A Dynamic Model of Demand for Private Health Insurance in Ireland
|
|
|
- MargaretMargaret Burke
- 10 years ago
- Views:
Transcription
1 DISCUSSION PAPER SERIES IZA DP No A Dynamc Model of Demand for Prvate Health Insurance n Ireland Clare Fnn Colm Harmon November 2006 Forschungsnstut zur Zukunft der Arbe Instute for the Study of Labor
2 A Dynamc Model of Demand for Prvate Health Insurance n Ireland Clare Fnn Unversy College Dubln Colm Harmon Unversy College Dubln and IZA Bonn Dscusson Paper No November 2006 IZA P.O. Box Bonn Germany Phone: Fax: E-mal: [email protected] Any opnons expressed here are those of the author(s) and not those of the nstute. Research dssemnated by IZA may nclude vews on polcy, but the nstute self takes no nstutonal polcy posons. The Instute for the Study of Labor (IZA) n Bonn s a local and vrtual nternatonal research center and a place of communcaton between scence, polcs and busness. IZA s an ndependent nonprof company supported by Deutsche Post World Net. The center s assocated wh the Unversy of Bonn and offers a stmulatng research envronment through s research networks, research support, and vsors and doctoral programs. IZA engages n () orgnal and nternatonally competve research n all felds of labor economcs, () development of polcy concepts, and () dssemnaton of research results and concepts to the nterested publc. IZA Dscusson Papers often represent prelmnary work and are crculated to encourage dscusson. Caton of such a paper should account for s provsonal character. A revsed verson may be avalable drectly from the author.
3 IZA Dscusson Paper No November 2006 ABSTRACT A Dynamc Model of Demand for Prvate Health Insurance n Ireland * The Irsh health care system offers a tax fnanced, unversal entlement to publc care at a nomnal user fee, nonetheless 50% of the Irsh populaton purchase prvate health nsurance. Ths paper emprcally models the propensy to nsure as a functon of ndvdual and household characterstcs usng panel data analyss and compares three alternate approaches; a statc, chamberlan-mundlak and dynamc specfcaton. Usng panel data from 1994 to 2000, we consder whether propensy to nsure s n fact a functon of heterogeney or of state dependence. A range of ndvdual and household characterstcs s shown to nfluence propensy to nsure. Overall the posve effect of educaton and ncome and the negatve effect of poor health status reman robust across three specfcatons. In movng toward a dynamc specfcaton, we show that persstence s a hghly sgnfcant determnant of demand for prvate health nsurance and also that reduces the sze of the coeffcents on the regressors. The latter pont hghlghts that whle educaton, ncome and, to a lesser extent, health status have very large effects on probably of nsurng, these effects are overestmated where no attempt s made to control for unobserved heterogeney or state dependence. JEL Classfcaton: G22, I10, D01 Keywords: health nsurance, dynamcs, panel, unobserved heterogeney, state dependence Correspondng author: Clare Fnn UCD Geary Instute Unversy College Dubln Belfeld Dubln 4 Ireland E-mal: [email protected] * We would lke to acknowledge fundng from the Irsh Research Councl for the Humanes and Socal Scences, the Endeavour Australa Postgraduate and Post-doctoral Research Fellowshp Programme and PRTLI-HEA, as well as the hospaly of the Centre for Health Economcs and Research Evaluaton (CHERE), UTS and the Department of Economcs at Unversy of Melbourne; We are grateful to Dense Doron, Lsa Farrell, Tmothy Fry, Jane Hall, Namh Hardman, Bran Nolan, Carol Propper, Elzabeth Savage, Mchael Shelds and semnar partcpants at Royal Melbourne Instute of Technology, CAER 3 rd Summer Workshop n Health Economcs, the Geary Semnar Seres and members of the UCD-ESRI HRB Health Economcs and Health Gan Programme for helpful comments, as well as the organsers and partcpants of the Mare Cure Tranng Course n Health Economcs 2006, York.
4 1. Introducton Although the Irsh health care system offers a tax fnanced, unversal entlement to publc care at a nomnal user fee, nearly 50% of the Irsh populaton have prvate health nsurance. Whle orgnally establshed n the late 1950 s to provde cover for the top 15% of earners (those nally excluded from the nomnal fee, publc health sector entlement) the proporton of the populaton choosng prvate health nsurance contnues to grow. Irsh health polcy actvely supports the prvate health sector, whch benefs from a number of cross-subsdes, both drect and ndrect, from publc fnances. Ths paper s motvated therefore by two concerns. Frstly what are the socoeconomc and households characterstcs of those who buy? Secondly, what role does polcy play? Most governments espouse equy of access as a goal of health systems, ndeed the WHO use equy n fnancng & accessbly as crera by whch health system performance s judged, from a equy perspectve then could polcy be contrbutng to nequales n access and qualy of healthcare? Tax relef on prvate health nsurance s justfed on the grounds that allows those wh chronc health condons to benef from nsurance at a reasonable cost (Department of Health and Chldren 1999). However, ths polcy reduces the cost of nsurance to everyone, not just the chroncally ll. There s evdence to suggest that the Irsh nsurance system does not suffer from adverse selecton. In fact would seem that those wh poor health status are less lkely to be nsured than those wh good health 1. Ths suggests that tax relef may not be havng the desred effect. Posng the queston, are the systems structures, as they pertan 1 See Doron, Jones and Savage for a fuller dscusson of the relatonshp between SAHS and the purchase pf prvate health nsurance. 2
5 to, adequately protectng some of the more vulnerable, less healthy segments of the populaton? Only a small lerature pertanng to the Irsh nsurance system exsts (Nolan and Wley 2000; Harmon and Nolan 2001; Watson and Wllams 2001). Only one Harmon & Nolan (2001) models the effects of ndvdual and household characterstcs on propensy to nsure focusng on a cross-secton analyss of propensy to nsure n A number of ndvdual and household characterstcs are shown to nfluence the nsurance decson. In lght of the avalably of new data & ndeed of the huge economc growth experenced n Ireland throughout the latter half the 1990 s nto the 2000, ths paper ams to expand on ths exstng work wh a more detaled panel data analyss. The method of estmaton as allowed us explo the panel nature of the data whle attemptng to control for unobserved ndvdual specfc effects. It has also faclated the ncluson of a lagged dependent varable to consder the possble role of state dependence, that s persstence perhaps due to changng preferences or costs assocated wh the nsurance state. Drawng on lerature relatng to utly and nsurance, a theoretcal framework supportng an emprcal nvestgaton of the effect of ndvdual and household characterstcs on the propensy to nsure s provded. Whle patterns of assocaton are evdent emprcally, the extent to whch they are functon of observed heterogeney or state dependence reman unclear. The robustness of certan ndvdual and household effects to the ncluson of a lagged dependent varable s therefore of nterest. The paper proceeds as follows. As the bass for the emprcal nvestgaton, a theoretcal framework s establshed n Secton 2. Secton 3 s a short overvew of the Irsh health 3
6 and nsurance system. Data and prelmnary statstcs are descrbed n Secton 4. Secton 5 focuses on estmaton. Whle results and conclusons are reported n Secton Theoretcal framework: Utly theory and Insurance The decson to nsure has been wdely consdered n theoretcal lerature pertanng to nsurance n general and, more specfcally, to health nsurance (Arrow 1963; Feldsten 1973; Van De Ven and Van Praag 1981; Propper 1989; Besley 1991; Hopkns and Kdd 1996; Besley, Hall et al. 1998; Besley, Hall et al. 1999). The decson to nsure s one of a dscrete choce, to purchase prvate health nsurance or not. A comparson of expected utly under nsurance to expected utly under no nsurance wll nform the nsurance decson (Besley, Hall et al. 1999; Propper 2000). It s antcpated that expected utly gan or loss from the decson to nsure wll be a functon of determnants pertanng to materal well-beng (ncome, educaton, age, famly characterstcs) and medcal need (age, sex, famly characterstcs, health status) (Van De Ven and Van Praag 1981; Propper 1989; Hopkns and Kdd 1996; Besley, Hall et al. 1999). The mpact of materal well-beng and medcal need on the propensy to demand prvate health nsurance s therefore of nterest. To expand on ths n a less formal dscusson, the utly of havng prvate health nsurance (or not) s nfluenced by expected medcal consumpton or probably of sckness. Medcal need s assocated wh uncertanty and as Arrow notes demand for medcal servces s rregular and unpredctable and affords satsfacton only n the event of llness (Arrow 1963). In lne wh ths Propper (1989) observes, health nsurance can only be used n states of ll health. The utly of prvate health nsurance 4
7 when sck therefore s greater than when well 2. Certan ndvdual and household characterstcs are assocated wh a hgher or lower rsk of medcal need, or to put n another way, wh a hgher or lower rsk vulnerably (Hopkns and Kdd 1996). Supported by the theory of adverse selecton, s expected that those wh hgh-rsk vulnerably are more lkely to nsure. Thus we mght expect that those wh poor health status or a chronc condon the elderly (due to decreasng health), the presence of chldren (due to hgher expected medcal consumpton) and females (due to expected future consumpton related to chldbrth) wll all have a posve effect on demand for prvate health nsurance. Spltng rsk vulnerably nto drect and ndrect rsk vulnerably faclates some further consderaton of the ssues n the context of the Irsh nsurance system. Drect rsk vulnerably refers to varables drectly assocated wh medcal need such as past or present health status/healthcare consumpton 3. On the other hand, ndrect rsk vulnerably s assocated wh characterstcs that do not reflect past or present health status but are ndrectly assocated wh medcal need, as such, those wh hgh ndrect rsk vulnerably have a hgher expectaton of medcal need (e.g as suggested earler, older people, females and those wh chldren). Under communy ratng, as exsts n the Irsh system, there s no dfferentaton n premum prce based on rsk. We would expect therefore that those wh hgh ndrect rsk vulnerably (gven that there s no fnancal penaltes for ther ncreased rsk) would be more lkely to nsure. On the other hand, although those wh hgh drect rsk vulnerably (poor past or present health) are not prohbed from nsurng or ndeed 2 Ignorng that gven the state of uncertanty about states of health there may be some utly n beng nsured, partcularly for the rsk adverse 3 For example, low drect rsk vulnerably descrbes good health status whereas hgh drect rsk vulnerably descrbes poor health status or a chronc condon 5
8 charged hgher prema under communy ratng, under a pre-exstng condon clause any health condons experenced pror to the purchase of nsurance are excluded from the terms of the cover for a perod of fve years. Thus the utly of prvate nsurance take-up s less for those who already have specfc health problems as they wll have to wa a consderable perod of tme before they can use. In essence mght be suggested that the nsurance system creates an ncentve for those wh hgh ndrect rsk vulnerably and a dsncentve for those wh hgh drect rsk vulnerably (those who already suffer from specfc medcal condons) to nsure. Thus whle factors such as age, the presence of chldren and females are all expected to have a posve effect on demand for prvate health nsurance, s antcpated, at least n an Irsh context, that poor health status mght be assocated wh a lower propensy to nsure. There s evdence of selecton nto nsurance by ncome (Van De Ven and Van Praag 1981; Propper 1989; Besley, Hall et al. 1999). Income can determne the probably of purchasng prvate nsurance n two ways. The frst s the ntuve expectaton that the hgher the ncome the less the opportuny cost assocated wh the purchase of prvate health nsurance n pure monetary terms (Hopkns and Kdd 1996). The second relates to the opportuny cost of tme. Wh respect to the latter, Propper (1989) notes that one of the costs assocated wh the publc sector s the cost of wang (Propper 1989). To elaborate on ths cost n the context of loss of tme, s noted that the cost of wang s the opportuny cost of healthy tme. It s assumed for those on a wang lst, stock of health s at less than s usual capacy. Therefore llness reduces the amount of healthy tme avalable. The opportuny cost of healthy tme, accordng to Propper (1989) s a functon of ncome, source of ncome and the extent to whch 6
9 ndvduals re-allocate ther use of tme. Both Hopkns and Kdd (1996) and Propper (1989) recognse that those wh a greater constrant on uses of tme have a hgher the opportuny cost and conclude that the value of tme s probably hgher for those who are employed rather than unemployed or not n the labour force, and those on hgher ncomes rather than lower ncomes (Hopkns and Kdd 1996) 4. Besley et al (1998) also focus on the tme loss assocated wh the publc sector wa; they show a posve relatonshp between publc sector wang tme and prvate nsurance take-up. Despe such evdence, Feldsten (1973) notes, wh respect to the US, that ncome mght negatvely affect the decson to nsure, remarkng for a gven probably dstrbuton of health expenses, hgher ncomes tends to make famles more wllng to assume rsk whch n turn reduces ther demand for prvate nsurance and concludes therefore n relaton hs research that the effect of ncome s ndetermnate. He notes that ths nsgnfcant ncome effect may represent a balancng of posve and negatve ncome effects (Feldsten 1973). Hopkns also nvestgated ths possbly and found ths not to be the case n an Australan context. Educaton may mpact drectly on the nsurance decson va s role n health decson-makng (Hopkns and Kdd 1996). Ths explanaton follows from the assumpton that educaton ncreases the effcency of producton of health (Grossman 1972). In short, those who are better educated may not only have greater knowledge and understandng of health nformaton, but are also capable of makng better health- 4 Ths s supported by the theory of household producton as noted by Becker, G. (1965). "A Theory of the Allocaton of Tme." The Economc Journal Vol. LXXV(No. 299): , who recognses that tme use s affected by fnancal resource, and thus that the extent to whch a tme constrant bnd s n the end an ncreasng functon of the opportuny cost of tme of an ndvdual and other members of hs/her household. The explc predcton s that otherwse dentcal people who s ncome are greater wll feel more rushed for tme (Hamermesh 2004). Ths hypothess s emprcally supported by Hamermesh (2004), they fnd that those wh hgher ncomes are more stressed for tme, and conclude that people wh a hgher value of tme are more stressed for tme, not only because they may work more, but because the command that they possess over goods makes them busy spendng ther ncomes Hamermesh, D. (2004). "Subjectve Outcomes n Economcs." Southern Economc Journal Vol
10 related decsons or formulatng better mxtures of health nputs, of whch nsurance mght be one. Ths mght be one explanaton as to why educaton s an mportant correlate of good health (Grossman 1999). Hopkns and Kdd (1996) and Van Praag (1981) both dentfy educaton and ncome as varables that mpact on materal wellbeng. Indeed those wh hgher educaton are also assocated wh beng more future orentated. Becker and Mullgan (1997) argue causaly between schoolng and tmepreference, that s, schoolng causes tme-preference for the future to rse. As such educaton fosters hgher future tme preference among ndvduals and by nducng nvestments that lower the rate of tme preference for the present (n ths case, prvate health nsurance), they may potentally mprove ther future health. Ths leads to an expectaton that those wh a hgher future tme-preference may forgo ncome now n favour of better qualy and faster access to health servces (n the prvate sector) n the future. Age and famly characterstcs, are also assocated wh materal well-beng through ther mpact on both stock of wealth. Age may act as an mportant determnant of propensy to nsure not only because s a varable assocated wh hgh ndrect rsk vulnerably and thus ncreased expected medcal consumpton but also because s also assocated wh ncreased stock of wealth. Stock of wealth generally ncreases as ndvduals/famles get older and as both Van De Ven (1981) and Hopkns and Kdd (1996) note younger ndvduals and famles are generally less well off. Specfc famly characterstcs, relatng to both famly sze and makeup, may nfluence medcal need and materal welfare. The composon of the famly un may mpact on the decson makers attude to rsk (Hopkns and Kdd 1996)- for example, presence of a spouse/and or chldren may make an ndvdual less lkely to assume rsk. In addon, 8
11 also noted by Propper (1989) (Propper 1989) and Ngu et al (Ngu and Burrows 1990) that the health of one famly member may affect the utly of other famly members, leadng them to conclude that famly composon may be a determnant of the decson to nsure. On the other hand the presence of more famly members, partcularly dependents, may lead to a lower famly wealth stock and hence lower the propensy to nsure. In emprcal lerature maral status tends to be posvely assocated and dependent chldren negatvely assocated wh the propensy to nsure (Propper 1989; Hopkns and Kdd 1996; Harmon and Nolan 2001). 3. Overvew of Irsh health and health nsurance system The Irsh health care system s an amalgam of publc and prvate provson and fnancng. Its charactersaton as a tax fnanced, unversal publc health system s somewhat msleadng as devates from ths model n a number of mportant ways. For those who qualfy for a means-tested medcal card, approxmately 30% of the populaton, general practoner servces, publc out and npatent hospal servces, and prescrbed medcaton are all provded free of charge. For the remander of the populaton, GP servces are charged to the user on a fee-per-vs bass, publc out and npatent hospal servces are heavly subsdsed by the state and provded at a nomnal user fee. Prescrbed medcaton s pad for by the user up to a maxmum monthly amount, the excess s then pad for by the state. Another dstnctve feature of the Irsh health system s the hgh proporton of the populaton wh prvate health nsurance, ths despe the exstence of a prmarly tax-fnanced, unversal publc alternatve. The prvate health nsurance system n Ireland was establshed n 1957, provded by a state-backed, non-prof, monopoly nsurer, VHI, was desgned to cater for the 9
12 top 15% of earners who, at the tme, were excluded from an entlement to free or subsdsed publc health servces. In 1987 nomnal charges for outpatent and npatent hospal care were ntroduced by the state, ths was followed n the early 1990 s wh an extenson of publc healthcare entlement to the whole populaton. Snce s ncepton prvate health nsurance n Ireland has operated under communy ratng. Income tax relef on prema has also been a sgnfcant feature of the system; however tax relef prevously pad at the top-rate tax rate pad was reduced to the standard tax rate n the md 1990 s. Under the 1994 Health Insurance Act (Nolan & Wley, 2000) the nsurance market was opened to competon, however communy ratng was retaned and n addon, rsk equalsaton was provded for. Whle theoretcally came nto effect n January 2003, ths has been postponed and contnues to be an ssue of contenton between the two man nsurers, VHI and BUPA, and the government. In 1996 BUPA entered the market and more recently, n 2004, a newly formed health nsurance company, Vvas. VHI, however, retans a sgnfcant market share, 82% to BUPA s 13% (HIA 2003).Wh s orgns to offer hospal cover to those not entled to publc care, prvate health nsurance has manly developed to provde cover for acute hospal care and typcally covers all or most npatent hospal expenses. Prvate outpatent hospal appontments are pad by the user on a pay-per-vs bass, as are GP servces. There remans a hgh deductble on both these servces; therefore nsurance makes a neglgble, f any, fnancal contrbuton to them. 5 In short, prvate health nsurance cover n Ireland s synonymous wh acute hospal npatent care; cover rarely extends 5 Recently both nsurers have begun to offer polces that make some contrbuton to GP, outpatent and other prmary care costs but ther mpact on the market thus far has been neglgble. Also s mportant to note that our data set precedes the avalably of ths type of nsurance polcy. 10
13 beyond the costs assocated wh an acute hospal stay. Although the Irsh health nsurance system s one orgnally desgned to cater for a small proporton of the populaton, there has been a steady and substantal rse n the numbers prvately nsured snce the 1950 s and currently close to 50% of the populaton are nsured. Ths relatvely hgh demand for prvate health nsurance s perhaps not altogether surprsng gven that Irsh health polcy actvely supports the prvate health sector n a number of ways. It s government polcy to contnue to faclate arrangements for prvate healthcare as the cost would otherwse fall on the state (Department of Health and Chldren 1999) A central tenet of the argument supportng ths polcy s not just the cost-savng to the publc system when patents seek treatment n the prvate sector (thereby forgong ther publc entlements and n theory, freeng up of resources n the publc sector for those that reman) but also the transfer of drect revenue to the publc system from the prvate (va nsurance payments, for example). However there s no research assessng the scale of these purported benefs to the publc system (Nolan 2004). It s clear that prvate health nsurance n Ireland s provded at below the true economc cost (Department of Health and Chldren 1999) and n turn, nsurance companes are not charged the full economc cost for prvate patents n publc hospals. 6 In addon to ths there are a number of other more ndrect ways n whch the publc system supports the prvate sector; va tax relef on nsurance purchase at 20%, tax breaks for prvate hospals, a 20:80 publc-prvate bed desgnaton whch s not strctly adhered to 7, a consultant contract whch does not 6 Although attempt to measure the extent of cross-subsdy n monetary terms has been under taken by Nolan and Wley (2001), ths work concentrates manly on areas of drect subsdy e.g. the dfference between what the actual hotel cost of a prvate patent and that charged to the nsurer. 7 Nolan & Wley (2001) note whle there s substantal crossover of prvate patents to publc beds the flow n the oppose drecton s much smaller. 11
14 specfy the extent of tme to publc patents 8 plus rembursement rules that favour the prvate patent. 9. Furthermore, no fees are charged for use of publc hospal equpment & premses when treatng prvate patents. Fnally, more generally acknowledged problems wh prvate medcne are also recognsed. The publc system absorbs cost of professonal tranng, publc hospal development & ndeed, accdent and emergency costs. Hence there seems to exst a szeable resource transfer from publc to prvate 4. Data & Prelmnary Statstcs In explorng the demand for prvate health nsurance n Ireland the Lvng n Ireland Survey s used. Ths s the frst to eghth wave of the Irsh verson of the European Communy Household Panel (ECHP). As ths s a household panel ndvduals from the same household could not be consdered ndependent. Therefore we use a subset of the sample and focus on the household reference person (HRP) as dentfed by the survey 10. Detaled nformaton categorsng ndvduals by nsurance source s avalable. However followng Besley et al (1999), who argue that provded ndvduals face the same costs n purchasng nsurance, eher out of pocket or lower wages, s consdered legmate not to dfferentate between them (Besley, Hall et al. 1999). The econometrc analyss requres consecutve observatons for the ncluson of lagged nsurance and a common date of entry (to the panel) for nal condons. Hence the panel s unbalanced wh absorbng attron. That s, ndvduals reman n the 8 Whle the type-2 consultant contract desgnates a 33-hour week to the publc sector, does not specfy the extent of ther tme commment to publc patents. In practce, publc patents are typcally left to be tended to by nonconsultant hospal doctors (the salary of whom s also pad by from publc health sector fnances), faclatng specalsts wh ths type of contract to concentrate on ther prvate patents. The Health Servce Executve however has recently deemed ths type two contract untenable n the future. 9 In addon to ths, the manner of consultant rembursement ads the preferental treatment of prvately fnanced patents. Although salared for publc, specalsts are pad on a fee per vs for prvate. As observed by Street and Duckett (1996) publc health systems have done ltle to alter the underlyng ncentves whereby those wh the greatest control over the condons of supply are rewarded rather than penalsed for mantanng wang lsts. 10 An alternatve approach mght be to examne the nsurance decson n the context of the famly un as suggested by Propper (1989) 12
15 sample at subsequent waves untl they have mssng nformaton on nsurance status or are not ntervewed at partcular wave, and drop out due to attron, ndvduals may ex the sample but no new ndvduals are added and ndvduals wh mssng data n 1994 are excluded. An examnaton of the characterstcs of those n our sample wh and whout prvate health nsurance provdes some nterestng nsghts 11. There exsts a clear dspary wh respect to percentage nsured across both educatonal and ncome levels, suggestng that the selecton nto nsurance by both those wh hgher ncomes and better educaton s substantal. Only 20% of those wh prmary educaton have prvate health nsurance, whle 86% of those who completed thrd level are nsured. Smlarly, of those n the lowest ncome quartle only 16% were nsured compared wh 59% of those n the hghest. Ths renforces the expectaton that propensy to nsure s assocated wh both educaton and ncome. ABOUT HERE- Table 1: percentage nsured Self-reported health status also regsters a notable dfference n percentage nsured. Those wh good health status are almost twce as lkely to be nsured (49%) compared wh those wh poor health status (23%). Ths does not seem to support the exstence of adverse selecton. Also worthy of note, those marred or wh partners have a hgher percentage nsured than those who are never marred or whout partners. Fnally the year old age group has the hghest percentage nsured. 11 The percentages reported here are from the HRP sample, however almost dentcal results are found for the full sample. See Appendx for a comparson of HRP and full sample descrptve statstcs. 13
16 ABOUT HERE- Table 2: Varable means A comparson of varable means for both the full and HRP sample shows that the HRP sample s a ltle less educated, older, less lkely to be eher sngle (unmarred) or whout a partner and wh slghtly poorer health status. In the full sample at wave 1 n % of the nal sample were nsured. The percentage nsured contnues to grow reachng over 48% by 2001, reflectng approxmately the proporton nsured n the populaton (Nolan and Wley, 2001). Ths also reflects closely the percentage of the populaton nsured n the sample of household representatve person used here. Furthermore an examnaton of transonal probables suggests consderable persstence n the nsured state. The frst row (Table 3) (for both samples) shows the probably of beng nsured n tme t condonal on beng nsured n t-1. Whle the second row represents the probably of beng nsured n tme t condonal on beng unnsured n tme t-1. The results suggest that havng nsurance n year t-1 s a good ndcator of whether you wll nsure n year t. ABOUT HERE- Table 3: ncdence and persstence 5. Statstcal framework In ths secton we model the decson to nsure n a lmed dependent varable framework. Usng a bnary ndcator, the ndependent varable,, represents the nsurance state of ndvdual n tme t, takng the value of 1 f the ndvdual nsures I and 0 otherwse. As such, I 1 ( * I > 0) = where 1(.) s the ndcator functon takng a 14
17 value of uny f the expresson n parenthess s true and zero otherwse. The basc model s specfed as: I * = β + βx + ν, where =1,2.N, t=1,2 7. (1) o A vector of ndvdual and household characterstcs as ndcated by X. The represents the compose error, s composed of unobserved heterogeney or unobserved tme fxed effects a and the dosyncratc error term u. ν v = a + u, (2) To control for ndvdual specfc unobserved effects, an unobserved effects bnary response model was consdered. In specfcatons (3) and (4) a represents the unobserved ndvdual specfc effect. Under random effects (3) we assume that a and u are normally dstrbuted and ndependent of X. * I = β + β X + a + u, where =1,2, 3.N; t=1, 2,3.N (3) 0 It s also assumed that there s no seral correlaton n and the X. Independence between the, are necessary for consstent results. If the assumptons for RE hold then random effects model s the most effcent and thus the preferred estmator. u a 15
18 If s assumed that the unobserved ndvdual effects are correlated wh the a X then a fxed effect specfcaton s approprate. Under the fxed effect specfcaton correlaton between the a and one or more of the X s assumed. * I = β X + a + u, where =1, 2,3.N; t=1, 2,3.N (4) Followng from ths, a Hausman test to determne random or fxed effects usng log rejects a random effects specfcaton (3) and suggests a legmate concern about unobserved effects. However, whle a fxed effects specfcaton would allow us control for unobserved effects correlated wh the explanatory varables, there s a problem wh usng a fxed effects specfcaton n ths nstance; drops tme nvarant effects, both observed and unobserved, from the model. Thus tells us nothng about varables that do not vary over tme, many of whch are varables of nterest, such as educaton. As such, f our am s to examne the ndvdual and household effects, a model that tells us ltle about tme-nvarant effects s not deal. Ths prompted us to take another approach. To control for unobserved ndvdual effects whle also ncludng explanatory varables such as educaton (whch for the most part s not expected to change over tme) we ntroduce a thrd specfcaton (5), ths takes the form of Mundlak-Chamberlan s Random Effects Model. Ths approach, dealng wh ndvdual effects correlated wh the regressors, specfes the E ( X a ) (Chamberlan 1984). A specal case, assocated wh earler work by (Mundlak 1978), uses the whn-ndvdual means of the regressors. If the assumpton of ndependence between the a and X s volated the results wll be nconsstent. We relax the 16
19 assumpton that a s ndependent of the X and attempt to control for ndvdual specfc unobserved effects correlated wh the X wh the ncluson of x, hence modelng the dependence between the a and X. Typcally ths has taken the form of a vector of tme means of tme-varyng varables, the assumpton beng that the regresson functon of the a s lnear n the tme means of tme-varyng varables (Propper and Burchardt 1999; Arulampalam, Booth et al. 2000; Propper 2000). In ths case we use only ncome as ths varable s expected to do reasonably well at capturng unobserved personaly tras assocated wh the decson to nsure, lke taste for qualy etc. In specfcaton (5), x and a represent unobserved ndvdual heterogeney. The x s the part of the unobserved ndvdual heterogeney correlated wh the X whch we attempt to model and the effects specfcaton. a n now not correlated as n a normal random * I = β + β X + x + a + u, where =1, 2,3.N; t=1, 2,3.N (5) 0 Whle certan observable characterstcs are shown to nfluence propensy to nsure, demand for prvate nsurance may persst for other reasons. Unobserved heterogeney, such as attude to rsk or taste for qualy, mght affect demand for nsurance, so mght state dependence. The latter refers to a causal relatonshp between past and current nsurance status. In short, an ndvdual unnsured n year t-1 wll behave dfferently n year t to an otherwse dentcal ndvdual nsured n year t-1. Ths mght result from an ndvdual changng preferences due to past experence of prvate nsurance or as noted by Propper (2000) from the cost of nformaton assocated wh changng nsurance 17
20 status. Descrptve statstcs gve ndcatons of very hgh nsurance persstence. To faclate examnaton of the effect of havng nsurance n year t-1 on propensy to nsure n year t and thereby consder the effect of persstence, a dynamc random effects prob model, (ncludng lagged dependent varable and correcton term whle also controllng for unobserved heterogeney) s consdered (Orme 1996; Arulampalam, Booth et al. 2000; Propper 2000). The model s specfed as follows. * β o + βx + ρi, t 1 + a1 I = X + δe + a + u, where =1,2,.N; t=2,3.n (6) As before * I represents the nsurance state of ndvdual n tme t. A vector of contemporaneous ndvdual and household characterstcs as ndcated by. The represents unobserved ndvdual heterogeney and the dosyncratc error term. X u a I s a lagged dependent varable representng the nsurance state n the prevous, t 1 year. The x s mean ncome, the tme-varyng explanatory varable, of ndvdual over tme and s ncluded to pck-up possble correlaton between the tme-varyng regressors and any unobservable heterogeny (followng Chamberlan (1984), Mundlak (1978) and outlned n Arulampalam, Booth et al (2000), Propper (2000)), allows us to model the dependence between the a and the X by assumng the regresson functon s lnear n the means of the tme varyng covarates, n ths case ncome. δ e sgnfes the nal condon correcton term. The a represents unobserved ndvdual heterogeney; s the ndvdual specfc and tme-nvarant random component. 18
21 In addon, n dynamc panel modellng, wh a lmed number of tme perods, correlaton between the a (unobserved heterogeney) and the nal observaton may result n nconsstent results (Hsao 1986; Propper 2000). To correct for ths a correcton term s added. In the spr of Heckman s standard selecton model, a reduced form equaton for the nal condon s modelled (Heckman 1981a; Heckman 1981b; Orme 1996; Arulampalam, Booth et al. 2000; Propper 2000). Ths process nvolves two man steps. Frstly an estmaton of a reduced form model for nal observaton I 1. Ths ncludes vector z, all the explanatory varables ncludng tme varyng means but also addonal presample nformaton followng Propper (2000), who uses the argument that parental factors may nfluence nal demand but not subsequent changes, we use occupaton of parent man breadwnner, age and sex. η represents a compose error term. For a further dscusson of ths methodology see Booth (2000). The reduced from model s specfed as follows. I * = λ ' z + η where =1,2,.N; t=1 (6) 1 From the reduced form equaton, a generalsed prob error, correcton term e, s generated. Ths takes the form e = 2y 1) φ( λ' z ) /( Φ({2 y 1} λ' z ) (7) (
22 Ths s used as an addonal repressor n the dynamc model to account for the correlaton between the nal condon and unobserved heterogeney Results & Dscusson In explong the panel, the mpact of certan ndvdual and household characterstcs to prvately nsure were examned usng both a fxed effect model and the random effect prob models as descrbed above. A test for selecton bas (whch works for both random and fxed effect specfcatons), ncludng a lead of selecton ndcator, as outlned by (Verbeek and Njman 1992) and n Wooldrdge (2002) was also ncluded. Ths determned that attron was not a problem. 5.1 Fxed Effects The fxed effect analyss whle controllng for unobserved effects tells us nothng about tme nvarant effects. It ncludes only those who change ther nsurance status durng the 8 years. That s, those who moved for 0 to 1 or 1 to 0 (1 ndcatng those wh prvate health nsurance, 0 ndcatng those whout). Hence only two varables are ncluded n the model and both reman sgnfcant. Income (log, equvalsed) shows a strong posve and sgnfcant effect as does poor self assessed health status. However as fxed effects models do not produce coeffcents for tme fxed effects and ths s not necessarly desrable for our purposes, an approprate strategy was to use a random 12 The new error component has Var ( a I ), whch s heteroscedastc. However accordng to a 1 Orme (and followng Orme (1996), Arulampalam (2000) and Propper (2000)) n cases of small values of rho there s no need to worry about nconsstent parameter estmates. The usual t-test for the nal condons term s a test for non-zero rho. 20
23 effect specfcaton that controls for unobserved heterogeney. If problems of omted varable bas can be solved whn a random effects specfcaton, ths s more effcent. ABOUT HERE- Table 4: Fxed effects model 5.2 Random Effects Model The prob coeffcents for three specfcatons of the random effect model are now consdered. As * I s a latent varable and nherently unobservable, s not measured n any knd of natural uns renderng the nterpretaton of the coeffcents to assume a qualatve meanng (Jones, Rce et al. Forthcomng). Thus the focus frst wll be on the sgn, relatve sze and sgnfcance of the regressor coeffcents. For the dynamc model we wll also consder, usng partal effects, the effect of the covarate on the probably of havng prvate health nsurance Statc Model The frst specfcaton (3) s a random effect prob, s assumed that there s no correlaton between the unobserved ndvdual specfc effect and the X. Consonant wh the theoretcal dscusson, the coeffcent on educaton, as wh ncome s large, posve and very sgnfcant. (Income s defned here as the aggregaton of net dsposable ncome for all household members, the ncome varable ncluded n ths analyss s log equvalsed ncome). The propensy of choosng prvate nsurance seems to rse substantally wh educaton, wh a stark contrast n propensy to nsure for those wh no qualfcatons or prmary to those wh thrd-level qualfcatons. A smlar relatonshp s shown to be the case wh respect to ncome, self a posve 21
24 correlate of educaton (Van De Ven and Van Praag 1981). Ths perhaps not unsurprsng gven that the opportuny cost of purchasng nsurance s less for those who are better off. Greater demand for medcal servces durng reproductve years assocates females wh a hgher rsk vulnerably and therefore an expectaton of hgher propensy to nsure (Hopkns and Kdd 1996). Research from the UK however fnds that despe the fact that females are more lkely to demand health servces they are less lkely to be nsured (Propper and Burchardt 1999). Whle females are shown here to have a lower propensy to nsure n ths case, s worth notng the nature of our sample and the context of our focus, whch s on the household representatve person. Although 50% of those nsured n the populaton are female, there s a much hgher proporton of males to females ( 3:1) n ths sample. Females n the HRP sample have a lesser propensy to nsure than ther male counterparts. In general, adverse selecton s a common feature of nsurance markets. Despe ths n many health systems those wh poor health tend to have a lower probably of beng prvately nsured (Doron, Jones et al. 2006). It has been suggested ths result mght be drven eher by unobserved heterogeney or cream skmmng. However the fxed effect specfcaton and the chamberlan random effects model (specfcaton (4)) attempts to control for ths, nonetheless n both models the coeffcent on self assessed health remans negatve and sgnfcant. One explanaton muted n our theoretcal framework s the jont mpact of both communy ratng and the pre-exstng condon clause, whch n the context of the Irsh nsurance market mght result n those wh poorer health havng a lower propensy to nsure. Indeed the results tend to bear ths 22
25 out; the analyss suggests that those wh poor SAHS are less lkely to be nsured than those wh good or very good SAHS. Those who are sngle (never marred) and those whout partners have a lower propensy to nsure than those who are or have been marred and those who have partner respectvely. Ths suggests that both the presence of a partner n a cohabng arrangement are more lkely to be nsured as are those who are or have a one stage been marred. The numbers of chldren, adults and elderly (over 65 s) n the household all have a negatve and sgnfcant effect on propensy to nsure. Intuvely the presence of chldren mght be expected to have a posve effect of propensy to nsure, for example f parenthood ncreases rsk adversy. However a negatve coeffcent for chldren as been found elsewhere (Hopkns and Kdd 1996; Harmon and Nolan 2001). One suggested reason for ths s unlke for the adult populaton treatment of chldren tends to be more unform between publc and prvate sectors. The mpact of dependents on both famly ncome or stock of wealth may contrbute to the latter result. Our results show that the propensy to nsure ncreases wh age, however does take on a quadratc form that suggests a smoothng of the curve, or a slower growth, as age ncreases. We have already consdered the presence of unobserved effects may have rendered random effects prob coeffcents nconsstent. Nonetheless they provde an nterestng pont of comparson wh the results for other two models (Specfcaton 5 and 6) where we attempt to control for unobserved effects usng, a random effects specfcaton, the Mundlak-Chamberlans Random Effects Model and fnally for state dependence va for the ncluson of a lagged dependent varable. 23
26 5.2.2 Mundlak-Chamberlan Model It s found that coeffcents specfcally on educaton and ncome, partcularly large usng specfcaton (3), are szeably reduced when unobserved effects (5) and then state dependence are controlled for (6). Focusng frst on specfcaton (5), Chamberlans random effects model, a tme mean of ncome s assumed to capture any ndvdual unobserved effect assocated wh the explanatory varables. By defnon s assumed that these unobserved effects are assocated wh ncome. Such unobserved effects mght nclude taste for qualy, rsk adversy or stock of wealth. Controllng thus for unobserved ndvdual specfc effects, the coeffcents on many covarates, whle remanng sgnfcant, are smaller n sze. Ths s partcularly true of the varable for ncome, but not unsurprsng gven that ntroducton of the tme mean for ncome. There are two possble nterpretatons; the Mundlak-Chamberlan nterpretaton vews ths varable as presentng unobserved ndvdual effects leavng the coeffcent on ncome to represent that true ncome effect when unobserved effects are controlled for. However another nterpretaton would be to vew the tme mean as a permanent or long run ncome effect, wh the coeffcent on the ncome varable as the effect of a transory ncome shock or current ncome. The am here s to specfcally to control for unobserved heterogeney and thus we wll veer toward the Mundlak-Chamberlan nterpretaton. However n nterpretng the true effect of ncome, some acknowledgement that longer ncome effects, vewed as stock of wealth perhaps, are captured by the tme mean along wh other unobserved effects. Although the sze of the coeffcent on ncome has halved compared to the random effect specfcaton (3), the effect of current ncome s both posve and sgnfcant. The coeffcent of tme mean representng unobserved effects s very large, posve 24
27 and sgnfcant, suggestng that unobserved ndvdual heterogeney s an mportant determnant of propensy to nsure. Precsely what unobserved effects mght be drvng ths result however s unknown. However the man focus here s not the tme mean varable self but to what extent the other regressors reman robust to s ncluson. Educaton remans robust to the ncluson of unobserved effects. Whle there s some reducton n the overall sze of the coeffcents, the effect remans large across all levels of educaton; thrd-level educaton remans the strongest determnant of propensy to nsure across educaton levels and across the covarates. The sze, sgn and sgnfcance of the coeffcent on health status reman more or less equal to that of the prevous specfcaton, wh only very small reducton n coeffcent sze. Those who have poor health status have a lower propensy to nsure to those wh good health status. Combned wh the fxed effects specfcaton, ths result seems to provde rather strong evdence that ndvdual unobserved heterogeney s not drvng ths result. Somewhat contrary to fndngs that suggest ndvdual unobserved heterogeney as a reason for the negatve effect of health status on demand for prvate health nsurance (Doron, Jones et al. 2006). A more n-depth examnaton of the queston to shed further lght on ths matter wh respect to the Irsh data would be useful, certanly the use of a better measure of health mght help. Age remans a posve and sgnfcant, whle the coeffcent on female remans negatve, however the sgnfcance of the coeffcent on the latter s weaker then s specfcaton (1) remanng sgnfcant only at the 10% level. Those never marred or no partner are stll less lkely to have nsurance. However the coeffcents on number of chldren and adults are now nsgnfcant n the presence of unobserved effects, whle the number of elderly, remanng sgnfcant has an even stronger negatve effect. 25
28 5.2.3 Dynamc Model Supported by evdence of consderable persstence from transonal probables examned earler, Specfcaton (6) ncludes not only the tme mean of ncome to model the dependence between the a and the X, but also a lagged dependent varable (to capture state dependence) and a correcton term (to correct for the nal condon problem). We also report margnal effects. Despe controllng for both unobserved effects and state dependence, wh the ncluson of a lagged dependence varable, the effect of educaton s stll large and sgnfcant. Ths suggests that over and above havng nsurance last year (year t-1), educaton matters and sgnfcantly so. The probably of beng prvately nsured ncreases substantally wh levels of educaton. The margnal effects of educaton level report a stark contrast between those who have prmary or no qualfcatons to those wh thrd-level qualfcatons. Compared wh those who have prmary or no qualfcatons the probably of havng prmary nsurance ncreases by 16% for those wh lower second level or junor cert qualfcatons, 29% for those wh upper second or leavng cert qualfcatons and fnally by 43% for those wh thrd level. Ths s a substantal ncrease n probably to nsure gven educatonal attanment. Indeed of those who were not nsured last year, the better educated are more lkely to nsure ths year. Smlarly, for those who were nsured, the better educated are most lkely to retan. It also s nterestng to note, despe the strong persstence effect, whch we wll dscuss shortly, educaton stll matters. Gven that we control for other soco-economc and famly characterstcs such as ncome, ths mght be explaned by a number of possbles. For example, the hgher educated may formulate a better 26
29 mxture of health-generatng nputs, of whch nsurance mght be one (Grossman, 1972). They mght also have preferences for hgher qualy healthcare, whch s perceved as a benef of nsurance. Smlarly there may be some unobserved effect relatng to educaton, but not to ncome, nfluencng the result 13. Current ncome s also posve and sgnfcant. Those wh hgh current ncome have a hgher propensy to nsure. An 11% ncrease n the probably of havng prvate health nsurance s reported f current ncome s doubled. Whle ths ndcatve of an ncome-gradent wh respect to probably to nsure, also suggests that an mmedate short-term ncrease n ncome would not result n substantal ncreases n demand for prvate health nsurance. However we must qualfy ths latter statement and the seemngly modest result by acknowledgng that unobserved effects varable, the tme mean for ncome, mght nclude longer-term ncome effects. If nterpreted n ths way the ncome effect rses sharply wh a 42% ncrease n nsurng. Fnally, whle a posve and sgnfcant effect of ncome on propensy to nsure s not unexpected, some argument has been made to support the hypothess that hgh-ncome famles are more lkely to assume rsk and not nsure because f necessary they can afford out-of-pocket payments for prvate health servces. The posve and sgnfcant coeffcent does not lend self to ths explanaton. In lne wh the fndngs of Harmon and Nolan (2001) who do not fnd sgnfcant self-selecton of those wh poor health nto nsurance, the effect of poor health status on propensy to nsure remans both negatve and sgnfcant. Those wh 13 Another explanaton mght be the nfluence of some employment effect. Although the numbers wh employer-bought nsurance s small, employer-bought health nsurance s assocated wh employment requrng hgher levels of educaton 27
30 poor health have a 10% less probably of beng nsured than those wh good health. 14 Ths coeffcent s not really affected by the ncluson of the lag and perhaps not unexpectedly. Overall the results show that those wh poor health status are less lkely to nsure than those wh good health status. However when we focus specfcally on those wh poor health, those wh no nsurance n t-1 are much less lkely to be nsured that those nsured n year t-1. Ths s suggestve that nsurance status n year t-1 affects whether those who have poor health status n year t are nsured or not. Ths may be explaned by the nsurance state changng preferences or a hgh cost for changng status, n the context of the Irsh system there exsts a sx-month wang tme for all condons and a fve-year wa for preexstng condons f you drop out and then decde to re-nsure at a later date. Fnally, gven that we mght expect the utly of nsurance for those wh poor health to be hgher; those wh poor health are stll less lkely to retan nsurance than those wh good health. Indeed, as already mentoned, nsurance s prmarly used n the acute hospal sector, so those who are nsured typcally have no costs for an npatent care. Ths suggests for some, gven ther specfc health condon, there s no benef or utly n remanng nsured. Ths mght pertan to those wh chronc llness who prmarly requre care non-acute/outpatent care. Furthermore f the treatment for ther poor health n the publc sector equates to that n the prvate sector then there s perhaps no utly for beng nsured. In addon, although tax relef s granted on the grounds that enables the chroncally ll to aval of prvate health nsurance at a reasonable cost, those wh poor health status are shown to have lower probably of beng prvate nsured than those wh good or very good health. From the perspectve of the nsurance 14 A further exploraton of ths counter-ntuve result and the exact nature of ths relatonshp would requre better ndcators of health than a subjectve health measure. 28
31 company the rules of provson work well and seem to guard adequately aganst adverse selecton. The coeffcent for the lagged dependent varable representng state dependence s very large, posve and sgnfcant ndcatng a strong persstence effect. Indeed the margnal effect of prvate nsurance n t-1 shows a very strong persstence effect, those who had nsurance last year, holdng everythng else equal, are 67% more lkely to nsure ths year. A hgh degree of persstence reveals that once nsured those wh prvate nsurance tend to keep. What s clear also that an ndvdual unnsured n year t-1 wll behave dfferently n year t to an otherwse dentcal ndvdual nsured n year t-1. Thus the state of nsurance n year t-1 nfluences your state n t. Agan ths mght be due to past experence of nsurance creatng a change of preference, the cost of nformaton assocated wh changng preference or the cost of changng from nsurance to no nsurance, whch n Ireland takes the form of loss of a sx month entry wa or a fve year wa for pre-exstng condons (Propper 2000). Although the sze of the coeffcent s large, wh transonal probables (Table 3) above 90% we mght have expected an even greater effect. One explanaton mght be for those wh certan characterstcs n the populaton of household representatves, the better educated, wealther, healther and nsured last year have an extremely hgh probably of beng nsured n year t. In short these characterstcs are almost fully predctve of the nsured state. On the other hand whle nsurance n t-1 s hghly predctve of nsurance n year t, the fact remans that that those nsured n t-1 wh the lowest educaton, lower ncome and poor health status are much less lkely to be nsured n t, than ther better-off counterparts. On another note, ths hgh degree of persstence may also have some polcy uses. For example a short-term ncentve to 29
32 entce adults under 30 nto the nsurance market may be que effectve. Recent worres regardng the age profle of the nsurance pool has led to a proposal of a small change n the rules of provson; those who nsure after 30 wll be charged an extra 2% levy per year. What the result here shows that f the non-nsured can be entced to purchase prvate nsurance the probably of them retanng n the future s extremely large. Overall the posve effect of educaton and ncome and the negatve effect of poor heath status reman robust across three specfcatons. The better educated, hgher ncome and healther have a hgher propensy to nsure. What s clear also s that there s a hgh degree of persstence. 6. Conclusons In ths paper we show that the nsured are better educated, wealther and healther than the unnsured. Ths n self s not necessarly pecular. However n the Irsh context poses some partcular ssues for the polcymaker. Much of prvate medcne s carred out whn the publc hospal sector and s perceved to be of a hgher qualy 15. There are potentally large drect and ndrect subsdes from publc fnances n support of prvate health nsurance and the prvate health sector. These subsdes are justfed on the grounds that nsurance s, thus, both generally affordable and ads those such as the chroncally ll n nsurance take-up. However polcy amed at promotng nsurance takeup (through ncentves such as communy ratng and cross-subsdzaton), as a way n 15 The prvate sector s perceved as havng a hgher qualy than that offered by the publc sector (wh wang tmes to specalst supportve of ths) (Watson, D. and J. Wllams (2001). "Perceptons of the Qualy of Health Care n the Publc and Prvate Sectors n Ireland: Report to the Centre for Insurance Studes Graduate Busness School, UCD." Books and Monographs Seres, ESRI No
33 whch to delver qualy healthcare, needs to address that s prmarly successful at encouragng those at the hgher ends of the educaton and ncome dstrbutons, and those wh better health to nsure. In movng toward a dynamc specfcaton we show that persstence s a hghly sgnfcant determnant of demand for prvate health nsurance and also that reduces the sze of the coeffcents on the regressors. The latter pont hghlghts that whle educaton, ncome and, to a lesser extent, health status have very large effects on probably of nsurng, these effects are overestmated where no attempt s made to control for unobserved heterogeney or state dependence. From an nsurance provder perspectve ths hgh degree of persstence mght have some polcy uses. Recent worres regardng the age profle of the nsurance pool has prompted attempts to exact a small change n the rules of provson, whch would result n those who nsure after 30 to be charged an extra 2% levy per year on prema. What our results suggests s that f the non-nsured can be entced to purchase prvate nsurance the probably of them retanng n the future s extremely large (ths agan mght be due to a change n preferences or the costs assocated wh changng status.e. cost of nformaton or wang). As such, a short-term ncentve to entce adults under 30 nto the nsurance market may be que effectve. A lower probably of havng nsurance for those wh poor health suggests that adverse selecton s not a problem. Ths despe the fact that the system s communy rated. For nsurance companes ths s good news and suggests that a pre-exstng condon crera desgn to combat the problem s workng well. For the unnsured however specfcally those wh poor health or a chronc condon ths s more worrsome. Some cross-subsdes from publc to prvate, such as tax-relef on nsurance 31
34 prema, have been justfed on the grounds that they ad those wh chronc llness purchase nsurance. What does n fact do s lower the prce of nsurance to the nsured, usng taxpayer revenue that everyone pays, ncludng many of those unnsured wh ll health. It could be argued that government polcy thus faclates targetng. In short, n so much as can under a communy rated system, renforces the nsurer preference, and thus works n the nterest of the nsurance company. In ths system, wh stated equy goals, the close ntertwnng of the publc and prvate sector, the problematc nature of the dual wang lsts system, the perceved qualy dfferences n the publc and prvate sector coupled wh an nsurance system that seems to encourage a certan profle of nsured, polcy mght be vewed as contrbutng to neques to access and qualy of care. 32
35 BIBLIOGRAPHY Arrow, K. J. (1963). "Uncertanty and the Welfare Economcs of Medcal Care." The Amercan Economc Revew 53(No. 5): Arulampalam, W., A. L. Booth, et al. (2000). "Umemployment Persstence." Oxford Economc Papers 52(1): Becker, G. (1965). "A Theory of the Allocaton of Tme." The Economc Journal Vol. LXXV(No. 299): Becker, G. and C. Mullgan (1997). "The Endogenous Determnaton of Tme Preference." The Quarterly Journal of Economcs 112(3): Besley, T. (1991). "Publc Provson of Prvate Goods and the Redstrbuton of Income." Amercan Economc Revew 81(4): Besley, T., J. Hall, et al. (1998). "Prvate and publc health nsurance n the UK." European Economc Revew Vol. 42: Besley, T., J. Hall, et al. (1999). "The demand for prvate health nsurance: do wang lsts matter?" Journal of publc economcs 72: Chamberlan, G. (1984). "Panel Data." Chapter, n Handbook of Econometrcs, Volume 1, Z. Grlches and M.D Intrlgator: Department of Health and Chldren (1999). Whe Paper, Prvate Health Insurance, Dubln: Government of Ireland. Doron, D., G. Jones, et al. (2006). "Healthy, wealthy and nsured? The role of selfassessed health n the demand for prvate health nsurance." CHERE Workng Paper 2006/1, CHERE, Sydney, Feldsten, M. S. (1973). "The Welfare Loss of Excess Health Insurance." The Journal of Polcal Economy 81(2, part 1): Grossman (1972). "On the concept of health capal and the demand for health." Journal of polcal economy 80: Grossman, M. (1999). "The Human Capal Model of Demand for Health." NBER Workng Paper Hamermesh, D. (2004). "Subjectve Outcomes n Economcs." Southern Economc Journal Vol
36 Harmon, C. and B. Nolan (2001). "Health Insurance and Health Servce Utlsaton n Ireland." Health Economcs 10: Heckman, J. (1981a). "The Incdental parameters problem and the problem of nal condons s estmatng a dscrete tme-dscrete data stochastc process." n Structural analyss of dscrete data wh econometrc applcatons, Mansk, C.F. and McFadden, D. (eds): Heckman, J. J. (1981b). "Statstcal Model for Dscrete Panel Data." n Structural analyss of dscrete data wh econometrc applcatons, Mansk, C.F. and McFadden, D. (eds). HIA (2003). The Prvate Health Insurance Market n Ireland, Dubln: Health Insurance Authory. Hopkns, S. and M. Kdd (1996). "The determnants of the demand for prvate health nsurance under Medcare." Appled Economcs 28: Hsao, C. (1986). Analyss of Panel Data, Cambrdge Unversy Press: Cambrdge. Jones, A., N. Rce, et al. (Forthcomng). Appled Health Economcs. Mundlak, Y. (1978). "On the poolng of tme-seres and cross-secton data." Econometrca 46(1): Ngu, M. and C. e. a. Burrows (1990). "Health Insurance Choce: An econometrc analyss of ABS health and health nsurance surveys, economcs and health." Proceedng from the Australan Conference of Health Economsts: Nolan, B. (2004). "Health Insurance n Ireland: Issues and Challenges." Econome Publque No /1. Nolan, B. and M. Wley (2000). Prvate Practce n Irsh Publc Hospals, Oak Tree Press. Orme, C. D. (1996). "The Inal Condons Problem and Two-Step Estmaton n Dscrete Panel Data Models." Unversy of Manchester, School of Economc Studes, Dscusson Paper Seres. Propper, C. (1989). "An econometrc analyss of the demand for prvate health nsurance n England and Wales." Appled Economcs 21: Propper, C. (2000). "Demand for prvate health care n the UK." Journal of Health Economcs 19: Propper, C. and T. Burchardt (1999). "Does the UK have a prvate welfare class?" Jnl Soc. Pol 28(4):
37 Van De Ven, W. and B. Van Praag (1981). "The demand for deductbles n prvate health nsurance." Journal of Econometrcs 17: Verbeek and Njman (1992). "Selectvy Bas n Panel Data Models." Internatonal Economc Revew 3(No.3). Watson, D. and J. Wllams (2001). "Perceptons of the Qualy of Health Care n the Publc and Prvate Sectors n Ireland: Report to the Centre for Insurance Studes Graduate Busness School, UCD." Books and Monographs Seres, ESRI No
38 Table 1: Percentage nsured Full Sample HRP Sample % % N % % N Insured Not Insured Insured Not Insured Overall , ,666 Prmary , ,452 Lower Second , ,419 Upper Second , ,751 Thrd level , ,982 Female , ,064 Male , ,602 Sngle , ,994 (never marred) Marred , ,672 No partner Partner , , , ,378 Income , ,974 Income , ,624 Income , ,345 Income , ,701 Age , ,243 Age , ,761 Age , ,104 Age , ,558 Poor Health , ,921 Good Health , ,713 Chronc condon , ,976 No chronc , ,593 *HRP; Household Representatve Person 36
39 Table 2: Varable means Sample FULL HRP N Have nsurance Lower second Upper second Thrd level Age Female Sngle No partner Chldren Adults Elderly Income (equv) Poor SAH Chronc Table 3: Transonal Probables Wave Percent nsured Full HRP Condonal Probables Full Prob(Yt=1/Yt-1=1) Prob(Yt=1/Yt-1=0) HRP Prob(Yt=1/Yt-1=1) Prob(Yt=1/Yt-1=0) Sample sze Full (44886) HRP (20666)
40 Table 4: Fxed Effects Log Insurance status Coeff Std. Err. Log equv. ncome ** Poor health status *** Tme Dummes YES log lkelhood = Number of obs = 3564 Table 5: Random Effects Prob Statc (3) Chamberlan (5) Dynamc (6) Insurance status Coeff. Std. Err Coeff. Std. Err Coeff. Std. Err lower second 0.87 *** *** *** 0.06 Upper second 2.00 *** *** *** 0.07 Thrdlevel 3.21 *** *** *** 0.10 Age 0.22 *** *** Age squared 0.00 *** *** Female *** ** Sngle *** *** ** 0.10 Nopartner *** *** *** 0.08 No. of chldren *** No. of adults *** ** 0.02 No. of elderly *** *** *** 0.06 Log equv. ncome 1.07 *** *** *** 0.07 Poor health status *** *** *** 0.06 Prvate t *** 0.06 correcton *** 0.05 Tme mean *** *** 0.10 Tme Dummes YES YES YES Regons YES YES YES _cons *** *** 0.41 Log lkelhood Number of obs rho *** Sgnfcant at the 1% level, ** Sgnfcant at the 5% level, * Sgnfcant at the 10% level 38
A dynamic model of demand for private health insurance in Ireland
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
HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA*
HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA* Luísa Farnha** 1. INTRODUCTION The rapd growth n Portuguese households ndebtedness n the past few years ncreased the concerns that debt
benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).
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
Can Auto Liability Insurance Purchases Signal Risk Attitude?
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
An Alternative Way to Measure Private Equity Performance
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
Analysis of Premium Liabilities for Australian Lines of Business
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
! # %& ( ) +,../ 0 1 2 3 4 0 4 # 5##&.6 7% 8 # 0 4 2 #...
! # %& ( ) +,../ 0 1 2 3 4 0 4 # 5##&.6 7% 8 # 0 4 2 #... 9 Sheffeld Economc Research Paper Seres SERP Number: 2011010 ISSN 1749-8368 Sarah Brown, Aurora Ortz-Núñez and Karl Taylor Educatonal loans and
Gender differences in revealed risk taking: evidence from mutual fund investors
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
Criminal Justice System on Crime *
On the Impact of the NSW Crmnal Justce System on Crme * Dr Vasls Sarafds, Dscplne of Operatons Management and Econometrcs Unversty of Sydney * Ths presentaton s based on jont work wth Rchard Kelaher 1
LIFETIME INCOME OPTIONS
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
PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIGIOUS AFFILIATION AND PARTICIPATION
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
Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur
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..
Staff Paper. Farm Savings Accounts: Examining Income Variability, Eligibility, and Benefits. Brent Gloy, Eddy LaDue, and Charles Cuykendall
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
Answer: A). There is a flatter IS curve in the high MPC economy. Original LM LM after increase in M. IS curve for low MPC economy
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.
The demand for private health care in the UK
Journal of Health Economcs 19 2000 855 876 www.elsever.nlrlocatereconbase The demand for prvate health care n the UK Carol Propper ) Department of Economcs, CASE and CEPR, UnÕersty of Brstol, Brstol BS8
1.1 The University may award Higher Doctorate degrees as specified from time-to-time in UPR AS11 1.
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
Traffic-light a stress test for life insurance provisions
MEMORANDUM Date 006-09-7 Authors Bengt von Bahr, Göran Ronge Traffc-lght a stress test for lfe nsurance provsons Fnansnspetonen P.O. Box 6750 SE-113 85 Stocholm [Sveavägen 167] Tel +46 8 787 80 00 Fax
Health Insurance and Household Savings
Health Insurance and Household Savngs Mnchung Hsu Job Market Paper Last Updated: November, 2006 Abstract Recent emprcal studes have documented a puzzlng pattern of household savngs n the U.S.: households
How To Calculate The Accountng Perod Of Nequalty
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.
The impact of hard discount control mechanism on the discount volatility of UK closed-end funds
Investment Management and Fnancal Innovatons, Volume 10, Issue 3, 2013 Ahmed F. Salhn (Egypt) The mpact of hard dscount control mechansm on the dscount volatlty of UK closed-end funds Abstract The mpact
Traditional versus Online Courses, Efforts, and Learning Performance
Tradtonal versus Onlne Courses, Efforts, and Learnng Performance Kuang-Cheng Tseng, Department of Internatonal Trade, Chung-Yuan Chrstan Unversty, Tawan Shan-Yng Chu, Department of Internatonal Trade,
How To Study The Nfluence Of Health Insurance On Swtchng
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
ADVERSE SELECTION IN INSURANCE MARKETS: POLICYHOLDER EVIDENCE FROM THE U.K. ANNUITY MARKET *
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
------ Evidence from a Natural Experiment in China
Are Assets n Medcal Savngs Accounts Dscounted? ------ Evdence from a Natural Experment n Chna Maoyong Fan, Zhen Le and Guoen Lu November 23rd, 2010 Abstract: In Chna, Medcal Savngs Accounts (MSAs) are
Day-of-the-Week Trading Patterns of Individual and Institutional Investors
Day-of-the-Week Tradng Patterns of Indvdual and Instutonal Investors Joel N. Morse, Hoang Nguyen, and Hao M. Quach Ths study examnes the day-of-the-week tradng patterns of ndvdual and nstutonal nvestors.
Tuition Fee Loan application notes
Tuton Fee Loan applcaton notes for new part-tme EU students 2012/13 About these notes These notes should be read along wth your Tuton Fee Loan applcaton form. The notes are splt nto three parts: Part 1
Using Series to Analyze Financial Situations: Present Value
2.8 Usng Seres to Analyze Fnancal Stuatons: Present Value In the prevous secton, you learned how to calculate the amount, or future value, of an ordnary smple annuty. The amount s the sum of the accumulated
DEFINING %COMPLETE IN MICROSOFT PROJECT
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,
Macro Factors and Volatility of Treasury Bond Returns
Macro Factors and Volatlty of Treasury Bond Returns Jngzh Huang Department of Fnance Smeal Colleage of Busness Pennsylvana State Unversty Unversty Park, PA 16802, U.S.A. Le Lu School of Fnance Shangha
HARVARD John M. Olin Center for Law, Economics, and Business
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
Evaluating the Effects of FUNDEF on Wages and Test Scores in Brazil *
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
To manage leave, meeting institutional requirements and treating individual staff members fairly and consistently.
Corporate Polces & Procedures Human Resources - Document CPP216 Leave Management Frst Produced: Current Verson: Past Revsons: Revew Cycle: Apples From: 09/09/09 26/10/12 09/09/09 3 years Immedately Authorsaton:
Statistical Methods to Develop Rating Models
Statstcal Methods to Develop Ratng Models [Evelyn Hayden and Danel Porath, Österrechsche Natonalbank and Unversty of Appled Scences at Manz] Source: The Basel II Rsk Parameters Estmaton, Valdaton, and
Student Performance in Online Quizzes as a Function of Time in Undergraduate Financial Management Courses
Student Performance n Onlne Quzzes as a Functon of Tme n Undergraduate Fnancal Management Courses Olver Schnusenberg The Unversty of North Florda ABSTRACT An nterestng research queston n lght of recent
Testing Adverse Selection Using Frank Copula Approach in Iran Insurance Markets
Journal of mathematcs and computer Scence 5 (05) 54-58 Testng Adverse Selecton Usng Frank Copula Approach n Iran Insurance Markets Had Safar Katesar,, Behrouz Fath Vajargah Departmet of Statstcs, Shahd
Hollinger Canadian Publishing Holdings Co. ( HCPH ) proceeding under the Companies Creditors Arrangement Act ( CCAA )
February 17, 2011 Andrew J. Hatnay [email protected] Dear Sr/Madam: Re: Re: Hollnger Canadan Publshng Holdngs Co. ( HCPH ) proceedng under the Companes Credtors Arrangement Act ( CCAA ) Update on CCAA Proceedngs
When Talk is Free : The Effect of Tariff Structure on Usage under Two- and Three-Part Tariffs
0 When Talk s Free : The Effect of Tarff Structure on Usage under Two- and Three-Part Tarffs Eva Ascarza Ana Lambrecht Naufel Vlcassm July 2012 (Forthcomng at Journal of Marketng Research) Eva Ascarza
TESTING FOR EVIDENCE OF ADVERSE SELECTION IN DEVELOPING AUTOMOBILE INSURANCE MARKET. Oksana Lyashuk
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
The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis
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 protected] Abstract.
Financial Instability and Life Insurance Demand + Mahito Okura *
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
Returns to Experience in Mozambique: A Nonparametric Regression Approach
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
PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12
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
Searching and Switching: Empirical estimates of consumer behaviour in regulated markets
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
Military Conscription and University Enrolment: Evidence from Italy
DISCUSSION PAPER SERIES IZA DP No. 4212 Mltary Conscrpton and Unversty Enrolment: Evdence from Italy Gorgo D Petro June 2009 Forschungsnsttut zur Zukunft der Arbet Insttute for the Study of Labor Mltary
Multiple-Period Attribution: Residuals and Compounding
Multple-Perod Attrbuton: Resduals and Compoundng Our revewer gave these authors full marks for dealng wth an ssue that performance measurers and vendors often regard as propretary nformaton. In 1994, Dens
Marginal Returns to Education For Teachers
The Onlne Journal of New Horzons n Educaton Volume 4, Issue 3 MargnalReturnstoEducatonForTeachers RamleeIsmal,MarnahAwang ABSTRACT FacultyofManagementand Economcs UnverstPenddkanSultan Idrs [email protected]
SIMPLE LINEAR CORRELATION
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.
1. Measuring association using correlation and regression
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
Fixed income risk attribution
5 Fxed ncome rsk attrbuton Chthra Krshnamurth RskMetrcs Group [email protected] We compare the rsk of the actve portfolo wth that of the benchmark and segment the dfference between the two
Heterogeneous Paths Through College: Detailed Patterns and Relationships with Graduation and Earnings
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
Number of Levels Cumulative Annual operating Income per year construction costs costs ($) ($) ($) 1 600,000 35,000 100,000 2 2,200,000 60,000 350,000
Problem Set 5 Solutons 1 MIT s consderng buldng a new car park near Kendall Square. o unversty funds are avalable (overhead rates are under pressure and the new faclty would have to pay for tself from
SUPPLIER FINANCING AND STOCK MANAGEMENT. A JOINT VIEW.
SUPPLIER FINANCING AND STOCK MANAGEMENT. A JOINT VIEW. Lucía Isabel García Cebrán Departamento de Economía y Dreccón de Empresas Unversdad de Zaragoza Gran Vía, 2 50.005 Zaragoza (Span) Phone: 976-76-10-00
Exhaustive Regression. An Exploration of Regression-Based Data Mining Techniques Using Super Computation
Exhaustve Regresson An Exploraton of Regresson-Based Data Mnng Technques Usng Super Computaton Antony Daves, Ph.D. Assocate Professor of Economcs Duquesne Unversty Pttsburgh, PA 58 Research Fellow The
CHAPTER 14 MORE ABOUT REGRESSION
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
Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic
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
Small pots lump sum payment instruction
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
The OC Curve of Attribute Acceptance Plans
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
! ## % & ( ) + & ) ) ),. / 0 ## #1#
! ## % & ( ) + & ) ) ),. / 0 12 345 4 ## #1# 6 Sheffeld Economc Research Paper Seres SERP Number: 2006010 ISSN 1749-8368 Pamela Lenton* The Cost Structure of Hgher Educaton n Further Educaton Colleges
Course outline. Financial Time Series Analysis. Overview. Data analysis. Predictive signal. Trading strategy
Fnancal Tme Seres Analyss Patrck McSharry [email protected] www.mcsharry.net Trnty Term 2014 Mathematcal Insttute Unversty of Oxford Course outlne 1. Data analyss, probablty, correlatons, vsualsaton
Impact of Attribution Metrics on Return on Keyword Investment. in Paid Search Advertising
Impact of Attrbuton Metrcs on Return on Keyword Investment n Pad Search Advertsng Hongshuang (Alce) L 1 P. K. Kannan Sva Vswanathan Abhshek Pan June 3, 2014 1 Hongshuang (Alce) L s Assstant Professor of
Wage inequality and returns to schooling in Europe: a semi-parametric approach using EU-SILC data
MPRA Munch Personal RePEc Archve Wage nequalty and returns to schoolng n Europe: a sem-parametrc approach usng EU-SILC data Marco Bagett and Sergo Sccchtano Unversty La Sapenza Rome, Mnstry of Economc
Forecasting the Direction and Strength of Stock Market Movement
Forecastng the Drecton and Strength of Stock Market Movement Jngwe Chen Mng Chen Nan Ye [email protected] [email protected] [email protected] Abstract - Stock market s one of the most complcated systems
Are Women Better Loan Officers?
Are Women Better Loan Offcers? Ths verson: February 2009 Thorsten Beck * CentER, Dept. of Economcs, Tlburg Unversty and CEPR Patrck Behr Goethe Unversty Frankfurt André Güttler European Busness School
Problem Set 3. a) We are asked how people will react, if the interest rate i on bonds is negative.
Queston roblem Set 3 a) We are asked how people wll react, f the nterest rate on bonds s negatve. When
A DYNAMIC ANALYSIS OF
A DYNAMIC ANALYSIS OF THE DEMAND FOR LIFE INSURANCE Andre P. Lebenberg (contact author) Faculty of Fnance The Unversty of Msssspp Oxford, MS 38677 [email protected] Tel: 662.915.3844 James M. Carson
Chapter 8 Group-based Lending and Adverse Selection: A Study on Risk Behavior and Group Formation 1
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
The Current Employment Statistics (CES) survey,
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,
What is Candidate Sampling
What s Canddate Samplng Say we have a multclass or mult label problem where each tranng example ( x, T ) conssts of a context x a small (mult)set of target classes T out of a large unverse L of possble
STAMP DUTY ON SHARES AND ITS EFFECT ON SHARE PRICES
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
Risk-based Fatigue Estimate of Deep Water Risers -- Course Project for EM388F: Fracture Mechanics, Spring 2008
Rsk-based Fatgue Estmate of Deep Water Rsers -- Course Project for EM388F: Fracture Mechancs, Sprng 2008 Chen Sh Department of Cvl, Archtectural, and Envronmental Engneerng The Unversty of Texas at Austn
Adverse Selection, Welfare and the Optimal Pricing of Employer- Sponsored Health Plans
Adverse Selecton Welfare and the Optmal Prcng of Employer- Sponsored Health Plans Carolne Carln Unversy of Mnnesota Robert Town Unversy of Mnnesota and NBER Aprl 009 Revson n Process Comments Welcome Abstract
Does Higher Education Enhance Migration?
DISCUSSION PAPER SERIES IZA DP No. 7754 Does Hgher Educaton Enhance Mgraton? Mka Haapanen Petr Böckerman November 2013 Forschungsnsttut zur Zukunft der Arbet Insttute for the Study of Labor Does Hgher
Job satisfaction among US Ph.D. graduates: the effects of gender and employment sector
Job satsfacton among US Ph.D. graduates: the effects of gender and employment sector Phlppe Moguérou 1 IREDU, CNRS-Unversté de Bourgogne (Djon, France) and SPRU, Unversty of Sussex (Brghton, UK) Frst draft,
An Empirical Study of Search Engine Advertising Effectiveness
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
Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting
Causal, Explanatory Forecastng Assumes cause-and-effect relatonshp between system nputs and ts output Forecastng wth Regresson Analyss Rchard S. Barr Inputs System Cause + Effect Relatonshp The job of
Management Quality and Equity Issue Characteristics: A Comparison of SEOs and IPOs
Management Qualty and Equty Issue Characterstcs: A Comparson of SEOs and IPOs Thomas J. Chemmanur * Imants Paegls ** and Karen Smonyan *** Current verson: November 2009 (Accepted, Fnancal Management, February
What should (public) health insurance cover?
Journal of Health Economcs 26 (27) 251 262 What should (publc) health nsurance cover? Mchael Hoel Department of Economcs, Unversty of Oslo, P.O. Box 195 Blndern, N-317 Oslo, Norway Receved 29 Aprl 25;
14.74 Lecture 5: Health (2)
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,
CHAPTER 5 RELATIONSHIPS BETWEEN QUANTITATIVE VARIABLES
CHAPTER 5 RELATIONSHIPS BETWEEN QUANTITATIVE VARIABLES In ths chapter, we wll learn how to descrbe the relatonshp between two quanttatve varables. Remember (from Chapter 2) that the terms quanttatve varable
Transition Matrix Models of Consumer Credit Ratings
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
A Model of Private Equity Fund Compensation
A Model of Prvate Equty Fund Compensaton Wonho Wlson Cho Andrew Metrck Ayako Yasuda KAIST Yale School of Management Unversty of Calforna at Davs June 26, 2011 Abstract: Ths paper analyzes the economcs
Calculating the high frequency transmission line parameters of power cables
< ' Calculatng the hgh frequency transmsson lne parameters of power cables Authors: Dr. John Dcknson, Laboratory Servces Manager, N 0 RW E B Communcatons Mr. Peter J. Ncholson, Project Assgnment Manager,
17 Capital tax competition
17 Captal tax competton 17.1 Introducton Governments would lke to tax a varety of transactons that ncreasngly appear to be moble across jursdctonal boundares. Ths creates one obvous problem: tax base flght.
High Correlation between Net Promoter Score and the Development of Consumers' Willingness to Pay (Empirical Evidence from European Mobile Markets)
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
An Interest-Oriented Network Evolution Mechanism for Online Communities
An Interest-Orented Network Evoluton Mechansm for Onlne Communtes Cahong Sun and Xaopng Yang School of Informaton, Renmn Unversty of Chna, Bejng 100872, P.R. Chna {chsun,yang}@ruc.edu.cn Abstract. Onlne
Financial Mathemetics
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,
PUBLIC VS. PRIVATE HEALTH CARE SERVICES DEMAND IN ITALY. A COUNT DATA ANALYSIS ON SHAW DATA
XIV CONFERENZA IL FUTURO DEI SISTEMI DI WELFARE NAZIONALI TRA INTEGRAZIONE EUROPEA E DECENTRAMENTO REGIONALE coordnamento, competzone, mobltà Pava, Unverstà, 4-5 ottobre 2002 PUBLIC VS. PRIVATE HEALTH
