The policymakers view of the great recession - a dynamic AD-AS analysis

I am a big fan of Tyler Cowen and Alex Tabarrok's "Dynamic AD-AS model". I introduce it to around 400 people a year. If you're not familiar with it, I advise you to read their excellent principles textbook. I also have a youtube video, and a presentation.

The main difference with the standard AD-AS model is as follows:

  • Instead of showing the price level and real GDP on the two axes, it shows inflation and real GDP growth

  • A Solow curve instead of a Long Run Aggregate Supply (LRAS) curve

  • An Aggregate Demand (AD) curve based on the (dynamic form of the) equation of exchange instead of Pigou's wealth effect, Keynes's interest-rate effect, and Mundell-Fleming's exchange-rate effect

  • A Short Run Aggregate Supply curve (SRAS) based on the signal extraction problem rather than labour markets

Lars Christensen (2013) has utilised the dynamic AD-AS model to provide an explanation of the Turkish demonstrations. My quibbles with his account would be (1) he argues that demonstrations will cause temporary disruption to production, and models this as a negative shock to SRAS. I would claim this is a negative shock to the Solow curve. (2) He argues that regime uncertainty will damage the potential growth rate of the economy, and models this as a negative Solow shock. I would argue that this is a secondary shock, and that the primary effect of regime uncertainty would be a negative AD shock.

In this post I just want to present a very simple “policymaker” view of the great recession. This involves a comparison of two time periods: Q1 of 2008 and Q2 of 2009. In 2008 AD is growing at 5.5%, which is split between 3% inflation and 2.4% real GDP growth. We can assume that in early 2008 the economy is healthy, and on the Solow curve. Then, by the second quarter of 2009 a series of problems have affected the economy. These include the collapse of the US subprime industry, massive declines in global confidence, and negative wealth effects. Students are asked to summarise these events, and should identify them as negative AD shocks (in other words NGDP was allowed to contract). This implies that inflation and real GDP should fall, and this can be corroborated with the actual data. Figure 1 shows the “solution”.

2008 Q1 – 2009 Q2

2008 Q1 – 2009 Q2

This can generate a discussion about appropriate policy responses, and students will probably mention monetary easing or fiscal stimulus as means to boost AD. Indeed in 2011 Q1 inflation was back at 3%, real GDP growth was 2.1%, and so the economy was close to returning to the original position and avoided the “laissez-faire” outcome of substantial deflation and a SRAS that shifts downwards.[1] Policymakers might indeed argue that successful (albeit belated) injections of demand worked.

However this would be too simplistic. There are plenty of other important shocks that could be incorporated, and inflation has generally been much higher than a negative AD shock would suggest.[2] More importantly, a lot of the story is lost if we start in 2008. It is telling that in Scott Sumner’s attempt to use the dynamic AD-AS model to explain the US recession he starts in July 2008 (Sumner 2009). This captures a key difference between Austrian and monetarist analysis. As Garrison (2001) has said,

“did the collapse occur (a) in the midst of a period of healthy growth because of sheer ineptness of the central bank or (b) near the end of a policy-induced boom that was unsustainable in any event and in the midst of confusion about just what the problem was and how best to deal with it?”

We need to start the analysis before the crisis, in case it was indeed an unsustainable boom. But we also need to allow for the possibility that central bank ineptness makes things even worse. I've done just that in a longer article, published by the Journal of Private Enterprise.

[1] Absent further shocks if AD remained at -3.9% then once inflation expectations fully adjust the implied rate of inflation would be -6.3% (i.e. this is where the AD curve intersects the Solow curve).

[2] This implies something has also affected the Solow curve, and so we need to use the dynamic AD-AS model to unpick this.