Setting health plan premiums to ensure efficient quality in health care: minimum variance optimal risk adjustment

Size: px
Start display at page:

Download "Setting health plan premiums to ensure efficient quality in health care: minimum variance optimal risk adjustment"

Transcription

1 Journal of Publc Economcs 84 (2002) locate/ econbase Settng health plan premums to ensure effcent qualty n health care: mnmum varance optmal rsk adustment Jacob Glazer *, Thomas G. McGure a, b a Tel Avv Unversty, Faculty of Management, Tel Avv, Israel b Harvard Medcal School, Boston, USA Abstract Rsk adustment refers to the practce of payng health plans a premum per person (or per famly) based on a formula usng rsk adusters, such as age or gender, and weghts on those adusters. One role of rsk adustment s to make sure plans have an ncentve to accept all potental enrollees. Another role, at least as mportant n our vew, s to lead health plans to choose the effcent level of qualty of care for the varous servces they offer. Most of the research and polcy lterature on rsk adustment focuses on the frst problem. Ths paper proposes a new way to calculate weghts n a rsk adustment formula that contends wth both problems. For a gven set of adusters, we dentfy the weghts that mnmze the varance n plan predctable health care costs that are not explaned by rsk adustment (addressng the access problem), subect to the payments satsfyng condtons for an optmal rsk aduster (makng sure plans provde the effcent qualty). We call the formula mnmum varance optmal rsk adustment (MVORA) Elsever Scence B.V. All rghts reserved. 1. Introducton Managed health care plans nsurance and servce organzatons responsble for provdng medcally necessary health care enroll about 75 percent of the U.S. populaton and large shares of European populatons. These health plans receve *Correspondng author. E-mal address: glazer@post.tau.ac.l (J. Glazer) / 02/ $ see front matter 2002 Elsever Scence B.V. All rghts reserved. PII: S (01)

2 154 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) ndvdual or famly premums, pad by governments (e.g. Medcare or Medcad n the U.S.), or by a combnaton of employers and consumers n U.S. employmentbased health nsurance. One of the maor concerns wth the health nsurance/ managed care health care polcy s adverse selecton (Enthoven, 1993). Generally, plans may take actons to dscourage or encourage potental enrollees. For one thng, they may refuse some applcants, although overt actons to dscourage ndvduals are normally prohbted and may be readly montored. More troublesome s that plans may dstort the mx of the qualty of health care they offer to dscourage hgh-cost persons from onng the plan. As a number of papers have observed, decsons about what care s medcally necessary are fundamentally outsde the scope of drect regulaton (Mller and Luft, 1997; Newhouse, 1996). In ths paper, we consder how rsk adustment of premums pad to health plans can 1 address what we refer to as the ndvdual access problem and the qualty problem. Rsk adustment refers to the practce of payng health plans a premum for each person or famly based on a formula usng rsk aduster varables, such as age, gender, or ndcators of pror health care use, and weghts on those adusters. Rsk adustment researchers have sought to fnd the formula that does the best ob of fttng the dstrbuton of health care costs n a populaton. Rsk aduster varables are chosen that are lkely to be related to costs, and weghts on those varables are chosen by regresson technques to mnmze the sum of the squared resduals, a 2 practce we wll refer to here as conventonal rsk adustment. The ntutve appeal of regresson coeffcents as weghts on rsk adusters derves from a desre to address the ndvdual access problem. Matchng payments wth ndvdual costs as closely as possble may dscourage plans from denyng membershp to some and aggressvely recrutng others. When the problem beng addressed s plans dstorton of the qualty of servces 1 A comprehensve dscusson of health nsurance premums and rsk adustment would nclude analyss of the settng of the premums charged to enrollees. Except n some ndvdual nsurance markets, ndvduals receve some subsdy of ths premum. In the case of government programs n the U.S. and some European countres, ths subsdy s complete and the person or famly pays nothng. In the employment-related health nsurance context n the U.S., the employer pays a porton of the premum. Authors have consdered how to set the consumer contrbuton n order to gve consumers the ncentves to on the rght plans. Ths ssue can also be consdered wthn an adverse selecton framework and may nvolve rsk adustng consumer premum contrbutons. The most mmedate effect of the person-pad component of premums s to nfluence the consumer s choce of plan. In ths paper, as wll be made clear n the next secton, we wll analyze a stuaton n whch the consumer s contrbuton s zero. Ths s emprcally accurate for many people n the U.S., ncludng a large share of the employed populaton, as well as the entre populaton enrollng n managed health care n Medcare and Medcad. It s also common n Europe. 2 Age and gender are commonly used n regresson formulae, but alone explan only about 1% of the varance n actual costs. Research on mprovng the explanatory power of these regressons has focussed on usng sophstcated clncal algorthms to defne new varables based on the dagnoses from pror health care use. Researchers have been able to run regressons explanng up to 10% of the varance n actual costs, but rsk adustment systems presently n use explan much less. See van de Ven and Ells (2000) for a revew of the emprcal lterature. Keenan et al. (2001) descrbes rsk adustment practces by maor publc and prvate payers n the U.S.

3 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) to affect the decsons of groups of potental applcants, however, the regressonbased approach lacks ntutve foundaton. Fgurng rsk adustment weghts to mtgate qualty dstortons requres a concepton of how health plans set the qualty of servces, and a concepton of what optmal qualty means n ths context. In ths paper we propose an approach to dervng weghts on rsk aduster varables that begns wth an explct statement of health plan behavor n a market wth heterogenety n the demand for health care servces, and an explct statement of the economc effcency problem. Wth two effcency problems n vew, ndvdual access and qualty dstortons, and one polcy nstrument, the formula for the premum pad to plans, t s natural to expect there to be some trade-off n meetng the two competng obectves. Although t has nspred most emprcal rsk adustment research, ndvdual access to health plans s not the maor socal effcency problem n the health plan market. In the U.S., Europe, Israel and Latn Amerca, governments and employers requre contractng plans to offer perodc open enrollments. Durng open enrollment perods (once a year for most U.S. employers, and every month n Medcare), plans must accept any applcant. It seems clear that open enrollment regulaton works well to ensure access. Smulaton research demonstrates that exstng rsk adustment systems leave some ndvduals as bg losers or wnners (Chapman, 1997; Shen and Ells, 2000), but there s no evdence that plans act to deny membershp to ndvduals. By way of contrast, qualty of care n managed care health plans s the maor polcy focus n the U.S. and elsewhere. We beleve that the qualty dstorton problem s more mportant and less amenable to other regulatory solutons than the ndvdual access problem, and s therefore the proper prmary target of rsk adustment polcy. Therefore, n the analyss below, when t comes to a tradeoff between polcy obectves, we put more weght on mantanng qualty than on ndvdual access. Plans ncentves to overprovde the qualty of some servces and underprovde the qualty of others derves from the dstrbuton of health care demands n the populaton. Wth data about the dstrbuton of those demands, the regulator can antcpate plans ncentves, and mpose a system of correctve taxes and subsdes at the person level, usng the many observable varables avalable for each ndvdual. Wthn ths framework we derve a smple rule that characterzes optmal rsk adustment: the relaton between the covarance between spendng on a servce and the rsk adusted premum and the covarance between spendng on that servce and the sum of spendng on all servces must be the same for all servces that the plan provdes. In general, there wll be many combnatons of weghts on rsk aduster varables that satsfy the rule for optmal rsk adustment. By choosng from among the set of optmal rsk adustment weghts the combnaton that leads to the best ft of premums to costs, we ntegrate our approach wth tradtonal regresson-based methods, n effect desgnng a rsk adustment formula that contends wth the two forms of adverse selecton problems dentfed above. We wll refer to our proposed method for rsk adustment as mnmum varance optmal rsk adustment (MVORA), optmal

4 156 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) because t solves the qualty dstorton problem, and mnmum varance because t does so n a way that mnmzes the sum of the squares of the devatons between premums and costs at the ndvdual level. In lne wth the lterature on rsk adustment, we argue that ths statstcal crteron s motvated by a concern for ndvdual access. Our paper s related to several lnes of research n health and publc economcs. Many papers n health economcs are concerned wth effcency problems due to selecton n nsurance markets (see Cutler and Zeckhauser, 2000; Encnosa, 1999 and van de Ven and Ells, 2000 for revews). Researchers have long been aware that one sgnfcant adverse selecton-related problem assocated wth competton among health plans s servce competton to attract the good rsks/deter the bad (Newhouse, 1996). Lttle research has been done, however, on the mplcatons of ths concern for a rsk adustment formula. So far, rsk adustment of ths type has only been characterzed n smple cases. Glazer and McGure (2000) consder the two-servce case and one rsk aduster, and show that the best rsk adustment weghts are the soluton to a par of equatons. The weghts obtaned by ths procedure are generally dfferent from those obtaned from regresson coeffcents. From the perspectve of publc economcs, t s natural to vew rsk adustment as queston of optmal taxes and subsdes on the prces pad for health nsurance, a framework adopted n several papers (Glazer and McGure, 2000; Neudeck and Podczeck, 1996; Selden, 1999). By explct characterzaton of the dstortons emergng n markets for health plans, we are able to dentfy, Pgouvan fashon, the set of taxes and subsdes necessary to algn prvate ncentves to maxmze proft wth the socal obectve of producton of the effcent qualty of health care. In the context of rsk adustment, the taxes and subsdes to correct ncentves work n a unusual fashon. Normally, a tax or subsdy apples drectly to the actvty ntended to be affected: for example, a tax on the volume of polluton s ntended to reduce polluton. Obvously though, there are other ways to tax/ subsdze to ht a partcular polluton target. If the actual level of polluton were not verfable, for example, a regulator famlar wth the effect of polluton (on, say water qualty) could put a tax on ths effect n order to manpulate the frm nto choosng the target level of polluton. The dea n ths paper s the same. Qualty provded by health plans s usually not verfable and, hence, cannot be taxed or subsdzed drectly. However, the regulator can magnfy or dmnsh the revenue consequences to the health plan s qualty choces by choce of the rsk adustment formula. If older people value and on a plan n response to good care for cardac problems and the regulator s concerned that the qualty offered for these dseases 3 s too low, the regulator can nduce hgher qualty by subsdzng older people. 3 The health payment lterature has used ths nsght prevously, as n Rogerson (1994) where the optmal hosptal per dscharge payment was fgured n order to nduce the desred non-verfable qualty. In ths case, qualty s one-dmensonal, and the level of the payment was the only nstrument necessary. Another early paper along the same lnes s Ma (1994).

5 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) A central nsght of ths paper s that there s an nstruments/targets feature of rsk adustment polcy. A plan sets qualty for many servces. The regulator has many varables n a rsk adustment formula. When the regulator has more nstruments than t needs, the qualty problem can be addressed. The fnal choce of rsk adustment formula can then be made n order to contend also wth ndvdual access. The paper s organzed as follows: Secton 2 presents our basc model of health plan behavor n an envronment wth no uncertanty and symmetrc nformaton. In Secton 3 we present the condtons for optmal rsk adustment and n Secton 4 we solve for the Mnmum Varance Optmal Rsk Adustment formula. An example of how to fgure MVORA s contaned n Secton 5. Secton 6 extends our analyss to the case of uncertanty and asymmetrc nformaton. 2. Health plan behavor Assume there are N ndvduals. Each one of them s about to choose a health plan. In ths secton, we wll analyze the behavor of one (representatve) health plan, takng the behavor of the others as gven. In Secton 3 we wll analyze ths behavor wthn a symmetrc equlbrum. The model presented n ths secton s based on that n Frank et al. (2000), hereafter referred to as FGM. The health plan s pad a premum (possbly rsk-adusted) for each ndvdual that enrolls. Indvduals dffer n ther need/demand for health care, and choose a plan whch maxmzes ther expected utlty. Health care s not a sngle commodty, but a set of servces maternty, mental health, emergency care, cardac care, and so on. A health plan chooses a ratonng or allocaton rule for each servce. The plan s choce of rules wll affect whch ndvduals fnd the plan attractve and wll therefore determne the plan s revenue and costs. We assume that the plan must accept every applcant, and we are nterested n characterzng the plan s ncentves to raton servces Utlty and plan choce The health plan offers S servces. Let ms denote the amount the plan wll spend on provdng servce s to ndvdual, f he ons the plan, and let: m 5 hm,m,...,m. The dollar value of the benefts ndvdual gets from a plan, 1 2 S u (m ), s composed of two parts, a valuaton of the servces an ndvdual gets from the plan, and a component of valuaton that s ndependent of servces. We assume these enter addtvely n utlty. Thus, where, u (m ) 5 v (m ) 1 m (1)

6 158 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) v (m ) 5O v (m ) s s s s the servce-related part of the valuaton and s tself composed of the sum of the ndvdual s valuatons of all servces offered by the plan. v s(? ) s the ndvdual s valuaton of spendng on servce s, also measured n dollars, where v 9s. 0, v 99 s, 0. Assume for the moment that the ndvdual knows v (m ) wth certanty. (Ths assumpton wll be relaxed n Secton 6). The non-servce component s m, an ndvdual-specfc factor (e.g. dstance or convenence) affectng ndvdual s valuaton, known to person. From the pont of vew of the plan, m s unknown, but s drawn from a dstrbuton F (m ). We assume that the premum the plan receves has been predetermned and s not part of the strategy the plan uses to nfluence selecton. The plan wll be chosen by ndvdual f u. u, where u s the valuaton the ndvdual places on the next preferred plan. We analyze the behavor of a plan whch regards the behavor of all other plans as gven, so that u can be regarded as fxed. Gven m and u, ndvdual chooses the plan f: m. u 2 v (m ). For now, we assume that, for each, the plan has exactly the same nformaton as ndvdual regardng the ndvdual s servce-related valuaton of ts servces, v, and regardng the utlty from the next preferred plan, u. For each ndvdual, the plan does not know the true value of m but t knows the dstrbuton from whch t s drawn. Therefore, for a gven m and u, the probablty that ndvdual chooses 4 the plan, from the pont of vew of the plan, s: n (m ) F (u 2 v (m )) (2) 2.2. Managed care Managed care ratons the amount of health care a patent receves. Followng Keeler et al. (1998) and FGM, let qs be the servce-specfc shadow prce the plan sets determnng access to care for servce s. A patent wth a beneft functon for servce s of v (? ) wll receve a quantty of servces, m determned by: s v 9 (m ) 5 q s s s Let the amount of spendng determned by the equaton above be denoted by m s(q s). Note that Eq. (3) s smply a demand functon, relatng the quantty of servces to the (shadow) prce n a managed care plan. s (3) 4 An alternatve nterpretaton s that ndex descrbes a group of people wth the same v (m ) functon and n (m ) s then the share of ths group that ons the plan.

7 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) Proft and proft maxmzaton Let q 5 hq,q,...,q be a vector of shadow prces the plan chooses and 1 2 S m (q) 5 hm (q ),m (q ),...,m (q ) be the vector of spendng ndvdual gets S S by onng the plan. Defne n (q) ; n (m (q)). Expected proft, p(q), to the plan wll depend on the ndvduals the plan expects to be members, the revenue the plan gets for enrollng these people, and the costs of each member. p(q) 5O n (q)[r 2O m (q )] (4) s s s where r s the rsk-adusted revenue the plan receves for ndvdual. The plan wll choose a vector of shadow prces to maxmze expected proft, Eq. (4). Defnng M5 os m s(q s) to be total spendng on a person, proft per person s p5 r2 M. Assumng that ndvduals share the same elastcty of demand for 5 any servce (allowng common elastctes to dffer across servces), the proft maxmzng condton for q s, s 5 1,2,...,S, becomes (see FGM): O nm s qs 5 ]]]]. (5) O F 9m sp 2.4. The effcency crteron The health plan allocates resources effcently when t acts so as to equalze the degree of ratonng across all S servce areas. Ths s (second-best) effcent n the sense that the value of a dollar of health care spendng s equalzed across all possble uses. Formally, ths requres qs 5 q, for all s, where q s some constant characterzng the overall degree of ratonng. When q 5 1, a frst-best effcency s acheved. When q. 1, second-best effcency s acheved but the overall budget for health s too small, and when q, 1, t s too bg. There s, however, no partcular reason to expect proft maxmzaton represented by Eq. (5) to lead to the same shadow prce for each servce s, unless the rsk adustment system s able to equalze the relatve ncentves to supply each servce. 3. Optmal rsk adustment In the analyss of plan behavor ust dscussed, the rsk adustment formula was taken as gven. We now consder how to structure the rsk adustment formula to 5 Note that the fact that all ndvduals share the same elastcty of demand for a certan servce does not mply that ther demand curves are dentcal.

8 160 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) nduce the plan to provde servces at the (second-best) effcent qualty. A rsk adustment formula that acheves ths goal wll be referred to as optmal rsk adustment. The analyss carred out here s very much n the sprt of the standard prncpal-agent lterature. In that lterature, the prncpal constructs an ncentve scheme so that the agent, actng n ts own nterest, wll behave accordng to the prncpal s goals. In what follows the regulator (the prncpal) sets a rsk adustment formula (an ncentve scheme) to nduce the proft maxmzng plan (the agent) to provde the effcent qualty (the regulator s goal). We have n mnd a stuaton where there are several dentcal health plans. Thus, Eq. (5) above descrbes the proft maxmzng behavor of each of the health plans n equlbrum (takng the behavor of the other plans as gven). In a symmetrc equlbrum, each ndvdual has the same probablty of beng n the plan, so n5 n. Assume that F s unform and the same for all. Then, we can smplfy Eq. (5) to say that second-best effcency requres that for each s, Eq. (6) holds. O m s ]]]]]] 5 l, (6) O msr 2O msm for some l. 0. Eq. (6) can be rewrtten as cov(m s,r) 2 cov(m s,m) 1 ]]]]]]] 5 (M 2 r 1 ]) (7) m l where covarances are defned as wth s 1 cov(m,r) 5 ] O (m 2 m )(r 2 r ) and s N s s 1 cov(m,m) 5 ] O (m 2 m )(M 2 M ), s N s s O m s ; m s, M ; O M, and r ; O r ] ] ] N N N Eq. (7) embodes the man result of ths paper. To equalze ncentves to raton all servces, the covarance of the rsk aduster wth the use of every servce must track the covarance of total predcted costs wth the servce s use. Intutvely, the optmal rsk adustment formula must have the property that by spendng on a

9 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) servce, the cost consequences to a plan (from servng the demands of new members) relates to the revenue consequences (from the premums they brng n) n the same way across all servces. When ths set of condtons holds, the plan has an ncentve to raton all servces wth the same strngency. It s worthwhle to contrast the condtons represented by Eq. (7) for the optmal way to set premums to the usual approach based on statstcal ft. Conventonal rsk adustment seeks to brng per person premums as close as possble to per person costs. The frst contrast to be made s that we have S condtons characterzng optmal rsk adustment rather than one crteron based on ft. If a plan chooses the qualty of S servces, t s necessary to address the condtons for effcency for each servce, so more than one crteron seems evdently necessary. Second, the statstcal crteron of ft of premums, r, to costs, M, s replaced by crtera represented by covarances that lead proft maxmzaton decsons of health plans to be dentcal wth condtons for effcency. When the extra revenue brought n by spendng on servce s (captured by the covarance between ms and r) s n the same relaton to the extra total costs ncurred by spendng on servce s (captured by the covarance between ms and M) for all servces, proft-max- mzaton s effcent. We are now ready to consder how to use a rsk adustment formula to set premums r n order to satsfy condtons for optmal rsk adustment. Suppose that a set of J rsk adusters (varables) s avalable to use n a rsk adustment formula. Call ths set X 5 (X 1,...,X J). The varable x represents the value of the aduster for person (x could be the age or gender or health status of ndvdual ). r s the rsk-adusted premum payment made on behalf of person. A rsk adusted premum can be wrtten as r 5 X b (8) where b s a vector of weghts on the rsk aduster varables. As can be seen from Eq. (8), for a gven set of rsk aduster varables, the ssue of desgnng a rsk adustment formula s a matter of settng the rght weghts b. Rather than usng a regresson, we fnd the b s as solutons to optmalty condtons. Usng Eq. (8), each of the S equatons n Eq. (7) can be wrtten as O bcov(m s,x ) 2 cov(m s,m) 1 ]]]]]]]] 5 M 2 r 1 ] (9) m l s Note that Eq. (9) s lnear n the coeffcents of the rsk adusters b. The RHS of Eq. (9) s the same for all S equatons. It s nterestng to note what Eq. (9) mples for the relatonshp between the

10 162 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) covarance of the x s wth expected costs on a servce and the b s. If the cov(m s,x) 6 s small, n order to equalze ths to a target covarance, the b s must be larger. Solvng for M 2 r 1 1/l, we are left wth S 2 1 lnear restrctons on the b s. Each of these restrctons takes the form of Eq. (10). In addton, a gven budget b n the form of an average rsk adusted premum yelds another lnear restrcton. We defne optmal rsk adustment to be the weghts b 1,...,bJ that satsfy the set of S 2 1 equatons (10), and the budget constrant n Eq. (11). We label any set of o b s satsfyng these s condtons b. o o O b cov(m s,x ) 2 cov(m s,m) O b cov(m S,x ) 2 cov(m S,M) ]]]]]]]]] 5 ]]]]]]]]] m m (10) s S s 5 1,...,S 2 1 OO b x ;O b x 5 b (11) 1 o o ] N The system of Eqs. (10) and (11) wll have at least one soluton f a rank condton s satsfed, here, f J. S. We have assumed that the plan makes S ndependent ratonng decsons, and there are J rsk adusters, none of whch s a lnear combnaton of the others. It s unclear what should be sad about the magntude of S, the number of ratonng decsons made by a plan. On the one hand, a plan offers a very large number of dfferent servces. But on the other hand, from the standpont of what management may practcally pay attenton to, there may be many fewer decson varables. For purposes of management, servces mght be grouped n ust a few categores: ratonng may be done for example at the level of prmary care, pedatrcs, mental health, cardac care, and so on, leadng to perhaps groups. (There are 19 Maor Dagnostc Categores, MDCs, for nstance, that underle the DRG system). As for the number of rsk adusters, J, ths number could be very large. Consder age and gender alone. Whle only two varables there are many potental degrees of freedom n age and gender. Medcare explots these by creatng age-gender cells. In a populaton ranged n age from say 1 65, a payer mght feasbly create ten age 3 two gender cells for 20 mutually exclusve varables to use for rsk adustment. Addng zone of resdence and elgblty status (among workng people: employee, spouse, other dependent), geometrcally enlarges J. Furthermore, dagnoses from pror health care use have been grouped nto scores of categores n recently developed dagnoss-based classfcaton systems (Ells et al., 1996; Wener et al., 1996). It 6 Ths s a generalzaton of a fndng of Glazer and McGure (2000), where t was shown that wth one sgnal (rsk aduster varable), as the nformatveness of the sgnal deterorates, the weght on the sgnal ncreases n the optmal rsk adustment formula.

11 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) thus seems qute safe to proceed on the assumpton that the number of rsk aduster varables, J, exceeds the number of decsons made by management, S. 4. Mnmum varance optmal rsk adustment (MVORA) For a gven per person budget b and a set X of rsk adusters, Mnmum Varance Optmal Rsk Adustment (MVORA) s a vector of rsk adustment weghts b * 5 ( b *,...,b *) that solves the followng (constraned mnmzaton) problem: 1 J 1 2 b,...,b ] 1 J N Mnmze O (M 2O b x ) (12) F cov(m s,x ) cov(m S,x ) cov(m s,m) cov(m S,M) s.t. O b ]]] 2]]] 5]]] 2]]] m m m m G s S s S s 5 1,...,S 2 1 (13) and O b x 5 b (14) The set of S 2 1 equatons n Eq. (13) s a rewrtng of the S 2 1 equatons n Eq. (10). MVORA contends wth both problems caused by adverse selecton. The constrants guarantee the rsk adustment formula s optmal, n the sense that no qualty dstorton wll take place. Mnmzng the sum of the square of the devatons addresses problems that may arse when payments devate from costs at the ndvdual level. The lterature on conventonal rsk adustment does not contan a formal statement of why conventonal rsk adustment s the soluton to an adverse selecton problem n the market for health plans. No one, so far as we know, has suppled the argument for why the economc loss assocated wth a devance between the rsk adusted payment and the predcted cost should go up accordng to the square of the dfference. In ths secton, we follow the exstng rsk adustment lterature and smply presume that a better statstcal ft of predcted to 2 actual costs, as measured by an R statstc, advances the cause of dealng wth the selecton problem related to ndvdual access. Before we go on to dscuss the soluton for MVORA, t may be worthwhle to present conventonal rsk adustment n the context of our model. Conventonal rsk adustment derves weghts from the soluton to an ordnary least squares

12 164 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) regresson of M on X. Thus, conventonal rsk adustment s, n fact, the soluton to Eq. (12) wthout the constrants n Eq. (13) and s gven by: c 21 b 5 (X9X) X9M The conventonal rsk adustment formula s therefore, c r 5 X b c We can express the set of S lnear constrants n Eq. (13) n matrx form as Ab 5 C, where A s an S 3 J matrx wth elements cov(m s,x ) cov(m S,x ) ]]] 2 ]]] s 5 1,...,S 2 1 a 5 ms m s S 5x s5 S b s a J 3 1 vector of weghts on the aduster varables and C s an S 3 1 vector of constants wth element cov(m s,m) cov(m S,M) ]]] 2 ]]] s 5 1,...,S 2 1 Cs 5 ms ms 5 b s5 S Note that the soluton for MVORA descrbed above s the estmated coeffcents from a regresson of M on X (.e. as n conventonal rsk adustment), constraned by the S lnear restrctons n Eqs. (13) and (14) (Thel, 1961). Thus, we can wrte MVORA as c c b* 5 b 1 (X9X) A9[A(X9X) A9] (C 2 Ab ) (15) One can see that f the constrants n Eq. (13) are bndng, MVORA wll be dfferent than conventonal rsk adusters. One specal case n whch the constrants do not bnd s when conventonal rsk adustment s perfect n the sense of completely capturng the varance n health care costs. In ths case, o bx5 M, and condtons (13) are exactly satsfed. Except n ths extreme case, the constrants (13) wll be bndng and MVORA wll dffer from conventonal. We can also observe that addng a new rsk aduster varable x wll mprove the 7 A smple regresson of M on X gnores the dstrbutonal characterstcs of M. For many observatons, M 5 0 (no health care spendng). Furthermore, among the postve observatons, M s skewed to the rght. Most of the emprcal lterature studyng the effect of economc factors on health care use has employed a two-part model orgnally developed for the RAND Health Insurance Experment (Duan et al., 1983). In the frst part, a use no use equaton s estmated wth a logt or a probt regresson. In the second part, an OLS regresson s performed on a transformaton (square root, log) of the dependent varable for the observatons wth postve spendng. See Mannng (1998) and Mullahy (1998) for recent dscusson. The rsk adustment lterature reles prmarly on one-part lnear regressons. The man ratonale appears to be a desre for smplcty n the rsk adustment formula.

13 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) regulator s ablty to mnmze Eq. (12). Addng a new rsk aduster wth at least some correlaton wth some element of cost gves the regulator another nstrument and must do at least as well n mnmzng Eq. (12). As the presentaton of MVORA makes clear, the rsk adustment formula we propose depends on plan proft maxmzaton. Strctly speakng, MVORA s the mnmum varance rsk adustment formula that sustans the effcent expendture decsons by plans as proft maxmzaton. The queston arses as to how to fgure MVORA when expendtures are smply data from some pattern of servce expendtures generated by some other process, for example a fee-for-servce system n whch there s no health plan pad by captaton, and n whch proft 8 maxmzaton by a health plan plays no role. Any change n the payment system, for example, ntroducton of a rsk-adusted captated system, wll change ncentves and change the pattern of servces. A MVORA system can be calculated wth frst and second moment nformaton from the data from the FFS system, but then ntroducton of the payment system would change the use patterns. It seems lkely that a recalculaton of the MVORA formula wth the shft n use patterns, and perhaps a further recalculaton, etc., would lead towards optmal servce patterns, but such a queston of the dynamc adustments to rsk adustment s not formally confronted here. We are makng only a weaker clam, that f the health plan s provdng the effcent level of servces whle proft maxmzng, MVORA wll keep t there. (The conventonal rsk adustment scheme would move the plan away from optmalty). 5. An example: conventonal rsk adustment and MVORA Lke conventonal rsk adustment, MVORA can be found by usng regresson technques n data. In ths secton we calculate conventonal rsk adustment and MVORA n a very smple emprcal example, to llustrate how MVORA works, and to show how t dffers from conventonal n the way premums are fgured. Suppose data on per person spendng comes from the followng smple envronment. There are two servces a and d. The populaton s dvded nto three age groups, age 1, age 2 and age 3. The proporton of the populaton n age, sa. Age s the only avalable rsk aduster. For each ndvdual, x 5 1 f the ndvdual s n age, and x 5 0 otherwse. If an ndvdual needs servce a, the dollar value of servces he wll receve s m, regardless of hs age group, and f an ndvdual needs servce d the dollar a value of servces he wll receve s m, regardless of hs age group. The proporton d of ndvduals who need servce a s the same n all three age groups and 8 Medcare s rsk adustment system to pay HMOs s, for example, calbrated on data from the FFS sector.

14 166 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) normalzed to 1 and the proporton of ndvduals who need servce d n age s g, 5 1,2,3. In fgurng MVORA, the nformaton assumpton we make n ths example s the followng. There are two types of people, the healthy and the sck, dfferentated by ther use of servce d. The healthy do not need d at all, the sck always need d. We assume that people know ther type wth certanty. That s, healthy people can accurately forecast that they wll use m of servce a and none of servce d, and the sck type accurately forecast that they wll use ma and md of servce d. So, g s the share of the sck types n age group. a of servce a The age categores wll be correlated wth type so long as the g dffers for each age category. Age wll thus have value as a rsk adustor. Conventonal rsk adustment wll be unable to fully solve the qualty problem snce under conventonal rsk adustment prvate nformaton remans about type and a health plan would want to underprovde servce d and overprovde servce a to attempt to attract the healthy wthn each age category. MVORA fxes ths problem. The calculatons below are smplfed because the age categores are (0,1) and the antcpated use of md by the healthy type s zero. Ths allows us to avod explct summatons over types n the covarance terms below. c Let r denote the conventonally rsk adusted premum pad for an ndvdual n group, then: c a d r 5 m 1 g m (16) Conventonal rsk adustment smply pays for each ndvdual the average cost of an ndvdual n the age group the person belongs to. Movng to optmal rsk adustment, we need to defne some covarances for ths example. One should frst notce that snce all ndvduals n all age groups use the same level of servce a, and covsm a,xd5 0, for 5 1,2,3 (17) covsm,md5 0, (18) a where M denotes the total dollar value of servces the ndvdual receves. As for servce d, under our assumptons: where: covsm d,xd5 asmd2 mdd for 5 1,2,3 (19) m 5 g m d d (20) s the average dollar value of servces d used by an ndvdual n age, and

15 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) m 5O a m d d 51 s the populaton average use of servce d. Furthermore, snce m s the same for all ndvduals, we get that: a (21) covsm,md5 varsm d (22) d d In order to calculate the optmally rsk adusted premum we assume that the plan breaks even namely, b 5 m 1 m (23) a d where b s the budget per person. Note that under the conventonally rsk adusted premum above the plan also breaks even, so the comparson between the two payments schemes wll be on an equal bass. Pluggng Eqs. (17) (23) nto Eqs. (10) and (11) we obtan that optmal rsk o o o adustment s a trple sb,b,b d that solves the followng two equatons: and Ob a(m 2 m ) 2Var m d5 0 (24) 51 o d d s d 3 o O ba5 ma1 m d (25) 51 The MVORA s the trple sb *,b * d that solves Mn O agm f s a1 md2 bd 1s1 2 gdsma2 b d g (26) b 1,b 2,b351 s.t. Eqs. (24) and (25). Table 1 presents a specal case of the example above. The frst column n Table 1 specfes the three age groups, the second specfes Table 1 An example of Conventonal and Mnmum Varance Optmal Rsk Adustment c Age a g ma md r r* group

16 168 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) a, the proporton of each group n the populaton, the thrd column specfes g, sckness d probablty, the fourth column specfes m a, the dollar value of servce a, the ffth column specfes m d, the dollar value of servce d. The last two columns c compare r, the conventonally rsk adustment premum, wth r*, the MVORA premum, for each age group. As can be seen from the table, the conventonal and MVORA premums are qute dfferent. For the relatvely healthy age group 1 (the young ) MVORA pays much less than conventonal rsk adustment whereas for the other two age groups t pays much more. As s well-known, snce age s an mperfect ndcator of type, a conventonally rsk adusted premum wll leave a health plan wth some ncentves to attract the healthy types (those that would use only servce a) wthn each age group. The conventonal premums would therefore nduce a plan to oversupply treatment for servce a and undersupply treatment for servce d. MVORA premums counteract ths ncentve. By undersupplyng servce d, the plan would tend to dscourage relatvely more of the age 3 people. By payng more for the age 3 group, MVORA gves the plan a strong ncentve to keep them. Indeed, MVORA fgures the rsk-adusted weghts so that the ncentve to supply both servce a and d s ust balanced to effcently allocate a fxed budget. The marked dsparty between the conventonal rsk adustment and MVORA s a consequence of the extreme assumpton we make about prvate nformaton. If ndvduals could not perfectly forecast ther health care use, MVORA would be closer to conventonal. One can also return to Eq. (5) above to examne shadow prces f plans were pad by conventonal rsk adustment. In our extreme example, the stuaton would be dsastrous for servce d. Snce everyone who has postve spendng on servce d s a loser (negatve p), the denomnator of Eq. (5) s so small t s negatve! Essentally, a plan has an ncentve wth conventonal rsk adustment to reduce servce d to as low a level as possble. 6. Uncertanty and asymmetrc nformaton The analyss so far has assumed that both the plan and the ndvdual know wth certanty the spendng n dollars on each of the health care servces that wll be used by the ndvdual, m s. We shall now extend our model to the more realstc case where both the ndvdual and the plan only hold some belefs about the future health care needs of the ndvdual, and these belefs may not the same. We wll see that the two crucal elements n the analyss here are the plan s belefs about each ndvdual s expected use of each servce s, whch we denote by m s, and the plan s belefs about ndvdual s belefs of how much of each of the servces he s expected to use, denoted by m ˆ s. In Appendx A, we analyze plan s behavor n ths case and show that the proft maxmzng shadow prces are gven by:

17 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) O nm s qs 5 ]]]]]]. (59) O F 9 m ˆ (r 2 M s ) A plan s expected proft for an ndvdual s the product of the probablty the person ons the plan tmes the proft or loss once he s a member. The probablty that he ons depends on what the person beleves he wll receve n terms of servces n the plan. The proft once he s enrolled depends on revenue and use n the plan. It s the plan s expectatons about these two elements that wll govern plan behavor n ratonng care. Note that wth respect to the decson to on a plan, the plan must form belefs about what the consumers expect. Ths s m. ˆ Once n the plan, and from the pont of vew of the plan, what consumers expect s no longer relevant what matters s what the plan expects a person to use. Ths s m. The placement of mˆ and m n Eq. (59) reflects these consderatons. See Appendx A for dervaton. Now MVORA can be redefned to be the soluton to the followng problem: 1 2 ] N b 1,...,bJ Mnmze O (M 2O b x ) (129) F cov(m ˆ,x ) cov(m ˆ,x ) cov(m ˆ,M ) cov(m ˆ,M s S s S ) s.t O b ]]] 2]]] 5]]] 2]]]] m m m m G s S s S s 5 1,...,S 2 1 (139) and O b x 5 b (149) The obectve functon n Eq. (129) s a modfcaton of Eq. (12) to take nto consderaton plan s belefs about ts expected loss (proft) on each ndvdual. Condton (129) s meant to capture the problem of ndvdual access. As such, t s plan s expectatons about what persons wll cost n the plan that govern plans decsons about access. We beleve t s reasonable to suppose that f M dffers from r, access problems may emerge. Therefore, mnmzaton of the sum of the squares of the devatons of r from plan s expectatons of costs, M, s a reasonable crteron. The constrant (139) comes from Eq. (59) n a way smlar to the way we have obtaned Eq. (13) from Eq. (5). The soluton for MVORA descrbed above s the estmated coeffcents from a regresson of M on X, constraned by the S lnear restrctons. Thus, we can wrte MVORA as b* 5 b 1 (X9X) A9[A(X9X) A9] (C 2 Ab ) (159) where, b 5 (X9X) 21 X9M and, A s an S 3 J matrx wth elements

18 170 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) cov(m s,x ) cov(m S,x ) ]]] 2 ]]] s 5 1,...,S 2 1 a 5 ms m s S 5x s5 S b s a J 3 1 vector of weghts on the aduster varables and C s an S 3 1 vector of constants wth element cov(m s,m) cov(m S,M) ]]] 2 ]]] s 5 1,...,S 2 1 Cs 5 ms ms 5 b s5 S In the case of uncertanty and asymmetrc nformaton the conventonal rsk adusters wll be dfferent than MVORA for two reasons. Frst, even f the c constrants (139) were not bndng, the soluton to MVORA wll be b and not b. Second, f the constrants are bndng, the soluton wll be gven by the constrants and wll not come from an unconstraned regresson. 7. Dscusson The qualty of health care offered by competng managed health care plans s the foremost concern of health polcy n the U.S. and many other countres. Many elements of qualty cannot be controlled by regulaton, leavng open the opportunty for plans to manpulate the qualty they offer n an effort to acheve a proftable mx of enrollees. Another polcy concern s that plans may take actons to dscrmnate aganst partcular ndvduals. The method of rsk adustment proposed here contends wth both the qualty and the access problem n an emprcally mplementable way. Condtons for optmal rsk adustment can be derved from the means and covarances among rsk aduster varables and elements of health care spendng. A statstcal ft crteron can then readly be appled to select among the weghts satsfyng condtons for effcent qualty. Applcaton of our methodology requres that the dstrbuton of health care spendng n a populaton be avalable at the level of the servce, not ust n total. The concept of a servce represents the level of aggregaton at whch management of a health plan makes decsons about resource allocaton. In most health care clams data sets, nformaton on the dagnoss, locaton of servces (offce, hosptal, etc.), procedure conducted, and specalty or type of provder s typcally ncluded on the clam. Informaton s certanly avalable to classfy health care encounters nto the servces called for n our paper; the ssue s, however, how all ths data should be used to do so. Ths requres new work on classfcaton of health care use from the pont of vew of resource management. Theoretcal work can consder the mplcatons of usng too fne or too coarse methods of aggregaton for purposes of dervng condtons for optmal rsk adustment. Our method also calls for nformaton about expected health care costs,

19 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) specfcally, plans expectatons about the consumers costs, and plans belefs about consumers expectatons. Whle ths may seem a dsmayng prospect, n ths respect we are no worse off than conventonal rsk adustment. Careful dscussons about the goals of conventonal rsk adustment have recognzed for some tme that the obectve of conventonal rsk adustment s to explan predctable costs, yet t s actual costs that we see n the regressons dervng recommended weghts. The same fall back of usng the dstrbuton of actual spendng as opposed to expected spendng s avalable to our method as well. Another way to look at the ssue s as a topc for research. Health plans make the decsons about resource allocaton and access on the bass of expectatons. Regulators attemptng to counter the adverse consequences of these decsons would beneft by knowng how plans make decsons and based on what nformaton. The salency of plans expectatons to the polcy ssues related to qualty and access makes ths a very mportant topc for emprcal research. Acknowledgements Ths paper was wrtten whle McGure was at Boston Unversty. Research support from grants R01 MH59254 from the Natonal Insttute of Mental Health and P01 HS10803 from the Agency for Health Research and Qualty are gratefully acknowledged. We are grateful to Margarta Alegra, Ana Balsa, Zhun Cao, Davd Cutler, Randy Ells, Rchard Frank, Alberto Holly, Mathas Kfmann, Kevn Lang, Joseph Newhouse, Mark Satterthwate, Manuel Tratenberg, and Alan Zaslavsky for comments on an earler draft. Remanng errors are the authors responsblty. Appendx A. Proft maxmzaton wth uncertanty and asymmetrc nformaton Let T denote the set of possble health states of each ndvdual and let t denote an element n T. Let vt5 hv t1(m t1),vt2sm t2d,...,v ts(m ts) denote the vector of S valuaton functons for the S servces, f an ndvdual s health state s realzed to be t. We assume that for each t and s, v ts(? ) satsfes the propertes dscussed earler. Let xˇ be some random varable, the value of whch depends on the state t, and let k be a dstrbuton functon defned over T. Let E k[x] ˇ denote the expected value of xˇ wth respect to the dstrbuton k. The order of moves s as follows: at the frst stage, the plan chooses ts level of shadow prces q 5 (q 1,q 2,...,q S). When choosng q, the plan s uncertan about each ndvdual s health state t, and t holds pror belefs about t, denoted by (the dstrbuton over T ) g. At the second stage, ndvduals observe the plan s level of shadow prces and decde whether or not to on. At ths stage, the ndvdual s

20 172 J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) uncertan about hs true health state t and he holds some pror belefs about t. Let f be a dstrbuton functon over T whch denotes the plan s belefs about ndvdual s belefs about hs health state t. Thus, both g and f are plan s belefs, the frst s the plan s belefs about ndvdual s health state and the second s ts belefs about what the ndvdual thnks hs health state to be. At the thrd stage, when servces are provded, the ndvdual s true health state s already known. Hence, for a gven shadow prce qs and a valuaton functon v ts, the plan s expendtures on ths ndvdual n servce s wll be mtssq sd, gven by: v 9ts(m ts(q s)) 5 q s. Let v t(q) 5 os v ts(m ts(q s)). Let u t denote the ndvdual s utlty f hs health state s t and he chooses the alternatve plan. The plan s assgned probablty that the ndvdual wll on s gven by (we drop momentarly the subscrpt from the analyss): n (q) F(E [u 2 v ˇ ]) f f t t The plan s expected proft on the ndvdual s S F GD p (q) 5 n (q) r 2 E O m ˇ (q ). fg f g ts s s Dfferentatng wth respect to q s9 yelds dp fg(q) ]] 5 F 9Ef vˇ 9ts9 mˇ 9ts9 r2 Eg O m ˇ ts(q s)] 2 ne[m f g ˇ ts9 9 ] dq s9 f gs F GD Usng v9ts 5 qs and m9ts 5 (esm ts)/qs for all t, where es s the elastcty of demand (common to all ndvduals) for servce s (see FGM, 1999), the rght-hand sde becomes S nm f s9 e F 9m ˆ (r 2 M ) 2]] s9 s9 q s9 D where mˆ s9 5 E f[m ˇ ts9] s the plan s belef about the ndvduals predcton of the health resources he wll consume, m 5 E g[m ˇ ts9] s the plan s predcton of ts expendture on the ndvdual, and M 5 os m s. Wth a populaton of N ndvduals, the proft-maxmzng qs wll therefore be 9 (59) from text. s 9 Inducng a plan to supply servces requres some proft. In the case of a perfect ft between a rsk-adusted payment r and M, expected proft on each person would be zero. In ths case, some fxed costs are necessary (as n any model of monopolstc competton).

21 References J. Glazer, T.G. McGure / Journal of Publc Economcs 84 (2002) Chapman, J.D., Based Enrollment and Rsk Adustment for Health Plans, unpublshed PhD dssertaton, Harvard Unversty. Cutler, D., Zeckhauser, R., The anatomy of health nsurance. In: Culyer, A., Newhouse, J. (Eds.), Handbook of Health Economcs. North Holland. Duan, N. et al., A comparson of alternatve models for the demand for medcal care. Journal of Economc and Busness Statstcs 1, Ells, R.P., Pope, G., Iezzon, L. et al., Dagnoss-based rsk adustment for medcare captaton payments. Health Care Fnancng Revew 17 (3), Encnosa, W., The Economc Theory of Rsk Adustng Captaton Rates, Agency for Health Research and Qualty, unpublshed. Enthoven, A.C., The hstory and prncples of managed competton. Health Affars 12 (Supplement), Frank, R., Glazer, J., McGure, T., Measurng adverse selecton n managed health care. Journal of Health Economcs 19, Glazer, J., McGure, T.G., Optmal rsk adustment of health nsurance premums: an applcaton to managed care. Amercan Economc Revew 90 (4), Keeler, E., Carter, G., Newhouse, J., A model of the mpact of rembursement schemes health plan choce. Journal of Health Economcs 17 (3), Keenan, P., Beeuwkes, B.M., McGure, T., Newhouse, J., The prevalence of formal rsk adustment, Fall, Inqury, Harvard Unversty. Ma, C.-T.A., Health care payment systems: cost and qualty ncentves. Journal of Economcs and Management Strategy 3 (1), Mannng, W.G., The logged dependent varable heteroskedastcty and the retransformaton problem. Journal of Health Economcs 17 (3), Mller, R.H., Luft, H.S., Does managed care lead to better or worse qualty of care? Health Affars 16 (5), Mullahy, J., Much ado about two: reconsderng retransformaton and the two-part model n health econometrcs. Journal of Health Economcs 17 (3), Neudeck, W., Podczeck, K., Adverse selecton and regulaton n health nsurance markets. Journal of Health Economcs 15, Newhouse, J., Rembursng health plans and health provders: effcency n producton versus selecton. Journal of Economc Lterature 34, Rogerson, W., Choce of treatment ncentves by a nonproft hosptal under prospectve prcng. Journal of Economcs and Management Strategy 3 (1), Selden, T., Premum subsdes for health nsurance: excessve coverage versus adverse selecton. Journal of Health Economcs 18, Shen, Y., Ells, R.P., Cost Mnmzng Rsk Adustment, unpublshed, Boston Unversty. Thel, H., Economc Forecasts and Polcy, 2nd Edton. North-Holland. van de Ven, W.P.M.M., Ells, R.P., Rsk adustment n compettve health plan markets. In: Culyer, A.J., Newhouse, J.P. (Eds.), Handbook of Health Economcs. North-Holland. Wener, J., Dobson, A., Maxwell, S. et al., Rsk adusted captaton rates usng ambulatory and npatent dagnoses. Health Care Fnancng Revew 17 (3),

An Alternative Way to Measure Private Equity Performance

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

More information

Can Auto Liability Insurance Purchases Signal Risk Attitude?

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

More information

Section 5.4 Annuities, Present Value, and Amortization

Section 5.4 Annuities, Present Value, and Amortization 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

More information

How To Calculate The Accountng Perod Of Nequalty

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.

More information

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

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

More information

DEFINING %COMPLETE IN MICROSOFT PROJECT

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,

More information

The OC Curve of Attribute Acceptance Plans

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

More information

Analysis of Premium Liabilities for Australian Lines of Business

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

More information

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic

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

More information

1. Measuring association using correlation and regression

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

More information

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

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..

More information

THE METHOD OF LEAST SQUARES THE METHOD OF LEAST SQUARES

THE METHOD OF LEAST SQUARES THE METHOD OF LEAST SQUARES The goal: to measure (determne) an unknown quantty x (the value of a RV X) Realsaton: n results: y 1, y 2,..., y j,..., y n, (the measured values of Y 1, Y 2,..., Y j,..., Y n ) every result s encumbered

More information

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

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.

More information

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis

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 wangtngzhong2@sna.cn Abstract.

More information

Marginal Benefit Incidence Analysis Using a Single Cross-section of Data. Mohamed Ihsan Ajwad and Quentin Wodon 1. World Bank.

Marginal Benefit Incidence Analysis Using a Single Cross-section of Data. Mohamed Ihsan Ajwad and Quentin Wodon 1. World Bank. 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

More information

ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING

ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING 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,

More information

Addendum to: Importing Skill-Biased Technology

Addendum to: Importing Skill-Biased Technology Addendum to: Importng Skll-Based Technology Arel Bursten UCLA and NBER Javer Cravno UCLA August 202 Jonathan Vogel Columba and NBER Abstract Ths Addendum derves the results dscussed n secton 3.3 of our

More information

17 Capital tax competition

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.

More information

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek 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

More information

Chapter 11 Practice Problems Answers

Chapter 11 Practice Problems Answers Chapter 11 Practce Problems Answers 1. Would you be more wllng to lend to a frend f she put all of her lfe savngs nto her busness than you would f she had not done so? Why? Ths problem s ntended to make

More information

CHAPTER 14 MORE ABOUT REGRESSION

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

More information

Calculation of Sampling Weights

Calculation of Sampling Weights 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

More information

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence 1 st Internatonal Symposum on Imprecse Probabltes and Ther Applcatons, Ghent, Belgum, 29 June 2 July 1999 How Sets of Coherent Probabltes May Serve as Models for Degrees of Incoherence Mar J. Schervsh

More information

Efficient Project Portfolio as a tool for Enterprise Risk Management

Efficient Project Portfolio as a tool for Enterprise Risk Management Effcent Proect Portfolo as a tool for Enterprse Rsk Management Valentn O. Nkonov Ural State Techncal Unversty Growth Traectory Consultng Company January 5, 27 Effcent Proect Portfolo as a tool for Enterprse

More information

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12

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

More information

PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIGIOUS AFFILIATION AND PARTICIPATION

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

More information

1.1 The University may award Higher Doctorate degrees as specified from time-to-time in UPR AS11 1.

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

More information

Underwriting Risk. Glenn Meyers. Insurance Services Office, Inc.

Underwriting Risk. Glenn Meyers. Insurance Services Office, Inc. Underwrtng Rsk By Glenn Meyers Insurance Servces Offce, Inc. Abstract In a compettve nsurance market, nsurers have lmted nfluence on the premum charged for an nsurance contract. hey must decde whether

More information

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by 6 CHAPTER 8 COMPLEX VECTOR SPACES 5. Fnd the kernel of the lnear transformaton gven n Exercse 5. In Exercses 55 and 56, fnd the mage of v, for the ndcated composton, where and are gven by the followng

More information

Forecasting the Direction and Strength of Stock Market Movement

Forecasting the Direction and Strength of Stock Market Movement Forecastng the Drecton and Strength of Stock Market Movement Jngwe Chen Mng Chen Nan Ye cjngwe@stanford.edu mchen5@stanford.edu nanye@stanford.edu Abstract - Stock market s one of the most complcated systems

More information

How To Study The Nfluence Of Health Insurance On Swtchng

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

More information

Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting

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

More information

Using Series to Analyze Financial Situations: Present Value

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

More information

Support Vector Machines

Support Vector Machines Support Vector Machnes Max Wellng Department of Computer Scence Unversty of Toronto 10 Kng s College Road Toronto, M5S 3G5 Canada wellng@cs.toronto.edu Abstract Ths s a note to explan support vector machnes.

More information

Traffic-light a stress test for life insurance provisions

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

More information

Joe Pimbley, unpublished, 2005. Yield Curve Calculations

Joe Pimbley, unpublished, 2005. Yield Curve Calculations Joe Pmbley, unpublshed, 005. Yeld Curve Calculatons Background: Everythng s dscount factors Yeld curve calculatons nclude valuaton of forward rate agreements (FRAs), swaps, nterest rate optons, and forward

More information

Course outline. Financial Time Series Analysis. Overview. Data analysis. Predictive signal. Trading strategy

Course outline. Financial Time Series Analysis. Overview. Data analysis. Predictive signal. Trading strategy Fnancal Tme Seres Analyss Patrck McSharry patrck@mcsharry.net www.mcsharry.net Trnty Term 2014 Mathematcal Insttute Unversty of Oxford Course outlne 1. Data analyss, probablty, correlatons, vsualsaton

More information

14.74 Lecture 5: Health (2)

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,

More information

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College Feature selecton for ntruson detecton Slobodan Petrovć NISlab, Gjøvk Unversty College Contents The feature selecton problem Intruson detecton Traffc features relevant for IDS The CFS measure The mrmr measure

More information

Searching and Switching: Empirical estimates of consumer behaviour in regulated markets

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

More information

Problem Set 3. a) We are asked how people will react, if the interest rate i on bonds is negative.

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

More information

IDENTIFICATION AND CORRECTION OF A COMMON ERROR IN GENERAL ANNUITY CALCULATIONS

IDENTIFICATION AND CORRECTION OF A COMMON ERROR IN GENERAL ANNUITY CALCULATIONS IDENTIFICATION AND CORRECTION OF A COMMON ERROR IN GENERAL ANNUITY CALCULATIONS Chrs Deeley* Last revsed: September 22, 200 * Chrs Deeley s a Senor Lecturer n the School of Accountng, Charles Sturt Unversty,

More information

NBER WORKING PAPER SERIES CROWDING OUT AND CROWDING IN OF PRIVATE DONATIONS AND GOVERNMENT GRANTS. Garth Heutel

NBER WORKING PAPER SERIES CROWDING OUT AND CROWDING IN OF PRIVATE DONATIONS AND GOVERNMENT GRANTS. Garth Heutel 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

More information

Cahiers de la Chaire Santé

Cahiers de la Chaire Santé 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

More information

On the Optimal Control of a Cascade of Hydro-Electric Power Stations

On the Optimal Control of a Cascade of Hydro-Electric Power Stations On the Optmal Control of a Cascade of Hydro-Electrc Power Statons M.C.M. Guedes a, A.F. Rbero a, G.V. Smrnov b and S. Vlela c a Department of Mathematcs, School of Scences, Unversty of Porto, Portugal;

More information

The Greedy Method. Introduction. 0/1 Knapsack Problem

The Greedy Method. Introduction. 0/1 Knapsack Problem The Greedy Method Introducton We have completed data structures. We now are gong to look at algorthm desgn methods. Often we are lookng at optmzaton problems whose performance s exponental. For an optmzaton

More information

J. Parallel Distrib. Comput.

J. Parallel Distrib. Comput. J. Parallel Dstrb. Comput. 71 (2011) 62 76 Contents lsts avalable at ScenceDrect J. Parallel Dstrb. Comput. journal homepage: www.elsever.com/locate/jpdc Optmzng server placement n dstrbuted systems n

More information

Staff Paper. Farm Savings Accounts: Examining Income Variability, Eligibility, and Benefits. Brent Gloy, Eddy LaDue, and Charles Cuykendall

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

More information

CHAPTER 5 RELATIONSHIPS BETWEEN QUANTITATIVE VARIABLES

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

More information

Section 5.3 Annuities, Future Value, and Sinking Funds

Section 5.3 Annuities, Future Value, and Sinking Funds Secton 5.3 Annutes, Future Value, and Snkng Funds Ordnary Annutes A sequence of equal payments made at equal perods of tme s called an annuty. The tme between payments s the payment perod, and the tme

More information

Leveraged Firms, Patent Licensing, and Limited Liability

Leveraged Firms, Patent Licensing, and Limited Liability Leveraged Frms, Patent Lcensng, and Lmted Lablty Kuang-Cheng Andy Wang Socal Scence Dvson Center for General Educaton Chang Gung Unversty and Y-Je Wang Department of Economcs Natonal Dong Hwa Unversty

More information

Risk-based Fatigue Estimate of Deep Water Risers -- Course Project for EM388F: Fracture Mechanics, Spring 2008

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

More information

How to Sell Innovative Ideas: Property right, Information. Revelation and Contract Design

How to Sell Innovative Ideas: Property right, Information. Revelation and Contract Design Presenter Ye Zhang uke Economcs A yz137@duke.edu How to Sell Innovatve Ideas: Property rght, Informaton evelaton and Contract esgn ay 31 2011 Based on James Anton & ennes Yao s two papers 1. Expropraton

More information

L10: Linear discriminants analysis

L10: Linear discriminants analysis L0: Lnear dscrmnants analyss Lnear dscrmnant analyss, two classes Lnear dscrmnant analyss, C classes LDA vs. PCA Lmtatons of LDA Varants of LDA Other dmensonalty reducton methods CSCE 666 Pattern Analyss

More information

Analysis of Demand for Broadcastingng servces

Analysis of Demand for Broadcastingng servces 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

More information

Kiel Institute for World Economics Duesternbrooker Weg 120 24105 Kiel (Germany) Kiel Working Paper No. 1120

Kiel Institute for World Economics Duesternbrooker Weg 120 24105 Kiel (Germany) Kiel Working Paper No. 1120 Kel Insttute for World Economcs Duesternbrooker Weg 45 Kel (Germany) Kel Workng Paper No. Path Dependences n enture Captal Markets by Andrea Schertler July The responsblty for the contents of the workng

More information

Portfolio Loss Distribution

Portfolio Loss Distribution Portfolo Loss Dstrbuton Rsky assets n loan ortfolo hghly llqud assets hold-to-maturty n the bank s balance sheet Outstandngs The orton of the bank asset that has already been extended to borrowers. Commtment

More information

DO LOSS FIRMS MANAGE EARNINGS AROUND SEASONED EQUITY OFFERINGS?

DO LOSS FIRMS MANAGE EARNINGS AROUND SEASONED EQUITY OFFERINGS? DO LOSS FIRMS MANAGE EARNINGS AROUND SEASONED EQUITY OFFERINGS? Fernando Comran, Unversty of San Francsco, School of Management, 2130 Fulton Street, CA 94117, Unted States, fcomran@usfca.edu Tatana Fedyk,

More information

Linear Circuits Analysis. Superposition, Thevenin /Norton Equivalent circuits

Linear Circuits Analysis. Superposition, Thevenin /Norton Equivalent circuits Lnear Crcuts Analyss. Superposton, Theenn /Norton Equalent crcuts So far we hae explored tmendependent (resste) elements that are also lnear. A tmendependent elements s one for whch we can plot an / cure.

More information

Criminal Justice System on Crime *

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

More information

AN APPOINTMENT ORDER OUTPATIENT SCHEDULING SYSTEM THAT IMPROVES OUTPATIENT EXPERIENCE

AN APPOINTMENT ORDER OUTPATIENT SCHEDULING SYSTEM THAT IMPROVES OUTPATIENT EXPERIENCE AN APPOINTMENT ORDER OUTPATIENT SCHEDULING SYSTEM THAT IMPROVES OUTPATIENT EXPERIENCE Yu-L Huang Industral Engneerng Department New Mexco State Unversty Las Cruces, New Mexco 88003, U.S.A. Abstract Patent

More information

Factors Affecting Outsourcing for Information Technology Services in Rural Hospitals: Theory and Evidence

Factors Affecting Outsourcing for Information Technology Services in Rural Hospitals: Theory and Evidence Factors Affectng Outsourcng for Informaton Technology Servces n Rural Hosptals: Theory and Evdence Bran E. Whtacre Department of Agrcultural Economcs Oklahoma State Unversty bran.whtacre@okstate.edu J.

More information

Statistical Methods to Develop Rating Models

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

More information

CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK. Sample Stability Protocol

CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK. Sample Stability Protocol CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK Sample Stablty Protocol Background The Cholesterol Reference Method Laboratory Network (CRMLN) developed certfcaton protocols for total cholesterol, HDL

More information

What is Candidate Sampling

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

More information

MARKET SHARE CONSTRAINTS AND THE LOSS FUNCTION IN CHOICE BASED CONJOINT ANALYSIS

MARKET SHARE CONSTRAINTS AND THE LOSS FUNCTION IN CHOICE BASED CONJOINT ANALYSIS MARKET SHARE CONSTRAINTS AND THE LOSS FUNCTION IN CHOICE BASED CONJOINT ANALYSIS Tmothy J. Glbrde Assstant Professor of Marketng 315 Mendoza College of Busness Unversty of Notre Dame Notre Dame, IN 46556

More information

Hollinger Canadian Publishing Holdings Co. ( HCPH ) proceeding under the Companies Creditors Arrangement Act ( CCAA )

Hollinger Canadian Publishing Holdings Co. ( HCPH ) proceeding under the Companies Creditors Arrangement Act ( CCAA ) February 17, 2011 Andrew J. Hatnay ahatnay@kmlaw.ca Dear Sr/Madam: Re: Re: Hollnger Canadan Publshng Holdngs Co. ( HCPH ) proceedng under the Companes Credtors Arrangement Act ( CCAA ) Update on CCAA Proceedngs

More information

CS 2750 Machine Learning. Lecture 3. Density estimation. CS 2750 Machine Learning. Announcements

CS 2750 Machine Learning. Lecture 3. Density estimation. CS 2750 Machine Learning. Announcements Lecture 3 Densty estmaton Mlos Hauskrecht mlos@cs.ptt.edu 5329 Sennott Square Next lecture: Matlab tutoral Announcements Rules for attendng the class: Regstered for credt Regstered for audt (only f there

More information

Chapter 7: Answers to Questions and Problems

Chapter 7: Answers to Questions and Problems 19. Based on the nformaton contaned n Table 7-3 of the text, the food and apparel ndustres are most compettve and therefore probably represent the best match for the expertse of these managers. Chapter

More information

The Application of Fractional Brownian Motion in Option Pricing

The Application of Fractional Brownian Motion in Option Pricing Vol. 0, No. (05), pp. 73-8 http://dx.do.org/0.457/jmue.05.0..6 The Applcaton of Fractonal Brownan Moton n Opton Prcng Qng-xn Zhou School of Basc Scence,arbn Unversty of Commerce,arbn zhouqngxn98@6.com

More information

To manage leave, meeting institutional requirements and treating individual staff members fairly and consistently.

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:

More information

Credit Limit Optimization (CLO) for Credit Cards

Credit Limit Optimization (CLO) for Credit Cards Credt Lmt Optmzaton (CLO) for Credt Cards Vay S. Desa CSCC IX, Ednburgh September 8, 2005 Copyrght 2003, SAS Insttute Inc. All rghts reserved. SAS Propretary Agenda Background Tradtonal approaches to credt

More information

RESEARCH DISCUSSION PAPER

RESEARCH DISCUSSION PAPER Reserve Bank of Australa RESEARCH DISCUSSION PAPER Competton Between Payment Systems George Gardner and Andrew Stone RDP 2009-02 COMPETITION BETWEEN PAYMENT SYSTEMS George Gardner and Andrew Stone Research

More information

What should (public) health insurance cover?

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;

More information

1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP)

1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP) 6.3 / -- Communcaton Networks II (Görg) SS20 -- www.comnets.un-bremen.de Communcaton Networks II Contents. Fundamentals of probablty theory 2. Emergence of communcaton traffc 3. Stochastc & Markovan Processes

More information

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 * 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

More information

Recurrence. 1 Definitions and main statements

Recurrence. 1 Definitions and main statements Recurrence 1 Defntons and man statements Let X n, n = 0, 1, 2,... be a MC wth the state space S = (1, 2,...), transton probabltes p j = P {X n+1 = j X n = }, and the transton matrx P = (p j ),j S def.

More information

An Interest-Oriented Network Evolution Mechanism for Online Communities

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

More information

Valuing Customer Portfolios under Risk-Return-Aspects: A Model-based Approach and its Application in the Financial Services Industry

Valuing Customer Portfolios under Risk-Return-Aspects: A Model-based Approach and its Application in the Financial Services Industry Buhl and Henrch / Valung Customer Portfolos Valung Customer Portfolos under Rsk-Return-Aspects: A Model-based Approach and ts Applcaton n the Fnancal Servces Industry Hans Ulrch Buhl Unversty of Augsburg,

More information

Project Networks With Mixed-Time Constraints

Project Networks With Mixed-Time Constraints Project Networs Wth Mxed-Tme Constrants L Caccetta and B Wattananon Western Australan Centre of Excellence n Industral Optmsaton (WACEIO) Curtn Unversty of Technology GPO Box U1987 Perth Western Australa

More information

Exhaustive Regression. An Exploration of Regression-Based Data Mining Techniques Using Super Computation

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

More information

TESTING FOR EVIDENCE OF ADVERSE SELECTION IN DEVELOPING AUTOMOBILE INSURANCE MARKET. Oksana Lyashuk

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

More information

Multiple-Period Attribution: Residuals and Compounding

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

More information

Extending Probabilistic Dynamic Epistemic Logic

Extending Probabilistic Dynamic Epistemic Logic Extendng Probablstc Dynamc Epstemc Logc Joshua Sack May 29, 2008 Probablty Space Defnton A probablty space s a tuple (S, A, µ), where 1 S s a set called the sample space. 2 A P(S) s a σ-algebra: a set

More information

Fuzzy Regression and the Term Structure of Interest Rates Revisited

Fuzzy Regression and the Term Structure of Interest Rates Revisited Fuzzy Regresson and the Term Structure of Interest Rates Revsted Arnold F. Shapro Penn State Unversty Smeal College of Busness, Unversty Park, PA 68, USA Phone: -84-865-396, Fax: -84-865-684, E-mal: afs@psu.edu

More information

Return decomposing of absolute-performance multi-asset class portfolios. Working Paper - Nummer: 16

Return decomposing of absolute-performance multi-asset class portfolios. Working Paper - Nummer: 16 Return decomposng of absolute-performance mult-asset class portfolos Workng Paper - Nummer: 16 2007 by Dr. Stefan J. Illmer und Wolfgang Marty; n: Fnancal Markets and Portfolo Management; March 2007; Volume

More information

HARVARD John M. Olin Center for Law, Economics, and Business

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

More information

Health Insurance and Household Savings

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

More information

SUPPLIER FINANCING AND STOCK MANAGEMENT. A JOINT VIEW.

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

More information

STATISTICAL DATA ANALYSIS IN EXCEL

STATISTICAL DATA ANALYSIS IN EXCEL Mcroarray Center STATISTICAL DATA ANALYSIS IN EXCEL Lecture 6 Some Advanced Topcs Dr. Petr Nazarov 14-01-013 petr.nazarov@crp-sante.lu Statstcal data analyss n Ecel. 6. Some advanced topcs Correcton for

More information

An Evaluation of the Extended Logistic, Simple Logistic, and Gompertz Models for Forecasting Short Lifecycle Products and Services

An Evaluation of the Extended Logistic, Simple Logistic, and Gompertz Models for Forecasting Short Lifecycle Products and Services An Evaluaton of the Extended Logstc, Smple Logstc, and Gompertz Models for Forecastng Short Lfecycle Products and Servces Charles V. Trappey a,1, Hsn-yng Wu b a Professor (Management Scence), Natonal Chao

More information

Data Mining from the Information Systems: Performance Indicators at Masaryk University in Brno

Data Mining from the Information Systems: Performance Indicators at Masaryk University in Brno Data Mnng from the Informaton Systems: Performance Indcators at Masaryk Unversty n Brno Mkuláš Bek EUA Workshop Strasbourg, 1-2 December 2006 1 Locaton of Brno Brno EUA Workshop Strasbourg, 1-2 December

More information

Implied (risk neutral) probabilities, betting odds and prediction markets

Implied (risk neutral) probabilities, betting odds and prediction markets Impled (rsk neutral) probabltes, bettng odds and predcton markets Fabrzo Caccafesta (Unversty of Rome "Tor Vergata") ABSTRACT - We show that the well known euvalence between the "fundamental theorem of

More information

Optimal Bidding Strategies for Generation Companies in a Day-Ahead Electricity Market with Risk Management Taken into Account

Optimal Bidding Strategies for Generation Companies in a Day-Ahead Electricity Market with Risk Management Taken into Account Amercan J. of Engneerng and Appled Scences (): 8-6, 009 ISSN 94-700 009 Scence Publcatons Optmal Bddng Strateges for Generaton Companes n a Day-Ahead Electrcty Market wth Rsk Management Taken nto Account

More information

When Talk is Free : The Effect of Tariff Structure on Usage under Two- and Three-Part Tariffs

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

More information

Oservce Vs. Sannet - Which One is Better?

Oservce Vs. Sannet - Which One is Better? o rcng n Compettve Telephony Markets Yung-Mng L nsttute of nformaton Management Natonal Chao Tung Unversty, Tawan 886-3-57111 Ext 57414 yml@mal.nctu.edu.tw Shh-Wen Chu nsttute of nformaton Management Natonal

More information

Quasi-Hyperbolic Discounting and Social Security Systems

Quasi-Hyperbolic Discounting and Social Security Systems Quas-Hyperbolc Dscountng and Socal Securty Systems Mordecha E. Schwarz a and Eytan Sheshnsk b May 22, 26 Abstract Hyperbolc countng has become a common assumpton for modelng bounded ratonalty wth respect

More information

The Use of Analytics for Claim Fraud Detection Roosevelt C. Mosley, Jr., FCAS, MAAA Nick Kucera Pinnacle Actuarial Resources Inc.

The Use of Analytics for Claim Fraud Detection Roosevelt C. Mosley, Jr., FCAS, MAAA Nick Kucera Pinnacle Actuarial Resources Inc. Paper 1837-2014 The Use of Analytcs for Clam Fraud Detecton Roosevelt C. Mosley, Jr., FCAS, MAAA Nck Kucera Pnnacle Actuaral Resources Inc., Bloomngton, IL ABSTRACT As t has been wdely reported n the nsurance

More information

A Novel Methodology of Working Capital Management for Large. Public Constructions by Using Fuzzy S-curve Regression

A Novel Methodology of Working Capital Management for Large. Public Constructions by Using Fuzzy S-curve Regression Novel Methodology of Workng Captal Management for Large Publc Constructons by Usng Fuzzy S-curve Regresson Cheng-Wu Chen, Morrs H. L. Wang and Tng-Ya Hseh Department of Cvl Engneerng, Natonal Central Unversty,

More information

A Lyapunov Optimization Approach to Repeated Stochastic Games

A Lyapunov Optimization Approach to Repeated Stochastic Games PROC. ALLERTON CONFERENCE ON COMMUNICATION, CONTROL, AND COMPUTING, OCT. 2013 1 A Lyapunov Optmzaton Approach to Repeated Stochastc Games Mchael J. Neely Unversty of Southern Calforna http://www-bcf.usc.edu/

More information