Brad DeLong is an esteemed economist at U.C. Berkely, but I am confused by evidence he uses in a recent blog post. Brad uses the following graph to argue that the level of deficit spending through the current crisis does not even approach the level during WWII which he views as necessary to reduce the unemployment level:
The most perplexing thing, to me, is that DeLong says this graph shows that we are “far” from a fiscal boost a “quarter to a third” as large as the WWII one – but the height of the 2009 spike, about 15%, is more than a third of the WWII spike of about 40%.
That aside, however, does this graph really show what he’s trying to measure? It displays the annual change in debt held by the public, as a percent of GDP. It is important, however, to differentiate between “World War II-style deficit spending” and a “fiscal boost”, terms DeLong uses interchangeably in his argument. This graph is well suited to identifying the degree of “boost”, as spikes represent sudden capital infusions (or large annual changes). However, it does not do a good job of showing the actual level of deficit spending. In fact, it is the first derivative of the level.
Here is another graph, from a different part of DeLong’s website:
As this clearly shows, the government is currently borrowing on an order only seen during World War II. A spike of 40%, which DeLong uses the first chart to argue in favor of, would take us to nearly 125% of GDP – an unsupportable level. So in terms of whether or not we will see a fiscal boost on the order of WWII, the answer is probably not; we are already at such a high level that we can not sustain an increase of that magnitude.
Just to throw a wrench in things, here’s another way of looking at this: the WWII spike lifted debt from 45% of GDP to 85%, or by just less than 2x. The recent cumulative spike is from 35% to 60% – also 2x. Thus, depending how you cut it, we’ve increased spending the same amount as in WWII (proportionate to the prior level); about half as much (in an absolute sense); or about a third as much (in terms of absolute annual change).
All of these are at or above DeLong’s target level, making me wonder why he thinks “we are not there”?
Felix Salmon’s chart of the day comes by way of wcw and shows household assets as a percentage of GDP:
Felix notes:
If this line reverts to the long-term mean, the extra evaporation of housing equity would be utterly devastating. But is there any reason to believe that it won’t?
But I wonder, what is the long-term mean of this statistic?
On the one hand, there is the regime that prevailed for almost the first 30 years, where real estate comprised a bit more than 80% of GDP. Returning to those levels would indeed be devastating in the near term.
On the other hand, there is the actual average of the values exhibited since 1950, which is closer to 110%. Still worrisome, but not as bad.
Finally, we must consider the real possibility that the steady state of this number has naturally inflated with time. As banks offered higher LTV’s, it became easier for Americans to invest the same part of their income in more valuable homes. This suggests that home values as a % of GDP could reasonably be increasing. Additionally, more developed capital markets have made it easier to unlock that value (and measure it), furthering the relative growth of home values.
If you naively detrend the graph, we are near to or potentially even below the mean value. However, a simple detrend would be insufficient because the reason I’ve outlined would not allow unlimited growth; they would merely account for a rise in growth since 1950. It may not be appropriate to extrapolate similar growth for the future.
Moreover, even if we had crossed a time-sensitive mean, the sharpness of the decline makes it highly likely that we will overshoot the mean rather than simply return to it. That over-correction will be painful no matter what the mean-reverting level is.
Finally, it’s also interesting to note that GDP has increased because of house values rising – homeowners have been able to extract increasingly large portions of their home equity and transform that into consumption. In the last decade in particular, a large part of consumption has been driven by this mechanism. Thus, the denominator of the graph is growing as well, meaning the growth in real estate assets is even more dramatic than one might otherwise think!
So, what do you think is the “normal” level of real estate as a percentage of GDP? I’d write “long-term mean” but as you can see, I’m not convinced that the mean is stable over the long term at all.
Floyd Norris on China’s reported GDP figures:
China is becoming the world’s most energy-efficient economy.
Or maybe its statisticians are just the most creative.
Essentially, for the past decade China’s GDP has grown at a slightly slower rate than its electricity usage (which makes sense, since production requires energy). This year, GDP growth contracted sharply but remained positive at +6.1%. Norris notes that “[China] has not yet reported March electricity consumption, but during the first two months of the year electricity use was down 9.2% from 2008’s first quarter.”
Draw your own conclusions regarding the validity of the state-reported GDP number…