Jump Main Menu. Go directly to the main content

Sección de idiomas

EN

Fin de la sección de idiomas

Access / Registration

Sección de utilidades

Fin de la sección de utilidades

MENU
Secondary menu End of secondary menu

Research projects

Start of main content

Learning from the Customer: Online Reviews and Corporate Investment

19th National Competition for Economic Research Grants

Economic analysis

Senior Researcher : Miguel Antón Sancho

Research Centre or Institution : IESE Business School. Universidad de Navarra

Abstract

Our project investigates how online customer reviews on Amazon.com can be a powerful source of granular information that can help companies internalize customer opinions to make better investment decisions in the medium and long term. The first stage of the project has consisted of building a bridge between product and brand names on the Amazon marketplace with the respective publicly listed owners, to assign product reviews to the ultimate owner of the product. This will allow us to link customer reviews with firms’ financial data from the CRSP database. On the Amazon side we have access to a dataset of products and reviews that spans from 1996 to 2018, whereas for products and brands owned by public entities thorough research has established that a comprehensive list does not exist, neither publicly nor privately. Therefore, to overcome this challenge we have so far relied on four alternative sources of firm brands and products: the Factset Revere product database, Google searches, S&P’s Capital IQ database and SEC 10-Ks.

The results obtained from data extraction procedures from these sources have been matched with our Amazon dataset, resulting in a match of around 190.000 reviews for 9876 brands. Given our small reviews sample, we have run regressions of firms’ returns on customer reviews, with noisy results, that at the moment are difficult to reconcile with findings in Huang (2018). We have also run regressions of CAPX on different measures of average ratings, with still some noise in the results given the small matched sample.

We are currently working on improving our matching exercise to have a more ample spectrum of reviews. To this end, we are extracting data from the United States Patent and Trademark office database, as patents and trademarks will proxy brands. Then, patents and trademarks data will be matched to respective owners exploiting the Bing Web Search API, as there are peculiarities in this type of data that do not allow us to take this link for granted.

Once this exercise is completed, we will have a satisfactory dataset with brands and firms that will be linked to financial data and reviews to perform meaningful economic analysis. With this information, using panel data econometrics and multivariate regressions, we will be able to estimate the marginal Q using granular, going beyond existing literature that is limited by peculiarities of accounting and financial data. The results will also shed light on mergers and acquisitions dynamics.

  • Activities related
  • Projects related
  • News related
  • Publications related

see all

End of main content