Is FIT the best fit? What the Bank of Canada framework renewal can tell us about the RBA review
It’s time for some public choice theory. Plus, this week’s Australian Q3 CPI
An article in The Saturday Paper on ‘What Drives Philip Lowe?’ contained two direct quotes that Lowe is, wait for it, ‘the perfect neoliberal RBA governor’ and ‘the first openly Keynesian Reserve Bank governor for some time.’ Both those suggestions have very little explanatory power for Australian monetary policy and tell us more about what drives the quoted individuals than the governor.
Public choice theory is a better guide to what drives the governor and has the advantage of being independent of personality. The governor responds to incentives like everyone else, so it is incumbent on government to make sure the incentives are the right ones. Bruce Preston has some Twitter threads (which I have unrolled here, here and here) arguing the current incentives are not good. Improving those incentives should be an important focus for the RBA review.
To get a sense of where the RBA review might land, it is worth considering the Bank of Canada’s most recent Monetary Policy Framework Renewal, which reported in December 2021. Canada’s economy is more like Australia’s than just about any other. It has also pursued a very similar approach to inflation targeting since the early 1990s. So the lessons from the BoC’s review should be broadly applicable to Australia. One of the RBA review panelists, Carolyn Wilkins, was Senior Deputy Governor at the BoC from 2014 to 2020 and was involved in the BoC framework renewal. See this 2018 speech for her reflections on the early stages of that process.
There are important differences in the scope of the two reviews. The BoC’s framework renewal occurs every five years and is largely an internally driven process, although culminates in a joint statement with the Government of Canada. It is roughly equivalent to reviewing the Statement on the Conduct of Monetary Policy between the RBA Governor and Treasurer, although with a great deal more rigor than has been applied to the process by successive Australian governments.
By contrast, the terms of reference for the RBA review are more far reaching, covering everything from statutory mandate and governance, all the way down to management and organisational culture. The BoC’s framework renewal doesn’t speak directly to many of those issues, although does make for an instructive process comparison.
The BoC’s latest review does speak to the appropriateness of flexible inflation targeting (FIT) in an economy very similar to Australia’s. The review endorsed the BoC’s existing inflation target with a 2% mid-point and a 1-3% inflation-control range. As part of the review, the BoC conducted a horse race exercise comparing FIT with alternative targeting frameworks, including:
average inflation targeting (AIT) (notionally the Fed’s framework);
price‐level targeting (PLT);
an unemployment‐inflation dual mandate; and
nominal gross domestic product (NGDP) growth and level targeting.
The different targeting regimes were simulated in the BoC’s main policy simulation and projection model, as well as other models. The results from those simulations are broadly consistent with the findings of the literature dating back to the Brookings Institution project on policy rules in the early 1990s, of which John Taylor’s eponymous rule was a part. Flexible inflation targeting outperformed in terms of the unconditional volatility of key macroeconomic variables such as the inflation and unemployment rates and the output gap, which is the standard approach to measuring the efficiency of different policy rules.
The review concluded that FIT, AIT and the dual mandate were the most robust frameworks. The subsequent framework agreement with the Government of Canada accordingly sought to augment the traditional FIT framework with elements of AIT and the dual mandate. The intention behind these changes was to provide for additional flexibility in pursuing maximum sustainable employment, as well as providing flexibility for policy instruments to be held lower for longer given the (then) low interest rate environment.
Importantly, the BoC is required by the joint statement to explain when it is using that flexibility. I suspect the RBA review will have a lot to say about making Australia’s FIT framework more rigorous and similarly transparent.
The horse race found that ‘FIT, AIT and the dual mandate all outperform NGDP level targeting with respect to all variables except for real household debt growth, where the margin in question is quantitatively small. As a result, it is fair to assume that most ranking systems would place NGDP level targeting somewhere behind FIT, AIT and the dual mandate.’ Nominal GDP growth targeting fared the worst (which is consistent with Scott Sumner’s long-standing opposition to growth rather than level targeting). Zac Gross has been modelling various policy rules in the context of the RBA’s Martin model and has reached some similar, although as yet unpublished, conclusions. It would not be surprising to see the RBA review perform a similar horse race calibrated to Australian data.
The fact that the policy rule simulation literature generally lands on similar conclusions about alternative targeting frameworks can be taken as an endorsement of FIT. The big caveat to that work is whether those simulations are robust to the Lucas critique when calibrated based on historical data in which FIT was the actual policy rule. These exercises are also typically based on historically calibrated interest rate smoothing parameters. I don’t view these simulation exercises as necessarily fatal to the case for nominal income level targeting, although they do place an increased burden on its advocates in making their case (that would be me).
Former BoC Governor Mark Carney made a case for NGDP level targeting at the effective lower bound (ELB) in a speech back in 2012:
The BoC’s framework renewal horse race also suggested that history-dependent targeting regimes like NGDP level targeting performed well against the ELB. Steve Ambler has made a case for nominal income targeting in a Canadian context, although his arguments are general ones and not really specific to Canada.
The big problem for the RBA is not that a FIT calibrated on Australian data will lose a horse race against other policy rules, but that the data will show that the actual cash rate has not followed the optimal FIT-implied rule. That was the finding of Zac Gross and Andrew Leigh’s examination of the leaning against the wind (LAW) episode, which highlighted the substantial welfare costs associated with the departure from the optimal policy rule under LAW. That finding is consistent with a much broader literature showing that LAW has much larger costs than benefits. The fact that NGDP level targeting outperformed FIT in terms of financial stability measures in the BoC’s horse race, despite underperforming on other dimensions, is particularly devastating for the RBA’s position.
The RBA has defended its existing framework on the basis that it is best practice on the part of other central banks. It has also claimed that other central banks have if anything converged on the RBA’s approach over time. The review will be an interesting test of those claims. Even if the review endorses FIT as the best fit for Australia, it may also conclude that the RBA’s conduct of monetary policy has departed from that approach in recent years.
That is where public choice theory comes in. Bruce Preston argues that addressing institutional design will fix the problems relating to objectives and framework. Indeed, there is no point in changing the targeting framework if the RBA is then going to ignore it. I am more inclined to see this as a joint problem, but if I am to be consistent in arguing that NGDPLT is not a radical departure from FIT, then I also have to accept that this is also not a strong argument to change to a new targeting regime.
Even if the review endorses some version of FIT, it is likely to argue for a much more rigorous approach to articulating its parameters and a regime for holding the RBA more accountable for its performance.
Australia Q3 inflation forecast
The US September CPI has not changed our model’s expectation for Australia’s Q3 trimmed mean inflation rate on a rounded basis. The Cleveland Fed’s trimmed mean was steady in September at 0.6%, yielding an average 0.5% increase over the quarter, below the 0.7% average increase for Q2. This leaves the local expected Q3 trimmed mean inflation rate at 1.4% q/q, taking the annual rate to 5.5% compared to 4.9% previously. This would represent a slight moderation on the 1.5% rise in Q2, despite the acceleration over the year. The median market expectation for the trimmed mean measure is another 1.5% q/q print, so we marginally below market for this release.
ICYMI
The C.D. Howe Institute Monetary Policy Council calls for Bank of Canada to raise the Overnight Rate to 3.75 Percent
Memes and themes: