Policy interest rates needed to increase from their pandemic lows, but there is a question of how far central banks really need to lean against apparently strong demand. What appears to be demand-driven could be a ripple effect from supply shocks affecting other products.
When Cyclone Larry hit North Queensland in 2006, most of Australia’s banana crop was knocked out. Banana prices increased around 400%. This was enough to add more than ½ a percentage point to inflation in the June quarter of that year. It was also one of the cleanest examples of a supply shock that monetary policy should look through. (And at the time resulted in me uttering the sentence, “Of course we can see through bananas!”. But that’s another story.)
Less well known is that prices of other fruit increased in that period, even though those crops were not damaged by the cyclone. This happened because, at those prices, people stopped buying bananas and substituted into other fruit, pushing their prices up. As banana supply normalised, prices of those fruit normalised too.
The question is, should we interpret the impact on prices of those other fruit as “demand-driven”? After all, their prices increased and so did the quantity consumed, the pattern you would expect to see with a positive demand shock.
Through the pandemic period and following Russia’ s invasion of Ukraine, supply shocks were prominent. Factories and other businesses were closed, shipping was disrupted, and domestic logistics were constrained. But there was also stimulus to demand coming from the extraordinary fiscal and monetary policy support. If the prescription for monetary policy is to look through temporary supply shocks to the extent one can without inflation expectations rising, but respond to demand-driven inflation, how do you tell the difference in this situation? This issue becomes especially relevant given the RBA’s assessment that inflation is becoming “increasingly homegrown and demand driven”.
One approach to this is simply to follow the logic that if price went up and quantity consumed went down, it must be supply-driven, while if both increased, it was demand-driven. This is the essence of a methodology developed at the Federal Reserve Bank of San Francisco and used by the RBA. It implies that about half the surge in inflation of recent years stemmed from supply shocks, with the rest coming from strong demand. Different methods using model-based estimates attribute a bit more of the increase in inflation to supply shocks, but the conclusions are qualitatively similar. But is this the right way to interpret recent price moves? Or, with apologies to Jeanette Winterson, is this another case of bananas not being the only fruit?
The complicated way to assess this would be to estimate how much of the price movements in the “demand-driven” components of the CPI reflected demand shifting from supply-constrained components, and how much was from stronger underlying demand. This would become intractable faster than you can say “estimate cross-elasticities of demand”. Another way would be to estimate how much of the increase in overall inflation is explained by the usual drivers of demand, such as income, for example using standard demand-driven models such as a Phillips Curve. The RBA have done this calculation using their own models. As noted above, these exercises do suggest that supply shocks were a bit more important than the San Francisco Fed approach, with demand-driven inflation playing a smaller, but still important, role. But these estimates are only as good as your model.
The simpler way is to remember that inflation is an aggregate phenomenon that should abstract from relative price shifts. It is the balance of aggregate demand and supply that matters. One way to assess aggregate demand for consumption goods and services is by looking at total consumption. Unlike in some peer economies, it has been weak of late in Australia, especially in per capita terms. After initially bouncing back to its pre-pandemic trend when the economy opened up, more recently consumption has been weaker than many expected. This is largely because real household incomes have gone backwards, squeezed by inflation itself, rising taxation and higher interest rates. There are clearly pockets of strong demand and some households are still cashed up with pandemic-related extra savings. But there is always a diversity of experience across households, and these distributional issues should not entirely overthrow an assessment of the aggregate situation.
Last week we saw the November monthly CPI indicator surprise a little on the downside. As our Westpac Economics colleague Senior Economist Justin Smirk noted, the categories that were lower than expected included personal services categories that the RBA had previously been pointing to as evidence of resilient demand.
Slower demand for discretionary services might be a response to the squeeze on household incomes, including from high inflation. But it could also be partly an adjustment as overseas travel patterns normalise. The international borders reopened in 2022, but airline capacity has taken longer to recover. On some metrics, it has still not done so completely. As long as supply in this industry is constrained and airfares higher than pre-pandemic, we should not be surprised to see domestic holidaying and other domestic services demand pushed up as a substitute. And just like the other fruit, we should not be surprised if demand for these services slows as supply constraints on international travel subside.
One upside surprise in the month was in homebuilding costs, which reached 35% above their level four years previously in November 2019. Growth in this component actually picked up in the month. Given the low level of dwelling investment, it is hard to attribute this to strong demand. But neither can we point to global supply chain issues as the culprit anymore. In Canada, one of the few other countries that include homebuilding costs in their CPI, this component has been falling for some months and is up 24% over the same four-year period. Rather, the domestic homebuilding industry remains supply-constrained. The backlog of already approved projects remains high. Meanwhile some capacity has exited, either through bankruptcy or the bid from infrastructure and other non-residential construction projects.
The rapid inflation of rents can, by contrast, be seen as the effect of strong demand. The population surge has strained this market. Vacancy rates are low, and it is hard to see how supply could have kept up, even if the building industry had not been as constrained as it is. But we do not expect this surge to be repeated in 2024. While still elevated by historical standards, demand pressures will gradually subside. Over time this will take some pressure off rental inflation without further action from monetary policy.
Even if the inflation surge had been driven solely by supply shocks, central banks would still have needed to raise policy rates. Both fiscal and monetary policy had been extremely accommodative. This extraordinary pandemic-era support was no longer needed. It was also important to demonstrate resolve and prevent inflation expectations rising. In thinking about how contractionary policy needs to be – and for how long – this expectations channel is important, and there is more to say about it. But it is just as important to be clear on how much of the current inflation is truly demand-driven. Otherwise, policymakers risk squashing the other fruit, to no real benefit.
A Black Friday Technical Aside
This “other fruit” issue is different from the well-known “substitution bias” issue in CPIs. Substitution bias arises because the CPI is compiled using fixed weights for each category. It does not allow for shifting spending patterns in response to large price changes, at least not at first. The spike in banana prices therefore fed through to the CPI as if people were still buying the same number of bananas. The ABS does reweight the CPI to account for these changes, and in recent years it has done so more often than it did back when Cyclone Larry hit, especially for fruit and vegetables, where it now has access to supermarket scanner data. But even with this improvement in data collection methods, measured CPIs tend to overstate people’s experience of this kind of price spikes.
A similar issue is becoming more prominent recently, with the increasing shift of pre-Christmas spending from December into the Black Friday sales in November. This time, the substitution is across time not fruit: people are buying the cheaper on-sale item instead of the full-price item the following month. Retail sales data show that spending is increasingly shifting into the discounted period. Even if household goods prices were being measured monthly, a quarterly CPI would miss the dampening effect on prices of spending being increasingly concentrated in the sales periods. The situation is even worse than that in Australia, though: many goods prices are currently only measured in the first month of the quarter, so Australia’s monthly and quarterly CPI data entirely miss the discounting in November. Prices come back in December, so longer-term trends are less distorted. But these measurement issues will matter for interpretation of December quarter CPI results.
