Building a Better Bootstrap
Estimates based on economic data have significant influence on public policy decisions made concerning a wide range of issues. For example, determination of poverty and inequality within society are vital to formulating economic policies that can affect the lives of millions. Unfortunately, estimates of these types of measures are quite unreliable when based on common statistical methods. Frequently, policies are put in place based on little more than good faith because reliable methods of evaluating the results of existing policies are not available.
The focus of Dr. Russell Davidson's work over the past two decades has been on developing new approaches to making inferences based on statistical data. Among his achievements have been substantial improvements to results derived from application of the so-called bootstrap method of estimating the probability distribution of a statistic. Working in conjunction with James MacKinnon, Dr. Davidson developed a way of greatly simplifying and shortening the computations needed to implement the bootstrap method. This allows the bootstrap to be applied in situations where the difficulty of the required computations might otherwise make it infeasible.
His program at McGill University will build on his achievements, extending the use of the bootstrap method to a wide variety of econometric applications, including inequality and poverty measurement. He will also focus on developing computational methods that make his new bootstrap techniques easy to use.
Dr. Davidson expects that his new work will have its greatest impact on everyday econometric practice. His primary goal is to discover techniques and methods that improve econometrics by increasing efficiency and ease of application. He believes that only through additional work will the bootstrap method reach its potential as a tool to influence policy decisions.