The UK Phillips Curve

 I have become increasingly conscious that this blog contains scant macroeconomic analysis, perhaps making the subtitle misleading. This piece therefore intends to summarise the findings from my recently submitted undergraduate thesis, but will hopefully still be accessible.

I recently submitted my undergraduate dissertation, entitled: 'The UK Phillips Curve Since the Great Moderation: A Single-Equation Econometric Analysis'. The basic idea was to employ the widespread single-equation approach to the Phillips curve in analysing UK macroeconomic data during the Great Moderation, the Great Recession and thereafter. But why was this analysis necessary?

First and foremost, as Simon Wren-Lewis has recently argued, the Phillips curve represents a fundamental economic intuition: when more people are in work, aggregate demand is stronger, meaning it is more likely to be profit-maximising for individual firms to raise prices, leading to inflation. A.W. Phillips, in his seminal 1958 paper, highlighted that this relationship held steady statistically in the UK between 1861-1957. What became known as the 'Phillips' curve therefore came to represent a positive relationship between economic activity (x) and inflation (π), taking the general form: 

π(t) = E(t)*π(t) + λ*x(t)        (1)

where E(t) represents expectations and x normally some measure of the output gap. The 1970s however demonstrated that λ need not always be positive, and R.J. Gordon notably argued for the inclusion of 'supply' shock factors to take this variation into account. Generally, Phillips curve vary in the assumptions made in the model regarding both expectations and the economic activity variable. 

During the 'Great Moderation' (c.1992-2008), in the UK both economic activity and inflation have been stable. However, in response to the Great Recession (where the UK output gap surpassed -5%), inflation remained remarkably stable, only falling as far to 1.1% and not further towards deflation as was feared. This has been dubbed as the case of "missing deflation", and has led many to reinvestigate the Phillips curve (see papers by Ball and Mazumder, 2011; Blanchard, Cerutti and Summers, 2015; Blanchard, 2016). In particular, I wished to investigate an apparent tension between those who advocate a purely backward-looking Phillips curve such as R.J. Gordon (2011), where expectations are replaced with realised past values, and those who argue for some version of the standard New Keynesian Phillips Curve (NKPC), one of the three microfounded equations that form the canonical New Keynesian model. In addition to trying to pin the Phillips curve down, I wished to discern whether single-equation models taking the form of (1) could be useful for informing future macroeconomic policy. 

The central econometric findings of the paper were as follows: 

• Inflation in the UK has demonstrated a particularly high level of persistence between 1992-2016. This was the case for both types of Phillips curve model estimated. Moreover, both types of model suggest that backward-looking behaviour of some sort has become more important, hence the 'missing deflation' puzzle. This is generally interpreted in the literature as evidence of so called 'inflation anchoring', in response to the Bank of England credibly targeting a 2% inflation. There is general consensus around 'anchoring' for US data, and the UK appears somewhat similar in this sense.

• Forward-looking behaviour of some kind is still present, though it was not clear whether this is in the rational expectations sense or in a heuristic 'learning' sense.

• For all models tested, economic activity was almost always statistically insignificant in determining UK inflation dynamics in this sample period, sometimes with the incorrect sign. This supports the 'flat' Phillips curve hypothesis - the idea that credible inflation targeting has led to a statistical vanishing of the Phillips curve - again around which there is considerable agreement (for instance, see Chris Dillow). 

• Single-equation models, whilst they shouldn't be discarded as Simon and Blanchard have recently argued, are not appropriate for modelling the UK Phillips curve, and amount to mis-specifications. 

From a policy perspective, there are two contrasting things to be drawn here. The first is that independent central banks appear to have been highly successful in controlling inflation by anchoring inflation expectations by reinstating policy credibility. It is therefore highly desirable that this continues. Secondly, and probably more importantly, the problem of an insignificant relationship between economic activity and inflation challenges the main assumption of modern macroeconomic stabilisation policy - that stabilising inflation is the channel through which the business cycle can be stabilised, termed the divine coincidence by Jordi Galí and Olivier Blanchard in 2007 paper. A major problem for the Bank of England is how to reliably target economic activity under such uncertain conditions. This has led to a debate in the current literature regarding the raising of inflation targets, which in theory may allow policymakers to effectively 'rediscover' the Phillips curve, however Guido Ascari has noted that doing this may de-anchor and unleash inflation expectations, which may potentially have greater costs to economic welfare. 

The general conclusion of the paper, however, is not an emphatic one. It would be an error to argue that the Phillips curve no longer exists, especially given its chequered and transient history. As Haldane and Quah demonstrated in their model of monetary policy credibility, the perceived attitudes of policymakers factor into inflation dynamics too. The Phillips curve perhaps disappears statistically precisely because policymakers have credibly denied the existence of such a trade-off. As soon as such attitudes change then, the Phillips curve is likely to rear its head once again.