Goldman's Magnum Opus On The Economic Impact From Japan's Earthquake

Goldman's Dominic Wilson has just released his magnum opus analysis on the situation in Japan and its aftermath. Unlike the lunatic drivel disseminated each and every week by GSAM's Jim O'Neill, which would interpret the imminent collapse of the Earth into a neutron star as the most bullish event in the (soon to be over) history of mankind, this report is actually one of more biased we have read from the Goldman strategists. That said, the natural bias to spin everything in a positive light still dominates the report (to which we retort: why not just blow up unpopulated pieces of America and rebuild them over and over: it does miracles for the Chinese "GDP" - it should work just as well in the US - plus Krugman would be in a constant state of Keynesian extasy). We obviously disagree: never before has there been the added precedent of nuclear fallout, or a pyschological intangible, to the mostly superficial infrastructural repairs that have to be undertaken. Goldman does acknowledge this as follows: "The ongoing nuclear risk at Fukushima has the potential to magnify the impact in other ways, although at the time of writing these risks seem to be fading." Based on what? Manipulated data reporting out of Japan, TEPCO and everyone else who is either guilty of strategic mismanagement, or is desperate to avoid a worldwide panic. It is this type of rushing to conclusions based on a complete lack of facts, that drives objective market observers furious with blind rage at the idiocy of the authors (and yes, "idiots" is what Sean Corrigan called all those who only see the upside in the Japanese catastrophe and ignore the massive human, economic and financial downside). Yet for all those who can bottle their rage for 15 minutes, we recommend reading the enclosed report as it is the very best that the Kool Aid crowd can serve at this point. Which, when applying some common sense, is not very much.

From Goldman Sachs

The Economic Impact of Japan’s Earthquake


As markets assess the impact of the recent tragic earthquake and tsunami in Japan, we attempt to place this issue into a broader perspective. While there can be no perfect analogue in such a disaster, studying previous ones may provide a sense of the probability distribution around the macro and market impacts of such events, and their trajectory.

Natural disasters tend to result in a strong contemporaneous reduction in average GDP growth followed by a quick rebound in the following quarter. Even in the case of very large disasters (in a ‘global’ rather than ‘local’ sense), the impact on economic growth fades quickly from the data, in part because in most (but not all) cases the boost from government spending offsets the fall in activity in other parts of the economy. The impact in asset markets has generally been more muted than the economic impact.

That said, the impact of the East-Japan earthquake and tsunami may be greater. This is likely to be one of the most costly disasters in global GDP terms; the ongoing nuclear risk has the potential to magnify the impact in other ways; power disruptions could last longer than normal; and supply chain disruptions may be an issue. While our Japan forecasts are under review, we are not planning any significant changes to our global forecasts.

From a global markets perspective, the broader macro context is important, and at the margin, positive. The Middle East crisis remains a key source of uncertainty. But data is likely to confirm global cyclical strength, European sovereign risks have calmed a little, and there are tentative signs that the most intense EM tightening phase may pass.

We extend our deepest sympathy to all those whose lives have been affected by the recent East-Japan earthquake.

As markets assess the impact of the recent catastrophic East-Japan earthquake and tsunami, in this week’s commentary we try to place this issue into a broader perspective. While there can be no perfect analogue in such a disaster, studying previous large disasters may help to gain a sense of the probability distribution around the macro and market impacts of such events, and their trajectory.

Natural Disasters in History

We have focused on three types of disasters (earthquakes, floods and storms) that are similar in nature: exogenous shocks with immediate destructive impact. We obtained disaster data from the EM-DAT International Disaster Database provided by the Centre for Research on the Epidemiology of Disasters (CRED). Over 6,612 disasters of these types have occurred since 1980, so we narrowed our sample by focusing on those in which the estimated direct economic damage (to buildings, production facilities, etc.) was greater than 1% of the preceding year’s GDP. We also eliminated disasters in countries without sufficient macroeconomic data available during the crisis period. This leaves us with a sample of 40 disasters in 23 different countries, nine of which occurred in the developed world (DM) and the remaining 31 in emerging market (EM) countries. Some familiar recent disasters in this group include the Hanshin earthquake in Kobe, Japan (1995), Hurricane Katrina in the US (2005), and the Indonesian earthquake and tsunami (2004).

Earthquakes are by far the most destructive in terms of both human impact and direct economic damage. The human impact is much larger in natural disasters in EMs, where nearly three times as many people are affected. But the direct property damage (relative to GDP) is much larger in the developed world, likely because there is a more valuable capital stock already in place when disasters strike. Table 1 contains some basic summary statistics about the magnitude of these crises, as measured by the estimated property damage and the number of people affected and/or killed by the disaster.

A Short-Lived Hit to Growth on Average

In order to explore the short-term economic impact of a large-scale natural disaster, we looked at quarterly and (where available) monthly data for a range of economic and market variables. Looking at averages does, of course, mask a great deal of variation, but some clear lessons still apply.

Turning first to GDP, we find two main results:

  • Natural disasters result in a strong contemporaneous reduction in average GDP growth followed by a quick rebound in the following quarter. Sequential growth is reduced by an average of approximately 3ppts (annualised) in the quarter in which the disaster occurs, although this result is driven mainly by EMs. Output falls by an average of 0.55ppts below its pre-disaster trend before rebounding nearly all the way back in the next quarter.
  • More destructive disasters produce a larger hit to growth. The size of the one-quarter hit to growth is strongly and negatively correlated with the size of the natural disaster on all three of the measures presented in Table 1.

The short-lived impact on economic activity is also confirmed by a range of other variables and is consistent with recent academic studies. Sequential IP growth falls sharply in the disaster month but then rebounds in the subsequent month. Export and import growth also both fall, in line with the overall reduction in output growth, before recovering in the next quarter. Investment growth dips but then accelerates strongly in the next 1-2 quarters, which suggests that the rebuilding of destroyed capital stock is an important component of post-disaster GDP resilience.

Policy rates tend to be reduced marginally on average, although this response generally is delayed until a few months after the disaster and there is a large degree of variation across cases. Equity returns dip in the month of the disaster but remain firmly in positive territory (at around 1%mom non-annualised on average), and then rebound sharply in the next month. Disasters have historically had a negligible impact on average on the short-term dynamics of inflation, government budget deficits and either nominal or real TWI exchange rates.

The Largest Global Disasters Tell a Similar Story

In addition to the lessons from the aggregate sample of disasters considered above, it is also helpful to look more closely at some of the globally most expensive disasters in recent history—since they are likely to provide the closest analogue to the East-Japan earthquake and tsunami. We sorted our sample of disasters by the estimated damage as a percentage of global (rather than local) GDP and, on this basis, the big five in descending order are the Hanshin (Kobe) earthquake of January 1995, Hurricane Katrina in August and September 2005, the Irpinia earthquake in Southern Italy in November 1980, the Sichuan earthquake in China in May 2008 and the earthquake in Los Angeles in 1994. The damage in each of these disasters exceeded a tenth of a percent of global GDP in that year (see Table 2). Restricting the sample in this way results in a sample biased towards the large advanced economies (with the exception of China in 2008). On the other hand, these are not among the most severe disasters within the context of local GDP, as Table 2 also shows.

The conclusions from looking at these big five disasters are qualitatively very similar to the results of the full sample. But the scale of economic and market moves in these cases may provide a better reference point for the East-Japan earthquake compared with the EM-heavy sample analysed above:

  • Even in the case of these large disasters in a ‘global’ rather than ‘local’ sense, the impact on economic growth fades quickly from the data. Notwithstanding the intense human and social costs of such large global disasters, the economic cost barely registers in quarterly economic data. In all four cases with available data, GDP growth was positive in the quarter in which the disaster happened. Output subsequently returned to its pre-disaster trend within one quarter, with the exception of the Chinese Sichuan earthquake of 2008 (although this likely reflects the concurrent global downturn).
  • The impact of the disaster is more clearly discernible in monthly activity data. Industrial production is the most reliably available across countries and over time. In the case of the Hanshin earthquake in Kobe, Japanese industrial production fell -2.6% mom in January 1995, but positive growth (+2.2%) was restored in the very next month. For the three months after the January earthquake, IP growth in Japan averaged 1.5%mom, double the average of the three months prior to the earthquake. And industrial production was above pre-earthquake levels within two months, by the end of March. In the case of Hurricane Katrina in 2005, IP growth was flat in August and down -2% in September, but growth was positive in October, and the pre-hurricane level of industrial production was surpassed by the end of November. The other three episodes were even less impactful: IP growth dipped to 0.5%mom in the month after the LA earthquake, and was low but positive in the months of the China and Southern Italy earthquake. Of course, the economic impact is greater if one zooms in on the region most directly affected by the natural disaster. For example, large-scale retail sales in the Hyogo and Osaka area fell sharply over the January-March 1995 period (-6.7%yoy, -3.4%yoy and -1.7%yoy). But even here, there was positive growth by April, and after drifting sideways for much of the rest of the year, large-scale retail sales rebounded strongly in January-March 1996.
  • Part of the reason that growth recovered quickly is that in most (but not all) cases, the boost from government spending offsets the fall in activity in other parts of the economy. In three of the five large episodes—the Hanshin Earthquake, the Southern Italian earthquake and Hurricane Katrina—government spending grew strongly in the quarter when the natural disaster occurred or in the quarter just after. The contrast is especially clear in the case of the Hanshin earthquake in Japan. In 1995Q1, real GDP grew 0.8%qoq, within which both private consumption and investment spending fell, but government spending increased by 2.8%qoq.
  • From a markets perspective, the impact has generally been even more muted than the economic impact. We only find a clear impact on equities in the largest disasters, the Hanshin earthquake and Hurricane Katrina, and in general there are few if any persistent moves in rates and FX markets. In the case of the Hanshin earthquake, the Nikkei 225 fell about 8% in the week following the disaster, but it soon rebounded and had recovered more than half its losses in the week thereafter. Bond yields (and the Yen) barely moved over this period, and continued a sustained downtrend, largely owing to the gradual economic slowdown in the quarters thereafter.

Key Differences with the East-Japan Earthquake

The evidence from historical disasters is helpful to gain a sense of the probability distribution around possible economic outcomes and their likely trajectory. But there are unique features of the East-Japan earthquake and tsunami currently, most of which suggest that the impact may be greater. We highlight five key points:

  • Even as the full extent of the damage in the East-Japan earthquake and tsunami is being assessed, it is already clear that this is likely be one of the most costly disasters in global GDP terms. According to the estimates by our Japan economics team, the total damage will be about ¥16trn, or around 1.6 times greater than the Hanshin earthquake, which was the most costly disaster before this in our sample.
  • The ongoing nuclear risk at Fukushima has the potential to magnify the impact in other ways, although at the time of writing these risks seem to be fading. A significant nuclear risk event, apart from the negative impact in the immediate affected area, is likely to affect consumer sentiment more broadly in Japan and potentially in other countries too. So the second round of economic impact from these types of developments could be significant. The nuclear dimension has also added a channel for international contagion as countries such as Germany have accelerated the shutdown and review of certain ageing nuclear power plants, putting more pressure on gas and oil prices, with the consequent deleterious effects on growth.
  • A prolonged disruption to the power network in Japan is a significant risk to Japanese growth and a material difference from many previous large disasters where power disruptions were mostly limited and local. The situation is evolving daily but damage from the earthquake has caused a shutdown of about 10%-12% of power station capacity throughout Japan. If power outages do not extend beyond end-April (our current base case), then after a contraction in 2011Q2, we expect +2% growth in Q3—a pattern not dissimilar to the historical evidence. However, if in a worst-case scenario, power disruption persists into late summer or even to end-December, our Japan economists estimate that GDP could decline until the end of the year.
  • Although the affected regions are somewhat further removed from Japan’s industrial heartland relative to the Hanshin earthquake, the fact that Japan is such a key part of the global industrial supply chain means that disruptions in specific industries and output could be material. Our equity analysts believe that sectors where there is likely to be a significant impact on sales include semiconductors, cellphones, digital cameras, petrochemicals and autos, whereas they see a relatively smaller impact on the machinery and steel sectors (Assessing earthquake impact risk on production in key industries, Shin Horie, March 21, 2011). Mike Buchanan and team have combined this micro industry level data with country trade linkages and see modest downside risks to growth in Singapore, Taiwan, Thailand, the Philippines and Korea, but little impact on China or India. A key mitigating factor is that in many industries inventories are sufficient to meet component demand for around six weeks. Still, disruption that lasts much longer will mean more risks of a kind that a purely ‘macro’ perspective may miss.
  • Lastly, part of the reason why the typical economic growth impact from natural disasters is fairly short-lived is the offsetting boost from government spending. We expect this offset this time around too. Chiwoong Lee’s latest note (Japan Economic Morning Pitch: ‘Financing Earthquake Reconstruction Still Uncertain’, March 22) lays out some of the options currently being considered in the media: (i) using the FY2010 and FY2011 emergency funds (¥1.2trn together), (ii) revising the FY2011 DPJ manifesto (¥3.3trn in total), and (iii) using government reserves (¥2.5trn). What cannot be covered by these sources will require increased issuance of new JGBs, but the scope for policy flexibility is more restricted today at least relative to the 1995 Hanshin earthquake. In 1995, interest rates were still above 3% and the size of the fiscal deficit was smaller (92% of GDP) than it is today (221% GDP). Japanese financial conditions have tightened since the earthquake, though the joint intervention on the JPY by major central banks has provided an important interruption to that dynamic.

Our Japanese forecasts are currently under review, although the key uncertainty around the growth picture centres around the longevity and magnitude of power disruptions. Reflecting the downside risks to earnings, Kathy Matsui and our Asian Portfolio Strategy team have already shaved 12% off their FY2012 earnings estimate for Japan’s equity markets, and we now expect returns here to be more back-loaded.

At this stage, we are not planning any significant revisions to growth forecasts outside Japan, although any changes there would mechanically influence the global forecast. The impact through export channels from temporarily lower Japanese demand is likely to be relatively small, even in non-Japan Asia where the linkages are tightest. And the other main source of transmission—through global financial conditions—is so far not registering as significant.

From a Global Markets Perspective, the Broader Macro Context is Mostly Constructive

While the earthquake has dominated headlines, it is important to keep an eye on other key developments that have occurred over the last two weeks in assessing the broader context in which these impacts are playing out.

Alongside the Japanese disaster, the other source of ‘headline’ risk for markets in recent weeks has come from developments in the Middle East. Here too, we have seen significant news lately. Gulf forces have entered Bahrain and protests continue both there and in Yemen. In addition, a UN Security Council resolution authorising a no-fly zone and escalating sanctions against Libya have already resulted in military strikes against government forces. In both cases, this could set the stage for greater stability in the region. But it is still difficult to be sure exactly what paths the regional crisis will take and interventions so far do not provide a decisive resolution. This is therefore a source of uncertainty that may remain with us even if concerns over the consequences of the Japanese earthquake subside.

The major transmission to the broader outlook remains through oil supply disruptions. We have had a structural view that oil prices are likely to be under upward pressure in 2011. For this reason, we have generally sought out energy exposure, directly and within equities and FX. Disruptions to nuclear capacity in Japan may ultimately reinforce those structural pressures. But our central forecast remains that the upward trend in oil prices will not prove to be a binding constraint on growth. That said, further supply shocks would certainly see fresh spikes in oil prices given low inventory buffers, so we remain vigilant about this source of risk. Our ‘oil-adjusted’ Financial Conditions Index in the US—a simple way of weighting the impact of financial conditions and oil prices on growth—tightened significantly in late February but has so far been fairly stable in March.

Beyond the Middle East, three other developments that are high on our radar screens look a little more benign. The first is the broader cyclical landscape. Last week's Philly Fed release was extremely strong and, as we move through the most macro-intensive part of the calendar, we expect the bulk of the data to reinforce a message of robust underlying growth. While the literal tracking of US GDP growth has been a little softer than our current forecast, the survey data are consistent with stronger underlying growth, so we will likely see some reacceleration in Q2. As we have shown recently (and as displayed in Chart 6), because the growth recovery has been the major story behind the rally since the end of August, the market has generally performed better in the data-intensive part of the month (between the Philly Fed and payrolls), with losses on average outside that period. If our positive growth view is borne out, this pattern could continue, as so far in March.

The second development that has been pushed off the headlines but that we expect to return this week is the continuing push for further measures to calm sovereign fears in Europe. The March 9 European Council meeting was, as we said at the time, a somewhat mixed result that has left many details still to be hammered out. But the fact that a deal was reached early was positive at the margin, as was the upsizing of the European Financial Stability Facility (EFSF) funding and the renegotiation of Greek borrowing terms, even if progress on Portugal and Ireland was more disappointing. And with little fanfare, we have seen Spanish and Italian spreads tighten meaningfully, the latter to levels not seen since last August.

This week’s EU summit on Thursday and Friday is expected to ratify those decisions, but expectations for any fresh developments beyond that are low. The risk is that markets will focus once again on what has not yet been resolved and the vagueness of the future structure of penalties for fiscal lapses. But, as Francesco Garzarelli has argued in the past, there is ‘endogenous risk’ in the Euro-zone crisis and the relaxation seen lately in European credit markets may itself have lowered the risk that this issue becomes a source of greater shocks for the market in the next few months.

Finally, we continue to watch the EM tightening dynamic closely. Since early November, inflation pressures and tightening responses in China and more broadly across EM have kept us away from EM equity assets, despite a relatively constructive medium-term view. We have argued in several places before that while the sharp underperformance of EM equities since then is leading us to ‘warm up’ to the asset class, some key ingredients of a more positive view have so far been absent—such as signs that the peak in the tightening cycles is at hand, easing of sequential inflation pressure and more concrete signs of slower growth.

While we still do not have clear evidence that these conditions have been met, we may be moving further down that road. The past few weeks have provided evidence that China’s economy is slowing, suggesting that the tightening that our Financial Conditions Index has been signalling may be starting to bite. More intriguing is the break in the upward trend in agricultural prices, which Themos Fiotakis discussed in a Global Markets Daily last week, with recent market damage reinforcing a drop in global agricultural prices that had begun on the first evidence of modest inventory builds. Damien Courvalin continues to remind us that March 31 is the first real news day for the new crop and that, while high prices should incentivise a supply response, the crop outlook will not be clear until well into the summer, with substantial fragility to sub-par weather given low inventory buffers. But with ‘normal’ weather, crop prices are likely to fall and the inflationary impulse from food prices may be peaking for now. If that dynamic continues, it would greatly add to the attractiveness of EM equity positions. And while it is early to be drawing any firm conclusion here, we think the market is paying too little attention to this dynamic.