IT innovation is allowing enterprises to find new ways to harness the power of information assets for decision making. This thesis presents three econometric method applications to marketing and management decisions. The first chapter empirically studies retail network product assortment decisions under uncertain underlying demand parameters using structural estimation. I use detailed data from a beverage vending machine network in Tokyo and find that agents increase the expected total revenue of the network by 19.6% than the baseline, where 12.3% is attributable to learning from the sales data, and 7.3% is attributable to agents’ informative initial belief. However, it is below the revenue when the demand parameters are known, which is 45.5% higher than the baseline. Furthermore, if the principal company could precisely process the sales data, the expected total revenue could be 39.6% higher even if the initial beliefs are no more informative than the rational expectation. The last observation indicates that there are some costs for the principal associated with the development and utilisation of sales data processing capabilities. The second chapter studies the causal effects of product recommendation by conducting a field experiment using many vending machines in railway stations that programmatically offer recommendations for consumers after recognising their characteristics via a built-in camera. We study the effects of recommending popular products and unpopular products, and ask how the effects differ across times of day and consumer characteristics. We find that both popular and unpopular product recommendations increase vending machine sales and choice probability of recommended products. But unpopular product recommendations cause opposite effects in the morning. The negative effects are mainly from male customers in crowded vending machines. We attribute the decrease in morning vending machine sales to the congestion created by recommendations. We conjecture that the negative effect on choice probabilities in the morning is because of social pressure from the surrounding consumers. In the third chapter, I derive a necessary condition for stochastic rationalisability by a set of utility functions with a unique maximiser, which I name the strong axiom of revealed stochastic preference (SARSP). I propose a test of rationality based on the SARSP that allows for any type of heterogeneity. The test can be implemented at low computational cost. Monte Carlo simulation shows that the test has an empirical size below the nominal level and relatively strong power.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:658144 |
| Date | January 2015 |
| Creators | Kawaguchi, Kohei |
| Publisher | London School of Economics and Political Science (University of London) |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://etheses.lse.ac.uk/3109/ |
Page generated in 0.0013 seconds