Statistical Methods to Develop Rating Models

Size: px
Start display at page:

Download "Statistical Methods to Develop Rating Models"

Transcription

1 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 Stress Testng Bernd Engelmann, Robert Rauhmeer, Robert (Edtors), 376 p., Sprnger Verlag, Berln 2006.

2 I. Statstcal Methods to Develop Ratng Models Evelyn Hayden and Danel Porath Österrechsche Natonalbank 1 and Unversty of Appled Scences at Manz 1. Introducton The Internal Ratng Based Approach (IRBA) of the New Basel Captal Accord allows banks to use ther own ratng models for the estmaton of probabltes of default (PD) as long as the systems meet specfed mnmum requrements. Statstcal theory offers a varety of methods for buldng and estmaton ratng models. Ths chapter gves an overvew of these methods. The overvew s focused on statstcal methods and ncludes parametrc models lke lnear regresson analyss, dscrmnant analyss, bnary response analyss, tme-dscrete panel methods, hazard models and nonparametrc models lke neural networks and decson trees. We also hghlght the benefts and the drawbacks of the varous approaches. We conclude by nterpretng the models n lght of the mnmum requrements of the IRBA. 2. Statstcal Methods for Rsk Classfcaton In the followng we defne statstcal models as the class of approach whch uses econometrc methods to classfy borrowers accordng to ther rsk. Statstcal ratng systems prmarly nvolve a search for explanatory varables whch provde as sound and relable a forecast of the deteroraton of a borrower's stuaton as possble. In contrast, structural models explan the threats to a borrower based on an economc model and thus use clear causal connectons nstead of the mere correlaton of varables. The followng sectons offer an overvew of parametrc and nonparametrc models generally consdered for statstcal rsk assessment. Furthermore, we dscuss the advantages and dsadvantages of each approach. Many of the methods are descrbed n more detal n standard econometrc textbooks, lke Greene (2003). 1 The opnons expressed n ths chapter are those of the author and do not necessarly reflect vews of the Österreschsche Natonalbank.

3 2 Evelyn Hayden and Danel Porath In general, a statstcal model may be descrbed as follows: As a startng pont, every statstcal model uses the borrower s characterstc ndcators and (possbly) macroeconomc varables whch were collected hstorcally and are avalable for defaultng (or troubled) and non-defaultng borrowers. Let the borrower s characterstcs be defned by a vector of n separate varables (also called covarates) x = x 1,...,x n observed at tme t - L. The state of default s ndcated by a bnary performance varable y observed at tme t. The varable y s defned as y = 1 for a default and y = 0 for a non-default. The sample of borrowers now ncludes a number of ndvduals or frms that defaulted n the past, whle (typcally) the majorty dd not default. Dependng on the statstcal applcaton of ths data, a varety of methods can be used to predct the performance. A common feature of the methods s that they estmate the correlaton between the borrowers characterstcs and the state of default n the past and use ths nformaton to buld a forecastng model. The forecastng model s desgned to assess the credtworthness of borrowers wth unknown performance. Ths can be done by nputtng the characterstcs x nto the model. The output of the model s the estmated performance. The tme lag L between x and y determnes the forecast horzon. 3. Regresson Analyss As a startng pont we consder the classcal regresson model. The regresson model establshes a lnear relatonshp between the borrowers characterstcs and the default varable: y = β' x + u (1) Agan, y ndcates whether borrower has defaulted (y = 1) or not (y = 0). In perod t, x s a column vector of the borrowers characterstcs observed n perod t L and β s a column vector of parameters whch capture the mpact of a change n the characterstcs on the default varable. Fnally, u s the resdual varable whch contans the varaton not captured by the characterstcs x. The standard procedure s to estmate (1) wth the ordnary least squares (OLS) estmators of β whch n the followng are denoted by b. The estmated result s the borrower s score S. Ths can be calculated by S y ( x ) = b x = E '. (2) Equaton (2) shows that a borrower s score represents the expected value of the performance varable when hs or her ndvdual characterstcs are known. The score can be calculated by nputtng the values for the borrower s characterstcs nto the lnear functon gven n (2).

4 I. Statstcal Methods to Develop Ratng Models 3 Note that S s contnuous (whle y s a bnary varable), hence the output of the model wll generally be dfferent from 0 or 1. In addton, the predcton can take on values larger than 1 or smaller than 0. As a consequence, the outcome of the model cannot be nterpreted as a probablty level. However, the score S, can be used for the purpose of comparson between dfferent borrowers, where hgher values of S correlate wth a hgher default rsk. The benefts and drawbacks from model (1) and (2) are the followng: OLS estmators are well-known and easly avalable. The forecastng model s a lnear model and therefore easy to compute and to understand. The random varable u s heteroscedastc (.e. the varance of u s not constant for all ) snce Var ( u ) = Var( y ) = E( y x ) [ 1 E( y x )] = b' x ( 1 b' x ). (3) As a consequence, the estmaton of β s neffcent and addtonally, the standard errors of the estmated coeffcents b are based. An effcent way to estmate β s to apply the Weghted Least Squares (WLS) estmator. WLS estmaton of β s effcent, but the estmaton of the standard errors of b stll remans based. Ths happens due to the fact that the resduals are not normally dstrbuted as they can only take on the values b x (f the borrower does not default and y therefore equals 0) or (1 b x ) (f the borrower does default and y therefore equals 1). Ths mples that there s no relable way to assess the sgnfcance of the coeffcents b and t remans unknown whether the estmated values represent precse estmatons of sgnfcant relatonshps or whether they are just caused by spurous correlatons. Inputtng characterstcs whch are not sgnfcant nto the model can serously harm the model s stablty when used to predct borrowers rsk for new data. A way to cope wth ths problem s to splt the sample nto two parts, where one part (the tranng sample) s used to estmate the model and the other part (the hold-out sample) s used to valdate the results. The consstency of the results of both samples s then taken as an ndcator for the stablty of the model. The absolute value of S cannot be nterpreted. 4. Dscrmnant Analyss Dscrmnant analyss s a classfcaton technque appled to corporate bankruptces by Altman as early as 1968 (see Altman, 1968). Lnear dscrmnant analyss s based on the estmaton of a lnear dscrmnant functon wth the task of separatng ndvdual groups (n ths case of defaultng and non-defaultng borrowers) accordng to specfc characterstcs. The dscrmnant functon s S = β' x. (4)

5 4 Evelyn Hayden and Danel Porath The Score S s also called the dscrmnant varable. The estmaton of the dscrmnant functon adheres to the followng prncple: Maxmzaton of the spread between the groups (good and bad borrowers) and mnmzaton of the spread wthn ndvdual groups Maxmzaton only determnes the optmal proportons among the coeffcents of the vector β. Usually (but arbtrarly), coeffcents are normalzed by choosng the pooled wthn-group varance to take the value 1. As a consequence, the absolute level of S s arbtrary as well and cannot be nterpreted on a stand-alone bass. As n lnear regresson analyss, S can only be used to compare the predcton for dfferent borrowers ( hgher score, hgher rsk ). Dscrmnant analyss s smlar to the lnear regresson model gven n equatons (1) and (2). In fact, the proportons among the coeffcents of the regresson model are equal to the optmal proporton accordng to the dscrmnant analyss. The dfference between the two methods s a theoretcal one: Whereas n the regresson model the characterstcs are determnstc and the default state s the realzaton of a random varable, for dscrmnant analyss the opposte s true. Here the groups (default or non-default) are determnstc and the characterstcs of the dscrmnant functon are realzatons from a random varable. For practcal use ths dfference s vrtually rrelevant. Therefore, the benefts and drawbacks of dscrmnant analyss are smlar to those of the regresson model: Dscrmnant analyss s a wdely known method wth estmaton algorthms that are easly avalable. Once the coeffcents are estmated, the scores can be calculated n a straghtforward way wth a lnear functon. Snce the characterstcs x are assumed to be realzatons of random varables, the statstcal tests for the sgnfcance of the model and the coeffcents rely on the assumpton of multvarate normalty. Ths s, however, unrealstc for the varables typcally used n ratng models as for example fnancal ratos from the balance-sheet. Hence, the methods for analyzng the stablty of the model and the plausblty of the coeffcents are lmted to a comparson between tranng and hold-out sample. The absolute value of the dscrmnant functon cannot be nterpreted n levels. 5. Logt and Probt Models Logt and probt models are econometrc technques desgned for analyzng bnary dependent varables. There are two alternatve theoretcal foundatons. The latent-varable approach assumes an unobservable (latent) varable y* whch s related to the borrower s characterstcs n the followng way:

6 I. Statstcal Methods to Develop Ratng Models 5 * y = β' x + u (5) Here β, x and u are defned as above. The varable y * s metrcally scaled and trggers the value of the bnary default varable y : * 1 f y > 0 y = (6) 0 otherwse Ths means that the default event sets n when the latent varable exceeds the threshold zero. Therefore, the probablty for the occurrence of the default event equals: P ( y = ) = P( u > β' x ) = 1 F( β' x ) = F( β' x ) 1. (7) Here F(.) denotes the (unknown) dstrbuton functon. The last step n (7) assumes that the dstrbuton functon has a symmetrc densty around zero. The choce of the dstrbuton functon F(.) depends on the dstrbutonal assumptons about the resduals (u ). If a normal dstrbuton s assumed, we are faced wth the probt model: β' x t 2 1 F( β' x ) = e 2 dt 2π (8) If nstead the resduals are assumed to follow a logstc dstrbuton, the result s the logt model: β' x e F( β' x ) = (9) β' x 1+ e The second way to motvate logt and probt models starts from the am of estmatng default probabltes. For sngle borrowers, default probabltes cannot be observed as realzatons of default probabltes. However, for groups of borrowers the observed default frequences can be nterpreted as default probabltes. As a startng pont consder the OLS estmaton of the followng regresson: p = b' x + u (10) In (10) the ndex denotes the group formed by a number of ndvduals, p s the default frequency observed n group and x are the characterstcs observed for group. The model, however, s nadequate. To see ths consder that the outcome (whch s E(y x ) = b x ) s not bounded to values between zero and one and therefore cannot be nterpreted as a probablty. As t s generally mplausble to assume that a probablty can be calculated by a lnear functon, n a second step the lnear expresson b x s transformed by a nonlnear functon (lnk functon) F:

7 6 Evelyn Hayden and Danel Porath p ( b' x ) = F. (11) An approprate lnk functon transforms the values of b x to a scale wthn the nterval [0,1]. Ths can be acheved by any dstrbuton functon. The choce of the lnk functon determnes the type of model: wth a logstc lnk functon equaton (11) becomes a logt model, whle wth the normal dstrbuton (11) results n the probt model. However, when estmatng (10) wth OLS, the coeffcents wll be heteroscedastc, because Var(u ) = Var(p ) = p(x ) (1-p(x )). A possble way to acheve homoscedastcty would be to compute the WLS estmators of b n (10). However, albet possble, ths s not common practce. The reason s that n order to observe default frequences, the data has to be grouped before estmaton. Groupng nvolves consderable practcal problems lke defnng the sze and number of the groups and the treatment of dfferent covarates wthn the sngle groups. A better way to estmate logt and probt models, whch does not requre groupng, s the Maxmum-Lkelhood (ML) method. For a bnary dependent varable the lkelhood functon looks lke: 1 y [ ] y ( ) = P( b' x ) 1 P( b' x ) L b. For the probt model P(.) s the normal densty functon and for the logt model P(.) s the logstc densty functon. Wth equaton (12) the estmaton of the model s theoretcally convncng and also easy to handle. Furthermore, the ML-approach lends tself for a broad set of tests to evaluate the model and ts sngle varables (see Hosmer and Lemeshow (2000) for a comprehensve ntroducton). Usually, the choce of the lnk functon s not theoretcally drven. Users famlar wth the normal dstrbuton wll opt for the probt model. Indeed, the dfferences n the results of both classes of models are often neglgble. Ths s due to the fact that both dstrbuton functons have a smlar form except for the tals, whch are heaver for the logt model. The logt model s easer to handle, though. Frst of all, the computaton of the estmators s easer. However, today computatonal complexty s often rrelevant as most users apply statstcal software where the estmaton algorthms are ntegrated. What s more mportant s the fact that the coeffcents of the logt model can be more easly nterpreted. To see ths we transform the logt model gven n (9) n the followng way: P = e 1 P β' x The left-hand sde of (13) are the odds,.e. the relaton between the default probablty and the probablty of survval. Now t can be easly seen that a varaton of a sngle varable x k of one unt has an mpact of e β k on the odds, when β k denotes the coeffcent of the varable x k. Hence, the transformed coeffcents e β are called (12) (13)

8 I. Statstcal Methods to Develop Ratng Models 7 odds-ratos. They represent the multplcatve mpact of a borrower s characterstc on the odds. Therefore, for the logt model, the coeffcents can be nterpreted n a plausble way, whch s not possble for the probt model. Indeed, the most mportant weakness of bnary models s the fact that the nterpretaton of the coeffcents s not straghtforward. The strengths of logt and probt models can be summarzed as: The methods are theoretcally sound The results generated can be nterpreted drectly as default probabltes The sgnfcance of the model and the ndvdual coeffcents can be tested. Therefore, the stablty of the model can be assessed more effectvely than n the prevous cases. 6. Panel Models The methods dscussed so far are all cross-sectonal methods because all covarates are related to the same perod. However, typcally banks dspose of a set of covarates for more than one perod for each borrower. In ths case t s possble to expand the cross-sectonal nput data to a panel dataset. The man motvaton s to enlarge the number of avalable observatons for the estmaton and therefore enhance the stablty and the precson of the ratng model. Addtonally, panel models can ntegrate macroeconomc varables nto the model. Macroeconomc varables can mprove the model for several reasons. Frst, many macroeconomc data sources are more up-to-date than the borrowers characterstcs. For example, fnancal ratos calculated from balance sheet nformaton are usually updated only once a year and are often up to two years old when used for rsk assessment. The ol prce, nstead, s avalable on a daly frequency. Secondly, by stressng the macroeconomc nput factors, the model can be used for a form of stress-testng credt rsk. However, as macroeconomc varables prmarly affect the absolute value of the default probablty, t s only reasonable to ncorporate macroeconomc nput factors nto those classes of models that estmate default probabltes. In prncple, the structure of, for example, a panel logt or probt model remans the same as gven n the equatons of the prevous secton. The only dfference s that now the covarates are taken from a panel of data and have to be ndexed by an addtonal tme seres ndcator,.e. we observe x t nstead of x. At frst glance panel models seem smlar to cross-sectonal models. In fact, many developers gnore the dynamc pattern of the covarates and smply ft logt or probt models. However, logt and probt models rely on the assumpton of ndependent observatons. Generally, cross-sectonal data meets ths requrement, but panel data does not. The reason s that observatons from the same perod and observatons from the same borrower should be correlated. Introducng ths correlaton n the estmaton procedure s cumbersome. For example, the fxed-effects estmator known from panel analyss for contnuous dependent varables s not avalable for the

9 8 Evelyn Hayden and Danel Porath probt model. Besdes, the modfed fxed-effects estmator for logt models proposed by Chamberlan (1980) excludes all non-defaultng borrowers from the analyss and therefore seems napproprate. Fnally, the random-effects estmators proposed n the lterature are computatonally extensve and can only be computed wth specalzed software. For an econometrc dscusson of bnary panel analyss, refer to Hosmer and Lemeshow (2000). 7. Hazard Models All methods dscussed so far try to assess the rskness of borrowers by estmatng a certan type of score that ndcates whether or not a borrower s lkely to default wthn the specfed forecast horzon. However, no predcton about the exact default pont n tme s made. Besdes, these approaches do not allow the evaluaton of the borrowers rsk for future tme perods gven they should not default wthn the reference tme horzon. These dsadvantages can be remeded by means of hazard models, whch explctly take the survval functon and thus the tme at whch a borrower's default occurs nto account. Wthn ths class of models, the Cox proportonal hazard model (cf. Cox, 1972) s the most general regresson model, as t s not based on any assumptons concernng the nature or shape of the underlyng survval dstrbuton. The model assumes that the underlyng hazard rate (rather than survval tme) s a functon of the ndependent varables; no assumptons are made about the nature or shape of the hazard functon. Thus, the Cox s regresson model s a semparametrc model. The model can be wrtten as: h β' x ( t x ) h ( t) e, (14) = 0 where h (t x ) denotes the resultant hazard, gven the covarates for the respectve borrower and the respectve survval tme t. The term h 0 (t) s called the baselne hazard; t s the hazard when all ndependent varable values are equal to zero. If the covarates are measured as devatons from ther respectve means, h 0 (t) can be nterpreted as the hazard rate of the average borrower. Whle no assumptons are made about the underlyng hazard functon, the model equaton shown above mples mportant assumptons. Frst, t specfes a multplcatve relatonshp between the hazard functon and the log-lnear functon of the explanatory varables, whch mples that the rato of the hazards of two borrowers does not depend on tme,.e. the relatve rskness of the borrowers s constant, hence the name Cox proportonal hazard model. Besdes, the model assumes that the default pont n tme s a contnuous random varable. However, often the borrowers fnancal condtons are not observed contnuously but rather at dscrete ponts n tme. What s more, the covarates are

10 I. Statstcal Methods to Develop Ratng Models 9 treated as f they were constant over tme, whle typcal explanatory varables lke fnancal ratos change wth tme. Although there are some advanced models to ncorporate the above mentoned features, the estmaton of these models becomes complex. The strengths and weaknesses of hazard models can be summarzed as follows: Hazard models allow for the estmaton of a survval functon for all borrowers from the tme structure of hstorcal defaults, whch mples that default probabltes can be calculated for dfferent tme horzons. Estmatng these models under realstc assumptons s not straghtforward. 8. Neural Networks In recent years, neural networks have been dscussed extensvely as an alternatve to the (parametrc) models dscussed above. They offer a more flexble desgn to represent the connectons between ndependent and dependent varables. Neural networks belong to the class of non-parametrcal methods. Unlke the methods dscussed so far they do not estmate parameters of a well-specfed model. Instead, they are nspred by the way bologcal nervous systems, such as the bran, process nformaton. They typcally consst of many nodes that send a certan output f they receve a specfc nput from the other nodes to whch they are connected. Lke parametrc models, neural networks are traned by a tranng sample to classfy borrowers correctly. The fnal network s found by adjustng the connectons between the nput, output and any potental ntermedary nodes. The strengths and weaknesses of neural networks can be summarzed as: Neural networks easly model hghly complex, nonlnear relatonshps between the nput and the output varables. They are free from any dstrbutonal assumptons. These models can be quckly adapted to new nformaton (dependng on the tranng algorthm). There s no formal procedure to determne the optmum network topology for a specfc problem,.e. the number of the layers of nodes connectng the nput wth the output varables. Neural networks are black boxes, hence they are dffcult to nterpret. Calculatng default probabltes s possble only to a lmted extent and wth consderable extra effort. In summary, neural networks are partcularly sutable when there are no expectatons (based on experence or theoretcal arguments) on the relatonshp between the nput factors and the default event and the economc nterpretaton of the resultng models s of nferor mportance.

11 10 Evelyn Hayden and Danel Porath 9. Decson Trees A further category of non-parametrc methods comprses decson trees, also called classfcaton trees. Trees are models whch consst of a set of f-then splt condtons for classfyng cases nto two (or more) dfferent groups. Under these methods, the base sample s subdvded nto groups accordng to the covarates. In the case of bnary classfcaton trees, for example, each tree node s assgned by (usually unvarate) decson rules, whch descrbe the sample accordngly and subdvde t nto two subgroups each. New observatons are processed down the tree n accordance wth the decson rules' values untl the end node s reached, whch then represents the classfcaton of ths observaton. An example s gven n Fgure 1. Sector Constructon Other Years n busness Less than 2 EBIT. Equty rato Less than 15% More than 15%. Rsk class 2 Rsk class 3 Fgure 1. Decson Tree One of the most strkng dfferences of the parametrc models s that all covarates are grouped and treated as categorcal varables. Furthermore, whether a specfc varable or category becomes relevant depends on the categores of the varables n the upper level. For example, n Fgure 1 the varable years n busness s only relevant for companes whch operate n the constructon sector. Ths knd of dependence between varables s called nteracton. The most mportant algorthms for buldng decson trees are the Classfcaton and Regresson Trees algorthms (C&RT) popularzed by Breman et al. (1984) and the CHAID algorthm (Ch-square Automatc Interacton Detector, see Kass, 1978). Both algorthms use dfferent crtera to dentfy the best splts n the data and to collapse the categores whch are not sgnfcantly dfferent n outcome. The general strengths and weaknesses of trees are:

12 I. Statstcal Methods to Develop Ratng Models 11 Through categorzaton, nonlnear relatonshps between the varables and the score can be easly modelled. Interactons present n the data can be dentfed. Parametrc methods can model nteractons only to a lmted extent (by ntroducng dummy varables). As wth neural networks, decson trees are free from dstrbutonal assumptons. The output s easy to understand. Probabltes of default have to be calculated n a separate step. The output s (a few) rsk categores and not a contnuous score varable. Consequently, decson trees only calculate default probabltes for the fnal node n a tree, but not for ndvdual borrowers. Compared to other models, trees contan fewer varables and categores. The reason s that n each node the sample s successvely parttoned and therefore contnuously dmnshes. The stablty of the model cannot be assessed wth statstcal procedures. The strategy s to work wth a tranng sample and a hold-out sample. In summary, trees are partcularly suted when the data s characterzed by a lmted number of predctve varables whch are known to be nteractve. 10. Statstcal Models and Basel II Fnally, we ask the queston whether the models dscussed n ths chapter are n lne wth the IRB Approach of Basel II. Pror to the dscusson, t should be mentoned that n the Basel documents, ratng systems are defned n a broader sense than n ths chapter. Followng 394 of the Revsed Framework from June 2004 (cf. BIS, 2004) a ratng system comprses all the methods, processes, controls, and data collecton and IT systems that support the assessment of credt rsk, the assgnment of nternal ratngs, and the quantfcaton of default and loss estmates. Compared to ths defnton, these methods provde one component, namely the assgnment of nternal ratngs. The mnmum requrements for nternal ratng systems are treated n part II, secton III, H of the Revsed Framework. A few passages of the text concern the assgnment of nternal ratngs, and the requrements are general. They manly concern the ratng structure and the nput data, examples beng: a mnmum of 7 ratng classes of non-defaulted borrowers ( 404) no undue or excessve concentratons n sngle ratng classes ( 403, 406) a meanngful dfferentaton of rsk between the classes ( 410) plausble, ntutve and current nput data ( 410, 411) all relevant nformaton must be taken nto account ( 411). The requrements do not reveal any preference for a certan method. It s ndeed one of the central deas of the IRBA that the banks are free n the choce of the

13 12 Evelyn Hayden and Danel Porath method. Therefore the models dscussed here are all possble canddates for the IRB Approach. The strengths and weaknesses of the sngle methods concern some of the mnmum requrements. For example, hazard rate or logt panel models are especally suted for stress testng (as requred by 434, 345) snce they contan a tmeseres dmenson. Methods whch allow for the statstcal testng of the ndvdual nput factors (e.g. the logt model) provde a straghtforward way to demonstrate the plausblty of the nput factors (as requred by 410). When the outcome of the model s a contnuous varable, the ratng classes can be defned n a more flexble way ( 403, 404, 406). On the other hand, none of the drawbacks of the models consdered here excludes a specfc method. For example, a bank may have a preference for lnear regresson analyss. In ths case the plausblty of the nput factors cannot be verfed by statstcal tests and as a consequence the bank wll have to search for alternatve ways to meet the requrements of 410. In summary, the mnmum requrements are not ntended as a gudelne for the choce of a specfc model. Banks should rather base ther choce on ther nternal ams and restrctons. If necessary, those components that are only needed for the purpose to satsfy the crtera of the IRBA should be added n a second step. All models dscussed n ths chapter allow for ths. References Altman EI (1968), Fnancal Indcators, Dscrmnant Analyss, and the Predcton of Corporate Bankruptcy, Journal of Fnance. BIS (2004), Internatonal Convergence of Captal Measurement and Captal Standards, Basel Commttee on Bankng Supervson, June Breman L, Fredman JH, Olshen RA, Stone SJ (1984), Classfcaton and Regresson Trees, Wadsworth. Chamberlan G (1980), Analyss of Covarance wth Qualtatve Data, Revew of Economc Studes 47, Cox DR (1972), Regresson Models and Lfe Tables (wth Dscusson), Journal of Royal Statstcal Socety, Seres B. Greene W (2003), Econometrc Analyss, 5 th ed., Prentce-Hall New Jersey. Hosmer W, Lemeshow S (2000), Appled Logstc Regresson, New York, Wley. Kass GV (1978), An Exploratory Technque for Investgatng Large Quanttes of Categorcal Data, Appled Statstcs 29 (2), pp

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

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

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

Latent Class Regression. Statistics for Psychosocial Research II: Structural Models December 4 and 6, 2006

Latent Class Regression. Statistics for Psychosocial Research II: Structural Models December 4 and 6, 2006 Latent Class Regresson Statstcs for Psychosocal Research II: Structural Models December 4 and 6, 2006 Latent Class Regresson (LCR) What s t and when do we use t? Recall the standard latent class model

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

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

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

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

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

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

Single and multiple stage classifiers implementing logistic discrimination

Single and multiple stage classifiers implementing logistic discrimination Sngle and multple stage classfers mplementng logstc dscrmnaton Hélo Radke Bttencourt 1 Dens Alter de Olvera Moraes 2 Vctor Haertel 2 1 Pontfíca Unversdade Católca do Ro Grande do Sul - PUCRS Av. Ipranga,

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

How To Understand The Results Of The German Meris Cloud And Water Vapour Product

How To Understand The Results Of The German Meris Cloud And Water Vapour Product Ttel: Project: Doc. No.: MERIS level 3 cloud and water vapour products MAPP MAPP-ATBD-ClWVL3 Issue: 1 Revson: 0 Date: 9.12.1998 Functon Name Organsaton Sgnature Date Author: Bennartz FUB Preusker FUB Schüller

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

Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network

Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network 700 Proceedngs of the 8th Internatonal Conference on Innovaton & Management Forecastng the Demand of Emergency Supples: Based on the CBR Theory and BP Neural Network Fu Deqang, Lu Yun, L Changbng School

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

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

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

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

Survival analysis methods in Insurance Applications in car insurance contracts

Survival analysis methods in Insurance Applications in car insurance contracts Survval analyss methods n Insurance Applcatons n car nsurance contracts Abder OULIDI 1 Jean-Mare MARION 2 Hervé GANACHAUD 3 Abstract In ths wor, we are nterested n survval models and ther applcatons on

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

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

How To Evaluate A Dia Fund Suffcency

How To Evaluate A Dia Fund Suffcency DI Fund Suffcency Evaluaton Methodologcal Recommendatons and DIA Russa Practce Andre G. Melnkov Deputy General Drector DIA Russa THE DEPOSIT INSURANCE CONFERENCE IN THE MENA REGION AMMAN-JORDAN, 18 20

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

IMPACT ANALYSIS OF A CELLULAR PHONE

IMPACT ANALYSIS OF A CELLULAR PHONE 4 th ASA & μeta Internatonal Conference IMPACT AALYSIS OF A CELLULAR PHOE We Lu, 2 Hongy L Bejng FEAonlne Engneerng Co.,Ltd. Bejng, Chna ABSTRACT Drop test smulaton plays an mportant role n nvestgatng

More information

Forecasting and Stress Testing Credit Card Default using Dynamic Models

Forecasting and Stress Testing Credit Card Default using Dynamic Models Forecastng and Stress Testng Credt Card Default usng Dynamc Models Tony Bellott and Jonathan Crook Credt Research Centre Unversty of Ednburgh Busness School Verson 4.5 Abstract Typcally models of credt

More information

Customer Lifetime Value Modeling and Its Use for Customer Retention Planning

Customer Lifetime Value Modeling and Its Use for Customer Retention Planning Customer Lfetme Value Modelng and Its Use for Customer Retenton Plannng Saharon Rosset Enat Neumann Ur Eck Nurt Vatnk Yzhak Idan Amdocs Ltd. 8 Hapnna St. Ra anana 43, Israel {saharonr, enatn, ureck, nurtv,

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

Transition Matrix Models of Consumer Credit Ratings

Transition Matrix Models of Consumer Credit Ratings Transton Matrx Models of Consumer Credt Ratngs Abstract Although the corporate credt rsk lterature has many studes modellng the change n the credt rsk of corporate bonds over tme, there s far less analyss

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

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

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

Binomial Link Functions. Lori Murray, Phil Munz

Binomial Link Functions. Lori Murray, Phil Munz Bnomal Lnk Functons Lor Murray, Phl Munz Bnomal Lnk Functons Logt Lnk functon: ( p) p ln 1 p Probt Lnk functon: ( p) 1 ( p) Complentary Log Log functon: ( p) ln( ln(1 p)) Motvatng Example A researcher

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

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

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

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

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

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

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

Method for assessment of companies' credit rating (AJPES S.BON model) Short description of the methodology

Method for assessment of companies' credit rating (AJPES S.BON model) Short description of the methodology Method for assessment of companes' credt ratng (AJPES S.BON model) Short descrpton of the methodology Ljubljana, May 2011 ABSTRACT Assessng Slovenan companes' credt ratng scores usng the AJPES S.BON model

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

ESTIMATING THE MARKET VALUE OF FRANKING CREDITS: EMPIRICAL EVIDENCE FROM AUSTRALIA

ESTIMATING THE MARKET VALUE OF FRANKING CREDITS: EMPIRICAL EVIDENCE FROM AUSTRALIA ESTIMATING THE MARKET VALUE OF FRANKING CREDITS: EMPIRICAL EVIDENCE FROM AUSTRALIA Duc Vo Beauden Gellard Stefan Mero Economc Regulaton Authorty 469 Wellngton Street, Perth, WA 6000, Australa Phone: (08)

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

NEURO-FUZZY INFERENCE SYSTEM FOR E-COMMERCE WEBSITE EVALUATION

NEURO-FUZZY INFERENCE SYSTEM FOR E-COMMERCE WEBSITE EVALUATION NEURO-FUZZY INFERENE SYSTEM FOR E-OMMERE WEBSITE EVALUATION Huan Lu, School of Software, Harbn Unversty of Scence and Technology, Harbn, hna Faculty of Appled Mathematcs and omputer Scence, Belarusan State

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

Discussion Papers. Support Vector Machines (SVM) as a Technique for Solvency Analysis. Laura Auria Rouslan A. Moro. Berlin, August 2008

Discussion Papers. Support Vector Machines (SVM) as a Technique for Solvency Analysis. Laura Auria Rouslan A. Moro. Berlin, August 2008 Deutsches Insttut für Wrtschaftsforschung www.dw.de Dscusson Papers 8 Laura Aura Rouslan A. Moro Support Vector Machnes (SVM) as a Technque for Solvency Analyss Berln, August 2008 Opnons expressed n ths

More information

Scale Dependence of Overconfidence in Stock Market Volatility Forecasts

Scale Dependence of Overconfidence in Stock Market Volatility Forecasts Scale Dependence of Overconfdence n Stoc Maret Volatlty Forecasts Marus Glaser, Thomas Langer, Jens Reynders, Martn Weber* June 7, 007 Abstract In ths study, we analyze whether volatlty forecasts (judgmental

More information

Reporting Forms ARF 113.0A, ARF 113.0B, ARF 113.0C and ARF 113.0D FIRB Corporate (including SME Corporate), Sovereign and Bank Instruction Guide

Reporting Forms ARF 113.0A, ARF 113.0B, ARF 113.0C and ARF 113.0D FIRB Corporate (including SME Corporate), Sovereign and Bank Instruction Guide Reportng Forms ARF 113.0A, ARF 113.0B, ARF 113.0C and ARF 113.0D FIRB Corporate (ncludng SME Corporate), Soveregn and Bank Instructon Gude Ths nstructon gude s desgned to assst n the completon of the FIRB

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

The Current Employment Statistics (CES) survey,

The Current Employment Statistics (CES) survey, Busness Brths and Deaths Impact of busness brths and deaths n the payroll survey The CES probablty-based sample redesgn accounts for most busness brth employment through the mputaton of busness deaths,

More information

Evaluating credit risk models: A critique and a new proposal

Evaluating 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 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

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

Study on Model of Risks Assessment of Standard Operation in Rural Power Network

Study on Model of Risks Assessment of Standard Operation in Rural Power Network Study on Model of Rsks Assessment of Standard Operaton n Rural Power Network Qngj L 1, Tao Yang 2 1 Qngj L, College of Informaton and Electrcal Engneerng, Shenyang Agrculture Unversty, Shenyang 110866,

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

Calculating the high frequency transmission line parameters of power cables

Calculating the high frequency transmission line parameters of power cables < ' Calculatng the hgh frequency transmsson lne parameters of power cables Authors: Dr. John Dcknson, Laboratory Servces Manager, N 0 RW E B Communcatons Mr. Peter J. Ncholson, Project Assgnment Manager,

More information

Chapter XX More advanced approaches to the analysis of survey data. Gad Nathan Hebrew University Jerusalem, Israel. Abstract

Chapter XX More advanced approaches to the analysis of survey data. Gad Nathan Hebrew University Jerusalem, Israel. Abstract Household Sample Surveys n Developng and Transton Countres Chapter More advanced approaches to the analyss of survey data Gad Nathan Hebrew Unversty Jerusalem, Israel Abstract In the present chapter, we

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

Economic Interpretation of Regression. Theory and Applications

Economic Interpretation of Regression. Theory and Applications Economc Interpretaton of Regresson Theor and Applcatons Classcal and Baesan Econometrc Methods Applcaton of mathematcal statstcs to economc data for emprcal support Economc theor postulates a qualtatve

More information

ENTERPRISE RISK MANAGEMENT IN INSURANCE GROUPS: MEASURING RISK CONCENTRATION AND DEFAULT RISK

ENTERPRISE RISK MANAGEMENT IN INSURANCE GROUPS: MEASURING RISK CONCENTRATION AND DEFAULT RISK ETERPRISE RISK MAAGEMET I ISURACE GROUPS: MEASURIG RISK COCETRATIO AD DEFAULT RISK ADIE GATZERT HATO SCHMEISER STEFA SCHUCKMA WORKIG PAPERS O RISK MAAGEMET AD ISURACE O. 35 EDITED BY HATO SCHMEISER CHAIR

More information

1 De nitions and Censoring

1 De nitions and Censoring De ntons and Censorng. Survval Analyss We begn by consderng smple analyses but we wll lead up to and take a look at regresson on explanatory factors., as n lnear regresson part A. The mportant d erence

More information

A DYNAMIC CRASHING METHOD FOR PROJECT MANAGEMENT USING SIMULATION-BASED OPTIMIZATION. Michael E. Kuhl Radhamés A. Tolentino-Peña

A DYNAMIC CRASHING METHOD FOR PROJECT MANAGEMENT USING SIMULATION-BASED OPTIMIZATION. Michael E. Kuhl Radhamés A. Tolentino-Peña Proceedngs of the 2008 Wnter Smulaton Conference S. J. Mason, R. R. Hll, L. Mönch, O. Rose, T. Jefferson, J. W. Fowler eds. A DYNAMIC CRASHING METHOD FOR PROJECT MANAGEMENT USING SIMULATION-BASED OPTIMIZATION

More information

Vasicek s Model of Distribution of Losses in a Large, Homogeneous Portfolio

Vasicek s Model of Distribution of Losses in a Large, Homogeneous Portfolio Vascek s Model of Dstrbuton of Losses n a Large, Homogeneous Portfolo Stephen M Schaefer London Busness School Credt Rsk Electve Summer 2012 Vascek s Model Important method for calculatng dstrbuton of

More information

8 Algorithm for Binary Searching in Trees

8 Algorithm for Binary Searching in Trees 8 Algorthm for Bnary Searchng n Trees In ths secton we present our algorthm for bnary searchng n trees. A crucal observaton employed by the algorthm s that ths problem can be effcently solved when the

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

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

An Empirical Study of Search Engine Advertising Effectiveness

An Empirical Study of Search Engine Advertising Effectiveness An Emprcal Study of Search Engne Advertsng Effectveness Sanjog Msra, Smon School of Busness Unversty of Rochester Edeal Pnker, Smon School of Busness Unversty of Rochester Alan Rmm-Kaufman, Rmm-Kaufman

More information

Intra-year Cash Flow Patterns: A Simple Solution for an Unnecessary Appraisal Error

Intra-year Cash Flow Patterns: A Simple Solution for an Unnecessary Appraisal Error Intra-year Cash Flow Patterns: A Smple Soluton for an Unnecessary Apprasal Error By C. Donald Wggns (Professor of Accountng and Fnance, the Unversty of North Florda), B. Perry Woodsde (Assocate Professor

More information

Gender differences in revealed risk taking: evidence from mutual fund investors

Gender differences in revealed risk taking: evidence from mutual fund investors Economcs Letters 76 (2002) 151 158 www.elsever.com/ locate/ econbase Gender dfferences n revealed rsk takng: evdence from mutual fund nvestors a b c, * Peggy D. Dwyer, James H. Glkeson, John A. Lst a Unversty

More information

Statistical algorithms in Review Manager 5

Statistical algorithms in Review Manager 5 Statstcal algorthms n Reve Manager 5 Jonathan J Deeks and Julan PT Hggns on behalf of the Statstcal Methods Group of The Cochrane Collaboraton August 00 Data structure Consder a meta-analyss of k studes

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

Scaling Models for the Severity and Frequency of External Operational Loss Data

Scaling Models for the Severity and Frequency of External Operational Loss Data Scalng Models for the Severty and Frequency of External Operatonal Loss Data Hela Dahen * Department of Fnance and Canada Research Char n Rsk Management, HEC Montreal, Canada Georges Donne * Department

More information

Risk Model of Long-Term Production Scheduling in Open Pit Gold Mining

Risk Model of Long-Term Production Scheduling in Open Pit Gold Mining Rsk Model of Long-Term 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 information

BERNSTEIN POLYNOMIALS

BERNSTEIN POLYNOMIALS On-Lne Geometrc Modelng Notes BERNSTEIN POLYNOMIALS Kenneth I. Joy Vsualzaton and Graphcs Research Group Department of Computer Scence Unversty of Calforna, Davs Overvew Polynomals are ncredbly useful

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

Logistic Regression. Lecture 4: More classifiers and classes. Logistic regression. Adaboost. Optimization. Multiple class classification

Logistic Regression. Lecture 4: More classifiers and classes. Logistic regression. Adaboost. Optimization. Multiple class classification Lecture 4: More classfers and classes C4B Machne Learnng Hlary 20 A. Zsserman Logstc regresson Loss functons revsted Adaboost Loss functons revsted Optmzaton Multple class classfcaton Logstc Regresson

More information

Stress test for measuring insurance risks in non-life insurance

Stress test for measuring insurance risks in non-life insurance PROMEMORIA Datum June 01 Fnansnspektonen Författare Bengt von Bahr, Younes Elonq and Erk Elvers Stress test for measurng nsurance rsks n non-lfe nsurance Summary Ths memo descrbes stress testng of nsurance

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

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

HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA*

HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA* HOUSEHOLDS DEBT BURDEN: AN ANALYSIS BASED ON MICROECONOMIC DATA* Luísa Farnha** 1. INTRODUCTION The rapd growth n Portuguese households ndebtedness n the past few years ncreased the concerns that debt

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

Estimating Total Claim Size in the Auto Insurance Industry: a Comparison between Tweedie and Zero-Adjusted Inverse Gaussian Distribution

Estimating Total Claim Size in the Auto Insurance Industry: a Comparison between Tweedie and Zero-Adjusted Inverse Gaussian Distribution Estmatng otal Clam Sze n the Auto Insurance Industry: a Comparson between weede and Zero-Adjusted Inverse Gaussan Dstrbuton Autora: Adrana Bruscato Bortoluzzo, Italo De Paula Franca, Marco Antono Leonel

More information

Face Verification Problem. Face Recognition Problem. Application: Access Control. Biometric Authentication. Face Verification (1:1 matching)

Face Verification Problem. Face Recognition Problem. Application: Access Control. Biometric Authentication. Face Verification (1:1 matching) Face Recognton Problem Face Verfcaton Problem Face Verfcaton (1:1 matchng) Querymage face query Face Recognton (1:N matchng) database Applcaton: Access Control www.vsage.com www.vsoncs.com Bometrc Authentcaton

More information

Covariate-based pricing of automobile insurance

Covariate-based 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) Covarate-based prcng of automoble nsurance Abstract Ths

More information

Traffic State Estimation in the Traffic Management Center of Berlin

Traffic State Estimation in the Traffic Management Center of Berlin Traffc State Estmaton n the Traffc Management Center of Berln Authors: Peter Vortsch, PTV AG, Stumpfstrasse, D-763 Karlsruhe, Germany phone ++49/72/965/35, emal peter.vortsch@ptv.de Peter Möhl, PTV AG,

More information

Assessing the Fairness of a Firm s Allocation of Shares in Initial Public Offerings: Adapting a Measure from Biostatistics

Assessing the Fairness of a Firm s Allocation of Shares in Initial Public Offerings: Adapting a Measure from Biostatistics Assessng the Farness of a Frm s Allocaton of Shares n Intal Publc Offerngs: Adaptng a Measure from Bostatstcs by Efstatha Bura and Joseph L. Gastwrth Department of Statstcs The George Washngton Unversty

More information

Optimal Customized Pricing in Competitive Settings

Optimal Customized Pricing in Competitive Settings Optmal Customzed Prcng n Compettve Settngs Vshal Agrawal Industral & Systems Engneerng, Georga Insttute of Technology, Atlanta, Georga 30332 vshalagrawal@gatech.edu Mark Ferguson College of Management,

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

Discount Rate for Workout Recoveries: An Empirical Study*

Discount Rate for Workout Recoveries: An Empirical Study* Dscount Rate for Workout Recoveres: An Emprcal Study* Brooks Brady Amercan Express Peter Chang Standard & Poor s Peter Mu** McMaster Unversty Boge Ozdemr Standard & Poor s Davd Schwartz Federal Reserve

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

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

Realistic Image Synthesis

Realistic Image Synthesis Realstc Image Synthess - Combned Samplng and Path Tracng - Phlpp Slusallek Karol Myszkowsk Vncent Pegoraro Overvew: Today Combned Samplng (Multple Importance Samplng) Renderng and Measurng Equaton Random

More information

Fixed income risk attribution

Fixed income risk attribution 5 Fxed ncome rsk attrbuton Chthra Krshnamurth RskMetrcs Group chthra.krshnamurth@rskmetrcs.com We compare the rsk of the actve portfolo wth that of the benchmark and segment the dfference between the two

More information

ECONOMICS OF PLANT ENERGY SAVINGS PROJECTS IN A CHANGING MARKET Douglas C White Emerson Process Management

ECONOMICS OF PLANT ENERGY SAVINGS PROJECTS IN A CHANGING MARKET Douglas C White Emerson Process Management ECONOMICS OF PLANT ENERGY SAVINGS PROJECTS IN A CHANGING MARKET Douglas C Whte Emerson Process Management Abstract Energy prces have exhbted sgnfcant volatlty n recent years. For example, natural gas prces

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

Quantification of qualitative data: the case of the Central Bank of Armenia

Quantification of qualitative data: the case of the Central Bank of Armenia Quantfcaton of qualtatve data: the case of the Central Bank of Armena Martn Galstyan 1 and Vahe Movssyan 2 Overvew The effect of non-fnancal organsatons and consumers atttudes on economc actvty s a subject

More information

The Racial and Gender Interest Rate Gap. in Small Business Lending: Improved Estimates Using Matching Methods*

The 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 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