Treasury Revisits the NAIRU

Don’t mention Martin Place

Treasurer Josh Frydenberg’s suggestion that fiscal policy will remain expansionary until the unemployment rate falls below 5% was accompanied by Treasury releasing new estimates of the NAIRU at around 4.5% to 5%, which are consistent with those estimated by the RBA using a somewhat different approach, most notably, the choice of dependent variable.

Treasury’s interest in the NAIRU lies in forecasting wages and not inflation. Inflation is the RBA’s problem, even if the RBA under Phil Lowe tries to make it Treasury’s problem. NAIRU-type models work somewhat better in forecasting wages than inflation. This should not be surprising. It is worth recalling Bill Phillips’ original model was in wages, not prices.

Under a fully credible inflation targeting regime, the only thing that should forecast inflation is inflation expectations. While it is unlikely an economy would ever completely satisfy that benchmark, it is noteworthy that those countries that have adopted inflation targeting have seen a dramatic flattening in their Phillips curve relationships, consistent with this expectation. One interpretation of these quantitatively small Phillips curve relationships in inflation is that they are a residual due to monetary policy imperfectly endogenising economic conditions.

Putting inflation expectations into a NAIRU-type model involves trying to estimate an unobservable equilibrium value with an unobservable, or at least imperfectly sampled, explanatory variable. What could go wrong? This is one of the reasons I’m a fan of nominal GDP targeting as a conditioning variable for monetary policy. Nominal GDP is directly observable, albeit with a lag, although has very good real-time proxies.

There is also the problem that inflation, inflation expectations and output/unemployment gap measures are somewhat observationally equivalent. Econometric estimation can help us sift through their respective contributions, but there’s lots of scope for the explanatory power of one measure to be confused with that of another. Treasury’s working paper concedes as much with respect to the contribution of inflation and inflation expectations. It was inflation expectations that played havoc with the RBA’s model, leading the Bank to over-estimate the NAIRU until recently. For all we know, the latest estimates could also be out.

Treasury’s presents its preferred model and the decomposition of annual wages growth in terms of the explanatory variables as follows:

As the text notes, and the blue bits make clear, ‘wages growth is still largely determined by past inflation.’ This is a problem if you are relying on a pick-up in wages growth to raise the inflation rate, as the RBA likes to suggest. Treasury is not so impolite as to mention 65 Martin Place explicitly in this context, but they could be forgiven. Internal Treasury documents released under FOI have been more pointed.

For those who think wages do not reflect productivity growth, the Treasury approach imposes the restriction that wages fully reflect productivity growth (albeit with some dynamics not unlike those I find in my work). An equivalent way of approaching this question is to assume that inflation and inflation expectations must be equal in the long-run, so nominal wages must grow at the same rate as productivity growth plus inflation (Treasury’s model imposes the assumption that the elasticities of expected and actual inflation sum to 1). While we can question these assumptions, to the extent that the data accept the overall model, it also rejects claims of a lack of pass through of productivity growth to wages. Sorry Paul.

One could also add that the elasticity of aggregate compensation of employees to nominal GDP is close to one, no restrictions required. But you knew that.

NAIRU estimates drop out of these models, which is another way of saying that the NAIRU is an equilibrium condition, not a forecast. When the RBA says that wages growth needs to be 3% or higher for inflation to return to target, it is just describing what equilibrium looks like, not how you get there.

Weak nominal wages growth has a real component and a nominal component. Public policy can do something about both. Addressing the real side is the hard yakka of structural reform. Addressing the nominal component requires the RBA to meet its mandate. Treasury’s model is consistent with this division of labour. The way Governor Lowe talks about wages, inflation and the contribution of monetary policy often is not. Nominal wages growth will increase when inflation and inflation expectations increase. Inflation will only increase to be more consistent with target when monetary policy is more accommodative than it has been.