The Effect of Early Flood Warnings on Mitigation and Recovery During the 2010 Pakistan Floods - Reducing Disaster: Early Warning Systems for Climate Change - page 253

Table 13.6 Tobit: effect of warnings on household losses through mitigation action taken
Explanatory variables Value of household loss

Female −42670.4

(27,755.82)

Age 369.8491

(1,070.612)

Received any flood warning 250,922.3 ***

(38,973.59)

Took any mitigation action 191,057.4 **

(89,257.46)

(Received any flood warning) * (Took any mitigation action) −199,484 **

(93,602.24)

Flood propensity score 24,199.28

(71,034.4)

Ln savings −5,074.22

(3,429.241)

Constant −204,781

Observations 65,118.5

384

Standard errors in parentheses
*** p < 0.01; ** p < 0.05; * p < 0.1

possessions. Time preference (preferring immediate payment to deferred income 8 )
significantly reduces the likelihood of mitigation activity. Meanwhile, holding an
insurance product (not shown here) is negative but not significant, and receiving
external assistance (e.g., from neighbors) with mitigation or being related to power-
ful people (patronage) is not significant.
Next, we test whether receiving a warning and taking mitigation action impacted
the actual value of household losses. As shown in Table 13.6 , loss value is positively
correlated with both receiving warnings and taking mitigation action, as households
with high flood exposures are more likely to receive warnings, take mitigation
action,orexperienceloodlosses.However,thecoeficientontheinteractionterm
of warning and mitigation is significantly negative, indicating that increases in flood
warnings make mitigation behaviors more likely to reduce losses.
Finally, we tested whether taking mitigation actions increased household resil-
ience. 9 Table 13.7 shows that mitigation activities significantly increase the likelihood
of having recovered or replaced household possessions at the time of the survey.
This is consistent with our expectations, since the mitigation activities named in the
survey mostly relate to household possessions and structure (reinforcing structure,
using sandbag barriers, moving possessions to higher level/ground).
Our results provide a substantial contribution to the literature on early warnings.
In contrast to the popular view that flood-exposed Pakistani residents ignored

8 This variable was measured with a hypothetical survey question asking about their preferences for
a sum of money today or a larger sum six months in the future.
9 Our survey included measures of (self-reported) resilience in terms of losses experienced, loss
recovery, and time to recovery of various possessions and livelihoods.