Recent data released by the Federal Reserve Bank of New York show mortgage credit has not expanded much since the beginning of the current economic expansion. Unlike many other loans products, Mortgage and home equity line of credit have not grown at the same pace that they used to before the Great Recession. Economists at the New York Fed expected mortgage debt to increase as fast as house prices do, which is a trend they observed during the expansion right before the Great Recession. However, mortgage debt has not done so. Instead, researchers at the bank found plausible that student loans might have outstripped mortgages loans over the last three years. This article takes on the issue and concludes that it is too premature to say that such is the case.
The Fed’s analysis:
The Fed’s analysis goes like the following. William C Dudley, CEO of the Bank, starts by flagging the situation. In his words, “there are other difficult challenges that many households face, particularly with respect to a subject we’ve discussed on previous occasions – student loans”. Andrew Haughwout, Head of Microeconomics Studies at the bank, seconds him by noting that this time around houses prices are up more than one third, whereas mortgages debt has barely grown by one percent since early 2012. Haughwout focuses in explaining data on mortgages, for which he claims there is “a stark contrast to last expansion” in which “both prices and debt roughly doubled” between the years of 2000 and 2006. Both economists pointed towards student loans to explain partially the current situation of the household balance sheet. In other words, the fact that mortgages are not adding debt into the Household balance sheets, begs the question of what is indeed doing it.
Google’s search terms may help out in complementing:
This article takes on the issue by looking at a similar but higher frequency data. In order to expand what economists at the New York’s Fed found, this article uses a time series of the Google’s search terms “mortgage calculator” and “student loans”. I assume both terms reveal the willingness of the American population to at least apply for either of the two lines of credit. In other words, I believe Google’s search terms unveil the interest random people have on such a products over time. Working with these two search terms implies that households face some leisure-labor model constraint. This constraint means that given the deterioration of economic conditions under the Great Recession, households were forced into the school and had to choose to study rather than work. Thus, technically, those two choices became “exclusive” during the recovery from the Great Recession.
That being said, I split the data in two to show how this time around the situation is different. First, a period right before the Great Recession stretching from 2004 until 2009; and a second period right after the Great Recession spanning from 2009 towards 2016. The outcome of splitting the data on those two cycles works for showing how the relationship has changed since 2009.
Graph 1 shows the two search terms over time. It is clear how “mortgage calculator” has declined from about the half of the length period. The term “student loans” instead has kept up over time, even while the economy entered the Great Recession.
Graph 2 presents us with the behavior of the data during the first period ranging from 2004 towards 2009. The term “mortgage calculator” surpasses the term “student loans” by the end of the period length. Otherwise, Graph 3 shows how the term “student loans” outstripped “mortgage calculator” apparently by the end of the period.
When I run the regression, the results are somewhat similar to the graphical analysis. Table 1 summarizes the model and the mentioned two breakdowns of the data. The “all time” model covers data starting on 2004 until what has forgone of 2016. The first breakdown covers 2004 until 2009 while the second breakdown covers 2009-2016. The data on this first regression are the natural logarithms of the Google’s search terms, for which the first difference was applied. The estimated beta for the “all time” 2004-2016 model is .66. On contrast, the estimated coefficient for the first break-down of the data is .97; whereas the second breakdown of the data shows a coefficient of .81.
The length period 2004-2009 shows an almost parallel growth between both terms. On the other hand, the length period 2009-2016 shows a slower rate of change of roughly four-fifths in the relationship. Apparently, there appears to be a deceleration of the “mortgage calculator” term relative to the “student loans” term. However, although the data show some contrasts across periods, it is still too premature to conclude that “student loans” have outstripped “mortgage calculator”, which in our theory equals to say that student loans have outperformed mortgage loans. The reason for stating cautiously this is the fact that the “all-time” estimated beta is considerable lower (.66) than the estimated beta of the second period 2009-2016. Therefore, as of today and by using these “Big data” sources, it is hard to conclude that student loans have surpassed mortgage loans in the balance sheets of American households.