Analyzing the Relationship Between Organic and Sponsored Search Advertising: Positive, Negative or Zero Interdependence? 1

Size: px
Start display at page:

Download "Analyzing the Relationship Between Organic and Sponsored Search Advertising: Positive, Negative or Zero Interdependence? 1"

Transcription

1 Analyzng the Relatonshp Between Organc and Sponsored Search Advertsng: Posve, Negatve or Zero Interdependence? Sha Yang Stern School of Busness New York Unversy Anndya Ghose Stern School of Busness New York Unversy Abstract The phenomenon of pad search advertsng has now become the most predomnant form of onlne advertsng n the marketng world. However, we have ltle understandng of the mpact of search engne advertsng on consumers responses n the presence of organc lstngs of the same frms. In ths paper, we model and estmate the nter-relatonshp between organc search lstngs and pad search advertsements usng a unque panel dataset based on aggregate consumer response to several hundred keywords over three months collected from a major natonwde retaler store chan that advertses on Google. In partcular, we focus on understandng whether the presence of organc lstngs on a search engne s assocated wh a posve, negatve or no effect on the clck-through rates of pad search advertsements, and vce-versa for a gven frm. We frst buld an ntegrated model to estmate the relatonshp between dfferent metrcs such as search volume, clck-through rates, converson rates, costper-clck and keyword ranks. A Herarchcal Bayesan modelng framework s used and the model s estmated usng Markov Chan Monte Carlo (MCMC) methods. Our emprcal fndngs suggest that clckthroughs on organc lstngs have a posve nterdependence wh clck-throughs on pad lstngs, and vce-versa. We also fnd that ths posve nterdependence s asymmetrc such that the mpact of organc clcks on ncreases n utly from pad clcks s 3.5 tmes stronger than vce-versa. Usng counterfactual experments, we show that on an average ths posve nterdependence leads to an ncrease n expected profs for the frm rangng from 4. % to 6.5 % when compared to profs n the absence of ths nterdependence. To further valdate our emprcal results, we also conduct and present the results from a controlled feld experment. Ths experment shows that total clck-through rates, conversons rates and revenues n the presence of both pad and organc search lstngs are sgnfcantly hgher than those n the absence of pad search advertsements. The results predcted by the econometrc model are also corroborated n ths feld experment, whch suggests a causal nterpretaton to the posve nterdependence between pad and organc search lstngs. Gven the ncreased spendng on search engne based advertsng, our analyss provdes crcal nsghts to managers n both tradonal and Internet frms. Keywords: Pad search advertsng, Organc search lstngs, Search engnes, Clck-through rates, Converson rates, Electronc commerce, Internet markets, Monetzaton of user - generated content, Herarchcal Bayesan modelng. The authors are lsted n reverse alphabetcal order and contrbuted equally. They are grateful to the assocate edor, and two anonymous referees for extremely helpful comments. The authors also thank Susan Athey, Mchael Baye, Erc Bradlow, Erk Brynjolfsson, Mark Schankerman, and semnar partcpants at Carnege Mellon Unversy, Columba Unversy, McGll Unversy, Federal Trade Commsson (FTC), New York Unversy, Purdue Unversy, Unversy of Calgary, Unversy of Connectcut, Unversy of Calforna at Irvne, Unversy of Goethe-Frankfurt, Unversy of Pennsylvana, Unversy of Washngton, Mcrosoft Research, the 008 Internatonal I.O. Conference, the 008 Marketng Scence Instute conference, the 008 NET Instute Conference, the 008 Internatonal Symposum on Informaton Systems (ISIS), the 008 Workshop on Informaton Technology and Systems (WITS), the 008 Workshop on Informaton Systems Economcs (WISE) and the 009 Toulouse Conference on The Economcs of the Internet and Software for useful suggestons. Anndya Ghose acknowledges the generous fnancal support from NSF CAREER Award IIS The usual dsclamer apples.

2 . Introducton Over the past few years, search engnes lke Google, Yahoo and MSN have dscovered that as ntermedares between consumers and frms, they are n a unque poson to sell new forms of advertsements. Ths has led to the prolferaton of sponsored search advertsng where advertsers pay a fee to Internet search engnes to be dsplayed alongsde non-sponsored or organc web search results. Because search engne based advertsng s drectly related to users search queres, s consdered by users as beng far less ntrusve relatve to other forms of onlne advertsng. From the frm sde, ths knd of advertsng leads to more qualfed prospects snce the ads are dsplayed n response to user-orgnated search behavor. From the consumer perspectve, ths s a much more relevant form of advertsng snce the keywords and the ad message are typcally matched wh user-generated queres. These features have lead to a wde-spread adopton of ths form of Web.0 meda by frms, wh the global pad search market expected to reach almost $0 bllon by the end of 009. How does ths mechansm work? In sponsored search, frms who wsh to advertse ther product or servces on the Internet create text-based ads and subm that nformaton n the form of keyword lstngs to search engnes. A keyword s a combnaton of words or terms that best descrbes the product, brand or retaler beng advertsed. Bd values are assgned to each ndvdual keyword and then search engnes p advertsers aganst each other n second-prce aucton-style bddng for the hghest posons on search engne result pages. When users search for that keyword on a search engne, the relevant ad along wh the advertsers web page appears as a sponsored lnk on the top and the rght sde of the organc search results. When users clck on the sponsored ad, they are taken to the advertser s webse. An mportant determnant of the effectveness of sponsored search advertsng for a gven advertser s the lkelhood of the same advertser appearng n the natural or organc lstngs of the search engne, and s poson on the organc lstngs for a gven keyword. Organc rankngs of advertsers webses are based on a complex and propretary ndexng algorhm devsed by the search engne nvolvng the qualy of the webse and the webse's relatve mportance wh respect to other lnks. Thus, consumers often face two competng lst of results that may both be relevant to ther search query: () the sponsored search lstng and () the organc search lstng. Advertsers have been grapplng wh the trade-offs n each of these two forms of referrals. On the one hand, because a frm can control the message of pad search ads, one would expect hgher conversons from them. On the other sde, because people value the perceved 'edoral ntegry' of organc lstngs, one would expect hgher conversons from them. Some anecdotal evdence suggests that there s a potental dsconnect between the percepton of sponsored lstngs by busness and users, wh Pad Search Ad Spend Wll H $0 Bllon By 009, 03/07/06,

3 consumers havng a posve bas towards organc search lstngs. For example, Hotchkss et al. (005) fnd that more than 77% of partcpants favored organc lnks more than the sponsored lnks, as offerng sources of trusted, unbased nformaton. A smlar set of fndngs are reported n Greenspan (004). On the other hand, Jansen (007) fnds that sponsored lnks are more relevant than organc lnks n the context of ecommerce search queres, and ths fndng s robust to usage across Google, Yahoo and MSN s search engnes. Moreover, there s also some anecdotal evdence that suggests that pad search may lead to hgher conversons than organc search. 3 Ths study looked at search engne vss on Google, Yahoo and MSN and showed a medan order converson rate of 3.4 % for pad search compared to a converson rate of 3.3 % for organc search results durng the same tmeframe. These mxed fndngs then motvate the queston that to what extent should frms nvest n sponsored search advertsements when they also appear n the organc lstngs for a gven search query n that search engne. After all, frms ncur costs whle engagng n pad search advertsng. Despe the growth of search advertsng, we have ltle understandng of the mpact of sponsored search advertsng on consumers overall reacton n the presence of organc search lstngs for a gven frm. Our key objectve n ths paper s to compare and analyze the nterdependence between pad ad lstngs wh organc lstngs for a gven advertser. Hence, we focus on the followng questons: How do dfferent keyword-level attrbutes mpact performance metrcs such as average clck-through rates and converson rates n pad search advertsng as compared wh those n organc search lstngs? What s the nature of the relatonshp between average clck-through rates on organc and pad lstngs for a gven set of keywords? Are the combned clck-through rates, converson rates and revenues from sponsored and organc lstngs hgher or lower than those from organc lstngs only (.e., when pad search advertsng s paused)? Whle an emergng stream of theoretcal lerature n sponsored search has looked at ssues such as mechansm desgn n auctons, no pror work has emprcally analyzed these knds of questons. By modelng the assocaton between pad and organc clcks and vce-versa, we am to examne f there s a posve, negatve or zero assocaton between them. Towards ths goal, we use a unque panel dataset of aggregate consumer response to several hundred keywords over three months collected from a large natonwde retaler that advertses on Google. To be clear, we only have aggregate keyword-level data, not dsaggregate user-level data. We propose a Herarchcal Bayesan modelng framework n whch we model consumers aggregate responses jontly wh the advertser s and search engne s decsons. Our paper s the frst academc study that estmates the effect of sponsored search advertsng on consumer search, clck and converson behavor n the presence of organc lstngs of the same frm for the same set of keywords at the same tme. We am to make the followng four contrbutons: 3 3

4 Frst, we buld an ntegratve smultaneous equatons-based model to emprcally estmate the mpact of ad rank and other keyword-level attrbutes on consumer aggregate responses, and advertser and search engne decsons n the presence of both pad and organc search lstngs. Specfcally, we examne the relatonshp of these attrbutes wh average search, clck-through and purchase propenses as well as wh the advertser s cost-per-clck (CPC) and the search engne s decson wh respect to allocatng keyword ranks n accordance wh nstutonal practces. We fnd that on average the presence of retalerspecfc nformaton and brand-specfc nformaton s a sgnfcant predctor of converson rates n pad and organc lstngs, respectvely. A smultaneous equatons model as proposed n our study makes possble for us to descrbe current phenomena and prescrbe some recommendatons to advertsers. It also provdes two key benefs. On one hand, we are able to account for the potental endogeney of keyword rank and CPC. On the other hand, ths allows us to do some smple polcy smulatons to nfer the optmal CPC for dfferent keywords, and thereby mpute the mpact of sponsorng ads n the presence of organc lstngs on frms profs. Second, we nvestgate the value to frms from partcpatng n such sponsored search advertsng by examnng the nature of nterdependence between organc and pad lstngs usng an aggregate keyword-level dataset of clcks and conversons on pad and organc lnks. We also explore the asymmetrc nature of the relatonshp between these two forms of search engne lstngs. Based on our model, we fnd that average clck-throughs on organc search lstngs have a posve nterdependence wh average clck-throughs on pad lstngs, and vce-versa. Ths posve nterdependence s also asymmetrc such that the mpact of organc clcks on ncrease n utly from pad clcks for the same frm s 3.5 tmes stronger than vce-versa. We provde detals of alternatve econometrc models that we buld and estmate to explore ths nterdependence. All models provde qualatvely smlar results. Thrd, we conduct some counterfactual experments usng polcy smulatons to hghlght the magnude of the posve nterdependence between these two forms of advertsng. We fnd that on an average, ths posve nterdependence leads to an ncrease n expected profs for the frm rangng from 4. % to 6.5% when compared to profs n the absence of eher of these. Further, the posve nterdependence s the strongest n the case of the least competve keywords (such as retaler-specfc keywords) and weakest n the case of the most competve keywords (such as brand-specfc and generc keywords). Therefore, the proposed parsmonous modelng framework can help advertsers make optmal decsons and nvestgate the value from partcpatng n such sponsored search advertsng n the presence of natural or organc lstngs n search engnes. Fnally, we descrbe a smple feld experment that sheds further lght on the predctons from our emprcal model and analyss. It shows that gven a set of keywords, combned clck-through rates, converson rates and total revenues accrung to the frm n the presence of both pad and organc lstngs s 4

5 hgher than those n the absence of pad search advertsements. We fnd that although the presence of pad search advertsements takes some traffc away from organc lstngs for some keywords, for a vast majory of keywords n our sample, the average clck-through rate of organc lstngs when pad search was on was hgher than the average clck-through rate of organc lstngs when pad search was nactve. Ths result shows that the posve nterdependence effect predcted by the econometrc model s also corroborated n controlled experments and potentally suggests a causal nterpretaton to the results obtaned from the model. To evaluate the consstency of results between the feld experment and the estmated model, we conduct analyss usng the specfc sample of keywords for the duraton for whch the experment was run wh pad lnks. The results from these analyses are consstent wh those from the ntegrated model, and show that on an average, pad clck-through rates are posvely assocated wh organc clck-through rates, and vce-versa. Further, the posve nterdependence s also asymmetrc n nature. These results reman qualatvely the same even when we estmate addonal model specfcatons such as the autologstc model and the smultaneous move game structural model. The remander of ths paper s organzed as follows. Secton gves an overvew of the dfferent streams of lerature related to our paper. Secton 3 presents a smultaneous model of consumer search, clck, and purchase, the advertser s keyword prcng decson, and the search engne s decson of assgnng keyword ranks. Secton 4 presents an emprcal applcaton of the proposed model along wh a descrpton of the data that s used. We also present a dscusson of a number of robustness checks we have conducted usng alternatve model specfcatons such as an autologstc model and a smultaneous move game model. The detals are n the Onlne Appendx. We descrbe the counterfactual experments conducted through some smple polcy smulatons n Secton 5. Secton 6 presents a controlled feld experment that also analyzes the effect of organc search lstngs on pad search performance. Secton 7 presents some manageral mplcatons and concludes the paper wh a dscusson of some lmatons.. Lerature and Theoretcal Foundaton Our paper s related to several streams of research. Frst, relates to recent research n onlne advertsng. A number of approaches have modeled the effects of advertsng based on aggregate data (Tells 004). However, much of the exstng academc research (e.g., Gallagher et al. 00, Dreze and Hussherr 003) on advertsng n the onlne world has focused on measurng changes n brand awareness, brand attudes, and purchase ntentons as a functon of exposure. Other studes measure (ndvdual) exposure to advertsng va aggregate advertsng dollars (e.g., Mela et al. 998, Ilfeld and Wner 00). Because banner ads have been perceved by many consumers as beng annoyng, tradonally they have had a negatve connotaton assocated wh. Moreover, was argued that snce there s consderable 5

6 evdence that only a small proporton of vss translate nto fnal purchase (Moe and Fader 003, Chatterjee et al. 003), clck-through rates may be too mprecse for measurng the effectveness of banners served to the mass market. Interestngly however, Manchanda et al. (006) found that banner advertsng actually ncreases purchasng behavor, n contrast to conventonal wsdom. These studes therefore hghlght the mportance of nvestgatng the mpact of other knds of onlne advertsng such as search keyword advertsng on actual purchase behavor, snce the success of keyword advertsng s also based on consumer clck-through rates. Our study s also related to other forms of pad placements avalable to retalers on the nternet such as sponsored lstngs on shoppng bots. For example, Baye and Morgan (00), Montgomery et al. (004), Baye et al. (009) have studed the role of shoppng bots as nformaton gate keepers and estmated the mpact of retalers rank durng placement on clck-through rates. From a theoretcal perspectve, our paper has nterestng parallels to the tradonal product placement based advertsng next to edoral content n the mass meda. Shapro et al. (997) report a study of magazne advertsng n whch partcpants were requred to read a magazne artcle delvered on a computer screen. The artcle was flanked by target ads desgned to receve mnmal reader attenton. They found that users were more lkely to nclude the product featured n the ad n ther consderaton set compared to the control partcpants who had not vewed the ads. Geeradyn and Fauconner (000) dscuss the advent of advertorals whch are prnted advertsng message but have the look and content of an ordnary newspaper or magazne artcle. In creatng the ad, the edoral form of the medum and artcle content next to whch the ad s to be placed, s taken nto account. Research n consumer behavor has theorzed about these effects. For example, the truth effect (the ncreased belef n an ad clam due to a prevous exposure, Hawkns and Hoch, 99; Law, and Braun, 000), and the mere exposure effect (the formulaton of a posve affect from exposure to a bref stmulus, Janszewsk 993) both llustrate changes n consumer behavor followng a sngle exposure to a stmul whout awareness of the pror exposure. There s also an emergng theoretcal stream of lerature exemplfed by Edelman et al. (007), Feng et al. (007), Varan (007), Lu et al. (009) who study mechansm desgn n sponsored keyword auctons. Athey and Ellson (008) buld a model that ntegrates consumer behavor wh advertser decsons. Wlbur and Zhu (008) examne the ncentves of search engnes to prevent clck-fraud. Katona and Sarvary (007) buld a model of competon n sponsored search and fnd that the nteracton between search lstngs and pad lnks determne equlbrum bddng behavor. Xu et al. (009) fnd that whle organc lstng may hurt search engne revenue, could nduce hgher socal welfare and sales dversy. Despe the emergng theory work, very ltle emprcal work exsts n onlne sponsored search advertsng. Exstng work has so far focused on search engne performance (Bradlow and Schmtlen 6

7 000, Telang et al. 004) and examned ssues related to adverse selecton (Anmesh et al. 009) and prcng dfferences based on ad context (Goldfarb and Tucker 007). Our paper s closely related to an emergng stream of work that uses frm-level data from search engne advertsers. Rutz and Buckln (007) study converson probably, for hotel marketng keywords n Los Angeles. Rutz and Buckln (008) show that there are spllovers between search advertsng on branded and generc keywords, as some customers may start wh a generc search to gather nformaton, but later use a branded search to complete ther transacton. Ghose and Yang (008a) compare pad search advertsng to organc lstngs wh respect to predctng order values and profs. Ghose and Yang (009) quantfy the mpact of keyword attrbutes on consumer search and purchase behavor as well as on advertser s cost-per-clck and the search engne s rankng decson for dfferent ads. They fnd that keyword rank nfluences not only ad clck-through rates but also the fnal converson rates from the advertser's webse, thus mplyng that the value per clck s not unform across slots on the search engne's results page. They also show that keywords that have more promnent posons on the search engne results page, and thus experence hgher clck-through or converson rates, are not necessarly the most profable ones profs are often hgher at the mddle posons than at the top or the bottom ones. Agarwal et al. (008) provde quantatve nsghts nto the profably of advertsements assocated wh dfferences n keyword poson. Ghose and Yang (008b) buld a model to map consumers searchpurchase relatonshp n sponsored search advertsng. They provde evdence of horzontal spllover effects from search advertsng resultng n purchases across other product categores. Song and Mela (009) buld a dynamc structural model to explore how the nteracton of consumers, search engnes and advertsers affects consumer welfare and frm profs. Gerstmeer et al. (009) dscuss some nterestng bddng heurstcs and hghlght whch of these leads to hgher profs for the advertser. Jerath et al (009) fnd that a superor frm may obtan a poson below the nferor frm, but stll obtan more clcks than the nferor frm. Thus, may not always want to be n the topmost poson, a fndng consstent wh the rank-profably relatonshp n Ghose and Yang (009). 3. An Integratve Model of Consumer Response and Frms Decsons We next present a model that ntegrates consumer search and purchase behavor wh frms decson-makng behavor such as prce of a keyword and the rank of a keyword ad after the aucton. Ths model consders the smultaneous presence of both pad and organc search lstngs. A Herarchcal Bayesan modelng framework s used and the models are estmated usng Markov Chan Monte Carlo (MCMC) methods (Ross and Allenby 003). 7

8 Let us denote N as the total number of searches for keyword n week t. We model the total number of searches over tme as a log normal regresson specfed as follows: 4 ln( N ) Tme (a) Retaler Brand 3Length 4 ~ N(, ) (b) The covarates, Brand, Retaler and Length, n the above equaton are descrbed below. Pror work (Broder 00) has analyzed the goals for users web searches and classfed user queres n search engnes nto three categores of searches: navgatonal (for example, a search query consstng of a specfc frm or retaler), transactonal (for example, a search query consstng of a specfc product) or nformatonal (for example, a search query consstng of longer words). Search engnes not only sell nonbranded or generc keywords as advertsements, but also well-known product or manufacturer brand names as well as keywords ndcatng the specfc advertser n order for the frm to attract consumers to s webse. 5 Moreover, advertsers also have the opton of makng the keyword generc or specfc by alterng the number of words contaned n the keyword. Hence, we focus on the three mportant keywordspecfc characterstcs for a frm (the advertser) when advertses on a search engne. Ths ncludes whether the keyword has () retaler-specfc nformaton (for example, Wal-Mart, walmart.com ) or not, () brand-specfc nformaton (for example, Sealy mattress ) or not, () and the length (n words) of the keyword (that determnes how narrow or broad the consumer search s). Based on these factors, we construct three keyword-specfc characterstcs denoted by Brand, Retaler and Length. The frst two varables are coded as dummy varables. The ndustry dynamc effects are controlled for by addng a tme trend n all equatons. 6 Ths s consstent wh pror work n ths area (for example, Ghose and Yang 009). For a gven keyword, whle some searches do not lead to any clcks at all, some searches lead to clcks on eher organc or pad lstngs, and some searches lead to clcks on both lstngs. Ths dual-clck search behavor s also shown by Jansen et al. (007), and hence s mportant to ncorporate such dualclck search behavor n the model and analyss. Let same keyword only n the organc lstng,,0 N denote the number of clck-throughs on the 0, N denote the number of clck-throughs on the same keyword 4 We use a log normal regresson on the total search volume because of the exstence of outlers, whch do not make the normal dstrbuton a good one to apply. 5 For example, a consumer seekng to purchase a dgal camera s as lkely to search for a popular manufacturer brand name such as NIKON, CANON or KODAK on a search engne as searchng for the generc phrase dgal camera. Smlarly, the same consumer may also search for a retaler such as BEST BUY or CIRCUIT CITY n order to buy the dgal camera drectly from the retaler. 6 In addon to addng a tme trend, we have also explored the specfcaton of correlated error terms over tme. However, the autoregressve coeffcents dd not come out sgnfcant for any of the seven equatons. Therefore, we present the results wh tme trend effect. 8

9 only n the pad lstng, and N, denote the number of clck-throughs on the same keyword n both the 0,0,0 0,, organc and pad lstngs. The remanng N N N N N searches lead to zero clck-throughs. Based on ths, the lkelhood functon s multnomal:,0 0,, 0,0,0 0,,,0 N 0, N, N 0,0 p p p p f ( N, N, N, N ) () Let us further assume the probably of a clck-through takes the log form exp( )/[ exp( )] for organc lstngs and exp( )/[ exp( )] for pad lstngs. Here denotes the latent utly of an average clck on organc lstngs, and denotes the latent utly of an average clck on pad lstngs. Then by assumng the condonal ndependence of a clck through on organc lstngs and a clck through on pad lstngs, we can wre down the four probables specfed n equaton () as follows: p p p p exp( ),0 p(, ) d d (3a) exp( ) exp( ) exp( ) 0, p(, ) d d (3b) exp( ) exp( ) exp( ) exp( ), p(, ) d d (3c) exp( ) exp( ) 0,0 p(, ) d d (3d) exp( ) exp( ) We further model the nterdependent relatonshp of the two latent utles assocated wh the two types of clck through for the same keyword as follows: (4a) (4b) The above two equatons ncorporate the noton that the latent utly of an organc search and a pad search s dependent on both s ntrnsc utly ( ) and the extrnsc utly from each other ( and ). Wh respect to the ntrnsc utly, we model to be dependent on the knd of keyword that s dsplayed n response to a search query. Ths s modeled as: s s s Ranks s Retaler sbrand s3length s4tme Wh respect to the extrnsc utly, N 0,0 and ndcate the effect that maps the, s, (5) nterdependence between pad lstngs and organc lstngs for keyword. A posve sgn on and suggest a posve nterdependency or complementary relatonshp between the clck-throughs va organc and pad lstngs. That s, the clck-through on the organc (pad) lstng tends to ncrease the utly of a 9

10 clck-through on the pad (organc) lstng. Smlarly, a negatve sgn on and suggest a negatve nterdependency or substutve relatonshp between the clck-throughs va organc and pad lstngs. Fnally, a zero value of and suggest ndependence between the clck-through va the organc and pad lstngs. Fnally, we model the unobserved heterogeney across keywords as: [,, ]'~ MVN(, ) (6a) [,, ]'~ MVN(, ) (6b) So far, we have modeled the clck-through rates. Next, we model the converson behavor condonal on the clck-through. Denote from organc searches, searches, and q ( q M as the total number of conversons for keyword n week t M as the total number of conversons for keyword n week t from pad ) as the converson probably for organc (pad) searches condonal on a clckthrough. Then assumng the conversons are ndependent events, we can wre down the lkelhood of M and f ( M f ( M N N M as follows:,0,0, N, N 0, 0,, N, N,,, N, N 0,0 0,0 M q q,0, N N M ) (7a) M q q 0,, N N M ) (7b) Pror work (Brooks 005) has shown that there s an ntrnsc trust value assocated wh the rank of a lstng on a search engne, whch leads to the converson rate droppng sgnfcantly wh an ncrease n the rank (.e., wh a lower poson on the screen). Another factor that can nfluence converson rates s the qualy of the landng page of the advertser s webse. Anecdotal evdence suggests that f onlne consumers use a search engne to drect them to a product but do not see addressed adequately on the landng page, they are lkely to abandon that se. Dfferent keywords lead to dfferent knds of landng pages. In keepng wh the nstutonal practces of search engnes, we use the clck-through rates (CTR) (standardzed n our emprcal analyss) to control for the landng page qualy score 7, where clck-through rate s defned as the number of clcks over the number of searches. Furthermore, dfferent keywords are assocated wh dfferent products. It s possble that some product-specfc characterstcs nfluence consumer converson rates, and thus, s mportant to control for the unobserved product characterstcs that may nfluence converson rates once the consumer s on the webse of the advertser. Hence, we 7 Google computes a qualy score for each keyword as a functon of the relevancy, transparency, and navgably of nformaton on the landng page and the past clck-through rate of that keyword (Ghose and Yang 009). The key dea s to provde a hgher user experence after a clck-through to the advertser se from ther search engne. However, snce we do not know have nformaton on the landng page qualy scores, we use the clck-through rate as the proxy for landng page qualy scores. Further nformaton on ths metrc s avalable at 0

11 nclude the three keyword characterstcs to proxy for the unobserved keyword heterogeney stemmng from the dfferent products sold by the advertser. These factors lead us to model the converson probables on organc lstngs (s=) and on pad lstngs (s=) as follows: q s exp( cs cs Ranks sctr s sretaler s3brand s4length s5tme s ) (8a) exp( c c Rank CTR Retaler Brand Length Tme ) s s s s s s c [ c c ]'~ MVN( c s, ) (8b) s, s s Next, we model the cost-per-clck (CPC) of the ad keywords posted n the sponsored search lst. 8 Because dfferent keyword attrbutes determne whether s a generc or branded keyword, the advertser s CPC for a gven keyword wll depend on these attrbutes. The advertser decdes on bd prce by trackng the performance of a keyword over tme such that the current bd prce s dependent on past performance of that keyword. It does ths n two ways. 9 Frst, the keyword s current bd prce s a functon of the rank of the same keyword n the prevous perod, n both the pad and organc lstng. Second, the keyword s bd prce s also based on the extent of profs from conversons through pad and organc lstngs, respectvely, n the prevous perod. Here prof s defned as the revenue from advertsng net of the varable cost of the product beng sold through that keyword mnus the costs of placng that advertsement for the frm (the advertsement cost s equal to the total number of clcks tmes cost per clck). Hence, we nclude both these sets of covarates. 0 Fnally, we also control for possble competve effects of other advertsers by ncludng the maxmum bd prce for a gven keyword (denoted by Competor_Prce). Ths leads to the followng equaton: ln( CPC ) Rank, t, Rank, t, s3 ln( Prof, t, s4 ) ln( Prof, t, Retaler Brand Length Competor _ Prce Tme ~ N(, ) (9b) Fnally, we model the search engne s decson on assgnng ranks for the sponsored keyword. Search engnes lke Google, MSN and Yahoo decde on the ranks durng the keyword aucton by takng nto account both the current bd prce and the qualy score. Snce the qualy score s most affected by the pror clck-through rates, and more recent CTR s gven hgher weghtage by the search engne n computng ths score, we use the one perod lagged value of CTR (standardzed n our emprcal analyss) 4 9 s5 ) s (9a) 8 Snce we do not have data on actual bds, we use the actual cost-per-clck as a proxy for the bd prce. These two are hghly correlated. 9 Ths nformaton about current bd prces beng based on the two metrcs of past performance (lagged prof and lagged rank) was gven to us by the advertser. Our results are robust to the use of eher one of these two metrcs as well as to ther excluson also. Gerstmeer et al (009) also dscuss that current perod bd can be a functon of past profs from that keyword. 0 Our results are robust to the use of gross profs n whch we consder only the advertsement revenues and advertsement-related costs as well net profs n whch we consder the varable costs of the products.

12 as a control varable. As before, we use the three keyword attrbutes to proxy for the dfferent unobserved characterstcs of the landng page. We control for possble competve effects of other advertsers by ncludng the max bd prce for a gven keyword, Competor_Prce. Ths leads to the followng equaton for the rank of a keyword n sponsored search: ln( Rank ) ln( CPC ) CTR, t, Retaler Brand Length Competor _ Prce Tme v (0a) ~ N(, ) (0b) We allow the error terms n the fve levels of decsons to be correlated. That s, (,,,,,, )'~ MVN(0, ). () 3. Econometrc Issues and Identfcaton The specfcaton of the covarance of the error terms s mportant to help control the potental endogeney n the keyword ranks for pad searches and conversons. In order to show ths endogeney ssue and the dentfcaton of the proposed system of smultaneous equaton model, we provde a sketch of the model below. The proposed seven equatons, n essence, can be wrten as follows: Rank f CPC, X, ) (a) ( CPC f X, ) (b) ( q f Rank, X, ) (c) 3( 3 3 q f Rank, X, ) (d) 4( 4 4 f Rank, X, ) (e) 5( 5 5 f Rank, X, ) (f) 6( 6 6 N f X, ) (g) 7( 7 7 In the above smultaneous equatons system, X X 7 are the exogenous covarates assocated wh the seven equatons, respectvely. Note that X 5 ncludes one covarate that s the nstrument of, that s, ˆ. Smlarly, X 6 ncludes one covarate that s the nstrument of, that s, ˆ.,..., 7 are the error terms assocated wh the seven equatons, respectvely. These error terms are manly capturng nformaton that s observed by the decson makers (consumer, advertser, and search engne) but not There are three reasons for usng the log (rank). Frst, rank s measured as weekly average rank for a keyword and therefore s a contnuous varable rather than an nteger. Second, the log transformaton makes s dstrbuton closely mmc a normal dstrbuton and ths mgates the effect of outlers. Thrd, we have also tred an alternatve lnear model and found that the log transformed model performs slghtly better than the lnear model for both the nsample f and out-sample f.

13 observed by the researcher. The proposed system of smultaneous equatons presented n equatons (a), (b), (d) and (f) closely resembles the trangular system n standard econometrc textbooks (page 679 of Greene 999). Ths model s dentfed based on the followng argument. Frst, equaton (b) s a classcal regresson model whose parameters can be naturally dentfed. To see ths more clearly, CPC s modeled as exogenously determned (modeled as the advertser s decson and a functon of the advertser s past performance wh the same keyword n both pad and organc lstngs, and other keyword related characterstcs as specfed n equaton b). CPC, n turn, affects the search engne s rankng decson for pad ads, Rank, and fnally Rank affects both clck-through and the converson probables. Thus, equatons (b) (g) can be dentfed accordngly. In fact, f the correlaton between and each of,..., 7 s equal to zero, then we can estmate the seven equatons separately. Now, f the correlaton between and any of,..., 7 s not equal to zero, then Rank wll be endogenous and estmatng these equatons separately wll lead to nconsstent estmates. We gve a smple example below. Suppose that and have a non-zero correlaton. Then n equaton (b), Rank wll be correlated wh 6 because Rank s correlated wh and s correlated wh 6. The way to account for ths endogeney problem s to smultaneously estmate equatons (a) to (g). Snce we are not able to predct the correlaton structure n the proposed smultaneous equatons model,.e. do not know whch correlaton s zero and whch correlaton s not zero, we estmate the full covarance matrx and let the data nform us. As shown n Lahr and Schmdt (978) and dscussed n Greene (999), a trangular system of smultaneous equatons can be dentfed whout dentfcaton constrants such as nonlneary or correlaton restrcton. In partcular, the dentfcaton of such a trangular system comes from the lkelhood functon. Ths s also noted by Hausman (975) who observes that n a trangular system, the Jacoban term n the lkelhood functon vanshes so that the lkelhood functon s the same as for the usual seemngly unrelated regressons problem (Hausman 975). Hence, a GLS (generalzed least squares) or SUR (seemngly unrelated regresson) based estmaton leads to unquely dentfed estmates n a trangular system wh a full covarance on error terms as shown by Lahr and Schmdt (978). Furthermore, we dd a smulaton analyss and found that our estmaton procedure accurately recovers the true parameter values. Ths suggests that dentfcaton s not a problem. 4. An Emprcal Applcaton In ths secton, we present an emprcal applcaton of the proposed model usng a unque panel dataset of aggregate keyword-level data on clcks and conversons collected from a Fortune 500 frm that advertses on Google. We frst descrbe the data generatng process and the data used n the estmaton, 3

14 then dscuss our emprcal fndngs, and fnally conduct polcy smulatons based on the proposed model and parameter estmates. 4.. Data When an Internet user enters a search query nto a search engne, he gets back a page wh results, contanng both the organc lnks most relevant to the query and the sponsored lnks,.e., pad advertsements that are ranked sequentally by the search engne. The servng of a text ad n response to a query for a certan keyword s denoted as an mpresson. If the consumer clcks on the ad, he s led to the landng page of the advertser s webse. Ths s recorded as a clck, and advertsers pay the search engne on a per clck bass. Ths s known as the cost-per-clck. In the event that the consumer ends up purchasng a product from the advertser, ths s recorded as a converson. Our data contans weekly nformaton on pad search advertsng from a large natonwde retal chan, whch advertses on Google and s smlar to the data used n Ghose and Yang (009). The data span all keyword advertsements by the company durng a perod of three months n the frst quarter of 007, specfcally for the 3 calendar weeks from January to March 3. The data are based on an exact match between the user query and sponsored ad. Note that search engnes only provde aggregate level daly or weekly data to advertsers. The use of exact match (nstead of broad or phrase match) prevents any concern from possble aggregaton bases arsng due to the absence of data from every sngle aucton that occurred n a gven week or n a gven day. Moreover, the frm provdng us the data for ths study had confrmed that for an overwhelmng majory of the keywords n our sample, there was very ltle varaton n the number of competors for a gven keyword across the tme perod of our data. Ths feature also mnmzes the mpact of competon on the extent of varaton n the rank for a gven keyword, further allevatng any concern of bases from data aggregaton. The data conssts of keyword ads from all sx categores of products that ths natonwde chan retaler sells (beddng, bath, dnng, kchen, electrcs, and home décor). These keyword ads encompass all the 40 departments subsumed whn these sx product categores. Between them, the keywords represent several hundred unque SKUs. We have 06 unque brands represented by these keywords and for the same advertser we have several dfferent combnatons of s name, each represented by a unque keyword. The frm s a Fortune-500 frm wh a strong natonal and nternatonal presence but due to the nature of the data sharng agreement between the frm and us, we are unable to reveal the name of the frm. 4

15 Each keyword n our data has a unque advertsement ID. For any gven keyword search query, the retaler s lnked through to both organc and pad search results on the search engne. In a gven search sesson, whle the majory lead to clcks on eher organc or pad (or no clcks at all), some searches lead to clcks on both organc and pad lstngs durng the same search sesson. We obtaned the data on the total number of searches for a gven keyword n a gven week from the Google Keyword Metrcs tool. 3 The data provded had nformaton on the number of clcks on pad lstngs only, number of clcks on organc lstngs only, and number of clcks on both organc and pad lstngs for any gven keyword on a daly bass that was aggregated by the advertser to a weekly bass. The number of searches that do not lead to any clcks s calculated by subtractng the sum of the aforementoned three types of clcks from the total number of searches. Smlar to data on conversons through pad search, we have data on conversons through organc lstngs for any gven keyword n the same week. Note that these are all aggregate keyword-level data, not dsaggregate user-level data. Data on organc search rankngs for the same set of keywords was obtaned based on a web crawler that gathered nformaton on where the advertser s lnk would appear on Google s organc lstngs. The web crawler was constructed n PERL. To get a more precse estmate of the rank of the organc lstng of the advertser, we retreved ths data once every week over a sx-week perod from Google. 4 Snce there was very ltle change n the rank of the organc lstng (the standard devaton across ranks for a gven keyword was very low), we used the weekly average rank of a keyword n the organc lstng for the purpose of our analyss. Fnally, to control for competve bd prces n our estmatons, we collected data from Google s keyword prcng tool, whch gves estmates of advertsers maxmum cost per clck for any gven keyword. Google s keyword estmator tools gve two key peces of nformaton: The estmated upper and lower range for the cost per clck of that keyword (roughly correspondng to the prce of appearng ranked frst and thrd on the sponsored lnks related to that keyword). We take the average of these two values to construct the Competor_Prce varable. Then for both pad and organc lstngs, we have nformaton on the number of clcks, number of conversons, the total revenues from a converson for a gven keyword for a gven week, the average cost per clck (CPC) n pad search, the maxmum CPC for a gven keyword across competors, and the rank of the keyword. Whle a search can lead to an mpresson, and often to a clck, may not lead to an actual purchase (whch we defne as a converson). The product of CPC and number of clcks gves the total 3 Ths tool s avalable at and retreves both daly and monthly searches for a gven keyword on Google. We computed the weekly number based on the daly data. As a robustness check, we also computed the weekly numbers based on the monthly data and fnd that yelds smlar results. We used the total number of searches to compute the clck-through rates snce we do not have data on the total number of mpressons on natural lstngs for a gven keyword. 4 The web crawler performed searches wh the exact keyword that we have n our data to search for the page number and rank of the organc lnk of the advertser that s retreved by the search engne. 5

16 costs to the frm for sponsorng a partcular advertsement. We have data on the contrbuton margn of each keyword based on the sub-product category (department) that represents. Based on the contrbuton margn and the revenues from each converson through a pad search advertsement, we compute the gross prof per keyword from a pad search converson. The dfference between gross profs and keyword advertsng costs (the number of clcks tmes the cost-per-clck) gves the net profs accrung to the retaler from a sponsored keyword converson. Ths gves us the Pad_Prof varable. Smlarly, the Organc_Prof varable s computed based on the contrbuton margn of the keyword and the revenues from each converson through an organc search lstng. 5 Our sample only ncludes those keywords for whch we had access to data on total weekly searches for those keywords on Google. Ths resulted n a dataset wh 46 unque keywords and a total of 400 observatons. Table reports the summary statstcs. Not surprsngly, a majory of the keywords have brand-specfc nformaton (65%) whle only 8.8% of the keywords have retaler-specfc nformaton. Interestngly, we note that the mean clck-through rate was 6.6% and.77%, respectvely, from pad and organc searches. The mean converson rate was 5.7% and.67%, respectvely, from pad and organc searches. Fnally, note that the mean prof from pad search advertsements was 3. tmes hgher than that from organc search lstngs. = = Insert Table Here = = 4. Emprcal Fndngs Frst, note from table that all three keyword-specfc characterstcs (Retaler, Brand and Length) sgnfcantly predct the search volume. Specfcally, retaler-specfc keywords are assocated wh a hgher volume of searches whle keywords contanng nformaton on brands (manufacturer or product) are assocated wh a lower volume of searches. Moreover, the volume of searches decreases wh an ncrease n the length of the keyword,.e., an ncrease n the specfcy of the search. = = Insert Table Here = = From tables 3a and 3b we see that the average magnude of nterdependence (the parameter, ) between pad clcks and organc clcks s posve and statstcally very sgnfcant. Therefore, a hgher probably (number) of clcks on organc lstngs s correlated wh a hgher probably (number) of clcks on pad search ads, and vce versa. 6 Ths hghlghts that the presence of organc search lstngs has a posve assocaton wh the average clck-through rates n pad search advertsements, and vce-versa. We also fnd that the magnude of ths posve nterdependence between pad search and organc search s asymmetrc. On average, the average mpact of organc clcks on ncreases n the utles of a pad clck 5 The CPC for clcks on organc lstng s always zero. We also ran all the estmatons whout factorng n the contrbuton margn of the dfferent keyword. All our results are robust to the use of gross margns only. 6 The three characterstcs of a keyword (Retaler, Brand, and Length) are all mean centered, and hence the ntercept can be vewed as the mean effect. 6

17 (0.98) s 3.5 tmes stronger than the average mpact of pad clcks on ncreases n organc clck (0.8). In,0 0, addon, from equatons (3a-3d), we also computed the ndvdual probables p,, p, p and the lft n probables from a un ncrease n utly based on the emprcal estmates and the summary statstcs. We found that un ncrease n pad utly ncreases organc clck-through probably by.5 tmes more than vce-versa. As predcted, Rank has an overall negatve relatonshp wh CTR n both pad and organc lstngs as seen n Table. The poson of the advertsement lnk on the search engne results page clearly plays an mportant role n nfluencng clck-through rates. Ths knd of prmacy effect has also been seen n other emprcal studes of the onlne world (Ansar and Mela 003, Brynjolfsson et al. 004, Rutz and Buckln 007, Baye et al. 009, Ghose and Yang 009). Interestngly, the magnude of the effect of rank on clck-through rates s dfferent for pad searches from organc searches. The magnude of the Rank coeffcent s smaller (-0.0) for pad searches than for organc searches (-0.06) suggestng that keyword s poson on the screen plays a relatvely more mportant role n nfluencng clcks n organc search compared to clcks n pad search. We fnd that neher the presence of a brand name nor the retaler s own name n the search keyword has a statstcally sgnfcant effect on clck-through rates n pad lstngs. However, n the case of organc lstngs we fnd that the coeffcent of Brand s posve and sgnfcant (0.9) whle that for Retaler s negatve and sgnfcant (-0.58). The coeffcent of Length s sgnfcantly posve n organc search (0.7) suggestng that longer keywords that typcally represent more goal-orented searches for specfc products, tend to experence hgher clck-through rates on organc lstngs. As shown n Table, many of the estmated varances (unobserved heterogeney) of the ntercept, the nterdependence effect, and the Rank coeffcent are sgnfcant n both organc and pad search clck-through probables. Ths suggests that the baselne clck-through rates and the way that keyword rankng predcts the clck-through rates are dfferent across keywords, drven by factors beyond the three observed keyword characterstcs. = = Insert Table 3 Here = = Next, consder Table 4 wh fndngs on converson rates. We fnd that Rank has a negatve relatonshp wh converson rates from both pad and organc searches but s statstcally sgnfcant only for organc lstngs. Ths mples that the lower the Rank (.e., hgher the poson of the sponsored lstng on the screen), the hgher s the Converson Rate n organc search. On the other hand, CTR has a posve effect on the Converson Rate n pad search but a statstcally nsgnfcant effect n organc search converson rates. Overall, ths suggests that there s an ndrect effect of Rank on the converson probably for pad search (through s effect on clck-through rate), but a drect effect for organc search. Our analyss also reveals the coeffcent of Retaler s posve and sgnfcant for conversons n pad search but statstcally nsgnfcant for organc search. As shown n Table 4, many of the estmated 7

18 varances (unobserved heterogeney) for the ntercept and the Rank coeffcent are sgnfcant n both organc and pad search converson rate. Ths suggests that the baselne converson rates and the way that keyword rankng predcts the converson rates are dfferent across keywords, drven by unobserved factors beyond the observed characterstcs. = = Insert Table 4 Here = = The analyss of CPC reveals that Brand has a statstcally sgnfcant and negatve effect on keyword cost-per-clck. The coeffcent of Lag Pad_Rank s negatve and statstcally sgnfcant whle that for Lag Organc_Rank s negatve but statstcally nsgnfcant. Smlarly, the coeffcent of Lag Prof s statstcally sgnfcant only n the case of organc search lstngs. These fndngs suggest that whle past performance metrcs (lagged pad rank and lagged organc prof) are beng ncorporated by the frm pror to bddng for keywords, may not be bddng n an optmal manner. Ths s further evdent n the negatve correlaton of the Brand dummy wh CPC. Gven that brand-specfc keywords are more competve, one would expect ths assocaton to be the other way around. = = Insert Table 5 Here = = On the analyss of keyword rank, we fnd that Brand and Retaler have a statstcally sgnfcant and negatve relatonshp wh Rank, suggestng that keywords that have brand-specfc nformaton or retaler-specfc nformaton generally tend to be lsted hgher up on the screen. Both CPC and Lag CTR are statstcally sgnfcant and nversely related to Rank as expected (see for example, Ghose and Yang 009). Further, the effect of maxmum bd prce across all competng frms on Rank s also posve and statstcally sgnfcant thereby confrmng that the extent of competon from other advertsers plays a key role n the sponsored search auctons outcomes. = = Insert Table 6 Here = = Fnally, s worth notng n Table 7 that the unobserved covarance between several varables s statstcally sgnfcant. Ths suggests that keyword rankng s endogenous and a frm s bds for a gven keyword are lkely to be based on the same keyword s past performance. = = Insert Table 7 Here = = 4.3. Robustness Checks: Alternate Model Specfcatons In order to demonstrate the robustness of our man results usng the ntegrated model, we also explore two alternatve model specfcatons. In partcular, we buld two non-nested models. Frst, we buld and estmate an autologstc model n order to examne the sgn of the nterdependence between pad and organc clck-throughs. Thereafter, we explore the nature of ths nterdependence to examne f there s some knd of asymmetry n the relatonshp usng a smultaneous move game structural model. For 8

Analyzing the Relationship Between Organic and Sponsored Search Advertising: Positive, Negative, or Zero Interdependence?

Analyzing the Relationship Between Organic and Sponsored Search Advertising: Positive, Negative, or Zero Interdependence? Unversy of Pennsylvana ScholarlyCommons Operatons, Informaton and Decsons Papers Wharton Faculty Research 7-00 Analyzng the Relatonshp Between Organc and Sponsored Search Advertsng: Posve, Negatve, or

More information

An Empirical Analysis of Search Engine Advertising: Sponsored Search in Electronic Markets

An Empirical Analysis of Search Engine Advertising: Sponsored Search in Electronic Markets Unversty of Pennsylvana ScholarlyCommons Operatons, Informaton and Decsons Papers Wharton Faculty Research -9 An Emprcal Analyss of Search Engne Advertsng: Sponsored Search n Electronc Markets Anndya Ghose

More information

An Empirical Analysis of Search Engine Advertising:Sponsored Search in Electronic Markets 1

An Empirical Analysis of Search Engine Advertising:Sponsored Search in Electronic Markets 1 An Emprcal Analyss of Search Engne Advertsng:Sponsored Search n Electronc Markets Anndya Ghose Stern School of Busness New York Unversty aghose@stern.nyu.edu Sha Yang Stern School of Busness New York Unversty

More information

WISE 2004 Extended Abstract

WISE 2004 Extended Abstract WISE 2004 Extended Abstract Does the Internet Complement Other Marketng Channels? Evdence from a Large Scale Feld Experment Erc Anderson Kellogg School of Management, Northwestern Unversty Erk Brynjolfsson

More information

Volume 30, Issue 4. Who likes circus animals?

Volume 30, Issue 4. Who likes circus animals? Volume 30, Issue 4 Who lkes crcus anmals? Roberto Zanola Unversty of Eastern Pedmont Abstract Usng a sample based on 268 questonnares submtted to people attendng the Acquatco Bellucc crcus, Italy, ths

More information

International Trade and California Employment: Some Statistical Tests

International Trade and California Employment: Some Statistical Tests Internatonal Trade and Calforna Employment: Some Statstcal Tests Professor Dwght M. Jaffee Fsher Center for Real Estate and Urban Economcs Haas School of Busness Unversty of Calforna Berkeley CA 94720-1900

More information

1 Basic concepts for quantitative policy analysis

1 Basic concepts for quantitative policy analysis 1 Basc concepts for quanttatve polcy analyss 1.1. Introducton The purpose of ths Chapter s the ntroducton of basc concepts of quanttatve polcy analyss. They represent the components of the framework adopted

More information

The link between immigration and trade in Spain

The link between immigration and trade in Spain 1 The lnk between mmgraton and trade n Span José Vcente Blanes-Crstóbal March, 2003 Abstract Ths paper tests for the mpact of mmgraton on blateral trade usng Spansh data. It also explores some possble

More information

Extended Abstract for WISE 2005: Workshop on Information Systems and Economics

Extended Abstract for WISE 2005: Workshop on Information Systems and Economics Extended Abstract for WISE 5: Workshop on Informaton Systems and Economcs How Many Bundles?:An Analyss on Customzed Bundlng of Informaton Goods wth Multple Consumer Types Wendy HUI Ph.D. Canddate Department

More information

Experiments with Protocols for Service Negotiation

Experiments with Protocols for Service Negotiation PROCEEDINGS OF THE WORKSHOP ON APPLICATIONS OF SOFTWARE AGENTS ISBN 978-86-7031-188-6, pp. 25-31, 2011 Experments wth Protocols for Servce Negotaton Costn Bădcă and Mhnea Scafeş Unversty of Craova, Software

More information

A Longer Tail?: Estimating The Shape of Amazon s Sales Distribution Curve in Erik Brynjolfsson, Yu (Jeffrey) Hu, Michael D.

A Longer Tail?: Estimating The Shape of Amazon s Sales Distribution Curve in Erik Brynjolfsson, Yu (Jeffrey) Hu, Michael D. A Longer Tal?: Estmatng The Shape of Amazon s Sales Dstrbuton Curve n 2008 1. Introducton Erk Brynjolfsson, Yu (Jeffrey) Hu, Mchael D. Smth The term The Long Tal was coned by Wred s Chrs Anderson (Anderson

More information

Consumer Choice of Service Plan: Switching Cost and Usage Uncertainty

Consumer Choice of Service Plan: Switching Cost and Usage Uncertainty Consumer Choce of Servce Plan: Swchng Cost and Usage Uncertanty Png Xao Tat Chan Chakravarth Narasmhan 1 May, 007 1 Png Xao s a doctoral student n Marketng, Tat Chan s an Assstant Professor of Marketng,

More information

A Two-Echelon Inventory Model for Single-Vender and Multi-Buyer System Through Common Replenishment Epochs

A Two-Echelon Inventory Model for Single-Vender and Multi-Buyer System Through Common Replenishment Epochs A Two-Echelon Inventory Model for Sngle-Vender and Mult-Buyer System Through Common Replenshment Epochs Wen-Jen Chang and Chh-Hung Tsa Instructor Assocate Professor Department of Industral Engneerng and

More information

Stay Out of My Forum! Evaluating Firm Involvement in Online Ratings Communities Neveen Awad and Hila Etzion

Stay Out of My Forum! Evaluating Firm Involvement in Online Ratings Communities Neveen Awad and Hila Etzion Stay Out of My Forum! Evaluatng Frm Involvement n Onlne Ratngs Communtes Neveen Awad and Hla Etzon. INTRODUCTION A growng number of onlne retalers are enablng and encouragng consumers to post revews of

More information

Do not turn over until you are told to do so by the Invigilator.

Do not turn over until you are told to do so by the Invigilator. UNIVERSITY OF EAST ANGLIA School of Economcs Man Seres PG Examnaton 016-17 FINANCIAL ECONOMETRICS ECO-7009A Tme allowed: HOURS Answer ALL FOUR questons. Queston 1 carres a weght of 5%; queston carres 0%;

More information

Evaluating the statistical power of goodness-of-fit tests for health and medicine survey data

Evaluating the statistical power of goodness-of-fit tests for health and medicine survey data 8 th World IMACS / MODSIM Congress, Carns, Australa 3-7 July 29 http://mssanz.org.au/modsm9 Evaluatng the statstcal power of goodness-of-ft tests for health and medcne survey data Steele, M.,2, N. Smart,

More information

Modeling Consumer Footprints on Search Engines: An Interplay with Social Media 1

Modeling Consumer Footprints on Search Engines: An Interplay with Social Media 1 Modelng Consumer Footprnts on Search Engnes: An Interplay wth Socal Meda 1 Anndya Ghose Stern School of Busness, New York Unversty, aghose@stern.nyu.edu Panagots G. Iperots Stern School of Busness, New

More information

Prediction algorithm for users Retweet Times

Prediction algorithm for users Retweet Times , pp.9-3 http://dx.do.org/0.457/astl.05.83.03 Predcton algorthm for users Retweet Tmes Hahao Yu, Xu Feng Ba,ChengZhe Huang, Haolang Q Helongang Insttute of Technology, Harbn, Chna Abstract. In vew of the

More information

The ranks of Indonesian and Japanese industrial sectors: A further study

The ranks of Indonesian and Japanese industrial sectors: A further study Journal of Physcs: Conference Seres PAPER OPEN ACCESS The ranks of Indonesan and Japanese ndustral sectors: A further study To cte ths artcle: Ubadllah Zuhd 2017 J. Phys.: Conf. Ser. 820 012029 Vew the

More information

RELATIONSHIP BETWEEN BUSINESS STRATEGIES FOLLOWED BY SERVICE ORGANIZATIONS AND THEIR PERFORMANCE MEASUREMENT APPROACH

RELATIONSHIP BETWEEN BUSINESS STRATEGIES FOLLOWED BY SERVICE ORGANIZATIONS AND THEIR PERFORMANCE MEASUREMENT APPROACH RELATIONSHIP BETWEEN BUSINESS STRATEGIES FOLLOWED BY SERVICE ORGANIZATIONS AND THEIR PERFORMANCE MEASUREMENT APPROACH Carlos F. Gomes, School of Economcs, ISR- Insttute of Systems and Robotcs, Unversty

More information

Volume 29, Issue 2. How do firms interpret a job loss? Evidence from the National Longitudinal Survey of Youth

Volume 29, Issue 2. How do firms interpret a job loss? Evidence from the National Longitudinal Survey of Youth Volume 29, Issue 2 How do frms nterpret a job loss? Evdence from the Natonal Longtudnal Survey of Youth Stephen M. Kosovch Stephen F. Austn State Unversty Abstract Emprcal studes n the job dsplacement

More information

Impacts of supply and demand shifts

Impacts of supply and demand shifts Impacts of supply and demand shfts 1. Impacts of Supply shft S S S S Same sze of shft D D Elastc Demand Inelastc demand 2. Impacts of Demand shft D D S D D S Same sze of shft D Elastc Supply Inelastc demand

More information

Key Words: dairy; profitability; rbst; recombinant bovine Somatotropin.

Key Words: dairy; profitability; rbst; recombinant bovine Somatotropin. AgBoForum Volume 4, Number 2 2001 Pages 115-123 THE ESTIMATED PROFIT IMPACT OF RECOMBINANT BOVINE SOMATOTROPIN ON NEW YORK DAIRY FARMS FOR THE YEARS 1994 THROUGH 1997 Loren W. Tauer 1 Data from New York

More information

The Effect of Retail Grocery Coupons for Breakfast Cereals on Household Purchasing Behavior

The Effect of Retail Grocery Coupons for Breakfast Cereals on Household Purchasing Behavior The Effect of Retal Grocery Coupons for Breakfast Cereals on Household Purchasng Behavor Joshua Bernng A and Hualu Zheng B A Assstant Professor Department of Agrcultural & Resource Economcs Unversy of

More information

Innovation in Portugal:

Innovation in Portugal: Innovaton n Portugal: What can we learn from the CIS III? Innovaton and Productvty Pedro Moras Martns de Fara pedro.fara@dem.st.utl.pt Globelcs Academy 2005 25 May 2005 Introducton The study of the relatonshp

More information

CHICKEN AND EGG? INTERPLAY BETWEEN MUSIC BLOG BUZZ AND ALBUM SALES

CHICKEN AND EGG? INTERPLAY BETWEEN MUSIC BLOG BUZZ AND ALBUM SALES Assocaton for Informaton Systems AIS Electronc Lbrary (AISeL) PACIS 2009 Proceedngs Pacfc Asa Conference on Informaton Systems (PACIS) July 2009 CHICKEN AND EGG? INTERPLAY BETWEEN MUSIC BLOG BUZZ AND ALBUM

More information

Impact of Internet Technology on Economic Growth in South Asia with Special Reference to Pakistan

Impact of Internet Technology on Economic Growth in South Asia with Special Reference to Pakistan Pakstan Journal of Socal Scences (PJSS) Vol. 35, No. 2 (2015), pp. 777-784 Impact of Internet Technology on Economc Growth n South Asa wth Specal Reference to Pakstan RzwanaYasmeen FUUAST, School of Economc

More information

Battle of the Retail Channels: How Internet Selection and Local Retailer Proximity Drive Cross-Channel Competition

Battle of the Retail Channels: How Internet Selection and Local Retailer Proximity Drive Cross-Channel Competition Assocaton for Informaton Systems AIS Electronc Lbrary (AISeL) ICIS 2007 Proceedngs Internatonal Conference on Informaton Systems (ICIS) December 2007 Battle of the Retal Channels: How Internet Selecton

More information

Guidelines on Disclosure of CO 2 Emissions from Transportation & Distribution

Guidelines on Disclosure of CO 2 Emissions from Transportation & Distribution Gudelnes on Dsclosure of CO 2 Emssons from Transportaton & Dstrbuton Polcy Research Insttute for Land, Infrastructure and Transport June 2012 Contents 1. Introducton...- 3-1.1 Purpose and concept...- 3-1.2

More information

emissions in the Indonesian manufacturing sector Rislima F. Sitompul and Anthony D. Owen

emissions in the Indonesian manufacturing sector Rislima F. Sitompul and Anthony D. Owen Mtgaton optons for energy-related CO 2 emssons n the Indonesan manufacturng sector Rslma F. Stompul and Anthony D. Owen School of Economcs, The Unversty of New South Wales, Sydney, Australa Why mtgaton

More information

The Application of Uninorms in Importance-Performance Analysis

The Application of Uninorms in Importance-Performance Analysis The Applcaton of Unnorms n Importance-Performance Analyss BENOIT DEPAIRE, KOEN VANHOOF, GEERT WETS Department of Busness Studes Hasselt Unversty Campus Depenbeek, Agoralaan Gebouw D, 3590 Depenbeek BELGIUM

More information

Willingness to Pay for Beef Quality Attributes: Combining Mixed Logit and Latent Segmentation Approach

Willingness to Pay for Beef Quality Attributes: Combining Mixed Logit and Latent Segmentation Approach Wllngness to Pay for Beef Qualty Attrbutes: Combnng Mxed Logt and Latent Segmentaton Approach Chanjn Chung Department of Agrcultural Economcs Oklahoma State Unversty Stllwater, OK 74078 Emal: chanjn.chung@okstate.edu

More information

Product Innovation Risk Management based on Bayesian Decision Theory

Product Innovation Risk Management based on Bayesian Decision Theory Advances n Management & Appled Economcs, vol., no., 0, - ISS: 79-7 (prnt verson), 79-7 (onlne) Internatonal Scentfc Press, 0 Product Innovaton Rsk Management based on Bayesan Decson Theory Yngchun Guo

More information

Analyses Based on Combining Similar Information from Multiple Surveys

Analyses Based on Combining Similar Information from Multiple Surveys Secton on Survey Research Methods JSM 009 Analyses Based on Combnng Smlar Informaton from Multple Surveys Georga Roberts, Davd Bnder Statstcs Canada, Ottawa Ontaro Canada KA 0T6 Statstcs Canada, Ottawa

More information

Consumption capability analysis for Micro-blog users based on data mining

Consumption capability analysis for Micro-blog users based on data mining Consumpton capablty analyss for Mcro-blog users based on data mnng ABSTRACT Yue Sun Bejng Unversty of Posts and Telecommuncaton Bejng, Chna Emal: sunmoon5723@gmal.com Data mnng s an effectve method of

More information

Online Product Opinions: Incidence, Evaluation and Evolution

Online Product Opinions: Incidence, Evaluation and Evolution Onlne Product Opnons: Incdence Evaluaton and Evoluton Wendy W. Moe Assocate Professor of Marketng Robert H. Smth School of Busness Unversty of Maryland 3469 Van Munchng Hall College Park MD 20472 301)

More information

International Trade and California s Economy: Summary of the Data

International Trade and California s Economy: Summary of the Data Internatonal Trade and Calforna s Economy: Summary of the Data by Professor Dwght M. Jaffee Fsher Center for Real Estate and Urban Economcs Haas School of Busness Unversty of Calforna Berkeley CA 94720-1900

More information

Supplier selection and evaluation using multicriteria decision analysis

Supplier selection and evaluation using multicriteria decision analysis Suppler selecton and evaluaton usng multcrtera decson analyss Stratos Kartsonaks 1, Evangelos Grgorouds 2, Mchals Neofytou 3 1 School of Producton Engneerng and Management, Techncal Unversty of Crete,

More information

Appendix 6.1 The least-cost theorem and pollution control

Appendix 6.1 The least-cost theorem and pollution control Appendx 6.1 The least-cost theorem and polluton control nstruments Ths appendx s structured as follows. In Part 1, we defne the notaton used and set the scene for what follows. Then n Part 2 we derve a

More information

Sources of information

Sources of information MARKETING RESEARCH FACULTY OF ENGINEERING MANAGEMENT Ph.D., Eng. Joanna Majchrzak Department of Marketng and Economc Engneerng Mal: joanna.majchrzak@put.poznan.pl Meetngs: Monday 9:45 11:15 Thursday 15:10

More information

GETTING STARTED CASH & EXPENSE PLANNING

GETTING STARTED CASH & EXPENSE PLANNING FINANCE TO SCALE Brex has partnered wth Hubspot for Startups to provde a fnancal plannng and expense management resource for your startup. These materals are a framework for fnancal plannng pror to scalng

More information

Impact of ROI on Bidding and Revenue in Sponsored Search Advertisement Auctions

Impact of ROI on Bidding and Revenue in Sponsored Search Advertisement Auctions Impact of ROI on Bddng and Revenue n Sponsored Search Advertsement Auctons Boleslaw K. Szymansk and Juong-Sk Lee Optmaret Inc., Newtonvlle, NY, 228 Rensselaer Polytechnc Insttute, Computer Scence Department,

More information

Do Competing Suppliers Maximize Profits as Theory Suggests? An Empirical Evaluation

Do Competing Suppliers Maximize Profits as Theory Suggests? An Empirical Evaluation Unversty of Massachusetts Boston ScholarWorks at UMass Boston Management Scence and Informaton Systems Faculty Publcaton Seres Management Scence and Informaton Systems January 2015 as Theory Suggests?

More information

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and

This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and Ths artcle appeared n a journal publshed by Elsever. The attached copy s furnshed to the author for nternal non-commercal research and educaton use, ncludng for nstructon at the authors nsttuton and sharng

More information

The Capitalization of Area Payment into Land Rental Prices: Micro-evidence from Italy

The Capitalization of Area Payment into Land Rental Prices: Micro-evidence from Italy The Capalzaton of Area Payment nto Land Rental Prces: Mcro-evdence from Italy Guastella G. 1, Moro D. 2, Sckoka P. 2, and Venezan M. 1 1 Dpartmento d Economa Agroalmentare e DSES, Unversà Cattolca del

More information

Econometric Methods for Estimating ENERGY STAR Impacts in the Commercial Building Sector

Econometric Methods for Estimating ENERGY STAR Impacts in the Commercial Building Sector Econometrc Methods for Estmatng ENERGY STAR Impacts n the Commercal Buldng Sector Marvn J. Horowtz, Demand Research Angela Coyle, U.S. Envronmental Protecton Agency ABSTRACT The early stages of developng

More information

Effect of Adjacent Product Price on Customer s Willingness to Pay of Focal Brand: A Bayesian Approach

Effect of Adjacent Product Price on Customer s Willingness to Pay of Focal Brand: A Bayesian Approach Theoretcal Economcs Letters, 2017, 7, 1940-1949 http://www.scrp.org/journal/tel ISSN Onlne: 2162-2086 ISSN Prnt: 2162-2078 Effect of Adjacent Product Prce on Customer s Wllngness to Pay of Focal Brand:

More information

Price and Quality Effects of Generic Advertising: The Case of Norwegian Salmon

Price and Quality Effects of Generic Advertising: The Case of Norwegian Salmon Natonal Instute for Commody Promoton Research & Evaluaton October 2003 NICPRE 03-02 R.B. 2003-05 Prce and Qualy Effects of Generc Advertsng: The Case of Norwegan Salmon by: Øysten Myrland Unversy of Tromsø,

More information

The Spatial Equilibrium Monopoly Models of the Steamcoal Market

The Spatial Equilibrium Monopoly Models of the Steamcoal Market Advances n Management & Appled Economcs, vol.2, no.3, 2012, 125-132 ISSN: 1792-7544 (prnt verson), 1792-7552 (onlne) Scenpress Ltd, 2012 The Spatal Equlbrum Monopoly Models of the Steamcoal Maret Hu Wen

More information

Analysis Online Shopping Behavior of Consumer Using Decision Tree Leiyue Yao 1, a, Jianying Xiong 2,b

Analysis Online Shopping Behavior of Consumer Using Decision Tree Leiyue Yao 1, a, Jianying Xiong 2,b Advanced Materals Research Onlne: 2011-07-04 ISSN: 1662-8985, Vols. 271-273, pp 891-894 do:10.4028/www.scentfc.net/amr.271-273.891 2011 Trans Tech Publcatons, Swtzerland Analyss Onlne Shoppng Behavor of

More information

Why do we have inventory? Inventory Decisions. Managing Economies of Scale in the Supply Chain: Cycle Inventory. 1. Understanding Inventory.

Why do we have inventory? Inventory Decisions. Managing Economies of Scale in the Supply Chain: Cycle Inventory. 1. Understanding Inventory. -- Chapter 10 -- Managng Economes of Scale n the Supply Chan: Cycle Inventory Pros: Why do we have nventory? To overcome the tme and space lags between producers and consumers To meet demand/supply uncertanty

More information

THE SOCIAL CONTEXT OF ACTIVITY-SCHEDULING: A DISCRETE-CONTINUOUS MODEL OF THE RELATIONSHIP BETWEEN WITH WHOM AND EPISODE START TIME AND DURATION

THE SOCIAL CONTEXT OF ACTIVITY-SCHEDULING: A DISCRETE-CONTINUOUS MODEL OF THE RELATIONSHIP BETWEEN WITH WHOM AND EPISODE START TIME AND DURATION THE SOCIAL CONTEXT OF ACTIVITY-SCHEDULING: A DISCRETE-CONTINUOUS MODEL OF THE RELATIONSHIP BETWEEN WITH WHOM AND EPISODE START TIME AND DURATION Khandker M. N. Habb Department of Cvl & Envronmental Engneerng

More information

Evaluating The Performance Of Refrigerant Flow Distributors

Evaluating The Performance Of Refrigerant Flow Distributors Purdue Unversty Purdue e-pubs Internatonal Refrgeraton and Ar Condtonng Conference School of Mechancal Engneerng 2002 Evaluatng The Performance Of Refrgerant Flow Dstrbutors G. L Purdue Unversty J. E.

More information

SEARCH LESS, FIND MORE? EXAMINING LIMITED CONSUMER SEARCH WITH SOCIAL MEDIA AND PRODUCT SEARCH ENGINES

SEARCH LESS, FIND MORE? EXAMINING LIMITED CONSUMER SEARCH WITH SOCIAL MEDIA AND PRODUCT SEARCH ENGINES SEARCH LESS, FIND MORE? EXAMINING LIMITED CONSUMER SEARCH WITH SOCIAL MEDIA AND PRODUCT SEARCH ENGINES Completed Research Paper Anndya Ghose Stern School of Busness, New York Unversty 44 West 4th Street,

More information

An Analysis of Auction Volume and Market Competition for the Coastal Forest Regions in British Columbia

An Analysis of Auction Volume and Market Competition for the Coastal Forest Regions in British Columbia An Analyss of Aucton Volume and Market Competton for the Coastal Forest Regons n Brtsh Columba Susan Athey, Peter Cramton, and Allan Ingraham 1 Market Desgn Inc. and Crteron Auctons 0 September 00 The

More information

COAL DEMAND AND TRANSPORTATION IN THE OHIO RIVER BASIN:

COAL DEMAND AND TRANSPORTATION IN THE OHIO RIVER BASIN: COAL DEMAND AND TRANSPORTATION IN THE OHIO RIVER BASIN: ESTIMATION OF A CONTINUOUS/DISCRETE DEMAND SYSTEM WITH NUMEROUS ALTERNATIVES * by Kenneth Tran and Wesley W. Wlson December 2011 Abstract Coal-fred

More information

Planning of work schedules for toll booth collectors

Planning of work schedules for toll booth collectors Lecture Notes n Management Scence (0) Vol 4: 6 4 4 th Internatonal Conference on Appled Operatonal Research, Proceedngs Tadbr Operatonal Research Group Ltd All rghts reserved wwwtadbrca ISSN 00-0050 (Prnt),

More information

The Employment Effects of Low-Wage Subsidies

The Employment Effects of Low-Wage Subsidies The Employment Effects of Low-Wage Subsdes Krstna Huttunen Jukka Prttlä Roope Uustalo CESIFO WORKING PAPER NO. 3043 CATEGORY 4: LABOUR MARKETS MAY 2010 An electronc verson of the paper may be downloaded

More information

Bulletin of Energy Economics.

Bulletin of Energy Economics. Bulletn of Energy Economcs http://www.tesdo.org/journaldetal.aspx?id=4 Energy Intensty and Technology Sourcng: A Study of Manufacturng Frms n Inda Santosh Kumar Sahu a,, K. Narayanan b a Madras School

More information

A Group Decision Making Method for Determining the Importance of Customer Needs Based on Customer- Oriented Approach

A Group Decision Making Method for Determining the Importance of Customer Needs Based on Customer- Oriented Approach Proceedngs of the 010 Internatonal Conference on Industral Engneerng and Operatons Management Dhaka, Bangladesh, January 9 10, 010 A Group Decson Makng Method for Determnng the Importance of Customer Needs

More information

Customer Price Sensitivity and the Online Medium

Customer Price Sensitivity and the Online Medium Customer Prce Senstvty and the Onlne Medum Venkatesh Shankar* Arvnd Rangaswamy** Mchael Pusater*** February 1999 *Assstant Professor of Marketng, Robert H. Smth School of Busness, Unversty of Maryland,

More information

A DUOPOLY MODEL OF FIXED COST CHOICE. Charles E. Hegji*

A DUOPOLY MODEL OF FIXED COST CHOICE. Charles E. Hegji* 004 DUOPOY MODE O IXED OST HOIE harles E Hegj* INTRODUTION omparson of frms n ournot and Stackelberg eulbrum s a subject that has receved much attenton unversally mposed assumpton n most dscussons of the

More information

The Role of Price Floor in a Differentiated Product Retail Market

The Role of Price Floor in a Differentiated Product Retail Market Economc Analyss & Polcy, Vol. 40 No. 3, DECEMBER 2010 The Role of Prce Floor n a Dfferentated Product Retal Market Barna Bakó 1 Faculty of Economcs, Corvnus Unversty of Budapest Fovám tér 8, Budapest,

More information

An Empirical Study about the Marketization Degree of Labor Market from the Perspective of Wage Determination Mechanism

An Empirical Study about the Marketization Degree of Labor Market from the Perspective of Wage Determination Mechanism An Emprcal Study about the Marketzaton Degree of Labor Market from the Perspectve of Wage Determnaton Mechansm Qushuo He Shenzhen Insttute of Informaton Technology, Shenzhen 51809, Chna heqs@szt.com.cn

More information

6.4 PASSIVE TRACER DISPERSION OVER A REGULAR ARRAY OF CUBES USING CFD SIMULATIONS

6.4 PASSIVE TRACER DISPERSION OVER A REGULAR ARRAY OF CUBES USING CFD SIMULATIONS 6.4 PASSIVE RACER DISPERSION OVER A REGULAR ARRAY OF CUBES USING CFD SIMULAIONS Jose Lus Santago *, Alberto Martll and Fernando Martn CIEMA (Center for Research on Energy, Envronment and echnology). Madrd,

More information

Logistics Management. Where We Are Now CHAPTER ELEVEN. Measurement. Organizational. Sustainability. Management. Globalization. Culture/Ethics Change

Logistics Management. Where We Are Now CHAPTER ELEVEN. Measurement. Organizational. Sustainability. Management. Globalization. Culture/Ethics Change CHAPTER ELEVEN Logstcs Management McGraw-Hll/Irwn Copyrght 2011 by the McGraw-Hll Companes, Inc. All rghts reserved. Where We Are Now Relatonshps Sustanablty Globalzaton Organzatonal Culture/Ethcs Change

More information

The Long-Term Effects of Price Promotions on Category Incidence, Brand Choice and Purchase Quantity. Koen Pauwels. Dominique M.

The Long-Term Effects of Price Promotions on Category Incidence, Brand Choice and Purchase Quantity. Koen Pauwels. Dominique M. The Long-Term Effects of Prce Promotons on Category Incdence, Brand Choce and Purchase Quantty Koen Pauwels Domnque M. Hanssens S. Sddarth 1 January 11, 2002 1 Koen Pauwels s Assstant Professor at the

More information

EVALUATE THE IMPACT OF CONTEMPORARY INFORMATION SYSTEMS AND THE INTERNET ON IMPROVING BANKING PERFORMANCE

EVALUATE THE IMPACT OF CONTEMPORARY INFORMATION SYSTEMS AND THE INTERNET ON IMPROVING BANKING PERFORMANCE VOLUME 2, 2011 EVALUATE THE IMPACT OF CONTEMPORARY INFORMATION SYSTEMS AND THE INTERNET ON IMPROVING BANKING PERFORMANCE Akram Jalal Karm, Allam M. Hamdan Ahla Unversty, Kngdom of Bahran Ths paper examnes

More information

Development and production of an Aggregated SPPI. Final Technical Implementation Report

Development and production of an Aggregated SPPI. Final Technical Implementation Report Development and producton of an Aggregated SPP Fnal Techncal mplementaton Report Marcus Frdén, Ulf Johansson, Thomas Olsson Servces Producer Prce ndces, Prce Statstcs Unt, Statstcs Sweden 2010 ntroducton

More information

Effect of crop choice on split fertilizer application. Mira Nurmakhanova

Effect of crop choice on split fertilizer application. Mira Nurmakhanova Effect of crop choce on splt fertlzer applcaton Mra Nurmakhanova Department of Economcs Iowa State Unversty Ames, Iowa (515) 294-5051 mra@astate.edu Selected Paper prepared for presentaton at the Amercan

More information

TRANSPORTATION DEMANDS IN THE COLUMBIA-SNAKE RIVER BASIN

TRANSPORTATION DEMANDS IN THE COLUMBIA-SNAKE RIVER BASIN The Navgaton Economc Technologes Program March 1, 2006 navgaton economcs technologes TRANSPORTATION DEMANDS IN THE COLUMBIA-SNAKE RIVER BASIN US Army Corps of Engneers IWR Report 06-NETS-R-03 March 1,

More information

An Implicit Rating based Product Recommendation System

An Implicit Rating based Product Recommendation System An Implct Ratng based Product Recommendaton System Cheh-Yuan Tsa Department of Industral Engneerng and Management Yuan Ze Unversty, Taoyuan Cty, Tawan Tel: (+886) 3-463-8800 ext. 2512, Emal: cytsa@saturn.yzu.edu.tw

More information

Overeducation, undereducation and earnings

Overeducation, undereducation and earnings Overeducaton, undereducaton and earnngs Further evdence on the mportance of ably and measurement error bas 1 Deter Verhaest $ Hogeschool-Unverse Brussel (HUB) Faculty of Economcs and Management, Human

More information

The Labor Market Impacts of. Adult Education and Training in Canada

The Labor Market Impacts of. Adult Education and Training in Canada The Labor Market Impacts of Adult Educaton and Tranng n Canada Shek-wa Hu Department of Economcs Unversty of Western Ontaro shu@uwo.ca Jeffrey Smth Department of Economcs Unversty of Maryland smth@econ.umd.edu

More information

Labour Demand Elasticities in Manufacturing Sector in Kenya

Labour Demand Elasticities in Manufacturing Sector in Kenya Internatonal Journal of Busness and Socal Scence Volume 8 Number 8 August 2017 Labour Demand Elastctes n Manufacturng Sector n Kenya Anthony Wambugu Unversty of Narob School of Economcs P.O.Box 30197-00100

More information

Self Selection and Information Role of Online Product Reviews

Self Selection and Information Role of Online Product Reviews Self Selecton and Informaton Role of Onlne Product Revews Xnxn L Lorn M. Htt The Wharton School, Unversty of Pennsylvana Introducton Word of mouth has long been recognzed as a major drver of product sales.

More information

CONSUMER PRICE INDEX METHODOLOGY (Updated February 2018)

CONSUMER PRICE INDEX METHODOLOGY (Updated February 2018) CONSUMER PRCE NDEX METHODOLOGY (Updated February 208). Purpose, nature and use The purpose s to obtan country representatve data for the prces of goods and servces and to compute overall and group ndces

More information

IMPACT OF ADVERTISING ON DUOPOLY COMPETITION

IMPACT OF ADVERTISING ON DUOPOLY COMPETITION IMPACT OF ADVERTISING ON DUOPOLY COMPETITION Compettve paper n Marketng track POSTGRADUATE PAPER Presented at Irsh Academy of Management Annual Conference, Trnty College, Dubln nd 3 rd September 004 Malcolm

More information

Potato Marketing Factors Affecting Organic and Conventional Potato Consumption Patterns

Potato Marketing Factors Affecting Organic and Conventional Potato Consumption Patterns 1 Potato Marketng Factors Affectng Organc and Conventonal Potato Consumpton Patterns Yue, C. 1, Grebtus, C. 2, Bruhn, M. 3 and Jensen, H.H. 4 1 Unversty of Mnnesota - Twn Ctes, Departments of Appled Economcs

More information

Optimal Issuing Policies for Substitutable Fresh Agricultural Products under Equal Ordering Policy

Optimal Issuing Policies for Substitutable Fresh Agricultural Products under Equal Ordering Policy 06 Internatonal Academc Conference on Human Socety and Culture (HSC 06) ISBN: 978--60595-38-6 Optmal Issung Polces for Substtutable Fresh Agrcultural Products under Eual Orderng Polcy Qao- TENG,a, and

More information

Coupon Redemption and Its Effect on Household Cheese Purchases

Coupon Redemption and Its Effect on Household Cheese Purchases Natonal Insttute for Commodty Promoton Research & Evaluaton October 003 NICPRE 03-0 R.B. 003-04 Coupon Redempton and Its Effect on Household Cheese Purchases by: Dansheng Dong and Harry M. Kaser Cornell

More information

Supplementary Appendix to. Rich Communication, Social Preferences, and Coordinated Resistance against Divide-and-

Supplementary Appendix to. Rich Communication, Social Preferences, and Coordinated Resistance against Divide-and- Supplementary ppendx to Rch Communcaton, Socal Preferences, and Coordnated Resstance aganst Dvde-and- Conquer: Laboratory Investgaton (2013) by Tmothy N. Cason and Va-Lam Mu In the text, before proceedng

More information

Household Budget and Calorie Consume of Livestock Products: Evidence from Indonesia SUMMARY

Household Budget and Calorie Consume of Livestock Products: Evidence from Indonesia SUMMARY Household Budget and Calore Consume of Lvestock Products: Evdence from Indonesa M.A.U. Muzayyanah 1, S.Nurtn 1 & S.P. Syahlan 1 1 Departement of Soco-Economcs, Faculty of Anmal Scence, Gadjah Mada Unversty,

More information

An Empirical Analysis of the RFID Effect on Suppliers' Financial Performance

An Empirical Analysis of the RFID Effect on Suppliers' Financial Performance Assocaton for Informaton Systems AIS Electronc Lbrary (AISeL) ICIS 2008 Proceedngs Internatonal Conference on Informaton Systems (ICIS) 2008 An Emprcal Analyss of the RFID Effect on Supplers' Fnancal Performance

More information

Do Changes in Customer Satisfaction Lead to Changes in Sales Performance in Food Retailing?

Do Changes in Customer Satisfaction Lead to Changes in Sales Performance in Food Retailing? Do Changes n Customer Satsfacton Lead to Changes n Sales Performance n Food Retalng? Mguel I. Gómez Research Assocate Food Industry Management Program Department of Appled Economcs and Management Cornell

More information

The Online Medium and Customer Price Sensitivity

The Online Medium and Customer Price Sensitivity ebusness Research Center Workng Paper 04-1999 The Onlne Medum and Customer Prce Senstvty Venkatesh Shankar Arvnd Rangaswamy Mchael Pusater ebrc 117F Technology Center Buldng Research Park Unversty Park,

More information

RouteMaker Users Guide

RouteMaker Users Guide Gettng Started RouteMaker Users Gude Users Gude Fg. RouteMaker Man Screen RouteMaker s an addon opton for ProftMaker that helps you manage your Route busness. It prnts delvery Manfests for your drvers

More information

The Credit Risk Assessment Model of Internet Supply Chain Finance: Multi-Criteria Decision-Making Model with the Principle of Variable Weight

The Credit Risk Assessment Model of Internet Supply Chain Finance: Multi-Criteria Decision-Making Model with the Principle of Variable Weight Journal of Computer and Communcatons, 2016, 4, 1-11 http://www.scrp.org/journal/jcc ISSN Onlne: 2327-5227 ISSN Prnt: 2327-5219 The Credt Rsk Assessment Model of Internet Supply Chan Fnance: Mult-Crtera

More information

Deriving Value from Social Commerce Networks Andrew T. Stephen and Olivier Toubia*

Deriving Value from Social Commerce Networks Andrew T. Stephen and Olivier Toubia* Dervng Value from Socal Commerce Networks Andrew T. Stephen and Olver Touba* Andrew T. Stephen Doctoral Canddate n Marketng Columba Unversty Graduate School of Busness 3 Broadway Urs 3 New York NY 7 Emal

More information

Varunraj Valsaraj, Kara Kockelman, Jennifer Duthie, and Brenda Zhou University of Texas at Austin. Original Version: September 2007.

Varunraj Valsaraj, Kara Kockelman, Jennifer Duthie, and Brenda Zhou University of Texas at Austin. Original Version: September 2007. FORECASTING EMPLOYMENT & POPULATION IN TEXAS: An Investgaton on TELUM Requrements Assumptons and Results ncludng a Study of Zone Sze Effects for the Austn and Waco Regons Varunraj Valsaraj Kara Kockelman

More information

The Credit Risk Assessment Model of Internet Supply Chain Finance: Multi-Criteria Decision-Making Model with the Principle of Variable Weight

The Credit Risk Assessment Model of Internet Supply Chain Finance: Multi-Criteria Decision-Making Model with the Principle of Variable Weight Journal of Computer and Communcatons, 2017, 5, 20-30 http://www.scrp.org/journal/jcc ISSN Onlne: 2327-5227 ISSN Prnt: 2327-5219 The Credt Rsk Assessment Model of Internet Supply Chan Fnance: Mult-Crtera

More information

MODERN PRICING STRATEGIES IN THE INTERNET MARKET

MODERN PRICING STRATEGIES IN THE INTERNET MARKET The 11 th Internatonal Days of Statstcs and Economcs, Prague, September 14-16, 2017 MODERN PRICING STRATEGIES IN THE INTERNET MARKET Maksm Patshn Sergey Kulpn Abstract The artcle provdes an overvew of

More information

iafor The International Academic Forum

iafor The International Academic Forum The Study of Servce Qualy and Costs of Hospals n Socal Secury System n Thaland Nantarat Tangvoontham, Srnakharnwrot Unversy, Thaland Papusson Chawat, Kng Mongkut s The Asan Conference on the Socal Scences

More information

Influence of Selected Variables on the Price of Petroleum Products in the Republic of Croatia

Influence of Selected Variables on the Price of Petroleum Products in the Republic of Croatia Economy Transdscplnarty Cognton www.ugb.ro/etc Volume 6, Issue 2/203 24-30 Influence of Selected Varables on the Prce of Petroleum Products n the Republc of Croata Ton MILUN, Marna HORVATOVIĆ, Ivona POSARIĆ

More information

The Substitutability of Labor of Selected Ethnic Groups in the US Labor Market

The Substitutability of Labor of Selected Ethnic Groups in the US Labor Market DISCUSSION PAPER SERIES IZA DP No. 1945 The Substtutablty of Labor of Selected Ethnc Groups n the US Labor Market Martn Kahanec January 2006 Forschungsnsttut zur Zukunft der Arbet Insttute for the Study

More information

Farm-Specific Factors Affecting the Choice Between Conventional and Organic Dairy Farming

Farm-Specific Factors Affecting the Choice Between Conventional and Organic Dairy Farming Farm-Specfc Factors Affectng the Choce Between Conventonal and Organc Dar Farmng Cornels Gardebroek E-mal: koos.gardebroek@alg.aae.wau.nl Paper prepared for presentaton at the X th EAAE Congress Explorng

More information

Experimental Validation of a Suspension Rig for Analyzing Road-induced Noise

Experimental Validation of a Suspension Rig for Analyzing Road-induced Noise Expermental Valdaton of a Suspenson Rg for Analyzng Road-nduced Nose Dongwoo Mn 1, Jun-Gu Km 2, Davd P Song 3, Yunchang Lee 4, Yeon June Kang 5, Kang Duc Ih 6 1,2,3,4,5 Seoul Natonal Unversty, Republc

More information

Horizontal Product Differentiation in Auctions and Multilateral Negotiations *

Horizontal Product Differentiation in Auctions and Multilateral Negotiations * Horzontal Product Dfferentaton n Auctons and Multlateral Negotatons * Charles J. Thomas Bart J. Wlson John E. Walker Department of Economcs Economc Scence Insttute Clemson Unversty Chapman Unversty 222

More information