RDP 2004-09: Co-Movement of Australian State Business Cycles 2. Previous Research and Data Issues

In contrast to the lack of previous work using Australian data, there are a number of studies suggesting that quite close linkages exist between US regions. Kouparitsas (2002) and Carlino and Sill (1997), for example, find that correlations of per capita personal income cycles among regions of the US range from 0.6 to 0.8, and Owyang, Piger and Wall (2003) find that turning points in the economic cycles of US states occur at broadly the same time as turning points nationally. Similar results have also been found for regions within European countries (Barrios and de Lucio 2003; Barrios et al 2003), and across country blocs (Artis, Kontolemis and Osborn 1997; Hall, Kim and Buckle 1998; Stock and Watson 2003).

Although the nomenclature used varies, the literature broadly identifies two types of shocks to economic activity that might account for this observed co-movement:[3]

  • common shocks, which affect all regions simultaneously (examples of such shocks include changes in the exchange rate, monetary policy or world economic activity); and
  • idiosyncratic shocks, which are specific to individual regions (examples of which include changes in regional fiscal policy, regional droughts or local bank failures).

Models using this framework also typically include a mechanism for idiosyncratic shocks to spill over from the region in which they originate to other regions, through trade and investment channels, for instance.

If regional cycles are driven to a large extent by idiosyncratic shocks, they will tend to display individual dynamics (i.e. they will not co-move) unless those shocks then spill over to other regions. In contrast, if regional cycles are heavily influenced by common shocks, they will tend to display similar dynamics. Consequently, common shocks and spillovers create co-movement, whilst idiosyncratic shocks reduce co-movement. The available research suggests that common shocks are a major source of cyclical co-movement between regions, although spillovers are also non-trivial (Kouparitsas 2002). Cross-country research has yielded similar results although, consistent with the smaller number of common shocks, spillovers play a greater role in cross-country dynamics (Norrbin and Schlagenhauf 1996; Clark and Shin 2000).

The difficulty with examining these issues at a sub-national level is the scarcity of comprehensive measures of economic activity. The most comprehensive measure of state economic activity, gross state product (GSP), is only consistently available for Australian states on an annual basis (from 1989–1990), which limits its usefulness in assessing business cycle synchronicity. We have therefore chosen to use two alternative proxies of economic activity in our analysis: state final demand (SFD) and hours worked. Both measures have limitations. SFD, which measures total domestic spending in each state and is akin to domestic final demand at a national level, excludes important components of economic activity, particularly external trade (both international and interstate). This deficiency has implications for our assessment of the importance of spillovers, given that trade is often seen as a primary avenue for the transmission of shocks from one state to another. In contrast, hours worked data should capture changing conditions in a state's external sector, but information from the labour market is only an indirect estimate of economic activity, and shocks might be observed with a lag. In what follows, we use both measures to try to mitigate these deficiencies.


Several papers also include industry-level shocks, which could be assigned to either of these categories (depending on the degree of industrial similarity). [3]