Poverty & Equity Global Practice Working Paper 155



QUANTIFYING THE IMPACTS OF CAPTURING
TERRITORY FROM THE GOVERNMENT IN THE
REPUBLIC OF YEMEN




                                                     Sharad Tandon
                                                         May 2018
Poverty & Equity Global Practice Working Paper 155




  ABSTRACT
  This paper estimates the welfare change arising from the capture of the Republic of Yemen`s capital in 2014,
  using a multi-themed household survey conducted as the capital was captured. Despite the little violence in
  this setting, the increase in fragility resulted in a large decline in household welfare driven by both a decline in
  income and an increase in food prices. Beyond traditional welfare metrics, women were affected by the fragility
  more so than men, where there was a nearly universal drop in women’s decision-making ability that did not
  differ based on a woman’s bargaining position in the household. Furthermore, this decline in decision making
  was immediate, and did not continue to worsen in the months towards the end of the period when household
  welfare dropped the most. Lastly, the tumultuous setting had implications for individuals’ ability to report their
  subjective welfare in accordance with their unambiguous decline in traditional welfare metrics.




This paper is a product of the Poverty and Equity Global Practice Group. It is part of a larger effort by the World Bank to
provide open access to its research and contribute to development policy discussions around the world. The authors may be
contacted at standon3@worldbank.org.


 The Poverty & Equity Global Practice Working Paper Series disseminates the findings of work in progress to encourage the exchange of
 ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully
 polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions
 expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for
 Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank
 or the governments they represent.

                                                          ‒ Poverty & Equity Global Practice Knowledge Management & Learning Team


                         This paper is co-published with the World Bank Policy Research Working Papers.
    QUANTIFYING THE IMPACTS OF CAPTURING TERRITORY FROM THE
                       GOVERNMENT IN THE REPUBLIC OF YEMEN




                                              By Sharad Tandon∗




Keywords: Conﬂict, Poverty, Gender, Republic of Yemen

JEL classiﬁcation: D12, E24, J16, O12, O53




   ∗
     The World Bank, 1818 H St. NW, Washington, D.C. 20420, USA, standon3@worldbank.org, +1-202-458-4043. I
would like to thank Anna D‘Souza, Benu Bidani, Ruth Hill, Sailesh Tiwari, and Tara Vishwanath for comments on ealier
drafts. I would also like to thank audience members at the World Bank. All remaining errors are attributable to the
author. The views expressed here are those of the authors and may not be attributed to the World Bank.
1. Introduction

       Fragile and conﬂict settings are becoming increasingly important to poverty reduction, where the
share of the world’s poor living in fragile settings is expected to increase from 17 percent in 2017 to
approximately 60 percent by 2030.1 However, in some regions, conﬂict is already the primary driver of
extreme poverty. In the Middle East and North Africa, there have been civil wars in Iraq, Libya, the
Syrian Arab Republic, and the Republic of Yemen in the past decade that have profoundly aﬀected
welfare in each of these countries. The civil war in Syria has also aﬀected the welfare of neighboring
states- Lebanon and Jordan in particular- through a surge in refugees, a decline in security, and a
worsening economic situation (e.g., Krishnan et al. 2017). Additionally, there has been a surge in
terrorist activity since 2011 in the region in general, but the surge has disproportionately aﬀected Iraq
and Syria (e.g., Institute for Economics and Peace 2017). Of the entire Middle East and North Africa
region, approximately 25 percent of the population has been exposed to these violent or disruptive
shocks, and it is critical to better understand how welfare has changed amongst these populations.2
       There are numerous articles investigating the consequences of violent conﬂict on economic out-
comes. For example, the severity of conﬂict, usually measured in the number of fatalities in a region,
has been shown to be negatively correlated with ﬁrm performance (e.g., Abadie and Gardeazaba 2003;
Guidolin and La Ferrara 2007; Klapper et al. 2013; Singh 2013, Amodio and Di Maio 2017, etc.),
aggregate economic activity (e.g., Alesina et al. 1996; Eckstein and Tsiddon 2004; Chen et al. 2008;
and Cerra and Saxena 2008), and a few individual economic outcomes such as employment and income
(e.g., Collier and Duponchel 2013, Ksoll et al. 2015).3 Given the signiﬁcant impacts of violent conﬂicts
on economic activity, there has been a signiﬁcant emphasis placed on job creation in fragile settings,
where a number of development agencies, governments, and researchers have experimented with how
to support livelihood in these settings (e.g., Blattman and Ralston 2015).
       However, there are many issues that could beneﬁt from further analysis. First, there are signiﬁcant
diﬀerences in the form and shape of conﬂicts that could be missed by measures of severity of conﬂicts-
especially ones that focus on deaths alone. For example, of the four recent civil wars in the Middle
East and North Africa, the nature of violence diﬀers between the conﬂicts. Figure 1 presents estimates
of average number of deaths per year during the height of each conﬂict, and further separates deaths
by civilian versus non-civilian status. Syria and Iraq had the highest number of deaths, but the
nature of the deaths were signiﬁcantly diﬀerent. The number of civilians killed in Iraq far outnumbers
non-civilian deaths, whereas it is the opposite in Syria.
       Additionally, the Republic of Yemen is a clear outlier in the small number of deaths- both civil-
ian and non-civilian. However, the eﬀects on Yemenis have still been dire through other means-
the violence has eroded peoples earnings and contributed to the unreliable payment of public sector
employees, and intermittent blockades on Yemeni ports have prevented fuel, medical supplies, and
food from reaching an at-times desperate population (World Bank 2017). Thus, it is important to
better understand how welfare evolves in many more types of fragile settings, including those where
traditional measures of violence would indicate a low severity.
       In addition to analyzing a wider range of fragile situations, it could be beneﬁcial to to address a
   1
     See (accessed March 2018): http://www.worldbank.org/en/topic/fragilityconﬂictviolence/overview.
   2
     The share represents the sum of the populations of Iraq, Jordan, Lebanon, Libya, the Syrian Arab Republic, and
the Republic of Yemen, all divided by the sum of the population of all countries in the Middle East and North Africa
Region. Populations were obtained from the World Development Indicators for 2016.
   3
     For a survey of the literature investigating further consequences of conﬂict (particluarly civil war) in the longer term,
see Blattman and Miguel (2010).



                                                              2
wider range of household and intra-household issues more fully that have not been able to be addressed
given the scarcity of data in conﬂict settings. Many articles are forced to rely on data sources aside from
household surveys collected during conﬂict- such as businesses reporting after the conﬂict about their
economic activity during the conﬂict (e.g., Amodiao and Di Maio 2017) and also peoples’ recollection
of their experiences during conﬂict after the situation allows the possiblity of ﬁelding household surveys
(e.g., Blattman and Annan 2010). Although these data sources have yielded many valuable insights,
they could leave gaps in how households might have coped with conﬂict shocks and there could be
biases in people’s recollection of many details in such turbulent circumstances long after they have
occurred.
       This paper addresses some of these issues by investigating the consequences of the unexpected,
quick, and mostly non-violent capture of the Republic of Yemen‘s capital, Sana‘a, by Houthi forces in
September of 2014. This setting is unique, where the capture was very quick and accomplished with
little violence.4 The non-government forces placed the president under house arrest and immediately
took control of key government agencies- including those responsible for defense and ﬁnance. However,
because there was little violence, and because there was little concern for many of the less prominent
government agencies, the Central Statistics Oﬃce was able to complete an ongoing multi-themed
household survey as planned for the remaining three months of the year. This setting allows one to
analyze the degree to which the welfare and employment of households were aﬀected by the capture,
the ways in which households might have coped, and the potential diﬀerent impacts the capture might
have had on diﬀerent household members.5
       Given less violence than is typically associated with conﬂicts, one might expect there to be few
impacts of the capture of the Yemeni capital relative to other conﬂict settings. However, the capture
and the change in the rule of law did result in a decline in safety and an increase in uncertainty that
potentially had an impact on people’s ability to go out and seek a living and acquire goods and services
necessary for a minimum standard of living. For example, although the violence in this setting had
mostly ceased after the capture of Sana‘a, there were reports of arbitrary detentions of civilians by
Houthi forces and instances of targeted violence6 , and there was an increase in terrorist events.7 Both
of these factors might have made it more diﬃcult for people to leave their households. In addition to
the decline in safety leading to a disruption in economic activity, the prices of fresh food items and fuel
signiﬁcantly increased following the capture of the city (ﬁgure 2). These price changes are consistent
with disruptions in supply chains, and previous research has demonstrated that high food prices were
already a signiﬁcant impediment to consumption even prior to the capture of the capital (e.g., WFP
2011; World Bank 2017).
       The decline in safety is further corroborated by the ongoing household survey that was being
conducted as the capital was captured. The survey included a subjective welfare module, which
separately asked the head of the household and their spouse whether they were satisﬁed with security.
   4
     The city was captured after a 4-day siege that was mostly centered around a military base in the northern part of
the city, and abruptly ended when the internationally-recognized government ceded control of the city. See (accessed
July 2017): http://www.bbc.com/news/world-29380668.
   5
     Only in March, 2015, as the non-government forces marched past the capital and attempted to overtake the southern
part of the country, did a coalition of Gulf states begin airstrikes to reinstate the internationally-recognized government
(see (accessed July 2017): http://www.bbc.com/news/world-middle-east-29319423). Although the escalation of the
conﬂict in 2015 and beyond also deserves analysis, the unique opportunity of exploiting the ongoing household survey
was not available during that time period and the data available is consistent with data scarcity present in other conﬂict
settings.
   6
     See (accessed July 2017): https://www.hrw.org/news/2017/01/12/yemen-no-accountability-war-crimes.
   7
     See (accessed July 2017): https://www.theguardian.com/world/2014/oct/09/yemen-suicide-attack-kills-20.



                                                            3
Following the capture of the capital, there was a signiﬁcant decline in the share of households where
either a household head or spouse reported to be very satisﬁed with security that was not observed in
the rest of the country.8 Importantly, this decline in safety might have had a worse impact on women
than men, where not a single woman interviewed following the capture of the city (out of 235 total)
reported to be very satisﬁed with security, whereas women were able to report being satisﬁed with
other aspects of their life.
       Although the household survey continued in the following the capture of the capital, the identiﬁ-
cation strategy requires the sampling of households to not dramatically change following the capture.
The sampling design was such that respondents were randomized over space and time, and there should
be no diﬀerences in the households surveyed in the months before and the months after the capture of
the capital if the survey was conducted as planned (World Bank 2017). However, enumerators might
have both consciously and unconsciously avoided more dangerous areas, which might have aﬀected
the representativeness of the sample.
       Despite this potential challenge, there is little evidence that sampling was aﬀected dramatically.
The sampling was reported to have been carried out as planned- all enumeration areas and initial
households to interview were selected approximately a year before the city was captured, none of the
enumeration areas were replaced, and there were no reports of diﬃculties in completing the survey in
the ﬁnal quarter. Also, there is little diﬀerence in relatively ﬁxed household characteristics between
households surveyed prior to the capture to those surveyed after, corroborating the reports of the
sampling not being dramatically aﬀected.9
       Given this setting, this paper highlights three key ﬁndings. First, despite less violence than typi-
cally seen in other conﬂict settings, there was a nearly immediate decline in household expenditure,
wages, and food consumption, that was driven by larger and more robust declines as the length of
the occupation progressed. By the third month following the capture, expenditure in Sana’a was
approaching the level in the rest of the Republic of Yemen, which traditionally is much poorer than
the capital (World Bank 2017). However, only relatively better oﬀ households were able to initially
maintain their consumption, whereas households with worse socio-economic characteristics observed
an immediate decline in expenditure.
       However, in addition to the decline in household welfare, there is evidence that women fared
particularly poorly in this setting. The capture of the city had adverse consequences on nearly all
measures of women‘s empowerment reported in the survey, where women had less control over nearly
all measures of spending, and this control was transferred to primarily the male household head. The
decline in control over food expenditure was especially pronounced, which was also one of the areas of
spending where women in Sana‘a had the most control prior to the capture of the city. This decline in
   8
      Prior to the capture, approximately 11.6% of households had either a head or a spouse report to be very satisﬁed
with security; following the capture of the capital, only 1.0% of households had a head or spouse report to be very
satisﬁed with security.
    9
      Speciﬁcally, a wide variety of household characteristics are reported by month, as well as the average pre-capture
means of each of the variables. Figure 3 demonstrates that the 95 percent conﬁdence interval of each monthly estimate
contains the pre-capture mean of the variable in the vast majority of instances. Of all the 95 percent conﬁdence intervals
reported (180), the pre-capture mean is contained in 96.1 percent of them (173). Importantly, of those conﬁdence
intervals that do not contain the pre-capture mean, the distance from the bound of the interval to the pre-capture mean
is very small; there is no evidence that any particular variables had a large proportion of monthly conﬁdence intervals
that did not contain the pre-capture mean (the maximum for any variable is 2 of the 12 monthly conﬁdence intervals
not containing the pre-capture mean); there is little evidence that the post-capture months of October-December were
contributing the majority of the instances where the pre-capture means were outside of the conﬁdence intervals (2 of
the 7 total instances were from those months); and there is little evidence that the conﬁdence intervals were uniformly
larger following the capture of the capital.



                                                            4
decision making is consistent with women being less able to leave the household than men due to safety
concerns, where the decline in decision making was immediate and did not get worse as household
expenditure signiﬁcantly declined in the later months of the occupation.
       This decline in reported ability to make decisions was also corroborated by observable actions. In
particular, there was a signiﬁcantly larger share of women who were in the household and able to
answer the women’s decision-making module following the capture, which is consistent with women
being less able to leave the household following the capture of the city. Importantly, there is no
pattern suggesting that women with better bargaining positions or women from better-oﬀ households
fared better than other women, suggesting that the decline was nearly universal. Furthermore, in
speciﬁcations investigating changes in employment of women, there was slight evidence of a small
decline in employment, which is further consistent with women being less able to leave the household
than men.
       Lastly, this setting also had implications for the validity of subjective measures of welfare. The
2014 household survey in the Republic of Yemen asked both household heads and their spouses about
their satisfaction with a number of dimensions of welfare (e.g., food, income, etc.). Despite the nearly
universal decline in nearly all dimensions of household welfare documented above- especially in diet
and wages- households failed to report less satisfaction in each welfare dimension. Rather, households
reported more satisfaction in food consumption, while one can reject the hypothesis that there was a
large and negative change in satisfaction with employment and satisfaction with income. Thus, not
only were the subjective welfare measures unable to uncover the decline in more traditional reported
measures of welfare (e.g., employment, wages, consumption), they estimated the incorrect sign of the
welfare change in the case of food consumption. This ﬁnding is consistent with a model in which
non-consumption factors, such as safety or conﬁdence in the Houthi forces, become more salient to
household satisfaction than consumption.10
       These results are robust to two important concerns. First, one might be worried about the possi-
bility that the poor economic conditions might have led to the capture of the city by the Houthi forces,
and the above patterns are capturing this pre-existing trend (e.g., Miguel et al. 2004). However, the
Houthi capture of the Republic of Yemen‘s capital city is a setting in which this concern is minimized.
First, none of the independent variables of interest- household welfare, women’s decision-making, and
subjective satisfaction- were trending in the months prior to the capture of Sana’a. Only after the cap-
ital was captured did these measures begin to change. Second, some have suggested that the capture
of the city could have been facilitated by powerful politicians rather than the will of the population in
response to economic conditions. Although the Houthis took advantage protests in September11 , their
success seems more related to the fact that the majority of the army loyal to the former president did
not resist the assault.12 This is corroborated by the fact that the Houthis fought and lost numerous
wars to the government over the prior decade, but only once the army did not resist their advances
were they able to achieve more success.13
       Second, as is likely consistent with other conﬂict settings, the time immediately prior to the conﬂict
  10
     This pattern could also be explained by households lying to the enumerators out of fear of the Houthi forces. However,
this explanation is not consistent with the fact that household reported signiﬁcantly less satisfaction with safety, which
was one of the primary responsibilities of the Houthis after the capture. However, regardless of the exact cause, the
measures were not changing in similar ways of more traditional welfare measures.
  11
     See Al Jazeera (Accessed July 2017): http://america.aljazeera.com/articles/2014/9/25/houthi-yemen-takeover.html.
  12
     See (accessed July 2017): http://www.reuters.com/article/us-yemen-security-saleh-idUSKBN0MM1MV20150326)
  13
     See (accessed July 2017):        http://www.aljazeera.com/news/middleeast/2014/08/yemen-houthis-hadi-protests-
201482132719818986.html.



                                                            5
was turbulent. In particular, the government temporarily removed a large fuel subsidy for the month
of August that not only supported direct consumers of fuel, but also supported the transport of
goods across the country and the use of fuel as an intermediate input of production.14 However,
this temporary removal of the fuel subsidy did not appear to be driving the results. First, there was
little-to-no immediate impact on all the measures of household and intrahousehold welfare analyzed
here for that month, where these outcomes only changed following the capture of Sana’a. Second,
the subsidy removal was targeted at the entire country and not just Sana’a, whereas all the results
analyzed here show a pattern of a clear change in Sana’a coupled with little change in the rest of the
country.
    There are three primary contributions of these results. First, this paper demonstrates that the non-
violent capture of territory from an internationally-recognized government can lead to a signiﬁcant
decline in welfare across a number of dimensions. Although previous settings have demonstrated
that the severity of violence is correlated with economic activity (e.g., Abadie and Gardeazaba 2003;
Guidolin and La Ferrara 2007; Klapper et al. 2013; Singh 2013, Amodio and Di Maio 2017, etc.),
these ﬁndings demonstrate that even households and regions not undergoing signiﬁcant violence have
a marked decline in welfare during volatile times. Furthermore, this setting allows a more thorough
investigation of household welfare and coping strategies than previously possible given the collection
of a multi-themed household survey during the capture of Sana‘a.
    Second, the decline in women‘s decision-making demonstrates an important way in which conﬂict
might impact individuals within the household diﬀerently. Many news organizations and humanitarian
agencies report that, due to economic necessity, women become more involved in decision making and
economic activity during the conﬂict, but then revert back to the pre-conﬂict baseline once the conﬂict
ends. For example, news reports in the Republic of Yemen following the escalation of conﬂict in 2015
suggest that women are being forced to take a more active role in household decisions and income
generation as the humanitarian situation became more dire.15 However, to my knowledge there has
been little corroboration of these results.16 These results, along with the fact that there has been
little change in the small labor force participation of women in general in the Republic of Yemen
either during the capture of Sana’a or over the course of the more violent conﬂict between 2015 and
2016 (e.g., Gallup World Poll), contrast with the narrative of more responsibilities for women during
conﬂict.
    And lastly, these results suggest that subjective measures of welfare might not be suitable in
such tumultuous circumstances, and that individuals might have signiﬁcant biases in recounting their
experiences during conﬂict in surveys conducted well after the end of conﬂicts. Subjective levels
of welfare have been demonstrated to have large ﬂuctuations when respondents are prompted with
vignettes in non-tumultuous settings, but the correlates of welfare were largely similar regardless of
prompting (e.g., Ravallion et al. 2016). However, the present case demonstrates that particular
measures of subjective welfare might even be trending in the opposite direction of more traditional
measures in settings where non-consumption factors might become more important.
    This is unfortunate given that the beneﬁts of collecting such measures- their timely nature and low
  14
     The removal of the fuel subsidy was unpopular, where protests against the policy change formed in Sana’a, and the
government immediately reinstated the fuel subsidy by the end of the month.
  15
     For an example from the Republic of Yemen following the escalation in conﬂict in 2015, see
http://www.aljazeera.com/news/2015/12/yemen-widows-weakest-victim-war-151215061011411.html. However, these re-
ports do not suggest that women are necessarily treated better, but rather that their responsibilities have increased.
  16
     Mkutu (2008) demonstrates that an increase in widow-led households necessarily led to more participation of women
in Uganda in a pastoral setting.



                                                          6
cost- are particularly important in fragile situations. For example, in many fragile contexts, surveys
often ask about household coping strategies that include implicit reference points that could be in
ﬂux in such contexts.17 Additionally, the Living Standards Measurement Surveys at the World Bank
have been implementing conﬂict modules asking individuals to recall their experiences during conﬂict,
many times years after a conﬂict.18 However, if individuals’ satisfaction can be shown to be at odds
with actual household behavior and consumption, it is possible that there might be signiﬁcant biases
in what individuals might recollect in response to conﬂict. More evidence regarding the validity of
such survey modules could signiﬁcantly help welfare monitoring in fragile contexts.
   This paper is structured as follows. Section 2 describes the context and various manners in which
the capture of territory from the government might aﬀect household welfare in non-violent settings;
Section 3 describes the data; Section 4 describes the empirical strategy; Section 5 presents estimates
of the decline in welfare following the capture of the city; Section 6 presents estimates of the decline in
women‘s empowerment; Section 7 presents estimates of the change in subjective measures of welfare;
and Section 8 concludes.

2. The Capture of Sana‘a and the Potential Decline in Welfare

   Following the capture of Sana‘a, there are a number of manners in which welfare might have been
aﬀected. First, the supply chains providing goods and services to the capital city were likely disrupted.
Prior to the conﬂict, the Republic of Yemen relied on international sources for the vast majority of
its consumption, where approximately 90 percent of its food was imported from abroad (e.g., World
Bank 2017). Imports to the capital city primarily came through a port in Al Hudaydah, and were
then transported by truck to the capital city.19 Given the diﬃculty and uncertainty of transporting
food and other goods across new checkpoints where safety and access were not guaranteed likely made
it more diﬃcult for goods to reach the capital city.
   There is evidence that this indeed was a signiﬁcant problem in Sana‘a following the capture of the
city. Utilizing average prices collected at markets in the city by the World Food Programme (WFP)
each month for 14 food items and commodities, there was a signiﬁcant increase in the price of a
number of goods following the capture of the city. In particular, ﬁgure 2 demonstrates that fresh food
items and fuel, all of which are imported on a regular basis from abroad, had large price increases in
the months following the capture of the city that was not evident over the same time period in the
two years prior to the capture of the city.
   Thus, in response to the capture of the city, households had to deal with a signiﬁcant adverse
price shock. Price rises have a signiﬁcant impact on household welfare and a variety of coping strate-
gies might be employed (Barrett 2002). Empirically, a number of contexts have demonstrated that
household welfare signiﬁcantly declines in the face of an adverse food price shock, where households
reduce diet quality (e.g., D‘Souza and Jolliﬀe 2012; D‘Souza and Joliﬀe 2014), and potentially sell
a limited share of productive assets (e.g., Fafchamps et al. 1998). Based on these other contexts,
households in Sana‘a might also be forced to adopt a number of these coping strategies. However,
the coping strategies in this circumstance might be substantially diﬀerent. For example, households
might ﬁnd it diﬃcult to sell productive assets given the overall poor condition of economic activity;
  17
     Examples include whether households at less than they thought they should (FAO Food Insecurity Experience Scale),
and whether households consumed less preferred foods (World Food Program Coping Strategies Index).
  18
     See (accessed March 2018) http://econ.worldbank.org/WBSITE/EXTERNAL/EXTDEC/EXTRESEARCH/EXTLSMS/0,,content
SitePK:3358997,00.html.
  19
     See FAOs summary of food imports in the Republic of Yemen at (accessed July 2017):
http://www.fao.org/giews/countrybrief/country.jsp?code=YEM.


                                                    7
or, for example, it might be diﬃcult for households to seek secondary employment opportunities given
the security situation.
       In addition to the adverse price shock, there are other potential mechanisms by which household
welfare might be aﬀected. The uncertainty introduced by the lack of protection and the authority of
a new force likely signiﬁcantly impacted economic activity and investment. As discussed in the Intro-
duction, news reports suggest that safety was negatively impacted20 , which likely aﬀected household
members‘ ability to leave their households and seek employment opportunities. And even of those who
did venture out to seek work, the lack of economic activity likely further impacted their employment
opportunities. However, it is important to note that the government promised to continue to pay
government salaries during this time period, which was a major source of employment in the capital
city at the time of its capture.21 Thus, many of the most severe income impacts might have been
temporarily avoided, and the resulting estimates are an underestimate in the case that the government
is not in a position to continue to provide a livelihood to a large share of households.22
       Thus, at exactly the time that prices of necessary goods and services were rising, households likely
had a decline in household income. A vast literature across a multitude of contexts has investigated
the potential welfare impacts of adverse income shocks across a number of dimensions. For example,
it has been shown that adverse income shocks adversely aﬀect the quality of food consumption but
not overall calorie consumption (e.g., Block et al. 2004; Brinkman et al. 2009), can reduce school
attendance amongst school-aged children (e.g., Edmonds et al. 2009; etc.), and potentially force
households to sell a limited share of productive assets and reduce possible future income generation
(e.g., Fafchamps et al. 1998). But similar to coping with an adverse price shock, the set of available
coping strategies might be limited in this circumstance.
       Importantly, these shocks- both the potential tightening of household budgets and the decline in
safety- might have signiﬁcantly diﬀerent eﬀects on women in the household than men. For example,
increasing the constraints on households tends to lead to an excess burden being placed on vulnerable
members of the household in a variety of settings (e.g., Rose 1999; Khanna et al. 2003; Miguel 2005;
D’Souza and Tandon forthcoming; etc.), which could place an excess burden on women in the Republic
of Yemeni households given the country is typically ranked as having one of the highest (if not the
highest) gender gaps in the world (e.g., World Economic Forum 2017). Furthermore, the increase
in prices might similarly force households to substitute more home-produced goods and services for
market-produced goods and services, which would most likely place a higher time burden on household
women (e.g., World Development Report 2012). Additionally, the decline in safety might make it more
diﬃcult for women to leave the household and lead productive lives than men. In particular, women’s
movement and freedom are reported to be signiﬁcantly curtailed in the country due to safety and
family concerns (e.g., World Bank 2012), suggesting that they might even be less free to leave the
household and make decisions following the introduction of conﬂict.
       However, it is important to note that the eﬀects of these shocks on women are empirically ambigu-
ous. The decline in household resources might suﬃciently burden households that women are forced
to take up income generating possibilities and other responsibilities that they were unable to do so
before the conﬂict. Many news organizations and humanitarian agencies report that women are forced
to take on more responsibilities, both inside and outside the household, in response to conﬂict. For
  20
     See (accessed July 2017): https://www.hrw.org/news/2016/11/17/yemen-abusive-detention-rife-under-houthis.
  21
     See (accessed July 2017): http://www.reuters.com/article/us-yemen-security-salaries-idUSKBN15A1WW.
  22
     This eventually happened in the Republic of Yemen in September of 2016 when the government started to pay
employees infrequently.



                                                      8
example, news reports in the Republic of Yemen following the escalation of conﬂict in 2015 suggest
that women are being forced to take a more active role in household decisions and income generation as
the humanitarian situation became more dire.23 Alternatively, an increase in widow-led households in
more violent conﬂicts could also lead to higher labor force participation and empowerment for women.
However, the current setting is one in which there was little violence and no observed increase in the
share of widow-led households.

3. Data

    The 2014 Household Budget Survey (HBS) conducted by the Central Statistics Oﬃce of the Re-
public of Yemen with support from the World Bank was in progress as the Houthi forces captured the
city of Sana‘a. The survey‘s primary purpose was to update the oﬃcial poverty rate in the Republic of
Yemen, which had not been updated since 2005. Given the little violence in the capture of the capital
city, the survey was administered before, during, and after the capture of the city. However, given
the sampling design, the survey is able to be used to identify how households were aﬀected by the
capture. First, the sampling design was such that respondents were randomized over space and time,
and there should be no diﬀerences in the households surveyed in the months before and the months
after the capture of the capital if the survey was conducted as planned (World Bank 2017).24 Second,
the survey is representative at the governorate/capital city level.
    The 2014 HBS included a complete set of modules, including standard information on household
consumption, employment, education, health, etc. that allow a thorough investigation of many di-
mensions of household welfare. However, especially useful to the current context, the survey includes
a thorough module on women‘s empowerment in the household and a module on subjective welfare.
These modules allow an investigation of intra-household dynamics and an investigation of the appro-
priateness of subjective welfare measures in such tumultuous circumstances.
    Households in Sana‘a were far better oﬀ than the rest of the country. Poverty in Sana’a was
approximately one-quarter of the rate in the rest of the country, and all other measures of welfare
were similarly better in the capital city. Additionally, women were signiﬁcantly more empowered in
the capital city than in the rest of the country, but subjective welfare was approximately equal across
all regions of the country.
    Another advantage of the 2014 HBS is that there was a nearly identical 2005 HBS that allows one
to investigate whether changes in welfare were potentially due to seasonal factors, or due to a country-
wide shock as opposed to the capture of Sana‘a. In particular, the 2005 data allow an investigation of
the diﬀerence-in-diﬀerence estimate of welfare in Sana‘a following the capture of the city. Additionally,
the data further allow us to investigate the triple diﬀerence as well, where the diﬀerence-in-diﬀerence
estimate in the city of Sana‘a is compared to the diﬀerence-in-diﬀerence in the rest of the country.
However, the women‘s empowerment and subjective welfare modules were not implemented in the
2005 survey, and speciﬁcations analyzing those welfare measures can only analyze the diﬀerence in
Sana‘a following the capture of the city in 2014, and compare that diﬀerence to the change in the rest
  23
     For an example from the Republic of Yemen following the escalation in conﬂict in 2015, see
http://www.aljazeera.com/news/2015/12/yemen-widows-weakest-victim-war-151215061011411.html. However, these re-
ports do not suggest that women are necessarily treated better, but rather that their responsibilities have increased.
  24
     Roughly equal numbers of households were surveyed in each month, where 78 households were surved in October, 80
households were surveyed in November, and 77 households were surveyed in December. A total of 235 households were
surveyed after the capture of Sana’a was complete and 269 households were surveyed in the 3 non-Ramadan months
leading up to the capture. However, the responses to the women’s decision-making module were missing for one of the
households in the pre-period, leaving one less observation in those speciﬁcations than the others. However, all results
are robust to dropping that household from all speciﬁcations.


                                                          9
of the country.
       Although the household survey continued following the capture of the capital, the identiﬁcation
strategy requires the sampling of households to not dramatically change following the capture. Im-
portantly, given the little amount of violence, the survey was reported to have been carried out as
planned. All enumeration areas and initial households to interview were selected approximately a year
before the city was captured, none of the enumeration areas were replaced, and there were no reports
of diﬃculties in completing the survey in the ﬁnal quarter.
       Despite these reports of little trouble, it is always possible that enumerators consciously or un-
consciously avoided more dangerous areas when replacing households. However, investigating the
possibility, there does not appear to be any evidence that households surveyed before the capture
were signiﬁcantly diﬀerent in observable characteristics that are unable to adjust to the conﬂict. Ta-
ble 1 presents observable characteristics of households surveyed in the third and fourth quarter of
2014. Households surveyed in the third quarter all began their surveys prior to the capture, and
households surveyed in the fourth quarter all began their surveys after the city had been captured.
Consistent with the capture having little eﬀect on the sampling, column (3) demonstrates that there
is little diﬀerence between diﬃcult-to-adjust characteristics following the capture of Sana’a.25
       However, it is also possible that sampling changes might have been more observable as conﬂict
progressed, and then sampling quickly reverted to the planned sampling procedure when violence was
reduced. However, ﬁgures 3a-3c demonstrate that this does not appear to be the case. Speciﬁcally,
a wide variety of household characteristics are reported by month, as well as the average pre-capture
means of each of the variables.
       Figures 3a-3c demonstrate that the 95 percent conﬁdence interval of each monthly estimate of each
diﬃcult-to-adjust control variable contains the pre-capture mean of the variable in the vast majority of
instances. Of all the 95 percent conﬁdence intervals reported (180), the pre-capture mean is contained
in 96.1 percent of them (173). Importantly, of those conﬁdence intervals that do not contain the pre-
capture mean, the distance from the bound of the interval to the pre-capture mean is very small; there
is no evidence that any particular variables had a large proportion of monthly conﬁdence intervals
that did not contain the pre-capture mean (the maximum for any variable is 2 of the 12 monthly
conﬁdence intervals not containing the pre-capture mean); there is no evidence that the post-capture
months of October-December were contributing the majority of the instances where the pre-capture
means were outside of the conﬁdence intervals (2 of the 7 total instances were from those months);
and there is little evidence that the conﬁdence intervals were uniformly larger following the capture
of the capital.
       However, there is an important data issue to highlight that might aﬀect estimating the impact
of capturing Sana‘a. Ramadan occurred primarily in July in 2014, and this might bias expenditure
during the pre-period given that consumption patterns are signiﬁcantly diﬀerent during Ramadan (e.g.,
Jolliﬀe and Serajuddin 2015).26 Thus, in speciﬁcations that use only the 2014 survey, all households
surveyed during Ramadan are excluded from the sample, and the baseline speciﬁcation uses the three
non-Ramadan months prior to the capture of the capital as the pre-period to compare against the
three post-capture months of the survey. However, in speciﬁcations that compare the 2014 survey to
the 2005 survey, households surveyed during July (Ramadan in 2014) or October (Ramadan in 2005)
  25
    The households surveyed in the third quarter include those that were surveyed during Ramadan.
  26
    In a number of contexts, it has been demonstrated that food consumption is actually higher during Ramadan and
other religious festivals despite fasting during daylight hours given the number of feasts, etc. that occur. This is
corroborated in the case of the Republic of Yemen (e.g., World Bank 2017).



                                                        10
in either survey were excluded. However, it is important to note that the estimates are qualitatively
identical when all interviewed households are included.

4. Empirical Strategy

       Based on this setting, this paper utilizes a number of empirical strategies to investigate the eﬀects
of the capture of Sana‘a on household welfare, women’s empowerment, and subjective metrics of well-
being. In the baseline empirical strategy, this paper investigates how adjustable outcomes in Sana’a
changed in response to the capture of the capital, and compares this change in Sana’a to a number
of other regions and time periods. Figure 4 plots average annual household expenditure by month of
the survey, and presents an outline of the empirical strategy (along with the lack of change in non-
adjustable outcomes in ﬁgures 3a-3c).27 Expenditure in Sana’a was maintained in October of 2014 on
average, but by November and December, there was a sharp decline in expenditure such that there
was little diﬀerence in expenditure between the city of Sana’a and the rest of the Republic of Yemen
despite the poverty rate being approximately four times less in the city of Sana’a. Importantly, over
the same time period, there was little change in expenditure in the rest of the Republic of Yemen, and
there was little change in expenditure in Sana’a in the 2005 survey.
       Formalizing this empirical strategy, this paper ﬁrst estimates the change in household oucomes
over the course of the occupation in the following speciﬁcation:



(1)                             Outcomei = Σ3
                                            j =1 [βj P ost M onth ji ] + γXi +            i


where i refers to household, P ost M onth j are indicators equaling one if the household was inter-
viewed during the j’th month following the occupation28 , and X is a vector of individual control
variables.29 The baseline estimates restrict the households in speciﬁcation (1) to households from
the city of Sana’a, and also restricts the sample to households surveyed in the three non-Ramadan
months prior to the capture and the three months following the capture of the city that the survey
continued. Standard errors are clustered at the PSU level, and observations are weighted to make the
estimates representative of the population.30 The coeﬃcients of interest are βj , which represent the
the percentage point change in a variety of outcomes following the capture of the city relative to the
three months before the capture.31
       This paper further compares the simple change in Sana’a to the change in the rest of the Republic
of Yemen to determine whether the simple change might be capturing factors that aﬀected the entire
  27
      The measure of expenditure is the same as used to construct the national poverty estimates, and is restricted to
items for which household expenditure can be reliably estimated for all households. This omits expenditure on durable
goods, for which households only purchase intermittently.
   28
      Speciﬁcally, P ost M onth 1, P ost M onth 2, and P ost M onth 3 respectively are indicators equaling one if the house-
hold was surveyed in October, November, and December.
   29
      Control variables include an indicator for whether the household head is married, the household head ever attended
school, an indicator if the household head was a smoker, an indicator if the household head chews Qat, the years of
schooling of the household head, an indicator whether the household has a child, an indicator if the household’s dwelling
is make of cut stone, an indicator equaling one if the walls if the household’s dwelling were made of other material,
an indicator equaling one if the household’s roof that was made of reinforced concrete, an indicator equaling one if the
household’s roof was made of wood and concrete, an indicator equaling one if the household’s roof was made of mud, an
indicator if the household’s ﬂoor was made of concrete, an indicator equaling one if the household’s ﬂoor was made of
tile, and an indicator equaling one if the household had a ﬂush toilet.
   30
      All results discussed throughout the paper are robust to the exclusion of weights.
   31
      Although the baseline speciﬁcation is restricted to balance the number of pre and post-months and to use the pre-
months that are closest in time to the capture of the city, Apendix 2 demonstrates that the results are qualitatively
identical regardless of which months are used to construct the pre-period.


                                                            11
country as opposed to the capture of Sana’a. Speciﬁcally, this paper also estimates the following
diﬀerence-in-diﬀerence speciﬁcation:



(2)                Outcomeir = Σ3
                                j =1 [βj P ost M onth jir ∗ Capitalir ] + αYir + γXir + υir


where all variables are deﬁned as above, r denotes region, Capital denotes an indicator equal to
one if the household is from the city of Sana‘a, and Y includes all lower-order terms (P ost Conf lict
and Capital).32 The sample in speciﬁcation (2) is expanded to include households surveyed in the
rest of the Republic of Yemen during the same time period, and the speciﬁcations report the larger
of standard errors clustered at the governorate/capital city level or PSU level.33 Again, βj is the
coeﬃcient of interest and denotes how much more outcomes changed in the city of Sana‘a than in the
rest of the Republic of Yemen.
       In addition to the double diﬀerence investigated in speciﬁcation (2), this paper further reports a
diﬀerence-in-diﬀerence speciﬁcation comparing the change in Sana’a in 2014 to the change in Sana’a
in 2005. Although a lot of turbulent events happened between 2005 and 2014, including a political
revolution toppling the former president, the speciﬁcation can further demonstrate that the changes in
household outcomes are not driven by seasonal factors in Sana’a that aﬀect households in the fourth
quarter of the year. Speciﬁcally, the paper estimates the following speciﬁcation:



(3)              Outcomeit = Σ2
                              j =1 [βj P ost M onth jir ∗ P ost Surveyit ] + αYit + γXit + υit


where all variables are deﬁned as above, t refers to time, P ost Survey denotes an indicator equaling one
if the household was interviewed in the 2014 survey and equal to zero of the household was interviewed
in the 2005 survey, and Y denotes a vector of lower-order terms (P ost Conf lict and P ost Survey ).34
The sample is restricted to households surveyed in Sana’a, but includes all households from both the
2014 and 2005 surveys.35 Again, βj is the coeﬃcient of interest and in this speciﬁcation denotes how
many more percentage points household outcomes increased in 2014 than in 2005. However, since
the women’s empowerment module and the subjective welfare module were not included in the 2005
survey, only a subset of total outcomes can utilize speciﬁcation (3).
       Lastly, this paper also investigates speciﬁcations that simultaneously control for seasonal and
country-wide eﬀects. In particular, the following speciﬁcation is also estimated:



(4)       Outcomeirt = Σ2
                        j =1 [βj P ost M onth jir ∗ P ost Surveyirt ∗ Capitalirt ] + αYirt + γXirt + φirt


where all variables are as deﬁned above, and Y includes all lower-order terms.36 Again, the coeﬃcient
of interest is βj , which denotes how many more percentage points larger the diﬀerence-in-diﬀerence
estimate was in Sana‘a than the diﬀerence-in-diﬀerence estimate in the rest of the Republic of Yemen.
  32
     The control variables are the same as above, but now an indicator for whether the household resided in an urban
area is also included in speciﬁcations that compare the city of Sana’a to other regions.
  33
     There are 22 clusters when clustering at the governorate and capital city level.
  34
     This speciﬁcation only investigates two post-conﬂict months because October, one of the post-conﬂict months, was
Ramadan in 2005.
  35
     Standard errors are clustered at the PSU level.
  36
     This includes every possible second-order term and each of the individual components of the third-order term.



                                                         12
This speciﬁcation includes households from all of the Republic of Yemen in all time periods. Addi-
tionally, the larger of standard errors clustered at the PSU level and the governorate/capital city level
are reported, and observations are weighted to make them representative of the population.
   Given the long time period between the 2005 and 2014 surveys and the numerous events that
happened between- political transition and food price shock- this paper focuses on estimates from
speciﬁcation (2) that compare the change in household outcomes in the city of Sana’a to the rest of
the Republic of Yemen using only data from the 2014 survey. However, estimates of the double and
triple diﬀerence utilizing the 2005 survey are presented, and all results discussed in the text are robust
to using either of those estimation strategies for outcomes that are captured in both the 2014 and
2005 surveys.

5. Decline in Welfare Following the Capture of the Capital

   As demonstrated in Figure 4, there was a dramatic decline in household welfare following the
capture of the capital that was primarily driven by larger and more robust decreases as the occupation
lasted longer. Table 2 presents estimates of how annual per capita expenditure changed following the
capture. Estimates of speciﬁcation (1) are presented in columns (1)-(2). Column (1) presents the
simple percentage change; and column (2) adds control variables to absorb unobserved heterogeneity
and increase the precision of the estimates.
   In either speciﬁcation, expenditure following the capture of the city is less than in the three months
leading up to the capture. However, the change for households surveyed in October and November are
lower in magnitude and less precisely estimated than the change for households surveyed in December.
In the most complete speciﬁcation, households surveyed in December had an expenditure that was 37.7
percent lower than the average expenditure in the months leading up to the capture of the city, and
one can reject the hypothesis that the three post-conﬂict coeﬃcients jointly equaled zero at standard
levels of signiﬁcance (p-value of 0.007).
   Estimates of speciﬁcation (2) are presented in column (3) of table 3. The estimates demonstrate
a qualitatively similar pattern as the estimates in columns (1)-(2). Importantly, given the lack of any
change in expenditure in the rest of the Republic of Yemen during the same time period, the esti-
mates from speciﬁcation (2) are very similar to the estimates from speciﬁcation (1), where households
surveyed in December had 41.5 percent lower expenditure than households surveyed in the months
leading up to the capture. Thus, the baseline estimates do not appear to be capturing a shock that is
common to the entire country.
   Furthermore, there is very little evidence that the drop in expenditure was due to seasonal factors
present in the city of Sana’a. When comparing the drop in expenditure in the capital in 2014 to the
change over the same time period in 2005, the results are qualitatively identical to columns (1)-(3).
Expenditure in December was 37.5 percent lower than expenditure in the months before the capture.
Similarly, column (4) presents estimates of speciﬁcation (4), and demonstrates that the diﬀerence-in-
diﬀerence in Sana’a was much larger than the diﬀerence-in-diﬀerence in the rest of the Republic of
Yemen, and this change was driven by a larger and more precise estimate in December.
   However, it is important to verify that these changes in expenditure are being driven by the
capture of Sana‘a, and not uncovering a pre-existing change in expenditure. Thus, column (5) of table
2 reports estimates of a speciﬁcation similar to speciﬁcation (2). Rather than interacting the conﬂict
indicator with the indicators for post-conﬂict months, the speciﬁcation interacts the capital indicator
with indicators equal to one if the household was surveyed in the months leading up to the capture


                                                   13
of Sana’a, and restricts the households in the sample to those surveyed in the three months prior to
the capture.37 , 38 In this speciﬁcation, the coeﬃcient on the diﬀerence-in-diﬀerence estimates denotes
by how much larger the change in expenditure was in the period leading up to the capture of Sana’a
than earlier in the year.
       Estimates of this falsiﬁcation test are presented in column (5) of table 2. The magnitude of the
estimate is smaller than the change found in November and December, and one cannot reject the
hypothesis that expenditure was not trending diﬀerently in 2014 than in 2005 prior to the capture of
the city. Furthermore, column (6) re-estimates the baseline speciﬁcation, but includes terms capturing
diﬀerential pre-trends in expenditure. However, the baseline result of a large change in December is
unchanged, one still cannot reject the hypothesis that expenditure was not trending prior to the
capture of Sana‘a, and we can reject the hypothesis that the change in December equaled the change
in the pre-conﬂict months at traditional signiﬁcance levels (p-value= 0.015).
       Table 3 further investigates the diﬀerence-in-diﬀerence estimate of the expenditure change in Sana‘a
following the capture of the city, as well as another potential mechanisms and coping strategies.
Columns (1) and (2) re-estimate speciﬁcation (2), but use the natural logarithm of food and non-
food expenditure as the dependent variables respectively. Estimates from both speciﬁcations are
qualitatively identical to the baseline estimates, where there was a large decline that was primarily
driven by larger and more precisely estimate declines in expenditure as the occupation progressed.
       This ﬁnding is especially striking given the fact that ﬁgure 2 demonstrated that the price of fresh
foods increased, suggesting that the decline in food consumption might be even more severe than
suggested by expenditure. Columns (3) and (4) investigate this possibility by re-estimating the baseline
speciﬁcation, but using daily per capita calorie and protein consumption as the dependent variables.
Both speciﬁcations demonstrate a qualitatively identical pattern where both calories consumed and
diet quality had large and more precisely estimated declines in December. The decline in calorie
consumption is consistent with a severe income shock, where oftentimes during negative income or
price shocks individuals are able to smooth overall calorie consumption by sacriﬁcing diet quality (e.g.,
D’Souza and Jolliﬀe 2014).
       This paper further explores a potential mechanism driving this decline in expenditure. Investigat-
ing the potential employment impacts of the capture of the city, column (5) re-estimates the baseline
speciﬁcation but utilizes total household income from wages as the dependent variable. The pattern
in wages is also qualitatively identical to the baseline pattern, where there was a large decline in wages
driven by larger and more precisely estimated declines as the occupation progressed. Interestingly,
the magnitude of the decline in wages, expenditure, and diet quality are all very similar.
       In addition to investigating causes of the decline in welfare, this paper also investigates poten-
tial coping strategies. First, this paper ﬁnds evidence that households signiﬁcantly increased their
borrowing. Column (6) re-estimates the baseline speciﬁcation, but uses an indicator for whether the
household had a loan as the dependent variable.39 Households signiﬁcantly increased their borrowing
following the capture of the city. However, unlike the rest of the patterns analyzed in table 3, there
were not large diﬀerences in the borrowing patterns in the months immediately after the capture,
and the last month of the survey after the occupation had progressed for three months. These re-
sults suggest that one of the reasons some of the households might have been able to maintain their
  37
     The added month indicators are included in the lower-order terms.
  38
     The results are robust to using any month prior to the capture of Sana‘a as the time period to demarcate the
pre-preiod.
  39
     Results are qualitatively similar when utilizing the total amount borrowed as well.



                                                       14
consumption in the month after the capture is through borrowing money.
       However, there is further evidence that some households were able to weather the initial months
of the capture better than others. Speciﬁcally, when estimating the eﬀects of the capture separately
for diﬀerent socio-economic groups as measured by factors that are not adjustable in response to the
capture, this paper demonstrates signiﬁcant diﬀerences in the response. Column (7) demonstrates
that households where the head did not ﬁnish primary school- households which were likely less well
oﬀ than others- observed an immediate decline in expenditure that was similar in magnitude and
precision to those in the months after the occupation had progressed.
       Alternatively, column (8) demonstrates a qualitatively identical pattern to the baseline speciﬁ-
cation. Households where the head did ﬁnish primary school were able to initially maintain their
expenditure in October, but the estimates ﬂipped signs and the magnitudes signiﬁcantly increased
during November and December. In this specifcation, one can reject the hypothesis that the change in
October was equal to the decline in December at conventional levels of signiﬁcance (p-value=0.061).
However, in the case of less well oﬀ households in column (7), the declines in October and December
were indistinguishable (p-value=0.967).40 , 41
       Importantly, estimates in columns (6)-(8) suggest that relatively advantaged households- those
that are able to access credit markets and have higher value assets- are able to better cope with the
capture of the capital. Thus, at least initially, the capture likely exacerbated pre-existing inequalities
that were already prevalent in the country and the region (e.g., Krishnan et al. 2016). However, the
shock was suﬃciently disruptive that, even within a matter of months, very few households were able
to maintain their consumption.

6. Women’s Empowerment

       One of the advantages of this setting is the thoroughness of the 2014 HBS survey, which allows one
to investigate the changes to a number of intra-household dynamics. As discussed above, the 2014 HBS
also included a very thorough investigation of women‘s empowerment. The Republic of Yemen had
large inequities in opportunities within the household based on gender prior to the capture of Sana‘a
(e.g., Krishnan et al. 2016), and it is important to understand whether those with fewer opportunities
and resources within the household were forced to bear a signiﬁcantly larger burden of shocks as has
been shown in other contexts (e.g., Rose 1999; Khanna et al. 2003; Miguel 2005; D’Souza and Tandon
forthcoming; etc.).
       In particular, the survey inquired about the control women had over a number of diﬀerent types of
assets and spending in the household. However, this module was included only in the 2014 survey, and
thus one cannot investigate the diﬀerence-in-diﬀerence comparing the 2014 change to the 2005 change.
Thus, this paper analyzes the simple diﬀerence in women‘s empowerment in Sana‘a and compares this
diﬀerence to the rest of the Republic of Yemen.
       Speciﬁcally, this section re-estimates speciﬁcation (2), but uses an indicator equaling one if the
spouse or the head of women-headed households made the decision to make a number of diﬀerent
types of purchases alone. Table 4 presents estimates of the women‘s empowerment speciﬁcations.
Each cell represents an estimate of either the change in empowerment following the capture of Sana‘a
or the diﬀerence-in-diﬀerence in Sana‘a relative to the rest of the Republic of Yemen. Columns (1)
  40
     All other available socio-economic indicators that are not themselves able to adjust to the capture yield qualitatively
similar results. Speciﬁcally, all the indicators of dwelling quality, or other cut-oﬀs for education, all yield similar results
where previously advantaged households are able to maintain their consumption.
  41
     The results are identical when estimating a model with a triple interaction (Advantaged ∗ P ost M onth ∗ Capital).



                                                              15
and (2) present estimates of the simple diﬀerence following the capture of Sanaa (β from speciﬁcation
1). Column (1) restricts the sample to households from the city of Sanaa; and column (2) restricts
the sample to households from the rest of the Republic of Yemen. Column (3) reports estimates of
the diﬀerence-in-diﬀerence following the capture of Sana‘a (β from speciﬁcation 4). Each row utilizes
a diﬀerent measure of women‘s empowerment as the dependent variable.
   Table 4 demonstrates that there was a dramatic reduction in women‘s empowerment in one of the
three types of spending that women actually had a signiﬁcant role in making decisions. Speciﬁcally,
row (1) of table 4 demonstrates there was a very large reduction in the share of women that made
decisions about food purchases in Sana’a, that did not occur in the rest of the Republic of Yemen.
The most complete speciﬁcation in column (3) demonstrates that there was a 15.4 percentage point
decline in the share of women who made decisions about food purchases.
   This change represents over three-quarters of the women who were making decisions regarding
food purchases prior to the capture, and the share of women who controlled food purchases in the city
of Sana’a approached and even dropped below levels seen in the rest of the Republic of Yemen. This
fact is even more striking given the relative lack of power women had in food purchase decisions in
the rest of the Republic of Yemen.
   The other two areas where women had signiﬁcant control over purchases were clothing for women
and children. The estimates suggest that there was a decline in the control over those purchases as
well, but the estimates are not as precisely estimated. However, one potential reason for the lower
precision might have been the frequency with which those purchases have to be made. Food has to
be purchased at relatively frequent intervals, whereas if women were not able to leave the household
during a tumultuous period, the purchase of women’s and children clothing could be delayed. Thus,
households might not have yet grappled with who would make purchases in othe dimensions that
required less frequent purchasing.
   Similarly, the estimates in table 4 demonstrate that the estimated change in control over expen-
ditures was negative for nearly all dimensions (elderly care is the only exception). Although the
estimates are not as precisely estimated as the change in control over food spending, nearly one-third
of the diﬀerence-in-diﬀerence estimates in column (3) are statistically signiﬁcant at conventional levels.
   Table 5 investigates this change in women’s decision making more thoroughly. First, columns (1)-
(2) investigate who might actually be taking over for women in the control of food purchases. Column
(1) demonstrates that there is a signiﬁcant increase in men having control over the food purchasing
decision. Tellingly, the magnitude of the coeﬃcient is roughly equal to the magnitude of the estimate
in the ﬁrst row and third column of table 4. On the other hand, column (2) demonstrates that there
appears to be very little change in other adult members taking over food purchases in the household.
Thus, the change in women’s decision making in the household appears to be transferred to the male
head.
   Alternatively, one might be concerned exactly what the question about decision making is cap-
turing. Women’s control over spending decisions and empowerment in general could be diﬃcult to
capture in simple survey instrument, and the responses might not align well with actual behaviors.
However, column (3) presents evidence demonstrating that there did appear to be a change in behavior
that was consistent with the change in responses. Speciﬁcally, the women’s decision-making module
was supposed to be administered to the most senior woman (whether head, spouse, or otherwise), but
in the case that either the spouse was not there or the head would not allow the spouse to answer,




                                                   16
other household members could respond to the module.42 , 43
       Following the capture of the city, not only was there a substantial reported transfer of decision
making authority from women to men, there was also a signiﬁcant increase in the share of women
who responded to the module. This is broadly consistent with women being more likely to be in the
household and less able to move about freely following the capture of the city. Following the capture
of the city, the increase in the share of women who answered the module was 12 percent larger in
Sana’a than in the rest of the Republic of Yemen.44 Thus, it does not appear that the change in
decision making reported in table 4 is only describing a change in peoples’ attitudes with little change
in behavior.
       Additionally, there are signiﬁcant diﬀerences between the change in women’s decision making and
the change in welfare described in Section 5. Speciﬁcally, column (4) estimates the changes in decision
making but allows the eﬀect to diﬀer by month of the occupation. Unlike the welfare estimates
presented in tables 2 and 3, there is an immediate drop in women’s decision making following the
capture, and very few households had women continue to control food purchases. This decline is
relatively consistent throughout the entire occupation period analyzed, where one cannot reject the
hypothesis that the change relative to the months prior to the occupation was identical in each month
(p-value 0.458).
       Furthermore, there was little diﬀerence in the change based on women’s bargaining power inside
the household. Using assets owned exclusively by the spouse as a proxy of bargaining power, columns
(5) and (6) re-estimate speciﬁcation (2) but restricts the sample to the approximately 45 percent of
the households where the spouse owns gold and the rest of households where spouse does not own
gold. The estimates are qualitatively identical, which both types of households demonstrating similar
declines in the share of women who controlled food purchases.45
       Similarly, the socio-economic status of households also did not seem to matter to the response.
Columns (7) and (8) re-estimate speciﬁcation (2), but restrict the sample to those in which the head
ﬁnished and did not ﬁnish primary school respectively. In contrast to the welfare results presented in
table 3, there was little diﬀerence in the change in women’s decision making between the two types of
households.
       All of these results suggest that the decline in women’s decision-making in households in Sana’a
following the capture of Sana’a was nearly universal, and did not depend on either the status of the
household or the women in the household. Given the two potential causes of the change in women’s
decision making- a worsening of household resources and a decline in safety- these patterns are mostly
consistent with the decline in women’s decision making being driven by a decline in safety. The change
was immediate and did not seem to change as households’ budgets got signiﬁcantly worse in November
and December. Furthermore, there was minimal amounts of variation based household characteristics
and the standing of the spouse in the household. All of these factors similarly point to the initial
decline in safety as opposed to the evolving nature of household budgets.
       Alternatively, table 6 investigates how women’s employment changed as the conﬂict progressed.
  42
     Men were allowed to sit in on the module if they required it, and this was indicated in the survey. However, there
was little diﬀerence in the answers between women who answered alone and those that answered in the presence of a
man.
  43
     Spouses answered for the vast majority of the households, where approximately 81 percent of households had a
woman respond to the module prior to the capture.
  44
     The baseline estimates described in table 4 are qualitatively identical if the sample is restricted to only households
in which the women answered the module, and also qualitatively identical if women answered the module not in the
presence of a man.
  45
     Gold was the only asset that spouses owned in non-negligible numbers.


                                                            17
As described in the Introduction, a number of news agencies and humanitarian agencies report that
women actually have to increase their employment and income generation in times of conﬂict due
to increased need. However, in every potential speciﬁcation investigated so far, there is no evidence
that women actually increased employment. In the baseline speciﬁcation in column (3), the estimate
is actually negative, and the conﬁdence interval is relatively small. Thus, the estimate precludes a
large increase in employment as the city was captured, and certainly demonstrates that the potential
increase in employment at the bounds of the conﬁdence interval was signiﬁcantly less than the share of
women who lost their ability to make decisions about food purchases in tables 4 and 5. Furthermore,
the estimates broken up by month in column (4) suggest that that employment could actually have
declined following the capture, which is roughly consistent with women not being able to leave the
household as freely as before the conﬂict.
       Lastly, ﬁgure 5 demonstrates that this lack of increase in employment might not be an artifact
of the lack of violent conﬂict and lack of extreme deprivation experienced in some other conﬂicts in
the world. Speciﬁcally, using data from the Gallup World Poll, which randomizes between male and
female respondents over the age of 15, there is little evidence that employment signiﬁcantly increased
in the Republic of Yemen as the conﬂict signiﬁcantly escalated.46

7. The Response of Subjective Welfare

A. Salience and Subjective Welfare

       Subjective welfare measures have become increasingly popular in the past decade, where global
reporting and progress towards some of the Sustainable Development Goals can be tracked using
subjective welfare measures (e.g., food security). One reason for this surge in popularity is the low
cost to collect subjective welfare measures and the speed with which these surveys can be conducted.
For example, the Food and Agriculture Organization (FAO) was able to insert a food consumption
module in the Gallup World Poll to construct a subjective welfare measure of food access at a cost of
just over $8,000 per country, and the entire survey took 2 weeks to collect in each country (Ballard et
al. 2013).
       The beneﬁts of subjective welfare measures are particularly important in fragile countries, where
primary data collection is diﬃcult and the need for up-to-the-date information is vital. One example
of the need for timely data was the need to track food availability in a number of fragile countries
that were faced with the threat of famine during 2017.47 However, given the dramatic changes that
are taking place in fragile countries, there exists the possibility that people’s frames, references, and
interpretation of survey questions that underlie many subjective welfare measures might not be stable.
       Prior research has conﬂicting conclusions regarding the suitability of subjective welfare measures.
Some researchers suggest that they are too diﬃcult to properly interpret in any context (e.g., Bertrand
and Mullainathan 2001), which is bolstered by the lack of alignment between subjective measures
of food security and more traditional measures such as undernourishment obtained from the same
respondents (e.g., Maxwell et al. 2014). Alternatively, others ﬁnd that there exist biases in the
reporting of subjective welfare measures that aﬀects the level of the estimates, but there is evidence
that these biases have limited impacts on the correlates of welfare in a number of contexts (e.g.,
Ravallion et al. 2016).
  46
    This is also evident in a regression framework.
  47
    See Al Jazeera (Accessed April 2017):             http://www.aljazeera.com/news/2017/03/famine-united-nations-
170310234132946.html.



                                                         18
       The 2014 HBS included a module of subjective satisfaction with a number facets of individuals‘
lives. Investigating how well these subjective measures might capture trends in traditional measures
of welfare (e.g., expenditure, diet choice, etc.), this paper presents a simple framework of how individ-
uals report experiences that incorporates salience. Households with homothetic preferences choose a
consumption bundle (C ) based on prices p and wealth w. Following the choice of consumption bundle,
it is assumed that individuals are asked a simple ”yes” or ”no” question regarding whether they are
satisﬁed with their consumption.
       However, it is further assumed that non-consumption factors can aﬀect this satisfaction, and that
in particular, the salience of these factors is allowed to vary relative to consumption factors. In this
framework, salience refers to the degree to which an individual might weight the non-consumption
factor relative to consumption in their decision. Speciﬁcally, an individual responds that she is satisﬁed
with consumption if the following decision rule is positive:


                                       Di = (1 − σ ) ∗ gi (Ci) + σ ∗ Si + θi


where gi denotes a function by which individuals compare their consumption to their ideal consumption
bundle, and it is assumed that higher consumption leads to individuals being closer to their ideal
consumption at a diminishing rate (e.g., g > 0, g < 0 for consumption below their ideal bundle).
However, it is assumed that consumption factors are explicitly compared to non-consumption factors
Si , which capture all factors that would make it more likely an individual would report being satisﬁed
for any level of consumption (e.g., S ≥ 0).48 Importantly, the salience of non-consumption factors S
are allowed to vary relative to consumption factors through weights denoted by σ (bounded between
zero and one), where higher values of σ imply a higher salience for S . The interpretation of changes
in σ are such that people are re-evaluating what satisfaction might mean.
       Lastly, θ represents unexpected events that might aﬀect an individual‘s reporting of satisfaction
(e.g., current events, individuals‘ moods at the time of the decision, etc.). It is further assumed that
θ is drawn from a common distribution function Fθ , where the unexpected component has a mean of
zero and larger unexpected swings aﬀecting reporting are less likely than smaller ones.49 Given the
unexpected component θi , the probability that individual i will respond aﬃrmatively that they are
satisﬁed with their consumption and how that probability changes as consumption increases can be
expressed as50 :

                                                φ = 1 − Fθ [−E (Di )]
                                       δφ
                                      δCi   = (1 − σ ) ∗ fθ [(−E (Di )] ∗ gi (Ci )

                                                                  δφ
Given the assumptions outlined above, it follows that            δCi   ≥ 0.
       This simple framework nests the case where no non-consumption factors can aﬀect the reporting of
satisfaction with consumption (σ = 0). In this case, increasing consumption increases the probability
of an individual reporting to be satisﬁed with consumption. However, an individual that consumes
close to their ideal consumption might still report being unsatisﬁed with their consumption depending
on the size of the unexpected swing, such as a deviation from an individual‘s expected mood, that
  48
     One example of information that could be contained in S is perceptions of safety of individual i, which might make
somebody care less about consuming below their ideal consumption.
  49
     Speciﬁcally, the density fθ is increasing over the interval (−∞, 0) and decreasing over the interval (0, ∞).
  50
     The inequality follows from the assumption that σ ∈ [0, 1] and the assumption that g > 0 for consumption below
one‘s ideal consumption bundle.


                                                           19
aﬀects their reporting. Similarly, there exists a chance that an individual who even consumes less than
their perceived minimum threshold might report being satisﬁed with their consumption. Furthermore,
increasing the variance of the unexpected swings makes it more likely for these reversals to occur for
individuals who had consumption very close to, or very far below, their ideal consumption.
       However, in the case where non-consumption factors explicitly factor in the reporting of satis-
faction (σ > 0), it follows that the salience of non-consumption factors aﬀects the sensitivity of the
probability of reporting satisfaction with consumption. In the limiting case of only non-consumption
factors aﬀecting reporting (σ = 1), the link between consumption and the probability of reporting sat-
isfaction is obscured entirely, and increasing consumption does not change the probability of reporting
satisfaction. However, for cases in which 0 < σ < 1, it follows that:

                  δ2 φ                                               δE (Di )
                 δCi δσ   = −fθ [−E (Di )] ∗ (1 − σ ) ∗ gi (Ci ) ∗     δσ       + fθ [−E (Di )] ∗ [−gi (Ci )]

                                                             δ2 φ
One can show under certain assumptions that                 δCi δσ   < 051 , which suggests that the more salient
non-consumption factors are relative to consumption, the less sensitive the probability of reporting
satisfaction is to increasing consumption. Based on this inequality, it follows that as non-consumption
factors become more salient, there exists a suﬃciently high salience σ such that actual consumption
relative to perceived thresholds has little eﬀect on reporting. In such a case, subjective measures of
welfare might not signiﬁcantly change even if there is a decline in actual consumption.
       However, this overly simple framework has assumed that changes in σ and changes in consumption
are independent of each other. But there exists the possibility that the onset of a large event could
cause both σ and C to co-move in opposite directions. In such a case, non-consumption factors
would be weighted more at exactly the time that consumption and welfare might be changing, and
one could ﬁnd that subjective welfare measures move in exactly the opposite direction of traditional
consumption-based measures. The capture of Sana‘a by Houthi forces is one such potential case,
where the capture of the city caused consumption and welfare to decline. But it is possible that
the importance of safety increased at the same time, and that change in weighting might have made
individuals more satisﬁed with their situation by not having suﬀered violence despite the decrease in
consumption.
       One can see this through the following:

                      δφ                        δσ                                      δCi       δσ
                      δV    = fθ [−E (Di )] ∗ [ δV ∗ g (Ci ) + (1 − σ ) ∗ g (Ci ) ∗     δV    +   δV   ∗ Si ]

where V is the likelihood that the individual will face violence in subsequent periods following the
capture of Sana‘a. If an increase in the likelihood an individual faces violence actually causes causes
                                              δσ
safety to become more salient (e.g.,          δV   > 0 ), then such a change combined with Section 5 demon-
                                                    δC
strating that consumption declined (e.g.,           δV   < 0) could actually result in a case where an individual
has a higher probability of reporting to be satisﬁed with consumption despite the decline in actual
                       δφ
consumption (i.e.,     δV   > 0).52
                                                                                                                     2
  51                                                                                                                δ φ
     Speciﬁcally, without specifying the distribution function Fθ , one can note that the ﬁrst expression of δC      i δσ
                                                                                                                          is
potentially positive, while the second expression is always negative. However, the ﬁrst expression is decreasing in size
as σ rises since all components of the expression are bounded as σ rises. Thus, for a suﬃciently high σ  ¯ , the magnitude
of the second expression, which does not go to zero as σ rises, will outweigh the magnitude of the ﬁrst expression and
the inequality will follow. Alternatively, one can assume speciﬁc distributions for θ, such as the Uniform Distribution,
for which the derivative of the pdf is zero. In such a case, the ﬁrst expression is zero, and the inequality follows for all
values of σ ∈ [0, 1].
  52
     One can see this for households that would in expectation report being unsatisﬁed with consumption when only



                                                             20
B. Empirical Results

    The 2014 HBS asked both the household head and the head‘s spouse whether they were very
satisﬁed, somewhat satisﬁed, not very satisﬁed, or not at all satisﬁed with a number of dimensions of
the households welfare. The dimensions that refer to physical welfare included whether the household
was satisﬁed with their food consumption, income, job, dwelling, health, and security.53 However,
similar to the women‘s empowerment module, the subjective welfare module was only included in the
2014 survey. Thus, this section re-estimates the speciﬁcations analyzing women‘s empowerment by
looking at how satisfaction changed following the capture of Sana‘a, and compares this estimate to
how satisfaction changed in the rest of the Republic of Yemen.
    Table 7 presents estimates of the subjective welfare speciﬁcations. Each cell represents an estimate
of either the change in subjective welfare following the capture of Sana‘a or the diﬀerence-in-diﬀerence
in Sana‘a relative to the rest of the Republic of Yemen. Column (1) reports speciﬁcations where the
head and spouse report to be ”not at all satisﬁed,” and the satisfaction level is increasing for each
additional column. Columns (1)-(5) restrict the sample to households from the city of Sanaa; and
column (6) restricts the sample to households from the rest of the Republic of Yemen. Column (7)
reports estimates of the diﬀerence-in-diﬀerence following the capture of Sana‘a (β from speciﬁcation
2). Each row utilizes satisfaction with a diﬀerent dimension of subjective welfare as the dependent
variable.
    The results in table 7 demonstrate a pattern that is seemingly at odds with the change in other
consumption-based welfare measures described above. Rather than demonstrating a decline in satis-
faction in many dimensions of welfare, the estimates of extreme dissatisfaction were actually negative
for all dimensions aside from security. Although one cannot reject the hypothesis that dissatisfaction
was not equal to zero, the 95 percent conﬁdence interval bounds the eﬀect such that one can reject
the hypothesis that there was a large increase in dissatisfaction. Similarly, column (4) demonstrates a
signiﬁcant increase in extreme satisfaction with food consumption, and the 95 percent conﬁdence in-
tervals for the other dimensions precludes a large decline. This is despite the fact observable measures
of food consumption and wages sharply dropped following the capture of Sana’a.
    Based on the pattern evident in columns (1)-(4), columns (5)-(7) investigate whether the share
of households that were either “very satisﬁed” or “fairly satisﬁed” changed following the capture.
Column (5) demonstrates the pattern in Sana’a is similar to the pattern presented in columns (1)-
(4). However, column (6) demonstrates there appears to be no large change in satisfaction outside
of Sana‘a. The magnitude of the coeﬃcients are much smaller, vary in sign, and one cannot reject
the hypothesis that there was no change in satisfaction in any of the dimensions. Column (7) further
investigates the diﬀerence-in-diﬀerence comparing the change in satisfaction in Sana‘a to the change
in the rest of the Republic of Yemen. All estimates aside from security are positive, implying that the
increases in satisfaction were larger in Sana‘a than in the rest of the Republic of Yemen.
    One explanation for these results is that in tumultuous times, households reassess their determi-
nation of what it means to be satisﬁed with particular facets of their life as other facets of their life
consumption factors are weighted in the decision (i.e., g (Ci ) < 0) and maintain their safety (i.e., Si > 0). In such a case,
                                                                        δφ
the ﬁrst and third term of the bracketed expression on the RHS of δV       are positive, while the middle term is negative.
However, for a suﬃciently large change in salience ( δV ) relative to the change in consumption ( δC
                                                     δσ
                                                                                                      δV
                                                                                                        i
                                                                                                          ), the positive terms
will outweigh the negative term, and the likelihood of reporting satisfaction will have increased in response to the shock
despite the drop in consumption.
  53
     There are also questions about whether individuals were satisﬁed with a number of social aspects of their life, such
as friends, etc.



                                                              21
become more salient (e.g., safety). However, another possibility is that households were not truthfully
reporting their subjective welfare fearing retribution from a government-sponsored survey. Although
it is impossible to rule out this hypothesis, there is little evidence for this interpretation. The pri-
mary role that the Houthi forces took over from the government was security, and that would be the
dimension that individuals would have the most incentive to falsely claim things were just as good as
before. However, respondents reported a sharp drop in security. Furthermore, The Central Statistics
Oﬃce was independent of the Houthi forces and took great care to remain politically neutral even as
the conﬂict progressed. However, regardless of the cause of the biases, these results demonstrate that
subjective welfare measures can directly contradict more traditional welfare measures, and that there
are many potential reasons that subjective welfare measures might be too diﬃcult to interpret in such
tumultuous circumstances.

8. Conclusion

   This paper demonstrates that in response to a relatively non-violent capture of a city, welfare
measures immediately declined, a number of coping strategies were employed, there were changes to
intra-household dynamics, and the changes to subjective measures of welfare substantially diﬀered
from observed household behavior and choices. These ﬁndings are evident even though this setting
lacked a number of aggravating factors that are generally present in longer and more violent conﬂicts,
such as worse disruptions to supply chains (e.g., port blockades) and worse shocks to economic activity.
Thus, these estimates might generalize some of the previous estimates of violent conﬂict to less violent
conﬂicts.
   However, there are a number of questions unanswered by the current context. First, these are all
short-term changes, and it is unclear what might happen in the long term. For example, the coping
strategies uncovered in this setting, such as loans, might eventually become unavailable and the level
of deprivation might signiﬁcantly worsen; more sustained conﬂict might lead to more deprivation and
cause women to be more rather than less engaged in household decision making out of necessity; and
it is unclear how households might adapt after the cessation of conﬂict to avoid long-term adverse
impacts of coping strategies (e.g., the speed with which households can improve diets and acquire
basic health services following the conﬂict, etc.).
   Second, as is the case in all conﬂicts, the times surrounding the empirical investigation in this
setting were tumultuous. There were protests, there was a quickly reversed removal of fuel subsidies
prior to the capture of Sana‘a, and there was lingering political uncertainty from the political transition
that occurred in 2012. Thus, it is possible that the results described here might be context-speciﬁc, and
it is possible that other countries might not experience similar shocks from the capture of territory
from the internationally recognized government. However, the robustness of the empirical ﬁndings
suggests that it is unlikely that any of these other factors are driving the results.
   Third, it is important to emphasize that it is diﬃcult to generalize the results from this setting to
present-day situation. As noted before, the conﬂict is more violent now than at the time of the 2014
HBS, there are substantial impediments to importing food, fuel, and medical equipment that were not
present during the capture of Sana‘a, and the escalation of the conﬂict has been going on for more
than two years. Thus, it is likely that the challenges currently presented to households are likely to be
substantially diﬀerent from those described in Sana‘a in 2014, which could result in a diﬀerent set of
coping strategies (e.g., internal displacement) or place a diﬀerent burden on disadvantaged household
members.


                                                      22
   Additionally, this paper investigates households’ subjective assessment of their satisfaction rather
than a more complete battery of experiential questions. Thus, it is diﬃcult to gauge whether peo-
ple could self-report their speciﬁc experiences with deprivation in accordance with more traditional
measures (e.g., undernourishment) if they were asked speciﬁc questions without implicit reference
points (e.g., inquiring about the number of meals people ate). Although other contexts have shown
that experiential food security indicators that are unlikely to be misinterpreted still do not align well
with measures of undernourishment based on household consumption (e.g., Maxwell et al. 2014), it is
possible that more speciﬁc experiential indicators might respond to adverse shocks in similar ways as
more traditional measures of welfare.




                                                   23
References

Abadie, A., and J. Gardeazabal. 2003. The Economic Costs of Conﬂict: A Case Study of the Basque
Country. American Economic Review 93 (1), 113-132.

Alesina, A., S. Ozler, N. Roubini, and P. Swagel. 1996. Political Instability and Economic Growth.
Journal of Economic Growth 1 (2), 189-211.

Amodio, F., and M. Di Maio. 2017. Making Do with What You Have: Conﬂict, Input Misallocation,
and Firm Performance. The Economic Journal, forthcoming.

Barrett, Christopher. 2002. “Food Security and Food Assistance Programs,” in Handbook of Agricul-
tural Economics vol. 2 (B), eds. B. Gardner and G. Rauser, 2103-2190. Amsterdam: Elsevier.

Ballard, T., A. Kepple, and C. Caﬁero. 2013. ”The Food Insecurity Experience Scale: Develop-
ing a Global Standard for Monitoring Hunger Worldwide.” Technical Paper, Food and Agriculture
Organization of the United Nations, Rome.


Bertrand, M., and S. Mullainathan. 2001. Do People Mean What They Say? Implications for
Subjective Survey Data. American Economic Review 91 (2), 67-72.

Blattman, C., and E. Miguel. 2010. Civil War. Journal of Economic Literature 48 (1), 3-57.

Blattman, C., and L. Ralston. “Generating Employment in Poor and Fragile States: Evidence from
Labor Market and Entrepreneurship Programs.” Working Paper, Columbia University, New York.

Blattman, C., and J. Annan. 2010. “The Consequences of Child Soldiering.” The Review of Economics
and Statistics 92 (4), 882-898.

Block, Steven A., Lynnda Kiess, Patrick Webb, S. Kosen, Regina Moench-Pfanner, Martin W. Bloem,
C. Peter Timmer. 2004. “Macro Shocks and Micro Outcomes: Child Nutrition During Indonesia‘s
Crisis.” Economics and Human Biology 2(1), 21-44.

Brinkman, Henk-Jan, Saskia de Pee, Issa Sanogo, Ludovic Subran, and Martin Bloem. 2009. “High
Food Prices and the Global Financial Crisis Have Reduced Access to Nutritious Food and Worsened
Nutritional Status and Health.” Journal of Nutrition 140(1), 153-161.

Cerra, V., and S. Saxena. 2008. Growth Dynamics: the Myth of Economic Recovery. American
Economic Review 98 (1), 439-457.

Chen, S., N. Loayza, and M. Reynal-Querol. 2008. The Aftermath of Civil War. The World Bank
Economic Review 22(1), 63-85. Collier, P., and M. Duponchel. 2013. The Econmic Legacy of Civil

War: Firm-Level Evidence from Sierra Leone. Journal of Conﬂict Resolution 57(1), 65-88.

D‘Souza, A., and D. Joliﬀe. 2012. ”Rising Food Prices and Coping Strategies: Household-level
Evidence from Afghanistan.” Journal of Development Studies 48 (2), 282-299.

D‘Souza, A., and D. Joliﬀe. 2014. ”Food Insecurity in Vulnerable Populations: Coping with Food
Price Shocks in Afghanistan.” American Journal of Agricultural Economics 96 (3), 790-812.

D‘Souza, A., and S. Tandon. Forthcoming. Intra-Household Nutritional Inequities in Rural Bangladesh.


                                                24
Economic Development and Cultural Change.

Eckstein, Z., and D. Tsiddon. 2004. Macroeconomic Consequences of Terror: Theory and the Case of
Israel. Journal of Monetary Economics 51 (5), 971-1002.

Edmonds, E., N. Pavcnik, and P. Topalova. 2010. ”Trade Adjustment and Human Capital Invest-
ments: Evidence from Indian Tariﬀ Reform.” American Economic Journal: Applied Economics 2 (3),
42-75.

Fafchamps, M., C. Udry, and K. Czukas. 1998. Drought and Saving in West Africa: are Livestock a
Buﬀer? Journal of Development Economics 55, 273-305.

Guidolin, M., and E. La Ferrara. 2007. Diamonds are Forever, Wars are Not: Is Conﬂict Bad for
Private Firms? American Economic Review 97 (5), 1978-1993.

Institute for Economics and Peace. 2017. “Global Terrorism Index.” Report, Institute for Economcis
and Peace, New York.

Jensen, R., and N. Miller. 2008. ”A Revealed Preference Approach to Measuring Hunger and Under-
nutrition.” NBER Working Paper No. 16555.

Jollife, D., and U. Serajuddin. 2012. Estimating Poverty with Panel Data, Comparably: An Example
from Jordan. Policy Research Working Paper 7373, World Bank, Washington, D.C.

Khanna, R., A. Kumar, J. Vaghela, and J. Puliyel. 2003. “Community Based Retrospective Study of
Sex in Infant Mortality in India.” British Medical Journal 327: 126-129.


Klapper, L., C. Richmond, and T. Tran. 2013. Civil Conﬂict and Firm Performance: Evidence from
Cote DIvoire. Working Paper, the World Bank, Washington, D.C.

Krishnan, N., M. Biradavolu, K. Carey, M. Groh, M. Lucas, D. Masterson, A. Newby, N. Palaniswamy,
F. Russo Riva, D. Sharma, and T. Vishwanath. 2017. “Syrian Refugees and their Hosts in Jordan,
Lebanon, and the Kurdistan Region of Iraq: Lives, Livelihoods, and Local Impacts.” Report, World
Bank, Washington, D.C.

Krishnan, N., G. Lara Ibara, A. Narayan, S. Tiwari, and T. Vishwanath. 2016. Uneven Odds, Unequal
Outcomes. Working Paper, World Bank, Washington, D.C.

Ksoll, C., R. Macchiavello, and A. Morjaria. 2015. Guns and Roses: Flower Exports and Electoral
Violence in Kenya. Working Paper, Warwick University.

Maxwell, D., B., Vaitla, and J. Coates. 2014. How do Indicators of Household Food Insecurity Measure
Up? An Empirical Comparison from Ethiopia. Food Policy 47 (0), 107-116.

Miguel, E. 2005. “Poverty and Witch Killing.” Review of Economic Studies 72 (4): 1153-1172.

Miguel, E., S. Satyanath, and E. Sergenti. 2004. Economic Shocks and Civil Conﬂict: An Instrumental
Variables Approach. Journal of Political Economy 112 (4), 725-753.

Mkutu, K. 2008. Uganda: Pastoral Conﬂict and Gender Relations. Review of African Political Econ-
omy 35 (116), 237-254.



                                                 25
Ravallion, M., K. Himelein, and K. Beegle. 2016. Can Subjective Questions on Economic Welfare be
Trusted? Economic Development and Cultural Change 64 (4), 697-726.

Singh, P., and J. Pedersen. 2003. Copying with Conﬂict: Palestinian Communities Two Years into
the Intifada. FAFO Report 408, Oslo, Norway.

Sundberg, R., and E. Melander. 2013. ”Introducing the UCDP Georeferenced Event Dataset.” Journal
of Peace Research 50 (4), 523-532.

Rose, E. 1999. “Consumption Smoothing andExcess Femaile Mortality in Rural India.” Review of
Economics and Statistics 81 (1): 41-49.

Upton, J., J. Cisse, and C. Barrett. Food Security as Resilience: Reconciling Deﬁnition and Measure-
ment. Working Paper, Cornell University, Ithaca, NY.

World Bank. 2011. Evidence and Lessons from Impact Evaluations on Social Safety Nets. Independent
Evaluation Group, World Bank, Washington.

World Bank. 2012. “World Development Report 2012: Gender Equality and Development.” Report,
World Bank, Washington, D.C.

World Bank. 2017. Free Falling: Living Standards During Times of Conﬂict in Yemen. Working
Paper, World Bank, Washington, D.C.

World Economic Forum. 2017. “Global Gender Gap Report 2017.” Report, World Economic Forum,
Geneva.




                                                26
Table 1. Differences between Control Variables in the Third and Fourth Quarters

                                                               Sampled Sampled Difference
                                                                in Third in Fourth (Quarter 4 -
                                                               Quarter Quarter Quarter 3)
                               Household Size                     8.855    7.877     -0.978
                                                                 [0.673] [0.345]     [0.752]
                               Household Head Married             0.899    0.878     -0.021
                                                                 [0.021] [0.033]     [0.039]
                               Household is Urban                 1.000    1.000        -
                                                                 [0.000] [0.000]        -
                               Head Attended School               0.696    0.748      0.052
                                                                 [0.051] [0.043]     [0.066]
                               Head a Smoker                      0.384    0.378     -0.006
                                                                 [0.051] [0.041]     [0.065]
                               Head Chews Qat                     0.829    0.825     -0.004
                                                                 [0.031] [0.036]     [0.047]
                               Years of Schooling- Head           7.936    9.184      1.248
                                                                 [0.630] [0.612]     [0.873]
                               Household Has a Child              0.958    0.983      0.025
                                                                 [0.018] [0.007]     [0.019]
                               Walls Made of Cut Stone            0.306    0.227     -0.079
                                                                 [0.060] [0.046]     [0.075]
                               Walls Made of Other Stone          0.509    0.645     0.136*
                                                                 [0.062] [0.048]     [0.077]
                               Roof Made of Reinforced Concrete 0.562      0.540     -0.023
                                                                 [0.075] [0.058]     [0.094]
                               Roof Made of Wood and Concrete 0.268        0.283      0.015
                                                                 [0.064] [0.056]     [0.085]
                               Roof Made of Mud                   0.170    0.177      0.007
                                                                 [0.049] [0.055]     [0.073]
                               Floor Made of Concrete             0.232    0.297      0.065
                                                                 [0.046] [0.051]     [0.068]
                               Floor Made of Tile                 0.627    0.623     -0.004
                                                                 [0.066] [0.058]     [0.087]
                               Household has a Flush Toilet       0.309    0.225     -0.084
                                                                 [0.052] [0.045]     [0.068]
                               Notes: This table reports the average characteristics of surveyed households in the
                               third and fourth quarter in the city of Sana'a in 2014 respectively in columns (1) and (2),
                               and further reports that simple difference in column (3). The survey was designed such
                               that the households surveyed such that there should be no differences between
                               households between months of the survey, and thus if the samplings was done
                               according to plan, the two estimates should be similar. There were 266 households
                               surveyed in quarter 3 and 231 households surveyed in quarter 4. Estimates are
                               weighted so as to be representative of the population, and standard errors are
                               clustered at the PSU level. The statistical significance of the difference is reported for
                               column (3). *** denotes statistical significance at the 1 percent level; ** denotes
                               significance at the five percent level; and * denotes significance at the 10 percent
                               level.




                                                                           27
Table 2. Expenditure Changes in Response to the Capture

                                               Dependent Variable:


                                                                                     ln(Expenditure)
                                                            _______________________________________________________________
                                                                                              Double
                                                                               Double      Difference 2:
                                                                            Difference 1:   How Much
                                                                          How Much Larger Larger Effects
                                                                           Effects were in    were in
                                                                           Sana'a Relative  Sana'a in
                                                                            to the Rest of 2014 than in    Triple   Including Pre-Conflict
                                                  Single Difference         Yemen- 2014        2005      Difference         Trends
                                                   (1)          (2)               (3)           (4)          (5)        (6)        (7)
              Post-Capture Months
        Effect of Capture in October             -0.083        -0.069           -0.097                  -                -              -         -0.063
                                                 [0.191]      [0.187]          [0.233]                  -                -              -        [0.275]
        Effect of Capture in November            -0.223        -0.203           -0.265               -0.131           -0.299            -         -0.232
                                                 [0.159]      [0.142]          [0.174]              [0.161]          [0.221]            -        [0.228]
        Effect of Capture in December          -0.403***     -0.374***        -0.428***            -0.375**         -0.462**            -        -0.394*
                                                 [0.133]      [0.107]          [0.152]              [0.148]          [0.215]            -        [0.211]
             Pre-Capture Months
        Trend in August                             -             -                   -                 -                -          0.159          0.159
                                                    -             -                   -                 -                -         [0.246]        [0.244]
        Trend in September                          -             -                   -                 -                -          -0.075         -0.039
                                                    -             -                   -                 -                -         [0.252]        [0.254]
        Control Variables                           N             Y                   Y                 Y                Y             Y              Y
        Includes 2005 households                    N             N                   N                 Y                Y             N             N
        Excludes Households surveyed
                                                                                                                                                     N
        after the capture                           N             N                   N                 N               N              Y
        Includes Rest of Yemen                      N             N                   Y                 N                Y             Y              Y
        P-value of Test of Coefficients
        on all Post-Capture Effects             0.026**      0.006***           0.039**             0.039**          0.088*          0.626         0.217
        Equaling Zero
        P-value of Test of Coefficients
        on Pre-Trend Months Equaling                -             -                   -                 -                -              -        0.014**
        December Coefficient
        Observations                               504          504               4317                2207            9097           2189          4317
        Notes: This table estimates the change in expenditure following the capture of the Yemeni capital by rebels. Columns (1) and (2) include only
        households from Sana'a surveyed in 2014, and estimates the change in expenditure by month of capture; column (3) includes households from the
        rest of Yemen, and the estimates represent by how much more expenditure changed in Sana'a than the rest of Yemen in 2014; column (4) restricts
        the sample to households surveyed in Sana'a in 2014 and 2005, and the estimates represent by how much more expenditure changed in Sana'a in
        2014 than in 2005; column (5) includes all households surveyed in 2014 and 2005, and the estimates how much larger the difference-in-difference
        in expenditure was in Sana'a than in the rest of Yemen; column (6) restricts the sample to households surveyed in all of Yemen in 2014 but
        restricts the sample to those surveyed before the capture to estimate the change in expenditure in the months leading up to the capture relative
        to the baseline month; and column (7) re-estimates the difference-in-difference specificaiton in column (3), but includes the pre-capture trends.
        All specifications exclude households surveyed during the month of Ramadan- July in 2014 and October in 2005. For specifciations comparing
        2014 to 2005, households surveyed in both months were excluded from all surveys. All specifications include all lower-order terms and control
        variables. Control variables include all variables reported in Table 1 except for household size (the dependent variable is per-capita). Standard
        errors clustered at the PSU level are reported in parentheses in columns (1)-(2) and in column (4); columns (3) and columns (5)-(7) report the larger
        of standard errors clustered at the PSU level or clustered at the governorate level; all estimates are weighted so as to be representative of the
        entire population. *** denotes statistical significance at the 1 percent level; ** denotes significance at the five percent level; and * denotes
        significance at the 10 percent level.

                                                                                 28
Table 3. Changes in Other Measures of Welfare and Coping Strategies

                                             Dependent Variable:
                                                                                                                                  Indicator
                                                                       ln (Daily                ln(Daily                          for
                                                          ln (Non-     Household                Household                         whether
                                             ln (Food     Food         Calorie                  Protein      ln (Household        Household ln (Per Capita Daily
                                             Expenditure) Expenditure) Consumption)             Consumption) Income)              had Loan Expenditure)
                                                    (1)             (2)                (3)             (4)              (5)             (6)         (7)         (8)
            Post-Capture Months
Effect of Capture in October                      0.010           -0.104            0.215            0.024             -0.012         0.138       -0.459*      0.127
                                                 [0.216]         [0.235]           [0.224]          [0.221]           [0.147]        [0.104]      [0.242]     [0.277]
Effect of Capture in November                    -0.118           -0.278            0.141            -0.197            -0.202         -0.002     -0.418**     -0.247
                                                 [0.177]         [0.176]           [0.165]          [0.244]           [0.138]        [0.089]      [0.176]     [0.197]
Effect of Capture in December                   -0.381**        -0.433***         -0.346**         -0.349**          -0.334**        0.174**    -0.533***    -0.393**
                                                 [0.148]         [0.157]           [0.173]          [0.160]           [0.162]        [0.075]      [0.187]     [0.163]
Restrict Households to Those Where
Household Head Did Not Finish Primary               N                N                 N               N                N               N            Y           N
School
Restrict Households to Those Where
                                                    N                N                 N               N                N               N            N           Y
Household Head Finished Primary School
P-value of Test of Coefficients on Joint
                                                  0.060*          0.046**           0.147            0.130            0.062*         0.067*       0.025**     0.062*
Significance of Higher-Order Terms
P-value of Test of Coefficients on October
and December Higher-Order Terms Being             0.130            0.353            0.288            0.237           0.026**          0.135        0.760      0.065*
Equal
Observations                                      4317             4317             4317             4317              2209           4317         1636        2581
Notes: This table estimates the change in welfare and coping strategies following the capture of the Yemeni capital by rebels. Columns (1)-(6) utilize different
measures of welfare and coping strategies, and columns (7) and (8) estimate the change in expenditure for households of different socioeconomic status (whether
household head finished primary school) separately. All specifications exclude households surveyed during the month of Ramadan (July). All specifications include
all lower-order terms and control variables. Control variables include all variables reported in Table 1. The larger of standard errors clustered at the PSU level or
clustered at the governorate level are reported in parentheses; all estimates are weighted so as to be representative of the entire population. *** denotes statistical
significance at the 1 percent level; ** denotes significance at the five percent level; and * denotes significance at the 10 percent level.




                                                                                  29
Table 4. Changes in Women’s Decision Making in Response to the Capture
                                                                                       Coeffeicient on Conflict
                                                         Coeffeicient on Conflict           Indicator for         Coefficient Yielding How much larger
                                                       Indicator for Households in     Households in the Rest      the change in Sana'a was than the
                                                               Sana'a Only                    of Yemen                        Rest of Yemen
                                                               __________                    ___________                        _________
                Dependent Variable:                                 (1)                          (2)                               (3)
                Women Responsible for Purchase
                of Food                                         -0.136***                       0.021                          -0.154***
                                                                 [0.042]                       [0.022]                          [0.055]
                Women Responsible for Purchase
                of Heads Clothes                                  -0.012                        0.007                            -0.021
                                                                 [0.019]                       [0.015]                          [0.026]
                Women Responsible for Purchase
                of Womens' Clothes                                -0.074                        0.007                            -0.067
                                                                 [0.082]                       [0.034]                          [0.089]
                Women Responsible for Purchase
                of Childrens' Clothes                             -0.041                        0.006                            -0.039
                                                                 [0.041]                       [0.022]                          [0.049]
                Women Responsible for Purchase
                of Medicine for Spouse                            -0.036                       0.014**                          -0.048*
                                                                 [0.024]                       [0.007]                          [0.025]

                Women Responsible for Purchase
                of Medicine for Other Adult Women                 -0.016                       -0.000                            -0.019
                                                                 [0.020]                       [0.006]                          [0.020]
                Women Responsible for Purchase
                of Medicine for Children                          -0.031                        0.002                           -0.036*
                                                                 [0.019]                       [0.008]                          [0.021]
                Women Responsible for
                Expnediture on Son's Marriage                     -0.016                        0.001                           -0.018*
                                                                 [0.010]                       [0.003]                          [0.011]
                Women Responsible for
                Expnediture on Daughter's
                Marriage                                          -0.013                        0.005                            -0.019
                                                                 [0.010]                       [0.004]                          [0.012]
                Women Responsible for
                Expenditure on Boys' Education                    -0.011                       -0.000                            -0.014
                                                                 [0.008]                       [0.009]                          [0.014]
                Women Responsible for
                Expenditure on Girls' Education                   -0.019                        0.001                            -0.022
                                                                 [0.012]                       [0.009]                          [0.015]
                Women Responsible for
                Expenditure on Elderly Care                       0.005                         -0.008                           0.014
                                                                 [0.028]                       [0.023]                          [0.041]
                Women Responsible for Taking on
                Debt                                              -0.004                       -0.002                            -0.006
                                                                 [0.019]                       [0.007]                          [0.020]
                Observations                                       503                          3,789                            4,292
                Notes: This table estimates how women's decision-making ability changed following the capture of Sana'a. Each cell reports
                coefficients from a separate regression. Column (1) reports the simple difference in the share of households where women were
                responsible for the decision described in each row; column (2) reports the same specification but restricts the sample to the rest of
                Yemen; and column (3) presents estimates by how much larger the change was in the city of Sana'a than in the rest of Yemen. The
                difference-in-difference specifcation in column (3) includes all lower-order terms, and all specificaitons include control variables.
                Control variables include all variables reported in Table 1. Specifications in column (1) report standard errors clustered at the PSU
                level in parentheses; and columns (2)-(3) report the larger standard error from either a specifcaiton clustering the standard errors at
                the PSU level or clustering the standard errors at the governorate level. Estimates are weighted so as to be representative of the
                entire population. *** denotes statistical significance at the 1 percent level; ** denotes significance at the five percent level; and *
                denotes significance at the 10 percent level.


                                                                                  30
Table 5. A Closer Examination of Decision Making in the Household

                                                  Dependent Variable:
                                                               Indicator
                                                  Indicator if     if            Indicator
                                                    Head is     Neither          if Spouse
                                                  Involved in Spouse             Answered
                                                   the Food nor Head             Decision-
                                                   Purchase       are             Making              Indicator if Woman Responsible for Food
                                                   Decision Involved              Module                             Purchases
                                                   ________ _______               _______           ________________________________________
                                                         (1)            (2)            (3)           (4)           (5)           (6)           (7)           (8)
    Post-Conflict Indicator*Sana'a
    Indicator                                        0.167**          -0.013      0.120***      -    -0.182*                 -0.131** -0.132** -0.173**
                                                     [0.069]         [0.042]       [0.043]      -    [0.093]                  [0.051] [0.059] [0.072]
    Effect of Capture in October                        -                -            -     -0.163**    -                        -        -        -
                                                        -                -            -      [0.074]    -                        -        -        -
    Effect of Capture in November                       -                -            -    -0.196***    -                        -        -        -
                                                        -                -            -      [0.065]    -                        -        -        -
    Effect of Capture in December                       -                -            -      -0.101     -                        -        -        -
                                                        -                -            -      [0.067]    -                        -        -        -
    Restrict Sample to Those Where
                                                         N               N             N             N              Y            N             N             N
    Spouse Owned Gold
    Restrict Sample to Those Where
                                                         N               N             N             N             N              Y            N             N
    Spouse Did Not Own Gold
    Restrict Sample to those where Head
                                                         N               N             N             N             N             N              Y            N
    Did Complete Primary School
    Restrict Sample to those where Head
                                                         N               N             N             N             N              N            N             Y
    Did Not Complete Primary School

    P-value of Test of Coefficients on Joint              -              -              -         0.019**           -             -             -             -
    Significance of Higher-Order Terms
    P-value of Test of Coefficients on
    October and December Higher-Order                     -              -              -          0.368            -             -             -             -
    Terms Being Equal
    Observations                                       4292            4292          4292           4292         1000          3292          2668          1624
    Notes: This table investiages a number of different specifications better describing the change in women's decision making following the capture of Sana'a.
    Each coefficient represents how much larger the change in the dependent variable was in Sana'a than the rest of Yemen in 2014. Column (1) utilizes an
    indicator equaling one if the household head was involved in the food purchase decision; column (2) utilizes an indicator equaling one if niether the head nor
    the spouse is involved in the decision to purchase food; column (3) utilizes an indicator equaling one if the spouse (or head if the household head was a
    women) responded to the decision-making module; and columns (4)-(9) utilize an indicator if the spouse (or head if the household head was a woman) was in
    charge of the decision to purchase food. Columns (1)-(4) include the whole sample, while columns (5)-(8) restrict households based on measures of women's
    bargaining power and socioeconomic status. All specifications exclude households surveyed during the month of Ramadan (July). All specifications include all
    lower-order terms and control variables. Control variables include all variables reported in Table 1. The larger of standard errors clustered at the PSU level or
    clustered at the governorate level are reported in parentheses; all estimates are weighted so as to be representative of the entire population. *** denotes
    statistical significance at the 1 percent level; ** denotes significance at the five percent level; and * denotes significance at the 10 percent level.




                                                                                  31
Table 6. Changes in Women’s Employment in Response to the Capture

                            Dependent Variable: Indicator Equaling One if Spouse is Employed
                                                                                                    Restrict Sample                                      Restrict Sample
                                                                                           Restrict  to Women in                                          to Women in
                                          Restrict                              Restrict  Sample to   Households                                           Households
                             Restrict     Sample Difference-in- Estimate       Sample to Women who where Head                                            where Head Did
                             Sample      to Rest of Difference (col Change by Women who Do Not Own    Completed                                           Not Complet
                            to Sana'a     Yemen       (1)-col (2))    Month    Own Gold     Gold    Primary School                                       Primary School
                             _______      _______ ___________ __________ __________ __________ ____________                                              _____________
                                 (1)          (2)              (3)                 (4)               (5)               (6)                 (7)                    (8)
Post-Conflict Indicator       -0.005        0.005            0.005                  -              0.002             0.007               -0.000                 0.010
                              [0.028]      [0.006]          [0.006]                 -             [0.012]           [0.007]             [0.009]                [0.007]
Post-Conflict
Indicator*Sana'a
Indicator                         -            -             -0.014                 -              0.043             -0.037              -0.020                 0.013
                                  -            -            [0.029]                 -             [0.048]           [0.034]             [0.039]                [0.020]
Effect of Capture in
October                           -            -                -               -0.049*               -                 -                   -                      -
                                  -            -                -               [0.030]               -                 -                   -                      -
Effect of Capture in
November                          -            -                -                0.054                -                 -                   -                      -
                                  -            -                -               [0.051]               -                 -                   -                      -
Effect of Capture in
December                          -            -                -               -0.048*               -                 -                   -                      -
                                  -            -                -               [0.026]               -                 -                   -                      -
P-value of Test of
Coefficients on Joint
                                  -            -                -                0.055*               -                 -                   -                      -
Significance of Higher-
Order Terms
P-value of Test of
Coefficients on October
and December Higher-              -            -                -                0.984                -                 -                   -                      -
Order Terms Being
Equal
Observations                    504         3,813             4,317              4,317             1,000             3,317               2,681                  1,636

Notes: This table investiages a number of different specifications describing the change in women's employment following the capture of Sana'a. Columns (1) and (2) estimate a
simple change in employment following the capture of Sana'a in the city of Sana'a and the rest of Yemen respectively. Columns (3)-(8) estimate how much larger the change in
women's employment was in Sana'a than the rest of Yemen in 2014. Columns (1)-(4) utilze the entire sample, and columns (5)-(8) restrict households based on measures of
women's bargaining power and socioeconomic status. All specifications exclude households surveyed during the month of Ramadan (July). All specifications include all lower-
order terms and control variables. Control variables include all variables reported in Table 1. The larger of standard errors clustered at the PSU level or clustered at the
governorate level are reported in parentheses; all estimates are weighted so as to be representative of the entire population. *** denotes statistical significance at the 1
percent level; ** denotes significance at the five percent level; and * denotes significance at the 10 percent level.




                                                                                        32
Table 7. Changes in Satisfaction Following the Capture

                                                                                                             How much Larger the
                                  Increase                                                                     Increase in Share
                                  in Share  Increase Increase Increase                                           Reporting any
                                   "Not At  in Share in Share in Share                                        Satisfaction was in
                                      All  "Not Very "Fairly       "Very    Increase in Share Either "Very Sana'a than Rest of
                                 Satisfied"Satisfied" Satisfied" Satisfied" Satisfied" or "Fairly Satisfied"        Yemen
                                   ______    ______    ______     ______      ________________________          _____________
                                      (1)       (2)       (3)       (4)        (5)              (6)                    (7)
    Security                        -0.080    0.030      0.032 -0.073*** -0.125                0.037                 -0.178
                                   [0.079]   [0.077]   [0.078]    [0.024]    [0.094]          [0.040]               [0.112]
    Food                            -0.099    -0.009    0.060*    0.026**    0.067*           -0.027                 0.094*
                                   [0.063]   [0.056]   [0.034]    [0.013]    [0.036]          [0.036]               [0.052]
    Income                          -0.022    -0.085     0.066     0.031    0.121**            0.017                 0.089
                                   [0.043]   [0.057]   [0.047]    [0.033]    [0.055]          [0.040]               [0.075]
    Work                            -0.036    0.024      0.025     0.001      0.043           -0.012                 0.048
                                   [0.051]   [0.058]   [0.049]    [0.020]    [0.055]          [0.037]               [0.069]
    Health                          0.001     -0.029     0.026     0.019      0.033           -0.017                 0.052
                                   [0.072]   [0.062]   [0.030]    [0.015]    [0.039]          [0.026]               [0.045]
    Dwelling                      -0.127**    0.003      0.041     0.040      0.093           -0.036                 0.114
                                   [0.053]   [0.065]   [0.060]    [0.034]    [0.064]          [0.035]               [0.075]
    Restrict the Sample
                                       Y             Y              Y              Y             Y                    N                              N
    to the city of Sana'a
    Restrict the Sample
                                       N             N              N              N             N                     Y                             N
    to the rest of Yemen

    Observations in each
    of the regressions               504            504            504           504            504                 3813                           4317
    reported in the
    column
    Notes: This table estimates how peoples' self-reported satisfaction with a number of dimensions of welfare changed following the capture of Sana'a. Each cell
    represents a coefficient of a separate regression of an indicator equaling one if either the head or the head's spouse reported a particular type of satisfaction
    with a particular dimension on an indicator equaling one if the household was interviewed following the capture of Sana'a. Column (1) reports the change in
    the share of households that were "not at all satisfied"; column (2) reports the change in share of households that were "not very satisfied"; column (3) reports
    the change in the share of households that were "fairly satisfied"; and column (4) reports the change in the share of households that were "very satisfied."
    Columns (5)-(7) report the change in the share of households that were either "very satisfied" or "fairly satisfied." Columns (1)-(5) restrict the sample to
    households from Sana'a, and report the simple increase in the share of households that had a particular level of satisfaction. Column (6) re-estimates the
    same specification, but restricts the sample to households surveyed in the rest of Yemen. Column (7) reports estimates of how much larger the increase in
    satisfication was in Sana'a than the rest of Yemen. All specifications include control variables, and column (7) includes lower-order terms. Control variables
    include all variables reported in Table 1. Columns (1)-(5) report standard errors clustered at the PSU level in parentheses; and columns (6)-(7) reports the
    larger of standard errors clustered at the PSU level and standard errors clustered at the governorate level. All estimates are weighted so as to be
    representative of the entire population. *** denotes statistical significance at the 1 percent level; ** denotes significance at the five percent level; and *
    denotes significance at the 10 percent level.




                                                                                   33
Figure 1. Average Deaths per Year in Recent Middle East Conflicts



                           45000
                           40000
                           35000
                           30000
                           25000
                           20000
                           15000
                           10000
                            5000
                                 0
                                              Iraq                Libya (2011                 Syria                 Yemen
                                                                  Revolution)

                                                       Average Estimated Deaths per Year Civilian
                                                       Average Estimated Deaths per Year Non-Civilian

                         Notes: The figures for Iraq were obtained from Iraq Body Count, and only include the casualty figures for 2014;
                         the figures for Libya were obtained from United Nations High Commissioner for Human Rights and the Libyan
                         government; the figures for Syria were obtained from the Syrian Observatory for Human Rights; and the figures
                         for Yemen were obtained from the United Nations High Commissioner for Human Rights.




                                                                               34
Figure 2. Changes in Commodity Prices in Sana’a Following the Capture




                         Figure 2a. Commodity Prices in Sana’a in 2014
                          500
                          450
                          400
                          350
                          300
                          250
                          200
                          150
                          100
                           50
                            0




                                                   Price Before September           Price After September

                         Figure 2b. Commodity Prices in Sana’a in 2012-2013
                           500
                           450
                           400
                           350
                           300
                           250
                           200
                           150
                           100
                            50
                             0




                                                   Price Before September           Price After September

                         Notes: These figures shows average commodity prices in Sana’a markets in 2014, and the average for the two
                         years prior to the capture of the capital. These figures were obtained from the monthly database maintained
                         by the World Food Programme.




                                                                             35
Figure 3. Monthly Confidence Intervals of Control Variables

 Figure 3a- Summary Statistics by Month
 1) Household Size                                            2) Household Head Married
 15                                                                  1

 13                                                            0.95
                                                                0.9
 11
                                                               0.85
  9
                                                                0.8
  7                                                            0.75

  5                                                             0.7
          Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec                 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
                           Lower Bound- 95% CI                                               Lower Bound- 95% CI
                           Upper Bound- 95% CI                                               Upper Bound- 95% CI
                           Pre-Conflict Mean                                                 Pre-Conflict Mean

 3) Household Head Never Attended School                      4) Household Head a Smoker
   1.2                                                         0.7
      1                                                        0.6

   0.8                                                         0.5
                                                               0.4
   0.6
                                                               0.3
   0.4
                                                               0.2
   0.2                                                         0.1
      0                                                          0
           Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec               Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
                                  Lower Bound- 95% CI                                           Lower Bound- 95% CI
                                  Upper Bound- 95% CI                                           Upper Bound- 95% CI
                                  Pre-Conflict Mean                                             Pre-Conflict Mean

 5) Household Head Chews Qat                                  6) Years Schooling
  1.1                                                          13

      1                                                        11

  0.9                                                           9

                                                                7
  0.8
                                                                5
  0.7
                                                                3
  0.6                                                                    Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
           Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
                             Lower Bound- 95% CI                             Lower Bound- 95% CI       Upper Bound- 95% CI
                             Upper Bound- 95% CI                             Pre-Conflict Mean
                             Pre-Conflict Mean
 Source: 2014 Household Budget Survey, Author’s Calculation




                                                                             36
Figure 3b- Summary Statistics by Month
7) Household has a Child                                     8) Household has a Flush Toilet
       1                                                       0.6
   0.95
                                                               0.4
    0.9
   0.85                                                        0.2
    0.8
                                                                    0
                                                                         Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
                                  Lower Bound- 95% CI                                        Lower Bound- 95% CI
                                  Upper Bound- 95% CI
                                  Pre-Conflict Mean                                          Upper Bound- 95% CI
                                                                                             Pre-Conflict Mean

9) Walls Composed of “Cut Stone”                             10) Walls Composed of Other Material
   0.8                                                              1

                                                               0.8
   0.6
                                                               0.6
   0.4
                                                               0.4
   0.2                                                         0.2
                                                                         Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
       0
                                                                                           Lower Bound- 95% CI
           Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
                                                                                           Upper Bound- 95% CI
                               Lower Bound- 95% CI                                         Pre-Conflict Mean
                               Upper Bound- 95% CI
                               Pre-Conflict Mean

11) Roof Composed of Reinforced Concrete                     12) Roof Composed of Wood and Concrete
   1                                                          0.8
 0.8
                                                              0.6
 0.6
 0.4                                                          0.4
 0.2
                                                              0.2
   0
           Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec     0
                                                                        Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
                                 Lower Bound- 95% CI                                       Lower Bound- 95% CI
                                 Upper Bound- 95% CI                                       Upper Bound- 95% CI
                                 Pre-Conflict Mean                                         Pre-Conflict Mean
Source: 2014 Household Budget Survey, Author’s Calculation




                                                                         37
Figure 3c- Summary Statistics by Month
13) Roof Composed of Mud                                         14) Floor Composed of Concrete
  0.6                                                             0.6


  0.4                                                             0.4


  0.2                                                             0.2


                                                                    0
    0
                                                                        Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
         Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
                                                                                               Lower Bound- 95% CI
                                 Lower Bound- 95% CI                                           Upper Bound- 95% CI
                                 Upper Bound- 95% CI                                           Pre-Conflict Mean

                                             15) Floor Composed of Natural Tiles
                                       1

                                     0.8

                                     0.6

                                     0.4

                                     0.2
                                           Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
                                                               Lower Bound- 95% CI
                                                               Upper Bound- 95% CI
                                                               Pre-Conflict Mean
Source: 2014 Household Budget Survey, Author’s Calculation




                                                                             38
Figure 4. Change in Annual Expenditure Following the Capture




                         3500000

                         3300000

                         3100000

                         2900000

                         2700000

                         2500000

                         2300000

                         2100000

                         1900000

                         1700000

                         1500000
                                          Jun          Jul          Aug          Sep           Oct       Nov        Dec

                                                Sana'a-2014           Sana'a-2005              Rest of Yemen-2014

                        Source: Household Budget Survey 2005 and 2014, author’s calculation.




                                                                           39
Figure 5. Share of Women Employed Escalated



                          0.2
                         0.18
                         0.16
                         0.14
                         0.12
                          0.1
                         0.08
                         0.06
                         0.04
                         0.02
                            0
                                             2014                           2015                            2016

                                                          Full Time Work            Part Time Work

                       Source: Gallup World Poll, 2014-2016, author’s calculation. Figure summarizes the average employment status
                       of female respondents.




                                                                           40
              Poverty & Equity Global Practice Working Papers
                                         (Since July 2014)

The Poverty & Equity Global Practice Working Paper Series disseminates the findings of work in progress
to encourage the exchange of ideas about development issues. An objective of the series is to get the
findings out quickly, even if the presentations are less than fully polished. The papers carry the names of
the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in
this paper are entirely those of the authors. They do not necessarily represent the views of the
International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or
those of the Executive Directors of the World Bank or the governments they represent.

This series is co‐published with the World Bank Policy Research Working Papers (DECOS). It is part of a
larger effort by the World Bank to provide open access to its research and contribute to development
policy discussions around the world.

For the latest paper, visit our GP’s intranet at http://POVERTY.

   1     Estimating poverty in the absence of consumption data: the case of Liberia
         Dabalen, A. L., Graham, E., Himelein, K., Mungai, R., September 2014

   2     Female labor participation in the Arab world: some evidence from panel data in Morocco
         Barry, A. G., Guennouni, J., Verme, P., September 2014

   3     Should income inequality be reduced and who should benefit? redistributive preferences in Europe
         and Central Asia
         Cojocaru, A., Diagne, M. F., November 2014

   4     Rent imputation for welfare measurement: a review of methodologies and empirical findings
         Balcazar Salazar, C. F., Ceriani, L., Olivieri, S., Ranzani, M., November 2014

   5     Can agricultural households farm their way out of poverty?
         Oseni, G., McGee, K., Dabalen, A., November 2014

   6     Durable goods and poverty measurement
         Amendola, N., Vecchi, G., November 2014

   7     Inequality stagnation in Latin America in the aftermath of the global financial crisis
         Cord, L., Barriga Cabanillas, O., Lucchetti, L., Rodriguez‐Castelan, C., Sousa, L. D., Valderrama, D.
         December 2014

   8     Born with a silver spoon: inequality in educational achievement across the world
         Balcazar Salazar, C. F., Narayan, A., Tiwari, S., January 2015

                               Updated on June 2018 by POV GP KL Team | 1
9    Long‐run effects of democracy on income inequality: evidence from repeated cross‐sections
     Balcazar Salazar,C. F., January 2015

10   Living on the edge: vulnerability to poverty and public transfers in Mexico
     Ortiz‐Juarez, E., Rodriguez‐Castelan, C., De La Fuente, A., January 2015

11   Moldova: a story of upward economic mobility
     Davalos, M. E., Meyer, M., January 2015

12   Broken gears: the value added of higher education on teachers' academic achievement
     Balcazar Salazar, C. F., Nopo, H., January 2015

13   Can we measure resilience? a proposed method and evidence from countries in the Sahel
     Alfani, F., Dabalen, A. L., Fisker, P., Molini, V., January 2015

14   Vulnerability to malnutrition in the West African Sahel
     Alfani, F., Dabalen, A. L., Fisker, P., Molini, V., January 2015

15   Economic mobility in Europe and Central Asia: exploring patterns and uncovering puzzles
     Cancho, C., Davalos, M. E., Demarchi, G., Meyer, M., Sanchez Paramo, C., January 2015

16   Managing risk with insurance and savings: experimental evidence for male and female farm
     managers in the Sahel
     Delavallade, C., Dizon, F., Hill, R., Petraud, J. P., el., January 2015

17   Gone with the storm: rainfall shocks and household well‐being in Guatemala
     Baez, J. E., Lucchetti, L., Genoni, M. E., Salazar, M., January 2015

18   Handling the weather: insurance, savings, and credit in West Africa
     De Nicola, F., February 2015

19   The distributional impact of fiscal policy in South Africa
     Inchauste Comboni, M. G., Lustig, N., Maboshe, M., Purfield, C., Woolard, I., March 2015

20   Interviewer effects in subjective survey questions: evidence from Timor‐Leste
     Himelein, K., March 2015

21   No condition is permanent: middle class in Nigeria in the last decade
     Corral Rodas, P. A., Molini, V., Oseni, G. O., March 2015

22   An evaluation of the 2014 subsidy reforms in Morocco and a simulation of further reforms
     Verme, P., El Massnaoui, K., March 2015




                           Updated on June 2018 by POV GP KL Team | 2
23   The quest for subsidy reforms in Libya
     Araar, A., Choueiri, N., Verme, P., March 2015

24   The (non‐) effect of violence on education: evidence from the "war on drugs" in Mexico
     Márquez‐Padilla, F., Pérez‐Arce, F., Rodriguez Castelan, C., April 2015

25   “Missing girls” in the south Caucasus countries: trends, possible causes, and policy options
     Das Gupta, M., April 2015

26   Measuring inequality from top to bottom
     Diaz Bazan, T. V., April 2015

27   Are we confusing poverty with preferences?
     Van Den Boom, B., Halsema, A., Molini, V., April 2015

28   Socioeconomic impact of the crisis in north Mali on displaced people (Available in French)
     Etang Ndip, A., Hoogeveen, J. G., Lendorfer, J., June 2015

29   Data deprivation: another deprivation to end
     Serajuddin, U., Uematsu, H., Wieser, C., Yoshida, N., Dabalen, A., April 2015

30   The local socioeconomic effects of gold mining: evidence from Ghana
     Chuhan-Pole, P., Dabalen, A., Kotsadam, A., Sanoh, A., Tolonen, A.K., April 2015

31   Inequality of outcomes and inequality of opportunity in Tanzania
     Belghith, N. B. H., Zeufack, A. G., May 2015

32   How unfair is the inequality of wage earnings in Russia? estimates from panel data
     Tiwari, S., Lara Ibarra, G., Narayan, A., June 2015

33   Fertility transition in Turkey—who is most at risk of deciding against child arrival?
     Greulich, A., Dasre, A., Inan, C., June 2015

34   The socioeconomic impacts of energy reform in Tunisia: a simulation approach
     Cuesta Leiva, J. A., El Lahga, A., Lara Ibarra, G., June 2015

35   Energy subsidies reform in Jordan: welfare implications of different scenarios
     Atamanov, A., Jellema, J. R., Serajuddin, U., June 2015

36   How costly are labor gender gaps? estimates for the Balkans and Turkey
     Cuberes, D., Teignier, M., June 2015

37   Subjective well‐being across the lifespan in Europe and Central Asia
     Bauer, J. M., Munoz Boudet, A. M., Levin, V., Nie, P., Sousa‐Poza, A., July 2015




                          Updated on June 2018 by POV GP KL Team | 3
38   Lower bounds on inequality of opportunity and measurement error
     Balcazar Salazar, C. F., July 2015

39   A decade of declining earnings inequality in the Russian Federation
     Posadas, J., Calvo, P. A., Lopez‐Calva, L.‐F., August 2015

40   Gender gap in pay in the Russian Federation: twenty years later, still a concern
     Atencio, A., Posadas, J., August 2015

41   Job opportunities along the rural‐urban gradation and female labor force participation in India
     Chatterjee, U., Rama, M. G., Murgai, R., September 2015

42   Multidimensional poverty in Ethiopia: changes in overlapping deprivations
     Yigezu, B., Ambel, A. A., Mehta, P. A., September 2015

43   Are public libraries improving quality of education? when the provision of public goods is not enough
     Rodriguez Lesmes, P. A., Valderrama Gonzalez, D., Trujillo, J. D., September 2015

44   Understanding poverty reduction in Sri Lanka: evidence from 2002 to 2012/13
     Inchauste Comboni, M. G., Ceriani, L., Olivieri, S. D., October 2015

45   A global count of the extreme poor in 2012: data issues, methodology and initial results
     Ferreira, F.H.G., Chen, S., Dabalen, A. L., Dikhanov, Y. M., Hamadeh, N., Jolliffe, D. M., Narayan, A.,
     Prydz, E. B., Revenga, A. L., Sangraula, P., Serajuddin, U., Yoshida, N., October 2015

46   Exploring the sources of downward bias in measuring inequality of opportunity
     Lara Ibarra, G., Martinez Cruz, A. L., October 2015

47   Women’s police stations and domestic violence: evidence from Brazil
     Perova, E., Reynolds, S., November 2015

48   From demographic dividend to demographic burden? regional trends of population aging in Russia
     Matytsin, M., Moorty, L. M., Richter, K., November 2015

49   Hub‐periphery development pattern and inclusive growth: case study of Guangdong province
     Luo, X., Zhu, N., December 2015

50   Unpacking the MPI: a decomposition approach of changes in multidimensional poverty headcounts
     Rodriguez Castelan, C., Trujillo, J. D., Pérez Pérez, J. E., Valderrama, D., December 2015

51   The poverty effects of market concentration
     Rodriguez Castelan, C., December 2015

52   Can a small social pension promote labor force participation? evidence from the Colombia Mayor
     program
     Pfutze, T., Rodriguez Castelan, C., December 2015


                          Updated on June 2018 by POV GP KL Team | 4
53    Why so gloomy? perceptions of economic mobility in Europe and Central Asia
      Davalos, M. E., Cancho, C. A., Sanchez, C., December 2015

54    Tenure security premium in informal housing markets: a spatial hedonic analysis
      Nakamura, S., December 2015

55    Earnings premiums and penalties for self‐employment and informal employees around the world
      Newhouse, D. L., Mossaad, N., Gindling, T. H., January 2016

56    How equitable is access to finance in turkey? evidence from the latest global FINDEX
      Yang, J., Azevedo, J. P. W. D., Inan, O. K., January 2016

57    What are the impacts of Syrian refugees on host community welfare in Turkey? a subnational
      poverty analysis
      Yang, J., Azevedo, J. P. W. D., Inan, O. K., January 2016

58    Declining wages for college‐educated workers in Mexico: are younger or older cohorts hurt the
      most?
      Lustig, N., Campos‐Vazquez, R. M., Lopez‐Calva, L.‐F., January 2016

59    Sifting through the Data: labor markets in Haiti through a turbulent decade (2001‐2012)
      Rodella, A.‐S., Scot, T., February 2016

60    Drought and retribution: evidence from a large‐scale rainfall‐indexed insurance program in Mexico
      Fuchs Tarlovsky, Alan., Wolff, H., February 2016

61    Prices and welfare
      Verme, P., Araar, A., February 2016

62    Losing the gains of the past: the welfare and distributional impacts of the twin crises in Iraq 2014
      Olivieri, S. D., Krishnan, N., February 2016

63    Growth, urbanization, and poverty reduction in India
      Ravallion, M., Murgai, R., Datt, G., February 2016

64    Why did poverty decline in India? a nonparametric decomposition exercise
      Murgai, R., Balcazar Salazar, C. F., Narayan, A., Desai, S., March 2016

65    Robustness of shared prosperity estimates: how different methodological choices matter
      Uematsu, H., Atamanov, A., Dewina, R., Nguyen, M. C., Azevedo, J. P. W. D., Wieser, C., Yoshida, N.,
      March 2016

66    Is random forest a superior methodology for predicting poverty? an empirical assessment
      Stender, N., Pave Sohnesen, T., March 2016

67    When do gender wage differences emerge? a study of Azerbaijan's labor market
     Tiongson, E. H. R., Pastore, F., Sattar, S., March 2016


                           Updated on June 2018 by POV GP KL Team | 5
68   Second‐stage sampling for conflict areas: methods and implications
     Eckman, S., Murray, S., Himelein, K., Bauer, J., March 2016

69   Measuring poverty in Latin America and the Caribbean: methodological considerations when
     estimating an empirical regional poverty line
     Gasparini, L. C., April 2016

70   Looking back on two decades of poverty and well‐being in India
     Murgai, R., Narayan, A., April 2016

71   Is living in African cities expensive?
     Yamanaka, M., Dikhanov, Y. M., Rissanen, M. O., Harati, R., Nakamura, S., Lall, S. V., Hamadeh, N., Vigil
     Oliver, W., April 2016

72   Ageing and family solidarity in Europe: patterns and driving factors of intergenerational support
     Albertini, M., Sinha, N., May 2016

73   Crime and persistent punishment: a long‐run perspective on the links between violence and chronic
     poverty in Mexico
     Rodriguez Castelan, C., Martinez‐Cruz, A. L., Lucchetti, L. R., Valderrama Gonzalez, D., Castaneda
     Aguilar, R. A., Garriga, S., June 2016

74   Should I stay or should I go? internal migration and household welfare in Ghana
     Molini, V., Pavelesku, D., Ranzani, M., July 2016

75   Subsidy reforms in the Middle East and North Africa Region: a review
     Verme, P., July 2016

76   A comparative analysis of subsidy reforms in the Middle East and North Africa Region
     Verme, P., Araar, A., July 2016

77   All that glitters is not gold: polarization amid poverty reduction in Ghana
     Clementi, F., Molini, V., Schettino, F., July 2016

78   Vulnerability to Poverty in rural Malawi
     Mccarthy, N., Brubaker, J., De La Fuente, A., July 2016

79   The distributional impact of taxes and transfers in Poland
     Goraus Tanska, K. M., Inchauste Comboni, M. G., August 2016

80   Estimating poverty rates in target populations: an assessment of the simple poverty scorecard and
     alternative approaches
     Vinha, K., Rebolledo Dellepiane, M. A., Skoufias, E., Diamond, A., Gill, M., Xu, Y., August 2016




                          Updated on June 2018 by POV GP KL Team | 6
81   Synergies in child nutrition: interactions of food security, health and environment, and child care
     Skoufias, E., August 2016

82   Understanding the dynamics of labor income inequality in Latin America
     Rodriguez Castelan, C., Lustig, N., Valderrama, D., Lopez‐Calva, L.‐F., August 2016

83   Mobility and pathways to the middle class in Nepal
     Tiwari, S., Balcazar Salazar, C. F., Shidiq, A. R., September 2016

84   Constructing robust poverty trends in the Islamic Republic of Iran: 2008‐14
     Salehi Isfahani, D., Atamanov, A., Mostafavi, M.‐H., Vishwanath, T., September 2016

85   Who are the poor in the developing world?
     Newhouse, D. L., Uematsu, H., Doan, D. T. T., Nguyen, M. C., Azevedo, J. P. W. D., Castaneda Aguilar, R.
     A., October 2016

86   New estimates of extreme poverty for children
     Newhouse, D. L., Suarez Becerra, P., Evans, M. C., October 2016

87   Shedding light: understanding energy efficiency and electricity reliability
     Carranza, E., Meeks, R., November 2016

88   Heterogeneous returns to income diversification: evidence from Nigeria
     Siwatu, G. O., Corral Rodas, P. A., Bertoni, E., Molini, V., November 2016

89   How liberal is Nepal's liberal grade promotion policy?
     Sharma, D., November 2016

90   Pro-growth equity: a policy framework for the twin goals
     Lopez-Calva, L. F., Rodriguez Castelan, C., November 2016

91   CPI bias and its implications for poverty reduction in Africa
     Dabalen, A. L., Gaddis, I., Nguyen, N. T. V., December 2016

92   Building an ex ante simulation model for estimating the capacity impact, benefit incidence, and cost
     effectiveness of child care subsidies: an application using provider‐level data from Turkey
     Aran, M. A., Munoz Boudet, A., Aktakke, N., December 2016

93   Vulnerability to drought and food price shocks: evidence from Ethiopia
     Porter, C., Hill, R., December 2016

94   Job quality and poverty in Latin America
     Rodriguez Castelan, C., Mann, C. R., Brummund, P., December 2016

95   With a little help: shocks, agricultural income, and welfare in Uganda
     Mejia‐Mantilla, C., Hill, R., January 2017


                          Updated on June 2018 by POV GP KL Team | 7
96   The impact of fiscal policy on inequality and poverty in Chile
     Martinez Aguilar, S. N., Fuchs Tarlovsky, A., Ortiz‐Juarez, E., Del Carmen Hasbun, G. E., January 2017

97   Conditionality as targeting? participation and distributional effects of conditional cash transfers
     Rodriguez Castelan, C., January 2017

98   How is the slowdown affecting households in Latin America and the Caribbean?
     Reyes, G. J., Calvo‐Gonzalez, O., Sousa, L. D. C., Castaneda Aguilar, R. A., Farfan Bertran, M. G., January
     2017

99   Are tobacco taxes really regressive? evidence from Chile
     Fuchs Tarlovsky, A., Meneses, F. J., March 2017

100 Design of a multi‐stage stratified sample for poverty and welfare monitoring with multiple
    objectives: a
    Bangladesh case study
    Yanez Pagans, M., Roy, D., Yoshida, N., Ahmed, F., March 2017

101 For India's rural poor, growing towns matter more than growing cities
    Murgai, R., Ravallion, M., Datt, G., Gibson, J., March 2017

102 Leaving, staying, or coming back? migration decisions during the northern Mali conflict
    Hoogeveen, J. G., Sansone, D., Rossi, M., March 2017

103 Arithmetics and Politics of Domestic Resource Mobilization
    Bolch, K. B., Ceriani, L., Lopez‐Calva, L.‐F., April 2017

104 Can Public Works Programs Reduce Youth Crime? Evidence from Papua New Guinea’s Urban Youth
    Employment Project
    Oleksiy I., Darian N., David N., Sonya S., April 2017

105 Is Poverty in Africa Mostly Chronic or Transient? Evidence from Synthetic Panel Data
    Dang, H.‐A. H., Dabalen, A. L., April 2017

106 To Sew or Not to Sew? Assessing the Welfare Effects of the Garment Industry in Cambodia
    Mejía‐Mantilla, C., Woldemichael, M. T., May 2017

107 Perceptions of distributive justice in Latin America during a period of falling inequality
    Reyes, G. J., Gasparini, L. C., May 2017

108 How do women fare in rural non‐farm economy?
    Fuje, H. N., May 2017

109 Rural Non‐Farm Employment and Household Welfare: Evidence from Malawi
    Adjognon, G. S., Liverpool‐Tasie, S. L., De La Fuente, A., Benfica, R. M., May 2017




                          Updated on June 2018 by POV GP KL Team | 8
110 Multidimensional Poverty in the Philippines, 2004‐13: Do Choices for Weighting, Identification and
    Aggregation Matter?
    Datt, G., June 2017

111 But … what is the poverty rate today? testing poverty nowcasting methods in Latin America and the
    Caribbean
    Caruso, G. D., Lucchetti, L. R., Malasquez, E., Scot, T., Castaneda, R. A., June 2017

112 Estimating the Welfare Costs of Reforming the Iraq Public Distribution System: A Mixed Demand
    Approach
    Krishnan, N., Olivieri, S., Ramadan, R., June 2017

113 Beyond Income Poverty: Nonmonetary Dimensions of Poverty in Uganda
    Etang Ndip, A., Tsimpo, C., June 2017

114 Education and Health Services in Uganda: Quality of Inputs, User Satisfaction, and Community
    Welfare Levels
    Tsimpo Nkengne, C., Etang Ndip, A., Wodon, Q. T., June 2017

115 Rental Regulation and Its Consequences on Measures of Well‐Being in the Arab Republic of Egypt
    Lara Ibarra, G., Mendiratta, V., Vishwanath, T., July 2017

116 The Poverty Implications of Alternative Tax Reforms: Results from a Numerical Application to
    Pakistan
    Feltenstein, A., Mejia‐Mantilla, C., Newhouse, D. L., Sedrakyan, G., August 2017

117 Tracing Back the Weather Origins of Human Welfare: Evidence from Mozambique?
    Baez Ramirez, J. E., Caruso, G. D., Niu, C., August 2017

118 Many Faces of Deprivation: A multidimensional approach to poverty in Armenia
    Martirosova, D., Inan, O. K., Meyer, M., Sinha, N., August 2017

119 Natural Disaster Damage Indices Based on Remotely Sensed Data: An Application to Indonesia
    Skoufias, E., Strobl, E., Tveit, T. B., September 2017

120 The Distributional Impact of Taxes and Social Spending in Croatia
    Inchauste Comboni, M. G., Rubil, I., October 2017

121 Regressive or Progressive? The Effect of Tobacco Taxes in Ukraine
    Fuchs, A., Meneses, F. September 2017

122 Fiscal Incidence in Belarus: A Commitment to Equity Analysis
    Bornukova, K., Shymanovich, G., Chubrik, A., October 2017




                        Updated on June 2018 by POV GP KL Team | 9
123 Who escaped poverty and who was left behind? a non‐parametric approach to explore welfare
    dynamics using cross‐sections
    Lucchetti, L. R., October 2017

124 Learning the impact of financial education when take-up is low
    Lara Ibarra, G., Mckenzie, D. J., Ruiz Ortega, C., November 2017

125 Putting Your Money Where Your Mouth Is Geographic Targeting of World Bank Projects
    to the Bottom 40 Percent
    Öhler, H., Negre, M., Smets, L., Massari, R., Bogetić, Z., November 2017

126 The impact of fiscal policy on inequality and poverty in Zambia
    De La Fuente, A., Rosales, M., Jellema, J. R., November 2017

127 The Whys of Social Exclusion: Insights from Behavioral Economics
    Hoff, K., Walsh, J. S., December 2017

128 Mission and the bottom line: performance incentives in a multi-goal organization
    Gine, X., Mansuri, G., Shrestha, S. A., December 2017

129 Mobile Infrastructure and Rural Business Enterprises Evidence from Sim Registration Mandate in
    Niger
    Annan, F., Sanoh, A., December 2017

130 Poverty from Space: Using High-Resolution Satellite Imagery for estimating Economic Well-Being
    Engstrom, R., Hersh, J., Newhouse, D., December 2017

131 Winners Never Quit, Quitters Never Grow: Using Text Mining to measure Policy Volatility and its Link
    with Long-Term Growth in Latin America
    Calvo-Gonzalez, O., Eizmendi, A., Reyes, G., January 2018

132 The Changing Way Governments talk about Poverty and Inequality: Evidence from two Centuries of
    Latin American Presidential Speeches
    Calvo-Gonzalez, O., Eizmendi, A., Reyes, G., January 2018

133 Tobacco Price Elasticity and Tax Progressivity In Moldova
    Fuchs, A., Meneses, F., February 2018

134 Informal Sector Heterogeneity and Income Inequality: Evidence from the Democratic Republic of
    Congo
    Adoho, F., Doumbia, D., February 2018

135 South Caucasus in Motion: Economic and Social Mobility in Armenia, Azerbaijan and Georgia
    Tiwari, S., Cancho, C., Meyer, M., February 2018




                         Updated on June 2018 by POV GP KL Team | 10
136 Human Capital Outflows: Selection into Migration from the Northern Triangle
    Del Carmen, G., Sousa, L., February 2018

137 Urban Transport Infrastructure and Household Welfare: Evidence from Colombia
    Pfutze, T., Rodriguez-Castelan, C., Valderrama-Gonzalez, D., February 2018

138 Hit and Run? Income Shocks and School Dropouts in Latin America
    Cerutti, P., Crivellaro, E., Reyes, G., Sousa, L., February 2018

139 Decentralization and Redistribution Irrigation Reform in Pakistan’s Indus Basin
    Jacoby, H.G., Mansuri, G., Fatima, F., February 2018

140 Governing the Commons? Water and Power in Pakistan’s Indus Basin
    Jacoby, H.G., Mansuri, G., February 2018

141 The State of Jobs in Post-Conflict Areas of Sri Lanka
    Newhouse, D., Silwal, A. R., February 2018

142 “If it’s already tough, imagine for me…” A Qualitative Perspective on Youth Out of School
    and Out of Work in Brazil
    Machado, A.L., Muller, M., March 2018

143 The reallocation of district-level spending and natural disasters: evidence from Indonesia
    Skoufias, E., Strobl, E., Tveit, T. B., March 2018

144 Gender Differences in Poverty and Household Composition through the Life-cycle A Global
    Perspective
    Munoz, A. M., Buitrago, P., Leroy de la Briere, B., Newhouse, D., Rubiano, E., Scott, K., Suarez-Becerra,
    P., March 2018

145 Analysis of the Mismatch between Tanzania Household Budget Survey and National Panel Survey
    Data in Poverty & Inequality Levels and Trends
    Fuchs, A., Del Carmen, G., Kechia Mukong, A., March 2018

146 Long-Run Impacts of Increasing Tobacco Taxes: Evidence from South Africa
    Hassine Belghith, N.B., Lopera, M. A., Etang Ndip, A., Karamba, W., March 2018

147 The Distributional Impact of the Fiscal System in Albania
    Davalos, M., Robayo-Abril, M., Shehaj, E., Gjika, A., March 2018

148 Analysis Growth, Safety Nets and Poverty: Assessing Progress in Ethiopia from 1996 to 2011
    Vargas Hill, R., Tsehaye, E., March 2018

149 The Economics of the Gender Wage Gap in Armenia
    Rodriguez-Chamussy, L., Sinha, N., Atencio, A., April 2018



                          Updated on June 2018 by POV GP KL Team | 11
150 Do Demographics Matter for African Child Poverty?
    Batana, Y., Cockburn, J., May 2018

151 Household Expenditure and Poverty Measures in 60 Minutes: A New Approach with Results from
    Mogadishu
    Pape, U., Mistiaen, J., May 2018

152 Inequality of Opportunity in South Caucasus
    Fuchs, A., Tiwari, S., Rizal Shidiq, A., May 2018

153 Welfare Dynamics in Colombia: Results from Synthetic Panels
    Balcazar, C.F., Dang, H-A., Malasquez, E., Olivieri, S., Pico, J., May 2018

154 Social Protection in Niger: What Have Shocks and Time Got to Say?
    Annan, F., Sanoh, A., May 2018

155 Quantifying the impacts of capturing territory from the government in the Republic of Yemen
    Tandon, S., May 2018

156 The Road to Recovery: The Role of Poverty in the Exposure, Vulnerability and Resilience to Floods in
    Accra
    Erman, A., Motte, E., Goyal, R., Asare, A., Takamatsu, S., Chen, X., Malgioglio, S., Skinner, A., Yoshida,
    N., Hallegatte, S., June 2018

157 Small Area Estimation of Poverty under Structural Change
    Lange, S., Pape, U., Pütz, P., June 2018




                        For the latest and sortable directory,
        available on the Poverty & Equity GP intranet site. http://POVERTY

                       WWW.WORLDBANK.ORG/POVERTY



                           Updated on June 2018 by POV GP KL Team | 12