Why We Love Cars

March 21, 2014

An integral part of macroeconomics is understanding business cycles, i.e. figuring out what drives changes in aggregate output. We have devoted the last many years trying to understand why large declines in output and employment – as seen in the Great Recession – occur. In trying to answer such questions, a major obstacle in macroeconomics is the paucity of micro data. For example, while we know what happens to consumption in the aggregate, we do not have an accurate idea of movements in consumption at the neighborhood or even city level.

This is unfortunate because micro level data can greatly help us better understand the root causes of large falls in output and employment. For example, by comparing the neighborhoods that suffer the largest decline in consumption with neighborhoods that somehow escape recessionary consequences, we can hone in more precisely at the root cause of the problem. It is a bit like an epidemiologist comparing subjects that seem immune to a virus with those who quickly succumb to the epidemic. It is clear that such evidence is critical to treating and preventing future outbreaks.

As a result we are always on the lookout for measures of economic activity at finer geographical levels like zip codes or counties. One of our favorite variables in this regard – and one that we have used very often in our research and book – is the number of new automobiles bought by residents of a given zip code.

Why, you might ask, does this particular data exist? The reason is that anytime someone buys a new car, they need to get it registered at the local motor vehicle authority, which in turn records the owners residential address. We can hence track on a monthly frequency the number of new vehicles bought by household in every zip code in the United States.

Is zip code level data on purchase of new automobiles useful for macro-economists? First, the bad news. Sales of new vehicles comprised only 2.9% of total GDP at the start of the Great Recession.

So one might think that looking at the sale of new cars is not very useful given how small a share they hold in the overall economy. But this intuition would be wrong. Remember that it is the change in GDP that we are truly interested in. And it turns out that as far as change in GDP is concerned, sales of new vehicles explain a large fraction. For example, the decline in the sale of new automobiles comprise 28.1% of the peak to trough fall in GDP during the Great Recession.

We can see the usefulness of new car sales for macro-economists more clearly in the graph below. It plots log GDP (left axis) and log vehicle spending (right axis) over time, with time (in quarters) indexed to zero for the trough of the Great Recession (2009Q2). Two things are worth noting. First, vehicle spending strongly co-moves with GDP, and in fact tends to lead the cycle. Second, the variation in vehicle spending is ten times as large as the variation in GDP. That is why we had to plot the two graphs on separate scales.




The net result is that new vehicle sales provide an extremely useful lens for figuring out what caused the Great Recession and hence what we can do to prevent such events from happening again. You will find a lot of mention of new car sales in our book. Now you know why we love cars.

By the way, if you know if a data source that measures household spending or consumption at a micro level, let us know in the comments or via twitter!

Bookmark and Share

6 Responses to Why We Love Cars

  1. Jim on March 21, 2014 at 8:47 ami

    I’m a little confused. You are framing what caused the great recession, I thought the general consensus was that mortgages were the main contributor. Are you inferring that the decrease in automobile consumption was a symptom of the loss of income for consumers?

  2. Joseph on March 21, 2014 at 10:58 ami

    Jim: It’s not so much that there’s a theory that automobile spending drives the business cycle, but that if you know a given zip code’s change in automobile spending, you can infer its change in consumption, which is the largest component of GDP.
    For example, if we were to observe that changes in automobile spending in a zip code is inverserly related to changes in housing prices, then the theory you hinted at above would be called into doubt, even though at the national level there seems to be a correlation.

  3. Pete on March 21, 2014 at 11:19 ami

    The way I read it is automobiles are correlated with the aggregate economy, almost like a Leading Economic Indicator. But why would the automobile micro data lead GDP? My intuition would be that large capital goods would lag GDP. Households resupply income before they buy big ticket items.

  4. Anon on March 21, 2014 at 12:32 pmi

    “The reason is that anytime someone buys a new car, they need to get it registered at the local motor vehicle authority, which in turn records the owners residential address.” It’s also necessary to register the purchase of a used car. Presumably, your data distinguish between purchases of new and used cars. Yes?

  5. Anonymous on March 21, 2014 at 12:39 pmi

    Where would one find data on monthly car sales by zipcode?

  6. AJ on March 21, 2014 at 4:54 pmi

    Isn’t the term “Great Recession” a misnomer. The recession began Q4 2007 and ended Q2 2009 after having positive GDP growth. Since Q2 2009 we have been stuck in a output gap where consumers have been spending $250B less per quarter compared to if the recession never occurred. The Great Depression or what became known as the Great Depression consisted of 3 years of recession 1930-33 then 7 years of an output gap where real GDP never closed the gap of potential GDP until 1940 due to war spending. In fact the last time the US had any meaningful CPI deflation was in 1933, but when the Fed went off the domestic gold standard it created positive inflation up to I believe 8% from a starting point in 1933 of -12%. In fact nowhere in the world has their been a sustained deflation since 1933 not even Japan. The most likely course ahead is a severe inflation as it will be the only politically acceptable choice for the Rentiers to deflate the debt away.

    Record auto loan issuance is simply a symptom of ZIRP policy and the continued extension of loan maturities out to on average now 72 months and further. I am hearing loan officers being asked to extend auto loans out to 84 and 96 month loans, in fact the norm in the Northwest is now 84 month loans according to Capital One auto analysts.