<html>
<head>
<meta http-equiv="content-type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div>
<div>
<div>
<p class="MsoNormal">Of special interest....<br>
</p>
<p class="MsoNormal"><br>
</p>
<p class="MsoNormal"><br>
Patrick Bajari is Vice President and Chief Economist for
Amazon. He will be the speaker at the Heinz Faculty
Research Seminar on Monday, <b>December 7th, 2015 at 12:00
noon in room 1000, Hamburg Hall</b>. Below you will find
the details for his talk. Please distribute this
announcement to appropriate lists in your departments.<o:p></o:p></p>
</div>
</div>
<div>
<div>
<p class="MsoNormal"> <o:p></o:p></p>
</div>
</div>
<div>
<div>
<p class="MsoNormal">Title:<br>
<br>
<b>Demand Estimation in a High Dimensional Data and an
Application to Amazon’s Book Marketplace<br>
</b> <br>
Abstract:<br>
<br>
In this paper, we document some basic facts about Amazon’s
e-commerce platform using several unique datasets. There are
both sizeable applied theory and empirical literatures which
address e-commerce. We contribute to the empirical
literature by examining datasets from Amazon that contain
the entire population of product sales, product offers and
sellers from multiple countries and for multiple product
lines. We also include important control variables that
have been omitted from previous research. We begin by
examining a dataset of all physical books and e-books that
were buyable on Amazon as of November 2014. The selection
available to consumers is large - there were 44 million
unique physical book titles available for purchase in the US
and over 33 million unique physical book titles in the EU in
November 2014. Approximately two million e-books are
self-published accounting for two thirds of total selection.
Next, following the method in Brynjolfsson, Hu and Smith
(2003) we estimate the compensating variation for book
selection available on Amazon’s book marketplace. We apply a
novel estimation procedure to obtain elasticities required
for this calculation which incorporates a mixture of
traditional econometric approaches and machine
learning approaches. We estimate an annualized 2014 benefit
to consumers of $2.2 billion in the US and $2.1 billion in
the EU. We continue by documenting some basic facts about
the supply side of the platform. Specifically, we summarize
the number of offers by seller in select product categories,
summarize the number of distinct 3rd party sellers
listing products on Amazon, and provide estimates for the
number of employees at 3rd party sellers. An average book in
our sample has 20 competing offers. Consistent with a large
number of multiple offers we find that the median
residual elasticity ranges from -7 to -32 for several
product lines in multiple countries. There were over 1.1
million merchants with offers on books on November 2014 and
hundreds of thousands of merchants offering books in the EU.
Finally, we estimate the number of full time employees
directly attributable to all sellers in all product lines.
We estimate 414,550 employees at 3rd part sellers in the US
and 233,612 employees at 3rd party sellers in the EU in
2014.<br>
<br>
Bio:<br>
<br>
Pat Bajari is Vice President and Chief Economist for Amazon.
He earned his doctorate in economics in 1997 from
the University of Minnesota where he specialized in applied
econometrics and empirical industrial organization. He was
a member of the economics faculty at Harvard, Stanford, Duke
and Minnesota and is a research fellow at the NBER. <br>
<br>
Pat worked for the Boston Consulting Group from 1999 to 2010
in pricing, providing applied econometric work to
guide decision making in a variety of industries. He also
worked as a consultant to the research departments for the
Federal Reserve Banks of San Francisco and Minneapolis and
served as an antitrust expert witness. <br>
<br>
Pat joined Amazon in 2010 where he reports to Jeff Wilke,
the Senior Vice President for the North American
Consumer Platform. He has used economic and econometric
tools to build automated systems for pricing, forecasting,
ordering, marketing spend in addition to providing applied
econometric analysis on a host of other business problems.<o:p></o:p></p>
</div>
</div>
</div>
<p class="MsoNormal"><o:p> </o:p></p>
</body>
</html>