DIVA Pilot Programme · Matched Panel Analysis · March 2026
162 Matched Households —
Baseline to Endline
Panel analysis linking the Oct–Dec 2025 baseline phone survey to the March 2026 end-of-pilot survey using M-Pesa numbers as unique household identifiers. Only matched records are included.
01 — Food Security
Meal Frequency: Before & After
Meals per day — 162 matched households (129 with data at both points)
Baseline · Oct–Dec 2025
13%
eating 3 meals/day
17 of 129 with meal data at both points
86% on 2 meals or fewer (112/129)
→
Endline · March 2026
77%
eating 3 meals/day
99 of 129 with meal data at both points · 33 of remaining 33 HHs had no baseline meal record
↑ +64 percentage points (of 129 paired HHs)
Meal Frequency Distribution
Matched households — baseline vs endline · n=129
Household-Level Meal Change
Direction of change per household · n=129
Improved (more meals)
68 HHs
Worsened (fewer meals)
5 HHs
| Meal level | Baseline | Endline | Change |
|---|
| 1 meal/day | 1 | 2 | +1 |
| 2 meals/day | 111 | 61 | −50 |
| 3 meals/day | 17 | 99 | +82 |
82 additional households moved to 3 meals/day between baseline and endline. The number eating 2 meals halved from 111 to 61. Only 5 households reported a worsened meal frequency — shifting from 3 meals to 2 — suggesting a small group experienced a secondary hardship after initially recovering.
02 — Hunger Incidence
Days Without Food: Household Flows
71
Hunger Resolved
Had hunger before,
none after
20
Never Hungry
No hunger at
either point
59
Hunger Persisted
Still food insecure
at endline
12
New Hunger
No hunger before,
yes after
Hunger incidence — 162 matched households
Baseline · Oct–Dec 2025
80%
experienced hunger days
130 of 162 households · Sept worst month
→
Endline · March 2026
44%
experienced hunger days
71 of 162 households
↓ −36 percentage points
Hunger Incidence Comparison
% of matched households experiencing hunger · n=162
While hunger dropped from 80% to 44%, 59 households still experienced hunger days after the donation — the single most important unresolved issue. Additionally, 12 households who were not hungry at baseline reported hunger at endline, suggesting new shocks emerged post-donation. These 71 households (59 persisted + 12 new) should be prioritised for follow-up.
03 — Livestock
Herd Composition Changes
Note: Not all matched households had livestock recorded at both points. Sample sizes: Cows n=52, Goats n=80, Sheep n=45, Camels n=9. The endline records livestock at the time the donation was received — not at the time of survey — so endline figures may slightly understate current holdings.
Mean Herd Size Per Household
Baseline vs endline (at time of donation)
Individual Household Outcomes
Households that gained vs same vs lost each species
| Species | Baseline Mean/HH | Endline Mean/HH | % Change | HHs Gained | HHs Same | HHs Lost | n |
|---|
| 🐄 Cows | 2.02 | 0.87 | −57% | 10 | 6 | 36 | 52 |
| 🐐 Goats | 2.00 | 1.52 | −24% | 27 | 12 | 41 | 80 |
| 🐑 Sheep | 2.27 | 1.11 | −51% | 12 | 6 | 27 | 45 |
| 🐪 Camels | 1.22 | 0.00 | −100% | 0 | 0 | 9 | 9 |
Herd losses were severe across all species — driven by the prolonged Sept 2025 drought. All 9 camel-owning households lost their entire camel holdings. Goats showed the most resilience: 27 households actually increased their goat herd, making goat restocking the strongest candidate for a targeted recovery programme. On livestock deaths post-donation, 105 of 162 matched households (64.8%) reported no deaths, while 57 (35.2%) did report deaths — a significantly more difficult picture than the broader unmatched cohort, and consistent with the severity of drought losses captured in the herd size data above.
04 — Financial Wellbeing
Savings, Stress & Resilience After the Transfer
Savings Balance Change
Because of cash transfer · n=162
Financial Stress Change
Stress levels since transfer · n=162
Unexpected Expense Capacity
Ability to meet major expense · n=162
| Metric | Improved | No Change | Worsened |
|---|
| Savings balance | 67 (41%) | 94 (58%) | 1 (1%) |
| Financial stress | 138 (85%) | 23 (14%) | 1 (1%) |
| Unexpected expense capacity | 104 (64%) | 57 (35%) | 1 (1%) |
Financial stress improvement is the strongest metric here — 85% of matched households felt less financial stress after the transfer, with virtually no deterioration. Unexpected expense capacity improved for 64%. Savings improvement is more modest (41%) — with 58% reporting no change — consistent with the donation being spent on immediate needs rather than accumulated. Crucially, in all three metrics, fewer than 1% of households reported worsening.
05 — Spending Patterns
How the Donation Was Used
Spending Allocation
% of 162 matched households choosing each category
M-Pesa Withdrawal Method
How households accessed the transfer · n=162
83% spent on household food/water, 64% on livestock food/water — directly addressing the two survival pressures identified in the baseline. Notably, 9% spent on school fees (15 households) and 9% on other needs (14 households) — a slightly broader spending pattern than the first endline file suggested, indicating some households used a portion for additional priorities. The majority (62%) withdrew cash, with 16% spending entirely via M-Pesa and 22% using both methods.
06 — Programme Perception
Value, Trust & Advocacy
8.2/10
Programme Value Score
Baseline · n=160
Programme Value Distribution
Scale 1–10 · How valuable is proactive protection?
81.25
NPS Score
Promoters 82.5% − Detractors 1.25% · n=160
NPS Score Breakdown
Promoters (9–10) vs Passives (7–8) vs Detractors (0–6) · n=160
Livestock Insurance Interest
Baseline willingness to buy insurance · n=112
Not interested/ready
77 HHs
WTP range: KES 50–1,000/cow
Most common: KES 100–200/cow
Main barrier: Lack of funds (cited by majority of "No" responses)
Livestock improvement agreement — endline · n=162
Endline · March 2026
95%
Agreed livestock improved due to donation
154 of 162 households · 8 said No
Near-unanimous endorsement
|
Said No
5%
8 households
These 8 HHs warrant individual follow-up
With a baseline programme value score of 8.2/10 and a true NPS of 81.25 (Promoters 82.5% minus Detractors 1.25%), beneficiaries were already strong advocates before the donation was received. Post-donation, 95% of matched households agreed their livestock situation improved. The combination of high advocacy scores and experienced benefit creates a strong foundation for a second programme phase or insurance product introduction. Note: NPS is calculated as % Promoters (score 9–10) minus % Detractors (score 0–6). Passives (score 7–8) are excluded from the calculation. The NPS was collected at baseline, before donation delivery.
07 — Drought Context
The Baseline Crisis: Jun–Sep 2025
Drought Months Reported
Baseline survey · Jun–Sep 2025 · matched HHs
Income Distribution at Baseline
Monthly household income (KES) · n=160
Median household income at baseline: KES 2,650/month · Mean: KES 2,842/month · 96% of households earned between KES 1,000–5,000/month (154 of 160 with income data). September 2025 was the peak drought month — reported by 74 of 162 matched households. This provides critical context for interpreting livestock losses and hunger incidence as drought-driven rather than programme-driven.
08 — Summary
Matched Panel: Complete Results
| Outcome | Baseline (Oct 2025) | Endline (Mar 2026) | HH-level Change | Assessment |
|---|
| 3 meals/day |
13% (17/129 paired) |
77% (99/129 paired) |
↑ 68 improved · 5 worsened · 33 HHs no baseline meal record |
✅ Strong |
| Hunger-free |
20% (32/162) |
56% (91/162) |
71 resolved · 59 still hungry · 12 new |
⚠️ Partial |
| Cow herd mean/HH |
2.02 (n=52) |
0.87 |
↓ −57% · 36 of 52 lost cows |
❌ Drought-driven |
| Goat herd mean/HH |
2.00 (n=80) |
1.52 |
↓ −24% · but 27 HHs gained |
⚠️ Resilient |
| Sheep herd mean/HH |
2.27 (n=45) |
1.11 |
↓ −51% · 27 of 45 lost sheep |
❌ Drought-driven |
| Camels |
1.22 (n=9) |
0.00 |
↓ −100% · total loss |
❌ Total loss |
| Savings improved |
— |
41% (67/162) |
58% no change · 1% worsened |
✅ Positive |
| Stress improved |
— |
85% (138/162) |
14% no change · <1% worse |
✅ Strong |
| Expense capacity improved |
— |
64% (104/162) |
35% no change · <1% worse |
✅ Positive |
| No livestock deaths post-donation |
— |
64.8% (105/162) |
35.2% (57 HHs) reported deaths — drought-driven |
⚠️ Partial |
| Agreed livestock improved |
— |
95% (154/162) |
Near-unanimous |
✅ Strong |
| Programme value (1–10) |
8.2 mean |
— |
High pre-existing trust |
✅ Strong |
| NPS Score (0–100) |
81.25 · World-class |
— |
Promoters 82.5% − Detractors 1.25% |
✅ Strong |