RDP 2001-01: The Decline in Australian Output Volatility 2. Literature Review

A decline in the volatility of output has been noted in a number of recent papers in a number of countries. Gruen and Stevens (2000) note the decline in Australian output volatility. Without conducting a detailed analysis they suggest that smaller shocks may provide the best explanation for the recent smoothness of output growth. Overseas studies have also focused on the reduction in output volatility. McConnell and Quiros (2000) find compelling evidence of a structural break in US output growth volatility in 1984. Decomposing output to find the source of the volatility decline, they attribute it to a change in the behaviour of durable goods production and associated inventories behaviour. Simon (2000) also documents the decline in volatility of US output growth and decomposes the sources using a structural VAR methodology. This decomposition finds that the primary source of the better performance is the reduction in the shocks hitting the economy rather than improved structural stability. Furthermore, the primary reduction has been in demand shocks and a much smoother profile for household consumption. The data show that households are now behaving much more in line with the predictions of the permanent income hypothesis and smoothing temporary income shocks much more than they did in the 1960s.

In a related area, there was a vigorous debate within the field of economic history about declines in US output volatility pre- and post-WWII. Results from this research on stabilisation are equivocal. There are many technical data questions resulting from the lack of good quality data for earlier time periods. The initial consensus in the literature was that stabilisation, in the form of reduced economic volatility, had occurred in the post-WWII period. This was formalised in DeLong and Summers (1986). This view was challenged by Christina Romer in a number of papers questioning the validity of the early data (see, for example, Romer (1986)). This contention has, itself, been challenged by Weir (1986) and Balke and Gordon (1989) among others.

Work on the duration dependence of business cycles attempts to understand stability without becoming bogged down with data issues by concentrating on less equivocal data – generally business cycle lengths. Longer expansions are seen as a closely related consequence of lower volatility, indeed it is generally the longer expansions that have brought the issue of lower volatility into the picture. Despite hopes that the analysis would not be affected to the same degree by data problems, the findings are not clear. Diebold and Rudebusch (1990) find some evidence for duration dependence in whole cycles since the turn of the century, but no evidence in post-WWII expansions. Addressing the issue of post-WWII stabilisation, Diebold and Rudebusch (1992) find that the average length of an expansion is at least 9 months longer post-WWII than pre-WWII.

Many papers confined themselves to establishing that stabilisation has occurred; fewer concerned themselves with explaining why this has occurred. In a paper analysing 19^th century US data, James (1993) finds increased volatility after the US Civil War and provides a possible reason. He finds the increased volatility was a result of the changed structure of the economy rather than the result of an increase in the size of shocks hitting the economy. On the basis of a structural VAR, he attributes the increased volatility after the Civil War to the increased influence of monetary shocks. This separation of structure (propagation mechanism) from shocks will also be used in this paper.

The structural VAR methods that will be used in this paper are very much in the tradition of Blanchard and Quah (1989) and Gali (1992). Similar work in the Australian context has been conducted by Otto (1995, 1999). Otto (1995) tried to identify the effect of terms of trade shocks on Australian output. He found that output could be best explained by productivity shocks with other shocks having a relatively minor effect.