This issue of the STI Review focuses on the new rationale and approaches in technology and innovation policy.

The volume examines the state-of-the-art of productivity and efficiency analysis. It brings together a selection of the best papers from the 10th North American Productivity Workshop. By analyzing world-wide perspectives on challenges that local economies and institutions may face when changes in productivity are observed, readers can quickly assess the impact of productivity measurement, productivity growth, dynamics of productivity change, measures of labor productivity, measures of technical efficiency in different sectors, frontier analysis, measures of performance, industry instability and spillover effects. The contributions in this volume focus on the theory and application of economics, econometrics, statistics, management science and operational research related to problems in the areas of productivity and efficiency measurement. Popular techniques and methodologies including stochastic frontier analysis and data envelopment analysis are represented. Chapters also cover broader issues related to measuring, understanding, incentivizing and improving the productivity and performance of firms, public services, and industries.

7 take advantage of the panel structure of their data to control for possible errors of specifica tion in their models. It is interesting to note that the econometric and DEA methods may be closer than some of their respective advocates seem to believe. Several of the studies show that the former as well as the latter can be effectively used to assess the relative effi ciency of groups of firms or individual firms, and one of them explicitly compare results arising from both (Fecher et al.). Econometric techniques can also be nonparametric and applied to estimating cost or production frontiers (and not only "average" functions), while ultimately DEA should be amenable to statistical inference. Perhaps the most valuable feature of all the analyses is their care and ingenuity in putting together the data, measuring variables, and pulling out relevant information. Many of them are not content with an overall output measure, but endeavor to manage with less aggregated measures. Nearly all also include in the estimated models a number of auxiliary variables intended to control for specific attributes of outputs, inputs, or production techniques, and other characteristics of firms.