Is Consumer Spending Really “Driving” the Recovery?
According to the Bureau of Economic Analysis’s advance estimate, in the fourth quarter of 2010, GDP increased at an annual equivalent rate of 3.2 percent. Personal Consumption Expenditures (PCE) alone contributed a whopping 3 percent to this rate, as they grew by 4.4 percent in the quarter. The main drag on GDP growth came from changes in private inventory investment which, while marginally positive, dropped precipitously from their highest level in a decade in the third quarter of 2010. The growth in PCE was the big news, though. For 2010 as a whole its growth rate was 2.7 percent, the fastest over any four-quarter period since 2006.
Economists are very fond of their jargon and one example of it that we hear a lot these days is that “consumption must drive the recovery.” What this means is that because consumption expenditures are such a large share of GDP (roughly 70 percent currently), for GDP as a whole to grow at a healthy pace it had better be the case that consumption expenditure growth does not lag this pace too much. Notice that this does not mean that consumption expenditures should grow faster than GDP. People tend to smooth consumption, so it tends to decrease at a slower pace than GDP in recessions and increase at a faster pace during recoveries.
Nonetheless, given the recent seemingly stellar behavior of consumption expenditures, a question worth asking is whether they are doing better than in recoveries from past recessions. While there certainly are various ways of measuring such performance, one particularly useful one is to look at the evolution of the consumption share of GDP in recovery periods. The faster consumption recovers relative to GDP’s recovery as a whole, the higher this share becomes.
The figure below shows how this share has evolved from the recent recession’s trough up until the last quarter of 2010 and compares it to the average behavior of the same measure during all recoveries from NBER recessions since 1952.
This figure does show that consumption growth, relative to GDP, is stronger in this recession than in previous ones. Nonetheless, one should be careful in interpreting any sort of causation in this relationship, as both consumption and output are determined together. (They are, to use some more jargon, endogenous variables.) This means consumption is no more a driver of GDP than GDP is a driver of consumption; they are simply determined together. To be able to make one assertion or the other, one needs a theory (demand-side theories, like Keynesianism, for example, emphasize consumption as the driver.)
Since income is, broadly speaking, split between consumption and savings, it might just be that consumers have been saving less than in the average recovery. Indeed, looking at savings rates as a fraction of GDP (not disposable income) reveals that savings rates in this recession have been below average.
Nonetheless, when we look at households’ net worth as a fraction of GDP, we get exactly the opposite picture: households have been able to improve their balance sheets while consuming more and saving less (relative to previous recessions.)
The key to reconciling these differences lies with asset prices. A cursory look at the Flow of Funds table reveals that the values of tangible assets (mostly real estate) have not changed much since the trough of the recession, consistent with the view that housing prices have yet to increase. But tangible assets constitute only roughly one-third of total assets. In the aggregate, financial assets are much more important (I say in the aggregate because this differs considerably across households,)and their value has increased substantially.
How did this increase in asset values come about? Since savings have increased no more than in previous recessions, it must be that financial asset prices have been going up a lot. Noting the S&P 500 has roughly doubled since March 2009 seems to lend some credence to this hypothesis.
Note: This article has been revised substantially since it was first posted. In a previous version the author looked at the real share of consumption in GDP as opposed to the nominal. It turns out the behavior of the two is very different (the author thanks Bernd Weidensteiner for pointing this out.) While the former decreases relative to the average recession, the latter increases. Because the BEA uses a chain-weighted method to compute real GDP, as opposed to a fixed-weighted method, the nominal share ratio (not the real) is the more appropriate measure to look at.