← Portfolio

The Approval Gap

Do building plans passed and buildings completed move together — and what can the gap between those two parts of the construction pipeline actually tell us?

PUBLISHED CASE
Source
Stats SA P5041.1 · May 2026
Period analysed
Jan 2025 to May 2026 monthly window
Price basis
Current prices + constant 2019 prices
Created by
Nyashadzashe Munyati

A plan approved today is not a building completed today.

Construction statistics often report two different things. Building plans passed tell us about work that has received approval and may move into construction. Buildings completed tell us about work that has reached the other end of the pipeline. The completed building may have been approved many months earlier.

That creates a practical problem. If plans passed are rising while completions are falling, is the construction pipeline strengthening, slowing down, or simply showing the normal time delay between approval and completion? A same-month comparison cannot answer that on its own.

THE PROBLEM

Do not confuse an approval with a completion.

I wanted to compare the upstream and downstream sides of the building pipeline without pretending that one month’s completions came from the same month’s approvals.

WHAT I SET OUT TO DO

Measure the gap, then test whether a lag is visible.

I extracted the published monthly series, separated current-price and real-price values, compared building types and provinces, and then tested whether plans at month t aligned better with completions several months later.

TOOLS I USED

What each tool was doing

Stats SA P5041.1

Official source for plans passed and buildings completed.

Python / pandas

Structure the 17-month monthly series and calculate the exploratory lag correlations.

SQL

Reproduce monthly pipeline ratios and province/category comparisons.

Interactive charts

Show timing, real-price movement and geographic divergence without hiding the source limitation.

HEADLINE FINDING

In real terms, plans passed fell 2.9% in Jan–May 2026 — but buildings completed fell 8.0%.

The downstream side of the building pipeline weakened more sharply than the approval side. Non-residential work shows the biggest difference: real plans passed fell 6.4%, while real completions fell 20.7%.

Stats SA P5041.1 · constant 2019 prices · Jan–May 2026 versus Jan–May 2025
5.1 ppdifference between the total real movement in completions and plans passed
02 · DATA

What the source actually gives me

The official release contains both current-price values and constant 2019-price values. Current prices include inflation. Constant prices are more useful when I want to ask whether the actual volume/value of activity changed after removing price effects.

TIMEmonthJan 2025 → May 2026
UPSTREAMplans_passedcurrent + real values
DOWNSTREAMbuildings_completedcurrent + real values
BREAKDOWNSbuilding typeprovince
WHY I USE THE REAL SERIES FOR THE MAIN TREND

If construction prices rise, a current-price series can increase even when the underlying amount of building activity does not. Constant 2019 prices remove that price effect from the comparison.

03 · TRANSFORMATION

How I turned the publication tables into analysis-ready data

1Extract the monthly totals for plans passed and buildings completed.
2Keep current-price and constant-2019-price values in separate fields so nominal and real movement cannot be mixed accidentally.
3Convert the published R'000 values into numeric fields and create a controlled YYYY-MM month key.
4Validate Jan–May totals back to the Stats SA summary tables before calculating findings.
5Build separate province and building-type tables rather than forcing every breakdown into one wide spreadsheet.
6Calculate the lag test on the real total series, but keep the result labelled exploratory because the current working window is only 17 months.
17 monthly observations·9 provinces·3 major building categories·source totals reconciled
04 · CALCULATION

The calculations — and what they do NOT mean

REAL PIPELINE MOVEMENT2026 real value ÷ 2025 real value − 1

Plans passed: −2.9%
Buildings completed: −8.0%

The downstream completion series weakened 5.1 percentage points more than the approval series.
NON-RESIDENTIAL DIFFERENCECompleted real YoY − Plans real YoY

−20.7% − (−6.4%) = −14.3 pp

Non-residential completions deteriorated much more sharply than non-residential approvals.
SAME-PERIOD PIPELINE RATIO18.611bn ÷ 39.008bn × 100

47.7%

This is a descriptive ratio only. It is not a conversion rate because the completed buildings were generally approved in earlier periods.
EXPLORATORY LAG TESTcorr(plans at t, completions at t + lag)

Highest tested correlation: 0.497 at lag 0.

That is not evidence of “same-month conversion”. It means this short 17-month window does not reveal a convincing delayed peak. I do not force a lag conclusion from it.
05 · SEVERAL CHARTS

I test the finding from more than one angle

06 · INTERPRETATION

What I would say to a non-technical manager

Approvals are not completionsThe two series must not be matched month-for-month.

A project approved in May can finish much later. The current release is useful for pipeline direction, but not enough on its own to calculate a true conversion rate.

Downstream activity is weaker−2.9% real plans vs −8.0% real completions.

In the first five months of 2026, the completed side of the building pipeline weakened more sharply than plans passed.

Non-residential stands out−6.4% real plans vs −20.7% real completions.

This is the first category I would drill into when asking whether approved non-residential pipeline is translating into completed activity.

The lag is not proven yet17 months is too short for a defensible dominant-lag claim.

The responsible next step is to import the full official time series and test a wider 0–24 month lag window by building type.

07 · RECOMMENDATION

What I would do next

EVIDENCENEXT ACTIONWHY
17-month monthly windowImport the complete Stats SA building-statistics time series.A longer history is required to estimate a stable approval-to-completion lag.
Non-residential gap of 14.3 ppTest lag behaviour separately for residential, non-residential and additions/alterations.Different building types can have very different construction durations.
Province movements diverge sharplyKeep province as a reporting dimension and only move to municipality-level analysis if the source genuinely supports it.Do not invent geographic granularity that is not present in the published data.
Same-period ratio 47.7%Label it “same-period pipeline ratio”, never “conversion rate”.The numerator and denominator do not refer to the same cohort of projects.

Inspect the data and calculations

The page already shows the conclusion and its limitation. These files let somebody reproduce the work.