The First Half of the Story: From Loss to Reallocation
The subject at this point is the incumbent side of post-flood economic change: how a shock selects among plants, damages their capital, and produces different paths for those that close and those that remain. The evidence already supports a conditional chain from flood damage to selective exit and survivor adjustment. The judgment it supports is narrower than “disasters cleanse economies”: the first half of the mechanism is established, but it is not complete until we know whether released resources reach capable new users and whether financing lets them operate.
That distinction is useful because it turns a collection of findings into an audit. A convincing account must pass through each transition. The flood must change plant outcomes; exit must show some productive selection; survivors must adjust capital rather than merely disappear from the sample; and the resulting resources must have somewhere productive to go. If the chain stops at survivor averages, it describes selection and rebuilding, not yet reallocation.
What the incumbent-side evidence establishes
The pattern begins with a negative shock to production, but the important result is not simply that floods cause harm. The post-flood exit rate rises, and the increase is stronger among plants that were already in the bottom quartile of productivity in their narrowly defined industries. That combination says something about the composition of the incumbent population: the flood adds pressure to plants that were less productive before it arrived. The 3.6 percentage-point increase in the probability of exiting within two years and the additional 1.1 percentage-point disadvantage for the lowest-productivity group are evidence of a selective destructive margin.
The selection finding changes the interpretation of a later average. If the least productive plants are more likely to leave, the set of operating plants is no longer the same set that existed before the flood. A higher average can therefore arise partly because the population has been filtered. But the filter is not the whole mechanism. The plants that leave may have supplied jobs, relationships, and usable machines. Their equipment may be sold, damaged, or left idle. Selection tells us which production units disappeared; it does not tell us whether their resources improved anyone else’s production.
The survivor evidence adds a second transition. Plants that remain in business for at least three further years show more capital retirement, more repair and refuse-removal activity, and positive net investment. The coexistence of retirement and investment is economically meaningful. It indicates that survivors are not merely waiting for normal operations to resume. They are dealing with damaged capital and making new capital commitments. Net investment rises after accounting for the capital retired, and county-level investment results point in the same direction. The adjustment therefore appears broader than a statistical artifact created by a few unusually strong plants.
The later survivor TFP path completes the incumbent-side pattern but not the regional story.