Technology Differences over Space and Time (CREI Lectures in Macroeconomics) - Hardcover

Francesco Caselli

Hardcover

ISBN 10: 0691146020 ISBN 13: 9780691146027

Publisher: Princeton University Press, 2016

Synopsis

Technology Differences over Space and Time looks at how countries use their productive resources--such as workers, skills, equipment and structures, and natural resources. Francesco Caselli develops methods to assess the efficiency with which productive inputs are used, and how these efficiencies vary across countries and over time. Caselli finds that richer countries use skilled workers relatively more efficiently than unskilled workers, and equipment and structures relatively more efficiently than natural resources. They also are relatively more efficient users of labor than of capital. Technological change tends to make countries particularly efficient at using skills and less efficient at using capital. Technical change also favors experienced workers. In order to interpret and understand these findings, Caselli presents a theory of technology choice. In this theory, firms pick technologies that make the most efficient use of the most abundant production factors when these factors are good substitutes for the less abundant factors. Firms pick technologies that make the most of less abundant factors when other suitable factors are not available for substitution. For example, rich countries, where skilled workers are abundant, use skilled workers efficiently, as these are good substitutes for unskilled workers. This flexible framework can be applied to other pairs of inputs, over time, and across countries. Technology Differences over Space and Time has significant implications not only for the theoretical understanding of development and technological innovation, but also for government formulation of industrial policy and multinationals making decisions about what to invest in and where to make those investments.

About this title

About the Author

Francesco Caselli is the Norman Sosnow Professor of Economics at the London School of Economics, and a fellow of the British Academy.

From the Back Cover

"Francesco Caselli is one of the world's leading experts on economic growth. In Technology Differences over Space and Time, he provides a unified treatment of factor-biased technical change, brilliantly synthesizing earlier research and pushing the frontier forward. Among many insights, he shows how a common framework can help us understand both why skilled workers do so well over time and how this is particularly true in rich countries. Highly recommended!"--Charles I. Jones, Stanford Graduate School of Business

"Caselli shatters the view that technology can flow effortlessly across countries. He shows that the technology used in rich countries favors skilled workers, though not equipment and structures, whereas the technology used in developing countries surprisingly favors unskilled workers. Anybody who wants to understand how technology evolves with development--and how it shapes who gains the most from growth--will want to read this book."--Pete Klenow, Stanford University

"This book is obligatory reading for those interested in the state of the art of research on biased technology across countries and over time."--Antonio Ciccone, University of Mannheim

Excerpt. © Reprinted by permission. All rights reserved.

Technology Differences Over Space and Time

By Francesco Caselli

PRINCETON UNIVERSITY PRESS

Copyright © 2017 Princeton University Press
All rights reserved.

ISBN: 978-0-691-14602-7

Contents

Preface, ix, 
CHAPTER 1 Introduction and Preliminaries, 1, 
Part I Technology Differences across Space, 
CHAPTER 2 Skilled and Unskilled Labor, 17, 
CHAPTER 3 Natural and Reproducible Capital, 40, 
CHAPTER 4 Capital and Labor, 48, 
Part II Interpreting Technology Differences, 
CHAPTER 5 An Endogenous Technology Framework, 67, 
Part III Technology Differences over Time, 
CHAPTER 6 Skilled Labor, Unskilled Labor, and Experience over Time, 83, 
CHAPTER 7 Skills and Capital over Time and across Countries, 95, 
CHAPTER 8 Conclusions, 102, 
Appendix A. Proofs and Calculations, 105, 
Appendix B. A New Data Set on Mincerian Returns (with Jacopo Ponticelli and Federico Rossi), 108, 
References, 119, 
Index, 125, 

CHAPTER 1

INTRODUCTION AND PRELIMINARIES

1.1 A Nontechnical Overview

Economies rely on a rich set of different inputs to produce goods and services. The composition of the pool of productive resources varies dramatically across countries. Some countries have a lot of workers with little education and few highly educated workers, while others have an abundance of workers with many years of schooling. Some countries have a copious supply of natural resources, relative to their stock of productive machinery and structures, while others are resource-poor but have plenty of equipment. Some countries have a lot of people and little capital; others have high capital-labor ratios.

Do these big differences in the composition of the bundle of factors of production lead to correspondingly large differences in the way productive activities are organized? Do productive units in a country tailor their mode of production, choice of technology, and management practices to the particular patterns of scarcity and abundance of inputs in their country? And if so, how? Do they choose production methods which make the most of the abundant factors — giving up, so to speak, on those that are relatively limited in their countries? Or do they rather focus on maximizing the contribution of the inputs in limited supply, to make up for their scarcity?

Similar observations, leading to similar questions, can be made for a given country over time. In the space of years and decades, the relative supply of skills, equipment and structure, natural resources, and even workers belonging to different demographic groups can change substantially. Do modes of production change and adapt in response to these changes? If so, does the adaptation compound the changes, by increasing the reliance of production units on the inputs that are becoming more abundant; or does it lean against the wind, by strengthening the ability to contribute to output of the factors of production that are becoming relatively scarce?

This book reports on my attempts to provide some answers to these questions. It shows that the mode of production — or the "production technology" in the words I use throughout the book — does vary systematically across countries, depending on their endowments of different factors of production. The production technology also changes over time, as factor supplies change. As to whether this adaptation favors the abundant or the scarce inputs, the answer turns out to be "it depends." When the factors becoming scarce are not that different, in terms of the role they play in the production process, from those becoming abundant, the production technology adapts to maximize the impact of the abundant factors. On the other hand, when the scarce factors are special and are difficult to replace using the abundant factors, the production technology mutates to bolster the contribution of the scarce inputs.

Admittedly, none of the above sounds particularly surprising. What is perhaps more surprising is that these patterns of technology adaptation are completely ignored by the vast majority of economists' thinking on how the organization of production varies across countries and (to a slightly lesser extent) over time. The dominant view, instead, is that some countries just have "better" technology than others, regardless of differences in factor endowment. Therefore, regardless of factor endowments, countries with "poor" technology should strive to copy as best they can the modes of organization they observe in countries with "good" technology. Similarly, over time technology mostly just gets "better," lifting the productivity of all factors equally. This pervasive, and rather boring, view of technology is incompatible with the evidence presented in this book: technology differences over space and time are much more interesting than most economists make them out to be!

1.2 A Slightly More Technical Overview

Economists characterize the relationship between a country's productive resources and its GDP by means of the aggregate production function. The aggregate production function can be used to answer two types of questions: (i) If country A has x% more of a given input (say, labor) than country B, by how much will country A's GDP exceed country B's (everything else being constant)? (ii) If country A experiences an x% increase in a given input between years t and t + 1, by how much will its GDP increase between the two years?

In empirical applications, economists have long noticed that production functions are not stable. Namely, the mapping from inputs onto outputs changes both across countries and over time. It is customary to refer to this instability as technology differences (across countries) and technical change (over time).

A long tradition of studies attempt to quantify the pace of technical change. This endeavor is usually refereed to as growth accounting. Growth-accounting exercises usually find technical change to be very important in driving changes in GDP over time. There is also a more recent, but now well-established, strand on quantifying the magnitude of technology differences across countries. These development accounting studies tend to find that technology differences are very important in determining differences in GDP. Both these sets of findings have profoundly influenced the way economists think about economic performance in the long run.

While these traditions have been effective at quantifying the extent of technology differences and technical change, they have arguably been less successful at characterizing their nature. In the vast majority of cross-country empirical applications technology is assumed to be factor neutral. Roughly speaking (and I will of course be more precise below) factor neutrality implies that if the efficiency with which a country uses one input is x% higher than the same input's efficiency in another, then the efficiency of all other inputs is also x% higher. Discussions of growth-accounting exercises are sometimes more nuanced and more aware of the possibility of technical change that is not factor neutral. Still, the methodology delivers a single quantitative measure of the pace of technical change, and is unsuited to characterize its nature. For this reason, growth-accounting results are still almost universally interpreted as if technical change was factor neutral.

This factor-neutral representation of technology is incompatible with various facts about factor prices across countries and time. As I will explain, if that view was correct, skill premia in (skill-) poor countries would be much higher, compared to premia in (skill-) rich countries, than they are; and skill premia couldn't have grown over time the way they have. Also, the capital share in income would vary much more across countries and over time than it does in the data. Changes in the experience premium over time are also inconsistent with neutral technical change.

This book uses these and other observations to show that technology differences and technical change are factor biased: they change not only the overall efficiency with which a country exploits its bundle of productive inputs, but also the relative efficiency with which different factors contribute to production. In fact, in some cases evidence shows that as the efficiency with which one input increases, the efficiency of another decreases. Allowing factor-biased technical change helps explain, among other things, why skill premia are very similar in poor and rich countries; why they have been growing over time; why the experience premium has not declined in response to the maturing of the baby-boom generation; and why the capital share in income is fairly constant both across countries and over time.

More specifically, I show that in richer countries the efficiency with which skilled labor is used relative to unskilled labor is greater than in poorer countries; similarly, the efficiency with which reproducible capital (equipment and structure) is used relative to natural capital (mineral deposits, land, timber, etc.) is higher in rich countries; also, when comparing the efficiency of an overall bundle of labor inputs (appropriately combining skilled and unskilled labor) and an overall capital input (which combines reproducible and natural capital), rich countries use labor relatively more efficiently. Furthermore, the absolute efficiency with which physical capital is used appears to be not lower, and may even be higher, in poor countries.

Over time, I document (like others before me) an increase in the relative efficiency of skilled labor. I also find an increase in the relative efficiency of older workers relative to younger ones (holding skills constant). Finally, in an echo of the corresponding result in the cross section, the efficiency with which physical capital is used has been declining over time.

I interpret these findings by means of a simple theoretical model of endogenous technological choice. In the model, firms choose from a menu of technologies (production functions). The key consideration turns out to be the degree of substitutability among factors. When two factors of production are highly substitutable, firms choose technologies that maximize the efficiency of the cheaper factor (at the expense of the efficiency of the more expensive factor). Instead, when two factors are poor substitutes, firms choose to maximize the efficiency of the expensive factor.

To see how this framework sheds light on the empirical findings, consider skilled and unskilled labor. Rich countries have larger relative supplies of skilled labor, and hence skilled labor is relatively cheap there. Since skilled and unskilled labor are pretty good substitutes, firms in rich countries seek to make the most of skilled labor, and end up picking technologies that imply a high relative efficiency of skilled labor compared to poor countries. Rich countries also have larger relative supplies of physical capital (broadly construed to include natural and reproducible capital) compared to labor (broadly construed to account for the larger proportion of skilled workers). But labor and capital are (thought to be) poor substitutes, so in this case rich countries choose technologies that bolster the relative efficiency of capital. The other empirical patterns uncovered can be interpreted along similar lines.

The factor-neutrality approach implies a Manichean view where some countries "get it right" and others "get it wrong." Countries either make the most of their skilled and unskilled labor, reproducible and natural capital, or they fail to use any of these efficiently. One implication is that poor countries should strive to reproduce rich countries' technological choices, regardless of their factor endowments and other determinants of optimal technology choice. The nonneutrality findings in this book, and in the research on which this book is based, point to a more nuanced picture. To be sure, firms in poor countries lag far behind the technology frontier to which rich-country firms have access. But technology transfer and adoption should be selective and tailored to local conditions.

1.3 Aggregate Production Functions

The central — indeed the only — analytical tool used in this book is the aggregate production function. The aggregate production function is a mapping from a country's input quantities to a country's output, and we express it as

Yct = Fct(X1ct, X2ct, ...), (1.1)

where Yct is aggregate output in country c in year t, Xjct is the quantity of input j used in production, and Fct is the mapping in question. Note that the mapping carries subscripts c and t, indicating that the aggregate production function is country and time specific.

The empirical counterpart of output Yct is gross domestic product (GDP). More specifically, when we are concerned with cross-country comparisons, we focus on GDP at purchasing power parity (PPP GDP). PPP GDP adds up the quantities produced of all final goods and services using a common set of prices (PPPs) as weights. When making comparisons over time, constant-price series must be used.

This book is about how Fct varies across countries and over time. The factor-neutral case can be represented as

Fct(X1ct, X2ct, ...) = Act[??](X1ct, X2ct, ...). (1.2)

In this special case, the production functions F differ by, and only by, the multiplicative term A. There already is a lot of literature documenting that A contributes substantially to changes in GDP over time and very substantially to cross-country differences in GDP.

Factor neutrality is a natural first step in investigating cross-country technology differences as well as technical change, but a glance at equation (1.2) clearly shows that it is highly restrictive. The book focuses on the following conceptual generalization:

Fct(X1ct, X2ct, ...) = G(A1ct, X1ct, A2ct, X2ct, ...). (1.3)

In (1.3) the technology parameter Ajct augments factor j. A country (year) may have a relatively high value of one of the Ajcts without having a proportionally high value of another. In other words, technology differences need not be factor neutral — though neutrality is admitted as a possible special case. The book is about asking whether — and if so how — the Ajcts vary across countries and over time.

To this end, we must begin by identifying the list of relevant factors of production. I focus on four broad aggregates: unskilled labor, skilled labor, physical reproducible capital, and natural capital. The breakdown of the main factors of production into labor and capital is almost as old as economics, and the breakdown of labor into skilled and unskilled is also well established. The importance of accounting for reproducible and physical capital has recently been emphasized by Caselli and Feyrer (2007).

We must also specify a functional form for G. The book applies methods originally developed by Caselli and Coleman (2002, 2006) and Caselli (2005) which allow for identification of the Ajcts when the production function features constant elasticities of substitution (CES). Accordingly, in most of the book, I work with the following specification:

Yct = [(AKct Kct)s + (ALct Lct)s]1/s, (1.4)

Kct = [(ANct Nct)? + (AMct Mct)?]1/?, (1.5)

Lct = [(AUct Uct)? + (ASct Sct)?]1/?. (1.6)

Hence, the production process is represented by a sequence of nested CES aggregators. Beginning from the bottom, unskilled labor U and skilled labor S are combined into an aggregate labor input L with elasticity of substitution 1/(1 – ?). Similarly, natural capital N and reproducible capital M (M for machine) are combined into the aggregate K, with elasticity of substitution 1/(1 – ?). Finally, labor and capital are aggregated with elasticity 1/(1 – s) to produce output. Technology differences are captured by differences in the factor-augmenting terms AUct, ASct, ANct, AMct, AKct, and ALct, which are the object of this study.

(Continues...)

Excerpted from Technology Differences Over Space and Time by Francesco Caselli. Copyright © 2017 Princeton University Press. Excerpted by permission of PRINCETON UNIVERSITY PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.

Edition details

• Publisher: Princeton University Press
• Publication date: 2016
• Language: English
• ISBN 10: 0691146020
• ISBN 13: 9780691146027
• Binding: Hardcover
• Number of pages: 224