Identifying and Communicating Pro-Poor Policy
Messages to Policy Audiences: the Case of Chronic
Poverty in Uganda.
Kate Bird
Overseas Development Institute
Email: k.bird@odi.org.uk
Acknowledgements
The research on which this paper is based was conducted through the
Chronic Poverty Research Centre ( www.chronicpoverty.org ) with financial
support from DFID, research grant no. R7847. The field-based research in
3 "LADDER study sites" depended on access to the LADDER household
survey, survey instruments and sampling frame, village studies and
various un-published qualitative research outputs. Thanks are due to
Frank Ellis (team leader, ODG, University of East Anglia) and others in the
LADDER team for their openness in sharing their household survey
dataset and qualitative data with CPRC researchers. It was first written up
as Bird and Shinyekwa, 2003a.
Q-squared • Centre For International Studies • University Of Toronto
1 Devonshire Place, Toronto ON M5S 3K7 Canada
t: 416-885-7721 • f: 416-652-1678 • e: info@q-squared.ca
Q-Squared Working Paper No. 43
November 2007
1. Introduction
This paper discusses the processes surrounding the dissemination of findings from
Q2 work exploring the drivers and maintainers of chronic poverty in Uganda to a
range of policy audiences at the national and international level. It briefly introduces
the research design and research methods used and the findings produced before
discussing how an understanding of the challenges of getting certain issues on to
policy agendas helped in the identification and targeting of different actors in the
policy process with 'communication materials' of different types.
2. The research
The research presented here was undertaken in 2002 under the Chronic Poverty
Research Centre. The author selected three of the nine sites in rural Uganda where
an in-depth household survey of 315 households had been recently undertaken (in
2001, by LADDER1). The study sites were chosen on the basis of an initial analysis
of the LADDER dataset and following a review of the qualitative village reports
produced in 2002 by the LADDER team as a result of qualitative exercises with 2343
people in the nine study sites.
The nine LADDER study sites were selected purposively by the LADDER team to
represent all the main agro-climatic zones and natural resource based livelihoods in
Uganda. A wealth-ranking exercise was undertaken in each of the nine sites in order
to generate a sampling frame. These commonly placed households into five or six
strata which were combined by the LADDER research team to form three groups:
poor, middle and non-poor. Thirty-five households were selected2 from each village
for the in-depth household survey, ten households from each of the middle and nonpoor
groups and fifteen from the poor group.3
The LADDER sites were not selected based on their poverty characteristics and
cover an arc in the central west of the country to the central east, not touching on any
of the currently poorest or conflict affected areas. The selection was effective for the
original purposes of the LADDER research but limited our ability to understand the
dynamics of chronic poverty. A more comprehensive understanding was generated
by complementing the research in the 'LADDER villages' with additional field work in
remote Kisoro, SW Uganda.
1
The LADDER project, funded by DFID, has worked in Uganda, Tanzania, Kenya and
Malawi. Godfrey Bahiigwa, from EPRC, led the Uganda component of this work. For more
information about LADDER, see www.uea.ac.uk/dev/odg/ladder . This research depended on
access to the LADDER household survey, village studies, various un-published qualitative
research outputs. Thanks are due to Frank Ellis (team leader, ODG, University of East Anglia)
and others in the LADDER team for sharing their dataset and qualitative data with CPRC
researchers.
2
In the fishing villages stratified sampling ensured that roughly half of the households were
fisherfolk (e.g. 17 out of 35 in Kiribairya). This was to guarantee that the survey provided
adequate information on fish-based livelihoods. Unfortunately this makes other issues difficult
to explore with accuracy (e.g. mean village income figures, proportion of households involved
in fish-based livelihoods, proportion of landless households, proportion of households in
Kiribairya displaced by Teso ethnic cleansing who are owners of neither boats nor land). In
addition, researchers selected alternative households when it appeared that the sampling
process was generating an over-large number of households headed by a widow.
3
Analysis of the results from this survey shows that the wealth groupings correspond well to
household asset and income data although the picture is slightly less clear with regard to per
capita income figures.
1
The CPRC research team ch
e severity of poverty their residents were found to experience (Buwopuwa
ose two of their three sites from these nine, because of
4
in
Figure 1: Location of LADDER Study Districts5.
urvey provided an understanding of assets, livelihood
ortfolios, basic demographics, key shocks and responses to shocks and patterns of
change over the five years preceding the LADDER (round one) household survey6. A
rviews
tial
An
in-depth life histories (oral testimonies) were collected in each of the
e
th
Mbale District, and Kiribairya in Kamuli District), and the third, Kalangaalo (Mubende
District), because of its location in the Luwero Triangle and the depth of its
experience of conflict during the 1981-85 NRM (National Resistance Movement)
bush war. So, while these villages are not representative of Uganda as a whole they
can be said to be typical of the large numbers of poorer villages in rural Uganda
(Figure 1).
UGANDA
CONGO
R
i
v
e
r
N
i
l
Lake Kyoga
e
Lake Victoria
RWANDA
TANZANIA
KENYA
MUBENDE
KAMULI
MBALE
Kampala
Source: LADDER Project, 2002.
We assumed that the majority of the chronically poor would be found amongst the
most severely poor, but wanted to test whether this was true, as well as identifying
the key drivers, maintainers and interrupters of chronic poverty.
Analysis of the household s
p
mixed tool box of life histories, focus group discussions and key informant inte
was developed in order to generate evidence and analysis to help explain differen
poverty trajectories between study sites, cohorts and individual households and to
identify a range of explanatory variables for differential well-being outcomes.
Respondents for the 28 in-depth life history interviews were identified purposively.
average of 9
three villages. As a first step, the research team obtained full village listings from th
previous LADDER participative wealth ranking exercises in the three study sites. This
4
Buwopuwa was selected because the sampling frame for the poorest village in the M
sample, Bukhasusa, was no longer available. We re-analysed the dataset to develop
greater understanding of the three study villages, and the authors have undertaken
periods of field work in the selected study villages.
bale
a
two
5
For more information about LADDER, see www.uea.ac.uk/dev/odg/ladder
6
The survey has now been panelised and in the future it will be possible to provide a full Q2
analysis of the panel alongside life histories.
2
was cross-checked with the household survey data to enable the stratified random
selection of households. Households were selected, some from each wealth strata,
with an over-representation of poor and very poor households. This generated a
good mix of male and female household heads for interview. Subsequently, a
number of follow-on life history interviews were conducted in each site with
individuals selected purposively by their ascribed status (e.g. migrants, the destitute,
the disabled, member of a youth-headed household, younger widow, head of a 'gap-
generation
7
household').
The life histori
ich traced an
dividual's life through f
key
e-changing
ed to identify some of the
dvantages which protected them from falling into poverty, and to show in what way
the experiences of the
tly poor differed from those of the
transitorily poor. By talking to a pers
eir life, rather than taking a thematic
approach, we hoped t
in-
ory interviews
were also used to identify theme
p
discussions
Initial examination of the LADDER household survey, results from focus group
discussions (with grou
wealth and livelihood
grouping) and key infor
r exploration through life
histories an
rify the likely representativeness of
findings from individual life
r analysis of the LADDER
household survey was u
ews and to
triangulate findings, whe
, the results from analysis of the Uganda national household survey and
e first and second Participatory Poverty Assessments were used to contextualise
he findings from bo
DDER household survey and the
the richest and best
e other
tudy sites in income terms but that they had a higher proportion of female headed
es were collected during semi-structured interviews wh
rom their earliest childhood to the present day, including
in
lif
a
events. By talking also to the non-poor we hop
severely and persisten
on about th
o identify path-determination in individuals' lives and to p
point key moments of choice - or the absence of choice. The life-hist
s for further investigation through focus-grou
and key-informant interviews.
ps differentiated by age, gender,
mant interviews identified issues fo
d enabled the research team to ve
histories. Subsequently, furthe
sed to explore issues raised in life history intervi
re possible.
Furthermore
th
t
q
th the analysis of the LA
ualitative research at the three sites.
Tables 1 and 2 provide a quick overview of the three sites selected for in-depth
study. They range in wealth and remoteness, with Kalangaalo
connected and Kiribairya the poorest and least well connected. More detailed
descriptions can be found in Bird and Shinyekwa (2003) and LADDER (2001a, b, c).
2.1. Selective description of research findings8
Analysis of household data from the nine LADDER sites provided important
contextual information about livelihoods and asset portfolios and poverty. It showed
that the villages of Buwopuwa and Kiribairya were not just poorer than th
s
households and had a low proportion of tin rooved dwellings. However, mean land
and livestock ownership was higher in Buwopuwa than in many of the other study
sites, suggesting that these productive assets were not necessarily protective of
downward declines into poverty (see Table 1).
7
Older person/persons (grandparent(s)? ) in a household with a child/ children. No household
members aged 14-45.
8
This section of the paper draws strongly on Bird, K., & Shinyekwa, I. (2005) 'Even the 'rich'
are vulnerable: multiple shocks and downward mobility in rural Uganda.' Development Policy
Review, 2005. 23 (1):55-85.
3
Household wealth ranks were correlated with a number of variables (education of
household head, household education, land ownership, livestock ownership [in cattle
equivalent units], household assets, ownership of tools, migration, the household's
dependency ratio, receipt of transfers, and the livelihood diversity index). The
strength of correlation varied amongst the villages and the only variables important in
all three focal villages were land ownership and household assets. Education and the
ownership of livestock and tools were important in some but not all of the villages
(see Table 3). Looking at the dataset for all nine LADDER villages we found a
ignificant difference (at the .01 level) between poor and non-poor households in
marginal, leading to un- and under-
rlocked, and the constraints which prevented
s
terms of the ownership of land, livestock, tools and household assets.9 There was a
significant difference also (at the .05 level) in terms of the education level of the
household head and reliance on own production for subsistence.10 Which way
causality runs for each of these variables is difficult to determine.
Land ownership in the study villages is closely related to wealth grouping. The poor
have marginal holdings, and in Kiribairya only the non-poor own any land at all. Local
inheritance practice (whereby all sons receive a portion of their father's land on
marriage, rather than primogeniture where only the oldest son inherits) is resulting in
land fragmentation. Many holdings are
employment - particularly amongst the youth; reduced levels of aggregate and
individual well-being and increased differentiation driven by the distress sale of land
by households attempting to meet contingencies or clear debts.
This evidence was supplemented by analysis from participatory exercises (see Table
2, below) and the wealth-ranking information produced for the LADDER sampling
frame indicates that between 26% and 62% of households are considered poor in
their own communities.11
The life histories collected by the CPRC research team were found to provide a
strong starting point for understanding the trajectories into and out of poverty that
individuals and households follow. They enabled the identification of the most
common covariant and idiosyncratic shocks which triggered a decline into poverty
drivers), the way in which these inte
(
accumulation, investment and movement out of poverty (maintainers,12 illustrated in
Figure 2).
9
Poverty in this case is determined by whether per capita income is above or below the
poverty line for rural Uganda. Per capita income has been arrived at by calculating total
household income from all sources (e.g. crop and livestock sales, the value of agricultural
produce for home consumption, land rental, fisheries, casual labour, salaries and wages and
remittances) divided by the adult equivalent units per household.
10
Using a version of Dunn's test given in Siegel and Castellan (1988: 213-4).
11
Kalangaalo, 61%; Kansambya, 42%; Kabbo, 28%; Bunabuso, 53%; Buwopuwa, 31%,
iribairya, 26%; Iyingo, 61%; and Kinamwanga, 62%.
12
In this first round of collecting life histories we interviewed only household heads who had
been interviewed during the LADDER survey. This enabled us to move straight into the life-
history interview, without having to spend time collecting basic household asset, income and
e method that we hope to use) and other issues.
K
livelihood data. Due to the large number of female-headed households this did not lead to a
lack of women's stories, but may have biased our findings in other ways (e.g. inadequate
attention to 'youth', women within male-headed households, the situation of co-wives, the
chronic poverty of children, and so on). In subsequent research we plan to systematically
interview other household members to explore intra-household differentiation (see Bolt and
Bird, 2003 for an outline of th
4
Table 1: Descriptive Statistics of the LADDER Villages.
LADDER
Village
Per capita
Income
(US$)13
Mean
hh land
holding
(ha.)
Livestock
Ownership
in CEUs14
%
female
headed
hh
Building
materials of
houses (%)
Mean s.d.
Means.d.
Brick/
concrete
walls
Tin
roo
f
Mbale
Bukahasausa 191 193 0.54 0.54 0.90 17 0 60
Buwopuwa 180 188 2.17 1.72 2.24 23 46
6
Bunabuso 583 645 1.92 1.66 1.68 14 9 89
Kamuli
Iyingo 512 894 0.71 1.56 4.76 17 23 43
Kinamwanga 246 704 1.12 5.37 11.1911 49 63
Kiribairya 111 92 0.18 0.30 0.45 26 0 0
Mubende
Kabbo 229 164 1.54 3.77 10.3717 9 57
Kansambya 155 125 2.47 1.09 1.99 3 9 74
Kalangaalo 254 188 2.26 1.64 2.32 20 40 94
Evidence from the study sites shows that decline in well-being is associated with a
web of meso-level constraints and shocks which commonly combined negatively with
ousehold-level shocks
h
and socio-cultural or socio-psychological factors. Composite
shocks and loss of assets propelled a number of previously non-poor households into
severe and long-term poverty. Chronically poor households seldom faced a single
problem or constraint, and found few opportunities for accumulation and escape.
Those who were able to reduce the intensity of their poverty generally managed to do
so as a result of several serendipitous events or factors combining.
13
Ugandan Shilling figures converted to US dollars using the average rate during the first
quarter of 2001 (when the survey was conducted) of USh. 100,000 = US$57.85
14
CEU = cattle equivalent unit:Pigs=0.14; Goats=0.12; Sheep=0.10; Turkeys=0.04;
Chickens=0.02; Other=(given price/5% trimmed mean price for cattle)
5
Table 2: Key characteristics of the study sites
Kiribairya
Buwopuwa
Kalangaalo
District
Kamuli
Mbale
Mubende
Population
74 households
c. 520 people
204 households
c.1080 people
237
c. 1
hou
600
seh
peo
old
ple
s
Ethnic groups
Bakenya, Basoga, Banyoro, Iteso, Jaluo
arwanda, Bakiga, Bany
Bashoga, Bagishu, Basam
Bagishu
Bagand
Burndi,
a, B
Bot
any
oro,
oro,
ya
Physi
of are
cal de
a
scription
Lake side, drought affected
Crop land
cted by several streams
Flat plateau, moist
, bise
Accessibility
• Most remote of the three sites - very poor
road access, good access via Lake Kyoga
• 1½ - 2½ hour from Kamuli town (small town
poorly developed market and services)
o
ial
the th
it
,
• Good road access
• 1 hour from Mbale t
centre)
• Near Kenyan border
wn (substant regional
• Least remote
acc
of
ree ses - excellent
road ess
Level
provi
of ser
sion
vice
Highly inadequate
Go
Adequate
od
Access to markets
Poor
Excellent
Average
Main
activi
livelih
ties
ood
Fishing & fish processing, casual labour, some
agriculture, small numbers of livestock
, matooke, coffee, some
ri
Annual crop production, coff
or
ent, alcohol
a, brick makin
Annual crop production
livestock, boda boda, b
ck making
ee, livestock,
production and
g, taxis, and
fmal e
e, bo
constructio
mp
da-
loym
bod
n
sal
• 1986 - arrival of IDPs
• 1995 - borehole
• 2000 - flood swept away half of village
• 2002 - storm demolished primary school,
killing 2 children
• 1966 Kabaka exiled15 late 1960s -1970s
ngs
M bush
a
• 1986 peace (after this the lo
peop
es
offee wilt
beralisation
• 1994 Kabaka allowed to return
• 1990s - boreholes & clin
16
destroyed crops
• 2002 closure of local police
ime
Key events
2000 - borehole
-
wide
198
spr
1-85
ead
NR
conflict affected
killi
•
war - rea seriously
cal trading
ted)
cent
mid
199
re g
198
rew
0s c
0 market li
as
le inv
•
•
ic built in village
post - cr
• 199
199
6 - U
7 rains
PE
•
increased
Source: Life history and key informant interviews, August - November 2002.
15
Bugandan king
16
Universal Primary Education
6
Table 3: Correlating wealth rank groups with selected assets/ h'h characteristics.
Wealth ranking group with
Village name
owned
)
Livestock Tools
old
assets
Education
of HH Head
Hous
educ
Migrants Dependenc
y ratio
Spearman'
-.705
-.749** -.425*
tl
.000 .000 .000 .
KiribairyaSpearm
61
50*
.067 -.047
Sig
.130
063 .007 .701 .788
ear
s -.519**-.503** -. 311
-.571** .098 .1
.211 -.172
S
tailed)
.069
.5
Area
(ha.
Househ
ehold
ation
Buwopuwa
s -.776**-.756**
**
-.303 -.027 -.011
Sig. (2-.000
ai ed)
an's -.510**
-.255
-.2
011 .077 .878 .952
-.485** -.332 -.4*
. (2-.002 .140
tailed)
.003 .
KalangaaloSpman'
15
ig. (2-.001 .002
.000 . 581 12 .224 .323
** Correlation is sig
vel (2-tailed).
* Correlation is sig
N = 35
nificant at the . 01 le
nificant at the .05 level (2-tailed).
7
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative
Methods for Poverty Analysis in Decision-Making
2.1.1. Poverty trajectories
Evidence from focus group discussions and life histories in the three study sites
p
d u
mp
sig
poverty trajectories. Not surprisingly, when
t
bo
liv
n the individual and family
e
w
d
im
v
nd declines in well-being, and a
supplimentary study would be necessary to explore the importance of macroeconomic,
institutional and political economy factors.
Wound b
a
r family provides an individual with a number
o
sitive
utu
fo
vantages: good diet; access to education;
access to h
higher status; a network of friends and patrons in and outside
the village; the ability to travel outside the village, and therefore exposure to ideas;
more land or livestock to inherit; and more likelihood of marrying well. However,
t
dvantages ca
at an individual and their
h
old
bl
xtended periods of poverty.
Analysis of
h
ies and the LADDER dataset shows that most households
are vulnerable to downward movements of well-being, not just those in the so-called
'vulnerable groups'. Analysis of well-being changes between 1996 and 2001 in the
d
sh
holds felt that their situations had worsened over
t
iod
fe
co
dominance of downward movement.
O
8
ls
ived that they had experienced decreased
ddle or non-
duals felt that their well-
een stable. None felt that their well-being had improved.
fr
le
oo
s did not protect individuals and households
cli
ell-
nd
very poor interviewees, 6 described having
life non-poousehoand 6 in middle-ranking households. Interviews
eople illustrated that, in general, they had been much more prosperous
han they were now. Distorted recall may explain this, but the
ples tdivav
ded
de
s
ehold livelihood portfolios, asset holdings
ding
de
ty
holds and livestock numbers) and dietary
mmunity, its
involvement in village life and celebration
the size and composition of the
household now and at various points in the past. The latter included its ability to
a
ct and
rt mb ofendily and/or paid or unpaid servants
a
farm
hi
for
na
judgement to be made about the
r
ve sta
we
of
eho
estion.
rovide
alking a
vents
s w
ut t
hich
ith i
heir
ha
orta
es, the interviewees tended to focus o
led
nt in
to
hts
pro
into
ements a
e f
f po
that
an
ealth care;
mem
d m
ersh
ally
ip of
rein
non
rcing
-poo
ad
hese a
ouseh
nnot necessarily be used to predict th
e to avoid either short or e
istor
will
the life
be a
ataset
hat per
f the 2
ows that 64
. Th
indiv
% of house
histo
interviewed, 25 perce
e li
idua
riesnfirmed the pre
levels of well-being during their lifetime, 12 perceived a decline from mi
poor groups to being in the poorest group, and only 3 indivi
being had b
C
f
s
with older p
during their early lives t
e
i
(
patterns, an
oming
rom de
tarted
om m
nes
in
idd
in w
or n
bein
or h
on-p
g, a
r gr
of the 17
lds
oup
xam
nclu
inclu
tha
deta
hous
d of their household's comparative wealth within the co
in
iled
ing
idua
scri
sign
ls g
ption
and
e to
of
pe,
support their stories were convincing. They
hous
land
ing
s, a
ed
bled
ld
nd
fam
a
in qu
ttra
nd
elati
sup
labo
tus
po
ur. T
and
em
s ric
ll-be
ers
h in
ing
the
mat
the
ext
ion e
hous
8
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative Methods for Poverty Analysis in Decision-Making
Interlocking
problems
Alcohol
abuse
Theft
Conflict
Household
break-up
Do
vi
mestic
olence
Disability
Widowhood/
divorce
Remoteness/
isolation
Old age
Polygamy
Lack of co-operat
within the househ
io
o
n
ld
Fe
asse
w
t
s
Few livelihood
options
Complete distrust
of police & justice
system
Long-term
impact
Untr
tra
e
u
ated
ma
Land
fra
margin
le
g
a
ssness/land
mentation/
l landholdings
Lack
Figure 2: The interlocking drivers of chronic
No relief-to-development
phase for internally
poverty in rural Uganda
displaced people
of credit
Poo
lo
rl
c
y functioning
al markets
Abandonment
Theft
househ
sold a
used by
for alc
mist
prostit
wiv
hou
cons
of
ol
n
h
o
re
u
e
s
um
stored
d grain,
d cash
usbands
hol and
sses/
tes. By
s for
ehold
ption.
Property theft &
asset stripping
Poor access to
health care and
health information
High mortality
& morbidity
Detrimental
to others
Intra-househ
differentiation: re
& work/leisure. F
wives and their ch
non-favour
old
sourc
avo u
ildren
ed
e
red
vs
Beneficial to
some
Co-wives rarely see poly
as beneficial. Some offs
had liked having man
mothers & lots of child
their own age to play w
g
pri
y
re
ith
amy
ng
n
Assets &
well-being
decline for
many
Poor
education
High drop-out
Difficult
accessing in
output mar
ies
put and
kets
Low farm-gate
prices
Ill-health
HIV/AIDS
9
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative
Methods for Poverty Analysis in Decision-Making
10
poor as religious, hardworking (a lazy person is unlikely to survive) and deeply
concerned about the future (World Bank, 1996).
In sum, the research found that while geographical remoteness was a factor driving
poverty in the three study sites, isolation from markets, public services, information
and decision-making was regarded by many poor people as simply providing a
backdrop for the fundamental determinants of their poverty trajectories. Gender,
wealth, locality and life experience influenced people's interpretation of the
fundamental poverty drivers but within this rich picture, it was clear that sequenced
and composite covariate shocks combined with a range of idiosyncratic shocks to
made even the 'rich' within the study sites vulnerable to long term declines in well-
being.
Analysis of policy context, information use and
processes
is section of the paper provides a brief analysis of the political economy and policy
ntext around selected thematic areas; a identification and typology of policy actors,
fferentiated by level (international and national); an examination of their likely use of
idence in decision-making, and their (likely) preferred sources of information.
2.1.2. Drivers and maintainers of poverty
The often sharp declines described above illustrate the long-run negative impacts of
the Amin and Obote I and II regimes, the NRM bush war, acute land fragmentation
and the interlocking of other poverty drivers. The sustained economic growth of the
last decade occurred from a very low base, and there is a widespread perception that
this has yet to bring households and individuals back to the level of basic food
security and well-being that was enjoyed early in the twentieth century. This is
reflected in the stories of the majority of older life-history interviewees (from all wealth
groups) who indicated that life had become harder, over the long term, across a
range of dimensions.
The multidimensional and long-run impact of conflict is suggested by many of the life
histories collected in Kiribairya (Kamuli) and Kalangaalo (Mubende) and by the first
Uganda Participatory Poverty Assessment (UPPAP 1), during which insurgency was
cited as a key driver of poverty in 40% of the UPPAP sites (MFPED, 2000). The
national household survey has consistently shown that the incidence of poverty is
much higher in conflict-affected Northern Uganda (nearly 67% in rural areas) than in
Central Uganda (under 26% in rural areas).
Capabilities and functionings are also of profound importance, and both national
household data and UPPAP 1 indicate poor health, high dependency ratios and
inadequate skills and education as strongly linked with poverty (Okidi, 1999; Okidi
and Mugambe, 2002; Lawson, 2003). Alcohol dependence and problem drinking
were found by UPPAP 1 to be widespread, rating only second to poor health as a
perceived driver of poverty (MFPED, 2000; USAID, 2003). This is supported by our
fieldwork findings.
As an interpretation of focus-group data on cultural or personally-based determinants
of poverty, UPPAP 1 lists 'idleness and laziness' as the seventh most commonly
cited driver of poverty. But whose views are represented here? This conclusion
seems likely to be based on the local elite's derogatory perceptions of the poor (see
also Hossain and Moore, 1999; Woodhouse, 2003). Research in neighbouring Kenya
repeats this picture, with local government officers stereotyping the poor as lazy,
immoral and uncaring about the future, while the local people themselves see the
3.
Th
co
di
ev
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative
Methods for Poverty Analysis in Decision-Making
It then provides an analysis of the forms of 'communication tool' used to contribute in
pecific policy debates in particular arenas (e.g. round table discussions, journal
physical impairments or alcohol dependency taken seriously in
olicy decisions are not strongly evidence-based and even if evidence was available,
g
solidarity and group loyalty. Why is this the case? For example, an individual
aracteristic
strength of
an indi
ation with a particular group can therefore be diminished by
eir identification with and loyalty to another group. Individuals commonly 'own'
s
articles, power point presentations) and the differential weighting given to different
types (or combinations) of qualitative and quantitative evidence in these different
communication tools.
3.1. The political economy barriers to effective policy engagement around
selected issues17
This section explores the barriers to getting the problems facing people with mental
illnesses, mental and
low income developing countries. These issues were identified as important by the
Q2 research described above.
These issues are commonly ignored by political elites despite some of the problems
being severe, widespread and experienced across all wealth strata. This may be
because the scale and severity of the problems facing the poor, the chronically poor
and the marginalised are not adequately recognised. However, it may also be because
p
policy makers would be unlikely to change their stance. A contributing factor may be
the dominant discourse around economic growth, development and poverty reduction
which may have demarcated the 'space' for policy response in such a way that
substantial investments in social policies for such marginalised groups are unlikely.
3.1.1 Problem poorly understood
Evidence poorly communicated to policy makers
Policy makers may lack information about the nature and scale of the problems and,
as a result, they are seen as being small scale and of marginal interest. But the
availability of robust research evidence on a topic does not mean that policy makers
are aware of that evidence. Effective dissemination and communication can ensure
that the facts are known, but even then the issue will have to compete with others
and the recognition an issue receives may be heightened by effective representation.
here are numerous barriers that prevent adequate representation of people facin
T
complex social problems. Individuals may not acknowledge common interest, may
fail to identify common characteristics and may have interests which would clash with
other groups members. The costs of mobilisation can exceed benefits in countries
where potential members form a low proportion of the overall population, are widely
spread, have powerful competing group identities, are poor and do not have access to
telecommunications or ICTs18.
The development of active social movements lobbying for policy change in support of
poor, chronically poor and marginalised people is also commonly hampered by
ited
lim
labelled by others as 'disabled' might feel that their strongest identifying ch
is instead that that they are a farmer, mother, Banyoro or a Ugandan. The
viduals' identific
th
17
This section of the paper draws on Bird, K. , and Pratt, N. with O'Neil, T., and Bolt, V.J.
(2004a) 'Fracture Points in Social Policies for Chronic Poverty Reduction' October 2004. ODI
orking Paper 242.
W
www.odi.org.uk CPRC Working Paper 47 www.chronicpoverty.org
18
ICTs - information and communication technologies.
11
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative
Methods for Poverty Analysis in Decision-Making
numerous shifting and overlapping group identities. It is therefore likely that
individuals self-exclude where membership of a certain group is associated with
stigma or where the labelling of a group is regarded as pejorative.
Social movements representing category-based groups tend to suffer from weak and
oorly trained leadership which is either co-opted or inadequately networked. Limited
e
represe
roup leaders. For example, evidence from Uganda
uggests that ethno-linguistic and geographic identity can overwhelm the local
me categorisation weakens the coherence of such
ovements. In addition, there tends to be considerable internal heterogeneity with
be
nable to control the content or process of their engagement. This can lead to their
their engagement being manipulated to benefit
e interlocutor. Clientelistic relationships with individuals may alternatively provide
ublic funds, but are unlikely to result
the representation of their interests in public fora.
he political processes surrounding policy agenda setting and policy making are
was a disinclination to welfare, despite social protection being the third pillar of the
ent
t
p
leadership experience and the strength of alternative identities can weaken th
ntative power of g
s
government representatives of disabled people, who become co-opted by local elites
into representing the interests of their clan or local area rather than the interests of
disabled people. Limited vertical solidarity between elite or non-poor individuals and
poor people who share the sa
m
individuals strongly aware of their differences from each other. These can lead to
hierarchies, power asymmetries and 'internal' labelling which may affect how
leadership drawn from one sub-grouping represents the group as a whole.
Atomised and with limited leverage, even through electoral politics, political
movements may fail to take hold. Individuals may feel that their best hope is through
cultivating 'friendships' with patrons. NGO lobbies, parliamentarians or other elected
representatives, donors and development researchers may speak on their behalf.
Such relationships, however, are generally asymmetric, and the weaker party may
u
interests being misrepresented, and
th
them with some means of accessing (diverted) p
in
Evidence weakly used in policy formation
The political economy in many developing countries is such that the need to deliver
improved rights for marginalised and vulnerable groups is rarely seen to justify either
increased political attention or the devotion of increased resources to those groups.
Governments find it difficult to prioritise marginal groups and the chronically poor.
They are unlikely to develop and implement policies favouring these groups over
larger and more powerful groups, as they would have little to gain and much to lose
as a result. As a result, evidence may be weakly used in policy formation.
Dominant discourse demarcates 'space' for policy response
T
commonly influenced strongly by the dominant poverty and development discourses.
Policy discourse determines (and is determined by) what is sayable and thinkable
(Gasper and Apthorpe, 1996). National discourses are significantly influenced in
developing countries by donors.
For many years the Washington Consensus supported a neo-liberal agenda in which
the 'rolling back of the state' and the 'enabling' of enterprise were dominant. There
World Bank's poverty reduction strategy outlined in the 1990 World Developm
Report. This is only slowly shifting, and developing countries' poverty reduction
policies commonly still focus primarily on growth (without considering the nature of
th policies for social developmen
growth) and on the 'economically active poor', wi
12
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative
Methods for Poverty Analysis in Decision-Making
and asset creation for the poor concentrating largely on untargeted investments in
the health and education sectors.
Although there are exceptions, for example in India, where there is a stronger focus
port, elite
erceptions. These elite perceptions commonly reinforce categorisations of the poor
he problems of certain groups are seen as less important
he problems of certain groups appear to be seen by policy makers as unimportant.
s
bout what causes poverty, how economic growth occurs and how poor people can
, making it
ifficult for policy makers to help them. These issues are discussed briefly in the
nd the culture of poverty.
the combined result of decisions made by rationally
riented elites to protect themselves against 'public evils', plus the active
on reducing income poverty. This world view results in an expectation that growth will
result in significant and sustained poverty reduction with any remaining poverty being
largely residual. The dominance of this way of thinking does not encourage a focus
on investment or policy agendas 'outside the box'. The 'box' which delimits the areas
of accepted focus can also be described as the 'framework of possible thought'
(Chomsky, 1987). Issues which fall outside this box or framework are regarded as
subversive or irrelevant. Thus, for example, alcohol dependence and mental illness
are off the development map and development practitioners have been slow to take
disability seriously. Research findings which identify such issues are rigorously
interrogated and may even be intentionally and systematically undermined by the
knowledge communities allied with the dominant paradigm.
Dominant poverty and development narratives may interact with, and sup
p
as deserving and undeserving. These categorisations are used to justify the limited
attention and low budgetary allocations given to particular issues and groups. The
categorisations of deserving and undeserving poor, in turn, determine the framing of
certain research questions, so that some questions are emphasised and work on
them funded, while others are not even fully articulated. A lack of research funding
for these low priority areas limits the generation of empirical evidence which might
challenge their perceived unimportance. The framing of research questions and the
availability, or otherwise, of empirical evidence has an interactive relationship with
both agenda setting and policy formation.
T
T
This perceived lack of importance is linked to particular poverty and development
discourses, the emergence of which can be explained in a number of ways. People in
the country's the political, social and economic elite may have clear perception
a
escape from poverty. These 'elite perceptions' can draw on ideas of certain
categories of poor people as being 'undeserving'. The 'undeserving poor' may be
seen as lacking the entrepreneurial drive to make the most of opportunities in the
economy and their behaviour and culture may exacerbate their problems
d
following sections on elite perceptions, the politics of blame a
Elite perceptions
European elites historically felt threatened by the poor and viewed poverty reduction
measures as part of a 'civilising mission'. The development of welfare policies in
Western Europe can be seen as '
o
engagement of morally committed ideologues.' (Reis, 1999:135). This suggests that
pro-poor social policy is more likely to be formed and effectively implemented if
suitable combinations of ideological and pragmatic concerns can be identified.
Lessons from South Africa and Brazil suggest that the elites in highly unequal and
racially divided societies have limited vertical solidarity in poverty reduction (Kalati &
Manor, 1999; Reis, 1999). In both, elites feel that their personal security is threatened
by high levels of urban crime. In Bangladesh, which is (largely) ethnically and
13
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative
Methods for Poverty Analysis in Decision-Making
religiously homogenous, the elite sees the poor as the repository of social morality
(Hossain & Moore, 1999).
Elites may support policies which tackle the negative externalities of poverty if they
feel that such investments are in their own interest. But they do not face revolution or
the contagious epidemic diseases that European elites feared historically and which
drove the development of social programmes. Elites in South Africa and Brazil fear
violent crime but they invest privately in security rather than collectively (Kalati &
Manor, 1999; Reis, 1999).
While the practical or rational interests of elites support poverty reduction policies it is
not clear that either their fear of crime or their vertical solidarity creates a strong
enough 'push' for change. In the absence of this 'push', education is identified as the
best mechanism for reducing poverty in South Africa, Brazil, Bangladesh and
Uganda. It is seen as being a pain-free and non-zero sum solution. Through
ducation everyone can get richer without redistribution.
Lewis (1965; 1969) and
uggests that poverty is perpetuated through low levels of education, a lack of
eserving' and 'undeserving' poor, and suggests that it is inherently racist and
001). This view suggests that any attempt to eradicate or alleviate
overty among the 'underclass' is doomed to failure (ibid.). The opposing view
e
However, Hossain and Moore show that elite discourses in Bangladesh which
support universal primary education as a key anti-poverty intervention could be
shaped to widen support for investments in child nutrition (malnourished children
cannot concentrate in school), health care (to reduced absenteeism), adult literacy
(illiterate mothers cannot fully support the education of their children). A similar
analysis of elite perceptions of poverty and the anti-poverty discourses in other
countries could be used as the basis for understanding which could support the
creation of a supportive policy environment for pro-poor social policy.
The culture of poverty
The idea of a culture of poverty was developed by Oscar
s
participation in mainstream society and the inherent socio-psychological, political and
economic traits of the poor themselves. This theory suggests that the poor have a
different culture to the rest of society which is characterised by deviant attitudes,
values and behaviours (Patterson, 2000). These affect the way in which capital is
transmitted intergenerationally and may result in individuals being unable or unwilling
to take advantage of emergent opportunities (Moore, 2001).
Moore (2001) summarises the ongoing debate on the 'culture of poverty' theory and
shows that at one end of a continuum, there are those (e.g. Edward Banfield) that
believe that much if not most poverty is based upon the 'innate' characteristics of the
poor, sometimes called the 'underclass'. Moore links this approach to concepts of the
'd
classist (Moore, 2
p
suggests that poverty emerges and persists solely because of socio-economic
structures external to the value systems and behaviours of the poor. In between lie
those, like Lewis, who believe that 'cultures of poverty' have emerged over
generations to enable poor people to cope with their situation. Their values, beliefs
and behaviours were once an appropriate response to the social, economic and
political barriers that they faced but have become ossified and now limit the current
generation's ability to respond to opportunities.
Some research has suggested that social class has a powerful influence on
behavioural, social and psychological variables (Singh-Manoux & Marmot, 2005), but
contradictory findings suggest that it is " very difficult to make any comprehensive
14
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative
Methods for Poverty Analysis in Decision-Making
cross-cultural generalisations about the poor other than that they lack money and are
often socially and politically marginalised" (Rigdon, 1998:17, in Moore, 2001).
the development and the reproduction of an underclass
orcoran, 1995; Ario, et al ., 2004).
m those of welfare use or
eparated short-term and longer-term reliance on welfare.
vidence from low and middle income developing countries suggests that socio-
ocial factors associated with being poor and excluded or
iscriminated against can create their own barriers to the escape from poverty.
wages - parents who
orked when they were children may send their own children to work because they
ally facilitate survival in the midst of
ad or deteriorating socio-economic, political or environmental conditions, keeping
There has been a focus in the United States literature on the perverse incentives
generated by a welfare system. These include a reluctance to supply labour at low
wages, teenage out-of-wedlock births, and the low levels of aspirations associated
with continued welfare use which creates welfare dependency. By changing the
behaviour of both recipients and their children welfare systems are argued by some
in America as supporting
(C
Corcoran (1995) found that there was no evidence welfare receipt alters values or
attitudes or had a consistent effects on the labour supply and earnings of welfare
recipient's sons. However welfare-recipient parents were more likely to have
daughters who received welfare and had became unmarried teenage mothers. Few
studies disentangled the impacts of parental poverty fro
s
Findings from a study which analysed the National Longitudinal Survey of Youth
1979 cohort in the US suggest that the extreme and persistent racial divide in wealth
ownership in the US is at least partially the result of family processes during
childhood (Keister, 2004). This is because household disintegration can result in child
poverty, which in turn is associated with being a low income adult and the analysis of
this data shows that African American households are less stable over time.
E
cultural and psycho-s
d
Research in India found that coming from a group which has been discriminated
against over an extended period of time may have a persistent impact on the earning
potential of members of that group through its impact on individuals' expectations of
prejudice which suppresses motivation (Hoff & Pandey, 2004). Also some extended
family systems can hold back individuals who might progress more economically if
they took an individualistic approach to life. This can result in the kin system
becoming a poverty trap for its members (Hoff & Sen, 2005). Other evidence shows
that - despite the impact of working as a child on life long
w
feel that working will strengthen their characters or provide them with useful
experience (Emerson & Souza, 2005).
However, some of the socio-cultural traditions, institutions and value systems which
appear to entrench poverty - for example the gendered inheritance practices in South
Asia and traditions in parts of Africa which encourage the investment of profit in
social rather than financial or physical capital - are wide-spread across socioeconomic
groups and are "structural impediments that both the poor and the rich
must negotiate" (Moore, 2001).
Moore (2001) suggests that the growing literature on coping strategies provides an
alternative way of looking at 'culture of poverty' as it links the behaviour of the poor to
the context in which they live. It is likely that the coping and survival strategies
passed on from one generation to the next actu
b
the poor from destitution or death but often helping to reproduce the social and
economic structures that obstruct escape from poverty - a form of 'adverse
incorporation'. (Moore, 2001)
15
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative
Methods for Poverty Analysis in Decision-Making
However, the notion of the poor having a particular culture or set of cultures or
ehaviours which intensifies and perpetuates their poverty has been challenged.
here are many reasons why policy makers could ignore the issues facing poor,
roblems might be fully understood and the scale and severity of the problem
income
rowth either directly or through remittances and other transfers. However, the
rlap and the public sector is appropriated by
rivate interests. The failure to institutionalise power enables it to be personalised in
prevent the development and implementation of
ffective pro-poor policy. This is even more the case where the poor are drawn from
rawn largely from national elites. Research has shown
b
Corcoran (1995) argues that intergenerational poverty in the United States is
commonly the result of racial discrimination and a lack of opportunities that affect
individuals in each new generation rather than an intergenerationally transmitted
culture of poverty.
Not high on policy makers' agendas
T
chronically poor and marginalised people. Most simply, it may be a matter of political
arithmetic; there are not enough people affected by a particular issue in any one
constituency to create the demand to address the problem. However, following the
analysis above, this is unlikely to be the case; numbers of affected voters is at least
enough to mean that such issues should figure in the discourse among others.
P
recognised, but policy makers may still be reluctant to respond. This may be because
they make inaccurate assumptions about the distribution of benefits from growth and
the effectiveness of traditional safety nets. These assumptions may lead them to
believe that both the most efficient and the most effective way of generating
improvements in well-being for poor, chronically poor and marginalised people is
through economic growth, and that this will eventually result in widely spread
g
household characteristics of the most severely and chronically poor largely excludes
them from benefiting from economic growth in the wider economy and traditional
safety nets have only ever protected some people, some of the time. Under pressure
from the impacts of HIV/AIDS and societal change, these traditional systems cannot
be relied on to move people out of chronic poverty or prevent its transmission to the
next generation.
Alternatively, policy makers may be reluctant to respond because the national policy
process has been distorted by clientelism and neo-patrimonialism. In neo-patrimonial
states, a form of citizenship which binds individuals directly to the state above and
beyond the ties of kinship, community and faction has failed to develop (Chabal &
Daloz, 1999: 6). State service is personalised rather than bureaucratic and
functionaries do not necessarily see their roles as legally and professionally
distinctive. Private and public roles ove
p
this way (van de Walle, 2001: 117), and access to public state institutions is seen as
the main means of personal enrichment (Chabal & Daloz, 1999). Illiberal structures
and the weak institutionalisation of political practices, benefit the political elite but
undermine 'the developmental state' (van de Walle, 2001: 116). So, neo-
patrimonialism, a weak separation between public and private, and poor
institutionalisation of the state
e
marginalised groups.
Clientelism is a key feature of neo-patrimonial states. It involves the exchange of
gifts, favours and services, patronage and courtier practices (van de Walle, 2001:
118). Maintaining clientelistic relationships undermines representational relationships
between citizens and political leaders, and makes it more difficult for the needs of the
poor to be represented in policy fora or responded to adequately by government.
Policy makers are likely to be d
16
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative
Methods for Poverty Analysis in Decision-Making
that where there is limited vertical solidarity, elites are more likely to act to reduce
ively against policy innovations. For example,
olicies to reduce alcohol consumption may be countered by global beverage
s (2000) have shown that for problems to be identified as serious
nd in need of immediate remedy, policy makers must be linked with researchers in
e: not a political winner
poverty and inequality if they associate such changes with enhanced personal safety
and economic security (Hossain & Moore, 1999, 2002). Arguments in favour of
poverty reduction which are based on notions of equity and rights will appeal to policy
makers' altruistic qualities but often need to be supported with efficiency arguments
which illustrate that policy changes will support economic growth or other pressing
objectives.
Vested interests may argue persuas
p
manufacturers concerned about reduced sales revenue and by Ministries of Finance
concerned about lost tax revenue. Alternatively, policy makers' attention may be
diverted by lobbies which are more visible, vocal and persuasive of the urgency of
their case.
Not recognised as severe or large-scale problems
Issues may fail to be adopted by policy makers because they are not be seen to be
sufficiently severe or large scale and policy makers do not feel they can justify
allocating time or budget to the issue. As we have shown, this can be because the
issues are poorly understood or because other constituencies and interest groups
are more effective or more powerful and therefore more able to dominate the
attention of policy makers, it may also be that policy narratives and the 'framework of
possible thought' (Chomsky, 1987) are such that there is low demand for information
on these issues, and so little research has been undertaken or it has been poorly
disseminated. Donors play a substantial role in framing the possible - in low income
countries they bring resources and knowledge to the table. It is therefore likely that a
substantial proportion of the neglect of these issues can be attributed to their low
priority among donors.
Keeley & Scoone
a
active networks. Where researchers can demonstrate an easily articulated problem,
support their argument with empirical evidence of the scale and severity of the
problem and suggest practical and easily implementable solutions, so much the
better. Expressing complex or nuanced realities is less likely to gain the attention of
policy makers and researchers therefore tend to trade off rigour for impact by simplify
their message and putting into a 'black box' complex or disputed analysis (ibid.: 9).
The interest that policy makers have in particular research findings, and their
willingness to take them seriously and feed them into policy processes depends, in
part, on the researcher's membership of networks, but also on whether the research
topic and findings fit with the current discourses and lie within the 'framework of
possible thought'. If outside this area, the researcher can expect to have their
findings contested or ignored, blurring the distinction between the technical and the
political (ibid.: 4). What is identified as fact is therefore socially constructed (ibid.: 7).
Complex and expensiv
Policy responses to complex problems which are not amenable to single-agency
interventions are daunting to policy makers. Policy design is technically complex and
implementation is administratively demanding. Achieving success is likely to be
expensive, and where the policy is aimed at helping marginalised groups, personal
and career rewards can appear distant. These incentives will drive individuals to seek
easier sources of success.
17
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative
Methods for Poverty Analysis in Decision-Making
3.2. The political economy of poverty related policies in Uganda
A thorough understanding of the Ugandan political economy was necessary if
findings about the drivers and maintainers of chronic poverty in Uganda were to have
any hope of influencing policy. Thorough political analysis has shown that politics in
contemporary Uganda holds as many threats as opportunities for reducing long-term
overty (Hickey, 2005). Little effort has been made amongst development actors in
nd deeper group of supporters needs to be found, beyond the current
hampions' (ibid.).
.3. Mapping the political context of pro-poor policy making in Uganda
ed. After this, an analysis was undertaken to assess the
egree of influence each actor has over the policy process and the resources at their
his analysis was very helpful in developing a dissemination strategy for earlier work
p
Uganda to articulate the type of 'pro-poor' or redistributive growth that is likely to be
required to alleviate chronic poverty and key interest groups in Uganda are yet to be
convinced that measures should be targeted at the poorest groups and regions.
Many in Uganda also regard the poverty reduction agenda as externally imposed.
This may mean that if poverty reduction is to stay on the political agenda commitment
a broader a
'c
3
In this section we describe a 'quick and dirty' exercise undertaken to map of the
political context in Uganda. This built on the analysis presented above on the barriers
to pro-poor policy engagement and of the politics of poverty related policy making in
Uganda.
First, the main actors connected with policy processes in Uganda were identified.
See Figure 3, below. Then a political interests map was developed, using an
approach similar to that developed by Merilee Grindle (see Nash, Hudson and
Luttrell, 2006). This involved prioritising the policy areas for the different actors
identified in Figure 4, below. For this exercise some categories (e.g. national civil
society) were subdivid
d
disposal to influence policy outcomes. Lastly an assessment was made of actual and
potential alliances amongst actors.
This mapping exercise confirmed the relative power and influence of the Ministry of
Planning, Finance and National Development and certain Parliamentary Committees
(e.g. the Budget Committee) and highlighted the relative lack of influence of other
actors (e.g. Parliament, many national NGOs).
T
on the drivers and maintainers of chronic poverty in Uganda.
18
Q Squared in Policy: A conference on the Use of Qualitative and Quantitative
Methods for Poverty Analysis in Decision-Making
Figure 4: Key actors involved pro-poor policy processes in Uganda.
kly influences policy decision making
ers. They may feel that
dgement, experience, tacit knowledge and substantive expertise is of
greater weight than research-based evidence (ibid.). The values and self-belief of key
policy-makers and political context are all extremely important. Decision-making in
government and by the civil service, in many countries, is strongly affected by
organisational norms, habit and tradition. Much of this tends towards reinforcing
opinion-based decision-making.
Davies found that policy makers want both soft and hard evidence, including
evidence that we might consider anecdotal, which would indicate whether they were
'heading in the right direction'. UK-based policy makers had not heard of
experimental evidence and were not interested in it. Nor were they interested in the
findings from systematic reviews of evidence.
International actors
Multilateral and Bilateral donors
International Civil Society Organisations (e.g. NGOs, Trade Union Movement)
Private sector
Diplomatic representatives of other countries
International media
Regional bodies (African Union, East African Union, COMESA)
National actors
Parliament and Parliamentary Committees
Office of the President
Ministry of Planning, Finance and National Development
Education, Health, Agriculture and other Line Ministries
PEAP Steering Committees
Local government bodies
National civil society organisations
Political parties
Membership organisations (e.g. trade unions)
Social movements
Technical experts (sectoral specialists, academics)
Citizens
Media
Legislature
Police and army
3.4. Using evidence in decision-making
Running alongside this context mapping exercise was an attempt to assess how
important evidence is in policy making in Uganda. Findings from elsewhere suggest
hat evidence, including research evidence, wea
t
(Davies, 2005) (see also section 3.1., above)
esearch evidence is not alw