RDP 2015-07: A Multi-sector Model of the Australian Economy 6. What has Driven the Australian Business Cycle?

In this section, we explore the sources of Australian business cycles over recent decades. To do this, we construct a historical decomposition of Australian GDP growth and inflation over our sample, which broadly coincides with the inflation-targeting era in Australia. The idea behind a historical decomposition is as follows. The model attributes all deviations of growth and inflation (and other variables) from their steady-state values to the model's structural shocks. A historical decomposition recovers these shocks and shows the contribution of each to the evolution of GDP growth and inflation.

Figures 10 and 11 present the results of this exercise.[16] As in the variance decomposition section, we group similar structural shocks into broader categories to make the results more interpretable. In this exercise we separate out the risk premium shock (εψ) from the other world shocks. For GDP we also separate the investment-specific productivity shock (εϒ) from the other productivity shocks.

Figure 10: Historical Decompositions – GDP Growth
Figure 11: Historical Decompositions – Inflation

Over our estimation sample, the Australian economy has experienced two sustained expansions – in the mid to late 1990s and mid 2000s – and three mild slowdowns – in 1995, the early 2000s and during the global financial crisis (GFC). The period since the GFC is difficult to categorise. The economy grew strongly in the immediate aftermath of the GFC. But, on average, it has grown more slowly than in the preceding two decades.

The model identifies three main causes of the expansion in the 1990s. First, a sequence of strong productivity shocks that was spread fairly evenly across the decade. Second, positive investment shocks in the period shortly after the 1990s recession. Third, positive demand shocks in the early and again in the late 1990s.

For the most part, foreign shocks were also expansionary during this period. Risk premium and resource price shocks were generally small and often offsetting.

The model attributes the expansion of the mid 2000s largely to positive investment and resource price shocks (although foreign and demand shocks also contributed at various times). The positive investment shocks in the early years of the decade were concentrated in the non-mining sectors of the economy. The pick-up in mining investment (which was largely an endogenous response to higher resource prices) came later in the decade. In contrast, the model suggests that productivity shocks reduced the pace of GDP growth during this period, consistent with most accounts of this era (Eslake 2011; Kearns and Lowe 2011).

The model attributes the slowdown in 1995 largely to monetary policy shocks. The RBA increased interest rates by 275 basis points between August and December 1994 in order to limit an anticipated pick-up in inflation.[17] From the perspective of the model, which is estimated over the entire inflation-targeting era, the size of the monetary tightening appears unusually large. Hence, it labels at least some of these interest rate increases as policy ‘shocks’ rather than endogenous responses to strong economic conditions. However, as emphasised by Stevens (1999a), this tightening cycle occurred at a time when the RBA had recently adopted inflation targeting and inflation expectations remained weakly anchored. In such an environment, a more aggressive response to an anticipated inflation may be required to prevent the pick-up in inflation from occurring, as well as to reinforce credibility of the central bank's policy regime. Indeed, Stevens (1999b) identifies the 1994 tightening cycle as the key episode that allowed the RBA to lower interest rates in the face of a depreciating exchange rate during the Asian financial crisis of 1997.

In contrast, the model attributes the early 2000s slowdown largely to negative investment shocks, particularly in the non-traded sector. This result may indicate a degree of model misspecification as it is likely to be capturing the effect of the introduction of the GST in July 2000. This induced a large amount of building investment activity to be brought forward in the first half of 2000 and consequent reduction in activity in the second half of that year. Because the model features only lump-sum taxation, it does not account for the introduction of this tax explicitly and instead assigns its effects to the model's investment shocks.

According to the model, the slowdown associated with the GFC had several causes. Early on, the economy experienced a sequence of negative foreign, resource price and productivity shocks. These were followed a few quarters later by negative demand and investment shocks. This is consistent with the idea that bad news from abroad triggered a loss of domestic confidence which, in the model, manifests itself in demand and investment shocks.

Although the broad pattern of shocks appears plausible during the episode, one might have expected foreign shocks to play an even larger role, given that the crisis was largely triggered by economic developments offshore. The fact that they don't may be due the fact that the model lacks a rich financial sector, which was an important channel through which foreign disturbances may have affected the Australian economy in this episode.[18] During the crisis, the model suggests that the unusually large reduction in nominal interest rates was instrumental in preventing a more severe downturn.

In the period immediately following the GFC, rising resource prices made a substantial contribution to Australian GDP growth. Investment grew by even more than would be expected from rising resource prices alone and so the model's investment shocks also made a positive contribution. Also, the model's markup and productivity shocks, after several years of subtracting from GDP growth, contributed to higher growth once again. More recently, however, falling resource prices have subtracted from GDP growth. This has been followed by a period of unexpectedly weak investment growth and some large negative demand shocks. Although this latter period also coincided with an exchange rate depreciation, the models suggests that this can largely be explained by the other structural shocks affecting the Australian economy.

Underlying inflation has remained within the RBA's 2–3 per cent target band for much of the inflation-targeting era. Exceptions include a few quarters in 1995–96 and 2001–02, when it exceeded 3 per cent by a small amount, an episode in 2007–09, when it reached almost 5 per cent and a period in 1997–99, when it fell below 2 per cent.

The model attributes the low inflation outcomes of the 1990s largely to a sequence of productivity and mark-up shocks.[19] These forces were particularly strong in the latter half of the decade and, according to the model, explain the unusually low inflation outcomes in 1997–99.

The deflationary impact of productivity shocks is consistent with the strong productivity outcomes recorded during this era. The contribution of lower markups may reflect the lagged effect of earlier product market reforms (see Kent and Simon (2007) for a discussion of the timing of Australian product market reforms). These reforms may have have intensified competition in many industries and reduced firm mark-ups, as discussed in Forsyth (2000).[20]

The model attributes the pick-up in inflation in 1995–96 to a number of factors. There was a temporary reduction in the deflationary effects of the model's productivity and supply shocks. This coincided with a modest exchange rate depreciation and a positive contribution from the model's foreign shocks. In contrast, the model suggests that the small rise in inflation in 2001–02 was almost entirely due to the lagged effects of the exchange rate depreciation around the turn of the century, which was larger than the model can explain by fundamentals.[21]

The largest deviation of inflation from the RBA's target occurred in 2007-09. According to the model, this largely reflected a sequence of large mark-up shocks. There is some out-of-model evidence to support this finding. For example, the Australian Bureau of Statistics' measure of retail trade gross margins – which capture both changes in net profit and cost of doing business – increased strongly during this period. However, in the absence of a more compelling out-of-model story for why mark-ups should have increased in this episode, we find this explanation for the acceleration in inflation at this time incomplete. It may be that the model is attributing deviations in inflation that it cannot otherwise explain to mark-up shocks.

Since the GFC, inflation has remained within the RBA's target. These relatively stable inflation outcomes conceal a number of strongly countervailing influences on inflation. The model suggests that the appreciation of the exchange rate associated with the investment phase of the mining boom exerted a strong deflationary effect on the economy. Global economic influences more generally have also lowered inflation during this period. This could reflect the continued emergence of China as a source of manufacturing goods, which may have put downward pressure on the prices of these goods. It could also be due to weak economic conditions in the United States, Western Europe and Japan for much of this period. The model suggests that these deflationary forces were offset by weak productivity growth outcomes, which raised inflation. Although movements in Australian inflation are often attributed to developments in resource prices, the model suggests that these changes have had only a minimal effect on inflation over recent decades.

On the basis of these results, we believe that the model provides a plausible explanation for the behaviour of Australian GDP growth and inflation over recent decades. While many of the factors contributing to the outcomes for these variables have previously been identified elsewhere – for instance, strong productivity growth in the 1990s – the model adds value by quantifying the importance of these factors. It also helps us to separate out the contribution of shocks such as risk premium and monetary policy shocks to economic outcomes from the endogenous response of variables like exchange rates and interest rates to economic conditions.

Footnotes

In both figures the contributions sum to the deviation of year-ended GDP growth (Figure 10) and inflation (Figure 11) from its sample mean. [16]

For a description of monetary policy during this episode, see Debelle (1999) and Stevens (1999a). [17]

Finlay and Jääskelä (2014) find that credit shocks played an important role in the financial crisis in Australia, but not a dominant one. [18]

Although the shocks are not autocorrelated they may have persistent effects due to price rigidities. [19]

In theory, these reforms should lead to permanent changes in mark-ups and so should not be captured in our model, which focuses on cyclical variations. However, they may induce temporary transitional dynamics as the economy moves from a less competitive equilibrium to a more competitive equilibrium. The shocks in our model may be capturing these effects. [20]

This apparent departure of the Australian dollar's value from underlying fundamentals was noted by the RBA at the time, for instance in Macfarlane (2000). [21]