This was a meta-evaluation of evaluation five shelter projects funded by a large agency after the earthquake in Haiti. Despite verbal agreements to share the (anonymised) findings, this report was never made public. All the information presented here is in the public domain or is included with notional data to demonstrate the methodology.
We insisted on looking at the earliest decisions to allocate money to the shelter sector. A third of overall funding was needed to reach just 1% of people with housing support. Half of shelter funding was spent to reach fewer than 25% of households with a transitional shelter. Lower cost alternatives to transitional shelter might have been a better early decision.
The pie charts show households reached compared to money spent reaching them.
Costs and Quality: materials, people, logistics and delivery
This was also a unique opportunity not just for cost comparisons between different organisations but also cost breakdowns and the relative spending on staffing and material costs in relation to shelter quality and procurement decisions. This seemed to show that:
- Not all organisations could break down their costs;
- Material costs did not necessarily account for the highest per unit prices or variations in price;
- Higher quality could be achieved with lower spending on materials if there was higher spending on construction supervision;
- Every procurement route (for more or less identical products) was different so whether the procurement decisions were appropriate was probably a function of internal organisational capacity rather than other contextual factors;
- More had to be spend on trucks and vehicles in hard-to-reach locations.
'Medium' data for cross-referencing and checking baselines
We were also able to cross-reference data to check whether the proportion of tenants in the local population had: a) been assessed; and, then, b) was represented in the same proportion within the beneficiary group. This showed that households on temporary sites (sites that had been justified on the basis that they were needed for tenants) were in fact nearly 100% owners.