More on nominal GDP targeting

Michael Hatcher, University of Southampton. Posted on September 12th 2016.

In my last post, I put forward three reasons why nominal GDP targeting would be a bad idea for central banks. My reasons, in short, were as follows:

1. Nominal GDP targeting does not provide an anchor for inflation expectations;
2. Nominal GDP targeting would not be easy to communicate to the general public;
3. Credibility would be strained under nominal GDP targeting.

In this post I develop these ideas further with the use of some simple diagrams familiar to economists. I start out with the current regime of inflation targeting to give us a benchmark against which nominal GDP targeting can be compared.

Inflation targeting

Under inflation targeting, the central bank takes low and stable inflation as its primary objective and treats past deviations from the inflation target as ‘bygones’. Subject to price stability, an inflation targeter also has a secondary objective: to support economic growth and employment. This means that if price stability is achieved, then the central bank has the freedom to set interest rates to ensure (insofar as possible) that the economy is at or near its full capacity level. Thus, an inflation targeter is not an ‘inflation nutter’ – a policymaker who keeps inflation at target at any cost (see King,1997). Instead, deviations from target are acceptable as long as they support growth or overall economic stability and do not put price stability in jeopardy. It is for this reason that Bernanke and Mishkin refer to inflation targeting as a framework for monetary policy, not a strict policy rule.

The flexibility provided by inflation targeting can be seen in Figure 1. In an ideal world, the central bank would like to keep inflation equal to its target value of π* and output at its full capacity level Y*, as shown by the ‘bliss point’, B. But in practice this is neither possible nor desirable because the economy is continually hit by unforeseen shocks. Consequently, central banks practice a form of flexible inflation targeting: temporary deviations from the inflation target are permitted in order to prevent large fluctuations in output and employment. The monetary rule curve, MR^IT , shows all the combinations of inflation and output that the central bank will choose at times when supply shocks make the bliss point unattainable. Points on the MR^IT curve which are closer to the bliss point are preferred by the central bank, but big enough supply shocks will imply inflation and output combinations some distance from the bliss point (see points A and C).

Figure 1: Inflation targeting: response to supply shocks

Notes: Diagram shows the response of inflation and real to GDP to supply shocks under inflation targeting.

Crucially, inflation targeting is not confined to a single MR curve. Indeed, since inflation targeting central banks do not announce to the general public the rate at which they will trade-off inflation and output variations, they can ‘fine tune’ their response to supply shocks, being tougher on inflation at times when their anti-inflation credibility is called into question (see MR^IT,flat) and more lenient when allowing inflation to rise above target will help the economy to maintain output near its full capacity level (steeper MR curve, not shown). Thus, inflation targeting offers flexibility not just by permitting deviations of inflation from target, but by giving the central bank the freedom to respond differently when circumstances call for it. This flexibility would also be present under price-level targeting, albeit that in this case the central bank has a target for the aggregate price level rather than its growth rate. By contrast, nominal GDP targeting does not give the same flexibility.

Nominal GDP targeting

Under nominal GDP targeting, the central bank would no longer take inflation as its primary objective and past deviations from target would be corrected rather than treated as bygones. The central bank would give an equal weight of 1 to both the price level and output – or, equivalently, to inflation and real GDP growth – as we see from the formula P + Y = N*, where N* is the target level of nominal GDP (all variables are expressed in logs). In particular, the central bank is promising to make the price level and output move one-for-one, but in opposite directions (see Figure 2). Consequently, the monetary rule curve under nominal GDP targeting, MR^NGDP, will have a fixed slope of -1 and the central bank is committed, at any point in time, to a single MR curve . For example, in response to a large negative supply shock, it should conduct monetary policy so that inflation is at π₁ and output at Y₁ (point 1), since this will ensure that the nominal GDP target is met. By comparison to inflation or price-level targeting, the central bank would have little flexibility since supply shocks would not be a valid reason to deviate from the nominal GDP target. As a result, the central bank would be expected to meet the nominal GDP target even in the face of large supply shocks, despite the fact that this may be costly or even impossible.

Figure 2: Nominal GDP targeting: response to supply shocks and the ‘bliss curve’

Notes: Diagram shows the response of inflation and real GDP to supply shocks under a nominal GDP level targeting regime.

Since the past price level is predetermined (and could be set at 1) we can express the target in terms of inflation and real GDP.

This brings us on to a second point, namely that there are an infinite number of bliss points under nominal GDP targeting which coincide exactly with the MR curve. Hence, if the central bank announces today a nominal GDP target that rises by 5% per year, this could be met through 2% inflation and 3% real GDP growth, but also through any of the infinite combinations of inflation and growth that satisfy the equation: inflation + growth = 5%. In other words, the central bank is sending to the public the message that inflation of 9% and real GDP growth of -4% (point 1) is just as desirable as inflation of 2% and real GDP growth of 3% (point 2), which is just as desirable as real GDP growth of 9% and deflation of 4% (point 3). In other words, anything goes in terms of inflation and output, provided that the nominal GDP target is met. In my last post, I argued that for this reason nominal GDP targeting would neither anchor inflation expectations nor be easy to communicate to the general public. By contrast, inflation targeting has a single bliss point, so the central bank is sending a clear message about what inflation rate it would like to achieve. This, in turn, will help to keep actual inflation under control.

My third concern about nominal GDP targeting was that it would struggle to build and maintain a high level of credibility. I focused in my previous post on demand shocks, but here I take up the argument made by Scott Sumner and others that nominal GDP targeting would provide greater flexibility in response to supply shocks (see here). Comparing Figures 1 and 2, it is not clear that this argument is correct. Indeed, nominal GDP targeting is rather restrictive, for it implies that a shock which pushes up inflation by X% should be met with an output contraction of exactly X%, in order for nominal GDP to remain at target. This is rather problematic given that supply shocks can feed through to inflation and output at very different speeds, and need not abide by the inverse one-to-one relationship required by nominal GDP targeting. A nominal GDP targeting central bank would thus set itself a strict target from which it would frequently want to deviate, with the result that it would struggle to build and keep credibility. As argued above, either inflation or price-level targeting would provide more flexibility precisely because the central bank does not need to declare a strict rule that specifies exactly how it should respond to supply shocks.

Footnote: I only noticed Scott Sumner's reply to my last post after writing this piece. My reply can be found here.