Is There A Tradeoff between EmployerProvided Health Insurance and Wages?


 Maximillian Watts
 2 years ago
 Views:
Transcription
1 Is There A Tradeoff between EmployerProvded Health Insurance and Wages? Lye Zhu, Southern Methodst Unversty October 2005 Abstract Though most of the lterature n health nsurance and the labor market assumes a tradeoff between employerprovded health nsurance and wages, ts emprcal valdty has not been establshed. Employng Current Populaton Survey 2004 data, ths paper assesses the tradeoff hypothess n a dstrbutonal analyss framework usng stochastc domnance tests. In addton, t contrbutes to the prevous lterature by ncorporatng an ndrect effect of health nsurance on wages nto the analyss. Health nsurance not only drectly affects wages, but also ndrectly by mprovng ndvdual productvty. The results confrm the exstence of a tradeoff for full tme workng wves, and explan why the prevous lterature fals to do so. JEL: C12, C14, D31, I11, J3 Keywords: EmployerProvded Health Insurance, Tradeoff, Dstrbutonal analyss, Stochastc Domnance Tests I am grateful to Professor Danel Mllmet for hs profound supervson and contnuous help. Thanks to Professor Nathan Balke, Professor Tom Fomby and Professor Y Deng for helpful comments and advce. All errors are mne. Correspondng address: Lye Zhu, Department of Economcs, Southern Methodst Unversty, Dallas, TX Tel: Emal:
2 I. Introducton In contrast to most other developed countres, health nsurance n the US s both provded and fnanced predomnantly by employers, especally for workng aged ndvduals. Current Populaton Survey 2004 data show that 63% of Amercan adult populaton s covered by employerprovded health nsurance (HI) 1 and ths percentage changes only slghtly over tme. The magntude of employerprovded health nsurance coupled wth the nsttutons and rules for health nsurance provson have made health nsurance an mportant parameter of labor market decsons both for ndvduals and frms. Theory predcts that there s a tradeoff between employerprovded health nsurance and wages. Therefore, most studes of health nsurance effects on both labor force partcpaton and job choce as well as on other labor ssues are based on ths assumpton. However, the lnk has not been emprcally affrmed. The focus of the recent emprcal studes of the hypothess tests has centered on problems wth the data and the endogenety of HI. However, even usng "good" data and applyng methods, such as panel data or nstrumental varable (IV) models, to solve the endogenety problem, researchers stll cannot obtan strong results n support of the tradeoff hypothess. Most fnd a postve relatonshp; a few obtan results that support the tradeoff hypothess, but have certan shortcomngs n ther studes whch weaken the results. In addton to addressng these two dffcultes, ths paper also deals wth two other possble shortcomngs of the exstng lterature that may explan the nconsstent results. Frst, the basc lnear regresson model employed n most studes may be msspecfed. Second, the effect of HI on wages may be heterogeneous. In terms of model specfcaton, ths paper not only consders the drect effect of HI on wages, but also ncorporates the ndrect effect of HI nto the analyss. HI may ncrease ndvdual productvty, and therefore the returns to other determnants of productvty (such as educaton), by enhancng ndvdual health or boostng workng morale. Therefore, HI enters the wage equaton, not only from the drect tradeoff pont of vew, but also from the productvtyenhancng pont of vew. The drect effect of HI can be reflected by the coeffcent on HI n a wage equaton. The ndrect effect can be captured by the coeffcents on the nteractons between HI and 1 Employerprovded health nsurance s denoted as HI n ths paper. Therefore, HI n ths paper does not refer to whether ndvduals have nsurance or not, but refers to whether ndvduals have employerprovded health nsurance or not. Smlarly, NOHI refers to ndvduals wthout employerprovded health nsurance. It wll be specfed f health nsurance status s n use. 1
3 ndvdual or job characterstcs n a wage equaton. These two effects, the drect tradeoff effect and the ndrect productvtyenhancng effect, work n opposte drectons. The total effect of HI on wages depends on the magntudes of these two effects. Snce the coeffcents on the nteractons turn out to be sgnfcantly dfferent from zero, and the nteractons are hghly correlated wth other regressors, the prevous lterature whch omts the nteracton terms generates based results. Another possble reason that the prevous lterature cannot fnd sgnfcant tradeoff s the heterogenety problem. If ndvdual heterogenety exsts, the estmated tradeoff can be statstcally nsgnfcant when usng regresson analyss, whch s based on the condtonal mean, because the tradeoff wll dffer among ndvduals and may offset. To deal wth ndvdual heterogenety, the tradeoff hypothess s tested n ths paper usng a dstrbutonal analyss. The dstrbutonal analyss combned wth tests of stochastc domnance (SD) dffers from prevous emprcal regresson analyses. It utlzes all avalable nformaton to fnd and test for unform rankngs of dstrbutons for dfferent groups of people. Moreover, ths approach allows one to easly assess potental heterogenety n the magntudes of tradeoff across the dstrbuton of wages. Specfcally, the dstrbutonal analyss and stochastc domnance tests are performed n two cases: when HI s exogenous and when t s endogenous. In the frst case, ndvduals are classfed accordng to ther HI status (HI or NOHI). Dstrbutons of the followng outcomes are obtaned for each sample: the total wage, wages explaned by the constant and resduals, and wages explaned by ndvdual and job characterstcs. In addton, the hypothetcal wages of NOHI sample usng the HI regresson coeffcents are obtaned. Then, the correspondng HI and NOHI dstrbutons are compared by stochastc domnance tests. Wage dfferentals explaned by the constant and resduals correspond to the drect effect n the typcal regresson analyss. By ntroducng the hypothetcal wage, wage dfferentals explaned by characterstcs s further decomposed nto two gaps: characterstc gap, whch s solely due to dfferences n attrbutes, and compensaton gap, whch s due to dfferent wage returns for the same characterstcs across the two groups. The compensaton gap corresponds to the ndrect effect of HI, and s nterpreted as mproved ndvdual productvty through better health or hgher morale. Snce most of the prevous lterature predcts that HI s endogenous n the wage equaton, t s worthwhle to deal wth ths endogenety and thus the selecton bas problem. When HI s endogenous, followng the same 2
4 procedure as above, ndvduals are dvded nto two groups accordng to a bnary nstrument of HI. Accordng to Abade (2002), for the two groups dvded by a bnary IV, the dfference between the cumulatve dstrbuton functons (CDF) s proportonal to the dfference between the CDFs of the complers n each group under certan assumptons. Here, complers refer to those ndvduals who comply wth the IV,.e., whose HI status changes accordng to the IV. Thus, we can test for stochastc domnance among the complers by testng for domnance across the two groups dvded by a bnary IV. The nstruments, borrowed from Olson (2002), are tested for ther valdty. The endogenety of HI s also tested usng the vald nstruments. The results n ths paper, based on Current Populaton Survey (CPS) 2004 data on fulltme workng wves, not only suggest the exstence of a tradeoff, but also suggest the exstence of an ndrect effect of HI on wages. Ths explans why prevous studes fal to establsh a tradeoff. The paper s organzed as follows: Part II s the lterature revew. Part III ntroduces the methods used n ths paper. Part IV descrbes the data. Part V provdes the results. Part VI concludes. II. Lterature Revew A. Theoretcal Foundaton of the Tradeoff Hypothess The theory of the tradeoff between wages and health nsurance s that health nsurance s a frnge beneft employers provde to employees as compensaton. In a compettve product market, economc theory suggests that what matters to proft maxmzng frms s the value of the total compensaton package that they must offer to attract labor servces. If the compensaton level s too low, there wll not be enough labor attracted to the frm; f the compensaton level s too hgh, the frm cannot survve n the market. Thus, frms compensaton level to employees wll be smlar to that offered by other frms whch face the same labor pool. As a result, f we assume that the other benefts offered by frms do not change, to reman compettve, frms wll reduce wages by $1 for each dollar ncrease n health nsurance benefts. Indvduals wll then choose among frms offerng dfferent wage/health nsurance combnaton accordng to ther own preferences. Under ths scenaro, there exsts a tradeoff between wages and health nsurance. As a benchmark, the predct tradeoff s W / HI = 1. The actual tradeoff would dffer from ths n certan stuatons. For example, the tradeoff should be less than 1 f the health nsurance reduces job turnover and job turnover s a cost to the frm. 3
5 Another example, snce health nsurance costs are not taxable ncome for frms, but a cost before tax, the actual tradeoff should be 1 (1 t), where t s the tax rate. The fgures n Curre and Madran (1999) can llustrate the above theory. In Fgure 1 (A), frm 1 and frm 2 are two frms facng a compettve product market and same labor pool. Employee 1 and Employee 2 need the same total compensaton level, but have dfferent preferences over wage compensaton and health nsurance compensaton. We can see there s a tradeoff between wages and health nsurance: f the wage level s hgh, health nsurance wll be low (Employee 2); on the other hand, a lower wage level s combned wth a hgher health nsurance level (Employee 1). Thus, f all frms face the same tradeoff between wages and benefts n total compensaton, the wage/health nsurance bundles that are observed n the market wll reflect the sortng of employees across frms on the bass of ther heterogeneous preferences for health nsurance. Ths framework s the motvaton for much of the lterature on the tradeoff between wages and employerprovded health nsurance or other frnge benefts. B. Emprcal Tests of the Tradeoff Hypothess The emprcal mplementaton of the wagehealth nsurance tradeoff pctured n Fgure 1 (A) has typcally been the estmaton of y = α + X β + HI γ + µ (1) where y s labor market outcome of nterest (here, log hourly wages) for ndvdual, ndvdual and/or job characterstcs for ndvdual, coverage for ndvdual, and µ s ndvdual dsturbance. X s a vector of HI s ether the avalablty or value of health nsurance Condtonal on X and n the absence of tax consderatons, the theory predctsγ = 1f y represents hourly wages and HI s approprately measured n dollars. The emprcal valdty of Equaton (1) wth respect to wages, however, has been dffcult to establsh. The typcal estmates of γ are ether postve or nsgnfcant. The lterature has thus focused not on the magntude of the wagehealth nsurance tradeoff, but on the reasons why economsts cannot fnd evdence that there s one. 4
6 One man pont that researchers focus on s the lack of a sutable dataset. In order to estmate Equaton (1) data s requred on both compensaton and frnge beneft expendtures, that s, wage and health nsurance levels. The frmlevel datasets that nclude nformaton on benefts expendtures are usually aggregated at the frm level, but typcally do not nclude the types of human captal varables that mght allow researchers to control for the productvty of the workforce. The problem created by these omtted varables s llustrated n Fgure 1 (B). Employee 1 has hgher abltes and thus earns more wages as well as frnge benefts, such as health nsurance. Employer 2 has relatvely low abltes and thus earns a lower wage and lower frnge benefts. Thus, f total compensaton ncreases wth average worker productvty and both benefts and other consumpton goods are normal, a regresson usng such frmlevel data wll yeld a postve relatonshp between wages and benefts. Some researchers thus merge n average employer expendtures by ndustry from a frmlevel dataset to ndvduallevel datasets. Even so, such methods stll usually lead to a postve relatonshp. For example, Lebowtz (1983) uses the RAND Health Insurance Study to estmate the wage/frnge beneft tradeoff, but stll fnds a postve (although nsgnfcant) effect of employer health nsurance expendtures on wages. Researchers then deduce that the reason that they cannot fnd a tradeoff s that productvty s determned by both observed human captal varables and unobserved (to the econometrcan) ablty. Ths can be shown as y = α + X β + HIγ + δ + ~ µ (2) where δ s some unobservable varable or fxed effect, ~ µ s ndvdual dsturbance, and other varables are the same as n Equaton (1). Thus, f δ s omtted and s correlated wth health nsurance, the estmaton of γ wll be based. Ths mples that even condtonal on observed human captal varables, some frms employ hgher ablty workers and pay a hgher level of total compensaton. But, as shown n Fgure 1 (B), f ths hgher level of compensaton s allocated to both wages and benefts, a postve relatonshp between wages and frnge benefts wll be estmated despte usng observable human captal controls. Varous approaches have been taken to crcumvent ths problem of omtted varable bas. One common method s the dfferencng method,.e., purge the unobserved varable by dfferencng Equaton (2), ether across dfferent years for same person, or across dfferent job classfcatons wthn a frm. Buchmueller and Lettau (1997) use an employerlevel dataset that tracks compensaton and beneft expendtures for varous jobs 5
7 wthn the frm over a 4year perod. They purge the unobserved productvty dfferences by dfferencng Equaton (2) over tme, essentally examnng the mpact of the growth n health nsurance expendtures over tme on changes n wages over tme. Even so, they fnd no evdence of tradeoff between health nsurance and wages. Olson (1992), Mller (1995) and Ryan (1997) use panel datasets of workers to estmate the effect of changes n health nsurance coverage on changes n wages. However, the problem s that the majorty of changes n health nsurance coverage are generated by job changes and the unobserved job characterstcs that also mpact compensaton are unlkely to be constant followng a job change. The study by Olson s less subject to ths crtcsm as hs sample of dsplaced workers s exogenously selected by the closng of a plant or smlar event. Gruber (1994) explots a dfferent source of varaton, the changes of laws n many states n 1970s, whch requred employers who offered health nsurance to treat pregnancy and chldbrth the same as any other health ssue. He fnds that wages for those groups most lkely to beneft from the law fell n drect proporton to the antcpated cost of the beneft. Overall, hs results are consstent wth a full shftng of employer health nsurance costs onto wages. A recent development n the lterature to deal wth the endogenety problem s the IV method. Olson (2002), usng Current Populaton Survey (CPS) data, models the wages of marred women employed fulltme n the labor market. He uses husband s unon status, husband s frm sze, and husband s health nsurance coverage through hs job as nstruments for wfe s own employerprovded health nsurance benefts. The estmates suggest that wves wth own employerprovded health nsurance accept a wage about 20% lower than what they would have receved workng n a job wthout benefts. However, Olson does not fully test the valdty of the nstruments, especally the orthogonalty of the nstruments wth respect to wages. III. Methodology To test the tradeoff hypothess, ths paper not only consders the drect effect of HI on wages, but also ncorporates the ndrect effect of HI nto the wage equaton. In addton, ths paper deals wth ndvdual heterogenety problem and analyzes the tradeoff n a dstrbutonal framework. Fnally, to solve the endogenety due to omtted varables, the dstrbutonal analyss reles on the nstrumental varables framework ntroduced by Abade (2002). 6
8 A. Productvty Effect of HI If we consder health as one knd of human captal, health nsurance can be treated as an nvestment n ths human captal. Employees typcally start wth a large health endowment that must be contnuously replenshed as t deprecates. If employees have employerprovded health nsurance, they wll have guaranteed health care, and ther health rsk can be allevated. Therefore, HI can mprove employees health, whch can then rase ther productvty. Although employees wth NOHI may stll have other types of health nsurance, HI s usually more generous and covers hgher health care spendng. Moreover, employees productvty can also be enhanced by hgher morale due to better health care. Thus, HI enters the wage equaton not only from the drect tradeoff pont of vew, but also from the productvtyenhancng pont of vew. Fgure 1 (C) llustrates ths productvty effect of HI. When employee 1 obtans better health nsurance from her employer, her productvty mproves and therefore the frm s proft ncreases. On the graph, ths hgher proft can be reflected by an soproft curve, whch s hgher than the orgnal socost curve for the frm. In a compettve market, the frm wll pay hgher wages to employee 1 to reflect the ncrease n proft. Therefore, the actual wage of employee 1 s not W2 but W3. Movement from A to B reflects the drect tradeoff between HI and wages; movement from B to C reflects the ndrect effect of HI on wages. These two effects work n opposte drectons. The total effect of HI on wages s reflected by the movement from A to C, whch only has an nsgnfcant change of wages, snce the above two effects offset each other. In the regresson analyss, ths ndrect effect of HI can be captured by addng the nteracton terms between HI and other characterstc varables nto equaton (2) y = α + X β + HI γ + HI X θ + µ (3) where vector θ, the coeffcents on the nteracton terms, measures the dfferent returns to the same characterstcs for ndvduals wth HI and wth NOHI. As before, the coeffcent on HI,γ, captures the HI drect effect. The prevous regresson analyss gnores the ndrect effect and thus may result n based estmaton f θ 0, snce the nteracton terms are correlated wth HI. B. Dstrbutonal Analyss Assumng Exogenety of HI 7
9 To access the degree of heterogenety, ths paper tests the tradeoff hypothess usng dstrbutonal analyss by comparng the wage dstrbutons of wves wth HI and wves wth NOHI. Ths comparson between two dstrbutons can be formalzed usng stochastc domnance (SD). Several tests for SD have been proposed n the lterature; the approach heren s based on a generalzed KolmogorovSmrnov test. Suppose F 1 and F0 are CDFs for two groups of ndvduals, group 1 and group 0, then the followng defntons apply: F 1 Frst Order Domnates (FSD) F 0 f and only f F1 ( y) F0 ( y), y Y, wth strct nequalty for some y ; F 1 Second Order Domnates (SSD) F 0 f and only f y ( t) dt y F0 F 1 ( t) dt, y Y, wth strct nequalty for some y. The tests characterze the relatonshp between the dstrbutons. Therefore, f the NOHI wage CDF domnates the HI wage CDF, there exsts a tradeoff between employerprovded health nsurance and wages. To test for domnance, defne the emprcal CDFs for Y wth HI and wth NOHI respectvely as 1 F ˆ (4) N HI, N ( y) = I( YHI, y) N = 1 1 F ˆ (5) M NOHI, M ( y) = I( YNOHI, j y) M j= 1 where N (M) s the sze of the sample wth HI (NOHI). Now defne the followng functons of the jont dstrbuton d MN = mn sup[ FNOHI, M ( y) FHI, N ( y)] (6) M + N y MN y s = mn sup{ [ FNOHI, M ( t) FHI, N ( t)] dt} (7) M + N y where mn s taken over FNOHI FHI and HI FNOHI F (n effect performng two tests). The tests for FSD and SSD are based on the emprcal counterparts of d and s ( dˆ and ŝ ) usng the emprcal CDFs. The test for FSD requres: 8
10 () computng the values of Fˆ NOHI, M ( yq ) and Fˆ HI, N ( yq ) for y q, q = 1,..., Q, where Q denotes the number of ponts n the support Y utlzed ( Q = 500 n the applcaton), () computng the dfferences d = Fˆ ( y ) Fˆ ( y ), d = Fˆ ( y ) Fˆ ( y ), () fndng d ˆ = mn{max{ d }, max{ d }}. 1q HI, M q NOHI, N q 1q 2q 2q NOHI, M q HI, N q If d ˆ 0 (to a degree of statstcal certanty), then the null hypothess of frst order domnance s not rejected. Furthermore, f ˆ 0 d and max{ d 1 } < 0, then Y HI FSD Y NOHI as the value of the CDF for dstrbuton Y NOHI s at least as great as the correspondng value for dstrbuton Y HI at all y q, q = 1,..., Q. On the other hand, f d ˆ 0 and max{ d 2 } < 0, then Y NOHI FSD Y HI. The analogous test for SSD requres: () computng the values of Fˆ NOHI, M ( yq ) and Fˆ HI, N ( yq ) for y q, q = 1,..., Q, where Q denotes the number of ponts n the support Y utlzed ( Q = 500 n the applcaton), () computng the dfferences d = Fˆ ( y ) Fˆ ( y ), d = Fˆ ( y ) Fˆ ( y ), 1q HI, M q NOHI, N q 2q NOHI, M q HI, N q () calculatng the sums, fndng s q = = d 1 1 1q,, s = 2 q d = 1 2q,, q = 1,..., Q, ˆ 1q 2q (v) fndng s = mn{max{ s }, max{ s }}. q If s ˆ 0 (to a degree of statstcal certanty), then the null hypothess of second order domnance s not rejected. Moreover, f s ˆ 0 and max{ s 1q } <0, then Y HI SSD Y NOHI as the cumulatve value of the CDF for dstrbuton Y NOHI exceeds the correspondng value for dstrbuton Y HI at all y q, q = 1,..., Q ; otherwse, f max{ s 2q } <0, then Y NOHI SSD Y HI. Ths paper approxmates the emprcal dstrbuton of the test statstcs usng bootstrap technques. For each of 1000 bootstrap samples, dˆ and ŝ are computed. Thus, whether the emprcal dstrbutons are characterzed by FSD or by SSD can be reported. The bootstrap reported probabltes represent the crtcal levels assocated wth the nonrejecton regon. q 9
11 To ths pont, Y HI and Y NOHI have represented two uncondtonal varables. However, the magntudes of wages are not only decded by whether the employee has HI or has NOHI, but also decded by other ndvdual and job characterstcs. Therefore, we need to separate the HI effect on wages from the effects of other varables to correctly measure the tradeoff. From Equaton (3), the predcted wages respectvely for HI and NOHI groups are = ˆ α + X ˆ β + ˆ ε, k HI NOHI (8) yˆ k, k k, k k, =, The partal wage denotes the wage part explaned by the estmated ntercept and resduals for HI and NOHI, y d k, ˆ k k, = The CDFs of ths partal wage for HI and NOHI groups, denoted as = α + ˆ ε, k HI, NOHI (9) d d FNOHI and F HI, can be compared by domnance tests. The dfference between the two dstrbutons corresponds to the drect effect n the regresson analyss, and can be nterpreted as the tradeoff between HI and wages f HI s exogenous. The part left n wage equaton (8) stands for wages explaned by characterstc varables, whch s y = X ˆ β, k HI, NOHI (10) ch k, k, k = The correspondng CDFs, denoted as F and F ch NOHI ch HI, can also be compared by domnance tests. However, the dfference between the two dstrbutons reflects dfferences n X and dfferences n returns. We can decompose ths part usng a smlar method as the standard BlnderOaxaca decomposton. For a regresson, the standard BlnderOaxaca decomposton s n the form of the average value. Here, the author extends the standard Oaxaca decomposton to the dstrbutonal analyss. A hypothetcal wage CDF for ndvduals wth NOHI can be calculated usng the estmated coeffcents from the wage equaton of ndvduals wth HI, h.e., ˆ h F X β ) (shorted as F / ). Wth ths, decomposton smlar to the standard OaxacaBlnder ( NOHI HI decomposton can be performed, F ch HI NOHI HI ch ch h h ch F = F F ] + [ F F ] (11) NOHI [ HI NOHI / HI NOHI / HI The frst part of the expresson on the rght hand sde s the dfference n the earnngs dstrbutons explaned by dfferent characterstcs of ndvduals wth HI and NOHI. It can be called the characterstcs gap. The second part s the dfference due to the estmated parameters or wage structure. It s labeled the unexplaned part of the 10 NOHI
12 wage dfference. In ths paper, we call t the compensaton gap, the wage return dfferences between HI and NOHI ndvduals wth the same characterstcs. It reflects the ndrect effect of HI on wages. In general, FSD and SSD are tested for 5 pars of CDFs: the orgnal data, the partal wages, the predcted wage explaned by the characterstc varables, and the hypothetcal wages. Ths paper also presents graphs of the horzontal dfference between any two CDFs n consderaton. Specfcally, at every quantle, the wage dfference between two CDFs can be calculated and can be plotted aganst the accumulated probabltes. By ths, the pattern of the dfference across the entre wage dstrbutons can be analyzed. C. Dstrbutonal Analyss When HI Is Endogenous Snce most of the prevous lterature predcts that HI s endogenous n the wage equaton, and snce the wage equaton may not be correctly specfed, t s worthwhle to test the endogenety of HI. The reason s that f we dvde the data accordng to an endogenous varable, for example, HI, regresson and dstrbutonal analyss wll suffer from selecton bas. However, f we can fnd nstruments for HI, the tradeoff between HI and wages can be put nto the nstrumental varables envronment developed by Abade (2002). Let Y (0) be the potental outcome for ndvdual wthout treatment, or wth NOHI. Let Y (1) be the potental outcome for the same ndvdual wth HI. Let Z be a bnary nstrument for HI. Denote HI (0) the value that HI would have taken f Z = 0 ; HI (1) has the same meanng for Z = 1. In practce, for any partcular ndvdual we can not observe both HI (0) and HI (1). Instead the realzed treatment HI = HI 1) Z + HI (0) (1 Z ) s observed. Smlarly, only Y = Y 1) HI + Y (0) (1 HI ) s ( ( observed. Under the assumptons of: (1) ndependence of the nstrument: ( Y (0), Y (1), HI (0), HI (1) ) s ndependent of Z ; (2) frst stage: 0 < P ( Z = 1) < 1 and P ( HI (1) = 1) > P( HI (0) = 1) ; (3) monotoncty: P ( HI (1) HI (0)) = 1, Abade has the followng lemma for ndvduals whose treatment status s affected by varaton n the nstrument: HI ( 0) = 0 and HI ( 1) = 1(the subpopulaton of complers): 11
13 F C E[ I ( Y y) Z = 1] E[ I( Y y) Z = 0] ( y) F0 ( y) = = K ( F1 0 ), (12) E[ HI Z = 1] E[ HI Z = 0] C 1 F C C where F E[ I( Y (1) y) HI (1) = 1, HI (0) 0] and F E[ I( Y (0) y) HI (1) = 1, HI (0) 0], 1 = = 12 0 = = denotes the dstrbutons of complers; F E[ I( Y (1) y) Z 1] and F E[ I( Y (0) y) Z 0], the 1 = = 0 = = condtonal dstrbutons gven Z = 1 and Z = 0 ; K = ( E[ HI Z = 1] E[ HI Z = 0] ) <. Under assumpton (2), we know that K > 0. 1 Ths lemma states that for a bnary nstrument, the dfference between two dstrbutons of complers s proportonal to the dfference between the two dstrbutons categorzed by the nstrument. So, f we use the nstrument to solve the endogenety problem and dvde the ndvduals nto two groups accordng to the bnary nstrument, comparng F 1 and F 0 provdes the sgn of the dfference between from the comparson of the orgnal dstrbuton of complers, dfferent nstruments wll generate dfferent results. C C F 1 and F 0. Ths s dfferent F HI and F NOHI. Moreover, snce the IV results apply only to The nstruments used n ths paper are borrowed from Olson (2002): husband s frm sze, husband s health nsurance, and husband s Unon status. Olson (2002) qualtatvely analyzes the economc valdty of these three nstruments, but does not statstcally test ther valdty. For a varable to be the rght nstrument, t must be correlated wth the endogenous varable, and orthogonal to the error process. The relevance of the nstrument varable can be tested by examnng the frst stage regresson. The test statstcs relate to the explanatory power of the nstrument n the regresson. A statstc commonly used s the R squared of the frst stage regresson wth the ncluded exogenous varables partalled out. Alternatvely, ths may be expressed as the Ftest of the jont sgnfcance of all the nstruments n the frst stage regresson. The orthogonalty of the nstruments can be tested usng two stage least squares (2SLS) or general method of moment (GMM) technques. If the dsturbance s homoskedastc, the GMM estmator s equvalent to the 2SLS estmator. If t s heteoskedastc, the 2SLS estmator s neffcent but consstent, whereas the standard estmated covarance matrx s nconsstent. Ths paper tests the heteroskedastcty of the dsturbance n wfe s wage equaton usng Whte/Koenker nr2 statstc and PaganHall general statstc. In the context of GMM, the orthogonalty of the nstrument may be tested va the commonly employed Hansen s J statstc. Ths statstc s
14 the value of the GMM objectve functon, evaluated at the effcent GMM estmator. The J statstc s dstrbuted as 2 χ wth degree of freedom equal to the number of overdentfyng restrctons. A rejecton of the null hypothess mples that the nstruments are not satsfyng the orthogonal condtons requred for ther employment. Ths may be ether because they are not truly exogenous, or because they are beng ncorrectly excluded from the regresson. Snce the model s overdentfed, testng a subset of the overdentfyng restrctons s possble. In ths context, the C test allows us to test a subset of the orgnal set of orthogonalty condtons. The statstc s computed as the dfference between two J statstcs: that for the (restrcted, fully effcent) regresson usng the entre set of overdentfyng restrctons, versus that for the (unrestrcted, neffcent but consstent) regresson usng a smaller set of restrctons, n whch a subset of nstruments are removed from the set. The C statstc, dstrbuted 2 χ wth degrees of freedom equal to the loss of overdentfyng restrctons, has the null hypothess that the specfed varables are proper nstruments. Usng the vald nstruments, ths paper then tests whether wfe s HI s exogenous n the wage equaton. Both DurbnWuHausman (DWH) and the WuHausman tests are appled. For the DurbnWuHausman test, a quadratc form n the dfferences between the two coeffcent vectors, the IV estmator whch s fully effcent under the null but nconsstent f the null s not true and the OLS estmator whch s consstent under both the null and the alternatve hypotheses, scaled by the precson matrx, gves rse to a test statstc for the null hypothess that the OLS estmator s consstent and fully effcent. The test statstc s dstrbuted as 2 χ wth the degrees of freedom equal to the number of regressors beng tested for endogenety, whch equals 1 n ths paper. The asymptotcally equvalent WuHausman test s an Ftest of the sgnfcance of the frst sage resduals n the auxlary second stage regresson of 2SLS. One advantage of the WuHausman Fstatstc over the DurbnWu Hausman test s that wth certan normalty assumptons, t s a fnte sample test exactly dstrbuted as F. IV. Data The data used n ths paper come from the 2004 Current Populaton Survey (CPS) March Supplement dataset, and the 2004 CPS January Basc dataset. 13
15 The CPS s a monthly survey of a probablty sample of housng unts each month. The Annual Demographc Survey or March CPS Supplement s the prmary source of detaled nformaton on ncome and work experence n the Unted States. The labor force and work experence data from ths survey are used to profle the U.S. labor market and to make employment projectons. More mportantly, the March CPS Supplement provdes rch data on the ndvdual health nsurance nformaton. CPS has a varable HI, whch ndcates whether or not the ndvdual s covered by a health nsurance plan provded through current/former employer/unon. However, the 2004 March Supplement does not nclude data on tenure:.e., how long the worker has been n the current job; only the 2004 January Basc dataset has ths nformaton. So, ths paper frst merges these two data sets together. Ths step drops a lot of observatons, due to the structure of the CPS. 2 Then, marred ndvduals are chosen to match the husband and wfe. From ths couple dataset, the observatons are further restrcted to the followng: ndvduals aged 2560, employed full tme, n the prvate sector, whose man earnngs are wages and salary. The observatons wth the ncomplete nformaton 3 are dropped, as well as ndvduals wth hourly wages less than two dollars. After these restrctons, the sample sze s 1287 couples. The followng are the bref ntroducton of the varables used n ths paper. The logarthm of current man job hourly wage s the dependent varable. Ths paper chooses the current man job earnngs nstead of total earnng because currently a lot of part tme ndvduals work several jobs, but the nsurance coverage manly comes from the man job. The CPS only provdes the total wages and the earnngs from the other jobs. So, subtractng the latter from the former yelds the wages from the man job. The 2 The CPS s a monthly survey of a probablty sample of housng unts each month. It does not, however, survey a completely new set of housng unts each month. Rather, the sample s dvded nto eght representatve sub samples called rotaton groups, wth housng unts n each rotaton group beng ntervewed for four consecutve months, followed by an 8month break, and then by another four months of ntervews. Thus, CPS sample housng unts are each elgble for 8 dfferent monthly ntervews, and rotaton groups are referred to n CPS parlance by ther month n sample of MIS. In any gven monthly sample, approxmately oneeghth of sample unts wll be ntervewed for the frst tme (MIS=1), oneeghth for the second tme and so on. One eghth of the sample wll leavng the sample permanently (MIS=8), and oneeghth of the sample wll be leavng for the next eght months before beng rentervewed (MIS=4). These latter two rotaton groups, MIS=8 and MIS=4, are referred to as the outgong rotaton groups. So, 75% of the CPS sample s common from month to month (any consecutve two months); whle 50% of the CPS sample s common from one year to the next for the same month. However, because of oneresponse, mortalty, mgraton and recordng errors, there maybe stll some errors after we match the two surveys usng the household number and ndvdual lne number (whch dentfes the ndvdual n the household). Madran and Lefgren (1999) test the dfferent matchng method and gve us better matchng strategy. Frst match two datasets usng household dentfer (H_IDNUM), ndvdual lne number wthn the household (A_LINENO) and household number (H_HHNUM) whch equals 1 n the ntal ntervew and ncreased by 1 f the household s replaced by other n the next ntervew. After matchng usng H_IDNUM, A_LINENO and H_HHNUM, mposng addtonal merge crtera on gender, race and age. If gender and race dffer n two surveys for the same ndvdual, or f the dfference of age n tme t+1 and tme t s less than 1 or greater than 3, we delete these observatons. 3 Manly ncludes those ndvduals wthout health nsurance status nformaton, wthout tenure nformaton, and wthout unon membershp nformaton. 14
16 CPS also provdes the hours worked per week for the man job and weeks usually worked per year. Thus, we can obtan current man job hourly wage, and use the logarthm of t as the dependent varable. HI status s the target varable, HI = 1 f ndvduals have HI and 0 f not. The varable HI means that the ndvdual has employerprovded health nsurance, and NOHI means not. Race 4, Age, Age square, Educaton, Geographcal locaton, Number of kds under 18 and Husband s yearly earnngs. All of these varables are categorcal varables except age and husband s yearly earnngs, whch are contnuous. Race has three categores: whte, black, and other. Educaton s categorzed nto 7 groups: no dploma, hgh school dploma, some college, bachelor, master, professonal schools ncludng MD, and doctoral degree. The CPS classfes the states nto four regons: Northeast, Mdwest, South and West. The number of kds under 18 s classfed nto two categores: no kds under age 18 and some kds under age 18. Major ndustry 5, Major occupaton, Tenure, Tenure square, Current job frm sze, and Unon membershp. The CPS has 14 major ndustral codes and our dataset covers 13. CPS has 11 major occupaton codes, and ours ncludes 10. The excluded one s armed forces. Tenure nformaton comes from the January Basc dataset, and s calculated as the number of years the ndvdual have been employed n the current job. Frm sze equals one f the number of the employees exceeds 100, zero otherwse. Lastly, unon membershp equals one f the ndvdual s a unon member or the current job s covered by a unon, and zero otherwse. Husband s HI, Husband s frm sze), and husband s unon membershp are potental nstruments for wfe s HI status and are defed smlarly as wfe s varables. Table 1 provdes summery statstcs. From the table we can see that wfe s log hourly wage ranges from 0.80 to 5.93, the mean s Furthermore, t shows that the percentage of wves havng HI s relatvely low (0.59) compared to ths percentage of husbands (0.68). Table 2 provdes the correlaton between wfe s HI status and ts nstruments: husband s HI, husband s frm sze and husband s unon membershp. Ths gves us a rough vew of the relatonshps. From the table we have the followng fndngs. Frst, all three nstruments are negatvely correlated wth wfe s HI. Ths confrms that these three varables are potental nstruments. Second, husband s HI s strongly correlated wth wfe s HI; 4 Race turns out to be nsgnfcant n the wage equaton, so I run two sets of the regresson: wth and wthout race varables. Snce the results are smlar, I only provde the results wthout usng ths varable. 5 Industry also turns out to be nsgnfcant n the wage equaton, so I run two sets of the regresson: wth and wthout major ndustry code. Snce the results are smlar, I only provde the results wthout usng ths varable. 15
17 husband s frm sze and unon membershp are less correlated. Ths s reasonable snce the latter two varables are ndrectly correlated wth wfe s HI: they are correlated wth wfe s HI through the correlaton wth husband s HI. V. Results A. Regresson Results To begn, tests for heteroskedastcty are provded, usng the followng regressors: HI, age, age squared, educaton, geographcal locaton, number of kds under 18, major occupaton, tenure, tenure squared and husband s yearly earnngs. 6 The Whte/Koenker nr2 test statstc s equal to ( p = ). The PaganHall general test statstc 7 s equal to ( p = ). Therefore, the tests reject the null hypothess that the dsturbance s homoskedastc. Thus, n the followng, only the GMM results are presented. 8 The nstrumental relevance and orthogonalty test results are llustrated n Table 3. The results are lsted for dfferent combnatons of the three nstruments: husband s frm sze (Fsze), husband s health nsurance status (HHI), and husband s unon membershp (Unon). In general, unlke Olson (2002), ths paper fnds that husband frm sze and husband HI status are vald nstruments, but husband s unon membershp s not. Specfcally, for the relevance test, the second column of Table 3 s the partal Rsquared. The thrd column reports the Fstatstc; the pvalues are n brackets. We can see that the partal Rsquared s between 0.01 and The most relevant nstrument s husband s health nsurance, for whch the partal Rsquared s 0.07, followed by husband s frm sze whose partal Rsquared s 0.02 and husband s unon status whose partal R squared s The Fstatstcs are large, and all the pvalues equal These results are consstent wth our prevous fndngs n Table 2. For the orthogonalty tests, Table 3 reports the Hansen s J statstcs for GMM model. The pvalues are n the brackets. We can see that the Fsze and HHI combnaton has the lowest J, whch s 0.60 ( p = ). Thus, we do not reject the hypothess that Fsze and HHI are orthogonal to the dsturbance and that they are not n the 6 Industry and race are not statstcally sgnfcant, so they are dropped from the regresson. 7 The nstruments used for ths statstc are husband s frm sze and husband s health nsurance status, vald nstruments known from the followng test results. 8 The author also tests nstrument valdty for the 2SLS model, and fnds that the results do not dffer sgnfcantly. 16
18 regresson equaton. In contrast, models wth nstruments combned wth Unon have very hgh J statstcs; the pvalues are below 0.1. Ths suggests that the hypothess that Unon s orthogonal to the dsturbance should be rejected. Table 3 also reports the Cstatstc 9 of subnstruments tests gven the vald nstruments. However, snce the three nstruments are hghly correlated, we only have the Cstatstcs for Fsze, HHI and Unon gven all the three nstruments. We can see that Fsze has the lowest C, followed by HHI and Unon. For example, the C statstc for Fsze s 0.41 ( p = ); for HHI the C statstcs s 1.31 ( p = ); for Unon t s 5.57 ( p = ). Therefore, the Cstatstcs also suggest that Fsze and HHI pass the orthogonalty test, but not Unon. Moreover, Table 3 lsts the HI coeffcents of the IV regressons. The coeffcents are negatve, whch means that there are tradeoffs between employerprovded health nsurance and wages. However, they are not statstcally sgnfcant, except when usng Unon as nstrument. After testng the valdty of the nstruments, Table 4 reports the endogenety test results usng the vald nstruments. The WuHausman test statstcs for Fsze, HHI and both IVs are 2.19, 1.69 and 2.57 respectvely; the pvalues are 0.14, 0.19 and 0.11 respectvely. The DurbnWuHausman tests generate smlar results. Thus, we cannot strongly reject or accept the null hypothess that the HI varable s exogenous, snce the related p values are all between 0.10 and 0.20, the margn of rejecton and acceptance 10. So, n the followng, results assumng exogenety and endogenety are provded. Fnally, OLS and IV regresson results are provded n Table 5, comparng the regressons wth and wthout the nteractons. For the regressons wthout nteractons, column 2 of Table 5 s the HI coeffcent of the OLS regresson, wth the pvalue n brackets. Column 3 reports the HI coeffcent of the GMM model, wth the pvalue n brackets. For the regresson wth nteracton terms, column 4 and 6 reports the OLS and the GMM 11 coeffcents respectvely, wth pvalues n brackets; column 5 and 7 are the Fstatstcs of all the nteracton terms. In the regressons wth nteractons, not only s HI nstrumented, but also all the nteractons, snce they 9 The C statstc for a sngle nstrument s 0, snce there are no overdentfyng restrctons for one nstrument of one endogenous varable. 10 The author also test the Hausman test, plus Durbn s flavor of Hausman test and Wu s flavor of Hausman test. However, for these Hausman tests, the null hypothess s that the dfferences of the coeffcents of IV and OLS regresson are not symmetrc. So, strctly speakng, these are not endogenety tests, but rather tests of the dfference of two dfferent regresson methods: OLS vs. IV. So, although the tests suggest that the dfferences are not systematc (all the pvalue are 1), we stll cannot say that HI s exogenous. 11 The author also estmates 2SLS models wth the frst stage lnear, probt and logt respectvely, but they do not generate fundamentally dfferent results. 17
19 are also endogenous. Rows n Table 5 are for OLS model, the model wth Fsze as nstrument, and the model wth HHI as nstrument respectvely. From the table, we have the followng fndngs. Frst, there s large dfference of the HI coeffcents between whether or not we nclude the ndrect effect of HI. For example, for OLS regressons, the coeffcent on HI excludng the nteractons s 0.09, whle wth the nteractons, t s Though t s not statstcally sgnfcant, the sgn has changed from postve to negatve. For the IV regresson, the HI coeffcent s much bgger n absolute value for the model wth nteractons than the model wthout. For example, HI coeffcents n the IV regresson wth the Fsze as the nstrument are and respectvely for the nonnteracton and nteracton regresson. They are and for HHI as the nstrument. Second, for the regressons ncludng the ndrect effect of HI, the coeffcents on the nteractons are statstcally dfferent from zero for some models. For example, n the IV model wth HHI as the nstrument, the coeffcents on nteracton terms are statstcally dfferent from zero. Therefore, HI affects wages not only from the drect tradeoff pont of vew, but also from changng the returns of the ndvdual characterstcs. Thrd, although we have the expected sgn of the coeffcent on HI, they are not statstcally sgnfcant. Only the HI coeffcent of the OLS regresson wthout nteracton terms s statstcally sgnfcant at 10% level. All the other coeffcents are not statstcally sgnfcant. These results lead us to do the dstrbutonal analyss. B. Dstrbutonal Analyss Results Ths part reports the dstrbutonal analyss results when HI s exogenous and when t s endogenous. Fgures 24 provde the CDF comparsons n graph. Fgure 2 s the CDF comparsons assumng that HI s exogenous n wfe s wage equaton, and therefore the ndvduals are dvded nto HI and NOHI groups by the HI varable. Fgure 3 and 4 are CDF comparsons when wfe s HI s endogenous n her wage equaton. Here, ndvduals are dvded nto HI and NOHI groups by bnary nstruments. Fgure 3 s the results when usng husband s frm sze as the nstrument. Fgure 4 s the results when usng husband health nsurance as the nstrument. For each fgure, graphs n the frst column are CDFs; graphs n the second column are wage dfferentals of CDFs at dfferent quantles; graphs n the thrd column are accumulated CDFs. The graphs n the frst row are for total wages; graphs n the second row are for the partal wages, whch reflect the drect effect of HI; and graphs n the thrd 18
20 row are for the predcted wages explaned by ndvdual characterstcs ( for ndvduals wth NOHI usng the estmated coeffcents for ndvduals wth HI. X βˆ ), as well as the hypothetcal wages Correspondng to these graphs, Table 6 also provdes the quanttatve levels of these CDFs and dfferentals at the mean and at quantles 10, 25, 50, 75 and 90. The three rows are results for the followng dfferent models respectvely: wthout nstrument, husband s frm sze as the nstrument and husband s health nsurance status as the nstrument. The three columns are results of total wage and wage dfferentals, wages explaned by the constant and resduals as well as wage dfferentals, and the wages explaned by observable characterstcs ( X βˆ ) and Oaxaca decomposton results. Table 7 provdes the stochastc domnance test results correspondng to the graphs n Fgures 24. The three rows are results for the followng dfferent models respectvely: wthout nstruments, husband s frm sze as nstrument and husband s health nsurance status as nstrument. For each model, t also provdes n detal the stochastc domnance test results for the total wage, for the partal wage, and the wage explaned by the characterstc varables, whch s subcategorzed as X βˆ gap, X gap and the βˆ gap. The X βˆ gap stands for the total gap explaned by the characterstc varables; the X gap s the wage gap solely due to the characterstc dfferences; βˆ gap s the gap due to the compensaton dfference, the ndrect effect of HI. FSD and SSD test results are based on 1000 bootstrap repettons. are provded for FSD test; d 1 MAX, d 2 MAX, d, Pr( d * 0 1 ), Pr( d * 0), Pr( * ) 2 d 0 statstcs s 1 MAX, s 2 MAX, s, Pr( s * 0 1 ), Pr( s * 0), Pr( * ) 2 s 0 statstcs are provded for SSD test. The subscrpt 1 stands for the HI mnus NOHI; the subscrpt 2 stands for NOHI mnus HI. The superscrpt stands for the results for each bootstrap. Furthermore, the larger Pr( d * 0) s, the more possble that there s FSD; and the larger Pr( s * 0) s, the more possble that there s SSD. The followng are fndngs from the graphs n Fgure 24 and Tables 67. B. 1. Total Wages From the graphs of the frst row n Fgures 24 we can see that the CDFs of total wages do not reflect the tradeoff f we do not use an nstrument. In fact, the CDFs of HI even domnate the CDFS of NOHI. Usng the IV 19
21 approach, the stuaton reverses. The frst column n Table 6 clearly mples ths. The total mean wage dfference s 0.26 when not usng nstrument, but ends up wth and 0.07, respectvely, when usng husband s frm sze and health nsurance as nstruments. Table 7 lsts the stochastc domnance results. Except NOHI second order domnates HI when usng husband s frm sze as nstrument, the other relatonshps usng nstruments are not statstcally sgnfcant. Ths fndng explans why the prevous researchers cannot fnd the tradeoffs when they do not use proper nstruments. B. 2. Partal Wages From the second row graphs of partal wages n Fgures 24, we can see that there exsts a tradeoff between employerprovded health nsurance and wages, whch s the drect effect of HI on wages. When assumng the exogenety of HI, the tradeoff s very small. From Table 6 we can see that ths mean wage dfference explaned by constant and resduals s 0.02 n absolute value. When usng husband s frm sze as nstrument, the dfference s 0.24 n absolute value. When usng husband s health nsurance status as nstrument, the dfference s 0.51 n absolute value. However, these numbers depend on the magntude of K, so they do not reflect the true tradeoff. Nevertheless, ther sgns ndcate that there s a tradeoff between employerprovded health nsurance and wages for complers. The stochastc domnance test confrms the above drect fndngs from the graph. Table 7 suggests that for the partal wages, there s no frst or second order stochastc domnance for HI and NOHI groups when not usng any nstrument. For example, Pr( d * 0) s and Pr( s * 0) s When usng the nstruments, the NOHI not only second order stochastc domnates HI, but also frst order domnates. For example, Pr( d * 0) s and Pr( s * 0) s 0.993, and d 2 s when usng husband s frm sze as nstrument. When usng MAX husband s health nsurance status as nstrument, these three statstcs are 0.999, and respectvely. Table 6 also suggests that the total wages are explaned half by the characterstc varables and half by the constant and resduals. Ths s consstent wth the lterature, snce the Rsquared for wage equatons are at most 0.5. B. 3. Characterstc / Compensaton Gaps 20
22 For the wage explaned by the characterstc varables, the compensaton gap whch reflects the ndrect effect of HI s postve and much larger than the characterstc gap. From the thrd row graphs n Fgures 24, we can see that the hypothetcal CDFs of NOHI group usng HI regresson coeffcents are very close to the CDFs of the HI group, and far away from the CDFs of NOHI group. The last three columns of Table 6 llustrate ths quanttatvely. Except the one wthout nstrument, the numbers n W / postve and almost equal to the numbers n However, the numbers n the W / h noh h noh h noh column (compensaton gap) are Wh noh column (total gap); sometmes they are even larger. column (characterstc gap) are much smaller. Ths fndng tells us that the characterstc dfferences for HI and NOHI ndvduals are not that bg, at least for the characterstcs we can observe. For example, when usng husband s frm sze as nstrument, the mean wage dfference due to the compensaton dfference s 0.20, whle the total gap s 0.09 and gap due to the characterstc dfference s Whle for husband s health nsurance status as nstrument, the mean wage dfference due to compensaton dfference s 0.50 and the total gap s 0.47 and the gap due to the characterstc dfference s The test results n Table 7 confrm ths fndng. We can see from Table 7 that there are FSDs and SSDs for the βˆ gap, the compensaton gap or the ndrect effect of HI, but no obvous FSDs for the X gap,.e. the characterstc gap, when usng nstruments. For example, when usng husband s frm sze as nstrument, Pr( d * 0) for βˆ gap s 0.995, Pr( s * 0) s and d 1 s equal to However, these three statstcs MAX for X gap are respectvely 0.384, and Wth husband s health nsurance status as nstrument, the three statstcs for βˆ gap are 0.999, and respectvely. They are respectvely 0.044, and for X gap. Ths fndng means that the wage returns are hgher for ndvduals wth employerprovded health nsurance than those ndvduals wthout. In other words, the characterstcs are not too dfferent for HI and NOHI groups, at least for the characterstcs we can observe, but the wage return of the same characterstcs s very dfferent for the two groups. The ndvduals wth employerproved health nsurance not only get ther health nsurance from ther companes, they also get hgher pay. One possble explanaton s that HI mproves ndvdual s productvty through nvestment n health; another explanaton s that employees morale s hgher 21
Can Auto Liability Insurance Purchases Signal Risk Attitude?
Internatonal Journal of Busness and Economcs, 2011, Vol. 10, No. 2, 159164 Can Auto Lablty Insurance Purchases Sgnal Rsk Atttude? ChuShu L Department of Internatonal Busness, Asa Unversty, Tawan ShengChang
More informationbenefit 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 informationPSYCHOLOGICAL RESEARCH (PYC 304C) Lecture 12
14 The Chsquared dstrbuton PSYCHOLOGICAL RESEARCH (PYC 304C) 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 informationStudy on CET4 Marks in China s Graded English Teaching
Study on CET4 Marks n Chna s Graded Englsh Teachng CHE We College of Foregn Studes, Shandong Insttute of Busness and Technology, P.R.Chna, 264005 Abstract: Ths paper deploys Logt model, and decomposes
More informationAn 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 informationThe covariance is the two variable analog to the variance. The formula for the covariance between two variables is
Regresson Lectures So far we have talked only about statstcs that descrbe one varable. What we are gong to be dscussng for much of the remander of the course s relatonshps between two or more varables.
More informationPRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIGIOUS AFFILIATION AND PARTICIPATION
PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIIOUS AFFILIATION AND PARTICIPATION Danny CohenZada Department of Economcs, Benuron Unversty, BeerSheva 84105, Israel Wllam Sander Department of Economcs, DePaul
More informationMultivariate EWMA Control Chart
Multvarate EWMA Control Chart Summary The Multvarate EWMA Control Chart procedure creates control charts for two or more numerc varables. Examnng the varables n a multvarate sense s extremely mportant
More informationSIMPLE 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.
More informationThe Analysis of Covariance. ERSH 8310 Keppel and Wickens Chapter 15
The Analyss of Covarance ERSH 830 Keppel and Wckens Chapter 5 Today s Class Intal Consderatons Covarance and Lnear Regresson The Lnear Regresson Equaton TheAnalyss of Covarance Assumptons Underlyng the
More informationThe Probit Model. Alexander Spermann. SoSe 2009
The Probt Model Aleander Spermann Unversty of Freburg SoSe 009 Course outlne. Notaton and statstcal foundatons. Introducton to the Probt model 3. Applcaton 4. Coeffcents and margnal effects 5. Goodnessofft
More informationAnswer: 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 MultpleChoce Questons (30/00 ponts) Please, crcle the correct answer for each of the followng 0 multplechoce questons. For each queston, only one of the answers s correct.
More informationQuestions that we may have about the variables
Antono Olmos, 01 Multple Regresson Problem: we want to determne the effect of Desre for control, Famly support, Number of frends, and Score on the BDI test on Perceved Support of Latno women. Dependent
More informationMarginal 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 ramlee@fpe.ups.edu.my
More informationThe 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 informationHOUSEHOLDS 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
More informationCHAPTER 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 informationInequality and The Accounting Period. Quentin Wodon and Shlomo Yitzhaki. World Bank and Hebrew University. September 2001.
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 informationIntroduction to Regression
Introducton to Regresson Regresson a means of predctng a dependent varable based one or more ndependent varables. Ths s done by fttng a lne or surface to the data ponts that mnmzes the total error. 
More information9.1 The Cumulative Sum Control Chart
Learnng Objectves 9.1 The Cumulatve Sum Control Chart 9.1.1 Basc Prncples: Cusum Control Chart for Montorng the Process Mean If s the target for the process mean, then the cumulatve sum control chart s
More informationExhaustive Regression. An Exploration of RegressionBased Data Mining Techniques Using Super Computation
Exhaustve Regresson An Exploraton of RegressonBased Data Mnng Technques Usng Super Computaton Antony Daves, Ph.D. Assocate Professor of Economcs Duquesne Unversty Pttsburgh, PA 58 Research Fellow The
More informationDEFINING %COMPLETE IN MICROSOFT PROJECT
CelersSystems DEFINING %COMPLETE IN MICROSOFT PROJECT PREPARED BY James E Aksel, PMP, PMISP, MVP For Addtonal Informaton about Earned Value Management Systems and reportng, please contact: CelersSystems,
More information! # %& ( ) +,../ 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 17498368 Sarah Brown, Aurora OrtzNúñez and Karl Taylor Educatonal loans and
More informationCausal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting
Causal, Explanatory Forecastng Assumes causeandeffect relatonshp between system nputs and ts output Forecastng wth Regresson Analyss Rchard S. Barr Inputs System Cause + Effect Relatonshp The job of
More informationStaff Paper. Farm Savings Accounts: Examining Income Variability, Eligibility, and Benefits. Brent Gloy, Eddy LaDue, and Charles Cuykendall
SP 200502 August 2005 Staff Paper Department of Appled Economcs and Management Cornell Unversty, Ithaca, New York 148537801 USA Farm Savngs Accounts: Examnng Income Varablty, Elgblty, and Benefts Brent
More informationAnalysis 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 informationInstitute 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 informationGender 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
More informationH 1 : at least one is not zero
Chapter 6 More Multple Regresson Model The Ftest Jont Hypothess Tests Consder the lnear regresson equaton: () y = β + βx + βx + β4x4 + e for =,,..., N The tstatstc gve a test of sgnfcance of an ndvdual
More informationForecasting 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 informationx f(x) 1 0.25 1 0.75 x 1 0 1 1 0.04 0.01 0.20 1 0.12 0.03 0.60
BIVARIATE DISTRIBUTIONS Let be a varable that assumes the values { 1,,..., n }. Then, a functon that epresses the relatve frequenc of these values s called a unvarate frequenc functon. It must be true
More informationTHE 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 informationMarginal Benefit Incidence Analysis Using a Single Crosssection of Data. Mohamed Ihsan Ajwad and Quentin Wodon 1. World Bank.
Margnal Beneft Incdence Analyss Usng a Sngle Crosssecton 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 informationHYPOTHESIS TESTING OF PARAMETERS FOR ORDINARY LINEAR CIRCULAR REGRESSION
HYPOTHESIS TESTING OF PARAMETERS FOR ORDINARY LINEAR CIRCULAR REGRESSION Abdul Ghapor Hussn Centre for Foundaton Studes n Scence Unversty of Malaya 563 KUALA LUMPUR Emal: ghapor@umedumy Abstract Ths paper
More informationWhat 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 informationTrafficlight a stress test for life insurance provisions
MEMORANDUM Date 006097 Authors Bengt von Bahr, Göran Ronge Traffclght a stress test for lfe nsurance provsons Fnansnspetonen P.O. Box 6750 SE113 85 Stocholm [Sveavägen 167] Tel +46 8 787 80 00 Fax
More informationCHOLESTEROL 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 informationDoes 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
More informationHeterogeneous 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
More informationStatistical 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 informationReturns 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
More informationCalculation 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 twostage stratfed cluster desgn. 1 The frst stage conssted of a sample
More information8.5 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 informationCHAPTER 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 informationManagement Quality, Financial and Investment Policies, and. Asymmetric Information
Management Qualty, Fnancal and Investment Polces, and Asymmetrc Informaton Thomas J. Chemmanur * Imants Paegls ** and Karen Smonyan *** Current verson: December 2007 * Professor of Fnance, Carroll School
More informationChapter 8 Groupbased Lending and Adverse Selection: A Study on Risk Behavior and Group Formation 1
Chapter 8 Groupbased 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
More informationPROFIT RATIO AND MARKET STRUCTURE
POFIT ATIO AND MAKET STUCTUE By Yong Yun Introducton: Industral economsts followng from Mason and Ban have run nnumerable tests of the relaton between varous market structural varables and varous dmensons
More informationAn Analysis of Factors Influencing the SelfRated Health of Elderly Chinese People
Open Journal of Socal Scences, 205, 3, 520 Publshed Onlne May 205 n ScRes. http://www.scrp.org/ournal/ss http://dx.do.org/0.4236/ss.205.35003 An Analyss of Factors Influencng the SelfRated Health of
More informationSTAMP 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
More informationDeterminants of employmentbased private health insurance coverage in Denmark
Nordc Journal of Health Economcs Onlne ISSN: 18929710 Determnants of employmentbased prvate health nsurance coverage n Denmark ASTRID KIIL* k@sam.sdu.dk Unversty of Southern Denmark Abstract: Ths study
More informationTwo Faces of IntraIndustry Information Transfers: Evidence from Management Earnings and Revenue Forecasts
Two Faces of IntraIndustry Informaton Transfers: Evdence from Management Earnngs and Revenue Forecasts Yongtae Km Leavey School of Busness Santa Clara Unversty Santa Clara, CA 950530380 TEL: (408) 5544667,
More informationQuality Adjustment of Secondhand Motor Vehicle Application of Hedonic Approach in Hong Kong s Consumer Price Index
Qualty Adustment of Secondhand Motor Vehcle Applcaton of Hedonc Approach n Hong Kong s Consumer Prce Index Prepared for the 14 th Meetng of the Ottawa Group on Prce Indces 20 22 May 2015, Tokyo, Japan
More informationSTATISTICAL DATA ANALYSIS IN EXCEL
Mcroarray Center STATISTICAL DATA ANALYSIS IN EXCEL Lecture 6 Some Advanced Topcs Dr. Petr Nazarov 1401013 petr.nazarov@crpsante.lu Statstcal data analyss n Ecel. 6. Some advanced topcs Correcton for
More informationADVERSE 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, PerreAndre
More informationCambodian Child s Wage Rate, Human Capital and Hours Worked Tradeoff: Simple Theoretical and Empirical Evidence for Policy Implications
GSIS Workng Paper Seres ambodan hld s Wage Rate, Human aptal and Hours Worked Tradeoff: Smple Theoretcal and Emprcal Evdence for Polcy Implcatons Han PHOUMIN Sech FUKUI No. 6 August 2006 Graduate School
More informationAre 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
More informationLecture 10: Linear Regression Approach, Assumptions and Diagnostics
Approach to Modelng I Lecture 1: Lnear Regresson Approach, Assumptons and Dagnostcs Sandy Eckel seckel@jhsph.edu 8 May 8 General approach for most statstcal modelng: Defne the populaton of nterest State
More information14.74 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 information1. 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 informationSection 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 informationNumber 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
More informationThe Effects of Tax Rate Changes on Tax Bases and the Marginal Cost of Public Funds for Canadian Provincial Governments
The Effects of Tax Rate Changes on Tax Bases and the Margnal Cost of Publc Funds for Canadan Provncal Governments Bev Dahlby a and Ergete Ferede b a Department of Economcs, Unversty of Alberta, Edmonton,
More informationSolutions to First Midterm
rofessor Chrstano Economcs 3, Wnter 2004 Solutons to Frst Mdterm. Multple Choce. 2. (a) v. (b). (c) v. (d) v. (e). (f). (g) v. (a) The goods market s n equlbrum when total demand equals total producton,.e.
More informationManagement 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
More informationEvaluating the Effects of FUNDEF on Wages and Test Scores in Brazil *
Evaluatng the Effects of FUNDEF on Wages and Test Scores n Brazl * Naérco MenezesFlho Elane Pazello Unversty of São Paulo Abstract In ths paper we nvestgate the effects of the 1998 reform n the fundng
More informationThe Racial and Gender Interest Rate Gap. in Small Business Lending: Improved Estimates Using Matching Methods*
The Racal and Gender Interest Rate Gap n Small Busness Lendng: Improved Estmates Usng Matchng Methods* Yue Hu and Long Lu Department of Economcs Unversty of Texas at San Antono Jan Ondrch and John Ynger
More informationUnderstanding the Impact of Marketing Actions in Traditional Channels on the Internet: Evidence from a Large Scale Field Experiment
A research and educaton ntatve at the MT Sloan School of Management Understandng the mpact of Marketng Actons n Tradtonal Channels on the nternet: Evdence from a Large Scale Feld Experment Paper 216 Erc
More informationNasdaq Iceland Bond Indices 01 April 2015
Nasdaq Iceland Bond Indces 01 Aprl 2015 Fxed duraton Indces Introducton Nasdaq Iceland (the Exchange) began calculatng ts current bond ndces n the begnnng of 2005. They were a response to recent changes
More informationThe Development of Web Log Mining Based on ImproveKMeans Clustering Analysis
The Development of Web Log Mnng Based on ImproveKMeans Clusterng Analyss TngZhong Wang * College of Informaton Technology, Luoyang Normal Unversty, Luoyang, 471022, Chna wangtngzhong2@sna.cn Abstract.
More informationCHAPTER 7 THE TWOVARIABLE REGRESSION MODEL: HYPOTHESIS TESTING
CHAPTER 7 THE TWOVARIABLE REGRESSION MODEL: HYPOTHESIS TESTING QUESTIONS 7.1. (a) In the regresson contet, the method of least squares estmates the regresson parameters n such a way that the sum of the
More information5 Multiple regression analysis with qualitative information
5 Multple regresson analyss wth qualtatve nformaton Ezequel Urel Unversty of Valenca Verson: 913 5.1 Introducton of qualtatve nformaton n econometrc models. 1 5. A sngle dummy ndependent varable 5.3 Multple
More informationTransition 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
More informationThe Application of Fractional Brownian Motion in Option Pricing
Vol. 0, No. (05), pp. 738 http://dx.do.org/0.457/jmue.05.0..6 The Applcaton of Fractonal Brownan Moton n Opton Prcng Qngxn Zhou School of Basc Scence,arbn Unversty of Commerce,arbn zhouqngxn98@6.com
More informationA Novel Methodology of Working Capital Management for Large. Public Constructions by Using Fuzzy Scurve Regression
Novel Methodology of Workng Captal Management for Large Publc Constructons by Usng Fuzzy Scurve Regresson ChengWu Chen, Morrs H. L. Wang and TngYa Hseh Department of Cvl Engneerng, Natonal Central Unversty,
More informationA Multistage Model of Loans and the Role of Relationships
A Multstage Model of Loans and the Role of Relatonshps Sugato Chakravarty, Purdue Unversty, and Tansel Ylmazer, Purdue Unversty Abstract The goal of ths paper s to further our understandng of how relatonshps
More informationCapital asset pricing model, arbitrage pricing theory and portfolio management
Captal asset prcng model, arbtrage prcng theory and portfolo management Vnod Kothar The captal asset prcng model (CAPM) s great n terms of ts understandng of rsk decomposton of rsk nto securtyspecfc rsk
More informationDescriptive Statistics (60 points)
Economcs 30330: Statstcs for Economcs Problem Set 2 Unversty of otre Dame Instructor: Julo Garín Sprng 2012 Descrptve Statstcs (60 ponts) 1. Followng a recent government shutdown, Mnnesota Governor Mark
More informationRiskbased Fatigue Estimate of Deep Water Risers  Course Project for EM388F: Fracture Mechanics, Spring 2008
Rskbased 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 informationCovariatebased pricing of automobile insurance
Insurance Markets and Companes: Analyses and Actuaral Computatons, Volume 1, Issue 2, 2010 José Antono Ordaz (Span), María del Carmen Melgar (Span) Covaratebased prcng of automoble nsurance Abstract Ths
More informationTraditional versus Online Courses, Efforts, and Learning Performance
Tradtonal versus Onlne Courses, Efforts, and Learnng Performance KuangCheng Tseng, Department of Internatonal Trade, ChungYuan Chrstan Unversty, Tawan ShanYng Chu, Department of Internatonal Trade,
More informationEvaluating credit risk models: A critique and a new proposal
Evaluatng credt rsk models: A crtque and a new proposal Hergen Frerchs* Gunter Löffler Unversty of Frankfurt (Man) February 14, 2001 Abstract Evaluatng the qualty of credt portfolo rsk models s an mportant
More informationIDENTIFICATION 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 informationAn 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 RmmKaufman, RmmKaufman
More informationErrorPropagation.nb 1. Error Propagation
ErrorPropagaton.nb Error Propagaton Suppose that we make observatons of a quantty x that s subject to random fluctuatons or measurement errors. Our best estmate of the true value for ths quantty s then
More informationRisk Model of LongTerm Production Scheduling in Open Pit Gold Mining
Rsk Model of LongTerm Producton Schedulng n Open Pt Gold Mnng R Halatchev 1 and P Lever 2 ABSTRACT Open pt gold mnng s an mportant sector of the Australan mnng ndustry. It uses large amounts of nvestments,
More informationRecurrence. 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 informationA Probabilistic Theory of Coherence
A Probablstc Theory of Coherence BRANDEN FITELSON. The Coherence Measure C Let E be a set of n propostons E,..., E n. We seek a probablstc measure C(E) of the degree of coherence of E. Intutvely, we want
More informationFinancial 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 householdlevel data provded by the Postal Servces
More informationBrigid Mullany, Ph.D University of North Carolina, Charlotte
Evaluaton And Comparson Of The Dfferent Standards Used To Defne The Postonal Accuracy And Repeatablty Of Numercally Controlled Machnng Center Axes Brgd Mullany, Ph.D Unversty of North Carolna, Charlotte
More informationCommunication Networks II Contents
8 / 1  Communcaton Networs II (Görg)  www.comnets.unbremen.de Communcaton Networs II Contents 1 Fundamentals of probablty theory 2 Traffc n communcaton networs 3 Stochastc & Marovan Processes (SP
More informationWORKING PAPER SERIES TAKING STOCK: MONETARY POLICY TRANSMISSION TO EQUITY MARKETS NO. 354 / MAY 2004. by Michael Ehrmann and Marcel Fratzscher
WORKING PAPER SERIES NO. 354 / MAY 2004 TAKING STOCK: MONETARY POLICY TRANSMISSION TO EQUITY MARKETS by Mchael Ehrmann and Marcel Fratzscher WORKING PAPER SERIES NO. 354 / MAY 2004 TAKING STOCK: MONETARY
More informationHow 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 informationLuby s Alg. for Maximal Independent Sets using Pairwise Independence
Lecture Notes for Randomzed Algorthms Luby s Alg. for Maxmal Independent Sets usng Parwse Independence Last Updated by Erc Vgoda on February, 006 8. Maxmal Independent Sets For a graph G = (V, E), an ndependent
More informationBeating the Odds: Arbitrage and Wining Strategies in the Football Betting Market
Beatng the Odds: Arbtrage and Wnng Strateges n the Football Bettng Market NIKOLAOS VLASTAKIS, GEORGE DOTSIS and RAPHAEL N. MARKELLOS* ABSTRACT We examne the potental for generatng postve returns from wagerng
More informationTHE EFFECT OF PREPAYMENT PENALTIES ON THE PRICING OF SUBPRIME MORTGAGES
THE EFFECT OF PREPAYMENT PENALTIES ON THE PRICING OF SUBPRIME MORTGAGES Gregory Ellehausen, Fnancal Servces Research Program George Washngton Unversty Mchael E. Staten, Fnancal Servces Research Program
More informationMultiplePeriod Attribution: Residuals and Compounding
MultplePerod 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 informationSearching 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 informationStress test for measuring insurance risks in nonlife insurance
PROMEMORIA Datum June 01 Fnansnspektonen Författare Bengt von Bahr, Younes Elonq and Erk Elvers Stress test for measurng nsurance rsks n nonlfe nsurance Summary Ths memo descrbes stress testng of nsurance
More informationUNIVERSITA CATTOLICA DEL SACRO CUORE  Milano  QUADERNI DELL ISTITUTO DI ECONOMIA DELL IMPRESA E DEL LAVORO
UNIVERSITA CATTOLICA DEL SACRO CUORE  Mlano  QUADERNI DELL ISTITUTO DI ECONOMIA DELL IMPRESA E DEL LAVORO The Wage Effect of Workng n the Publc Sector When Educaton and Sector Choces Are Endogenous:
More informationADVERSE 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 Revsed May 2003 ABSTRACT In ths paper, we nvestgate
More information