The deleveraging-aggregate demand hypothesis & it’s implications for short-run macro
Mian & Sufi (2011, hereafter M&S) demonstrate that deleveraging has an impact on aggregate demand. This raises the question of how leverage and deleveraging can be incorporated into a short run framework of aggregate demand such as the IS/LM model. The Cubic IS curve is one attempt to show the effects of deleveraging, both through the cubic consumption function and the diagrammatic versions of the IS/LM.
The model and findings of M&S
M&S attempt to answer the question of whether or not deleveraging and the associated decline in consumer demand can explain the sharp reduction in employment in the US between 2007 and 2009. The key to their brilliant paper is the use of tradable and non-tradable industries. A negative consumer demand shock in a given location (from households deleveraging after a major period of leveraging), should reduce employment in industries producing non-tradable goods in that location, but should reduce employment in industries producing tradable goods across the country.
This is exactly what they found. Job losses in the non-tradable sector between 2007 and 2009 were significantly higher in high leverage countries of the US. Furthermore, the large decline in employment in the tradable sector was completely uncorrelated with the 2006 debt to income ratio.
Based on a clever theoretical and modelling insight, M&S are able to calculate the effect of the deleveraging-aggregate demand hypothesis on total employment. They quantify this effect as a total of 4 million job losses out of the 6.2 million jobs lost between 2007 and 2009. Thus, their results suggest that a decline in aggregate demand via household deleveraging is the dominant factor that explains the high and persistent unemployment of the economic crisis.
Could something else explain their results?
M&S confront three alternative explanations that could possibly explain the economic slump, which would render deleveraging to be much less important. Firstly, they address the uncertainty hypothesis – the notion that policy and regulatory uncertainty explains the fall in the economy. However, this does not explain the differences in employment losses across non-tradable and tradable industries in US counties. If it was business uncertainty, the employment losses should have been uniform across sectors and counties, and unrelated to the leverage of counties in 2006. Secondly, the structural employment hypothesis also fails to explain the data. This hypothesis emphasises the displacement of workers from industries like construction after the bubble burst. M&S have a number of counter-arguments, not least that they removed construction and real estate from the definition of non-tradable industries. This implies that the relationships and correlations that they found between leverage and declines in non-tradable employment is highly unlikely to be driven by construction.
Finally, M&S address the credit supply hypothesis. This hypothesis focuses on the possible link between the severe house price collapse, the consequential reduction supply of credit and the reduction of employment in these counties most affected. Again, this does not explain the differences in employment losses between tradable and non-tradable industries. This hypothesis would imply employment losses across all industries, which is not what is seen in the data. Therefore, the deleveraging-aggregate demand hypothesis is the best explanation for the extent and pattern of job losses in the US since the financial crisis.
The implications for a new IS/LM
The great work by M&S has been enhanced recently by the IMF, who dedicated an entire chapter of their World Economic Outlook (WEO) to household deleveraging. While this will be the subject of another post to add weight to the exploration of a cubic IS curve, the point of the chapter was that deleveraging has real consequences on the economy and needs be addressed through bold policy prescriptions.
Also, it seems to me that the renowned work of Nomura Chief Economist, Richard Koo, chimes perfectly with this topic. His conception of the balance sheet recession, first applied to Japan but now relevant to western economies, has deleveraging by different sectors of the economy at its core.
In light of the growing work and recognition of the importance of deleveraging and M&S framing their approach as the deleveraging – aggregate demand hypothesis, it seems appropriate to explore how this fits with traditional aggregate demand models like the IS/LM framework.
The IS/LM model has been a workhorse model in macroeconomic theory for a long time now. Although it receives criticisms from some quarters, the use of it when applied to the current US situation has arguably been the correct approach to understand the effects of monetary expansion. Alternative versions of the IS/LM model have been devised, to good effect, but given that leverage and deleverage usually results in the bubble bursting, it seems interesting to explore a model where debt can build up to a point where the equilibrium becomes unstable.
This is the reasoning for exploring a cubic IS curve, but before we get to a new consumption function that makes the curve possible, I want to go over the evidence from the IMF WEO, as well as the work of Koo and Paul Krugman on the liquidity trap. Subsequent posts on the cubic IS curve will address these areas.
*Mian, A & Sufi, A, 2011. What explains high unemployment? The Deleveraging-Aggregate Demand Hypothesis.