The book under review is a Ph.D. thesis defended at the Department of Economic History at Lund University in Sweden. The main purpose of this book is to examine long-term growth and business cycles in the Swedish manufacturing industry during the postwar period. As is argued in the introduction to the thesis, this period is of particular interest from the perspective of economic history since there are several economic regimes with falling and rising productivity growth including the productivity slowdown in the early 1970s. The main strength of the book is that the author demonstrates how new advances in time series analysis and cointegration theory can be used in order to learn more about the sources of technological advances and, thus, long-term economic growth.
After a short introduction, the first main chapter of the book provides a nontechnical account of endogenous growth models. This discussion serves as a background to the empirical analysis, in particular the distinction between different types of human capital and learning by investing, which is at the heart of the empirical modeling and testing in later chapters.
Chapter 3 presents the main data set and estimates of total factor productivity on a sectoral basis. The chapter starts with a detailed description of growth accounting methods including the familiar Solow growth accounting formula, which is used to compute total factor productivity. The data consists of annual output data at fixed prices for three types of industries, labor-intensive, capital-intensive, and knowledge-intensive industries. These output series are constructed using underlying two-digit branch level manufacturing statistics covering the period 1952–2001. Using these output series and corresponding data on the capital stock and working hours in the three sectors, it is straightforward to apply the Solow growth accounting formula to compute the total factor productivity (or the so-called Solow residual) which will be at the center of the empirical analysis in the remainder of the book.
Chapter 4 serves two purposes. First, it introduces the cointegrated VAR model in a nontechnical way and, second, it provides initial estimates of the main empirical models. The description of the cointegrated VAR model is not without imperfections and will be difficult to understand for readers unfamiliar with these models. My suggestion is that a reader should skip this discussion (and instead consult either a textbook in econometrics or the references given in this chapter) and move directly to section 4.3, where the first set of empirical results is discussed.
In section 4.4, the main empirical models are specified. Each sectoral model comprises five variables, total factor productivity computed in chapter 3, investments, the share of employees in organization, administration and sales, the share of employees in research, and export. It is stated in the text that all variables are in natural logarithms, but my guess is that only investment and export are in natural logs since the other three series are expressed as percentages (the data set is provided in an appendix). This raises a question about the assumptions made about total factor productivity. It seems natural to assume that the growth rate of total factor productivity cannot increase (or fall) indefinitely as it would if we assume non-stationarity. But for the empirical analysis to make sense and be interesting, we have to assume that technology growth is non-stationary. Unfortunately, these issues are not discussed and it is not clear, at least to this reader, how total factor productivity enters the models, i.e., if levels or growth rates are being used. I would have liked to see a thorough discussion of these issues combined with empirical tests of stationarity and exclusion from the steady state relations.
Having determined the cointegration rank (the number of steady state relations) of the models and estimated the unrestricted steady state relations, the author then turns to the economic hypotheses to be tested in chapter 5. The main novelty in this chapter is that the author shows how modern time series analysis can be used to analyze the sources of total factor productivity. However, I do think that more results could have been extracted from the data (for example, by exploring causal directions between physical and human capital on the one hand and total factor productivity on the other) and that some attempts should have been done in order to more clearly distinguish between the role played by different forms of human capital accumulation. As in the previous chapter, the arguments are not very clear when the results are interpreted. One potential problem is that estimated steady state relations are very similar (as is also mentioned in the text), making it very difficult to distinguish between the economic hypotheses. The author, therefore, also relies on economic reasoning when rejecting or not rejecting hypotheses. Unfortunately, this seems to introduce some inconsistencies between the economic arguments used to reject certain estimated steady state relations reported in table 5.1 and the jointly identified steady state relations presented in table 5.2.
The focus of the book changes somewhat in chapter 6, which starts out with a discussion about the real business cycle literature and its relation to endogenous growth. The focus now is on the relationship between total factor productivity, working hours and output, and the question of whether an increase in total factor productivity increases or decreases working hours. The author now specifies a stationary VAR model in order to analyze how productivity shocks affect working hours. The documentation of this analysis is not optimal, the VAR models are not specified and no test results other than impulse response functions are reported. This makes it very difficult to evaluate the results. The results indicate a positive relationship between technology shocks and working hours (and output). Further empirical testing using the steady state relations identified in chapter 5 seems to support this finding.
Chapter 7 is, in my opinion, the best chapter in the thesis. Here the author discusses three phases of economic growth in Swedish manufacturing. It is a well-written chapter and provides new knowledge about the particulars during the postwar period. The main result is that stable economic condition is a prerequisite for high economic growth and for stable and homogeneous relations between total factor productivity and the sources of productivity growth. Although this is not a new result, the empirical analysis suggests that the productivity slowdown, to at least some extent, can be explained by the distortions in the opportunity costs of investments in both physical and human capital introduced by the Swedish model during the 1960s and 1970s. The author also documents several interesting differences among the three sectors and provides a detailed discussion and interpretation of the empirical findings.
This book is not one I can recommend for readers interested in the sources of economic growth. The empirical analysis is not without imperfections, the models used are not fully documented, and there are a number of potential problems with the econometric analysis that makes it difficult to interpret the empirical results. This is unfortunate since chapter 7, the descriptive discussion about the phases of Swedish economic growth, is very interesting and should be essential reading for all economists interested in understanding the productivity slowdown in the Swedish economy during the 1970s.
