Topline data, on their own, can mislead. It pays to look below the surface to see the underlying trend.
- Headline data can sometimes mislead. One needs to delve into the detail to get the full story. Headline inflation was softer than expected in April, but underlying trends told a different story. Likewise, employment growth was weak in the month, but this was mostly seasonal and hours worked told a different story. Productivity growth is also likely to have been weak in Q1; again, business investment data give reason not to despair about the trend.
- The RBA has the knowledge and resources to delve into the detail and not be misled by the surface figures. It is clearly concerned about the possibility that the energy price shock will pass through to other prices at scale, keeping inflation high. New detailed datasets have become available that allow its staff to quantify some important relationships more accurately.
- The economy is a complicated thing, though, and it is possible that the RBA’s analysis of the outlook is missing some important trends and relationships. Structural change is especially hard to grapple with. Shifts in labour market trends and the consequences of the data centre boom might be being misunderstood, though hopefully not for long.
Recent data have tended to tell a different story than the headline figures would suggest. Headline inflation was soft in April, but underlying measures were still too high and, if anything, still drifting up. Employment was even weaker than seasonal patterns implied in the month, but the hours worked data told the opposite story. And productivity growth is likely to be reported to have been weak in the March quarter, but this is partly because much data centre investment is in the build phase, and will only add to GDP and productivity later.
These contrasts are all examples of why one can be misled by focusing on a few headline figures. Accurately assessing the state of the economy requires a considered view of the whole picture. There is no substitute for delving into all the data detail.
In doing so, one must also guard against seizing on the data detail that fits your preferred narrative. For example, we remain concerned that pass-through of fuel prices to the prices of other goods and services will be extensive, and we are seeing some evidence of pass-through in homebuilding costs and hospitality already in the April CPI. It is nonetheless also true that pass-through has so far been a bit less than we feared.
The RBA understands the subtleties of Australia’s economic data and has the resources to look below the surface properly. It will not jump to rash conclusions based on headline numbers. It will, however, want to take a pause to assess how dominant and lasting these below-the-surface trends will be.
To the extent that we can glean anything from the RBA’s latest communications, it is that – like us – they are worried about second-stage pass-through of energy costs. Recent research, which was presented to the May meeting of the Monetary Policy Board and discussed in a subsequent speech by the RBA’s chief economist, highlights that pass-through is larger and faster following a large shock than after a small one.
This is not exactly a new result. The RBA has known that the “Phillips curve” relationship between inflation and labour market slack is nonlinear for about three decades. Their main whole-economy model incorporates this nonlinearity, though it does not seem to be in the model used in the speech to highlight the different behaviour of inflation that it implies. And the linkage of this nonlinear relationship to the size of the shock is an implication of so-called “menu cost” models that has been known to other central banks for at least 20 years, both in theory and in other countries’ data, though this link wasn’t recognised when the original RBA nonlinearity paper was published.
Quantifying this effect accurately arguably had to wait until the dataset used in the recent research paper became available. The 1990s-era empirical result means it is nonetheless unlikely that the RBA’s existing forecasting framework would have entirely missed faster pass-through of larger supply shocks, even if the theory behind it was not fully recognised at the time. But the recent communication does speak to the balance of risks that the RBA perceives.
It is possible, though, that the RBA’s current analysis misses some other trends under the surface.
As highlighted by Westpac Economics colleague Ryan Wells’ note yesterday, as best as we can tell, official family forecasters’ views about the outlook for labour supply have diverged. Commonwealth Treasury shares our view that the underlying trend in the participation rate is up. This is a departure from its longstanding assumption that population ageing shrinks the workforce. As we have been highlighting for a while, the opposite is true in Australia and most other advanced economies. By contrast, the RBA’s forecast is for a flat participation rate this year and a decline next year as job opportunities diminish. The labour demand effect is assumed to dominate.
To be fair, Assistant Governor Hunter’s speech did acknowledge that extra labour supply is a possible response to cost-of-living pressures, citing a recent IMF working paper that used Australian data to show this. But they acknowledge it only as a risk, not as their base case. It is also a cyclical factor, not the underlying trend discussed above. Past RBA communication has acknowledged this upward trend, but the latest Statement on Monetary Policy did not mention it. As Ryan commented in his note, the labour demand effect will have to work hard to offset this trend and generate the decline the RBA assumes.
(As an aside, I find it interesting that this paper was released as an IMF Working Paper and not an RBA Research Discussion Paper. Yes, three of the authors are from the IMF, but one is the deputy head of the RBA’s Economic Research department and likely the only one authorised to access the detailed tax and Census linked dataset directly. Likewise, it is interesting that the speech cited the paper in the context of cost-of-living pressures, not the – admittedly small at a macro level – positive labour supply response to monetary policy that the paper focused on.)
It is also possible that the RBA’s forecasting framework has not fully absorbed the implications of the build phase of the AI boom. Capex data released recently shows a huge contribution from data centres and IT investment more broadly – bigger than even our top-of-market expectation. Business investment will be strong, but as Westpac Economics colleague Pat Bustamante points out today, the imported component is very large, especially in the fit-out stage. A fixed view of potential growth in supply capacity will interpret this growth as adding to capacity pressures, even though it is literally building capacity. And like the mining investment boom, the build phase actually drags on measured productivity, as workers labour to build capital stock that is not yet producing anything.
Will a weak outcome for measured labour productivity in Q1 nudge the RBA to become even gloomier about ongoing productivity trends, even though it is likely a temporary lag?
Let’s hope the team at the RBA remember to look below the surface.
