tailieunhanh - Quantitative Techniques for Competition and Antitrust Analysis by Peter Davis and Eliana Garcés_14

Tham khảo tài liệu 'quantitative techniques for competition and antitrust analysis by peter davis and eliana garcés_14', kinh doanh - tiếp thị, quản trị kinh doanh phục vụ nhu cầu học tập, nghiên cứu và làm việc hiệu quả | 478 9. Demand System Estimation and cross-price elasticities. In practice therefore MNL models are quite good for learning atom tire churucleristawhich tend to be associated with high or 10 levels of market shares but we recommend strongly against using MNL models in situations where we must learn aboutsubstitutionpatterns for merger simulation . There have been a number of respoogos totìie prreiemtihu titetatureharsdentified with MNL and we explore somnofthore thtponsei indihrnstotthissncSson. Tte important lesson from MNL and the pnohurtnhf mdpphndeecnof iiielevagiolterph-tives IIA is not that the MNL ill hopelecsmodolfhouga -hattspropnbly truei but rather that we can use the IIA ỊhopCTty toounahdontuhe sincnthhMNL mnoer unreasonable predictions about what win happoato ehidprfollnwinn natty if we observe what happens to merketsCaret Soiiowing tntty wewillOoahto truth data to reject MNL models and iCen-tfhparrmetr-rto rlaherhiscceteohoice mndule. Furthermore the literature has esownOsom hfNLmgdols dn Ah ot its tootrore most simply explained in that 00 6x0 1-6x 6 101 the 110x0 0001101 16 10 the introduction of unobserved predoctcharacteristioutn tie goroeal ol ilionxi I models but we shall see later thal ihh foo-htehhniquesUhreaaSuoino mofelswitf unobserved product characteristicocanfoushdingotuigoerdiscoote choine mtwts. Introducing UnobservefPhodcet AhaaheSsrihtise h- MNL Models A famous possibly true marketmc . is toad Ueiiiitr it cupholder experienced dramatically highsales-duatomers thought novel idea. Economists working with data from the time period however probably would not have had a variable ic thoirdataoetcalled cufholde ti wouldhave been a product characteristic drielohosdhu differeotihtingtf e pi would be observed by customers but sloebseuhedboouf itnalnoLSucn rsttuettuomurt be common. As a result Bultez andNanrtteh75t Nekuntcht ania-opet t et Berry 1994 and Berry et al. C9h hove eachrhueed-hr we .