21 July 2011

Some Good Nutrition News out of India?

A new paper by Eeshani Kandpal, forthcoming in the journal World Development reveals “unambiguous evidence that the ICDS (the Integrated Child Development Services programme) significantly reduces long term child malnutrition in India”.

ICDS is a child development programme with nearly a million centres throughout India. It is focused on welfare for children from birth to 6 years of age involving, feeding, health counselling, and education. Most studies have found ICDS to be ineffective in raising the levels of infant chronic undernutrition as measured by cumulative shortfalls of height for age from international standards (also known as stunting).

I feel that the paper slightly oversells its results, but maybe the temptation was strong, because the paper is very carefully done and methodologically sound and this is rare for ICDS, the largest nutrition-related programme in the world. In short, the paper is an important contribution.

The paper is noteworthy for a number of reasons:

• It uses the most recent nutrition data in India (2005-6)
• It properly uses propensity score matching to match up children in ICDS areas with those not in ICDS areas, but are identical in terms of observable factors that determine stunting rates and observable factors that determine participation in ICDS. The matching is vital, because ICDS is supposed to be targeted to areas in which stunting rates are highest. Without matching it is very possible that ICDS children would show higher stunting rates because of this targeting (and this is what the author actually finds)
• It compares matched and unmatched estimates of the effects of ICDS
• It breaks the results down by severely and moderately stunted children and by children under the age of 2 and under the age of 3, and by gender.


• For all children, the matched estimates show that ICDS has a positive and significant impact of 6% on child stunting. The unmatched estimates show a negative and significant impact of ICDS of about 7%. Matching matters, big time.
• The positive impacts seem stronger for the moderately stunted than for the severely stunted
• The results are stronger (magnitude and significance) for boys
• These analyses are run for the 1992-93 data and for the 1998-99 data and the results are generally zero. So ICDS is now having an overall positive impact on stunting, whereas before it did not.
• The 2005-06 results show stronger effects for those less than 3 years of age compared to those only less that 2 years of age
• A separate analysis of where and how ICDS centres are placed finds that placement is likely to be pro-poor, although areas with more educated mothers are favoured and areas with gender population imbalances are not favoured
• Another analysis of ICDS expenditures at the 29 state level indicates that stunting levels do not drive allocations, but votes for the political alliance that won the 2004 national election have a positive impact on the ICDS funds received

The paper shows that ICDS has an overall positive impact on stunting rates using the 2005-06 data (unlike with previous rounds of data), but that this impact is not particularly well focused on the most stunted or on under 2s.

The author suggests that the ICDS investment yields a 3.75 fold net return. The questions that remain: (a) did anything happen in ICDS design to result in this recent impact? (b) is a certain level of wealth needed before national integrated programmes such as ICDS can have an effect (as opposed to more focused cash transfers?), and (c) how much can the net benefit/cost ratio of 3.75 be improved (community based nutrition promotion reach ratios of 12:1--see Horton, Alderman and Rivera 2008, Copenhagen Consensus) through a more nutrition focused allocation of ICDS funds to states, through better placement of ICDS centres and through a better focus on under 2s, on those with severe stunting and on girls?

At last, we have a good nutrition story out of India (although not quite as good as the author would have us believe).

18 July 2011

Can the “commitment” to reduce hunger be measured? Should it?

The Irish Hunger Task Force was quite clear in its recommendations that someone should be measuring the commitment to reducing hunger – for rich and poor countries alike. It’s no longer acceptable to talk rather airily about “political will”. Wishful thinking on this is unacceptable, we need to now measure political commitment. But how? Any why?

The “why” is easier than the “how”. It is important to assess political commitment for several reasons:

• To hold governments to account
• To help mobilise society against hunger
• To bridge the divide between rhetoric and reality on political will
• To help government’s prioritise their own efforts to reduce hunger
• To highlight the political nature of hunger

How? An IDS initiative in partnership with Save the Children UK, ActionAid, Trocaire and Irish Aid has produced its first set of results in advance of a first report due out in late August. These were presented at a workshop at Action Aid today.

The Hunger Reduction Commitment Index (HRCI) initiative has operationalised “commitment” as a mix of publicly stated intent and action. It constructs an index from 9 secondary data indicators, 3 from each of 3 categories: policy focus on hunger, legal framework to support hunger reduction and public expenditure on hunger related sectors. The index needs to undergo further sensitivity analyses (and needs to figure out how to incorporate primary data collection on commitment) so I don’t want to talk too much about the specifics. Some preliminary observations:

• The preliminary rankings are quite different from the admirable ActionAid Hunger Free index which conflates hunger outcomes and hunger commitments and the IFPRI/Concern Global Hunger Index (which only looks at outcomes)
• When you focus solely on commitments, some countries rank a lot worse than they do on the HungerFree scorecard (China drops from 2 to 12 out of 21 countries) and some do better (Ethiopia moves up from 10 to 4)
• Brazil does well on commitments and outcomes, so it stays high in the rankings on this new index at number 1
• Malawi, comes out as 1 in the HRCI (number 4 in HungerFree index), which adds a twist to the current strained relationships with the UK
• The new index of commitment is compared against outcomes, administrative capacity, voice and wealth. It is interesting to see that several countries who have high commitment scores have medium admin capacity scores (e.g. Malawi) and low voice scores (e.g. Ethiopia), while some have low commitment scores and high admin capacity (e.g. Mozambique)

Many questions were raised by participants at today’s workshop:

• How are the primary data going to be incorporated into the index? How can perception indicators be made comparable across countries?
• How will the index, and the process of constructing it, be empowering for communities who are food insecure? How can their voices be contrasted with the voices of in-country “experts”?
• How do we reconcile a set of indicators that work for a range of countries with a set that are more meaningful at the country level?
• How do we avoid confusing messaging on hunger given the other indicators out there?
• How do we make the index the beginning of a dialogue rather than a name and shame exercise?

These are all issues that will be addressed in the next country-led phase of the initiative (assuming we can find resources). The opportunities for linking the next phase of the initiative into nationally owned processes around MDGs and right to food movements and into international debates (e.g. FAO reform process, and various accountability initiatives) seem plentiful.

I will give an update when the full report is available in late August.

17 July 2011

Is State Capacity a Political Choice? New Evidence from Bihar and Ethiopia

Two recent papers from IDS colleagues suggest that the answer to this question is “yes”. The paper by Santhosh Mathew and Mick Moore (2011) examines the weak capacity of the State of Bihar in India over the 1990-2005 period, and a paper by soon to be IDS Fellow Stephen Peterson (2011) on Public Financial Management in Ethiopia over the 1996-2010 examines the drivers of successful financial reform. The authors argue, In both cases, that deliberate and calculated political choices drove weak capacity in Bihar and strong capacity in Ethiopia.

Bihar is one of the poorest States in India, but is currently on one of the strongest upward trajectories when it comes to governance. It is also a key target of DFID resources. Mathew is a senior administrator in the State (he took a time out to complete PhD at IDS) and gives us a strong inside analysis of how and why such political choices were made.

He puts forward two principal reasons for Bihar’s poor performance:

• The type of leadership exercised by Chief Minister Yadav combined with the nature of his political coalition: to energise his political coalition of middle and lower caste groups he continually confronted the upper caste groups who were in control of state machinery

• This meant that the latter were not well disposed to helping him deliver “development” and the distribution of material resources to his core constituency. Even more importantly he undermined state capacity so as to exclude these historically powerful groups from the machinery. This meant that Bihar failed to compete for central Government money (won by other states with greater administrative capacity to fill out the forms) and that it often under-spent its own budget

Yadav’s crime was not incompetence, but the choice of where and how to deploy his considerable talents.

Ethiopia, in contrast, chose to aim for international standards when reforming its financial systems (it is now the third best in Africa and is managing the largest aid flows). In 1996 reporting on financial performance was experiencing a 6-7 year backlog, the Ethiopian Civil Service College was not teaching accounting or budgeting, and bookkeeping was single entry. That has all changed.

Peterson argues that the change was supported in the following ways:

• Ethiopian policymakers did not aspire to “summits of international best practice” but “consolidation of the basics of a firm financial plateau”
• Decentralisation, central to the survival of the ruling Tigrayan ethnic group, required effective public financial management reform, so the ownership of the reform was strong
• Public financial management was rightly regarded by the Government as a core function of the state and therefore a matter of sovereignty—an issue that donors should not drive
• The strategy of the reform did not stop at “recognise, improve and change”, but continued on to “sustain”

Both of these papers demonstrate:

• The value of the insider-outside perspective (Mathew, a skilled administrator stepping outside to a UK research organisation, and Stephenson, an American researcher, stepping inside an Ethiopian system)
• The dangers of apolitical analyses of state capacity
• The long time periods for things to get noticeably bad (Bihar) and noticeably good (Ethiopia)—there are no quick fixes
• The importance of leadership and discipline (to pursue developmental or non developmental ends)

Both papers are short and well written—I recommend them both.

01 July 2011

Violent Conflict and Incomplete Control

Yesterday I spent the day at a conference hosted by IDS on the Micro Analysis of Violent Conflict, organised by an EC supported research consortium, MICROCON.

The Director of MICROCON, IDS Fellow Patricia Justino, kicked things off by pointing out that the starting point in 2006 was that (a) most work on violence and conflict is at the state level, (b) we knew very little about the impacts of conflict and violence on ordinary people and (c) very little about how people can and cannot "navigate" (a term introduced by Philip Verwimp, one of the co-Directors of MICROCON) their ways through it--ways, which if uncovered, will provide important clues and cues to policy formulation in this area.

Some reflections

1. MICROCON has made a very conscious effort to collect data, sharing it as widely as possible. As Chris Cramer, one of the participants, said, if truth is the first casualty of war then evidence is one of the key battlegrounds--in other words, data can help shed light on dynamic, chaotic and uncertain contexts. Tilman Bruck (DIW in Germany) and Gary Milante (World Bank and an author of the latest WDR) spoke about how we can be more systematic about collecting data in conflict areas, using panels and more standard definitions of key variables.

2. The people focus is simple, but powerful. It helps delve into the household (do conflict and violence affect intrahousehold power dynamics in predictable ways? Do regular empirical observations --e.g. greater resources in the hands of women empowers them--hold when conflict is the shock?). It also helps us understand how institutions (in terms of norms, rules of the game and organisations) can lead to and be born from violent conflict (e.g. the importance of restoring health and education services---even if not focused on the poorest--is a way of signalling that normality is returning and hope invested in a peaceful solution is less likely to be squandered).

3. Incomplete control. We heard about Stathis Kalyvas' incomplete control hypothesis of when political actors use violence. When they have complete control (hence the Clash picture) they don't need to use violence and when they have no control violence is counterproductive as they have no information to make the violence selective.

4. Conflict can turn things upside down. Like in some parallel universe, university degrees are good things in peacetime but in conflict they can be a signal to help target elites.

5. Policy priorities. We heard from the policy-oriented participants how difficult it is to prioritise and sequence interventions in conflict affected and post-conflict settings, because everything seems to need addressing. The conclusion seemed to be "it's almost more important to prioritise something than what is that is prioritised". In the nutrition area, I highlighted how the presence of violence and conflict (and their apparently unpredictable distribution of impacts) made it even more important to protect the first 1000 days of an infants life, regardless of whether they are malnourished or not.

This is an exciting area. While I suspect the "impact of conflict" evidence is well on its way to being filled, the gap that is now yawning is the "impact of conflict prevention and mitigation interventions". That would build on the work of MICROCON version 1 and would be a good focus for MICROCON version 2.