ESTIMATES OF TECHNICAL INEFFICIENCY IN STOCHASTIC FRONTIER MODELS WITH PANEL DATA: GENERALIZED PANEL JACKKNIFE ESTIMATION

Size: px
Start display at page:

Download "ESTIMATES OF TECHNICAL INEFFICIENCY IN STOCHASTIC FRONTIER MODELS WITH PANEL DATA: GENERALIZED PANEL JACKKNIFE ESTIMATION"

Transcription

1 ESIMAES OF ECHNICAL INEFFICIENCY IN SOCHASIC FRONIER MODELS WIH PANEL DAA: GENERALIZED PANEL JACKKNIFE ESIMAION Panutat Satchacha Mchgan State Unversty Peter Schmdt Mchgan State Unversty Yonse Unversty Estmates of techncal neffcency based on fxed effects estmaton of the stochastc fronter model wth panel data are based upward. Prevous work has attempted to correct ths bas usng the bootstrap, but n smulatons the bootstrap corrects only part of the bas. he usual panel jackknfe s based on the assumpton that the bas s of order and s smlar to the bootstrap. / We show that when there s a te or a near te for the best frm, the bas s of order, not, and ths calls for a dfferent form of the jackknfe. he generalzed panel jackknfe s qute successful n removng the bas. However, the resultng estmates have a large varance. July 6, 009

2 . INRODUCION In ths paper we consder the stochastc fronter model wth tme-nvarant techncal neffcency n a panel data settng. hs model was frst consdered by Ptt and Lee 98), who estmated the model by MLE gven a dstrbutonal assumpton for techncal neffcency. Wthout such a dstrbutonal assumpton, Schmdt and Sckles 984) proposed fxed effects estmaton. In ths approach, the fronter ntercept s estmated as the maxmum of the estmated frm-specfc ntercepts, and a frm s level of neffcency s measured by the dfference between the fronter ntercept and the frm s ntercept. It s well understood that the max operaton causes the estmated fronter ntercept, and therefore the estmated neffcency levels, to be based upward. Schmdt and Sckles 984), Park and Smar 994) and Km, Km and Schmdt 007) dscuss ths problem. Hall, Härdle and Smar 995) show that the bootstrap s asymptotcally as wth N fxed) vald n ths settng, provded that there s a unque best frm no te for the largest populaton ntercept), and Km, Km and Schmdt 007) use the bootstrap to construct a bas-corrected estmate of the fronter ntercept and therefore of neffcency levels). he bootstrap s used to estmate the bas, whch s then subtracted from the orgnal estmate. In ther smulatons, Km, Km and Schmdt 007) found that the bas correcton was partally successful. It removed some but not all of the bas. Often t seemed to remove about half of the bas. Why t removed half of the bas, as opposed to some other fracton, s an nterestng puzzle. In ths paper we consder bas correctons based on the jackknfe. here are two motvatons for dong so. Our frst motvaton s that the jackknfe s thought to be smlar to the bootstrap, but t s analytcally smpler. herefore we use the jackknfe to explan why t s that under certan crcumstances we remove half of the bas, and so we at least partally resolve the

3 puzzle of the prevous paragraph. he second motvaton s to nvestgate whether the jackknfe s practcally useful as a bas-reducton technque n ths model. Here we are less successful, because the jackknfe does effectvely remove the bas of the estmate, but the varance and MSE of the jackknfe estmate are unfortunately rather large. he approprate form the jackknfe depends on the order of the leadng term n an expanson of the bas of the estmate. If the bas of the estmate s of order, the usual delete-one panel jackknfe estmator as n Hahn and Newey 004)) should remove the bas. However, ntutvely we would expect the jackknfe bas correcton to be smlar to the bootstrap bas correcton, whch was only partally successful. hus t would seem that the fnte-sample relevance of the bas beng of order may be questonable. In ths paper we analyze the case of an exact te for the best frm. In ths case the bootstrap s not asymptotcally vald. Furthermore, we show that the bas of the fxed effects estmate of the fronter ntercept s of order /, not. In ths case the usual delete-one panel jackknfe does not properly remove the bas. Indeed, we show that t removes approxmately) half of the bas. A dfferent form of the jackknfe, whch we call the generalzed panel jackknfe, does remove the bas n ths case. In the smulatons of Km, Km and Schmdt 007) there was not an exact te, and an exact te may also be unlkely n actual data. However, f there s nearly a te, n the sense that there s substantal uncertanty ex post about whch s the best frm, t s not clear whether asymptotcs that assume no te are more relevant than asymptotcs that assume an exact te. In order to further analyze a near te, we gve a specfc defnton nvolvng a local parameterzaton) of a near te, and we show that the bas s agan of order to successfully remove the bas. /, so that the generalzed panel jackknfe s needed 3

4 We then perform smulatons to assess the fnte-sample relevance of these results. he plan of the paper s as follows. In Secton, we defne some notaton and gve a bref revew of fxed effects estmaton of the stochastc fronter model wth panel data. In Secton 3 we show that the bas s of order / for the case of an exact te or a near te. Secton 4 descrbes the generalzed panel jackknfe that s approprate n ths crcumstance. In Secton 5 we explan the desgn of our Monte Carlo experments, and Secton 6 gves ts results. Fnally, Secton 7 contans our concludng remarks.. FIXED EFFECS ESIMAION OF HE MODEL Consder a sngle-output producton functon wth tme-nvarant techncal neffcency u 0. here are N frms, ndexed by,..., N, over tme perods, ndexed by t,...,. We consder the lnear regresson model of Schmdt and Sckles 984): y α + x β + v u,,..., N; t,...,, ) t t t where y t s the logarthm of output for frm at tme t ; x t s a vector of K nputs e.g., n logarthms for a Cobb-Douglas producton functon); β s a K vector of coeffcents; and v t s an..d. dosyncratc error wth mean zero and fnte varance. he v t represent uncontrollable shocks that affect the level of output, e.g., luck, weather, or machne performance. he tme-nvarant techncal neffcency u satsfes u 0 for all and u > 0 for some. here s no dstrbutonal assumpton on u except that t s one-sded. Defnng α α u, we can wrte ) as a standard panel data model: y α + x β + v. ) t t t 4

5 Obvously, α α snce u 0. When α and u ) s treatng as fxed, ) leads to a fxed effects estmaton problem n whch nether a dstrbuton for techncal neffcency nor the ndependence between techncal neffcency and x t or v t or both) s needed. We assume strct exogenety of the regressors x t n the sense that x,..., x ) s ndependent of v,..., v ). here s no restrcton on the dstrbuton of v t other than zero mean and fnte varance. o estmate β, we use the fxed effects estmate βˆ, whch can be estmated as least squares wth dummy varables, by regressng y t on x t and a set of N dummy varables, or as the wthn estmator, by regressng y y ) on x x ). Gven the estmate βˆ, the estmates t t can be recovered as the averages of the frm-specfc resduals,.e., ˆ α y x ˆ β where y t y t and x t x t, or equvalently as the coeffcents of the frm-specfc dummy varables. he wthn estmator βˆ s consstent as N or, and the frm-specfc ntercepts are consstent as. o estmate α and u, Schmdt and Sckles 984) suggested the followng estmators: ˆ α max ˆ α uˆ j, j,..., N ˆ α ˆ α,,..., N. 3) Park and Smar 994) show that these estmates are consstent as N,, and / ln N) 0. In ths paper, to mantan the connecton to the earler lterature on bootstrappng of ths model, and also the lterature on the jackknfe, we wll consder asymptotc arguments as wth N fxed. In ths case we can only measure neffcency relatve to the best of the N frms. 5

6 For ease of presentaton, we follow Km, Km and Schmdt 007) and rank the ntercepts α such that α α )... α N ), so that N) ndexes the frm wth the largest value of α among the N frms, whch we wll call the best frm. Smlarly, we rank the levels of techncal neffcency u n the opposte order such that )... N ) u u u. Obvously, α α u ) for all and specfcally α N ) α u N ). Now we defne the relatve neffcency measures u u u α α. 4) N ) N ) hese are the focus of ths paper snce, as wth N fxed, s a consstent estmate of α N ), not α, and u ˆ s a consstent estmate of u, not u. Although s consstent for α N ) as wth N fxed), t s based upward for fnte. hs s true because ˆ α ˆ α N ) and E ˆ α N ) ) α N ). hat s, the max operator n 3) nduces an upward bas: the largest s more lkely to contan postve estmaton error than negatve error. he upward bas n the estmate nduces an upward bas n the estmates of relatve techncal neffcency. hat s, E ˆ) α α N ) E u ) u ˆ. herefore we wll smply evaluate the bas of as an estmate of α N ) ; there s no need to separately evaluate the bas of the estmates of relatve techncal neffcency. he bas of as an estmate of α N ) corresponds to what Km, Km and Schmdt 007) call the frst-level bas. o correct ths frst-level bas, Km, Km and Schmdt 007) consder a bootstrap bas correcton for the fxed effects estmate. hey evaluate the second-level bas, boot E ˆ α ) ˆ α, and use t to correct the frst-level bas. hat s, f the second-level bas equals the frst-level bas, we would want to evaluate 6

7 boot boot ˆ α [ E ˆ α ) ˆ] α ˆ α E ˆ α ). 5) he feasble verson of ths s B b boot b) ˆ α ˆ ˆ BC α B α, 6) where b represents a sngle bootstrap replcaton and B s the total number of bootstrap replcatons. In ther smulatons see ther able 4), ths estmate removes some but not all of the bas n. Often t seems to remove about half of the bas. As noted n the ntroducton, the fact that half of the bas s removed s the puzzle that at least partally motvated our nterest n the jackknfe. 3. DERIVING HE ORDER IN PROBABILIY OF HE BIAS In ths secton, we show that the bas of s of order f there s no te for the best frm; that s, f α N ) s strctly larger than all of the other α. However, f there s a te for the best frm, or f there s a near te n a sense defned precsely below), the bas s of order For smplcty, we wll dscuss the smple case of no regressors: y /. α + v,,..., N; t,...,, 7) t t where v t are..d. wth mean zero and varance σ. hus y. he varous are ndependent and ˆ α α ) N0, σ ). However, the ncluson of regressors would not alter our results snce the wthn estmator of β s unbased, and our results really only requre that the vector whose th element s α ) ˆ α s normal wth mean zero and fnte varance matrx. See Hall, Härdle and Smar 995), Appendx ), equaton A.) for ths condton, whch would stll hold wth regressors. 7

8 3. he Case of No e that α α Suppose frst that there s no te for the best frm. hat s, there s a unque frm such N ). Hall, Härdle and Smar 995) show the equvalence of ) there s no te for the best frm, and ) the asymptotc dstrbuton of s normal. More precsely, they show that f there s no te, P ˆ α ˆ α ) as, so that the asymptotc dstrbuton of s the same as the N ) ˆ N asymptotc dstrbuton of α, the estmate of α N ) that would be used f the dentty of the best frm were known. Snce N ) s unbased, t follows that tmes the bas of must go to zero as. hus we conclude that the bas of s of an order smaller than /. We presume that t s of order. 3. he Case of an Exact e Suppose now that there s a te for the best frm the largest α ). Specfcally suppose that the frst k frms are ted, so that α N ) α α... α k for k N. Agan the dscusson n Hall, Härdle and Smar 995, Appendx )) apples. Wth a probablty that approaches one as, wll equal for some wth k, that s, the estmated best frm wll be one of the k truly best frms. herefore wth a probablty that approaches one, ˆ α α N ) ) max ˆ α α max{ ˆ α α N ) N ), ˆ α α ), N ) ˆ α α,..., ˆ α α k N ) ),..., N ) ) ˆ α k α N ) )} 8) 8

9 and therefore ˆ α N ) Z where Z s the maxmum of a set of k normals wth zero mean. α For k >, Z s not normal, and E Z) > 0. he bas of s therefore, for large, / E Z), whch s of order /. We can gve an explct expresson for the case of N k and the smple model above wth no regressors). We frst state the followng Lemma. Lemma Suppose X and X are..d. N μ,σ ),.e., X X ~ μ σ N, μ 0 0, σ then E [max X, X ) μ ] π ) σ. 9) Proof. Let Y Z X X X ~ μ σ N, 0 σ σ. σ So, ρ σ σ σ and E X X > X ) E Y Z > 0) μ + ) σλ0), where λ ) s the normal hazard functon μ + ) σ π ), snce λ0) φ0) Φ0)) π μ + π ) σ. Hence, E X X > X ) μ π ) σ and 9

10 E[max X, X )] E X X > X ) + E X X > X ), by symmetry E X X > X ), snce X and X are..d. herefore, bas E[max X, X )] μ π ) σ. In the present settng, X and X are ˆα and ˆα, μ α α, the varance s σ, and the bas of α max ˆ α, ˆ α ) equals π ) / σ. Clearly, ths s proportonal to /. ˆ σ 3.3 he Case of a Near e In the prevous sectons we saw that the bas of s of order best frm, whle t s of order f there s no te for the / f there s an exact te. It s not clear how relevant ether set of results wll be n fnte samples f there s n some sense) nearly a te. Intutvely that wll depend on how close we are to a te, whch depends not only on how close the α are to each other, but also on / σ, whch s the standard devaton of the. One way to model ths s by a local to te parameterzaton. So, to keep thngs smple, let N, α > α, and α / α c for c > 0, where c does not depend on. hen n our smple no regressors) model, ˆ α α ) N0, ). Also ˆ α α ) N0, ) and so σ / ˆ α α + c) N0, σ ), or ˆ α α) N c, σ ). hen σ [max ˆ α ˆ ) ] max[ ˆ ), ˆ, α α α α α α)] Z 0) 0

11 where Z s the max of a N 0, σ ) random varable and a N c, σ ) random varable. Clearly E Z) E N0, σ )) 0 and the bas of s agan for large ) / E Z), whch s of order /. A smlar analyss apples f α α γ c where c > 0 and γ. he value of c matters as above) when γ but t does not affect the lmt dstrbuton f γ >. So the asymptotcs for the case of a near te are very smlar to those for an exact te f a te s near enough. Once agan we can gve an explct expresson for the case of N k and the smple model no regressors). We state wthout proof the followng Lemma. Lemma Let X and X be ndependent normals, where X ~ N0, σ ) and X ~ N μ, σ ). hen where μ μ σ. E [max X, X )] [ Φ μ ) μ + φ μ )] σ, ) o apply ths to our model, X and X are ˆα and ˆα, σ s σ, μ / c, and / μ c / σ c. σ So the bas s whch s ndeed proportonal to / bas [ Φ c σ ) c σ ) + φ c σ )] σ, ) /. 4. CORRECING BIAS WIH HE PANEL JACKKNIFE AND HE GENERALIZED PANEL JACKKNIFE

12 4. he Panel Jackknfe Jackknfe estmaton s an automatc bas reducton tool under the assumpton of the exstence of a seres expanson for the bas of an estmator. Quenoulle 956) and ukey 958) show that usng the jackknfe estmates based on removng data and then recalculatng the estmator removes the frst order bas from an ntal estmator. For a basc background dscusson of jackknfe estmaton, see Mller 974). o descrbe the jackknfe n a general settng, let the data be ndexed by t,,...,. Let ˆ be the estmator based on all observatons, and let be the delete-observaton-t estmator that omts observaton t and uses the other observatons. hen the jackknfe estmator s hs estmator s sad to remove the bas of order hen So f the bas s of order. ˆ t ˆ J ˆ) ) ˆ. 3) t t ), n the followng sense. Suppose that 3 E ˆ) + B + D + O ). 4) E[ J ˆ)] + D + O ) + O ). 5), n the sense that 4) holds, the jackknfe leaves only the bas of order Hahn and Kuerstener 004), Hahn and Newey 004), and Fernández-Val and Vella 007) apply the jackknfe to nonlnear panel data models and dynamc panel data models. In the panel data settng, even though there are really N observatons, we treat the number of observatons n 3) as, and to calculate we delete the ˆ t th t perod observaton for each

13 cross-sectonal unt. hs s done because, n the models they consder, the bas s of order We refer to ths procedure as the panel jackknfe. Other smlar versons of the jackknfe can remove bas of order. For example, Dhaene, Jochmans and huysbaert 006) propose the half-panel jackknfe estmator:.) ˆ panel J half ) ˆ ˆ ˆ + ), 6) where ˆ s the fxed effects estmator based on the full sample; ˆ and ˆ are based on the frst- and second- halves of the panel sample, where each half-panel conssts of consecutve observatons over tme for all cross-sectonal unts. hey show that the half-panel jackknfe estmator also removes the bas of order that the bas s of order from the orgnal estmator. However, for the case, we wll consder only the standard panel jackknfe as descrbed above. It s obvous that when there s no te, the panel jackknfe wll remove the frst-level bas of the estmate of α N ) hence, the bas of the estmates of relatve techncal neffcency the bas s of order estmator that can handle bas of order generalzed jackknfe. u ) snce. For the cases of an exact te and a near te, however, we need a jackknfe /. he dfference n the order of the bas leads us to the 4. he Generalzed Jackknfe Schucany, Gray and Owen 97) were the frst to propose a jackknfe estmator that can handle a more general form of bas. It was not untl later that Gray and Schucany 97) gave t the name generalzed jackknfe. Gray and Schucany 97) defne the generalzed jackknfe as the followng. 3

14 Defnton Gray and Schucany 97) s Defnton.. Let ˆ and ˆ be two estmators for. hen, for any real number R, the generalzed jackknfe estmator G ˆ, ˆ ) s defned as ˆ ˆ ˆ ˆ R G, ). 7) R he usual Quenoulle) jackknfe corresponds to ˆ ˆ, ˆ ˆ t t), and R ). If we can express the bas of the estmators n terms of the sample sze and the true parameter, we can choose R so that the generalzed jackknfe s unbased. heorem Gray and Schucany 97) s heorem.. If the bas of the estmators ˆ and ˆ can be expressed as E ˆ ) + b k k, ), k, ; b, ) 0; and b, ) R b, ), then E G ˆ, ˆ )] [. 4

15 5 Proof. )., ), snce, ), ), )], [ )], [ )] ˆ, ˆ [ b Rb R Rb b R b R b G E In general, we do not have a bas expresson of the form of the prevous theorem, but we have a seres expanson of the bas wth a leadng term of known order. hen the generalzed jackknfe removes the leadng term of the seres expanson of the bas. heorem Gray and Schucany 97) s heorem.. If the bas of the estmators ˆ and ˆ can be expanded as an nfnte seres:, ),, ) ˆ + k b E k k and, ), ), b b R then. ), ), )] ˆ, ˆ [ R b R b G E + Proof. Smlar to the proof of heorem.

16 / 4.3 he Generalzed Panel Jackknfe When the Bas Is of Order We are specfcally nterested n the case that the bas of ˆ s of order the followng expanson holds: As before, we let ˆ ˆ observatons) and ˆ ˆ s equal to and the generalzed jackknfe s /. Suppose that / 3 / E ˆ) + B + D + O ). 8) t t). hen the weght R n heorem R B ) B ) 9). 0) G ˆ) ˆ ˆ t t) It s then easy to verfy that the bas of G ˆ ) s of order ˆ has been removed. ; that s, the / term n the bas of In the panel data case, once agan we treat the number of observatons as, and s calculated by deletng the ths the generalzed panel jackknfe. th t tme perod observaton for each cross-sectonal unt. We wll call he generalzed jackknfe removes bas more aggressvely than the usual jackknfe, n the ˆ t sense that the weghts attached to ˆ and to ˆ t t) are larger. For example, for 0 we have J ˆ) 0 ˆ 9 G ˆ) 9.5 ˆ 8.5 ˆ t t) ) ˆ t t) ). 6

17 Smlarly for 50 we have J ˆ) 50 ˆ 49 G ˆ) 99.5 ˆ 98.5 A detal that we do not pursue n ths paper s that t s possble to consder a second level ˆ t t) ) ˆ t t) ). of the jackknfe. If the bas of the orgnal estmate s of order /, the bas of ) G ˆ s of order. he usual panel jackknfe appled to the estmator ) G ˆ would remove the bas of order. he resultng estmator would be a lnear combnaton of the orgnal estmate, the varous drop one observaton estmates, and the varous drop two observatons estmates. 4.4 What If he Wrong Jackknfe Is Used? We have seen that the usual panel jackknfe s approprate when the bas s of order whereas the generalzed panel jackknfe s approprate when the bas s of order the queston of what happens f the wrong verson of the jackknfe s used., /. hs rases heorem 3 If the bas of ˆ s of order half of the bas. /, the usual panel jackknfe corrects approxmately Proof. We have E ˆ) + / B + hgher order terms. So, droppng the hgher order terms, we calculate B E[ J ˆ)] + B B +. ) + Comparng the bas n ths expresson to the orgnal bas of / B, we have removed about half of the frst-order bas term. 7

18 8 heorem 3 s our explanaton of the puzzle that the bootstrap and the jackknfe often correct half of the bas. Cases where ths occurs correspond to more or less an exact te. heorem 4 If the bas of ˆ s of order, the bas of the generalzed panel jackknfe s approxmately the negatve of the bas of the orgnal estmate. Proof. Suppose. ˆ) hgher order terms B E + + So, agan droppng the hgher order terms,. ) ] ) [ ] ) [ ] [ ] [ ˆ)] [ / / / / B B B B B B B G E t ) So the bas of ) ˆ G, B ), s approxmately the negatve of the orgnal bas, B. 5. DESIGN OF HE MONE CARLO EXPERIMENS In ths secton, we conduct Monte Carlo smulatons to nvestgate the fnte sample performance of the followng estmators of ) N α : ), the maxmum of the fxed effects

19 estmates; ) J ), the panel jackknfe estmate; ) G ) estmate; and v) boot BC, the bas-corrected bootstrap pont estmate., the generalzed panel jackknfe We are prmarly nterested n the bas of these estmators. However, we wll also report ther varance and mean square error. hese measures are defned precsely later n ths secton. he model s the smple panel data model wth no regressors, as gven n 7). hus, the data generatng process s y t α + v u α + v,,..., N; t,...,, t t 3) where α α u ; the u are..d. half-normal: u U where U ~ N0, σ ) ; and the v t are normal wth mean zero and varance σ v. hese dstrbutonal assumptons are not used n estmaton. hey just characterze the process that generates the data. he set of parameters s { α, σ v, σ u, N, } but ths can be reduced somewhat. All of the results bas, varance, and MSE) are nvarant wth respect to α, so we set t equal to one, wthout loss of generalty. Also, only ratos of varances matter. If we multply both σ u and σ v by a u constant q, the bases of the estmates change by q and the MSE s change by q. So we really only need to consder three parameters: N,, and a relatve varance parameter. Km, Km and Schmdt 007) used the relatve varance parameter γ σ ) [ σ + σ ) ], where u v u σ u ) var u) π ) π ) σ u. We wll use nstead the parameter μ defned by σ ). 4) u μ / σ v hs s not a matter of substance. We use μ because we fnd t easer to nterpret. It measures the 9

20 standard devaton of the α n unts of the standard devaton of the. Also, for reasons gven below, wth ths parameterzaton t turns out that does not matter very much. Only μ and N turn out to be mportant. So, n the end, our parameter space s { μ, N, }. We set scale by settng σ 0., whch for a gven determnes σ v. hen, for a gven μ, σ u ) s determned. We consder / 0,0,,0 / μ, and 0. Wth σ v 0., for a gven value of μ, the values of σ u ) and σ u are as follows: ) 0 μ 0. : σ ) 0. 00; σ ; u ) 0 / μ : σ ) 0. 0; σ ; u 3) μ : σ ) 0. u ; σ u ; 4) 0 / μ 3. 63; σ ) ; σ. 759 ; u 5) μ 0 : σ ) 0 u ; σ u We consder sample szes N,0, 0,50, and 00, and we set 0 u u u v. We also consdered 5,0,50, and 00, and the results for these values of are avalable n a supplementary set of tables, avalable from the authors on request. he basc outcomes that we would expect n the smulatons are as follows. Frst, bas wll be larger when N s larger, but the effect of N on the relatve performance of the varous bas-corrected methods s not obvous. Second, bas wll be larger when μ s smaller, snce then the varablty of the α s smaller relatve to the samplng varablty of the. We mght expect the ordnary panel jackknfe or the bootstrap to be better than the generalzed jackknfe when μ 0

21 s large we are farther from a te), and vce-versa. hrd, condtonal on μ, we do not expect to be very mportant. When we change n our experment, holdng constant μ and σ v, t means that σ v ncreases proportonally to, and σ u ) s unchanged. herefore nether the varablty of the α nor the samplng varablty of the changes. he only reason that should matter s that the jackknfe s weghts on ˆ and ˆ t t) depend on. We consder three dfferent varatons of the setup we have just descrbed. Experment I No e). he setup of ths experment s exactly as just descrbed. here are no restrctons on the α. hey just follow from the draws of the half-normal u. hs setup s very smlar to that of Km, Km and Schmdt 007). Experment II Exact e). We generate data as descrbed above. Now we the data generator) know whch frm s the best and the value α N ) of ts ntercept. We randomly select one of the other N ) frms and set ts ntercept also equal to α N ). herefore we have created an exact two-way te for the best frm. Experment III Near e). We start as n Experment II. However, once we have observed the best frm and α N ), we randomly select one of the other N ) frms and set ts ntercept equal to / α N α N α N ]. 5) [ So, for example, f 0, we have now created a new second-best frm whose ntercept s tmes closer to α N ) than the prevously second-best frm s ntercept. For each confguraton of { μ, N, }, we perform,000 replcatons. Wthn each replcaton, the bas-corrected bootstrap estmate s based on,000 bootstrap replcatons.

22 boot For each of the estmators α, J ˆ), α G ˆ), α ˆ α ) ˆ we calculate bas, varance and mean square error. he parameter beng estmated, α N ), vares across replcatons because of the random draws of the half-normal u that determne α α u. herefore we wll explctly state BC our defnton of bas, varance and MSE. Frst defne: ) NREP number of replcatons; ) r ndex of replcaton, r,..., NREP ; ) r value of α N ) n replcaton r ; v) ˆ r value of ˆ n replcaton r for any of the four estmators lsted above); and v) NREP r r ˆ and NREP ˆ. r r he defnton of bas s straghtforward: bas ˆ) NREP ˆ ) ˆ. r r 6) r hen we defne the mean squared error as MSE ˆ) NREP NREP r r ˆ ) r [ ˆ ) bas ˆ)] r r r + bas ˆ) 7) and the varance as var ˆ) MSE ˆ) bas ˆ) NREP r [ ˆ ) bas ˆ)] r r. 8) 6. RESULS OF HE MONE CARLO EXPERIMENS ables,, and 3 gve the results of Experment I n whch there s no te. All of these results are for 0. able gves the bas of the estmates, whle able gves varance and able 3 gves MSE. In all three tables, column ) gves results for ; column ) gves results for

23 the panel jackknfe J ) ; column 3) gves results for the generalzed panel jackknfe G ) ; and column 4) gves results for the bas-corrected bootstrap pont estmate boot BC. Consder frst able, whch gves the bas of the varous estmates as an estmate of α N ). hs s equvalent to the bas of estmated relatve techncal neffcency u ˆ as an estmate of As expected, the bas of s larger when N s larger the max s taken over more frms) and when μ s smaller we are closer to a te). he panel jackknfe and the bas-corrected bootstrap are less based than the fxed effects estmate. However, they only correct part of the bas. In most cases the jackknfe corrects more of the bas than the bas-corrected bootstrap. he generalzed panel jackknfe overcorrects so the orgnal upward bas now becomes a downward bas). When μ s very small, so that the varablty of the α s very small relatve to the samplng varablty of the, we are n a sense close to a te. In these cases the exact te asymptotcs appear to be relevant: the generalzed panel jackknfe s nearly unbased, and the panel jackknfe and also the bas-corrected bootstrap) corrects about half of the bas, as predcted by heorem 3. Conversely, when μ s large we are far from a te, the panel jackknfe and the bas-corrected bootstrap are nearly unbased, and the downward bas of the generalzed panel jackknfe s almost as large as the upward bas of, as predcted by heorem 4. able gves the varance of the varous estmates. hey are easy to summarze. he varance of the s less than the varance of the bas-corrected bootstrap pont estmate, whch s less than the varance of the panel jackknfe, whch s less than the varance of the generalzed panel jackknfe. he varance of the generalzed panel jackknfe s consderably larger than the varance of the other estmators. o properly nterpret these varances, remember that we are u. 3

24 ultmately nterested n estmatng the relatve sze of the u, whose varance s σ u ), and that n our setup σ ) 0.00, 0.0, 0.,, and 0 for μ 0,0 /,,0 /, and 0, respectvely. u So the varance of these estmators s large enough to be an ssue, except perhaps for the larger values of μ. able 3 gves the MSE of the estmates. In terms of MSE, the two varetes of the jackknfe are domnated by the bas-corrected bootstrap. he bas-corrected bootstrap s also generally better than the fxed effects estmate, except n those cases where the bas of s small.e., when N s small and μ s large). 6. Now we turn to Experment II, the case of an exact te. hese results are n able 4, 5, and In terms of bas, we see n able 4 that the generalzed panel jackknfe s clearly the best. It overcorrects the bas, but not by as much as the panel jackknfe and the bas-corrected bootstrap undercorrect. As expected from heorem 3, the panel jackknfe corrects about half of the bas. he bas-corrected bootstrap, whch s not vald asymptotcally n the case of an exact te, also appears to correct about half of the bas. In able 5, the varances of the estmates are rather smlar to the varances for the case of no te able ). he man dfference s that now the varance does not depend as strongly on μ, presumably because, once we have forced a te, the smlarty of the other α s not of as much mportance. he rankng of the estmators, n order of ncreasng varance, s stll the same as n able, bas-corrected bootstrap, panel jackknfe and generalzed panel jackknfe). In terms of MSE, we see n able 6 that the bas-corrected bootstrap stll domnates both varetes of the jackknfe. It s also generally better than the fxed effects estmate. hs 4

25 favorable performance of the bas-corrected bootstrap s perhaps surprsng, gven that t s not asymptotcally vald n the case of an exact te. Our last experment s Experment III, the case of a near te. he results for ths experment are gven n able 7, 8, and 9. As a general statement, the results are between those of Experment I and Experment II, whch s not surprsng. For small values of μ nearer te), the bas results n able 7 are qute smlar to those of able 4 for an exact te. In these cases the generalzed panel jackknfe has lttle bas, whle the panel jackknfe and the bas-corrected bootstrap correct about half of the bas. For large values of μ less near te), the panel jackknfe and the bas-corrected bootstrap stll correct only some of the bas, but the generalzed panel jackknfe overcorrects. Stll, t s generally true n able 7 that the generalzed panel jackknfe has the smallest bas. In terms of varance able 8) and MSE able 9), the results are farly smlar to those for both the case of no te and the case of an exact te. Once agan the bas-corrected bootstrap s generally the best, and the generalzed panel jackknfe s the worst. he last ssue we consder s the effect of changng. We consder the same three knds of experments as just descrbed, wth 5,0,50, and 00 n addton to 0, whch we have just dscussed). hese results are gven n a set of 36 Supplemental ables, avalable on request from the authors. In ths paper we wll dsplay the results only for μ and N 0. ables 0,, and gve the bas, varance, and MSE of the varous estmates. As dscussed n Secton 5, we do not expect changes n to be very mportant, because we are holdng constant N, μ, and σ v, or equvalently we are holdng constant N, σ u ), and σ v. Indeed, the motvaton for adoptng ths parameterzaton was that we expected t to 5

26 make one of the parameters unmportant. We expect changng to be more mportant for the jackknfe estmates than for the other two estmates, because the value of affects the weghts that the jackknfe puts on the orgnal estmate versus the average of the delete-one-observaton estmates. What we see n ables 0 s not surprsng. In able 0, the effect of changng on the bas of the estmates s very mnor. In able, changng does not affect the varance of the fxed effects estmate or the bas-corrected bootstrap pont estmate very much, but the varance of the jackknfe estmates ncreases notceably as ncreases. Correspondngly, n able the MSE of the jackknfe estmates ncreases as ncreases. However, t remans true that the value of s much less mportant than the values of N and μ n determnng the relatve performance of the varous estmates. 7. CONCLUDING REMARKS In the stochastc fronter model wth panel data, the fxed effects estmate of the fronter ntercept s based upward. Prevous work found that the bas-corrected bootstrap corrected only part of ths bas. hs paper has tred to explan that fndng and to see whether we can more successfully remove the bas usng the jackknfe. he bootstrap s known to be asymptotcally as wth N fxed) vald f there s no te for the best frm, and not vald f there s an exact te. So whether there s a te, and how close we are to havng a te f there s not an exact te, s a reasonable ssue to focus on. When there s an exact te, we show that the bas of the fxed effects estmate s of order / rather than. Not only s the bootstrap not vald, but the usual panel jackknfe, whch s based on the assumpton that the bas s of order, also does not work correctly. More 6

27 specfcally, we show that t removes approxmately) half of the bas. A dfferent form of the jackknfe, whch we call the generalzed panel jackknfe, s needed to remove the bas of order /. If there s no te, the bootstrap s vald and the panel jackknfe should also be effectve n removng bas, snce now the bas s of order. In ths case the generalzed panel jackknfe wll not work correctly, and ndeed we show that ts bas s the negatve of the bas of the fxed effects estmate; t reverses the bas. We also consder the case of a near te, whch we defne as the case that the dfference between the fronter ntercept and the ntercept of the second-best frm s O / ). In ths case the bas s agan of order / and so the generalzed panel jackknfe should remove t. Our smulatons support the fnte-sample relevance of these arguments. When there s a te or a near te, the generalzed panel jackknfe removes the bas effectvely, whereas the panel jackknfe and the bas-corrected bootstrap remove about half of the bas. When there s not a te, the generalzed panel jackknfe overcorrects the bas, and the panel jackknfe and the bas-corrected bootstrap are much better at removng the bas. he major drawback of the jackknfe s that ts varance s large. hs s true for both versons of the jackknfe but the varance s the largest for the generalzed panel jackknfe. here does not seem to be any good reason to prefer the panel jackknfe to the bas-corrected bootstrap, snce t has a larger varance and does not do a better job of correctng bas. However, whle the generalzed panel jackknfe s clearly domnated by the bas-corrected bootstrap n terms of MSE, t does do a very good job of removng bas when there s an exact te or a near te. Emprcally, presumably that corresponds to cases where the dentty of the best frm s n substantal doubt. he nablty of the generalzed panel jackknfe to beat the bas-corrected bootstrap n 7

28 terms of MSE when there s an exact or a near te s perhaps surprsng, snce the bootstrap s not vald f there s a te. However, not vald here has a specfc meanng, namely that we cannot clam that the dstrbuton of the bootstrap estmate around the orgnal estmate matches the dstrbuton of the orgnal estmate around the true parameter. Apparently the bas-corrected bootstrap s nevertheless a useful pont estmate. 8

29 REFERENCES Dhaene, G., K. Jochmans, and B. huysbaert, 006, Splt-Panel Jackknfe Estmaton of Fxed Effects Models Prevous tle: Jackknfe Bas Reducton for Nonlnear Dynamc Panel Data Models wth Fxed Effects), Workng Paper. Fernández-Val, I. and F. Vella, 007, Bas Correctons for wo-step Fxed Effects Panel Data Estmator, IZA Dscusson Paper 690, Insttute for the Study of Labor IZA) Gray, H.L., and W.R. Schucany, 97, he Generalzed Jackknfe Statstc, Marcel Dekker, Inc., New York. Hahn, J. and G. Kuerstener, 004, Bas Reducton for Dynamc Nonlnear Panel Models wth Fxed Effects, Unpublshed Manuscrpt. Hahn, J., and W. Newey, 004, Jackknfe and Analytcal Bas Reducton for Nonlnear Panel Models, Econometrca, 7, Hall, P., W.Härdle, and L. Smar, 995, Iterated Bootstrap wth Applcatons to Fronter Models, Journal of Productvty Analyss, 6, Km, M., Y. Km, and P. Schmdt, 007, On the Accuracy of Bootstrap Confdence Intervals for Effcency Levels n Stochastc Fronter Models wth Panel Data, Journal of Productvty Analyss, 8, Mller, R.G., 974, he Jackknfe A Revew, Bometrka, 6, -5. Park, B.U., and L. Smar, 994, Effcent Semparametrc Estmaton n a Stochastc Fronter Model, Journal of the Amercan Statstcs Assocaton, 89, Ptt, M.M., and L.F. Lee, 98, he Measurement and Sources of echncal Ineffcency n the Indonesan Weavng Industry, Journal of Development Economcs, 9, Quenoulle, M.H., 956, Note on Bas n Estmaton, Bometrka, 6, Schmdt, P., and R. Sckles, 984, Producton Fronters and Panel Data, Journal of Busness and Economc Statstcs,, Schucany, W.R., H.L. Gray, and D.B. Owen, 97, On Bas Reducton n Estmaton, Journal of the Amercan Statstcal Assocaton, 66, ukey, J.W., 958, Bas and Confdence n Not Qute Large Samples, Abstract), Annals of Mathematcal Statstcs, 8, 64. 9

30 ABLE EXPERIMEN I: NO IE 0 BIAS OF HE ESIMAES μ N ) E ˆ α α N ) ) E J ˆ α α N ) 3) E G ˆ α α N ) 4) boot E ˆ α α / / / / / / / / / / BC N ) 30

31 ABLE EXPERIMEN I: NO IE 0 VARIANCE OF HE ESIMAES μ N ) ) J 3) G 4) / / / / / / / / / / boot BC 3

32 ABLE 3 EXPERIMEN I: NO IE 0 MSE OF HE ESIMAES μ N ) ) J 3) G 4) / / / / / / / / / / boot BC 3

33 ABLE 4 EXPERIMEN II: EXAC IE 0 BIAS OF HE ESIMAES μ N ) E ˆ α α N ) ) E J ˆ α α N ) 3) E G ˆ α α N ) 4) boot E ˆ α α / / / / / / / / BC N ) Note: value of μ s rrelevant when N and there s an exact te. 33

34 ABLE 5 EXPERIMEN II: EXAC IE 0 VARIANCE OF HE ESIMAES μ N ) ) J 3) G 4) / / / / / / / / boot BC Note: value of μ s rrelevant when N and there s an exact te. 34

35 ABLE 6 EXPERIMEN II: EXAC IE 0 MSE OF HE ESIMAES μ N ) ) J 3) G 4) / / / / / / / / boot BC Note: value of μ s rrelevant when N and there s an exact te. 35

36 ABLE 7 EXPERIMEN III: NEAR IE 0 BIAS OF HE ESIMAES μ N ) E ˆ α α N ) ) E J ˆ α α N ) 3) E G ˆ α α N ) 4) boot E ˆ α α / / / / / / / / / / BC N ) 36

37 ABLE 8 EXPERIMEN III: NEAR IE 0 VARIANCE OF HE ESIMAES μ N ) ) J 3) G 4) / / / / / / / / / / boot BC 37

38 ABLE 9 EXPERIMEN III: NEAR IE 0 MSE OF HE ESIMAES μ N ) ) J 3) G 4) / / / / / / / / / / boot BC 38

39 ABLE 0 EFFEC OF CHANGING μ, N 0 BIAS OF HE ESIMAES Experment I NO IE II EXAC IE III NEAR IE ) E ˆ α α N ) ) E J ˆ α α N ) 3) E G ˆ α α N ) 4) boot E ˆ α α BC N ) 39

40 ABLE EFFEC OF CHANGING μ, N 0 VARIANCE OF HE ESIMAES Experment ) I NO IE II EXAC IE III NEAR IE ) J 3) G 4) boot BC 40

41 ABLE EFFEC OF CHANGING μ, N 0 MSE OF HE ESIMAES Experment ) I NO IE II EXAC IE III NEAR IE ) J 3) G 4) boot BC 4

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4)

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4) I. Classcal Assumptons Econ7 Appled Econometrcs Topc 3: Classcal Model (Studenmund, Chapter 4) We have defned OLS and studed some algebrac propertes of OLS. In ths topc we wll study statstcal propertes

More information

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity

LINEAR REGRESSION ANALYSIS. MODULE IX Lecture Multicollinearity LINEAR REGRESSION ANALYSIS MODULE IX Lecture - 30 Multcollnearty Dr. Shalabh Department of Mathematcs and Statstcs Indan Insttute of Technology Kanpur 2 Remedes for multcollnearty Varous technques have

More information

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction

The Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction ECONOMICS 5* -- NOTE (Summary) ECON 5* -- NOTE The Multple Classcal Lnear Regresson Model (CLRM): Specfcaton and Assumptons. Introducton CLRM stands for the Classcal Lnear Regresson Model. The CLRM s also

More information

4 Analysis of Variance (ANOVA) 5 ANOVA. 5.1 Introduction. 5.2 Fixed Effects ANOVA

4 Analysis of Variance (ANOVA) 5 ANOVA. 5.1 Introduction. 5.2 Fixed Effects ANOVA 4 Analyss of Varance (ANOVA) 5 ANOVA 51 Introducton ANOVA ANOVA s a way to estmate and test the means of multple populatons We wll start wth one-way ANOVA If the populatons ncluded n the study are selected

More information

x = , so that calculated

x = , so that calculated Stat 4, secton Sngle Factor ANOVA notes by Tm Plachowsk n chapter 8 we conducted hypothess tests n whch we compared a sngle sample s mean or proporton to some hypotheszed value Chapter 9 expanded ths to

More information

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur Analyss of Varance and Desgn of Exerments-I MODULE III LECTURE - 2 EXPERIMENTAL DESIGN MODELS Dr. Shalabh Deartment of Mathematcs and Statstcs Indan Insttute of Technology Kanur 2 We consder the models

More information

Discussion of Extensions of the Gauss-Markov Theorem to the Case of Stochastic Regression Coefficients Ed Stanek

Discussion of Extensions of the Gauss-Markov Theorem to the Case of Stochastic Regression Coefficients Ed Stanek Dscusson of Extensons of the Gauss-arkov Theorem to the Case of Stochastc Regresson Coeffcents Ed Stanek Introducton Pfeffermann (984 dscusses extensons to the Gauss-arkov Theorem n settngs where regresson

More information

STAT 3008 Applied Regression Analysis

STAT 3008 Applied Regression Analysis STAT 3008 Appled Regresson Analyss Tutoral : Smple Lnear Regresson LAI Chun He Department of Statstcs, The Chnese Unversty of Hong Kong 1 Model Assumpton To quantfy the relatonshp between two factors,

More information

Basically, if you have a dummy dependent variable you will be estimating a probability.

Basically, if you have a dummy dependent variable you will be estimating a probability. ECON 497: Lecture Notes 13 Page 1 of 1 Metropoltan State Unversty ECON 497: Research and Forecastng Lecture Notes 13 Dummy Dependent Varable Technques Studenmund Chapter 13 Bascally, f you have a dummy

More information

Global Sensitivity. Tuesday 20 th February, 2018

Global Sensitivity. Tuesday 20 th February, 2018 Global Senstvty Tuesday 2 th February, 28 ) Local Senstvty Most senstvty analyses [] are based on local estmates of senstvty, typcally by expandng the response n a Taylor seres about some specfc values

More information

Department of Statistics University of Toronto STA305H1S / 1004 HS Design and Analysis of Experiments Term Test - Winter Solution

Department of Statistics University of Toronto STA305H1S / 1004 HS Design and Analysis of Experiments Term Test - Winter Solution Department of Statstcs Unversty of Toronto STA35HS / HS Desgn and Analyss of Experments Term Test - Wnter - Soluton February, Last Name: Frst Name: Student Number: Instructons: Tme: hours. Ads: a non-programmable

More information

x i1 =1 for all i (the constant ).

x i1 =1 for all i (the constant ). Chapter 5 The Multple Regresson Model Consder an economc model where the dependent varable s a functon of K explanatory varables. The economc model has the form: y = f ( x,x,..., ) xk Approxmate ths by

More information

Econometrics of Panel Data

Econometrics of Panel Data Econometrcs of Panel Data Jakub Mućk Meetng # 8 Jakub Mućk Econometrcs of Panel Data Meetng # 8 1 / 17 Outlne 1 Heterogenety n the slope coeffcents 2 Seemngly Unrelated Regresson (SUR) 3 Swamy s random

More information

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X Statstcs 1: Probablty Theory II 37 3 EPECTATION OF SEVERAL RANDOM VARIABLES As n Probablty Theory I, the nterest n most stuatons les not on the actual dstrbuton of a random vector, but rather on a number

More information

Statistics for Economics & Business

Statistics for Economics & Business Statstcs for Economcs & Busness Smple Lnear Regresson Learnng Objectves In ths chapter, you learn: How to use regresson analyss to predct the value of a dependent varable based on an ndependent varable

More information

Chapter 2 - The Simple Linear Regression Model S =0. e i is a random error. S β2 β. This is a minimization problem. Solution is a calculus exercise.

Chapter 2 - The Simple Linear Regression Model S =0. e i is a random error. S β2 β. This is a minimization problem. Solution is a calculus exercise. Chapter - The Smple Lnear Regresson Model The lnear regresson equaton s: where y + = β + β e for =,..., y and are observable varables e s a random error How can an estmaton rule be constructed for the

More information

An (almost) unbiased estimator for the S-Gini index

An (almost) unbiased estimator for the S-Gini index An (almost unbased estmator for the S-Gn ndex Thomas Demuynck February 25, 2009 Abstract Ths note provdes an unbased estmator for the absolute S-Gn and an almost unbased estmator for the relatve S-Gn for

More information

Comparison of Regression Lines

Comparison of Regression Lines STATGRAPHICS Rev. 9/13/2013 Comparson of Regresson Lnes Summary... 1 Data Input... 3 Analyss Summary... 4 Plot of Ftted Model... 6 Condtonal Sums of Squares... 6 Analyss Optons... 7 Forecasts... 8 Confdence

More information

Econ Statistical Properties of the OLS estimator. Sanjaya DeSilva

Econ Statistical Properties of the OLS estimator. Sanjaya DeSilva Econ 39 - Statstcal Propertes of the OLS estmator Sanjaya DeSlva September, 008 1 Overvew Recall that the true regresson model s Y = β 0 + β 1 X + u (1) Applyng the OLS method to a sample of data, we estmate

More information

Properties of Least Squares

Properties of Least Squares Week 3 3.1 Smple Lnear Regresson Model 3. Propertes of Least Squares Estmators Y Y β 1 + β X + u weekly famly expendtures X weekly famly ncome For a gven level of x, the expected level of food expendtures

More information

Assignment 5. Simulation for Logistics. Monti, N.E. Yunita, T.

Assignment 5. Simulation for Logistics. Monti, N.E. Yunita, T. Assgnment 5 Smulaton for Logstcs Mont, N.E. Yunta, T. November 26, 2007 1. Smulaton Desgn The frst objectve of ths assgnment s to derve a 90% two-sded Confdence Interval (CI) for the average watng tme

More information

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur Analyss of Varance and Desgn of Experment-I MODULE VIII LECTURE - 34 ANALYSIS OF VARIANCE IN RANDOM-EFFECTS MODEL AND MIXED-EFFECTS EFFECTS MODEL Dr Shalabh Department of Mathematcs and Statstcs Indan

More information

Sampling Theory MODULE VII LECTURE - 23 VARYING PROBABILITY SAMPLING

Sampling Theory MODULE VII LECTURE - 23 VARYING PROBABILITY SAMPLING Samplng heory MODULE VII LECURE - 3 VARYIG PROBABILIY SAMPLIG DR. SHALABH DEPARME OF MAHEMAICS AD SAISICS IDIA ISIUE OF ECHOLOGY KAPUR he smple random samplng scheme provdes a random sample where every

More information

Testing for seasonal unit roots in heterogeneous panels

Testing for seasonal unit roots in heterogeneous panels Testng for seasonal unt roots n heterogeneous panels Jesus Otero * Facultad de Economía Unversdad del Rosaro, Colomba Jeremy Smth Department of Economcs Unversty of arwck Monca Gulett Aston Busness School

More information

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers

Psychology 282 Lecture #24 Outline Regression Diagnostics: Outliers Psychology 282 Lecture #24 Outlne Regresson Dagnostcs: Outlers In an earler lecture we studed the statstcal assumptons underlyng the regresson model, ncludng the followng ponts: Formal statement of assumptons.

More information

Maximizing the number of nonnegative subsets

Maximizing the number of nonnegative subsets Maxmzng the number of nonnegatve subsets Noga Alon Hao Huang December 1, 213 Abstract Gven a set of n real numbers, f the sum of elements of every subset of sze larger than k s negatve, what s the maxmum

More information

Interval Estimation in the Classical Normal Linear Regression Model. 1. Introduction

Interval Estimation in the Classical Normal Linear Regression Model. 1. Introduction ECONOMICS 35* -- NOTE 7 ECON 35* -- NOTE 7 Interval Estmaton n the Classcal Normal Lnear Regresson Model Ths note outlnes the basc elements of nterval estmaton n the Classcal Normal Lnear Regresson Model

More information

Lecture 10 Support Vector Machines II

Lecture 10 Support Vector Machines II Lecture 10 Support Vector Machnes II 22 February 2016 Taylor B. Arnold Yale Statstcs STAT 365/665 1/28 Notes: Problem 3 s posted and due ths upcomng Frday There was an early bug n the fake-test data; fxed

More information

2016 Wiley. Study Session 2: Ethical and Professional Standards Application

2016 Wiley. Study Session 2: Ethical and Professional Standards Application 6 Wley Study Sesson : Ethcal and Professonal Standards Applcaton LESSON : CORRECTION ANALYSIS Readng 9: Correlaton and Regresson LOS 9a: Calculate and nterpret a sample covarance and a sample correlaton

More information

A Comparative Study for Estimation Parameters in Panel Data Model

A Comparative Study for Estimation Parameters in Panel Data Model A Comparatve Study for Estmaton Parameters n Panel Data Model Ahmed H. Youssef and Mohamed R. Abonazel hs paper examnes the panel data models when the regresson coeffcents are fxed random and mxed and

More information

Chapter 11: Simple Linear Regression and Correlation

Chapter 11: Simple Linear Regression and Correlation Chapter 11: Smple Lnear Regresson and Correlaton 11-1 Emprcal Models 11-2 Smple Lnear Regresson 11-3 Propertes of the Least Squares Estmators 11-4 Hypothess Test n Smple Lnear Regresson 11-4.1 Use of t-tests

More information

January Examinations 2015

January Examinations 2015 24/5 Canddates Only January Examnatons 25 DO NOT OPEN THE QUESTION PAPER UNTIL INSTRUCTED TO DO SO BY THE CHIEF INVIGILATOR STUDENT CANDIDATE NO.. Department Module Code Module Ttle Exam Duraton (n words)

More information

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix

Lectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix Lectures - Week 4 Matrx norms, Condtonng, Vector Spaces, Lnear Independence, Spannng sets and Bass, Null space and Range of a Matrx Matrx Norms Now we turn to assocatng a number to each matrx. We could

More information

Linear Approximation with Regularization and Moving Least Squares

Linear Approximation with Regularization and Moving Least Squares Lnear Approxmaton wth Regularzaton and Movng Least Squares Igor Grešovn May 007 Revson 4.6 (Revson : March 004). 5 4 3 0.5 3 3.5 4 Contents: Lnear Fttng...4. Weghted Least Squares n Functon Approxmaton...

More information

The Granular Origins of Aggregate Fluctuations : Supplementary Material

The Granular Origins of Aggregate Fluctuations : Supplementary Material The Granular Orgns of Aggregate Fluctuatons : Supplementary Materal Xaver Gabax October 12, 2010 Ths onlne appendx ( presents some addtonal emprcal robustness checks ( descrbes some econometrc complements

More information

Limited Dependent Variables

Limited Dependent Variables Lmted Dependent Varables. What f the left-hand sde varable s not a contnuous thng spread from mnus nfnty to plus nfnty? That s, gven a model = f (, β, ε, where a. s bounded below at zero, such as wages

More information

More metrics on cartesian products

More metrics on cartesian products More metrcs on cartesan products If (X, d ) are metrc spaces for 1 n, then n Secton II4 of the lecture notes we defned three metrcs on X whose underlyng topologes are the product topology The purpose of

More information

Chapter 13: Multiple Regression

Chapter 13: Multiple Regression Chapter 13: Multple Regresson 13.1 Developng the multple-regresson Model The general model can be descrbed as: It smplfes for two ndependent varables: The sample ft parameter b 0, b 1, and b are used to

More information

ECONOMICS 351*-A Mid-Term Exam -- Fall Term 2000 Page 1 of 13 pages. QUEEN'S UNIVERSITY AT KINGSTON Department of Economics

ECONOMICS 351*-A Mid-Term Exam -- Fall Term 2000 Page 1 of 13 pages. QUEEN'S UNIVERSITY AT KINGSTON Department of Economics ECOOMICS 35*-A Md-Term Exam -- Fall Term 000 Page of 3 pages QUEE'S UIVERSITY AT KIGSTO Department of Economcs ECOOMICS 35* - Secton A Introductory Econometrcs Fall Term 000 MID-TERM EAM ASWERS MG Abbott

More information

Module 9. Lecture 6. Duality in Assignment Problems

Module 9. Lecture 6. Duality in Assignment Problems Module 9 1 Lecture 6 Dualty n Assgnment Problems In ths lecture we attempt to answer few other mportant questons posed n earler lecture for (AP) and see how some of them can be explaned through the concept

More information

Estimation: Part 2. Chapter GREG estimation

Estimation: Part 2. Chapter GREG estimation Chapter 9 Estmaton: Part 2 9. GREG estmaton In Chapter 8, we have seen that the regresson estmator s an effcent estmator when there s a lnear relatonshp between y and x. In ths chapter, we generalzed the

More information

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U)

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U) Econ 413 Exam 13 H ANSWERS Settet er nndelt 9 deloppgaver, A,B,C, som alle anbefales å telle lkt for å gøre det ltt lettere å stå. Svar er gtt . Unfortunately, there s a prntng error n the hnt of

More information

Negative Binomial Regression

Negative Binomial Regression STATGRAPHICS Rev. 9/16/2013 Negatve Bnomal Regresson Summary... 1 Data Input... 3 Statstcal Model... 3 Analyss Summary... 4 Analyss Optons... 7 Plot of Ftted Model... 8 Observed Versus Predcted... 10 Predctons...

More information

ANOMALIES OF THE MAGNITUDE OF THE BIAS OF THE MAXIMUM LIKELIHOOD ESTIMATOR OF THE REGRESSION SLOPE

ANOMALIES OF THE MAGNITUDE OF THE BIAS OF THE MAXIMUM LIKELIHOOD ESTIMATOR OF THE REGRESSION SLOPE P a g e ANOMALIES OF THE MAGNITUDE OF THE BIAS OF THE MAXIMUM LIKELIHOOD ESTIMATOR OF THE REGRESSION SLOPE Darmud O Drscoll ¹, Donald E. Ramrez ² ¹ Head of Department of Mathematcs and Computer Studes

More information

Linear Regression Analysis: Terminology and Notation

Linear Regression Analysis: Terminology and Notation ECON 35* -- Secton : Basc Concepts of Regresson Analyss (Page ) Lnear Regresson Analyss: Termnology and Notaton Consder the generc verson of the smple (two-varable) lnear regresson model. It s represented

More information

NUMERICAL DIFFERENTIATION

NUMERICAL DIFFERENTIATION NUMERICAL DIFFERENTIATION 1 Introducton Dfferentaton s a method to compute the rate at whch a dependent output y changes wth respect to the change n the ndependent nput x. Ths rate of change s called the

More information

Economics 130. Lecture 4 Simple Linear Regression Continued

Economics 130. Lecture 4 Simple Linear Regression Continued Economcs 130 Lecture 4 Contnued Readngs for Week 4 Text, Chapter and 3. We contnue wth addressng our second ssue + add n how we evaluate these relatonshps: Where do we get data to do ths analyss? How do

More information

j) = 1 (note sigma notation) ii. Continuous random variable (e.g. Normal distribution) 1. density function: f ( x) 0 and f ( x) dx = 1

j) = 1 (note sigma notation) ii. Continuous random variable (e.g. Normal distribution) 1. density function: f ( x) 0 and f ( x) dx = 1 Random varables Measure of central tendences and varablty (means and varances) Jont densty functons and ndependence Measures of assocaton (covarance and correlaton) Interestng result Condtonal dstrbutons

More information

This column is a continuation of our previous column

This column is a continuation of our previous column Comparson of Goodness of Ft Statstcs for Lnear Regresson, Part II The authors contnue ther dscusson of the correlaton coeffcent n developng a calbraton for quanttatve analyss. Jerome Workman Jr. and Howard

More information

β0 + β1xi. You are interested in estimating the unknown parameters β

β0 + β1xi. You are interested in estimating the unknown parameters β Ordnary Least Squares (OLS): Smple Lnear Regresson (SLR) Analytcs The SLR Setup Sample Statstcs Ordnary Least Squares (OLS): FOCs and SOCs Back to OLS and Sample Statstcs Predctons (and Resduals) wth OLS

More information

1. Inference on Regression Parameters a. Finding Mean, s.d and covariance amongst estimates. 2. Confidence Intervals and Working Hotelling Bands

1. Inference on Regression Parameters a. Finding Mean, s.d and covariance amongst estimates. 2. Confidence Intervals and Working Hotelling Bands Content. Inference on Regresson Parameters a. Fndng Mean, s.d and covarance amongst estmates.. Confdence Intervals and Workng Hotellng Bands 3. Cochran s Theorem 4. General Lnear Testng 5. Measures of

More information

e i is a random error

e i is a random error Chapter - The Smple Lnear Regresson Model The lnear regresson equaton s: where + β + β e for,..., and are observable varables e s a random error How can an estmaton rule be constructed for the unknown

More information

A Monte Carlo Study for Swamy s Estimate of Random Coefficient Panel Data Model

A Monte Carlo Study for Swamy s Estimate of Random Coefficient Panel Data Model A Monte Carlo Study for Swamy s Estmate of Random Coeffcent Panel Data Model Aman Mousa, Ahmed H. Youssef and Mohamed R. Abonazel Department of Appled Statstcs and Econometrcs, Instute of Statstcal Studes

More information

Chapter 4: Regression With One Regressor

Chapter 4: Regression With One Regressor Chapter 4: Regresson Wth One Regressor Copyrght 2011 Pearson Addson-Wesley. All rghts reserved. 1-1 Outlne 1. Fttng a lne to data 2. The ordnary least squares (OLS) lne/regresson 3. Measures of ft 4. Populaton

More information

Chapter 9: Statistical Inference and the Relationship between Two Variables

Chapter 9: Statistical Inference and the Relationship between Two Variables Chapter 9: Statstcal Inference and the Relatonshp between Two Varables Key Words The Regresson Model The Sample Regresson Equaton The Pearson Correlaton Coeffcent Learnng Outcomes After studyng ths chapter,

More information

UNR Joint Economics Working Paper Series Working Paper No Further Analysis of the Zipf Law: Does the Rank-Size Rule Really Exist?

UNR Joint Economics Working Paper Series Working Paper No Further Analysis of the Zipf Law: Does the Rank-Size Rule Really Exist? UNR Jont Economcs Workng Paper Seres Workng Paper No. 08-005 Further Analyss of the Zpf Law: Does the Rank-Sze Rule Really Exst? Fungsa Nota and Shunfeng Song Department of Economcs /030 Unversty of Nevada,

More information

III. Econometric Methodology Regression Analysis

III. Econometric Methodology Regression Analysis Page Econ07 Appled Econometrcs Topc : An Overvew of Regresson Analyss (Studenmund, Chapter ) I. The Nature and Scope of Econometrcs. Lot s of defntons of econometrcs. Nobel Prze Commttee Paul Samuelson,

More information

C/CS/Phy191 Problem Set 3 Solutions Out: Oct 1, 2008., where ( 00. ), so the overall state of the system is ) ( ( ( ( 00 ± 11 ), Φ ± = 1

C/CS/Phy191 Problem Set 3 Solutions Out: Oct 1, 2008., where ( 00. ), so the overall state of the system is ) ( ( ( ( 00 ± 11 ), Φ ± = 1 C/CS/Phy9 Problem Set 3 Solutons Out: Oct, 8 Suppose you have two qubts n some arbtrary entangled state ψ You apply the teleportaton protocol to each of the qubts separately What s the resultng state obtaned

More information

Introduction to Vapor/Liquid Equilibrium, part 2. Raoult s Law:

Introduction to Vapor/Liquid Equilibrium, part 2. Raoult s Law: CE304, Sprng 2004 Lecture 4 Introducton to Vapor/Lqud Equlbrum, part 2 Raoult s Law: The smplest model that allows us do VLE calculatons s obtaned when we assume that the vapor phase s an deal gas, and

More information

Now we relax this assumption and allow that the error variance depends on the independent variables, i.e., heteroskedasticity

Now we relax this assumption and allow that the error variance depends on the independent variables, i.e., heteroskedasticity ECON 48 / WH Hong Heteroskedastcty. Consequences of Heteroskedastcty for OLS Assumpton MLR. 5: Homoskedastcty var ( u x ) = σ Now we relax ths assumpton and allow that the error varance depends on the

More information

Chapter 5 Multilevel Models

Chapter 5 Multilevel Models Chapter 5 Multlevel Models 5.1 Cross-sectonal multlevel models 5.1.1 Two-level models 5.1.2 Multple level models 5.1.3 Multple level modelng n other felds 5.2 Longtudnal multlevel models 5.2.1 Two-level

More information

Lecture 3 Stat102, Spring 2007

Lecture 3 Stat102, Spring 2007 Lecture 3 Stat0, Sprng 007 Chapter 3. 3.: Introducton to regresson analyss Lnear regresson as a descrptve technque The least-squares equatons Chapter 3.3 Samplng dstrbuton of b 0, b. Contnued n net lecture

More information

Chapter 6. Supplemental Text Material

Chapter 6. Supplemental Text Material Chapter 6. Supplemental Text Materal S6-. actor Effect Estmates are Least Squares Estmates We have gven heurstc or ntutve explanatons of how the estmates of the factor effects are obtaned n the textboo.

More information

Department of Quantitative Methods & Information Systems. Time Series and Their Components QMIS 320. Chapter 6

Department of Quantitative Methods & Information Systems. Time Series and Their Components QMIS 320. Chapter 6 Department of Quanttatve Methods & Informaton Systems Tme Seres and Ther Components QMIS 30 Chapter 6 Fall 00 Dr. Mohammad Zanal These sldes were modfed from ther orgnal source for educatonal purpose only.

More information

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur

Dr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur Analyss of Varance and Desgn of Experment-I MODULE VII LECTURE - 3 ANALYSIS OF COVARIANCE Dr Shalabh Department of Mathematcs and Statstcs Indan Insttute of Technology Kanpur Any scentfc experment s performed

More information

Foundations of Arithmetic

Foundations of Arithmetic Foundatons of Arthmetc Notaton We shall denote the sum and product of numbers n the usual notaton as a 2 + a 2 + a 3 + + a = a, a 1 a 2 a 3 a = a The notaton a b means a dvdes b,.e. ac = b where c s an

More information

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity Week3, Chapter 4 Moton n Two Dmensons Lecture Quz A partcle confned to moton along the x axs moves wth constant acceleraton from x =.0 m to x = 8.0 m durng a 1-s tme nterval. The velocty of the partcle

More information

Temperature. Chapter Heat Engine

Temperature. Chapter Heat Engine Chapter 3 Temperature In prevous chapters of these notes we ntroduced the Prncple of Maxmum ntropy as a technque for estmatng probablty dstrbutons consstent wth constrants. In Chapter 9 we dscussed the

More information

Using T.O.M to Estimate Parameter of distributions that have not Single Exponential Family

Using T.O.M to Estimate Parameter of distributions that have not Single Exponential Family IOSR Journal of Mathematcs IOSR-JM) ISSN: 2278-5728. Volume 3, Issue 3 Sep-Oct. 202), PP 44-48 www.osrjournals.org Usng T.O.M to Estmate Parameter of dstrbutons that have not Sngle Exponental Famly Jubran

More information

9. Binary Dependent Variables

9. Binary Dependent Variables 9. Bnar Dependent Varables 9. Homogeneous models Log, prob models Inference Tax preparers 9.2 Random effects models 9.3 Fxed effects models 9.4 Margnal models and GEE Appendx 9A - Lkelhood calculatons

More information

Canonical transformations

Canonical transformations Canoncal transformatons November 23, 2014 Recall that we have defned a symplectc transformaton to be any lnear transformaton M A B leavng the symplectc form nvarant, Ω AB M A CM B DΩ CD Coordnate transformatons,

More information

Statistical Hypothesis Testing for Returns to Scale Using Data Envelopment Analysis

Statistical Hypothesis Testing for Returns to Scale Using Data Envelopment Analysis Statstcal Hypothess Testng for Returns to Scale Usng Data nvelopment nalyss M. ukushge a and I. Myara b a Graduate School of conomcs, Osaka Unversty, Osaka 560-0043, apan (mfuku@econ.osaka-u.ac.p) b Graduate

More information

A Robust Method for Calculating the Correlation Coefficient

A Robust Method for Calculating the Correlation Coefficient A Robust Method for Calculatng the Correlaton Coeffcent E.B. Nven and C. V. Deutsch Relatonshps between prmary and secondary data are frequently quantfed usng the correlaton coeffcent; however, the tradtonal

More information

Feature Selection: Part 1

Feature Selection: Part 1 CSE 546: Machne Learnng Lecture 5 Feature Selecton: Part 1 Instructor: Sham Kakade 1 Regresson n the hgh dmensonal settng How do we learn when the number of features d s greater than the sample sze n?

More information

See Book Chapter 11 2 nd Edition (Chapter 10 1 st Edition)

See Book Chapter 11 2 nd Edition (Chapter 10 1 st Edition) Count Data Models See Book Chapter 11 2 nd Edton (Chapter 10 1 st Edton) Count data consst of non-negatve nteger values Examples: number of drver route changes per week, the number of trp departure changes

More information

Rockefeller College University at Albany

Rockefeller College University at Albany Rockefeller College Unverst at Alban PAD 705 Handout: Maxmum Lkelhood Estmaton Orgnal b Davd A. Wse John F. Kenned School of Government, Harvard Unverst Modfcatons b R. Karl Rethemeer Up to ths pont n

More information

Computation of Higher Order Moments from Two Multinomial Overdispersion Likelihood Models

Computation of Higher Order Moments from Two Multinomial Overdispersion Likelihood Models Computaton of Hgher Order Moments from Two Multnomal Overdsperson Lkelhood Models BY J. T. NEWCOMER, N. K. NEERCHAL Department of Mathematcs and Statstcs, Unversty of Maryland, Baltmore County, Baltmore,

More information

Problem Set 9 Solutions

Problem Set 9 Solutions Desgn and Analyss of Algorthms May 4, 2015 Massachusetts Insttute of Technology 6.046J/18.410J Profs. Erk Demane, Srn Devadas, and Nancy Lynch Problem Set 9 Solutons Problem Set 9 Solutons Ths problem

More information

Computing MLE Bias Empirically

Computing MLE Bias Empirically Computng MLE Bas Emprcally Kar Wa Lm Australan atonal Unversty January 3, 27 Abstract Ths note studes the bas arses from the MLE estmate of the rate parameter and the mean parameter of an exponental dstrbuton.

More information

Simulated Power of the Discrete Cramér-von Mises Goodness-of-Fit Tests

Simulated Power of the Discrete Cramér-von Mises Goodness-of-Fit Tests Smulated of the Cramér-von Mses Goodness-of-Ft Tests Steele, M., Chaselng, J. and 3 Hurst, C. School of Mathematcal and Physcal Scences, James Cook Unversty, Australan School of Envronmental Studes, Grffth

More information

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification E395 - Pattern Recognton Solutons to Introducton to Pattern Recognton, Chapter : Bayesan pattern classfcaton Preface Ths document s a soluton manual for selected exercses from Introducton to Pattern Recognton

More information

Statistics for Managers Using Microsoft Excel/SPSS Chapter 13 The Simple Linear Regression Model and Correlation

Statistics for Managers Using Microsoft Excel/SPSS Chapter 13 The Simple Linear Regression Model and Correlation Statstcs for Managers Usng Mcrosoft Excel/SPSS Chapter 13 The Smple Lnear Regresson Model and Correlaton 1999 Prentce-Hall, Inc. Chap. 13-1 Chapter Topcs Types of Regresson Models Determnng the Smple Lnear

More information

DO NOT OPEN THE QUESTION PAPER UNTIL INSTRUCTED TO DO SO BY THE CHIEF INVIGILATOR. Introductory Econometrics 1 hour 30 minutes

DO NOT OPEN THE QUESTION PAPER UNTIL INSTRUCTED TO DO SO BY THE CHIEF INVIGILATOR. Introductory Econometrics 1 hour 30 minutes 25/6 Canddates Only January Examnatons 26 Student Number: Desk Number:...... DO NOT OPEN THE QUESTION PAPER UNTIL INSTRUCTED TO DO SO BY THE CHIEF INVIGILATOR Department Module Code Module Ttle Exam Duraton

More information

28. SIMPLE LINEAR REGRESSION III

28. SIMPLE LINEAR REGRESSION III 8. SIMPLE LINEAR REGRESSION III Ftted Values and Resduals US Domestc Beers: Calores vs. % Alcohol To each observed x, there corresponds a y-value on the ftted lne, y ˆ = βˆ + βˆ x. The are called ftted

More information

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal

Inner Product. Euclidean Space. Orthonormal Basis. Orthogonal Inner Product Defnton 1 () A Eucldean space s a fnte-dmensonal vector space over the reals R, wth an nner product,. Defnton 2 (Inner Product) An nner product, on a real vector space X s a symmetrc, blnear,

More information

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS Avalable onlne at http://sck.org J. Math. Comput. Sc. 3 (3), No., 6-3 ISSN: 97-537 COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

More information

/ n ) are compared. The logic is: if the two

/ n ) are compared. The logic is: if the two STAT C141, Sprng 2005 Lecture 13 Two sample tests One sample tests: examples of goodness of ft tests, where we are testng whether our data supports predctons. Two sample tests: called as tests of ndependence

More information

Joint Statistical Meetings - Biopharmaceutical Section

Joint Statistical Meetings - Biopharmaceutical Section Iteratve Ch-Square Test for Equvalence of Multple Treatment Groups Te-Hua Ng*, U.S. Food and Drug Admnstraton 1401 Rockvlle Pke, #200S, HFM-217, Rockvlle, MD 20852-1448 Key Words: Equvalence Testng; Actve

More information

ECONOMETRICS II (ECO 2401S) University of Toronto. Department of Economics. Winter 2017 Instructor: Victor Aguirregabiria

ECONOMETRICS II (ECO 2401S) University of Toronto. Department of Economics. Winter 2017 Instructor: Victor Aguirregabiria ECOOMETRICS II ECO 40S Unversty of Toronto Department of Economcs Wnter 07 Instructor: Vctor Agurregabra SOLUTIO TO FIAL EXAM Tuesday, Aprl 8, 07 From :00pm-5:00pm 3 hours ISTRUCTIOS: - Ths s a closed-book

More information

Difference Equations

Difference Equations Dfference Equatons c Jan Vrbk 1 Bascs Suppose a sequence of numbers, say a 0,a 1,a,a 3,... s defned by a certan general relatonshp between, say, three consecutve values of the sequence, e.g. a + +3a +1

More information

Bayesian predictive Configural Frequency Analysis

Bayesian predictive Configural Frequency Analysis Psychologcal Test and Assessment Modelng, Volume 54, 2012 (3), 285-292 Bayesan predctve Confgural Frequency Analyss Eduardo Gutérrez-Peña 1 Abstract Confgural Frequency Analyss s a method for cell-wse

More information

Chapter 5: Hypothesis Tests, Confidence Intervals & Gauss-Markov Result

Chapter 5: Hypothesis Tests, Confidence Intervals & Gauss-Markov Result Chapter 5: Hypothess Tests, Confdence Intervals & Gauss-Markov Result 1-1 Outlne 1. The standard error of 2. Hypothess tests concernng β 1 3. Confdence ntervals for β 1 4. Regresson when X s bnary 5. Heteroskedastcty

More information

Parametric fractional imputation for missing data analysis. Jae Kwang Kim Survey Working Group Seminar March 29, 2010

Parametric fractional imputation for missing data analysis. Jae Kwang Kim Survey Working Group Seminar March 29, 2010 Parametrc fractonal mputaton for mssng data analyss Jae Kwang Km Survey Workng Group Semnar March 29, 2010 1 Outlne Introducton Proposed method Fractonal mputaton Approxmaton Varance estmaton Multple mputaton

More information

β0 + β1xi and want to estimate the unknown

β0 + β1xi and want to estimate the unknown SLR Models Estmaton Those OLS Estmates Estmators (e ante) v. estmates (e post) The Smple Lnear Regresson (SLR) Condtons -4 An Asde: The Populaton Regresson Functon B and B are Lnear Estmators (condtonal

More information

Basic Business Statistics, 10/e

Basic Business Statistics, 10/e Chapter 13 13-1 Basc Busness Statstcs 11 th Edton Chapter 13 Smple Lnear Regresson Basc Busness Statstcs, 11e 009 Prentce-Hall, Inc. Chap 13-1 Learnng Objectves In ths chapter, you learn: How to use regresson

More information

Lecture 4 Hypothesis Testing

Lecture 4 Hypothesis Testing Lecture 4 Hypothess Testng We may wsh to test pror hypotheses about the coeffcents we estmate. We can use the estmates to test whether the data rejects our hypothess. An example mght be that we wsh to

More information

A Bound for the Relative Bias of the Design Effect

A Bound for the Relative Bias of the Design Effect A Bound for the Relatve Bas of the Desgn Effect Alberto Padlla Banco de Méxco Abstract Desgn effects are typcally used to compute sample szes or standard errors from complex surveys. In ths paper, we show

More information

= z 20 z n. (k 20) + 4 z k = 4

= z 20 z n. (k 20) + 4 z k = 4 Problem Set #7 solutons 7.2.. (a Fnd the coeffcent of z k n (z + z 5 + z 6 + z 7 + 5, k 20. We use the known seres expanson ( n+l ( z l l z n below: (z + z 5 + z 6 + z 7 + 5 (z 5 ( + z + z 2 + z + 5 5

More information

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems

Chapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems Numercal Analyss by Dr. Anta Pal Assstant Professor Department of Mathematcs Natonal Insttute of Technology Durgapur Durgapur-713209 emal: anta.bue@gmal.com 1 . Chapter 5 Soluton of System of Lnear Equatons

More information

The Ordinary Least Squares (OLS) Estimator

The Ordinary Least Squares (OLS) Estimator The Ordnary Least Squares (OLS) Estmator 1 Regresson Analyss Regresson Analyss: a statstcal technque for nvestgatng and modelng the relatonshp between varables. Applcatons: Engneerng, the physcal and chemcal

More information