BAYESIAN ESTIMATION AND MODEL SELECTION FOR
|
|
- Loreen Marshall
- 8 years ago
- Views:
Transcription
1 RODRÍGUEZ: BAYESIAN ESTIMATION AND MODEL SELECTION REVISTA FOR THE DE WEEKLY ECONOMÍA COLOMBIAN DEL ROSARIO, EXCHANGE 4 (DICIEMBRE RATE ) 43-7 BAYESIAN ESTIMATION AND MODEL SELECTION FOR THE WEEKLY COLOMBIAN EXCHANGE RATE Norbero Rodríguez Niño * Banco de la República nrodrini@banrep.gov.co ABSTRACT Tis documen reviews and applies recenly developed ecniques for Bayesian esimaion and model selecion in e conex of Time Series modeling for Socasic Volailiy. Afer e lieraure review on Generalized Condiional Auoregressive models, Socasic Volailiy models, and e relevan resuls on Markov Cain Mone Carlo meods (MCMC), an example applying suc ecniques is sown. Te meodology is used wi a series of Weekly Colombian-USA Excange Rae on seven differen models. Te GARCH model, wic uses Type-IV Pearson disribuion, is favored for e selecing ecnique, Reversible Jump MCMC, over oer models, including Socasic Volailiy Models wi a Suden- disribuion. Keywords: Bayesian Esimaion, Markov Cain Mone Carlo Meods (MCMC) GARCH Models, Socasic Volailiy, Model Selecion. JEL Classificaion: C, C3, C5, C, C5 I. INTRODUCTION Two differen and compeing ecniques are nowadays used for economericians and saisicians o model volailiy, as in reurn asses or excange raes: One of em, Auoregressive Condiional Heeroscedasic (ARCH), is generalizaion, GARCH, and muliple exensions ave proven o be very successful in modeling financial ime-varying volailiy series. Te compeing alernaive o GARCH models are Socasic Volailiy models, mainly reaed in e frequenis framework, wic ave more eoreical background. Tey appear in e financial lieraure on opion pricing as a generalizaion of e Black-Scoles model. * Tis work is based on e Maser Repor presened o Te Universiy of Texas a Ausin, by e auor, as a requiremen for e degree of Maser of Science in saisics.
2 44 REVISTA DE ECONOMÍA DEL ROSARIO, 4 (DICIEMBRE ) Usually e researcer faces e quesion of wa model o use. Several alernaives ave been proposed in e frequenis saisical framework o deal wi is, ranging from R, e radiional and exensively used Akaike Informaion Crierion (AIC), PRESS saisic, and many oers. In ose alernaives e model residuals are obained (usually observed minus adjused or peraps deviance residuals) and aggregaed o form e measures of adequacy. Unforunaely, classical approaces o model coice are limied. Te wellknown sandard Neyman-Pearson eory provides an opimal es roug e likeliood raio for e limied case of e comparison of wo disinc models. More generally, e likeliood raio es enables a coice beween models only in e nesed case, were ere is an unambiguous null ypoesis. Selecion is based upon an asympoic χ approximaion, wic usually is poor for small sample sizes. Frequenis eory does no offer muc for model selecion of non-nesed models, wic are no rare in pracice. (See Piorier, 995, and Gelfand, 995, for references). Tis documens uses MCMC, an inuiive, compuaionally easy-o-implemen, and inexpensive Bayesian alernaive, o decide among suiable models. Tis documen is organized as follows: Secion II presens a review of GARCH models and Socasic Volailiy models, especially ose models o be used laer in is paper; en i moves on o MCMC meods and Bayesian model selecion ecniques. Secion III deals wi ow o use MCMC o implemen esimaion of GARCH models, as Vronos e al. () sugges, and ow o esimae SV models. Secion IV deals wi e resuls and Secion V presens e main conclusions, some suggesions and limiaions. II. BACKGROUND Tis caper presens a sor review of some saisical and economerics model proposed in e lieraure o model ime-varying volailiy series. Afer a, Bayesian ecniques for model esimaion and selecion are briefly presened. A. Generalized Auoregressive Condiional Heerocedasic Models Time series models, radiionally fied in pracice, are supppose o ave consan variance, bu Wen working wi ig frequency ime series a is seldom e case, as can be seen in Figure, an Auoregressive Condiional Heerocedasic models (ARCH) ave been proposed in e lieraure o deal wi is problem, in e spiri of Engel (98) could be ried; is is y = ε or y = c + ε or y = c + ε () as some model for e levels of e observed ime serie, wi some funcion of lags of y or oer exogenous variables, or even an ARMA erm (see Box and Jenkins, 976); were ε ~ N(, ) and r = α + α ε =, for α i, i =,,..., r. i i i
3 RODRÍGUEZ: BAYESIAN ESTIMATION AND MODEL SELECTION FOR THE WEEKLY COLOMBIAN EXCHANGE RATE 45 Hence, e condiional variances are oug of as a funcion of e square of e previous observaional residuals. In e original and simple case a normaliy is assumed, e likeliood is given by: l T / ( y α,..., αr, ) = ( π ) exp ( ε / ) = () wen, in reurn financial ime series analysis, e usual mos imporan unknown o esimae and forecas is e volailiy, A Generalized ARCH (GARCH), a usually resuls in more parsimonious represenaions, as Bollerslev (986) proposed, assumes a e condiional variances follow an ARMA process, us: r s = + α αiε i + i = j = β j j (3) wi e analogous likeliood, as in e ARCH case, if normaliy for ε is assumed. Wi resricions α i, i =,,..., r, and β j, i =,,..., s, o guaranee a r s for all, and Ó i = α i + Ó j = β j < in order o assure saionariy. Weaker resricions can be required, in pracice, oug (See Nelson and Cao, 99). Anoer exension of (3) is using a Suden- disribuion, wi degrees of freedom o accoun for e eavier ails of e disribuion of e error process {ε } as was inroduced by Bollerslev (987), and by Baillie and Bollerslev (989). Te likeliood en is l( y α,..., α, β,..., β, r s n, ) = T = Γ Γ ( n / ) n+ ( ) [ π n ] / ε + n ( n+ ) / (4) An Exponenial GARCH (EGARCH), is inroduced by Nelson (99) o avoid imposing resricions on α i, β j in (3) wi ln( r s ε i ε i ε i ) = α + ϕ i E + α i + j ln( j ) β (5) i = i i i j = ε E for e condiional variance of {ε } and [ ] i i = Γ ( / ν ) Γ (/ ν ) Γ ( 3 / ν
4 46 REVISTA DE ECONOMÍA DEL ROSARIO, 4 (DICIEMBRE ) Addiionally, assuming a ε follows a Generalized Gaussian Disribuion, GED, en e likeliood is: l( y α, α, ϕ, β, ν, i i j ) = T = c exp.5 ε λ ν (6) were c = λ ν Γ ν ( + / ν ) and λ = / ( / ν ) Γ (/ ν ) Γ ( 3 / ν ) Te use of Bernoulli-Mixures of wo normal disribuions, proposed by Ball and Torous (983), was successfully implemened by Vlaar and Palm (993): r s = + α α i ε i + i = j = β j j for e condiional variances. Tey used an MA() erm o model e canges in levels of several excange raes, as: ε = λν θ ε y (7) and eac ε is disribued as a mixure of wo normals, ence ε ~ ( λ )N( λν, ) + λn (( λ ) ν, + δ ) were λ is e jump inensiy, v is e expecaion in e jump size, and δ e expeced cange in variance. Tis represenaion is useful and inuiive for economies wi arge or bands for eir excange raes, like Colombia or e European Economic Union. Te MA erm, θ ε is explained as allowing for mean reversion. Terefore, e likeliood is expressed as: l( y, θ, α, β, λ, δ, ν, i j ) T = = / λ ( ε + λν ) ( π ) exp + λ ( ε ( ) ) exp λ ν / + ) δ ( + δ ) ( (8) More recenly, Bera and Premarane () propose e use of e Pearson Type- IV Family Disribuions in order o model skewness and lepokurosis a are larger an usual. For a GARCH(,), e log-likeliood is:
5 47 RODRÍGUEZ: BAYESIAN ESTIMATION AND MODEL SELECTION FOR THE WEEKLY COLOMBIAN EXCHANGE RATE = = T ) y ( l ) y ( l Θ Θ = y y ln r y y arcan C ) ln( ).5 ln( ) ln( y l( α µ α µ δ α Θ (9) wi { }ψ δψ ψ π d exp sin C r = and ( ),,, r,,,, µ δ β α α Θ =. Tey used y y ε + + = and + + = β ε α for eir empirical applicaion, bu, of course, e model on y can be exended. Bu according o Nagaara (999), (9) sould be = y y ln r y y arcan ) r ( C ) ln( ).5 ln( ) ln( y l( α µ α µ δ α Θ wi [ ] [ ] + + = π ψ ψ δ ψ δ π r d ) r ( exp sin / ) r ( exp C Wen informaion arrives a random order and daa refers o close-o-close periods, Hsie (989) sowed a e use of a normal-lognormal mixure disribuion improves e fi over oer GARCH alernaives. Ta disribuion is no ried ere because i requires e compuaion of a defined inegral over one mus rely on ig numerical inegraion, owever, o provide a suiable soluion o a problem. ()
6 48 REVISTA DE ECONOMÍA DEL ROSARIO, 4 (DICIEMBRE ) B. Sae Space Models A differen alernaive in ime-series analysis is an Sae-Space (S-S), model, as Wes and Harrison (997) sae, ey exend and updae e seminal paper by Harrison and Sevenson (976). Te model is ypically represened by wo equaions: ' Observaion Equaion: Y = F θ + ν, v ~ N(, V ), () Sysem Equaion: θ = G θ + w, w ~ N(, W ) () were y is e observed (someimes laen) variable; θ is a vecor of unknown parameers, wic follows e firs order Markov process (); v is a vecor of unobserved and uncorrelaed socasic error erms; is a marix of known coefficiens; and w is an unobserved socasic erm, generally assumed uncorrelaed wi v. I is wor noing a any daa series for wic ere exiss a naural ordering of observaion, fis ino e dynamic framework, so e ime series no need be equally spaced, and missing daa problems can easily be andled in is conex. Pole e al. (994) presen applied meodology, as well as muliple examples. Usually, and wiou muc loss, V can be considered consan, and working in erms of e precision = V, i is possible o ge esimaions of W. Te likeliood for e S-S model is given below: l( y Θ ) = T l( y θ ) α T = = ( πv ) ( exp ' y F θ ) V (3) wic is e densiy of a normal wi mean ' F θ and variance V. Prior probabiliies can be se up on θ and a fully Bayesian analysis of e Sae-Space model can be run. A non-linear (non-gaussian) S-S model can be se up as follows, (See Harvey e al., 994 for deails): y = ε exp ( /) (4) as e non-linear observaion equaion, and η = γ + + η, η ~ N(, σ ), (5) σ η as e sysem equaion, were = ln ( )
7 RODRÍGUEZ: BAYESIAN ESTIMATION AND MODEL SELECTION FOR THE WEEKLY COLOMBIAN EXCHANGE RATE 49 Tis model, wic is known in e financial and economeric lieraure as e Socasic Volailiy model (SV for sor) can be ransformed o ge a linear observaional equaion, as ln ( y ) = + ln ( ε ) (6) were y is e mean correced reurn a ime ; is e log-volailiy a ime, wic is assumed o follow a saionary process, wi drawn from a saionary disribuion; ε and η are uncorrelaed sandard normal wie noise socks; is e persisence in e volailiy (wen for saionariy resricio < ); and σ η is e volailiy of e log-volailiy. Te likeliood, assuming normal disribuion, can be obained by using a Kalman Filer (See Jaquier e al, 994, Ap. B.). Te use of a v degrees of freedom Suden disribuion on as been considered, and e Kalman Filer needs some minor modificaions (See Ruiz, 994). C. Markov Cain Mone Carlo Meods Wen doing fully Bayesian analysis of complex or ig-dimensionaliy models, e researcer usually faces e problem of non-conjugacy, meaning a non-exac analyical poserior disribuion can be acieved. Tis leads o e necessiy of using simulaion approaces. Direc simulaion is ofen impossible, due o e complicaed maemaical form of e poserior disribuion in many applied models. Because of a, an exponenial rise in e ineres and applicaion of Markov Cain Mone Carlo (referred o by is acronym, MCMC) as a ool for numerical compuaion of complex inegrals, paricularly in Bayesian analysis, as emerged. Te key o Markov Cain simulaion is o creae a Markov process wose saionary disribuion is a specified π(θ y) and o run e simulaion long enoug a e disribuion of e curren draws is close enoug o e saionary disribuion. Once e simulaion algorim as been implemened, i sould be ieraed unil convergence as been reaced, or, if convergence is painfully slow, e algorim sould be alered. Hence, e sudy of MCMC as seen a corresponding ineres in e convergence properies of e resulan cains, wic may be assessed roug a suie of diagnosics borrowed from diverse areas suc as ime series, exploraory daa analysis (EDA), and cenral limi eory. Te mos widely used Markov Cain Mone Carlo meods are e Gibbs Sampler and e group of Meropolis-Hasings algorims and a good descripion of em can be found in Gamerman (997) or Gelman e al. (995), among many oers.. E.g., e use of digamma and rigamma funcions; Abramowiz and Segun (967) offer compuaional deails.
8 5 REVISTA DE ECONOMÍA DEL ROSARIO, 4 (DICIEMBRE ) D. Model Selecion Te mos damaging commen on e sandard pracice of coosing a single model, and en proceeding condiional on i, is a e researc s uncerainy is undersaed. Piorier (995, p. 65) For frequeniss e model selecion problem reduces o coosing one from a se of M models. Tis is usually e main aim of e analysis, and is done according o some model selecion crierion, as saed above. Boosrap meods ave been used for model selecion (See Maddala and Li, 996, sec 5, for references). For anoer perspecive see Poski and Tremayne (983). Essenially, wo alernaive approaces in e Bayesian conex are presened. Te firs was inroduced by Carlin and Cib (995) and considers all models in a formaion, called ere a supermodel. Te Markov Cain simulaion sceme for is supermodel is presened below. Te second approac presens sopisicaed simulaion ecniques using Markov cain wi jumps beween e differen models; i is referred o as Reversible Jumping, and i was inroduced by Green (995). I will be assumed rougou is secion a y is observed and i can be described according o a model M j wi parameers θ j of dimension d j, aking values in a parameer space Θ j R dj, j =,,..., M. Te value of M could be as, for insance, wen considering counable classes of models. m serves e purpose of indicaing a specific model. Assume for e momen a e poserior disribuion π(θ, j), e join disribuion of e super-parameer and e model indicaor are o be obained. However, e main ineres in inference is o obain e poserior disribuion of θ j m = j, j =,,...,M. Tese disribuions respecively provide e poserior inference wiin eac of e models and e poserior probabiliies of e models. Te supermodel approac provides a sample from is more general, peraps unnecessary poserior disribuion wereas e approac wi jumps only provides samples from θ j m = j, j =,,...,M, and m. Te presence of common parameers does no pose any problem ere.. Markov Cains for Super-Models Te join disribuion of all random quaniies is given by π(y, θ, j) = π(y θ,j) π(θ j)π j (7) were j is e value of m and π j = P(m = j). Given a m = j, e disribuion of y depends on θ only roug θ j, or maemaically, π(y, θ, j) = π(y θ,j) (8)
9 RODRÍGUEZ: BAYESIAN ESTIMATION AND MODEL SELECTION FOR THE WEEKLY COLOMBIAN EXCHANGE RATE 5 Assume also a e θ j are condiionally independen, given e value of m. Hence, M π( θ j ) = π( θ j i = i ) Noe a e prior disribuion π(θ i j), for i j does no make muc sense. I specifies e disribuion of e parameers of model i, condiioned on e fac a is is no e rue model. Carlin and Cib (995) refer o ese as pseudo-prior or linking disribuions. Due o e condiional independence (8), ese priors do no inerfere in e expressions of e marginal predicive densiies for eac model. Nevereless, ey are relevan for e consrucion of e cain and mus be specified. I follows from e above specificaion a M j, j ) π( θi j ) j = π( y, θ, j ) = π( y θ π wic is proporional o e join poserior disribuion of θ and m. A naural blocking is formed by grouping eac model s parameers and m. Te full condiional disribuions for θ, θ,..., θ M and m are obained as follows: For block θ j, j =,..., M, j π ( p j( θ j ) y θ, j ) π( θ j π( θ i ), i j j ) for m = j for m = i j For block m p M ( j ) = k π( y θ, j ) j M j = π( θ j ) π, j i j =,..., M a is a discree disribuion wi proporionaliy consan k = M M π( y θl, l ) π( θi l ) π l l = i =
10 5 REVISTA DE ECONOMÍA DEL ROSARIO, 4 (DICIEMBRE ) m can always be sampled direcly because i as a discree disribuion. Direc sampling from blocks θ j will depend on e conjugacy srucure for model m = j and e form of e pseudo prior disribuion. Wen direc sampling for some of e θ j s is no possible, Meropolis-Hasings seps may be used. Te above sceme saisfies e condiions of a convenional Markov Cain and erefore converges o e arge disribuion given by e poserior. Comparison beween models is based on e marginal poserior disribuion of m, p(j), j =,..., M. Tese probabiliies are esimaed by e proporion of values of m equal o j in e sample of size n. Te pseudo prior disribuions mus be carefully cosen, as ey affec e rae of convergence of e cain. Carlin and Cib (995) recommend e use of simple sandard approximaions based on univariae esimaes obained from pilo cains. Te same auors sugges using fairly vague prior disribuions, bu i is well-known a wen using is pracice on models wi differen dimensions e Bayes facors urn ou o be very sensiive. So, is prior seing may need furer jusificaion o saisfy poenial users. Finally, is approac is no applicable o e case of counable number of models under consideraion. Hence, e number of pracical and eoreical difficulies of is approac sugges i sould be used wi care. See Gamerman (997), were more deails can be found.. Markov Cains wi Jumps Green (995) inroduced a reversible-jump MCMC sraegy for generaing from e join poserior π(m,θ m y), based on e sandard Meropolis-Hasings approac. Te reversible-jump MCMC was also applied by Ricardson and Green (997) for an analysis of univariae normal mixure; by Nobile and Green (), for facorial experimens using mixure modeling; and Dellaporas and Foreser (999), for analysis of coningency ables. During reversible-jump MCMC sampling, e consruced Markov Cain moves wiin and beween models, so a e limiing proporion of visis o a given model is e required π(m y) In general, suppose a e curren sae of e Markov Cain a ime is (m, θ m ) were θ m as dimension d(θ m ) and a move is proposed a ime + o a new model m wi probabiliy j(m, m ) and corresponding parameer vecor θ m. Ten, a vecor is generaed from a specified proposal densiy q(u θ m, m, m ) and (θ m,, u ) = g m,m (θ m, u) is se for a specified inverible funcion g m,m suc a g m,m = g Noe m,m a d(θ m ) + d(u) = d(θ m ) + d(u ). Green (995) sowed a, if e new move is acceped as e nex realizaion of e Markov Cain wi probabiliy α = min{,r}, were π( y m', θ' m' ) π( θ' m' m' ) π( r = π( y m, θ ) π( θ m ) π( m m m' ) j( m', m )q( m ) j( m, m' u' θ' )q( u θ m m', m', m ), m, m' ) J
11 RODRÍGUEZ: BAYESIAN ESTIMATION AND MODEL SELECTION FOR THE WEEKLY COLOMBIAN EXCHANGE RATE 53 wi J = (θ m, u )/ (θ m, u) denoing e Jacobian of e ransformaion, en e cain saisfies e condiion of deailed balance and as e required limiing disribuion π(m, θ m y). Te condiion of deailed balance requires a e equilibrium probabiliy of moving from a sae (m,θ m ) o (m,θ m ) equals a of moving from (m,θ m ) o (m,θ m ) for deails, see Green (995). To implemen e reversible-jump MCMC, e probabiliies j(m,m ) need o be specified for every proposed move, as well as e proposal disribuions q(u θ m, m, m ), q(u θ m, m, m) and e funcion g m,m Tese coices do no affec e resuls in erms of models seleced bu may affec crucially e convergence rae of e Markov Cain. For e probabiliy j(m,m ) one non-informaive alernaive is j(m,m ) = (M ), for all m, m M, wen a eac sae of e cain a move from one model o oer one is always proposed. Vronos e al. () proposed a modificaion of Green s ecnique, wic ey ave successfully implemened in a series of experimens wi GARCH and EGARCH models; is is described as follows: Firs, ey sugges a all e parameers of e proposed model be generaed from a proposal disribuion. Consequenly, (θ m, u ) = (u,θ m ) wi d(θ m ) = d(u ) and d(θ m ) = d(u), q(u θ m, m, m ) = q(u m ), q(u θ m, m, m) = q(u m), and e Jacobian in (9) is. In is case, e probabiliy of accepance of e new move as e nex realizaion of e Markov cain is given by α = min{,r}, were π( y m', θ' m' r = π( y m, θ m ) π( θ' ) π( θ m m' m' ) π( m ) π( m' ) j( m', m )q( u' m ) m ) j( m, m' )q( u m' ) () Te proposal densiies q(u m ) and q(u m) can be cosen by invesigaion of a pilo run Tey sar e cain from e bes available saring values (e.g., e maximum likeliood esimaes wen available) and simulae e wiin-model Markov Cain many imes o obain approximae marginal poserior means and covariance marices for eac model parameer vecor. Tese esimaes are en used o consruc proposal densiies q(u m ) and q(u m) aken as mulivariae normal densiies. III. METHODOLOGY A. Te Daa In order o illusrae e esimaion and selecing meodologies, e weekly observaions of e US/Colombian (on spo) excange rae are used. Le er represen e Fridays excange raes, running from Ocober, 99, roug December 9 of 999; a made T = 48 observaions. Te daily excange rae corresponds o e weig average of rading, selling and buying, of U.S. dollars in e open marke. In e even of e marke being closed on a Friday, e observaion on e previous Tursday was used.
12 54 REVISTA DE ECONOMÍA DEL ROSARIO, 4 (DICIEMBRE ) Figure sows e raw daa; from is and wi e elp of a uni roo es, (See Enders, 995), i is easy o conclude a e excange rae as a uni roo, so, as usual, one works wi e firs difference of e naural logarim of excange raes, reurns, also known as coninuously compounded raes of reurn, (r = ln (er ) ln (er ) for =,..., T), wose represenaion is sown in Fig.. Figure Weekly Colombian excange rae From Figure 3, wic sows e isogram of reurns, i is noewory a e skewness and kurosis coefficiens for r wic are.858 and 5.687, respecively, are bo significanly posiive and muc larger an common, us sowing asymmery and lepokurosis. As poined ou by Vlaar and Palm (993), e skewness could be e resul of e asymmery in e movemens of e pariy adjusmens, and a ig kurosis could resul from a ime varying-variance. Tese wo resuls lead o considering ose disribuions differen from e normal, wose use is unlikely o yield appropriae resuls. Also, i is clear a e variance is no consan a all, wic can be seen from e Lunjg-Box s auocorrelaion saisic of e squared reurns: Q () = 8,3 and Q (4) = 44,86 for lags and 4, respecively.
13 RODRÍGUEZ: BAYESIAN ESTIMATION AND MODEL SELECTION FOR THE WEEKLY COLOMBIAN EXCHANGE RATE 55 B. Esimaion Based on e eoreical jusificaion of several models and aken in accoun e paricular Colombian economy, seven differen models were proposed for esimaion and selecion among em; Tey are described below..8 Figure Weekly reurn from colombian excange rae Wi e aim o implemen a fully Bayesian analysis, MCMCs were used o esimae e models, muc of em are esimaed by e firs ime from e Bayesian poin of view, wic is a new developmen of is work. Firs, an ARMA(,) model wi normal disurbances was fied; is model could be ried in pracice wen e scolar ignores e fac a e variance is no consan. In is repor, suc a model is also used as e reference model, agains wic oers are o be compared. Non-informaive Bea(,) from - o 3 priors were cosen for and θ. A N(,5) as prior for was used; finally, a nearly noninformaive bu proper Inverse-Gamma(.,.), (So, Expeced value and variance equal o /) was used for σ. 4. Maximum Likeliood Esimaion (MLE) was ried, bu problems geing precise esimaions of e variance-covariance marix, for some models, resriced is use. 3. In e sense a if X ~ Bea(,) en Y = X ~ Bea(,) from o. See Jaquier e al (999) 4. /σ as a prior for σ was ried oo, wi similar resuls
14 56 REVISTA DE ECONOMÍA DEL ROSARIO, 4 (DICIEMBRE ) 6 Figure 3 Hisogram of reurns. 4 # Frequency Midpoin Second, an ARMA(,) for e levels of reurn plus a GARCH(,) for e variance, so, c in () comes from an ARMA(,), and V(ε ) = from (3) wi r =, and s = ; sill assuming normaliy, e likeliood funcion is given by (). Noninformaive U(,) prior for and θ ; /α for α ; a N(,5) for α ; U(,) for α, and β, and in order o assure saionariy in (3), any proposal a did no fi α + β < is rejeced. Te same priors on as before was used. From iniial runs urned ou o be saisically equal o zero, us i is se o α / ( α ) as Nelson and Cao (99) propose for compuaional convenience. Tird, an ARMA(,) plus a GARCH(,) wi Suden- disribuion, and α + β >= are rejeced in (3) for saionariy purposes; n > in (4); was se as in before. Priors: /α for α ; /(n ) on n and α, β as in e second model,, and θ were reaed as in e firs model. Four, an ARMA(,) plus an EGARCH(,), insead of wi GED disribuion, erefore, e likeliood is given by (6). Priors:,, and θ as above; a N(,) was used as prior for α ; N(,5) for α for β an U(-,); ϕ is assumed idenically o zero; and U(,8) for υ. Any proposal wi β > was rejeced for saionariy purposes. Fif, an ARMA(,) plus an GARCH(,) wi a Mixure of wo normals as e error erm disribuion using (8) as e likeliood. Priors:,, and θ as before. /α for α for α, and β U(,); N(,) for υ; /(δ ) for δ ; finally, U(,) for λ. Six, an ARMA(,) plus an GARCH(,) wi Type-IV Pearson disribuion, ence, e likeliood is compued by using (). Priors: Inverse-Gamma(.,.) was used for α, U(,) for α, and β ; (r ) following an Inverse-Gamma (.,.); N(,) for δ and N(,5) for µ. Te disribuions used previously for,, and θ were used ere oo. Te resricion r > 3 is imposed, so e firs four momens from e Type-IV Pearson disribuion exis. Again, any proposal wi α + β >= is rejeced. α <. were rejeced for compuaional reasons.
15 RODRÍGUEZ: BAYESIAN ESTIMATION AND MODEL SELECTION FOR THE WEEKLY COLOMBIAN EXCHANGE RATE 57 Seven, and finally, a Socasic Volailiy model wi a Suden- disribuion on e error erm, expressed in (6) and (5). Non-informaive prior U(,) for, a N(,5) for γ; Inverse-Gamma(.,.) for σ ; and for υ a U(5,8) was used as η Jaquier e al. (999) did o assure e -Suden as a leas e firs-four momens. Aloug e use of single versus muliple (parallel) cains in MCMC is an open discussion, single cains for eac model are used in is job because of ime and compuing resources limiaions. For a recen discussion of is dilemma, e reader is referred o Mengersen e al. (999), wic conains poins in favor of eac alernaive. Rafery and Lewis (996) sraegy was implemened ere wi consan seleced in suc a way a e proporions of e proposals acceped were beween and 5%, as as become common pracice. Te updae is elemen-by-elemen and in random order. However, wen ig correlaions beween parameers in conjuncion wi slow convergence were found e blocking updae was implemened o improve convergence. In every case, a final cain of 8, was run and en seps of 5 o 3 were aken o avoid large auocorrelaions in e cains. In a way, firs-order auocorrelaions no larger an.55 were guaraneed. Convergence of eac cain is assessed by applying e Geweke s (99) crierion, wic null ypoesis is a saionariy as been reaced, and e es-saisic is suppose o follow a sandard normal disribuion under H. Tis es is implemened by using CODA (See Bes e al, 997). Te mean-vecor and Variance-covariance marices are o be obained in order o feed or implemen e RJMCMC, as explained laer. C. Model Selecion Te model selecion exercise consiss of applying e Reversible Jump MCMC algorim and e poserior probabiliies, running, ieraions and sowing e proporion of eac, a model m (m=,,... 7) is seleced. For cecking sabiliy, visual analysis is used. Aloug ere are some fres resuls abou assessing convergence in RJMCMC, eir value are no well-known ye, as menioned by Brooks and Guidici (999). For all seven models e same priors menioned in Secion B are o be used. Te proposal densiies q(u/m ) and q(u /m) for eac parameer were consruced by using e MCMC oupu of e separae model runs described above. Tese densiies are aken as mulivariae normals wi mean vecors, consising of e sample mean values and covariance marix equal o e corresponding sample mean vecor and covariance marix of e parameers in eac model.
16 58 REVISTA DE ECONOMÍA DEL ROSARIO, 4 (DICIEMBRE ) IV. RESULTS A. Esimaion 5 In e following presenaion e reurn raes are expressed in - scale, wic was used because of compuaional and presenaional avenues; oerwise, models wic use GARCH componen ge suck, i seems because values for go so close o zero a e algorim ge overwelmed. Table Geweke s convergence z scores for e seven models Model θ α α β σ n ARMA(,)Normal ARMA(,)+ GARCH(,):N ARMA(,)+ GARCH(,): Model θ α α β v ARMA(,)+ EGARCH(,):GED Model θ α α β λ σ n ARMA(,)+ GARCH(,):MixN Model θ α α β r σ n ARMA(,)+ GARCH(,):T-IV P Model γ υ σ η STOCHASTIC VOL.: Te Geweke s Convergence z Scores for e seven models are presened in Table, looking o e numbers i is clear a convergence as been reaced for almos every parameer in all models. Te esimaion resuls are presened in Table wi sandard errors in pareneses. 6 Aloug, parameer ransformaions 7 were ried for some models, convergence was no improved, ence i use was discarged. From Table i sould 5. I ook beween ours 4 minues and 56 ours and 48 minues, from e fases o e slowes model, o made all e ieraions, using a compuer wi a Penium I 33 MHz processor, and 64 MB RAM, running under WINDOWS- 98 Second Ediion. Times reduce o one ird using a Penium III 7 MHz processor, wi e same sofware and 9 MB RAM.
17 RODRÍGUEZ: BAYESIAN ESTIMATION AND MODEL SELECTION FOR THE WEEKLY COLOMBIAN EXCHANGE RATE 59 be said a excep for e non-inclusion of some parameers in mos of e models, no addiional work for e exclusion of non-significan parameers in any model was aemped because ime limiaions, and because i is no e main purpose of is work o improve every and/or one specific model. Fig. 4 presens e resuling cains diagrams and e isograms of e poserior sample of e parameers of e ARMA(,) model. Te sape of e poserior disribuion of and σ parameers indicae asymmery, ence deviaion from normaliy. Figs. 5 o do e same for e GARCH-N,GARCH-, EGARCH- GED, GARCH-MixN and GARCH Type-IV Pearson models, respecively. Table Esimaion resuls of e seven compeing models Model θ α α β σ n One (6.6e-3) (.) (.) (3-E-3) Two.95 Two (.4) (.79) (.6) (.4) (.4) (.4) Tree.53 Tree (.5) (.4) (.) (.8) (.) (.6) (.6) Model θ α α β v Four (.3) (.3) (.9) (.9) (.7) (.9) (.5) Model θ α α β λ υ δ Five (.3) (.5) (.44) (.9) (.) (.7) (.5) (.46) (.8) Model θ α α β r δ µ Six e-3..9e e-3 (.9) (-5) (.3) (3.8e-4) (.e-5) (.8e-5) (8.e-3) (4.3e-3) (7.4e-4) Model γ υ σ η Seven (.3) (.9) (.53) (.6) B. Model Selecion Te processing ime for, ieraions, using e same compuer as for esimaion, was 6 ours and 3 minues, for all e seven models, a oal of 47 parameers wic use above 3 MB of disk-space. Noe a, according o Fig., is is a conservaive run leng, and less an one-four of e run could be sufficien o acieve e same poserior disribuions. 6. Suc sandard errors refer o e ime-series esimaes wic are asympoic, e square roo of e specral densiy esimae divided by e sample size. 7. Like logarim or = ln(+)/( ), wen < ence R
18 6 REVISTA DE ECONOMÍA DEL ROSARIO, 4 (DICIEMBRE ) Figure 4 ARMA(,) Model θ σ Ieraions
19 RODRÍGUEZ: BAYESIAN ESTIMATION AND MODEL SELECTION FOR THE WEEKLY COLOMBIAN EXCHANGE RATE 6 Figure 5 GARCH-Normal Model θ α α β Ieraions
20 6 REVISTA DE ECONOMÍA DEL ROSARIO, 4 (DICIEMBRE ) Figure 6 GARCH- Model θ α α β η Ieraions
DOES TRADING VOLUME INFLUENCE GARCH EFFECTS? SOME EVIDENCE FROM THE GREEK MARKET WITH SPECIAL REFERENCE TO BANKING SECTOR
Invesmen Managemen and Financial Innovaions, Volume 4, Issue 3, 7 33 DOES TRADING VOLUME INFLUENCE GARCH EFFECTS? SOME EVIDENCE FROM THE GREEK MARKET WITH SPECIAL REFERENCE TO BANKING SECTOR Ahanasios
More informationDYNAMIC MODELS FOR VALUATION OF WRONGFUL DEATH PAYMENTS
DYNAMIC MODELS FOR VALUATION OF WRONGFUL DEATH PAYMENTS Hong Mao, Shanghai Second Polyechnic Universiy Krzyszof M. Osaszewski, Illinois Sae Universiy Youyu Zhang, Fudan Universiy ABSTRACT Liigaion, exper
More informationSPEC model selection algorithm for ARCH models: an options pricing evaluation framework
Applied Financial Economics Leers, 2008, 4, 419 423 SEC model selecion algorihm for ARCH models: an opions pricing evaluaion framework Savros Degiannakis a, * and Evdokia Xekalaki a,b a Deparmen of Saisics,
More informationTEMPORAL PATTERN IDENTIFICATION OF TIME SERIES DATA USING PATTERN WAVELETS AND GENETIC ALGORITHMS
TEMPORAL PATTERN IDENTIFICATION OF TIME SERIES DATA USING PATTERN WAVELETS AND GENETIC ALGORITHMS RICHARD J. POVINELLI AND XIN FENG Deparmen of Elecrical and Compuer Engineering Marquee Universiy, P.O.
More informationTerm Structure of Prices of Asian Options
Term Srucure of Prices of Asian Opions Jirô Akahori, Tsuomu Mikami, Kenji Yasuomi and Teruo Yokoa Dep. of Mahemaical Sciences, Risumeikan Universiy 1-1-1 Nojihigashi, Kusasu, Shiga 525-8577, Japan E-mail:
More informationJournal Of Business & Economics Research September 2005 Volume 3, Number 9
Opion Pricing And Mone Carlo Simulaions George M. Jabbour, (Email: jabbour@gwu.edu), George Washingon Universiy Yi-Kang Liu, (yikang@gwu.edu), George Washingon Universiy ABSTRACT The advanage of Mone Carlo
More informationVector Autoregressions (VARs): Operational Perspectives
Vecor Auoregressions (VARs): Operaional Perspecives Primary Source: Sock, James H., and Mark W. Wason, Vecor Auoregressions, Journal of Economic Perspecives, Vol. 15 No. 4 (Fall 2001), 101-115. Macroeconomericians
More informationThe Relationship between Stock Return Volatility and. Trading Volume: The case of The Philippines*
The Relaionship beween Sock Reurn Volailiy and Trading Volume: The case of The Philippines* Manabu Asai Faculy of Economics Soka Universiy Angelo Unie Economics Deparmen De La Salle Universiy Manila May
More informationDuration and Convexity ( ) 20 = Bond B has a maturity of 5 years and also has a required rate of return of 10%. Its price is $613.
Graduae School of Business Adminisraion Universiy of Virginia UVA-F-38 Duraion and Convexiy he price of a bond is a funcion of he promised paymens and he marke required rae of reurn. Since he promised
More informationModelling and Forecasting Volatility of Gold Price with Other Precious Metals Prices by Univariate GARCH Models
Deparmen of Saisics Maser's Thesis Modelling and Forecasing Volailiy of Gold Price wih Oher Precious Meals Prices by Univariae GARCH Models Yuchen Du 1 Supervisor: Lars Forsberg 1 Yuchen.Du.84@suden.uu.se
More informationPROFIT TEST MODELLING IN LIFE ASSURANCE USING SPREADSHEETS PART ONE
Profi Tes Modelling in Life Assurance Using Spreadshees PROFIT TEST MODELLING IN LIFE ASSURANCE USING SPREADSHEETS PART ONE Erik Alm Peer Millingon 2004 Profi Tes Modelling in Life Assurance Using Spreadshees
More informationHierarchical Mixtures of AR Models for Financial Time Series Analysis
Hierarchical Mixures of AR Models for Financial Time Series Analysis Carmen Vidal () & Albero Suárez (,) () Compuer Science Dp., Escuela Poliécnica Superior () Risklab Madrid Universidad Auónoma de Madrid
More informationMeasuring macroeconomic volatility Applications to export revenue data, 1970-2005
FONDATION POUR LES ETUDES ET RERS LE DEVELOPPEMENT INTERNATIONAL Measuring macroeconomic volailiy Applicaions o expor revenue daa, 1970-005 by Joël Cariolle Policy brief no. 47 March 01 The FERDI is a
More informationGOOD NEWS, BAD NEWS AND GARCH EFFECTS IN STOCK RETURN DATA
Journal of Applied Economics, Vol. IV, No. (Nov 001), 313-37 GOOD NEWS, BAD NEWS AND GARCH EFFECTS 313 GOOD NEWS, BAD NEWS AND GARCH EFFECTS IN STOCK RETURN DATA CRAIG A. DEPKEN II * The Universiy of Texas
More informationCointegration: The Engle and Granger approach
Coinegraion: The Engle and Granger approach Inroducion Generally one would find mos of he economic variables o be non-saionary I(1) variables. Hence, any equilibrium heories ha involve hese variables require
More informationThis document is downloaded from DR-NTU, Nanyang Technological University Library, Singapore.
This documen is downloaded from DR-NTU, Nanyang Technological Universiy Library, Singapore. Tile A Bayesian mulivariae risk-neural mehod for pricing reverse morgages Auhor(s) Kogure, Asuyuki; Li, Jackie;
More informationMeasuring the Downside Risk of the Exchange-Traded Funds: Do the Volatility Estimators Matter?
Proceedings of he Firs European Academic Research Conference on Global Business, Economics, Finance and Social Sciences (EAR5Ialy Conference) ISBN: 978--6345-028-6 Milan-Ialy, June 30-July -2, 205, Paper
More informationThe Real Business Cycle paradigm. The RBC model emphasizes supply (technology) disturbances as the main source of
Prof. Harris Dellas Advanced Macroeconomics Winer 2001/01 The Real Business Cycle paradigm The RBC model emphasizes supply (echnology) disurbances as he main source of macroeconomic flucuaions in a world
More informationDependent Interest and Transition Rates in Life Insurance
Dependen Ineres and ransiion Raes in Life Insurance Krisian Buchard Universiy of Copenhagen and PFA Pension January 28, 2013 Absrac In order o find marke consisen bes esimaes of life insurance liabiliies
More informationLIFE INSURANCE WITH STOCHASTIC INTEREST RATE. L. Noviyanti a, M. Syamsuddin b
LIFE ISURACE WITH STOCHASTIC ITEREST RATE L. oviyani a, M. Syamsuddin b a Deparmen of Saisics, Universias Padjadjaran, Bandung, Indonesia b Deparmen of Mahemaics, Insiu Teknologi Bandung, Indonesia Absrac.
More informationThe Economic Value of Volatility Timing Using a Range-based Volatility Model
The Economic Value of Volailiy Timing Using a Range-based Volailiy Model Ray Yeuien Chou * Insiue of Economics, Academia Sinica & Insiue of Business Managemen, Naional Chiao Tung Universiy Nahan Liu Deparmen
More informationSoftware implementation and testing of GARCH models
Sofware implemenaion and esing of GARCH models G F LEVY NAG Ld, Wilkinson House, Jordan Hill Road, Oxford, OX2 8DR, U.K. (email: george@nag.co.uk) Absrac: This paper describes he sofware implemenaion of
More informationSupplementary Appendix for Depression Babies: Do Macroeconomic Experiences Affect Risk-Taking?
Supplemenary Appendix for Depression Babies: Do Macroeconomic Experiences Affec Risk-Taking? Ulrike Malmendier UC Berkeley and NBER Sefan Nagel Sanford Universiy and NBER Sepember 2009 A. Deails on SCF
More informationStochastic Optimal Control Problem for Life Insurance
Sochasic Opimal Conrol Problem for Life Insurance s. Basukh 1, D. Nyamsuren 2 1 Deparmen of Economics and Economerics, Insiue of Finance and Economics, Ulaanbaaar, Mongolia 2 School of Mahemaics, Mongolian
More informationINTEREST RATE FUTURES AND THEIR OPTIONS: SOME PRICING APPROACHES
INTEREST RATE FUTURES AND THEIR OPTIONS: SOME PRICING APPROACHES OPENGAMMA QUANTITATIVE RESEARCH Absrac. Exchange-raded ineres rae fuures and heir opions are described. The fuure opions include hose paying
More informationHow Useful are the Various Volatility Estimators for Improving GARCH-based Volatility Forecasts? Evidence from the Nasdaq-100 Stock Index
Inernaional Journal of Economics and Financial Issues Vol. 4, No. 3, 04, pp.65-656 ISSN: 46-438 www.econjournals.com How Useful are he Various Volailiy Esimaors for Improving GARCH-based Volailiy Forecass?
More informationA Generalized Bivariate Ornstein-Uhlenbeck Model for Financial Assets
A Generalized Bivariae Ornsein-Uhlenbeck Model for Financial Asses Romy Krämer, Mahias Richer Technische Universiä Chemniz, Fakulä für Mahemaik, 917 Chemniz, Germany Absrac In his paper, we sudy mahemaical
More informationA Note on Using the Svensson procedure to estimate the risk free rate in corporate valuation
A Noe on Using he Svensson procedure o esimae he risk free rae in corporae valuaion By Sven Arnold, Alexander Lahmann and Bernhard Schwezler Ocober 2011 1. The risk free ineres rae in corporae valuaion
More informationThe naive method discussed in Lecture 1 uses the most recent observations to forecast future values. That is, Y ˆ t + 1
Business Condiions & Forecasing Exponenial Smoohing LECTURE 2 MOVING AVERAGES AND EXPONENTIAL SMOOTHING OVERVIEW This lecure inroduces ime-series smoohing forecasing mehods. Various models are discussed,
More informationDYNAMIC ECONOMETRIC MODELS Vol. 7 Nicolaus Copernicus University Toruń 2006. Ryszard Doman Adam Mickiewicz University in Poznań
DYNAMIC ECONOMETRIC MODELS Vol. 7 Nicolaus Copernicus Universiy Toruń 26 1. Inroducion Adam Mickiewicz Universiy in Poznań Measuring Condiional Dependence of Polish Financial Reurns Idenificaion of condiional
More informationAPPLICATION OF THE KALMAN FILTER FOR ESTIMATING CONTINUOUS TIME TERM STRUCTURE MODELS: THE CASE OF UK AND GERMANY. January, 2005
APPLICATION OF THE KALMAN FILTER FOR ESTIMATING CONTINUOUS TIME TERM STRUCTURE MODELS: THE CASE OF UK AND GERMANY Somnah Chaeree* Deparmen of Economics Universiy of Glasgow January, 2005 Absrac The purpose
More informationMarket Liquidity and the Impacts of the Computerized Trading System: Evidence from the Stock Exchange of Thailand
36 Invesmen Managemen and Financial Innovaions, 4/4 Marke Liquidiy and he Impacs of he Compuerized Trading Sysem: Evidence from he Sock Exchange of Thailand Sorasar Sukcharoensin 1, Pariyada Srisopisawa,
More informationAppendix D Flexibility Factor/Margin of Choice Desktop Research
Appendix D Flexibiliy Facor/Margin of Choice Deskop Research Cheshire Eas Council Cheshire Eas Employmen Land Review Conens D1 Flexibiliy Facor/Margin of Choice Deskop Research 2 Final Ocober 2012 \\GLOBAL.ARUP.COM\EUROPE\MANCHESTER\JOBS\200000\223489-00\4
More informationThe dynamics of international asset price linkages and their effects on German stock and bond markets
Te dynamics of inernaional asse price linkages and eir effecs on German sock and bond markes Dieric Domanski and Manfred Kremer 1 1. Inroducion Te financial marke urbulences in 1998, as oer crises previously,
More informationTHE NEW MARKET EFFECT ON RETURN AND VOLATILITY OF SPANISH STOCK SECTOR INDEXES
THE NEW MARKET EFFECT ON RETURN AND VOLATILITY OF SPANISH STOCK SECTOR INDEXES Juan Ángel Lafuene Universidad Jaume I Unidad Predeparamenal de Finanzas y Conabilidad Campus del Riu Sec. 1080, Casellón
More informationApplied Mathematical Modelling
Applied Maemaical Modelling 36 () 347 358 Conens liss available a SciVerse ScienceDirec Applied Maemaical Modelling journal omepage: www.elsevier.com/locae/apm Recursive sae esimaion for ybrid sysems Evgenia
More informationSkewness and Kurtosis Adjusted Black-Scholes Model: A Note on Hedging Performance
Finance Leers, 003, (5), 6- Skewness and Kurosis Adjused Black-Scholes Model: A Noe on Hedging Performance Sami Vähämaa * Universiy of Vaasa, Finland Absrac his aricle invesigaes he dela hedging performance
More informationRisk Modelling of Collateralised Lending
Risk Modelling of Collaeralised Lending Dae: 4-11-2008 Number: 8/18 Inroducion This noe explains how i is possible o handle collaeralised lending wihin Risk Conroller. The approach draws on he faciliies
More informationReal-time Particle Filters
Real-ime Paricle Filers Cody Kwok Dieer Fox Marina Meilă Dep. of Compuer Science & Engineering, Dep. of Saisics Universiy of Washingon Seale, WA 9895 ckwok,fox @cs.washingon.edu, mmp@sa.washingon.edu Absrac
More informationA PROPOSAL TO OBTAIN A LONG QUARTERLY CHILEAN GDP SERIES *
CUADERNOS DE ECONOMÍA, VOL. 43 (NOVIEMBRE), PP. 285-299, 2006 A PROPOSAL TO OBTAIN A LONG QUARTERLY CHILEAN GDP SERIES * JUAN DE DIOS TENA Universidad de Concepción y Universidad Carlos III, España MIGUEL
More informationAN ECONOMETRIC CHARACTERIZATION OF BUSINESS CYCLE DYNAMICS WITH FACTOR STRUCTURE AND REGIME SWITCHING * Marcelle Chauvet 1
AN ECONOMETRIC CHARACTERIZATION OF BUSINESS CYCLE DYNAMICS WITH FACTOR STRUCTURE AND REGIME SWITCHING * Marcelle Chauve Deparmen of Economics Universiy of California, Riverside 5 Universiy Avenue Riverside,
More informationHedging with Forwards and Futures
Hedging wih orwards and uures Hedging in mos cases is sraighforward. You plan o buy 10,000 barrels of oil in six monhs and you wish o eliminae he price risk. If you ake he buy-side of a forward/fuures
More informationOptimal Stock Selling/Buying Strategy with reference to the Ultimate Average
Opimal Sock Selling/Buying Sraegy wih reference o he Ulimae Average Min Dai Dep of Mah, Naional Universiy of Singapore, Singapore Yifei Zhong Dep of Mah, Naional Universiy of Singapore, Singapore July
More informationCrude Oil Hedging Strategies Using Dynamic Multivariate GARCH
Crude Oil Hedging Sraegies Using Dynamic Mulivariae GARCH Roengchai Tansucha * Faculy of Economics Maejo Universiy Chiang Mai, Thailand Chia-Lin Chang Deparmen of Applied Economics Naional Chung Hsing
More informationStock Price Prediction Using the ARIMA Model
2014 UKSim-AMSS 16h Inernaional Conference on Compuer Modelling and Simulaion Sock Price Predicion Using he ARIMA Model 1 Ayodele A. Adebiyi., 2 Aderemi O. Adewumi 1,2 School of Mahemaic, Saisics & Compuer
More informationII.1. Debt reduction and fiscal multipliers. dbt da dpbal da dg. bal
Quarerly Repor on he Euro Area 3/202 II.. Deb reducion and fiscal mulipliers The deerioraion of public finances in he firs years of he crisis has led mos Member Saes o adop sizeable consolidaion packages.
More informationTESTING NONLINEARITIES BETWEEN BRAZILIAN EXCHANGE RATE AND INFLATION VOLATILITIES *
TESTING NONLINEARITIES ETWEEN RAZILIAN EXCHANGE RATE AND INFLATION VOLATILITIES * Crisiane R. Albuquerque ** Marcelo S. Porugal *** Absrac Tere are few sudies direcly addressing excange rae and inflaion
More informationCALENDAR ANOMALIES IN EMERGING BALKAN EQUITY MARKETS
INTERNATIONAL ECONOMICS & FINANCE JOURNAL Vol. 6, No. 1, January-June (2011) : 67-82 CALENDAR ANOMALIES IN EMERGING BALKAN EQUITY MARKETS Andreas G. Georganopoulos *, Dimiris F. Kenourgios ** and Anasasios
More informationANALYSIS AND COMPARISONS OF SOME SOLUTION CONCEPTS FOR STOCHASTIC PROGRAMMING PROBLEMS
ANALYSIS AND COMPARISONS OF SOME SOLUTION CONCEPTS FOR STOCHASTIC PROGRAMMING PROBLEMS R. Caballero, E. Cerdá, M. M. Muñoz and L. Rey () Deparmen of Applied Economics (Mahemaics), Universiy of Málaga,
More informationPrincipal components of stock market dynamics. Methodology and applications in brief (to be updated ) Andrei Bouzaev, bouzaev@ya.
Principal componens of sock marke dynamics Mehodology and applicaions in brief o be updaed Andrei Bouzaev, bouzaev@ya.ru Why principal componens are needed Objecives undersand he evidence of more han one
More informationA DCC Analysis of Two Stock Market Returns Volatility with an Oil Price Factor: An Evidence Study of Singapore and Thailand s Stock Markets
Journal of Convergence Informaion Technology Volume 4, Number 1, March 9 A DCC Analysis of Two Sock Marke Reurns Volailiy wih an Oil Price Facor: An Evidence Sudy of Singapore and Thailand s Sock Markes
More informationWhy Did the Demand for Cash Decrease Recently in Korea?
Why Did he Demand for Cash Decrease Recenly in Korea? Byoung Hark Yoo Bank of Korea 26. 5 Absrac We explores why cash demand have decreased recenly in Korea. The raio of cash o consumpion fell o 4.7% in
More informationA comparison of the Lee-Carter model and AR-ARCH model for forecasting mortality rates
A comparison of he Lee-Carer model and AR-ARCH model for forecasing moraliy raes Rosella Giacomei a, Marida Berocchi b, Svelozar T. Rachev c, Frank J. Fabozzi d,e a Rosella Giacomei Deparmen of Mahemaics,
More informationThe Greek financial crisis: growing imbalances and sovereign spreads. Heather D. Gibson, Stephan G. Hall and George S. Tavlas
The Greek financial crisis: growing imbalances and sovereign spreads Heaher D. Gibson, Sephan G. Hall and George S. Tavlas The enry The enry of Greece ino he Eurozone in 2001 produced a dividend in he
More informationMorningstar Investor Return
Morningsar Invesor Reurn Morningsar Mehodology Paper Augus 31, 2010 2010 Morningsar, Inc. All righs reserved. The informaion in his documen is he propery of Morningsar, Inc. Reproducion or ranscripion
More informationNikkei Stock Average Volatility Index Real-time Version Index Guidebook
Nikkei Sock Average Volailiy Index Real-ime Version Index Guidebook Nikkei Inc. Wih he modificaion of he mehodology of he Nikkei Sock Average Volailiy Index as Nikkei Inc. (Nikkei) sars calculaing and
More informationThe Transport Equation
The Transpor Equaion Consider a fluid, flowing wih velociy, V, in a hin sraigh ube whose cross secion will be denoed by A. Suppose he fluid conains a conaminan whose concenraion a posiion a ime will be
More informationA CASE OF US AND INDIA ABSTRACT
A CASE OF US AND INDIA K. Kiran Kumar * and Ciranji Mukopadyay ABSTRACT Tis paper empirically invesigaes e sor run dynamic linkages beween NSE Nify in India and NASDAQ Composie in US during e recen 999-00
More informationMeasuring the Effects of Monetary Policy: A Factor-Augmented Vector Autoregressive (FAVAR) Approach * Ben S. Bernanke, Federal Reserve Board
Measuring he Effecs of Moneary Policy: A acor-augmened Vecor Auoregressive (AVAR) Approach * Ben S. Bernanke, ederal Reserve Board Jean Boivin, Columbia Universiy and NBER Pior Eliasz, Princeon Universiy
More informationOption Put-Call Parity Relations When the Underlying Security Pays Dividends
Inernaional Journal of Business and conomics, 26, Vol. 5, No. 3, 225-23 Opion Pu-all Pariy Relaions When he Underlying Securiy Pays Dividends Weiyu Guo Deparmen of Finance, Universiy of Nebraska Omaha,
More informationEquity market interdependence: the relationship between European and US stock markets
Equiy marke inerdependence: he relaionship beween European and US sock markes SANVI AVOUYI-DOVI, DAVID NETO Direcorae General Economics and Inernaional Relaions Economic Analysis and Research Direcorae
More informationWorking Paper No. 482. Net Intergenerational Transfers from an Increase in Social Security Benefits
Working Paper No. 482 Ne Inergeneraional Transfers from an Increase in Social Securiy Benefis By Li Gan Texas A&M and NBER Guan Gong Shanghai Universiy of Finance and Economics Michael Hurd RAND Corporaion
More informationCredit Index Options: the no-armageddon pricing measure and the role of correlation after the subprime crisis
Second Conference on The Mahemaics of Credi Risk, Princeon May 23-24, 2008 Credi Index Opions: he no-armageddon pricing measure and he role of correlaion afer he subprime crisis Damiano Brigo - Join work
More informationCHARGE AND DISCHARGE OF A CAPACITOR
REFERENCES RC Circuis: Elecrical Insrumens: Mos Inroducory Physics exs (e.g. A. Halliday and Resnick, Physics ; M. Sernheim and J. Kane, General Physics.) This Laboraory Manual: Commonly Used Insrumens:
More informationARCH 2013.1 Proceedings
Aricle from: ARCH 213.1 Proceedings Augus 1-4, 212 Ghislain Leveille, Emmanuel Hamel A renewal model for medical malpracice Ghislain Léveillé École d acuaria Universié Laval, Québec, Canada 47h ARC Conference
More informationTime Series Analysis Using SAS R Part I The Augmented Dickey-Fuller (ADF) Test
ABSTRACT Time Series Analysis Using SAS R Par I The Augmened Dickey-Fuller (ADF) Tes By Ismail E. Mohamed The purpose of his series of aricles is o discuss SAS programming echniques specifically designed
More informationA Bayesian framework with auxiliary particle filter for GMTI based ground vehicle tracking aided by domain knowledge
A Bayesian framework wih auxiliary paricle filer for GMTI based ground vehicle racking aided by domain knowledge Miao Yu a, Cunjia Liu a, Wen-hua Chen a and Jonahon Chambers b a Deparmen of Aeronauical
More informationBayesian Filtering with Online Gaussian Process Latent Variable Models
Bayesian Filering wih Online Gaussian Process Laen Variable Models Yali Wang Laval Universiy yali.wang.1@ulaval.ca Marcus A. Brubaker TTI Chicago mbrubake@cs.orono.edu Brahim Chaib-draa Laval Universiy
More informationRandom Walk in 1-D. 3 possible paths x vs n. -5 For our random walk, we assume the probabilities p,q do not depend on time (n) - stationary
Random Walk in -D Random walks appear in many cones: diffusion is a random walk process undersanding buffering, waiing imes, queuing more generally he heory of sochasic processes gambling choosing he bes
More informationPARAMETRIC EXTREME VAR WITH LONG-RUN VOLATILITY: COMPARING OIL AND GAS COMPANIES OF BRAZIL AND USA.
Perspecivas Globais para a Engenharia de Produção Foraleza, CE, Brasil, 13 a 16 de ouubro de 015. PARAMETRIC EXTREME VAR WITH LONG-RUN VOLATILITY: COMPARING OIL AND GAS COMPANIES OF BRAZIL AND USA. RICARDO
More informationTask is a schedulable entity, i.e., a thread
Real-Time Scheduling Sysem Model Task is a schedulable eniy, i.e., a hread Time consrains of periodic ask T: - s: saring poin - e: processing ime of T - d: deadline of T - p: period of T Periodic ask T
More informationChapter 8: Regression with Lagged Explanatory Variables
Chaper 8: Regression wih Lagged Explanaory Variables Time series daa: Y for =1,..,T End goal: Regression model relaing a dependen variable o explanaory variables. Wih ime series new issues arise: 1. One
More informationKey Words: Steel Modelling, ARMA, GARCH, COGARCH, Lévy Processes, Discrete Time Models, Continuous Time Models, Stochastic Modelling
Vol 4, No, 01 ISSN: 1309-8055 (Online STEEL PRICE MODELLING WITH LEVY PROCESS Emre Kahraman Türk Ekonomi Bankası (TEB A.Ş. Direcor / Risk Capial Markes Deparmen emre.kahraman@eb.com.r Gazanfer Unal Yediepe
More informationBid-ask Spread and Order Size in the Foreign Exchange Market: An Empirical Investigation
Bid-ask Spread and Order Size in he Foreign Exchange Marke: An Empirical Invesigaion Liang Ding* Deparmen of Economics, Macaleser College, 1600 Grand Avenue, S. Paul, MN55105, U.S.A. Shor Tile: Bid-ask
More informationAlgorithmic trading strategy, based on GARCH (1, 1) volatility and volume weighted average price of asset
IOSR Journal of Business and Managemen (IOSR-JBM) ISSN: 78-87X. Volume, Issue (Sep-Oc. ), PP 3-35 Algorihmic rading sraegy, based on GARCH (, ) volailiy and volume weighed average price of asse Simranji
More informationA Two-Account Life Insurance Model for Scenario-Based Valuation Including Event Risk Jensen, Ninna Reitzel; Schomacker, Kristian Juul
universiy of copenhagen Universiy of Copenhagen A Two-Accoun Life Insurance Model for Scenario-Based Valuaion Including Even Risk Jensen, Ninna Reizel; Schomacker, Krisian Juul Published in: Risks DOI:
More informationA General Approach to Recovering Market Expectations from Futures Prices With an Application to Crude Oil
A General Approac o Recovering Marke Expecaions from Fuures Prices Wi an Applicaion o Crude Oil Augus 19, 2015 Crisiane Baumeiser Universiy of Nore Dame Luz Kilian Universiy of Micigan Fuures markes are
More informationAnalysis of I-Series, An Appraisal and Its Models
Vol. No.2, pp.-, June 203 MODELING TO ANTICIPATE WORLD PRICE OF EACH OUNCE OF GOLD IN INTERNATIONAL MARKETS Mohammad Rikhegar Business Managemen, MA Suden Islamic Azad Universiy, a Souh Tehran Branch 009893632406
More informationReview of Middle East Economics and Finance
Review of Middle Eas Economics and Finance Volume 4, Number 008 Aricle 3 Transiory and Permanen Volailiy s: The Case of he Middle Eas Sock Markes Bashar Abu Zarour, Universiy of Paras Cosas P. Siriopoulos,
More informationUNDERSTANDING THE DEATH BENEFIT SWITCH OPTION IN UNIVERSAL LIFE POLICIES. Nadine Gatzert
UNDERSTANDING THE DEATH BENEFIT SWITCH OPTION IN UNIVERSAL LIFE POLICIES Nadine Gazer Conac (has changed since iniial submission): Chair for Insurance Managemen Universiy of Erlangen-Nuremberg Lange Gasse
More informationChapter 7. Response of First-Order RL and RC Circuits
Chaper 7. esponse of Firs-Order L and C Circuis 7.1. The Naural esponse of an L Circui 7.2. The Naural esponse of an C Circui 7.3. The ep esponse of L and C Circuis 7.4. A General oluion for ep and Naural
More informationMonetary Policy & Real Estate Investment Trusts *
Moneary Policy & Real Esae Invesmen Truss * Don Bredin, Universiy College Dublin, Gerard O Reilly, Cenral Bank and Financial Services Auhoriy of Ireland & Simon Sevenson, Cass Business School, Ciy Universiy
More informationA COMPARISON OF FORECASTING MODELS FOR ASEAN EQUITY MARKETS
Sunway Academic Journal, 1 1 (005) A COMPARISON OF FORECASTING MODELS FOR ASEAN EQUITY MARKETS WONG YOKE CHEN a Sunway Universiy College KOK KIM LIAN b Universiy of Malaya ABSTRACT This paper compares
More informationDistributing Human Resources among Software Development Projects 1
Disribuing Human Resources among Sofware Developmen Proecs Macario Polo, María Dolores Maeos, Mario Piaini and rancisco Ruiz Summary This paper presens a mehod for esimaing he disribuion of human resources
More informationCOMPUTATION OF CENTILES AND Z-SCORES FOR HEIGHT-FOR-AGE, WEIGHT-FOR-AGE AND BMI-FOR-AGE
COMPUTATION OF CENTILES AND Z-SCORES FOR HEIGHT-FOR-AGE, WEIGHT-FOR-AGE AND BMI-FOR-AGE The mehod used o consruc he 2007 WHO references relied on GAMLSS wih he Box-Cox power exponenial disribuion (Rigby
More informationREVISTA INVESTIGACIÓN OPERACIONAL Vol., 30, No. 1, 11-19, 2009
REVISA INVESIGACIÓN OPERACIONAL Vol., 30, No. 1, 11-19, 009 MULIVARIAE RISK-REURN DECISION MAKING WIHIN DYNAMIC ESIMAION Josip Arnerić 1, Elza Jurun, and Snježana Pivac, 3 Universiy of Spl Faculy of Economics,
More informationPrice, Volume and Volatility Spillovers among New York, Tokyo and London Stock Markets
INTERNATIONAL JOURNAL OF BUSINESS, 4(), 999 ISSN: 083-4346 Price, Volume and Volailiy Spillovers among New York, Tokyo and London Sock Markes Sangphill Kim and Meng Rui The dynamic relaionship among he
More informationIterated importance sampling in missing data problems
Ieraed imporance sampling in missing daa problems Gilles Celeux INRIA, FUTURS, Orsay, France Jean-Michel Marin INRIA, FUTURS, Orsay, France and CEREMADE, Universiy Paris Dauphine, Paris, France Chrisian
More informationChapter 8 Student Lecture Notes 8-1
Chaper Suden Lecure Noes - Chaper Goals QM: Business Saisics Chaper Analyzing and Forecasing -Series Daa Afer compleing his chaper, you should be able o: Idenify he componens presen in a ime series Develop
More informationPresent Value Methodology
Presen Value Mehodology Econ 422 Invesmen, Capial & Finance Universiy of Washingon Eric Zivo Las updaed: April 11, 2010 Presen Value Concep Wealh in Fisher Model: W = Y 0 + Y 1 /(1+r) The consumer/producer
More informationThe stock index futures hedge ratio with structural changes
Invesmen Managemen and Financial Innovaions Volume 11 Issue 1 2014 Po-Kai Huang (Taiwan) The sock index fuures hedge raio wih srucural changes Absrac This paper esimaes he opimal sock index fuures hedge
More informationMTH6121 Introduction to Mathematical Finance Lesson 5
26 MTH6121 Inroducion o Mahemaical Finance Lesson 5 Conens 2.3 Brownian moion wih drif........................... 27 2.4 Geomeric Brownian moion........................... 28 2.5 Convergence of random
More informationA General Pricing Framework for No-Negative-Equity. Guarantees with Equity-release Products: A Theoretical and
A General Pricing Framework for No-Negaive-Equiy Guaranees wih Equiy-release Producs: A Theoreical and Empirical Sudy Jr-Wei Huang 1 Chuang-Chang Chang 2 Sharon S. Yang 3 ABSTRACT We invesigae sochasic
More informationSTUDYING THE VOLATILITY OF THE ROMANIAN INVESTMENT FUNDS WITH THE ARCH AND GARCH MODELS USING THE "R" SOFTWARE
STUDYING THE VOLATILITY OF THE ROMANIAN INVESTMENT FUNDS WITH THE ARCH AND GARCH MODELS USING THE "R" SOFTWARE Anoniade-Ciprian ALEXANDRU Ecological Universiy of Buchares, Faculy of Economics Absrac In
More informationTHE EFFECTS OF INTERNATIONAL ACCOUNTING STANDARDS ON STOCK MARKET VOLATILITY: THE CASE OF GREECE
Invesmen Managemen and Financial Innovaions, Volume 4, Issue 1, 007 61 THE EFFECTS OF INTERNATIONAL ACCOUNTING STANDARDS ON STOCK MARKET VOLATILITY: THE CASE OF GREECE Chrisos Floros * Absrac The adopion
More informationBayesian model comparison with un-normalised likelihoods
Saisics and Compuing manuscrip No. (will be insered by he edior) Bayesian model comparison wih un-normalised likelihoods Richard G. Everi Adam M. Johansen Ellen Rowing Melina Evdemon-Hogan Each of hese
More informationAsian Economic and Financial Review VOLATILITY MEAN REVERSION AND STOCK MARKET EFFICIENCY. Hojatallah Goudarzi
Asian Economic and Financial Review journal homepage: hp://aessweb.com/journal-deail.php?id=500 VOLATILITY MEAN REVERSION AND STOCK MARKET EFFICIENCY Hojaallah Goudarzi Deparmen of Finance and Insurance,
More informationAn Online Learning-based Framework for Tracking
An Online Learning-based Framework for Tracking Kamalika Chaudhuri Compuer Science and Engineering Universiy of California, San Diego La Jolla, CA 9293 Yoav Freund Compuer Science and Engineering Universiy
More informationThe Economic Value of Volatility Transmission between the Stock and Bond Markets
The Economic Value of Volailiy Transmission beween he Sock and Bond Markes Helena Chuliá * Hipòli Torró Sepember 006 Keywords: Volailiy Spillovers, GARCH, Trading Rules JEL Classificaion: C3, C53, G11
More informationMultiprocessor Systems-on-Chips
Par of: Muliprocessor Sysems-on-Chips Edied by: Ahmed Amine Jerraya and Wayne Wolf Morgan Kaufmann Publishers, 2005 2 Modeling Shared Resources Conex swiching implies overhead. On a processing elemen,
More information