Bie HME HK DERE eB FA BRT eas GAOJE JULIANG JING JEXUE Advanced Econometrics RRR RAR RSM FEER SE ADA TR DAHER, RUPE T AT SE A WGK. A, PRAT A T RTE EBAY ARE AGE, EAR BOMB MR, RCRA PTI, ERE ‘BRAUER, SPT tha SHER AE ARASH. EVAL Bk EE PE AE CES ITE FO TERITROWARG RUG, SE TURE SER TER RE SENSE EAT BS A BR AR PAR AB (BR ALIA HS HOLS fl aah. AAS H RE ANE STK EERE, LE ER SUT RSE AE SIE. WIM, RRR OL DEAE Gem r A TLD PRAT, BORER E RS SME. RH SBA AMER A, WT, BAAS RO TS PBC SAT. IIb, AEST REI A TRA PT HORACE, DPR TEI FE ATED AAT TE SEE. GRRE BETTE. ORE. PEEL BHR SERIA IE A Hi HT AES PAR, LT PE RE REMI OEM SPERMS ST PA TERRE (CIP) che BRAS Ma. A: PRCT HTL, 2011.7 ISBN 978 ~7 — 04 032424 -2 1. Dit Wit IL OP RRR aE IV. F224. 0 5p ELAR PAIS TCP SHAT (2011) 8 099383 5 Tea BORIS FUE HeeTE Amit EE He MAI FA Sete EP SERIE AL fe UST SCRE HCH SFiAPEIS 400-810-0598 Hh Ae RI PK 4 fH http://www. hep. edu. en ita 100120 butp://wew. hep. com. en eo) ABR rnc PLAT ‘www. landraco. com FF 787x960 1/16 Jntp:// ew. landeaco. com. en Fl ok 20.5 - & 0TH ER 2011 ET AR Ne i md FD eR 2011 7 AS RED WRAL 01058581118 th 4.00% AHMAR . ETL, RARE, EWES eA TRAINER. LTA BLAS E 4 BS 32424-00FF Ill THREAT REEERM, ARBHYE EEFRERMTARE HEE FRRERG AREA. KHATFARES EH -ABALR, BELGRAFABHE, SQEPHAREREEZAEA. UESFERMS FEREARUOM, CLBRABDRBEERERN-REAIR, AB KRARMPA TRUBS HERB DAH, CTEABHE, OB % RUE, BARE, FCPURMAFARLAAE BRU ESHER BAM, HT EAMEHTESRERERT AIS PASS HE AGB -HENUEBHEA-MENB, PABRARAK SRA AEZRE, YH ERR (mathematical modeling) #4 % 4 iE (empirical validation), a Hitt EA HS ty — MEAT EE AP 4 fe Ao KERRY BE REAF, PHAR MUESHS A AFAR REAR PH. RE, UeTURAFSRSEPFAM ARE, HEALS KX MALEATE A NELLY, MUTA KLAR, BRB REAR. HOSP A-ROBAMPRARE. DADE —RPl SHREK E, ARXRHBEREAAAYERTRR, CURR MOTME. BR PREBHE PMA Zo AHERN IAHR (mean square error) Hit W* PERT RH. KERN IR RB ee ABB RN RR EMRHRE SMTA SAX ATT MAS it. HEENABRBRBAH. ABH — KP BRB T PAT RAIA A KU, RH PSRERRRAARAADHEH. B HIT MHD, CRYMUMR DANAE HE — HEBER E ANE CE BA ABN TB Ho PSR) = AR EL aL RWREALR AWA, RACK RARED RAS RR ERD Reiit. BORERCBHET ASASRREAM AH OMT RAR Ho RWORAAKMT MUNADRERDARAEATEARK, REBRR EO BRA PD AAARARRAH. KHRATARAMTH KR. MRED, FRERRMKEAAATRLAD TAMURA, TER, 4SRHARu Roa BADHELEEARH, ORHRAGAAZAE, EARANRT, BAW BAA ERT AE A. BH, WRAL MAT AGRE, REAR AE PRAARU AERP T, RH, PRR RAR SRR EAR it %% (heteroskedasticity-robust procedures). PEEHAEO ABLE APRA ARAKE. BH, AMARA POSPEAEA, RS, LAWRARERREA HA, MATA AAA PAW ARAS ORE TAP a8 Fh te (stationary ergodicity) HH fl FIRE. BRERKAPAROS RELAY, MHEAGR HERA BE ROB. KAT, ARBRE REA, AT UTARART AMA, & BAR AK MH — thy ¥4e it H (long-run variance-covariance estimator). HCRAAADIRAATLARE GH. ERED YS, TMB ZARUFRRESAAAR, TEROPRARSRAEREMA. ARB EE, we -KPUTRAER, A-AAARREA RM . ABB IT RADI RBH BHR AKUA Ko BNER BT SBA, CRATE BN Be EE RAHKAAE. RELHAK, LWBEAMRA REAR SRA, TH BRAKE. REGU RRREAT HU OSEEA GY HRD SH BNENBER TAGGED ARAL ORES ERT ESS RA Wy DME Tt (maximum likelihood estimation) 1s * M444 it (quasi- maximum likelihood estimation). KERR] #STEAKEAF, MTR DLE SH ERD T ARAMA HERMARADRT KR BPEMARHC NU ESAPERB RAY RUT RA, EES PATER E NH A. APHRET LSE A. B-,AEKFREMALN, PEA -TAMAK —HBRER, E-TRARERAMRRUT RARER. AEP HLH AGF AH, PRAR, AHS, HHERTAPE ZMH NER HRA. BX, KPRENUH ESF FSEALMBH, UN HEARA SM URAFEABREK. AM, ESAS —-HEEHU ESHER HEM, REY BBE, SHEHRAAT, RAV ESHFEH SH RWERE SA. B=, AEP ESHER, ERR ESM TARA aT FEAEAWM KR, LMHAPAD HW ARMA KAS EB ETA AAW A, EBT RAED HAR, FE BAER oe RR UESRFHM. HIM MRLEAESE SHES SRR MAR AA, tate i —+ He, Fw H [email protected], K#A RRS kA| um AW EBT RE, HEA EA Fe MS KAR EA HF th — MAAAAMBUESASE ALES (SOAS) MPR ESHSERA (http://www.wise.xmu.edu.cn/cean/) W%#24, HAE, AHEM HY FERRE E TESTER AH RPRRECHABEMRRAY, HEAR. LMRMAR. WKAE BERARAFSMNASSSLRAA ARR, BOER, b RAT FANRHTRSRENAM ST, MAPHORARABH. H RBGALGR-APRU HAF PER BS, PERRET HH ZF 2002 FERN, BERN HREN TH, RRB RCRRLDT TAARH. BNAFBHERARRRER, URAFSFHARRHERR KERR TH BRE TRH, HT KP KR Aa, MASE LHAHFRR MEERA, RWFRHEMAT URLS RR Ee OH MRA TAD, FRARD ARAL. Eh, RMMEMREROR Ho XP ETB RAE ea Et HE RR 2010 HBF Bie Pes He eel F 2011 FLIER T SEALY NG BTB-E WESHSSit aH A SUR TRAPIEROTHE - AB FIER —MEVAD TANABE aw ao Re" aon RE i Bao E aa PBB BESS HRB A ARIAT PRMMBYAASIASHHE eee e ee eee cece ee ee ee seen eee 20 SUE BAR ASAE UY eee cece eee eee eee ee ents este ee seen ease enes 8-H wa Hat aos RE BAT Ht WH sty WRI - AE RE A EL OLS (Bt REASTC OTE AIA OLS fait OLS fit 722 — BA eA SHBRHE --- NA RRE - PSG) SaitRs AGS 20S AUAPAMNRADRE HRD RH RG Bat BO BEG BG Bee BNE Ago Fie FAN R NHR EAD RH sot Rat aoe RE BRE RA BI BG BABE «+ BAR RARFHASMABARHRHREE AD a Bo Rae oy REG BI Roe BN MEP ++ YEE - OLS ff BiB - OLS (fit Bt RMT IE ASHE BL a SARE --- ENFGERE - +114 > 136 IDDUSHR RY cee eee eee eee eee eee tee eeeeene SPARS TEL PPR LB OLS fit —Bcte OLS (ait StF IE ASE We — TA Senne MEF TG EAD 8 EWA AEE FAVA NE + 154 177 Fae ESTA Y UR AEA AUB KMS — MA OLS (it — Bete OLS fH RAMEE AE SHRR PRAT HH KMD — He Cochrane-Orcutt 773% +++ + 182 186 - 196BAA KABA Sth LARSON 5 aoe aoe RG BA Bo BG BG BAB BNR PBA Re Boi Rae woe Bas BAG Roe BN BG Be SAE SAMRAT MPRA ATT aH fae BR BT Bee) — Seti 2SLS Be +--+ 2SLS ALE AE Fa — Wei ESSER S 1 SRBRHR «+ Hausman RY -- IRIS oe eee e cece eee e eee nee e nee e nee + 235 240 + 240 PSU AEE - TD RRTTE + GMM fii Hh BEEBE «+ + GMM fii} RAT EAHE - + 251 FADTEL GMM Seth fit LATE SABRE --- PHRERH - Nei eH BOWE ABRAM MLE/QMLE #)-3ct= RAED AT EERE RIN MLE (¥fi AE MLE #5728 — Beat 295Bo BIG BIT ae Bt BARA THSHRRHE -- Bot ARE - Baw Me - HR ERA RE TSMR EMER RRR QMLE Maia QMLE (MBEIIE — WHEE verre cere reece 311BO Ree WE: TEASED MCZRARRBAS, SRE ERE—A RAGA. HEBHES MMAR Fo FMB HSE — RMT ROKPLRSAMA PHLABRM AAI SRARE, AEE SHU ELH POBRARA k, HEBEL, AAR, RU EBHEPSRBBHRE, MBAR FSMAFHHRMNSRA, AK FERRI —AANBRERERERAAARREOH ESHER A Rl. KRW: KERMA, HEB. RABHR, MRM, ZS Ea. RIE S-P 5] SF TRAFERIRAREM ESAS SORA AS. EEA JAAP, HRA ERITH REP, BUTT AFTRPNUBARLAS, WAU. BH, HRA SE DOAHEV RUT RAES 5 BEAR PABA EZ. Wa, Ha RIVE REARS ARR BELEN MEANT EEL TAP AUREL SEF. EEG LIEN RL. URI RAR BE DUR MADART EH. FRAME BR, RUA PAR PRR. AEA HABIT (duration analysis). 38 SERED, BATT MAE PEER SBE i EP Se RAT RAPE B. Bis, OTe APAS BABE ER OT EA a FADES EAA Tak REMY Ht ak 2 J2 Boe Wee SIG PRATAAE BATHE THB FE TAREE EAA. BOA PEM RR, HB TI EER FUER, BATE RI SIA AE EA ER. DRA ERE TAR GRE RD ir BE Uf ZF AN REIN TRE PETA. OEE. — RTS, PRATER ROR: BMS, HOARE. Sek BFPMU BARE. Th, ARATS, MMAR LAT BARS EA first RH KASAM RAK EBS L EIT BR, eR EAE Sb AR BIT ARAKS BHRAKSS SOLER AER. ERAT BERFEAALERR. ATRRERA MTN, BANAT Re PRT BHLWRHBALR. AAKSMARAS KE, WRATEHRY BE FYE LAER (1) "CSRS A #4 Oe Sh AU YE SESE (empirical stylized facts). TBS REECE EARN NAT See. Le, AMAR, —t ARIE Fal SU RG GE A SJL A AR HR (Engel curve), SRIN Sai 7E fh LASS HAs Se EPS SUN YS TER SF BE; Ee LE, 7 J FEMS ETE ALM HAE (Phillips curve), HABA MAGE Pa A AL AR AAD SHAE; CEE RE, ASR A YY SR GE SES A SH TBF WARK (volatility clustering). GAH, FAR —MA Bea tw FF BB KARA ABA, SRB A FFE BR — PB) BA, IF ARAPRAMTALS HR. RAMAN LAPHAM. Pin, PALA MH BS OFF RE 2h F Nelson & Plossor (1982) #3EiE HR, HARMS BAMA Hw [AFI HE AR. (2) HRAFRRRATRM. RAAMHEERZ I, BAER PARRA HL DAR Se, EE MRA PON. WE b, ATRCRAN ANP RLEN T RPRABAM ED X, BREN LARA MPT, HMARHAT RBH Fe ih (3) AvP A 9% Wik (empirical verification). ABC Hw TAFREZ AN EHEKR, RARKAANRBER. CH —- SRR KA-TUUARNU RATER LED, BERBER, PORE ARE. HR, AE ake REM, FF BHAUEARTRALGES. MEANT RATERE ED VASA RFE AN— BeRa BF ae 3 (4) WA. ~A—-PU RAF SRA ABBE, CRT AKA BAAERL, HRATRCRAT A. DHARMA RARBR. HE PRENF. RFRA RN—MA HER, SIL Gujarati (2006, 8 1.3 4) F RHAAWBES RO WML. GR, HREM NARA RM — MERE MASCLASROTGR. HEL, -SARKERA TERETE RE EP MOAR T Li RT WBF ERR ITE HH LE, MRA EAA ERE: —P EB FACORERR, —-TRARRRNKAESD A (empirical analysis). 1X75 MEARE) RARER T RARE RATA A BAAR. B BAMA, EALCLAR ETA: —-TEBE REAR AKAN Batt (coherency), 3 —7* EH MAE BE. BEAT RATE FRED T AWA FDRA TARE S=7 RAB FATOMFRREARS PE. BEATA ERE MASH AY? AWS ARRLA (HATE, HRA. BPR) A ADR RA FHC, MELE —M. WRAL T EB BE, BCAA AR AT BRD EA, LEG BE RIE — A Ie BE. OTA fi, SAREE, EA, SO ARRAN, HEAD TERE — BEA ABB EAE, Aide e BABI T —M TBR. TE SOE WETS HEN, BAA ETE RN APES RARER A BAMA. Cournot (1938) HY UU MER NB SRE) LAFOTPRERARSNSEL—. REDRESS — id 64. (marginal revolution)-—#e HN eA BC, (dE, CHM AY WSIS, LMA. WEP. WM, SA SRS AMM. MAE R Walras (1874) TERI — MMB ON AR T “773%. Von Neumann & Morgenstern (1994) SH MWR, BEBE FAK, RE RARE EE NB AZ BRAACCREAS? MAM, REESE PER BEE Ao Bt, RAR SOR - TERME. GM, EMBASE ARE (teh GDP. WR. Ak. AWK. AR. RE) CMM AR. Keynes (1936) #1) J-Tt}RARAKGONEMAFEHY. AMS E OT AAA AA AR4 BM WRENS RRA Sek Y=C+I+G RH C=a+py SEH Y 22ILA, C RAB, T RAAB, G EBUTKM, o RAEINBAF, 6 RL AAD. RRB AER, IE RSM, FTG BIBER ARTA FRR, oy 1 eG 1-6 FAL, ULE A LS R= hy PIR SOK, BERANE BD TBE TAN BPEL FEM AUTH, CERES. OTT RL — Mal Fi ASST AT AFA FER HERA GREE RTE RH» ELLE, SF, BRO RALS MARIE AERIUN, Sh —/MAF RA Walras (1874) $2 HACC. WHAT SUPE MEAN, Bde Poe ete RMB AT EH, TH AAR ESA ER AS A ALS TH I TA BH SUATIE, BOCA n TM, i RIERA D.(P), HAN S,(P),n MA ROOTED P= (Py,--- Pp)! s SABE ATES EE (ald Po , SEAR IT A ROTI PS ny, BD: MBTAEN ¢ € {1,--- ,n}, Di(P*) = S,(P*) —fBES8Y OT HOE LAB, (LA oe RP, DO BR MRICS ATT AEE. Walras MAERLEMMARGR— HL. 1B SAE DUG, Ath Arrow & Debreu (1954) #1 TW AIMB SEA, AAD AU LM Soh AH ALIENS TOTO OTE. ROSIE ARETE AE ART CEA RAED. B=, ACEMALS TESOL BILE HAR, SAF AIO AURINEELE. MIRAI “BE” AAI ARDC MAM RIERA. ARH, ARATE HL a mT DLR A, SAAC RA MR HL AS ML UA AT SaO% 424% iE FRA RRRAT ENS — ME: ABMS. SHED Att CHEM? LEBEN REAP BS BE Bete, (ERE RE BRPRE— MAE. MRAM ROR, WRT REA ED PRAMAS SBRA, BARRA RS RM, CREAN, HEB HESO ee iE 5 SRA BIN. EMTS, BRATS, BRIO HG AD Be RAL, AHA UAR BRET LEI MARNAT RR FEM. KA, WAREAF LO EAR RSA RENE? Cb, ARAMA PRA NWRAREA EAR RAEN, HERA ARE. (EE, AIR APHCNRCS WUE Z LE BRMRAR EL. CAREFREE FARE, AWW TEE ERR ADT RAEN TIE. Hei Smith (1776) CARL) PRERARADT RAR SLD LIRR BI. SHR, BRAD PEVRA HH, OBA HM RRA A ET, ABUT BBA SAR, I AE A I TAR, BSED DAL GAURARSAES TA. WEARER ARAN STEN AR RE UTSHER WNBA. CHE 20 et 20 FREAK AEA, TEAFLES RUA BD HB EM TL —-—Gt BAR) (Econometrica) 4 1933 “FY BINH « EMAENLTEY, HARARE REBE, RALBAE: B—, CAR REBBEU RAAT RAETAF OTS WN GE. Bo, BRS HH RARHAREGE. B=, ABR ARR, eT SABE GE BUR BE th REBKS RN BARS RBA PO, UR RAE ie VEU FATABZ EM: (1) FEAT Ba A BEAR BT Be HR AA — 5 AE HE I 9 BE HLL FB (stochastic process) « (2) EMBAARA, HH RE MIER BSL, AL LL AE RE (data generating process) ft) XH (realizations). RNACABRERATAB. CMH RARERAT -TAFRH NEEM RRA. SR, HRENA NAAR RATERS ARAMA. LW, —KAFERHAR RRA LTE EM, oC HAP AS UABA ULI BE. PRT, SMA ER AMY MAH ZAM Grange & Terasvirta (1993) #l Lucas (1977) AHAB RAT TEARS AOE HE, FARE BUTS Lee EAA SEE BE HA AX NE HE. Led, Samuelson (1939) BR HH FO eS — Ha BEAR AF) — EE BP 2) OR A EMSA BA, RM aA EN BA AM. 1X RM PERAT RARE T BSA RN Am. (LE, —SPLRAN ERA WA BARA D THe RE OS ASE IH A, A EA AR AR He BO. Frisch (1933) i£8—44 #4 F438 61 iil (structural propagation mechanism) *J6 SS TRE DURE ASABE BA BRB LP ots 1 REY BS A SE Ba JRE PA BAN se EAB RE ET A. EE, ER Ae ER POERATLA. HAMMAR TERIAL “Bp ERB BEA ZS BB A — 7S BBE PE, AS ST eH HG BBFKAR, ROMENEMATS, GMAFLAAPHER. HRUR AP RAE AE. HT AIL ERIE PPS RL KR, AR PRBAR, RELA AMAA TT A. AUR, AEA BES HRS PARATAH AR. URARAN AMR MMSE PRA PERSIA LE. Aer HE HO FE YA ET ER a, SHEL il EAI #5 AER OE FE HL FERWNE, THAT SRE OAT. SAREE BE FRBARERREA ENR RLCK, WKS BARREN STAG. SET UARAF HE NRRUREE ER, AN TSANG RAB. AE, CNMEOR-TAFRREREBRARE. ATRRALAACAR X, DACA RARE. WRBRERAROTNBATIE. SAREE , BHC NRRL EEA EET BEE PET A Ee A EES UIA BEG, FEAR EEE ET, RAR. IK RELAY, TRARZ RAT RAEN. PREM ER AE NA Hit. LAATHAREE RRMA EH LO RWARRA, Bi eS Se FA SSE A — BR “ATE”. TRH ER, “Be BT Ae EB, EEE PT AE LI SEES Ae AS TE FRR. REA HPAI Rial, BARR a AR EAS TERT. PRT, 24 BT AUS A DU, 4ST A FE a A Se HH TY REFER LA BGE, SIITER. REL, FOREARM NU RAFAH ES LAM SRHOHRM, MPABM RR. HHA ZHP ERAT BCAA. SERRA RA PRET RS REN LA, RAAB A fee RF PE, HRAR LEARN OE AE. ob Hansen (1982) seth Ay” XMiT7H (generalized method of moment estimation) KRAEHit MAPLR aT fee oh SE PE A i eM BY, AA TD ST A RH, FARBEMLARR EN ABE A. FEAF, FEB RK UR IT. ALALAIEAYAT (Engle & Granger (1987), Phillips (1987)) & BBE Nelson & Plosser (1982) HGF, MTR S MMA A ASAE RANE 7 PUSIACTR ACA TE © CER AIESP BRIS TAP BFE SENET HEMT ATED, MOA BEE — 4h PT AAA REAR EN EP. Se BARA BRS EB BE Es eA Td JF SB A — AE, Et BF WAH NAR (volatility clustering). ERA. MIRHEMABEER. ARETE ARAB AR EARLS. Db, Se RBA ESE AGA RE TB Be Oh HR [al LE BER AT BEPC RA 5 EME RR) ET. BE ARE EA eT I AE CENT FD HY I SAK OR, SAT HAT SRW RARER. HOARE NATERRT SHARE R APM AERP EE. MALE, i, Be Fi) BEE TRAE, MBAR GH AS eS ET ie SHER, HHATAN REAM ARR. AT, HT ROA BE FRMER, AR T RR Ee FRE AEE (Le HUAIEYE (censoring) #1 PYZEHE (endogeneity) MII) Ne ZE, BOM BAY DR RAE RB PU eR. 4 Ct RAYS) (Econometrica) HTS, RA AES BA Fisher (1933) Mit RB GE ERA A RA RE YE EET TOE NBA: “WRAFSSNER ARAN EEM ERK, BEM PEMA S NA RATAN AR, (LUE RE Re Es BENTO ER RBS SUR I. (BORE REA REAL, FEAT — Fy YA EAR FEE, BASU RAP. A, Ht RAFEREAFAU YS, WARS AT RMA RH, RE A BS ABB RHE; AN, RARE RR RAPA SH MR, RAK SECA — EB A AE; SBE, StS. PIO, B= ABR ERE PAD BER HF, METER. AEA ERUAGS, THERES BMRA, TORT RAYS.” SED WAC HRAR LEB ATR IE, EARLE TILT: B, FHRREA BHR LASER (HNREEOAD RES). BO, REAFACALT RUN EME RARE. B=, WMWRRBH RRM, FEBREIN. AT ARRGRRA RA EES AP RE, ERAT OB PASE SORA T AA REN RAE ESB8 Be PE —, SANTEE. RAMA IL 18 FLSA OT FB 2 Y=C+h+Ge Cr = a+ BY +e StH, EAT, C, RAL AGHM, Tp EM ABLYE, G, RBUNSCHH, oI MARA. SKa M6 AARENBRES XL: o RATANRAKY, 8 BIL Bas BRA. BRESCIA BOR OF dL I BBA oY 1 0G, 1-8 AT MBB AME SH KG, BA ORK. BEEP LB FEAT PRAEAEAE 8% WASHER, HER AUR AAAI BOBO, EBL EE BRATS KAU BOE . BURSA i2>, HEMEL SEER BE ADRK AR, TUBS LS BY RR RH BRS, TR AR PP ET BUG, TORT UR BI BOK. ARLE OLE HEM RATA MBRZARAEMKKA, HMFR-BRARAW 6 HES, BB WHRVUEM ER. PIER TT SSN TAMAR, TOU, A AN STA. WREAK, EAA RMR, 3 ate aT BE Al. HEH, WRAP T WUBI hit BINT T A. BSE -E, PFO HAR RTE, PER I TT RAY, HARASS. HRB HT Se PERT ACI BB Bait, BARSAAH AW (BM Hairdle (1990), Pagan and Ullah (1999)). =. BERS AY Ee MB — AAR REE A A TBE Ht DBR FCF 2M (constants relative risk aversion utility), HAE IA a ore U=>0 suc) = So — RMB RH 6 > ORBAN AMURAT, y > 0 EAR RBSM, u(-) BARME ANT SAAS OD A BI, C, ALS ¢ NO. 1, RA MATA TAY fi, @ Re = Pi/Pry WH t— 1 WABI ¢ HNP CRS (BB ¢ — 1 Me POA Pra)e BART RES I (C.} DAI SSSR UAE 9 max E(U} max E(U) SRS eT Cet Pie < Wit Prge-a S'P ge ESS t HP, WE tA AA. EE 3% (marginal rate of intertemporal substitution) 4 du(Ci41) \ Cr \ urs = ard ~ (%) ac RP MMS AE O = (6,7). WER — REA E[OMRS.4:(0)Reslld] = 1 US A Bax PARE A Ck RS AR 1. SB RL ABFRENKHAE (ESATA Hansen & Singelton (1982)). BO APL ABs HK — A EAL 2a fH HE TH EEE? FL, HERAT SR AWM TIE A EE, Ae RE AR SUE» WADE ARH T SET KE ERR EN — Bt BK, KA YEAR a BEBE EH EB SR A I, EE CBI AE (Hansen (1982). BRRARA, MARRESR 7 OH), RRR Sy AUT Wh Sy We ASR ZF AY AE Ha, IL HY AL HE HZ AK (equity premium puzzle, Mehra & Prescott (1985)). Ju T MIX, APPA HE HST SINE, APUG AVE. Campbell & Cochrance (1999) RHMET AHR eR RRER PL. WARE T WM, WELT RATE NED, GARABHRM NAA, MRNA HCWRE. E, SP RR BRAM, AMEE ER—BAA L, MRA K,, AP Yi, Shi Rea i Pah. Aub i AEP BABE (Li, KG) BUF Y, NBT Yi = exp(e:)F(Li, Kz)10 BOR WREST SOP ec, LMALAR (tem, HRY, BAAR Hh, o RRA SA EE FUR HAT). — SEA AB EAR EPR EH A (constant return to scale, CRS), TAU FRR WFAAI A>0, AF(Li,K;) = F(AL:,AKi) AERIS FE BACH OEE ERE RET AAR, TOA SR A, WUT APE RE. TAR PRRAA A, BUT LAR WAM (regulation) Hifi. Alt, BICEP BEA RARE ERA EBS NL, A ATR BR AT DBE. TESTU ICP, LM BAT A — ER RITE PR BORD AR U AL Ui, Ai — RE ee F(Li, Ki) = AL?K? LET EP BRE, HEMMER SH (cr, 8) HSE Hyp:at+@=1 MR a+ PO > 1, WEP RRA HM MAD SB BON LSE HE tO. HE, RS REAR ABRERA BE MEMBA, PMRW BBA). RAPA F, t@ BARI, HARA White (1980) BUNA -RERHA ¥% (heteroskedasticity-consistent test procedure). ERAN, CRS StF SRB Ho: a+ B= 1 BURLAP RBA BAG — ERT AE BON ATE. SE BOERNE RL, HARAE CRS BEB — PARED. URI BLA TE, BEECH AP SOPRA A — WRT, MBA CRS SBMA Hy: a+B=1 PRES. KH, EP RPRY PARE RH RAFSSR AOR E ECSARMR NARA. DB. RHSFHSALEAR FEA IE — A RA — ETA Pe (ER TH) InYie = In Ay + ln Liz + Bln Ky + yBonus;z + SContractie + it Sop i RRL, ¢ REG. Bonusie HME tHE ABP RS A A HEL $l, Contracti: Ze i PAT TARA NTA Fe RAB Eel. XA —/ MARAE! (Hsiao (2003)).SSA TORRE IAL 1 20 THES 80 FEAR, SETI ste AVE TT Tel eB AD ASP Dg A RE. BOL AT, PA RA Ee IA SE. PHARM, AREA HT AB IE, ATE EY 47" (Groves, Hong, McMillan & Naughton (1994)). AT ARAMA ENAR, A WRF SRA Hy: y=5=0 EH, EMDR c, WAAR Ne, BH to Pe ABABA, AIAPE HE Vie ASR Bonus, BY ABFETE MA RRA: WALA FSS DRE HT, APR A EL YE Se RHE. HE RA Gen GR Bonusy ZH EEMK, Mii F BOG K ) — FH (ordinary least squares, OLS) {fi} RRERSERM y KH—BUH, TZ IN t RA FR WAAR. PEE, PRAT ERE T PE TK, BTR (HTK (instrumental variables estimation), AEA MHL MRS WAR HK, MADRS SHB, FY APR RREM Se ABR T OEP R. PARRA, CAESAR A Fr A RE I ABBE MRT SHU Hy: 6 = 7 = 0. MRK Hy : 6 = 7 =0 RK EA, AEM EAT AERARR. AWE LRT RIA eT, eR BORE RAE A MN, ROR AE RR — AT EAA. Led, te A RES AAAAMRANIGEP R (BHAA RAM AR) . Ak, SRB Hy 6 = 7 =0 RABE, RAB RA RDB LRG AT BRN TER, OA RE TE REAL. Hy BRA AD SA Th ic a EY PE AY, AG t PAM BR HEAR, Tha = {Yin Yi-2,..} Wt — 1 MA RR LW RB. SCA MTT IHU (efficient market hypothesis, EMH) $A 2 BBSRC i Gk AS PO ASR NB BD E(Y|h-1) = EY) KE, AWS ESL T,X RA, CRRA tC — 1 AAMT BAVA AS. ITE Y, SAAS, EEO HEHE Th HE (buy-and-hold trading strategy) MU TUS ah 3, FE HH AY THF ah GOR EME RST, HSCS Aa SA RT IN AH IE TB FH EMH f)—-*+ BEA MEEGAE SWRA HC, PRAM RD TAFE EAE TI ©12 ee eS #2 EMH BM FRB F » Yi = 00+ Do as¥e-g ter = SOP p REAR ENA EM, e LMP AT. EMH ROR Hy : ay = a2 =+++ =a, =0 TER RAM a; # O(1 0, a; 20,1<5< 9 dla; <1 j= {ar} ~ ii.d.(0,1) SHAY AAR Se Hk Th HB RAIA — Be AE SERRE HARROP RARE. BARRA, Leo Bollerslev (1986) fi" X ARCH 8) GARCH 8), 22s e i BRAT OBB TZ. SERRE 1, 7S Ee a) AE fe. IKI, BAR AL ae PRM E 0? BARE IEM, (BY, MRE RAM HY, A BRMLRAVT {2,)} HAAR ATTA AR TY. DA, HF BAMA fi cK (MLE) ARRESE IH. PRT, PT ABUL {2} REINS} A IRE IE ASDA (.2.d.N (0, 1)) RA RMS BERN DA (RAB TARMBSM). RB E, TAR BE La FY, RRA, A TTAT LAI MLE PreK HU MBSR, RAF {2} EB LIFR— BA L1.d.N (0,1), HUEMIRAAARA HS Aa VR — BOHM REMERON MLA REKE 2 DA BRIA MLE Ave BA UE A 2, IE a, RT ASB) BE TTT HEAR GY. SKA ATR AUR ih tt (Quasi-MLE, 5&2 QMLE, BIL White (1994)). 2 QMLE MUSH HEMTASEF MLE (St HED AS ADR, Hobs FU Ee RE QMLE PARE EA. E29 QMLE fii RNS 2 SIEM MLE i A A AY, XM, IE TESA 22 FOOLS fiche TE 22 BF REA I 22 FOOLS fit LE #, 2-FN. QMLE PRMAA AH RN SE, SP BRS HEMTAIS (HS tL White (1982, 1994)). +t. BRAM EARRAP, AMM HBB PB: — TAMA BEES KY TA AE RARE, AKRLARAP RZ MSS KN, BS TRINTTAL, PERE A EAE FAIS KMS TAL, {FAM AR Re Se SEL, (Lh fg A UR) BYE ERAT AST (duration analysis). FES, AATRT AEE BLA eS EF EES ACN TA BUR. FF THA ARES (hazard rate) ALM RES IE AS A, BAER14 SiR THA HOME. RRA HT APE RSI LER, BEE OR. ERAN Bee. ET, RPA PANTS RI Ta], FE EBON f (0), WEES BA F(t). WAR fERH (survival function) 4 S(t) = P(T; > t) =1- F(t) ABE Ptt) _ f(t) NO) = fig PES SEED = HOLE, UR d (t) EHR —TA BASIS Tet, EMT ¢ BRAN WRT. EB A(t) Al fe) MARA, Rt A(t) EET RE BERL f (¢) AR BLE. UA BREA AER, HY D (¢) BARA SARE. RUBE AT REALTOR. 2 T Pets A (47 HE HE (heterogeneity), RETA RE HOF RE RG EEE X,, ATES A(t) = exp(X{8)A(t) 3X — $A Hs Wy Ee BAB EAA AY (proportional hazard model), 4% #] HH Cox (1972) #8. BM _ dimx(t)__1_ A(t) =~8x, ~ XO Ox FY DERE Rt REE X; WER ARTI. TB SET SEIT AY CAE AAT PEE TY SA. HNN, ABRIL, FE AM i ADRES TAL, UT AS REBIT AT LAA AAT RAR HE, Het SEW. SBCA FEBE. PERI. ALEVE Rafal ERT TBE). RT ERDDTAPAH ERMA XM. SEMARRME X,, AIDE ¥; UA CER BE A) = Ax()S.(€) ‘ SAE S(t) = ewl- | d(s)ds], ATLLAU MLE Jett ar 3. Kicfer (1988) 97a MAINS OTT REVERIE. SFA HAMM, TSIM Lancaster (1990). ZEAE SHEKHAR, ASML E10 PURE ERB RSME CEES, RAT He ST SL SUE ARIAL. ESAS HN, —-MEDPR BEAT a: eH ASHES. EAT TFL ISL, URL AEM AM PF HE, TREST RPP SE 15 TYRE AEE ATE AE, MURS AR AE. FAY, UR FREE ADRS AA BASS TES TAN H AA, RAE WMARKRMERKA, RAR EANUNBAFERM, WS LA, xt RAE T EP EERE. BRD Tea TAHIR REARARNATECS BRN NAO, BARE N SM RISB AEE (Heme) ARR. Blan, APF Se RAR BR BEMAMA. BM Granger (2001) TEMA AMR ROE. BEATA? FUSCA SURMISE DT PE, RE EES KA HD BA FEMA ER PMT A” AT. ART, BHP IER EA RL ESAT EM, AREA BEET SORE. HEKS PARMAR LH AAEL ONE. RRIRT RARE ATR JARRE. BT5G, HEART ASE A , RLS A, 1 ME Ze SR BE Pe BE EA BRS EE DN, RT LU SH ec PH HER, SS CRE ALAC SN HI BRAS TT OLA, HCE FRE PES SE EE, (FE *p, AT SUA RR ET. GORA RAN, GRRE FY A LEE EN OH ERR ES. EAE PR, ME ER EI DURE AS, KS ES PE CIR EAE SC PENT, A BSR LSE Te AR Ae Ria EAT BH. FEEL HR A SEA POR BE FY DAY HE EAT AR Dr BE Sl, BB Py FEB FE HE HES PAB AT (Samuelson (2005)). RAZ, MMA HBS HBG RE, J HAN AR RATA. (AE, TRAE BIG A BR. EEG, AREAL REF 1B LADHE KAA URE T AS. MEL, TSE RAR, REM? NATE MHRA E MARNE HTT NG? HK, RLBFE—-TAT RAT RSNA. —S ARNG REL WIEN REATREM APE. be, SR METER GDP AEST KH {Ki}: Ya000___Y2001__—‘Ya002 Yoo0s__ Yao0a _Y200s__Y200e__-Y200r__Y2008__Y2009 84% 8.3% ——*9.1% 10.0% 10.1% 10.4% 11.6% 13.0% 9.0% 8.7%16 SMe RPE ARUEGIN GDP WK ENG RAED SER, Ake RLY, A SEMEL! WR-TMMRELA- TSA, RIERA TN. ERATOR 3B, RU DAT HAN BEA RA EN ARB OP” TA. WH THAFSEETAU AM, AERA BARA BRAT RRAA ARB RT TI RRS REE TEAS “GETARRGE” , TAT DOC AME Le AY Te USE AR BE, ETS AT. ESE LAE WA A 4th) “SEARHE” (stationarity) ak “[ALMHE” (homogeneity). ERLE, A DLAs LMU BEL JA FA — BRAT AB PTE A EMA PE ORIN. APPEAR PU RBRLRA T MTSE EME T REBRRA. HPE SC ES PEALE A AE TE FS TAF AS EN R=, BRAMAKACERAN RE. APRA HA. AOR ‘SBI BE SCHEAIAE AL, BLHIFEH (regime shifts) Al4;#4)2E44 (structural changes) 4: ERB ASTOIEMI SH. M—-TDARBE NAAR, AEMETE ASH (out-of- sample forecasts) FARRAH W . SAGARA BEM, BIE —S HE PTE ARMA RE, wR ERM RAE TR. EE BAJLT-SE AB, ERE FE AE Nb BS FR BAIN EEE a OT BE FR. Lei, Chow (1960) i TRRAFKAT ERED ARNAUD HK. Engle (1982) #2 tH Heh AY AYA FA Se a EY BABUTOM. LE, AAS SEN AP MT OR — KOR. RHARMPECEAA, ARAATAAMS AMER RE ERAN EIN. BO, MHRA. EA RA LOIN RAR ARE AR. BE, SHARE EPESHRE. ST REM RRA RS BE ES AFRL PRE ULNHAPFREL ARAN. —HARPR RA RARER FT EAA (unmeasurable), “24H RAEN AT MEERA. HFA, AE #44 (sample selection bias) BAM MRAM A. bein, EHR, THA TSE RL BA KDE BRA GDP ibs, Alt, GDP KA RR RURHATHEE. SR, ATH HMRANED. STARR A, BRR ATE. RH TR Ae A A ATT DR RABE ROBIE APFRRMAP ARAN LIME, AOE RIE MT HAE RE WHE, 8 MEDS RPA I BELT AY AE.& 17 Soe iW ABN EARORPUA RAPS NEAB SS MSE, Ure BRE SRAB HE, MERU ESHA EAN. BT TARAS ESP E, OCA RAISED HT. XE LRAT ERS RA FRA TNENER. FABRE PASIAN, WRB E ETB RPA. CMATHC TR BM, WASHER RY, FA LSA ATEN. HRTHRAFENTEOEN, ELARPA SRS PAE RESAWATUBARS SREP EA NA. Bai, AAR FETT SNL, CRUBAFLOT ARETE. FERRY #, RERRARRAT ESHER, BAR MES A. AMM EER HAR (2007) HLH. ARM RAR ANE Stik, EATS LETFR (2007, 2008, 2010) FAW XH. & 5 BR LL SRB HR PREAH BRS BIA HE 1.2 PBA ABLE TA? eA ea BE HE? HO EAT TE SX. 1.8 HET ARETE A? CATT AE HI? TL HERD I Ae fi YE AT BIR. 1.4 ERA ARB LEB ZR EP TT CSB NE.Bo RWS ARE HE AFPBIBAHHBLAM, GAPHAFRREY 5H WLEXLMAAORE AORTA, RPREY HAAR, X th ABREARMBES. XHAY ZMHOBARTATAR NX Hy BORAM X HS RAMNY . ANHMEHESRAH EH ARH 2 BAH SS, HALA wy JAVA E AIR (correct specification) #5 Et, RHRALATRMBFREAA TEE, AR: KHRETH, CHV, HR BR, RAB, w FM, RAR waar Bh RRR ABH, ACSA ( Lehn ¥) RRB IE RRR BLL, AVS ERE (ted y) ACT DEDLE LIM (realization). BEB (¥,X°) EAL, Se Y BABAR BIW?) < 00, X AE (+1) x 1 UE, RMSE — TE RIRN (AAD LR), 0 eR CN SH. HBT, APU BVIX) EAL. BLDLE BE X (His ak tt aeah) FY ( HeéneBIW He) 2 I XRT stn SMES AT MIGRE BL (VX) EER PI, JLB HE BON far (ey), TWX ALIRIESR ee! tela) =|" fev (esaydy eX = 2, Y HERES BAA _ fxy(z,y) fyx(ylz) = “jx(e) SOF fx(a) > 0. RARER EEE fyix(yle) RIT Y HX RBA (dependent relationship), BR# X MY BBA (predictive relationship). WX — RAPER EB, PY DLE MF RBBRERA 19 (a) ses E(Y|z) = E(Y|X =2) = [- wirxtuleyay @) at var(¥ |x) = var(¥|X = 2) = [oy - Ber tay fx ndy = B02) - (Or)? (8) APEANE (conditional skewness) El¥ = E(¥l2))*la] [var(Y [x))°/? (4) & PIE (conditional kurtosis ) S(¥ |x) = BUY = E(Y|2))*\2] [var(¥|2)]? (5) &¢E @ -4}4i2%(conditional a-quantile) Q(2,«) HF KT a K(¥|2) = PLY < Q(X,0)|X =2] =a€ (0,1) HELE, 4 a = 5 iM, Ql, 3) RHF (conditional median). EX = a, AEP A Q(a, > FERED RTE EER TH (threshold). SAREE BRERA fyyx(yle) I-AA (summary characterization). $3} A (+E Hy AUSF EK RE EE, BA, MAS LOVER? RAT HAN AGNAR. REET, St BTR ROR EET. be, EP OE RSE BF PP BF a, BT Ee A GE PS RETR. TORT FH Ah I, BF BT a, IN FSET PT REET RED AUTRE. RAVEMRA TERN EBAFSROPR HIRES. BURT FE FL BG EER, ANG EEMT VY ADREANE20 BoM RIB PABA PATA. UM, WRAS BALE RARE, BARA GEM RED ZETRARA MHD ER Y HARRAH RRA RRR REX SY ZAM ARRK (causal relationship), R AE Vt ZAM XK (predictive relationship). MAZ, FY AFA X ther ATS YR a BY AEE OE HIE, HR BAHAR VEAM X ORR SRY HAE. ARR RT ARH. APH MRE X HELLS SAY WEL, HX MY AAR Be, WRB SNARKARAREE, MA, XY AARRMHD. R DK, MRRMT XM Y AA RUGE, RHR X SY RAHA PFELEMARKA. Li, MAR X MARK Y ZIAD RE ECE EAR RAR, BA CH AR — FEE FH. REIT REA EEE PARKA, AVVRAT X RBM Y HIE. SoD RSH ESHAT SERRE B(YIX), EAL Y YX AIBA MK (regression function), $4? Y #2 SEBUM ARSE (regressand), X BU APRS (regressor). “IH (regression)” id RFI Y 5 X ZAI EINIER, FRR —KR WAR. EM 2.1 — [ EVABM (Regression Function)}: 444414 E(Y|X) #4 Y st X wea Ba, KSRAFRRADARCRS DT AMKAY ENS AY ERA (FS JR Cochrane (2001), Sargent & Ljungqvist (2002). $i da, 2s (9 4528 TH SBE WERK, BEF 5 S215 A BI BE PO EH a As Pe PR BA, AEWA, EAT Me Hr WE (Cochrane (2001)). & FAFACRGARRR SAY RFERN KA, BERABMARRRE SAAR ER. FAT, a AE BE EN. EVAR OT RN RATS PRM NRHA, EARS AEN. il, CUAF AHA RRZ NKR, WAI. MY POAT IS. PAL LF G21 SY AWR, X HAD MAMWA, MARR E(Y|X) = C(X) HAW RRR MAX SRY MR. MRAM MFR VARSAAS Sete SLD 21 (914 (marginal propensity to consume, MPC) MPC = 0"'(X) = aR) MPC 229UGS5 “FBC” TH — MER RIS. Boh, WY GE BATT TAR, AS HE BH RIAN (Engel’s law), MPC f& X MURR A. Bak, HY Lait C'(X) = Aenvix) BBW X (RRB ED IE me P22 YAP, X WMA, RAS RMR MIE, RUBS PR E(Y|X) = F(X) BAL RK, MEP RRSP HORAK. CA ARR RH {88 (CRS). CRS XH Xt FERN BH A > 0, AF(X) = F(AX) Bl 2.3 SRP X MAP RAW Y, WEAR E(Y|X) = C(X) BAK FRB. FEW ERR AT IL, SOUR ACEP HX ARE. BT dE(Y|X) _ OX) >0 @B(Y 1X) ax? JAE, BW ELAN RA BOE X BOSE RYE. Alb, Ze E FARMAN SH. FEM RAE E(Y|X) BZA, BE ECV |X) RHE. 312 2.1 E[E(Y|X)] = E(Y). =0"(X) <0 iE FAL Fi BALI (law of iterated expectations) BRIE. 5/52 2.2 | 9H1S5KM (Law of Iterated Expectations, LIE)]: st# RH TA Hd (X,Y), MBB E[O(X,Y)] A, M1 E[G(X,Y)] = E{E[G(X,Y)|X]}} TEAR RAPHE (YX) AL ESEBEBL RATT. ER a RIE I fay (eu) = frix (ule) fx(2), A22 Sok REE E(G(X,Y)| = | G(x, y) fry (a, y)dady J | J aa [cen pretolad tet dedy [Fee sear Ix (ede = {zee Y)|X = alfx (ade = BEEIGLY)XT SOP B(.|X) BAR X HAAR, TOHPTAY B(-) AER X PEE. B24 AAEM BY |X) HERB MNAT RE. BEY ROALR, X AHHH, CHME 1, HH 0. MA E(Y|X =1) = RHEL AWE E(Y|X = 0) = AtELAR POL WLAFYLEMA E(Y) = BIE(Y|X)] = P(X = 1)E(Y|X = 1) + P(X = 0)E(Y|X = 0) Sh P(X = 1) HHA PRERANL, P(X =0) RAR AML. Bh, ESE SPY SU YO BEF BY AAT SPR RSS TAPES He A Campbell, Lo & McKinlay (1997, $5 23 WL) MaeRhSe A fy Beas TT PENNA RE. AEMARK IA, FAI CJ, OT PARA Bat IZ, J SHES. RES RMU Y UT LRA RRMA eB EY |D) A E(Y|J). BMPBVAN A AAW BY |1) = BIE |J)i] RAZ, SARUM 1, ROUSE Y AS ERS ME 7 SER Y BRAC B07 |J) OBR. ERETE'S% BVI) — ELE(Y|J)IN = 0, BREA RER, ANMELDEN WEY - E(Y|J). FIRS EAPO, IRE (|X) HOES BMRB LL X WA ARE 9( X) WY. WT USE g(X) BE Y HYPER, & X—TVALRAEIM, HI877W (mean squared error, MSE) HEI.IA RASH ST 23 EX 2.2 [97K (Mean Squared Error, MSE)]: ik A Hide o(X) & BMY, WME o(X) HARE MSE(g) = E[Y — 9(X)P MSE £7HUiRZE (BUBUE g (X) SAR Y 22) APPA HEL. MR PK MSE Be), WTR Y REE. ARRAN MSE. 5 ‘$8 2.3 TENE E(Y|X) f) MSE Be). 3E32 2.3 [MSE Sti]: AH E(Y|X) AFA AMHR B(Y|X) = arg min MSE(o) = arg min IY — 9(X)? LPF RAHA TAM fe FF TR HAW HS (space of all measurable and square-integrable functions), Bl F= {a rRE | [eeors(aae < } TERA EX MSE(g) HATHA WE. & 9o(X) = E(Y|X) vy MSE(g) = B[Y ~ go(X) + 96(X) ~ 9(X)P = BLY ~ go(X)]°+£ [go(X) ~ 9( X)]P+2E {IY ~ g0(2)] [90(X)—9(X)]} = ELY ~ go(X)]? + Blgo(X) ~ 9(X))” Sth, HL EM E{[Y — g.(X)]|X} = 0, BY LATERIZE VE EALY — go(X)] [go(X) — 9(X)]} = 0 4€ MSE(g) S}R2GR, B—T L(Y — go(X)]? BRAEHA go(X) MIF WREWA, K-RSRABH o(X) HR. WOM Blg.(X) — g(X)? MH PFPA, EE g(X) AUER g(X) — go(X) WPAN. HE 2 ELV — 9(X)? ARPA ERLE, TTS BME g(-) MORERIE, BE #8 g(X) (8 MSE(g) AML PIER G(X) EM AEK AB He WR EE G(X) = go(X), Wize, WEIS MSE(g) Bek.24 So RIA er MSE 27M g (X) NAb Y BRE —, (EEE. SEL, MUR HL [Y — g(X)| AUER TBS AT RETR g (X) AFR fe. Hei, #4) 44xt4R2E (mean absolute error, MAE) thé--Ph A HEAHE a, MAE(g) = E|Y — 9(X)] {HJE, MSE(9) BD>AMi mR: BG, PTA; TOK, Ce HR EBA THRE, TE eR AA, ANT TE AN A. Ee MAE(g) BASLE REA EP LR m(X), MARR E(Y |X), BRE Y BOA RE A BY. Fee ALK m (x) KEE Pe 1 [i srxtoleray = 5 BENZ, WE X = 2, m(2) HART fy x (ale) ARTY. 12.5 BAL (X,Y) MIRAMAR HERI far (ey) =< e-¥, 0< 2 0 Inxtule) = BED — o--®, y¢ (2,00) Bue Bre) =| afvxtoleddy =e [we =l+a PRM B(Y|X)=14+X BX WRB.DIA PASS oT 25 AT HSE var(Y|X), WARP ABAR var(Y |X) = E(V|X)-[E(Y |X)? Bay B02) =|" Pixtoladdy = [ ye dy ae [vee = (09 (ron - [eva’) = (-e*) o — ae — al erry] =a + 2er | ye “dy =a+ 2 ye dy = 2? +2(1+2) var(¥ |x) = E(¥?|2) — [E(¥ |x)? =2+2(1+2)-(l+2)? =1 Fat Y BARE AERA X, BUX REY RT EA. YOOX WRE X HRB, HEY MAA TEARMF X. ERBZABHE ARR NEABRZ— (SMB=SM) . AS—MITE, 4X ALY ARATE AR A, tT LAB ASR Ee 2 A RE (BRABBAAB 2.2). EH 2.4 [| HAM (Regression Identity): 22 44+ M4k E(Y|X), & ca Y =E(¥|X) +e JH ec BAIA (regression disturbance) , WHA E(e\X) =026 Soe ARIA iEAR Me =Y-EA(Y|X). Wl Y =E(Y|X) +e seo BlelX) = E{Y - BW|X)X} = HUYIX) ~ Blge(X)X) = E(Y|X) ~ goX) =0 iE#. LER c AHRKAD ARR, AWE MATE ARWKR BY|X) WORSI. © RABE Y PRA BAHA EX) HRA AERS, th MAME (noise), MARR E(Y|X) HAWS (signal), TREE E(Y|X) HX RH Y. Ele|X) = 0 RAAVARAM « KAS RM Y RETA XWASHER. RAZ, HARSH Y HAH X Mi Rose BREE APA E(Y|X) Pe APE E(e|X) = 0 MFRBSRAA AMHR ARAM REM, BICMM MTT. M E(e|X) = 0, FEW c MFCR S Ele) = E[E(e|X)] =0 HHe 5X EX, E(Xe) = E[XE(e|X)] = E(X-0) =0 Mx b, e HX FETA WM RR h(X) WER, MFT AT MH AC), WA Eleh(X)] = 0. 2800, AREA EESLIN HK A(X) RW € RAE, HE RRREND LAREN RM h(X) Ac MAAR RUNED. BY DLE A, FEA AT BH AC.), Bleh(X)] = 0 45 Ble|X) = 0B SEAT AY. EL BU ART UE BA IX — EPP FE — MLA EY led Bl (SB MARA HH 2.10). eB, 4 E(elX) = 0 Hvar(e|X) WAM X HR. WR var(e|X) = o? > 0, RAK ¢ NAA 2 (conditional homoskedasticity).Bo APH ST 27 aint, X ASAEALE BIW Y ASAE 2, BA var(¥|X) =var(e|X). (2, WR AVF (ERM o? > 0, A var(e|X) # 0, MEIN EXER EEA ZH (conditional heteroskedasticity). ZEIX#PHAE F, PTD X UU Y ARN. TESA UHAFL TEER AANEEERERAEM AMAR. Hoo, EER FAR FB t ROA P REA (BBS RN tA FRB), ER AA RA TAAAN IRE OU. EMRE, SMUT EMT ie. B26 BR BAPLSE RL €= nV Bo + AX? HEP BELAE X Aly BAAN, A E(n) = 0,var(n) = 1. E(e|X) Fl var(e|X). (1) E(e|X) = E (nV +X? Ix) = VB + AX? E(n|X) = VB + AX?E(n) Bo + A,X? -0 (2) var(e|X) = E {{e — E(e|X)?|X} = E(e?|X) = Ef? (Bo + &.X?)|X] = (Bo + 1X?) E(7?|X) = (Bo +X?) -1 = Bot AX? RGR EX, e MARES, He HAH HEKMF X. FARRER HE? —KT, WR EY |X) HBF X, IBAA A RHE, var(Y|X) AUG AS FARA REM T X. HSE, AARP ET A. beh, BURA DML, SF Hy HD th ABEX. Granger & Machina (2006) MABE PRISM T BRERA AR PERERA.28 Bit RIS AE FRU BUF 220, Se Pe Seg 3S th BT oe a A: RCL AY BAL HB TT HE. $12.7 [BAHL #® (Random Coefficient Process)]: (Bt Y = fo + (Ai + bom) X +7 SEX Aly 40 HATA, A E(m) = 0, var(n) = 02. RK EWY|X) var(Y|X)» @ (1) E(¥|X) = Go + El(Gi + Ban) X|X] + E(n|X) = Bo + AX + BX E(n|X) + E(n|X) = Bo + AX + GX E(n) + E(n) = Bot AX + eX -0+0 = Bot Ax var(Y|X) = EB [(¥ - E(¥|X))?|X] = E [(B + (Gs + Bon) X + = Bo — B.X)*|X] E [(@2Xn+n)'|X] = E [(G.X +1)?n?|X] = (1+ XP E(n?|X) = (1+ 2X) E(1?) = (1+ 2X)? BAUDE, BELL RSAC EN SH A= i TF AEE tt RE He 2s Se BA RE OX BEATTIE. SESE, ATE Y WF RIE Y=StAXte eH €=(14+ eX)n FRPAT'H, E(e|X) = 0, (4 var(e|X) = (1+ 62.X)?o? 2 X HBR.aT _ Sete 29 B= AMER BEM, AH go(X) = ECV IX) BYOAIR (MSE) BSR, A 9o(X) = arg min IY — 9(X)? EPP LASHA TWA PH RRA, BD p= {ont oR | [ Pe2rte(e)ae < oo} ATT, EVAR BY |X) BX MARI. APB RAH X WY ZA EHERA (HMA RAR Y BaP X BS BO), A BRIER. iii, BEART BERRI 96(X). MA, UAT Go(X) ETRE? ET RARE H, —I2 A R e PP EY |X) Boe RA CAN BRECK, AASDRAMSH. AA BAAS ARR, MITRE FABRE RR BTE A RAP g0(X) HEL. AR PAD NRE ee. Pili, S565] A Rit Aco O75 PRA (affine functions) MMS. HM 2.3 [ HBR (Affine Functions)]: i X = (1,Xi,..., Xe)’, B= (Bo; Biy-++1Bk)'o MAH BRR LA k A= {osm oR | 9(X) = Bo + ¥> BjX;,B; € a jar = {a2 RR | ox = x9} KE, HHKOE SB OAARH, HFRKBR, BRBXA C4, Pa AE REX PHM PB MREBK, (iat BBR A BEER o(X) = X'B, BN g(X) MIRE X Al 6 REE PRB. SORE H HR A, 1H g(X) = X'B TRBAMBR 6 MAEM, (A TYREE X MARE. HM, RR = 1, CBR A ATLL FURY G(X) = Bo + Xi + BoXF A 9X) = Bo + Ain X,30 So RIE CERES, g( X) ARSENE X, AVESR PREM, LYRE 6 RUE. OE Sth, I X MW X = (1,X1, X?) MX = (1,In Xi), WA BHAT LIME. SORTA REY RET OER 6 Aa tR K, LAK ELMAR AR RM. BONAR RE EAN. AH HY NHL, SPAR HELLA 9,(X) RSM 6 MIEREEBME. Logistic UBL ARE MERE, HO GX) = 1 1+ exp(—X’B) Ae HE JE F/R At A LAT A, SRB) GE SHAE. SiH g(X) HUBBER AAAS ELF HR BCS ARAB, BR RSME B* UUROHA EM o(X) = X'S HY MSE Bee i 2 imi _ xg)? min B[Y — 9(X)} = min, BV X'By SRULAE g" (X) = X'B* WAY NRRRER-)— RAW (best linear least squares predictor), Mi RRB SUH 9 Hk BRAK) — RUA RH (best least squares approximation coefficient vector). 2H 2.5 [SRE RAM (Best Linear Least Squares Predic- tion)]: ik E(Y?) < 00, B (k +1) x (k+1) #M E(XX') RASH, H FRM AR ji —@fX)? = mi _ xg)? min BLY — 9(X)? = min, BW — x'Ay GOFF, PPR MA 9° (X) = X'" HP RAM OED B= [E(XX'1E(XY) TER BERS) BY -o(X)2 2 mi _ x'g)2 mip EV ~ o(X)]* = min, B(Y — X'B) HRA ag(y — X’p)? =0 de p=8"SH HERI 3 SRA ABO =X0F _ 5 [ex = x) = 2 [a0 - x'p) | =-26 [ow - xp) XD | = -2B[X(Y — X’A)]. BRE, BT EIX(Y —x'e"]=0 x E(XY) = E(XX’)8" ER E(XX') EE, A = (EXXYE(XY) AREA: Hessian AE @PE(Y — X'B)? dpdg’ oid E(XX') RARAFH, Hessian HPRIER (AttA?). Alt, 6* ERA bn fEHE 2.5 FH, REE (kh +1) x (k +1) RE = 28(XX') 1 E(X1) (Xa) +++ E(Xx) E(X:) E(X?) — E(XiX2) ++» E(X:X;) E(XX') = | E(X2) E(X2X:) E(X2) -+- E(X2Xx) E(Xx) E(X¢X1) E(XuXo) +++ E(XZ) AAPM A E(XY) HE, BA, HAE BY |X) eX MBAS ER, RRR) RUA Oo RAEN. RA, BRR FLA g*(X) = X'B* HABT EY |X). TEAMADE F, SOIREE 8° HE?32 Bo RAS HABE ATA 6 UE, Ao — MALMO: # B= (By, Ary’, X = (1, Xa), Ria AMARA BL . _ cov(¥, Xi) A= aria) 5 = BWV) ~ BBX) PRL, PA RAR EIR A) — FRE AM By 45 cov(Y, X1) MIE. HAZ, Bp RMB TH AA cov(Y, X1) RBI Y Al X, ZAREK ARBRE, BE ARHERWE cov(Y, X1) TERE Y AX, ZIMA EAR. Blk, BPE [AAA | 248X407 (correlation analysis) PR ORR EA) — FS ERIN. HE, EE [5] #4) (linear regression model). EX 2.4 | SHEANBI (Linear Regression Model): #4 Y=X'B+u, BER HAY HX HRA AMA, Hp u Zea RAI Z (regression model error), 4° k = 1, #4 =H wl /34RA (bivariate linear regression model) X HA 2 KH (straight linere gression model), Rk >1, MAASAR 244! (multiple linear regression model) . ZORUMAR Y = X'6 + uw RHEE, JI 8 FTMLEE RAL, Baus AVENUE RRNA. RHEL A Mee, IO HERE Xo BUMARERE Y, u = Y — Xe SE DISER Y chBLATNE X’B BUARA ASS. VORA u BOE 6 BMRB. ARB, JERR ORE FUE PRA aR JER HEA, BUTERA AER — SMH 8°, WEE (|X) = X'B°. Be Z, SABLA AT We A AOA IES AY LI ME g,(X) = ECV |X). (LA, BME go(X) TSE X RNGRAEOG, ALATA ¥ — X'8 4+ u TPAEBDEHIAR, HX SULTAN TT RELA — ie IED "RTH 52 2.6 SAT PE AI 5 CR Hh — STZ HMIIER. E26 RETR AWUHEHRL, > Y=X'Btu HH Bt =[E(XX)]* E(XY) ARRAUR Dik RA, BY p=pWEN Settee 33 SBRSAFERRERE E(Xu) =0 TERR AGAR HEA Y = X'6 + u, HRAG u=Y — X'S, Ai B(Xu) = E(XY) - E(XX)B (1) QSEHE: WOR 6 = B*, UTESC AREAL, Bl E(Xu) = E(XY) — E(XX')8" = E(XY) ~ E(XX’)[E(XX")|"1E(XY) =0 (2) FEAtE: MR E(Xu) =0 , WARE u = ¥ ~X'8, A E(Xu) = E(XY) — E(XX')8 =0 AAW E(XX’) BRN, B= [E(XX)|7E(XY) = 6° RIL, MRERK E(Xu) = 0A, WREA ARM NSAE 6 ETB RE) FGLWARK 0. WE. EEE 2.6 MRA E(Y|X) AEX WARE HOE LARA HE , SEEBI BURLY = X'8 + u MEARE E(Xu) = 0 BARI, 4A 6 = B"H. See BAR IE Se E(Xu) = 0 SRR) AE. RE, CHERAB RRA (MEH 25) BR. RAZ, RO STRRBEB)_ RAM 8", WREATH u WIRE X TE. AAR, WR X 5 u EA, MAHA RBG LRP Bb Se Th Bre AP LERH, 4X AW AAPL TIN, EZRA HE E(Xu) = 0 BORA E(u) = 0. WU E(u) =0 Rt, W E(Xu) =cov(X,u). Auk, ERR X Mu RAKE SOT. UE u PALA ET ATA X BRE BOR AM HD. PE RANE, RH E(Xu) = 0 45 BulX) =-0ORKEARN. BH WM AA, RZA RMIT. BN: HH E(ulX) = 0 WHE E(Xu) = 0, (4 E(Xu) =O, WHA E(ulX) £0. FRAT HM TIK-A. G28 Bus (X?—-1) +e, HX Me AMABILE, AISI HRERAA N(O, 1). E(u|X) = X?-14034 Doe RITE 1h E(Xu) = E[X(X? — DJ + E(Xe) = E(X3) - E(X) + B(X)B(6) =0 SOD FEM SHRBRE S(O TE a Be OER T A? EM 2.5 | PE AIRB AVE WIZE (Correct Model Specification in Conditional Mean)): 4121/22 Y=X'B+u, BER RHE E(Y|X) HEME, WRAL) RAH BE RH, E(Y|X) = X's? AAG, PRATEEM SAE B CREM E(Y|X) # X'B Sl Site ARAL st E(Y|X) HARI (misspecified). Stk FARA ARR, B— PRE BBO — MEIN B fA. SRE ARE EAE, HER — TMNT SBA Be ME RH X'p°, HEY|X) = X'S. KAZ, ERRENREA ARM SPAN BBRR E(Y|X) = X'p?. SRA 6° MAREAIRAN “MRS” , ECA FAR BRS SE 0 _ dE(Y|X) = ax BD 6° TIERED X Ht Y AHA RACAY (expected marginal effect). Hin, 4 Y AWK, X AWA, SH 6° Bid kay Be] (MPC), Be N— ALA Wt, FET HO BH Be be. URAL Bo WE? BERT Bo BIE Xo = 1 HRM. Xo = 1 SHEA HLM RE RADHA. RM Oo HAROLIMK X HBMLHY HOF SOMA”. Het, AFB LAF RHE AY HE AA HE ke OAL (capital asset pricing model, CAPM) B(Y|X) = 68 + BXSAUNT APRS HEBER 35 HY RNAS (MAAK R SEM BARZ A), XL E WAS ARMA AR (UTI SSR SG AIS 2), HR, AE SH 69 RAIA. 4 CAPM EM, 69 = 0. RRB: REMY 68 > 0, RU CAPM Atti T EVE Aas, IW FC LAER AK ECW |X) Ke F CAPM BMI 6X1. SUR 6g <0, JA CAMP tir TB yeH Aiba. SALOME B(V|X) MAGEE X RHE. Buse, 4 BBE REE IGM IOHE. Bu, Ten RR TO at, eR} REALE ERE X, DUES, RHEL ee EME Abie, FER SME 0°, (RARE URI u = Y — x’ 8° WE FRIAR E(ulX) =0 PRR SAS ERA EN, BURR u SAS RRB) Bie =Y — B(Y|X) BAZAAR. 2.7 PRABAIIRA Y=X'Btu Rat tH E(Y|X) HEAR, (1) BEAR 6° Fo— A MUR E €, HY = X'B? +e, HP E(e|X) =0; (2) Bt =. GEAR (1) SURRY BY |X) APE Bee EME, MARIE M 2.5, CEE AS SUE 6°, A E(Y|X) = X'B". ATW, BABE 2.4, BATISLY = B(Y|X)+e, Jt E(e|X) = 0. ¥ E(Y|X) = XB? RABIABSA, RBG (1). (2) H (1), E(Xe) = Bl(Xe)|X] = B[XE(e|X)] = E(X-0) =0 FAI, TAL Y = XB +c, ERE. HEH 2.6, AB = f°. EHD 2.765% (1) HH: WOR E(Y |X) RE EM, WUE TSA 8°, i E(Y|X) = X'B°. HERA F, RRM u = Y XB36 Bo RAIS HB WR B° MEET ICSE HY RBI © = Y — E(Y|X) . Hive (2) HH, WRAAE ARH Re ER, WA EY|X) SRRRERD— RAM g(X) = X'S BAM, AN, RRAEBDORTURK o STALSR 8°. PIA, 4 ( A) RAMA Re EY, B* ATURE X HY ERE NY. UR E(Y |X) AYE Y = XB + u AVIEZEAR HE E(Xu) = 0 RAL, AEM ERR, SUBS A? RAVE, WR Ee B(Xu) = 0 we 3L, (A E(u|X) A 0, KM SMB a? SY, BOSE RB E(Y|X) = X'B + E(ulX) # X'B BART u PHERES X ARM ASHAR, TUE X MY MR (HUF BE 7), (2A RE A kT 2 WA MSE HEM, —SREE BAS ESE BP TL, BE TE WY A, WAweH MSE Be). Bl 2.9 ARUP AGRA Bit (DGP) yaitix,+toe-1 alt 5% t+ Or) +e Se X, Me MAHAN N (0,1) BADER. (1) RAP BV |X1), WRX, EY AONE BR PLACATE al dE(¥ |X) ax; Y=f+AXitu =X'Bt+u PEAY LAR Ae RAHA, SEP X = (1, Xi)’ (2) RARE) — FAL 6" AUR HER) — FR BLL g"(X) = X'B (3) @u=Y —X'p", WiE B(Xu) =0. (4) OBOE REX, AYLI MARA BAN A A AE(Y|X1) _ 9 5, mae SOY — a,SADT APRESS 37 $R (1) AA X, Mle BIW, A L(e|X1) = Ele) =0. At BW IM) =14 5% + LF —-1) de(¥|%) 1,1 ax, ata BB, XY A X, RRR, RE R. (2) REERB DO RALWAR, A BY = [E(XX")* E(XY) -[; Ey o1 3 H =)1 2 go (X) = XB =1+ 5x RRR ANS FAME 9° (X) ROAR EY |X) AR ERRS, TT RT KH. (3) ARAB SE CM (2) BSE, u=Y-x's =Y — (65 + BX) = qt ~Ite AMD, ESCA APSE RIL, BD Fat. E(Xu) =E 1-G(XP 1) +6) (°] “Lo x dOd-) +e) fi : Bilal) = 2(x¢-1) 40 (4) 7%, ae(¥1X1) _ 1 1 wt ax, at gM ARH38 Boe — RISE XM Y OAS BBE X,, RES OF RRS Ot PPE A TW RIFE RE FRESH. HF RR RACKS BN, SH BERRI RHE. ERE SREP, H—-ARRTIEAM EY |X) HTEMSRCRN BRS, WE UMEAS RRA FESS KN RRR. KAPKA ESRI (nonparametric approach), EHEAA & fF F AEE BE AHR MN FIRS AIEB ROTA IEA BAB Aa T BAMA S RR DEAS AR, BRA ESE go(a) AB IPR 3X (smooth function). FY LAHI—“MARMEIEZE “3E” PRASK {),(z)}520 FEN BK go (c) BATRRET. Hk, RAE (Xu) = MT x € Support(X), go(x) = Lowe Ep (GM RK (Fourier coefficient) B= [ Go(x)p;(x)dz FBR go(a) TERA yj (7) LAY; IER LRM ITEZEICA, BD 1, i=j fr wterstene =a ={ § mt 3X 6 HH Kronecker delta BA. BI210 BBR g(x) =2? KH ze [n,n]. M ola) = rb ona) — 205) 4 Be) _ = x - eons 2 BY 2.11 BB -i1, -n Oo MIB 7G BP RAAM, OHM RK 6, BAR. RRA DAFT BL 2 Si wk AT Ay GEE DL Gol2) = D> Bis(2) 0 SUH p SETAE HE A HEE. BLD g(x) HY goa) INEZ B,(z) = go(2) — gp(a) = ¥ Bv@ i=ptt DIR p RBK, UMMA Bee. TSA, ABC (8,} Ra. Bl he, a BEA BE re HUBEBLAR AS (Yi, Xe} PALAU ee a BE Frit. ABE Pn Yi = DO Bi(X) tum, t= 1 an = SUP py = p(n) SERRA LAI. TRIES n 00 Bh, EAD B,, (2) QTE, Bit p, 00. HLA, RAIL p, MARA n MMA, AI HFAS HAMS, HERS REAK, BER, Yn — 00