Economic development
Theoretical econometrics
Applied econometrics

Econometrics
Economic Development
International Economics

May 1999 (Oral) Economic Development with Distinction
October 1998 (Oral) Econometrics with Distinction
May 1998 (Written) Microeconomics and Macroeconomic Theory with Distinction

Dynamic Pseudo-Panel Theory and analysis of Consumption in Taiwan and Mexico

Professor Peter C.B. Phillips
Professor Christopher Udry
Professor Donald W.K. Andrews
Professor T. Paul Schultz

May 2001

Ph.D., Yale University, expected May 2001
M.Phil., Yale University, May 2000
M.A., Yale University, May 1998
Bachelor of Commerce (Hons.) (Conjoint), The University of Auckland, New Zealand, September 1997
B.A. (Conjoint) in Statistics, The University of Auckland, New Zealand, September 1997
GPA: Auckland 9.0/9.0, Yale 4.0/4.0

Yale University
Cowles Foundation Prize 1998, 2000
Sasakawa Fellowship 1999
Yale University Fellowship 1997-2001

University of Auckland
Fowlds Memorial Prize, 1996.
(Most distinguished student at Honours or Masters level in the Commerce faculty)
Senior Scholarship in Economics (1995) and Statistics (1996)
Various Prizes in Accounting, Economics and Finance 1993-1995

Additional Awards of Significance
Top Overall Student in University Bursary/Entrance Scholarship Examinations 1992
(This is the examination taken by all students at the Grade 12 level in New Zealand in order to seek entrance to university.)
New Zealand Team, Asian Pacific Mathematical Olympiad 1991-92.

Instructor, Department of Economics, Yale University.
Time Series II (Graduate Course), Spring 2000

Teaching Assistant, Department of Economics, Yale University
Econometrics III (Graduate Course), assigned for Spring 2001
Econometrics I (Graduate Course), Fall 2000

Tutor/Teaching Assistant, Department of Economics, The University of Auckland.
Undergraduate Advanced Econometrics, First Semester 1996, 1997
(Recipient of University of Auckland Economics Department Tutor’s Award for Excellence in Tutoring based on Student Evaluations of Teaching in this class)
Undergraduate Financial Economics, Second Semester, 1996
Undergraduate Intermediate Microeconomics, First Semester, 1997

Statistics Demonstrator, Department of Mathematics and Statistics, University of Auckland 1995-96

Summer Internship, The World Bank, Washington D.C, 1999 – FINCR Division
Designed and estimated an econometric model of the determinants of sovereign default to the IBRD using dynamic panel logit analysis. Model is now used in country risk analysis.

Research Assistant for Dr. John Small, University of Auckland, 1996
This research consisted of comparing VAR, BVAR and Kalman Filter forecasts for a small-scale regional model of the Auckland economy.

Summer Research Analyst, BellSouth Network Centre, University of Auckland, 1995-1996.

• "Consumption Growth in a Booming Economy: Taiwan 1976-96", August 2000 [job market paper]
• "Asymptotic Theory for Heterogeneous Dynamic Pseudo-Panels", June 2000 [job market paper]
• "Estimation of Unequally-Spaced Dynamic Panels and Pseudo-Panels", August 2000
• "Household Consumption in a Volatile Economy: Mexico 1984-96" (in progress)
• "The Impact of Capital Controls on Growth Convergence", November 1999
• "An Econometric Analysis of IBRD Creditworthiness", World Bank Policy Research Working Paper (forthcoming, 2000)
• "Econometric Cost Structure Estimates for Cellular Telephony in the United States" (with John P. Small), Journal of Regulatory Economics 12(2): 147-157, 1997.

This dissertation studies household consumption behaviour in developing countries, using dynamic pseudo-panels constructed from repeated cross-sectional surveys. New econometric theory for the estimation of dynamic pseudo-panel models is developed and used to examine the impact of demographic change, individual behaviour, and the macroeconomic environment on consumption over time. Two developing countries are studied: Taiwan, which experienced rapid positive real GNP per capita growth at an average rate of 6.5% per annum over the period 1976-96; and Mexico, which averaged only 0.23% annual growth between 1984 and 1996 amidst volatile macroeconomic conditions and recurring crises. Taiwanese consumption growth is found to arise primarily from high levels of prudence, whereas other variants of consumption theory are shown to be incapable of explaining the Taiwanese data. Cohort-level models of consumption are estimated for each country, and are employed to ascertain the extent to which modern consumption theory explains differences in consumption and saving behaviour between the two. The results can be used to make inferences about the impact of an aging population on savings rates and consumption inequality in the world as a whole.

The first chapter, Asymptotic Theory for Heterogeneous Dynamic Pseudo-Panels, investigates the estimation of dynamic pseudo-panel models with inter-cohort parameter heterogeneity. A multidimensional limit theory is derived as both the cross-sectional and temporal dimensions of the data are allowed to pass to infinity, allowing for both stationary and non-stationary processes. This set-up contrasts with the existing literature, which analyzes unidimensional asymptotics under assumptions of stationarity and parameter homogeneity. Least squares (OLS) and instrumental variables (IV) estimators are considered. As the cross-sectional dimension passes to infinity for a fixed number of time periods, both OLS and IV are shown to be consistent and have asymptotically normal distributions, for which consistent covariance matrix estimators are derived. However, OLS is found to be inconsistent if the time dimension grows proportionally with, or faster than, the cross-sectional dimension. The IV estimator is found to be consistent under weaker conditions, but displays greater variability than OLS in small cross-sectional samples. In conjunction with Monte Carlo simulations on the finite-sample performance of the estimators in these two dimensions, the results lead to practical recommendations for the implementation of dynamic pseudo-panel methods.

Consumption and income have both grown rapidly in Taiwan over the past forty years, with younger birth cohorts experiencing faster growth. The long upward trend in consumption, coupled with the high savings rates of Taiwanese consumers, presents a strong challenge to the consumption smoothing predictions of the Permanent Income Hypothesis. The second chapter, Consumption Growth in a Booming Economy: Taiwan 1976-96, investigates the extent to which modern consumption theory can account for this trend. The methods developed in Chapter One are applied to annual household survey data from 1976-96, in order to evaluate the impact on consumption of migration, mortality, household composition, liquidity constraints, unanticipated aggregate shocks, hyperbolic discounting, habit formation and precautionary saving. Individual earnings are demonstrated to contain a strong stochastic trend for all cohorts and over ninety percent of each cohort is observed to save. This provides strong evidence against the presence of liquidity constraints being the main determinant of consumption growth, and it is argued that an active informal financial sector served to alleviate formal credit restrictions. Suggestions that consumers were repeatedly surprised by the high levels of income growth experienced are refuted by over twenty years of previous high growth experience, repeated official forecasts of continued high growth, evidence from roscas, and surveyed individual level expectations. Taiwanese consumption growth is instead found to primarily result from high levels of prudence, with the faster consumption growth of younger cohorts attributed to their greater participation in industries with higher earnings risk. The strong precautionary motive has important implications for the effects of economic growth and population aging on aggregate saving and inequality.

Mexican household surveys on consumption were taken in 1984, 1989, 1992, 1994 and 1996. This irregular-spacing, a common feature of many developing nation surveys, prevents the direct application of the methods of Chapter One. The third chapter, Estimation of Unequally-Spaced Dynamic Panels and Pseudo-Panels, establishes asymptotic results for the estimation of dynamic pseudo-panel models with such data. Nonlinear least squares, minimum distance, and one-step estimators are used to impose the nonlinear parameter restrictions which arise in dynamic models over unequally-spaced periods. A Monte Carlo simulation study corroborates the asymptotic results. These methods are also shown to be applicable to the estimation of dynamic models with irregularly spaced genuine panel data.

Mexico's economy experienced recurring economic crises in 1976, 1982, 1986 and 1994-95. The fourth chapter, Household Consumption in a Volatile Economy: Mexico 1984-96, is presently in progress and researches consumption behaviour in Mexico using the methods developed in the third chapter. Cohort-level models are estimated to examine the responsiveness of Mexican consumption to the sizable income shocks experienced. A comparison with the results for Taiwan reveals the extent to which modern consumption theory explains inter-country differences in savings rates and consumption behaviour.

The four chapters add to our understanding of the determinants of consumption and saving in developing nations. The econometric techniques developed for this purpose should be of more general interest, given the increasing availability of household surveys in several fields of economics.