Industrial Organization / Applied Microeconomics
Computational Economics
Economic Development
Economics of Technical Change

Industrial Organization
Microeconomics
Applied Econometrics
Economic Development

(Oral) October 1999, Industrial Organization and Development Economics
(Written) May 1998, Microeconomic and Macroeconomic Theory

Empirical Estimates of The Diffusion and Valuation of New Information Technologies

Professor Patrick Bayer
Professor Hanming Fang
Professor Gustav Ranis

May 2004

M. Phil., Economics, Yale University, December 2000
M.A., Economics, Yale University, December 1998

M.A., Economics, Jawaharlal Nehru University, New Delhi, India, May 1997
Specialization: International Trade and Development (graduated first in class)

B.Sc., Economics, Presidency College, Calcutta University, Calcutta, India, May 1995 (graduated third in University, high first class)

Yale University Graduate Student Fellowship, 1997-2001
Yale University Dissertation Fellowship, Fall 2001
Ford Foundation Scholarship, 1995-1997

Introductory Microeconomics (Spring 2000 and 2002, also Head TA Spring 2001)
Intermediate Microeconomics (Fall 1999)
Introductory Macroeconomics (Spring 2003, Fall 2003)
Economic Development (Fall 2000)

Research Assistant to Professor Gustav Ranis, Yale University, 1998-2000
Research Assistant to Professor Stephen Morris, Yale University, Summer 2001

Summer Internship: National Center for Applied Economic Research, New Delhi India, 1996

Estimating Diffusion Models Using Repeated Cross-sections: Quantifying the Digital Divide. (job market paper)

Estimating Neighborhood Effects in Technology Outcomes: An Ethnicity Based Approach. (in progress)

Estimating the Impact of Regulation on Demand for Home Internet Services.

Fiscal Decentralization and Human Development: Some Evidence from Argentina (econometrics paper requirement). [A different version of the collaboration work published below.]

With Nadir Habibi, Cindy Huang, Diego Miranda, Victoria Murillo, Gustav Ranis, and Frances Stewart, "Decentralization and Human Development in Argentina," Journal of Human Development, Vol. 4, No. 1, 2003

In my dissertation, a collection of three essays, I study the diffusion of the personal computer and the internet across households in the United States. Existing studies in this context are of a descriptive nature and suffer from serious methodological problems. They are therefore unable to investigate relevant policy questions, such as the predicted future dimension of the so-called digital divide and the dynamic impact of government programs initiated to bridge this divide. Using household data I find that the digital divide exists, i.e. substantial differences in access rates exist across different groups in the population. I am also able to predict that (ceteris paribus) this divide will not close in the near future, as some economists have argued. I forecast the exact magnitude of the divide across various dimensions such as income, education, race, etc. Second, I find strong evidence in favor of social learning or network effects that are commonly cited as justification for government interventions. Third, I estimate household level price elasticity for Internet access and find it to be negligible. This implies that Internet service at home should not be implicitly subsidized as is being currently done by the Internet Tax Freedom Act of 1998.

Duration models have been widely used to study diffusion of new technologies, since they allow for consumer heterogeneity in valuations. I use Current Population Survey (CPS) data which is provided in the form of repeated cross-sections and where actual adoption times (durations) are not observed. However, I show that a broad class of survival / duration models can be estimated using discrete data with extreme censoring (repeated cross-sections). Using Monte Carlo evidence I investigate the small sample properties of the maximum likelihood estimators in this context, and find them to be consistent and efficient. This finding should have broad applicability since it opens up new potential sources of data for such studies. I derive standard duration models from a forward-looking model of consumer behavior. I subsequently investigate the issue of the digital divide, which has been at the center of much heated debate in recent times with the government spending substantial amounts of money in order to bridge this divide. For example, the E-rate program uses revenue from long-distance phone calls to subsidize access to computers and the internet for schools, hospitals and libraries in under-privileged areas. Hausman (1998) finds that this program currently costs about four billion dollars to finance annually. Using estimated hazard rates for poorer targeted households I find that the program did indeed achieve its goal to an impressive extent.

Apart from consumer heterogeneity other models in the literature (epidemic models) have stressed social learning and/or network effects as alternative reasons driving the diffusion process for new technologies. The second essay in this volume investigates the existence of such phenomena in the context of new information technologies. Such phenomena have strong policy implications, for example information can be subsidized to obtain a social multiplier effect. Goolsbee and Klenow (2001) studied the same issue; however I show that their study suffers from serious methodological shortcomings that make their findings suspect. The main problem that arises in this context is that of defining and estimating relevant social networks. I utilize a more robust technique used by Bertrand et al (2001), and use ethnic origin of household members to define social networks. I find strong evidence in support of learning in social networks and/or network effects. A potentially serious issue that arises in this context is that of negative selection bias. I find that such a bias does exist in this case but in spite of it estimated social learning effects are highly significant.

The Internet Tax Freedom Act of 1998 (subsequently renewed) imposed a moratorium on the states’ ability to impose taxes on Internet access. The third essay in this volume estimates the price-elasticity of Internet access. I estimate a state level demand model as well as various discrete choice models at the individual household level. The crucial problem in this context (as in other studies of this nature) is that Internet access is a differentiated good, and prices and characteristics of all goods in the consumer choice set are not available to the investigator. This makes estimating the price elasticity of demand an onerous task. I follow Deaton (1987) in utilizing the geographic variation in prices to estimate this elasticity. I also obtain dynamic estimates of consumer surplus and deadweight loss from taxation (subsidies) in this context using various diffusion models. I find that price elasticity of access is negligible and therefore the impact of this Act is marginal.