There are still two to three days before the peak flooding – and damage – in Houston arrives, yet already Wall Street is trying to calculate how much the worst natural disaster in decades befalling the 4th largest American city will impact US GDP. Or rather boost it, because in two notes out late in the day Monday, one from Goldman and one from JPM, the authors come to two polar opposite conclusions: Goldman claims that the Houston natural disaster will reduce Q3 GDP by as much as 0.2%, while JPMorgan predicts that the “net impact on Q3 and Q4 GDP should be positive.”
How do the two arguably most sophisticated investment banks reach such diametrically opposing conclusions? Here is the answer, straight from the horse’s mouth. First, Goldman:
Preliminary estimates suggest that total losses could reach $30bn (0.16% of GDP), but the uncertainty around these figures is considerable, with estimates ranging from under $10bn to as high as $100bn. Assuming $30bn, Hurricane Harvey would become the 9th largest since World War II—and roughly 40% the size of Sandy in 2012 and 15% the size of Katrina in 2005—in terms of property damage as a percent of GDP (see Exhibit 1).
Exhibit 1: Hurricane Harvey Could Be the 9th Largest since World War II in
Terms of US Property Damage
These property damages will not be directly visible in most economic indicators, which generally focus on the flow of production, sales, and employment (as opposed to the stock of existing wealth). In terms of economic activity, natural disasters typically have two important and generally offsetting effects in the regions affected: (1) sharp, temporary declines in demand and production, followed by a rebound to the pre-disaster baseline, and (2) a boost to output from the rebuilding of damaged property and from a catch-up in economic activity displaced during the disaster.
Before we get to the bottom line impact on GDP, Goldman considers the impact on other aspects of the economy, such as labor, retail sales, industrial production and construction spending.
In past research, we’ve also found substantial effects on jobless claims, which, for example, increased by 80k in the week after Hurricane Sandy and slowly declined in subsequent weeks. Given the extent of the disruption and damages, we would expect at least some impact from Hurricane Harvey on jobless claims and on some of the August and September monthly data releases, potentially including retail sales, industrial production, and the various housing indicators. In terms of the August employment report released this Friday, the payroll survey period ended two weeks before the hurricane and should not be affected. We believe it is too early to estimate an impact on the September employment report.
Speaking of Sandy and other prominent U.S. hurricanes, the pattern appears to be uniform: the exhibit below revisits Goldman’s analysis of four major economic indicators around 18 major hurricanes, including Hurricanes Katrina and Sandy.
While even the medians are noisy, a slowdown often occurs in the month of the hurricane in retail sales, construction spending, and (to a lesser extent) industrial production. Growth in these indicators tends to rebound significantly to above-trend over the following 2-4 months. In contrast, hours worked—which reflect changes in both nonfarm payrolls and the average workweek—show less consistent effects.
Exhibit 2: Retail Sales, Construction Spending, and Industrial Production Growth
Often Slow Around Major Hurricanes, Relative to Recent Averages
So while superficially, one could assume a “Keynesian broken window fallacy” is all one needs to spin a natural disaster as bullish for the economy, Goldman is at least smart enough not to fall for that trap, and notes that one critical consideration in this particular case, is the importance of the energy sector in the regions currently affected.
Our commodities team found that Harvey has already significantly disrupted oil refining in Texas, shutting down over 16.5% of total US capacity. Oil and gas production in the region has also declined as a result of the storms. They note that while most of these outages are preventive, the slow-moving nature of the storm and the risk of damage to energy infrastructure suggest these disruptions could persist into September. In Exhibit 3, we estimate that declining energy-sector output could directly reduce Q3 GDP growth by as much as 0.2pp (qoq ar), affecting the inventories and/or consumption components. This assumes a gradual reopening of refineries and oil wells consistent with historical experience (we assume significant disruptions persist for four weeks).
Exhibit 3: Declines in Energy Production and Refining Could Reduce Q3 GDP
Growth by as Much as 0.2pp
Still, not even Goldman is immune to the fallacies of Keynesian econ 101, and writes that “we would expect the associated drag on Q3 growth to reverse in subsequent quarters (with a roughly offsetting boost to Q4 growth)” and adds that “in the longer run”, the overall impact of the hurricane on second-half growth is “uncertain, as the negative effects are likely to be offset by an increase in business investment and construction activity once the storm has passed.”
With the caveat that uncertainty remains high, we note the possibility that the hurricane could shift the composition of growth from Q3 to Q4 to some degree, particularly if rebuilding efforts take time to ramp up.
In other words, Goldman expects Harvey to cut Q3 GDP by roughly 0.2%, while boosting Q4 by a similar if not greater amount.
What about JPM?
Well, in this case, JPM’s chief economist Michael Feroli has decided to skip the entire “dip” phase (or any assessment of what the near halt of local energy infrastructure means for the broader economy), and has proceeded straight to the “broken window fallacy” nirvana: the growth that destruction somehow always guarantees in a Keynesian world (if that was the case, why not just nuke the US and rebuild it from scratch, assuing triple digit GDP growth for decades to come). Here is JPM’s base, and so far only, case:
Hurricanes can disrupt economic activity—which subtracts from GDP—and they can destroy capital—which over time will support the flow of GDP as the stock of capital is rebuilt. The disruptive aspect of hurricanes sometimes simply shifts activity from one period to another; this seems most likely the case for goods production and consumption. For example, any lost retail spending during the storm will probably be made up in later periods. Disrupted activity in service-producing sectors, in contrast, may represent real foregone activity not made up later; since services cannot be inventoried, it is harder to smooth consumption over different time periods. In any event, the output and income generated by service-producing sectors is still poorly measured in the US, and we are not convinced we would see the adverse effect of disrupted serviceproducing activity in the monthly or quarterly data used to estimate GDP.
Another common effect of hurricanes—and Harvey is no exception— is disruption to petroleum refining and natural gas distribution. Gasoline prices and, to a lesser degree, natural gas prices have already moved higher. This could crimp consumer spending power in the near term. However, if the past is any guide the impact on energy prices should be short-lived as refineries come back on line in coming days and weeks.
Of course, assuming they come back online in coming days and weeks. So what is JPM’s conclusion:
Any assessment of the economic impact of Hurricane Harvey will be inherently subject to a fair degree of uncertainty, given the recency of the event. As a general rule, hurricanes tend to be a short-run depressant and a medium-run boost to economic activity. (Here we should pause to emphasize the usual disclaimer: economic growth does not always correspond with economic well-being). Sources within the insurance industry as well as J.P. Morgan’s insurance industry research team estimate that the physical damage will be in the $10-$20 billion range. This would put Harvey in the top ten costliest storms, but still well below the $159 billion estimated damage from Katrina. The midpoint of the $10-$20 billion range is about 0.1%-pt of GDP. Total damage, and total rebuilding, should be greater than this amount, as invariably there will be uninsured losses that will be repaired. Even taking this into account, we believe the overall impact on GDP in Q3 and Q4 should be positive but very small, consistent with the historical experience. For this reason, we are not changing our top-line GDP forecast. While Harvey may not be a game-changer for the expansion, it may leave its mark on the high-frequency economic data.
Ironically, JPM leaves of with an optimistic outlook, despite admitting that the logistical impact to Houston, whose metropolitan area account for $500 billion in output, could be severe in both the short and long term:
We begin by observing how much economic activity normally occurs in the areas affected by the storm. For simplicity, we focus on the Houston metropolitan area given its relative importance in terms of output, but acknowledge that other areas were also affected by Harvey. The Houston metropolitan area accounted for about $500bn output in 2015 (Figure 1, the latest available data by metro area). This is undoubtedly a meaningful figure—this level of output was close to the total output of Norway as a whole and made the Houston area the fourth largest metropolitan area in terms of output for that year. But we should remember that US GDP was $18tr that year, with less than 3% of that total coming from the Houston area.
Following that assumption that several hundreds billions in lost output should be rounded down, we stopped reading and instead focused on the following twitter comments from an environment engineer on twitter, whose prediction of what may very well happen, is far more valuable than either Goldman’s of JPM’s: