Research Interest

As I am in my first year I don't have a very clear research area. I mainly interested in the following issues:

Marketing decision support systems
Retail management
Data mining
Applied econometrics
Discrete choice models

However, I am quite open minded. One of the advantages of being in a PhD program is that many people is doing lot of interesting things around you. It is quite possible that interacting with many smart people around me I could find an inspired new research interest.

Frequently I have several research ideas. But when I have to start a new research project I already forgot everything. So I will list here a list of alternative research ideas. I know that much of them should be quite fuzzy. Much of them could be bad ideas. But I do not care. The idea is simply to force myself to write something to do a little bit more structured and, the most importnant thing, if somebody think that he/she has a good complemetary idea, you can contact me to see to improve the original one. Of course you can also let me know if someone else alaready solve the problem that I am propposing.

Optimal minimal bid in a Internet Environment (2005)

Internet bidding web sites as eBay, Yahoo auctions or Liquidation.com usually offer auction of thousand of products evaryday. The auction rules are usually the same than used in the real world. However, in the internet environment there is an special factor that could make things different. Auction processes in an Internet bidding usually take several weeks so people do not have a complete information all time. Moreover, there exist non null costs of acces to the updated information. These costs are associated basically to the time required to acces the computer, enter the site and find the product.

In this particular setting is reasonable to think that optimal policy of consumers, is not enter to the site unil the final date of the auction is close enough to enter their best bid and wait for the final result of the game. Facing this behavior (that should be empirically tested), a couple of alternative mechanism could be designed to have different desired consumer behavior: maximizing seller surplus, maximizing number of bidders visits, etc.
For instance, introducing information access cost, it could be interesting to test the expected purchaser behavior under some of the following auction rules:

An increasing minimal jump in the bid (for instance, any new bid should be at least (1+x)% over the previous one.
The minimal bid that each consumer can do varies according the time passed from the previous bid (in order to avoid too small increasement).

An interesting application of these new alternative auction rules could appear if we assume that bidders can be automatic machines (in a B2B setting) or a mixture between machines and persons.

No serious literature review for this issue yet :(

Irrelevant attributes in assortment pereption (2005)

Literature reports that irrelevant attribute could be applied to differentiate is some market (Carpenter, Glazer and Nakamoto. 1994). In a retail market, an important store choice variable is the perception of variety (Hoch, Bradlow and Wansink. 1999). The perception of variety is determined by multiple factor as shelf space, number of SKU and availability of favorite items (Broniarczyk, Hoyer and McAlister. 1998). However, it seems plausible to think that there some irrelevant characteristic of the assortment of a given category that could change the variety perception. Let's suppose for example that in a given category a special exotic SKU from a foreign country is included every week in the category. It is quite plausible to think that the inclusion of this SKU would have a higher impact in the variety perception in the category than the inclusion of a product that share characteristic with other SKU already included in the category. Moreover, this positive impact should remain even when the product included is irrelevant in the sense that is not in the consideration set of the consumer. An alternative framework to this phenomenus could be found following the intertemporal choice (Hoch and Loewenstein. 1991 and Prelec and Loewenstein. 1997). If experimental data show that the propossed phenomenus is signifcant, we could explain it as negative discount rate in the sense that they always prefer to have the posibility to chose between many different products in the future even when they will never consider the posibility to change from his current favorite brand.

(Irrelevant) Attributes in overal store price perception (2005)

The overall price perception has been recognized as a key factor in store choice for years (Brown. 1971). However, there are strong evidence suggesting that consumer spend a short time comparing pricing and maybe as a consequence they not check the price of the item they selected (Dickson and Sawyer. 1990). Then, how they form an overal price store image in their minds? Can managers give them some specific clues to modify this overal image?

Literature reports several studies about how advertising, previous purchases and loyalty affect price sensitivity, but few research have been conducted to study which heuristic procedures are used by purchaser to build an overall price image (Alba, Broniarczyk, Shimp and Urbany. 1994).

Three hypotheses are propposed as alternative (and possible complementary) explanation of how people build his overall store-price image:

H1: The more frequently is purchased the product, the higher the impact in the price perception.
H2: The higher the value of the SKU, the higher the impact in the price perception.
H3: Price perception is not associated to the relevance of the price vector. Moreover it can be modified by aparently irrelevant characteristics as the price of first SKUs observed or the last in the list.

The methodology to conduct this research should be based in a set of experiments in which each hypothesis have to proved using ANOVA techniques.

Introducing Dynamic Effects to SKU-Level Demand Estimations (2005)

From the retailer's perspective, demand estimation at SKU level is a key element in operational decision. In fact, every day, retailers have to take decision about price, promotion and assortment. Most of these decisions should be taken for specific SKU. For instance, prices should be taken for every item. In the same way, store promotion are assigned to a single element instead a complete brand. However, most of research have been devoted to analyze brand level choices. This approach give a good marketing insight to the manufacturer in a tactical level leaving the operational retailer view with very few answers.

An approach that has been proposed to solve this problem is to use an attribute list to describe any SKU in the category (Fader and Hardie. 1996). On these approach, recent research had shown that estimation could be done parsimoniously (Bell, Bonfrer and Chintagunta. 2005). However, all these models are static in the sense that they don't include any explanatory variable to take into acount temporal changes in the preferences. The idea is include some dynamics effects on the attribute based approach that could give both, a better demand forecast and a useful insight about how attribute values change over the time. A brief review and a couple of slides are available describing how these topic could developed.

Science without religion is lame, religion without science is blind.
- Albert Einstein