Cycles, Growth and Structural Change

Lionello F Punzo

• Publisher: Routledge
• Year: 2003
• ISBN:
  9781134530007
• Edition: 1
• Topic: Business
• Subtopic: Business General
• Index: Business

This volume gathers together key new contributions on the subject of the relationship, both empirical and theoretical, between economic oscillations, growth and structural change. Employing a sophisticated level of mathematical modelling, the collection contains articles from, amongst others, William Baumol, Katsuhito Iwai and William Brock.

Part I
Facts and interpretations of growth and fluctuations

1 Economic cycles since 1870

Solomos Solomou

1.1 Introduction

Current research on business cycles has focused on the post-war period. The aim has been to derive the main ‘stylised facts’ of fluctuations. The approach here is very different. The emphasis is on the need for an historical perspective to business cycles. History inevitably provides us with a broader empirical basis that will allow us to formulate more general theoretical and empirical questions. It is argued that to focus on explaining the stylised facts of any one historical period to gain a general insight on economic fluctuations will lead to serious errors of interpretation.

An historical perspective is important not because history repeats itself but because history illustrates the evolutionary nature of business cycle behaviour and gives us an understanding of the factors that generate change. In economic systems that entail behavioural, institutional, structural and policy changes we can safely predict that the nature of business cycles will not be stable over time. The case for an historical perspective is founded on two important empirical features. First, the process of economic growth of modern economies has undergone massive structural change in the past two hundred years. One important aspect of structural change is the reallocation of production across sectors. For example, with industrialisation we would expect a change in the role of agricultural cycles in macroeconomic fluctuations. Given that these type of changes take place over a very low frequency, we need to observe business cycles over the long-term to gain insights on the resulting cyclical effects of structural change.

Second, policy regime changes have an observable effect on cyclical fluctuations. Business cycles during the rules-driven policy framework of the pre-1913 gold standard epoch were very different to those of the inter-war period. Similarly, the fluctuations of the Bretton Woods era were very different to those observed in the post-1973 era. Major policy regime changes have been few in number, once again, implying the need for a long-run perspective.

These observations raise serious doubts about the empirical relevance of the so-called ‘stylised facts’ approach to business cycles. This approach assumes that regularities exist over time and across countries (Lucas, 1981). However, many of these empirical regularities are usually derived from studies of very short time periods (often the post-1960 era) and a small selection of countries (usually the UK and the USA).

Even if we focus on a very limited set of empirical features we observe important changes. The average cyclical duration has changed over time. During 1870–1913 a number of variables (including aggregate investment, agricultural production and construction sector output) fluctuated with a long swing duration averaging about 20 years. During the inter-war period shorter fluctuations were observed. During the post-war ‘golden age’ the average cyclical period fell to 5 years. During the post-1973 era the average duration has once again lengthened, averaging approximately 10 years since the late 1970s. Cyclical amplitudes have also varied significantly over time. Low macroeconomic volatility during 1870–1913 gave way to high amplitude fluctuations during 1919–38; the stability of the post-war ‘golden age’ has been followed by the relatively more volatile post-1973 era. Patterns of co-movement of key variables have not been stable over time. Much of post-war research on business cycles has noted that prices and output have fluctuated contra-cyclically and has proceeded to explain this feature in terms of ‘real’ business cycle theory (Danthine and Donaldson, 1993). However, during the classical gold standard the relationship was not stable and during the inter-war period price and output fluctuations moved in a pro-cyclical manner (Cooley and Ohanian, 1991). Assuming universal stylised facts is not a realistic way of understanding business cycles.

This chapter surveys business cycle features across three historical epochs to illustrate some of the themes discussed in this introduction. Section 1 considers the pre-1913 epoch. Section 2 considers the changes observed in the inter-war period. Section 3 considers the post-war epoch.

1.2 1870–1913

The main institutional aspect of the period 1870–1913 is the gold standard. Between 1879–1913 Britain, France, Germany and America pegged their currencies to gold at a fixed rate. The relationship between this institutional-policy framework and economic fluctuations needs careful consideration. In particular, it is important to evaluate whether the rules-based policy framework modulated aspects such as the amplitude of fluctuations. The main structural aspect of the pre-1913 period is the large size of the primary sectors (such as agriculture and mining), as a percentage of the labour force and gross domestic product. Such a structure implies that supply-side shocks (such as weather shocks) are likely to have significant effects on sectoral and macroeconomic fluctuations. In terms of the world economy, the period is one of integration in trade, capital and labour flows between the industrial countries of Europe and the primary producing economies of the world. Such international linkages were important to determining the adjustment path to shocks in the industrial economies.

1.2.1 Describing cycles

Three different cycles have been identified by economic historians of the period: the Juglar trade cycle, the Kuznets swing and the Kondratieff wave. The Juglar cycle has an average period of seven to nine years. A number of studies have argued that this is the dominant cycle over this period (Aldcroft and Fearon, 1972; Lewis, 1978; Crafts, et al., 1989). For example, industrial production in the UK fluctuated with peaks in 1873, 1882, 1889, 1899, 1907 and 1913. The average peak to peak cycle is eight years. However, a number of data and conceptual problems should be noted before we accept this perspective to pre-1913 cycles. We should note a number of serious limitations in the historical industrial production series. For example the series for Britain (produced by Arthur Lewis) has been constructed using indicator variables and methods of extrapolation and interpolation. Foralarge proportion of the industrial production index Lewis has actually imposed a cycle of nine years on a priori grounds (Lewis, 1978; Solomou, 1994). A Juglar cycle is imposed on iron and steel products, commercial building, clothing, printing and chemicals, which account for over 28 per cent of Lewis’ total industrial production index and 35 per cent of the manufacturing and construction index. Thus, the Lewis industrial production index provides only limited independent information for the importance of a Juglar cycle during this period. If we only consider the path of those industries where the cycle is not imposeda priori, the evidence fora dominant Juglar cycle is very weak. The construction sector sees long-term fluctuations of twenty years (Thomas, 1973). Coal production showed variations in trend without any discernible regular short cycle (Catao and Solomou, 1993). Investment was dominated by a long cycle of 23 years (Cairncross, 1953). Agricultural production was dominated by a long cycle of 20 years (Solomou, 1994). This evidence gives some support to Hicks (1982), who noted that during 1875–1914 business cycles became far more irregular in duration.

The Kuznets swing refers to a variation in economic growth that is longer than the Juglar trade cycle. The swings observed are variations either in levels or in rates of growth. The actual length of the swings found varies with different studies, but something between 14 to 22 years is representative. This type of fluctuation is most clear in describing the path of capital formation in the leading economies. British and American investment fluctuated along a 20–year cycle in the level of investment whilst French and German investment fluctuated along a long cycle in the rate of growth (Solomou, 1987). Irregular Kuznets swings can also be observed in the level and growth of agricultural production, construction output, migration, the sectoral terms of trade and trade balances (Solomou, 1987; Lewis, 1978; Thomas, 1973; Cairncross, 1953; Rowthorn and Solomou, 1991).

The Kondratieff wave is a cycle of prices and output with an approximate period of fifty to sixty years. In some of the earlier literature the Kondratieff wave has been used as a framework for understanding epochs such as the ‘Great Depression’ of 1873–96 and the Belle-epoch inflation of 1899–1913 (Kondratieff, 1935; Schumpeter, 1939). The empirical evidence suggests that we need to tread with care in the evaluation of Kondratieff waves. Most macroeconomic variables (such as GDP, investment and industrial production) have not followed a longwave growth pattern (Solomou, 1987). However, long cycle adjustments in prices have been noted, particularly over the period 1873–1913 (Lewis, 1978; Rostow and Kennedy, 1979).

Given the historical discussion of multiple cycles it is useful to describe the relevance ofthis perspective using modern time-series methods. Solomou (1998) considers the empirical relevance of multiple cycle models by employing the Kalman filter to describe cycles in GDP for Britain, France, Germany and America. As can be observed in Figure 1.1a–1.1d the cyclical path of all these economies is depicted as the sum of short and long cycles. Although we have only presented the decomposition for GDP, multiple cycles are relevant at different levels of aggregation. This perspective provides an alternative to Hicks’ (1982) description of irregular fluctuations. One aspect of irregularity is that the epoch was influenced by cycles of different average periods. Accepting the idea of multiple cycles as reality suggests one way of capturing some of the observed irregularity.

1.2.2 Rules-driven policy framework and business cycles

The amplitude of business cycles was significantly lower during the gold standard period relative to the inter-war era (Sheffrin, 1988). The literature has attributed this to the rules-driven policy framework of the gold standard (Crafts and Mills, 1992). Such an idea offers only a partial explanation of this phenomenon. To understand the relatively low macroeconomic volatility of the period we need to consider the cyclical adjustment mechanisms operating over this period. Three adjustment mechanisms were of central importance. First, international labour mobility was exceptionally high. Second, capital mobility from Britain, France and Germany to the newly in...

Table of contents

• Cover Page
• Routledge Siena Studies in Political Economy
• Title Page
• Copyright Page
• Figures
• Tables
• Contributors
• Preface
• Part I: Facts and interpretations of growth and fluctuations
• Part II: The macroeconomy and its dynamics
• Part III: Dynamics by interaction
• Part IV: Challenges for quantitative methodologies