Before the anchor — Selection
Countries are chosen only on development similarity. Property performance is deliberately not consulted.
"Which countries looked like Kenya does now?"
Illustrative
Figures on this screen are labeled sample data used to demonstrate the workspace. Live connectors are not yet configured.The guiding principle
Do not treat property prices as an isolated signal. Country development comes first.
The pathway at a glance
Five stages in two halves. Stages 1–2 build the foundation and select a comparison set. Stages 3–5 observe what happened and translate it into a forecast. Selection is decided before any property outcome is examined.
- 1Foundation
Baseline scenario
Build Kenya's 5–10 year macro base case
Collect forward-looking views on Kenya's economy, demographics and social development over a five-to-ten-year horizon. Do not average the forecasts — interrogate each source's assumptions, methods and historical accuracy, then select the single most persuasive, internally consistent base case.
Output: A defensible Kenya baseline
- 2Foundation
Fixed analogue cohort
Select comparable countries; set the anchor date
Use Kenya's baseline to screen and rank countries that sat at a comparable development position at some point in their past. Select four to six countries and assign one common anchor date. This stage is selection only; it does not analyze property outcomes.
Output: A fixed country × time cohort
- 3Evidence to outlook
Market response
Observe the cohort's property response after the anchor
For each selected country, track what happened to real estate in the three-to-seven years after the anchor date, measured segment by segment before asking why. This is an observation step: no country reselection.
Output: Segment-level evidence
- 4Evidence to outlook
Causal hypotheses
Keep only drivers recurring across analogues
Turn the observed patterns into explanations. Retain a relationship only when it is supported across multiple analogues, not when it appears as a single-market correlation. Surviving mechanisms are the only things allowed into the forecast.
Output: Validated mechanisms
- 5Evidence to outlook
Forecast Kenya
Apply validated mechanisms to the baseline
Map the surviving mechanisms onto Kenya's baseline and state, for every segment, a probability of growth, a plausible value range, a confidence level and the named risks that could change the scenario.
Output: Decision-ready forecast
Selection ≠ observation — the rule that protects the model
The anchor date is a firewall. Everything before it is about choosing the comparison; everything after it is about measuring the result. The two must never inform each other.
After the anchor — Observation
With the cohort locked, the 3–7 year property response of each country is measured — winners and losers alike.
"What did property do next, and did it repeat?"
Prevents look-ahead bias
You cannot pick a country because you already know its market boomed — the boom is on the far side of the anchor date.
Forces full reporting
Disappointing analogues stay in the sample, so the base rate is honest.
Makes it repeatable
Anyone re-running the same screen and anchor date lands on the same cohort; the method is auditable.
Anchor placement: The anchor date sits 3–7 years in the past so a full post-event response window has already played out and can be measured — not forecast.
Decision rules — the short list
These rules are what keep the method disciplined. If a step is ever in doubt, return to them.
Select on merit, not average
Choose the best-reasoned macro scenario; do not blend forecasts into a mushy mean.
Country before price
The macro trajectory drives the view; prices are the outcome being explained.
Fix the cohort before looking
Countries are chosen on development similarity, with property outcomes unseen.
Small and well-matched beats large and loose
Four to six genuine comparables outperform a diluted set.
Keep every analogue
Do not drop the disappointments — that would inflate the base rate.
Cause over correlation
A driver enters the forecast only if it recurs across multiple analogues.
Forecast every segment explicitly
Each gets a probability, a range, a confidence level and named risks.
Re-anchor over time
Update scenarios, analogues and hypotheses as new data arrive.
Data sources and update cadence
The framework draws on public, well-documented sources so the base case and the analogue screen can both be audited and reproduced.
| Purpose | Primary sources | Refresh |
|---|---|---|
| Macro baseline (Stage 1) | World Bank, IMF WEO & Article IV, UN population/urbanization, AfDB, Oxford Economics | Annual; interim on major shocks |
| Analogue screening (Stage 2) | World Bank WDI, IMF, UN Habitat, national statistics offices | Annual |
| Property outcomes (Stage 3) | National statistics, central-bank house-price indices, private brokerage and research | Quarterly / annual |
| Mechanism validation (Stage 4) | Academic literature, cross-country studies, cohort observation | As evidence changes |
| Forecast & loop (Stage 5) | All of the above, re-anchored | Quarterly review |
Source lists are indicative; the exact providers depend on data availability for each analogue country.
Glossary
- Baseline scenario
- Kenya's chosen 5–10 year macro base case (Stage 1).
- Analogue cohort
- A small, fixed set of countries at a comparable development position (Stage 2).
- Anchor date
- The single point that splits selection from observation; set 3–7 years in the past.
- Response window
- The 3–7 year period whose property outcomes are measured (Stage 3).
- Transferable mechanism
- A driver that recurs across multiple analogues and is allowed into the forecast (Stage 4).
- Evidence weight
- How many analogues support a given mechanism.
- Validation loop
- The ongoing re-anchoring of scenarios, analogues and hypotheses as new data arrive (Stage 5).