Research shows how a disaster can impact an economy beyond the simple rebuilding process. The Sichuan earthquake induced a lifestyle shift in households toward greater spending and a structural shift of the economy away from industrial production.

The researchers looked at three behaviors: saving, drinking, and playing mahjong. They found that, on average, people who were spared by the earthquake but lived closer to the epicenter tended to decrease their saving, increase their drinking, and play mahjong more often. This was not due to changes in their income, which was not affected by proximity to the quake. Figure 1 shows that while a change in income was not affected by distance to epicenter, expenditures increased more for households that were closer to it. Results also held true after the researchers took into account potential confounders such as gifts given to affected households out of solidarity, price levels, or government aid.

Note: Taken from Filipski et al. (2019). Changes refer to the 2007–2009 period. Percent changes approximated from difference in logs.  Data come from the Sichuan Rural Household and Migration Survey and from Research Center for the Rural Economy data.

Note: Households who were spared by the earthquake still increased their expenditures, more so when closer to the epicenter.

All three of these behaviors seem to reflect an Epicurean attitude toward life. Households closer to the epicenter seem to have become more willing to take risks, both with their money (saving less) and their health (drinking more). As for mahjong, it serves as a fun pastime but also doubles as a risk-taking activity, as players usually put up small money stakes to spice up the game. All behaviors suggest that rather than safeguarding against the next earthquake, households who lived through the tremors became more inclined, as the saying goes, to “live like there’s no tomorrow.”

The consequences of such a shift in attitudes are not straightforward to evaluate. The additional spending might give a slight boost to the local economy, but the drinking and gambling might not. It also remains unclear what happens in the longer term. The authors are able to show that the effects persisted until 2011, which was three years after the quake and long after reconstruction was finished—but after that, there is no data. It is possible that people who formed habits of spending, drinking, and playing mahjong after the quake kept them in the long run, but it is equally possible that the fear of another earthquake kicked in after some time and lead people back to more frugal ways.

The earthquake not only affected people’s consumption and savings behavior, but also affected sectoral growth at the local level. The earthquake areas received massive aid for relief and reconstruction. The inflow of aid bid up the prices of local services and local wages. Consequently, the manufacturing sector lost competitiveness as their product prices and wages must be in line with the national market. After the earthquake, there was noticeable shrinking in the share of industrial production in the earthquake areas. However, factories and infrastructure might have collapsed in the earthquake counties, directly affecting industrial production. To isolate the impact of rising local services and wages on industrial production, the article first examines the neighboring counties of earthquake counties, which did not suffer the earthquake. As shown in Figure 2 below, these counties also suffered in industrial production compared to other non-disaster counties. There is still a likelihood that the production network of the neighboring counties suffered from the disruption in the earthquake counties, inhibiting industrial production in the neighboring counties. To rule out this possibility, the paper further investigates investment flows between paired earthquake counties and their neighboring counties and found there is no apparent pattern from 2001 to 2015.

Figure 2 Share of industrial GDP in two types of non-disaster counties (%)

Note: Taken from Bulte, Xu, and Zhang (2018). Data from 2002–2014 come from the Sichuan Statistical Yearbook published each year from 2003–2015. “Non-disaster neighbors of paired disaster” are defined as 1 if they border at least one of the paired disaster counties but did not suffer the earthquake, and 0 otherwise. “Other non-disaster” is defined as 1 if a county did not suffer the earthquake and does not neighbor any paired disaster counties, and 0 otherwise.

The finding supports the Dutch disease hypothesis. Rising prices of non-tradables as a result of massive inflows of transfers, mainly pairwise aid, attracted workers to the local service sector from the manufacturing sector, which could not easily adjust output prices. Consequently, the profitability of the manufacturing sector suffered a contraction in the short run.

[Kevin Chen, Zhejiang University and International Food Policy Research Institute (IFPRI); Mateusz Filipski, University of Georgia and IFPRI; Xiaobo Zhang, Peking University and IFPRI.]

This blog was written based on the following two papers and edited book:

Book: Chen, Kevin Z., ed.; Zhang, Qiang, ed.; Hsu, Claire, ed. 2016. Earthquake lessons from China: Coping and rebuilding strategies. Washington, D.C.: International Food Policy Research Institute (IFPRI). http://dx.doi.org/10.2499/9780896298743

Additional References: