What families lose when they hold on

We readily praise the resilience of families after a hurricane. But behind that word lies a sequence of sacrifices, each more costly than the last.

The previous posts dealt with the economy as a whole: growth trajectories, the measurement of cost, the recovery scenarios. This one changes scale and turns to households. In the weeks after a hurricane, an affected family has to make choices: drawing on savings, borrowing, asking a relative for help, selling a tool of the trade, sometimes taking a child out of school. These choices carry a cost, often an invisible one, and they help explain why the losses from a disaster last long after the event itself.

Solidarity, and the wall it runs into

Let’s start with what works. Mutual aid between relatives and neighbours is a reality, not a comforting cliché, and economists have documented it precisely.

The reference study remains Robert Townsend’s (1994), conducted in Indian villages. It shows that household consumption there varies rather little with a family’s own difficulties. When one household loses its harvest while its neighbours are spared, mutual aid makes up much of the loss: informal insurance does its job.

Townsend also establishes the limit of that mechanism, and it is the limit that concerns us here. The protection holds because misfortunes are staggered in time and spread across households: whoever helps today will be helped tomorrow. It disappears when everyone is hit at once. A cyclone is exactly that second case, since it reaches not one family but an entire territory, and those who might help have been struck themselves. This is not a matter of generosity but of pooling: a risk that hits everyone at the same moment cannot be shared within the group.

The diaspora exception

There is, however, one way out: money that comes from outside.

In the Philippines, Dean Yang and Hwa Jin Choi (2007) showed that relatives settled abroad send more money home when the family left behind runs into trouble. When a household’s income falls, transfers rise and offset roughly 60 percent of the loss. Unlike local mutual aid, this form of insurance withstands collective shocks, precisely because the person sending the money lives elsewhere and was not hit.

Two qualifications matter, though. First, this protection only applies to households that actually have a relative abroad, which creates yet another inequality between families, depending on whether one of their own has left. Second, even at 60 percent, the compensation remains partial. The hole is not filled, it is made smaller.

The hierarchy of sacrifices

What happens when neither mutual aid nor relatives abroad are enough? Households then follow a fairly regular order, observed across many countries.

You draw on savings first, when there are any. You borrow next, formally or otherwise. And when those options are exhausted, you sell. First comfort goods, then, as a last resort, productive assets: the vehicle used for work, the tools, the stock, the boat, the livestock. These sales happen under pressure, therefore at bad prices, often at the very moment when everyone else is selling the same thing.

It is this last step that changes the nature of the problem. As long as you are selling comfort goods, you lower your standard of living without compromising your future income. When you sell your tool of the trade, you cut into your future capacity to earn a living. The disaster then stops being a rough spell to get through: it durably alters the household’s trajectory.

That order is not universal, though. In Ethiopia, Michael Carter and his coauthors (2007) find that the poorest households often do the opposite: they cling to their few assets and cut back on food first. The logic is merciless. When a single draught animal is all that remains, selling it means giving up any possibility of getting back on your feet. Better to go hungry today than to have nothing at all tomorrow.

Nor does every household reach that last step at the same pace. After the 2005 Mumbai floods, Archana Patankar and Anand Patwardhan (2016) measured household losses, and the gradient is striking: for the poorest, they amounted to the equivalent of more than six months of income, against four to five months for low-income households, roughly three months for the middle class and two months for the better-off. In the same city and the same flood, the poorest households thus lost three times as much as the better-off, relative to their income. For them, the moment when the tool of the trade has to go comes much sooner.

And these averages still hide what matters most. In Bangladesh as in Pakistan, median losses are roughly half of mean losses: a minority of households concentrates most of the damage. Behind an aggregate figure that looks bearable, then, sit a few families for whom everything collapsed.

The threshold beyond which there is no way back

Economists have given that point of no return a framework. Michael Carter and Christopher Barrett (2006) proposed thinking about poverty in terms of assets rather than income, and this approach allows an essential distinction: between transitory poverty, which time resolves, and structural poverty, which it does not.

The idea is that there exists a threshold of assets below which rebuilding becomes almost impossible. Above it, an impoverished household gradually recovers. Below it, it no longer has enough assets to generate the surplus that would let it climb back, and it settles durably into poverty. Every distress sale moves a household closer to that threshold. Every additional disaster pushes a few more families across it.

That threshold has been measured in practice, after a hurricane. Carter and his coauthors (2007) tracked rural Honduran households hit by Mitch in 1998. Their first finding is counter-intuitive: 44 percent of households lost productive assets, and that share rises with wealth, from 22 percent in the poorest quartile to 68 percent in the richest. The wealthiest were therefore hit most often, simply because they had more to lose.

But what follows overturns that picture. Better-off households recovered within a few years, whereas among the poorest the effects lasted far longer and bit far harder. More painful still: before Mitch, poorer households were accumulating assets faster than richer ones, and the hurricane cut that catch-up short. The authors estimate a threshold at around $250 of assets, below which households become durably trapped in poverty. They run the same exercise in Ethiopia, among households hit by drought, and place the threshold there at the equivalent of barely more than half an ox. The authors themselves stress that these estimates rest on strong assumptions, but the direction of the result is clear.

Long-run data confirm this inertia. In Ethiopia, Stefan Dercon (2004) measures the scale of the asset collapse caused by the mid-1980s famine: between distress sales and animals lost, the share of households still holding significant livestock fell from around 75 percent before the crisis to about half by 1989. And the imprint of that shock can be read long afterwards. Rainfall shocks weigh on consumption growth a decade later, and the severity of the famine still explains trajectories in the 1990s. The past does not pass.

The cost that strikes before the disaster

Which brings me to the most surprising finding in this post, and to my mind the most important.

So far we have discussed what a disaster costs once it has happened. But merely living under the threat of disaster already carries a cost, and an enormous one. Chris Elbers, Jan Willem Gunning and Bill Kinsey (2007) studied Zimbabwean farmers exposed to climate risk. Their estimate: these households’ capital stock is 46 percent lower than it would be in the absence of risk.

The most striking part lies elsewhere. About two-thirds of that gap does not come from destruction actually suffered. It comes from behaviour adopted in the face of risk: you do not invest in the equipment that would raise your yield, you prefer less productive but safer activities, you keep cash rather than tie it up. In other words, most of the cost of risk is paid for disasters that never happened.

This is a complete reversal of perspective. We are used to counting damage after the event. Yet the bulk of the bill is levied continuously, in the form of investments never made, businesses never started, projects never attempted. If the finding carries over, an exposed island would be poorer not only because it gets hit, but because it lives in anticipation of being hit.

What actually works

Should we conclude that nothing can be done? No, and two recent findings show where to act.

The first concerns insurance, and it is subtler than it looks. In Kenya, Sarah Janzen and Michael Carter (2019) evaluated an index insurance scheme tied to livestock losses. It works, but not in the same way for everyone. Among wealthier households, whose main risk is losing their capital, it cuts the likelihood of selling assets massively, by 96 percentage points. Among poorer households, whose consumption barely covers the essentials, it cuts the likelihood of skipping meals by 49 percentage points. The same product protects the assets of some and the food of others. Judging such a scheme on a single criterion, asset sales or skipped meals, would therefore capture only half of what it delivers.

The second concerns cash transfers. Patrick Premand and Quentin Stoeffler (2022) evaluated multi-year programmes in the Sahel, paid out independently of any particular event. These transfers raise consumption by around 10 percent, but above all, for households later hit by drought, the welfare gain exceeds the amount of the transfer itself. The reason lies in the savings and income smoothing they make possible. A safety net already in place when the shock hits is worth far more than the same sum paid out in the emergency, once disaster has struck.

Resilience is not a virtue, it is an infrastructure

We often use the word resilience as a compliment, to salute the courage of those who absorb the blow. This post argues the opposite: what we take for courage is most often a sequence of costly sacrifices, and the admiration we express sometimes spares us from looking at what they cost.

Real resilience is not measured by the courage of families, but by what they have available at the moment of the shock: savings, insurance that compensates quickly, a social safety net already in place, access to credit that does not dry up. And those schemes still have to reach the people who need them most, which is far from guaranteed. Hallegatte and his coauthors (2017) note that in Nepal, after the 2011 floods and landslides, barely 6 percent of very poor households sought government support, against nearly 90 percent of well-off ones.

This is also what ties this post to the earlier ones. If well-being losses exceed asset losses, if trajectories diverge instead of converging, it is largely because at household level, each emergency adjustment erodes a little of their capacity to recover. Understanding these mechanisms precisely, and how existing protection schemes do or do not soften them, remains largely to be done. To my mind, it is one of the most useful lines of enquiry in disaster research.

References

Brouwer, R., Akter, S., Brander, L., & Haque, E. (2007). Socioeconomic Vulnerability and Adaptation to Environmental Risk: A Case Study of Climate Change and Flooding in Bangladesh. Risk Analysis, 27(2), 313-326.

Carter, M. R., & Barrett, C. B. (2006). The Economics of Poverty Traps and Persistent Poverty: An Asset-Based Approach. Journal of Development Studies, 42(2), 178-199.

Carter, M. R., Little, P. D., Mogues, T., & Negatu, W. (2007). Poverty Traps and Natural Disasters in Ethiopia and Honduras. World Development, 35(5), 835-856.

Dercon, S. (2004). Growth and shocks: evidence from rural Ethiopia. Journal of Development Economics, 74(2), 309-329.

Elbers, C., Gunning, J. W., & Kinsey, B. (2007). Growth and Risk: Methodology and Micro Evidence. World Bank Economic Review, 21(1), 1-20.

Hallegatte, S., Vogt-Schilb, A., Bangalore, M., & Rozenberg, J. (2017). Unbreakable: Building the Resilience of the Poor in the Face of Natural Disasters. Washington, DC: World Bank.

Janzen, S. A., & Carter, M. R. (2019). After the Drought: The Impact of Microinsurance on Consumption Smoothing and Asset Protection. American Journal of Agricultural Economics, 101(3), 651-671.

Kurosaki, T. (2012). Household-level Recovery after Floods in a Developing Country: Further Evidence from Khyber Pakhtunkhwa, Pakistan. PRIMCED Discussion Paper Series, No. 27.

Patankar, A., & Patwardhan, A. (2016). Estimating the uninsured losses due to extreme weather events and implications for informal sector vulnerability: a case study of Mumbai, India. Natural Hazards, 80(1), 285-310.

Premand, P., & Stoeffler, Q. (2022). Cash transfers, climatic shocks and resilience in the Sahel. Journal of Environmental Economics and Management, 116, 102744.

Townsend, R. M. (1994). Risk and Insurance in Village India. Econometrica, 62(3), 539-591.

Yang, D., & Choi, H. J. (2007). Are Remittances Insurance? Evidence from Rainfall Shocks in the Philippines. World Bank Economic Review, 21(2), 219-248.