaton between et Promoter Score and the Development of Consumers' Wllngness to Pay (Emprcal Evdence from European Moble Marets Ths paper shows that the correlaton between the et Promoter Score
Chapter 8 Group-based Lendng and Adverse Selecton: A Study on Rsk Behavor and Group Formaton 1 8.1 Introducton Ths chapter deals wth group formaton and the adverse selecton problem. In several theoretcal
PROMEMORIA Datum June 01 Fnansnspektonen Författare Bengt von Bahr, Younes Elonq and Erk Elvers Stress test for measurng nsurance rsks n non-lfe nsurance Summary Ths memo descrbes stress testng of nsurance
How Much s E-Commerce Worth to Rural Busnesses? Susan Watson, Assstant Professor O. John Nwoha, Program Assocate Gary Kennedy, Department Head and Assocate Professor Kenneth Rea, Vce Presdent for Academc