Data and working: What it took to own a home, 1986 vs 2026

This page holds the data and working behind What it took to own a home: 1986 vs 2026, issue 1 of What the Numbers Leave Out, published 20 November 2026. With it you can rebuild every number in the article from public sources, check my arithmetic, and see how the issue scores against the Healthy Wealthy Investor Standard v1.1.

Licence: the text and data on this page are CC BY 4.0. Use them freely, with credit to Healthy Wealthy Investor, What the Numbers Leave Out.

Download the files

What each figure is

Each row of the reading file is one figure. The columns are:

ColumnWhat it holds
measure_idThe measure, from the list below (for example M1).
region, periodWhere and when the figure applies.
value, unitThe figure and what it counts (years of one wage, % of two wages, ounces of gold).
inputsThe figures read from source that go into it.
workingThe sum, written out.
source, source_dateWho produced each input, and the date of that release.
statusVERIFIED (read from source), INFERRED (calculated by me) or BROKER (rests on my own broker pricing).
status_for_readersThe same status in the words the article uses: Read from source, Calculated, Broker pricing.
MeasureWhat it is
M1Median house price ÷ one year’s average full-time pay. Years of one wage.
M1-MMThe median multiple: median house price ÷ median household income before tax (Demographia’s measure).
M2Years to save a 20% deposit putting aside 15% of one wage, no interest.
M3The lender’s test: yearly repayment at the test rate on the average new loan, as a share of one wage and of two.
M4The actual repayment, at the standard variable rate and at the rate actually paid, as a share of one wage and of two.
M5Real rate: home loan rate minus yearly CPI inflation.
M6What a $100,000 debt is worth after ten years of steady inflation, in starting-year dollars.
M7Wage, house and loan priced in ounces of gold, and the wage in ounces of silver.
M9What’s left of a week’s pay after the repayment, before tax, in 2026 dollars.
M12Sydney house price in years of pay for workers a quarter, half and three quarters of the way up the full-time pay scale.
M13Pay spread: what the top tenth of full-time workers earn as a multiple of the bottom tenth.
LOAN, INC, REALSupporting figures: the average loans, the 1986 household incomes, and the real (after inflation) changes in pay and house prices.

The calculation, in plain words

Every calculated figure uses one of the sums below. Each has one worked example from the article. Figures are rounded only at the end.

FigureThe sumWorked example
Pay for a yearWeekly pay × 52.$2,083.70 a week × 52 = $108,352 a year (May 2026).
The 1986 average loanTotal lending ÷ number of loans.$9,011.3 million ÷ 197,492 loans = $45,629, about $45,600. The repayments are run on the rounded $45,600, as the article says.
Yearly repaymentPrincipal and interest, interest charged monthly. Monthly repayment = loan × r ÷ (1 − (1 + r) to the power −n), where r is the yearly rate ÷ 12 and n is the number of months. Yearly = 12 × monthly.$736,000 at 6.9% over 30 years: r = 0.00575, n = 360, monthly $4,847.30, yearly $58,168.
The test rateThe rate offered plus APRA’s 3-point buffer.6.9% + 3 = 9.9%. The yearly repayment at 9.9% on $736,000 is $76,855.
Share of one wage and of twoYearly repayment ÷ one year’s pay, and ÷ two years’ pay.$58,168 ÷ $108,352 = 53.7%; ÷ $216,704 = 26.8%.
Years of payPrice (or loan) ÷ one year’s pay.Sydney 2026: $1,529,308 ÷ $108,352 = 14.1 years.
Years to save a deposit20% of the price ÷ 15% of one year’s pay.1986: $19,665 ÷ $3,332 = 5.9 years.
The median multipleMedian house price ÷ median household income.1986: $98,325 ÷ $22,000 = 4.5. With the Sydney income estimate ($22,056 × 35,544 ÷ 30,445 = $25,750): $98,325 ÷ $25,750 = 3.8.
What’s left each weekWeekly pay minus (yearly repayment ÷ 52). For 1986, the result is then multiplied by the CPI ratio 103.64 ÷ 30.85 = 3.359 to put it in 2026 dollars.1986: ($427.20 − $138.88) × 3.359 = $969. 2026 at 6.9%: $2,083.70 − $1,118.62 = $965.
Real changeThe rise in dollars ÷ the rise in the CPI (December quarter 1986 to June quarter 2026: 102.31 ÷ 30.85 = 3.316).Pay rose 4.878 times; 4.878 ÷ 3.316 = 1.47, so about 47% more in real terms.
Ounces of gold or silverDollars ÷ the price of one ounce in Australian dollars.Sydney 2026: $1,529,308 ÷ $5,511.99 = 277.45, so 277 oz. 1986: $98,325 ÷ $547.37 = 179.6, so 180 oz. 277.45 ÷ 179.63 = 1.545, about 54% more.
The 1986 gold and silver pricesThe average of the Perth Mint’s twelve monthly averages for 1986.Gold: the twelve monthly averages add to $6,568.45; ÷ 12 = $547.37.
Real rateHome loan rate minus yearly CPI inflation.1986: 15.5 − 9.7 = 5.8 points.
A debt after ten years of inflation$100,000 ÷ (1 + inflation) to the power 10.At 9.7%: $100,000 ÷ 2.524 = $39,600.
Pay at a point on the pay scaleThe ABS prints how many full-time workers fall in each pay band. Count up the bands to the one holding the point you want (a quarter, half, three quarters), then move through that band in a straight line. This is the method the ABS uses for its own deciles, and on the 1986 table it lands on every decile the ABS printed.1986, a quarter of the way up: a quarter of 4,249.0 thousand workers is 1,062.25 thousand. 793.8 thousand earn under $280, and the $280 to $320 band holds 626.5 thousand, so the point is $280 + (1,062.25 − 793.8) ÷ 626.5 × $40 = $297 a week.

Run the script and it prints every figure beside the article’s, with OK or DIFF. On 20 November 2026 it printed: 91 of 91 figures match the article to its rounding.

Every source, with the date it was read

The numbers match the article’s source notes. Every link was opened and read on 20 November 2026.

  1. Reserve Bank of Australia, Statistical Table F5, Indicator Lending Rates, standard variable owner-occupier rate (FILRHLBVS), published 6 November 2026. www.rba.gov.au, read 20 November 2026.
  2. Reserve Bank of Australia, Statistical Table F6, Housing Lending Rates, f06hist.xlsx, series FLRHOFVA (new owner-occupier variable loans, 6.22% September 2026), published 6 November 2026. www.rba.gov.au, read 20 November 2026.
  3. ANZ, home loan rate changes page, standard variable index rate (principal and interest) 9.14% from 13 November 2026. www.anz.com.au, read 20 November 2026.
  4. Australian Bureau of Statistics, Average Weekly Earnings, Australia 1941 to 1990, cat. 6350.0, Table 3 (November 1986, full-time adults, ordinary time, persons). Released 11 February 1992. www.ausstats.abs.gov.au, read 20 November 2026.
  5. Australian Bureau of Statistics, Average Weekly Earnings, Australia, May 2026 (released 13 August 2026). www.abs.gov.au, read 20 November 2026.
  6. Abelson, P. and Joyeux, R., Housing prices and rents in Australia 1980-2023, ANU Tax and Transfer Policy Institute, September 2023, Table A.1. crawford.anu.edu.au, read 20 November 2026.
  7. Cotality, Home Value Index tables, July 2026 data (released 3 August 2026). discover.cotality.com, read 20 November 2026.
  8. Reserve Bank of Australia, Statistical Table G1, Consumer Price Inflation (from ABS 6401.0), series GCPIAG, published 29 October 2026. Index 30.85 (December quarter 1986), 102.31 (June quarter 2026), 103.64 (September quarter 2026). www.rba.gov.au, read 20 November 2026.
  9. Australian Bureau of Statistics, Housing Finance for Owner Occupation, Australia, June 1988, cat. 5609.0, released 22 August 1988, Table 2. Average calculated by the author from total commitments and number of loans. www.ausstats.abs.gov.au, read 20 November 2026.
  10. Australian Bureau of Statistics, Lending Indicators, September quarter 2026 (released 11 November 2026), average loan size for owner-occupier dwellings. www.abs.gov.au, read 20 November 2026.
  11. Australian Prudential Regulation Authority, “APRA maintains current macroprudential policy settings”, 28 May 2026. www.apra.gov.au, read 20 November 2026.
  12. The Perth Mint, historical metal prices, monthly averages 1968 to 2015 (monthly.csv), Perth Mint spot, AUD. An archive file with no release date printed. 1986 figures are the average of the twelve monthly averages. www.perthmint.com, read 20 November 2026.
  13. The Perth Mint, “October 2026 sales update”, published 5 November 2026, month-end gold and silver prices in AUD. www.perthmint.com, read 20 November 2026.
  14. Cotality, Housing Affordability Report, Australia, November 2025 (data to September 2025), p.3. discover.cotality.com, read 20 November 2026.
  15. Cotality, “Housing downturn broadens as Brisbane and Perth lead the decline”, 2 November 2026. www.cotality.com, read 20 November 2026.
  16. Australian Institute of Health and Welfare, Home ownership and housing tenure, Australia’s welfare, 16 October 2025. www.aihw.gov.au, read 20 November 2026.
  17. Broker pricing, November 2026. The author’s own current pricing for typical owner-occupier borrowers (6.9%) and self-employed or alt-doc borrowers (7.0%). Not a published series. No public copy. The rates are given here (6.9%, and 7.0% for self-employed or alt-doc borrowers, November 2026), so every figure that uses them can be rebuilt.
  18. Australian Bureau of Statistics, Weekly Earnings of Employees (Distribution), Australia, August 1986, cat. 6310.0, released 1 May 1987, Tables 9 and 10 (full-time employees aged 20 and over, earnings in all jobs). Quartiles calculated by the author from the printed earnings bands; the method reproduces the printed deciles. www.ausstats.abs.gov.au, read 20 November 2026.
  19. Australian Bureau of Statistics, Employee Earnings, August 2025, cat. 6337.0, released 12 December 2025, Table 7 (full-time employees, earnings in main job). Quartiles calculated by the author from the earnings bands; the calculated median is within 1% of the published $1,741 a week, which is the figure used here. The 1989 and 2025 top-tenth and bottom-tenth figures were also calculated by the author from this table, which runs on the same basis from 1989, its earliest year. The 1986 and 2025 surveys differ slightly: 1986 covers full-time employees aged 20 and over, pay from all jobs; 2025 covers full-time employees of all ages, pay from their main job. www.abs.gov.au, read 20 November 2026.
  20. NSW Department of Communities and Justice, Rent and Sales Report, sales tables, March 2026 quarter, Table 3, Greater Sydney non-strata (houses): first quartile $1,194,000, median $1,550,000, third quartile $2,275,000. The report removes the top and bottom 5% of sales in each area before calculating these figures. No release date is printed in the file. dcj.nsw.gov.au, read 20 November 2026.
  21. Demographia International Housing Affordability, 2026 edition, Frontier Centre for Public Policy, released 11 June 2026 (data July to September 2025). Median multiple: median house price divided by gross median household income. frontiercentre.org, read 20 November 2026.
  22. Australian Bureau of Statistics, Census 86: Australian Families and Households, cat. 2506.0, published 1989, p. 32 (median household income, 1986, $22,000 a year, Australia-wide; no Sydney figure was found for 1986). The ratio of about 4.5 is the author’s calculation: $98,325 divided by $22,000. www.ausstats.abs.gov.au, read 20 November 2026.
  23. Australian Bureau of Statistics, Census Characteristics of New South Wales, 1991 Census, cat. 2710.1, published 1993 (median household income, 1991: Sydney $35,544, NSW $30,445 a year). The 1986 Sydney estimate (about $25,750, ratio about 3.8) is the author’s calculation and assumes Sydney’s 1991 margin over NSW also held in 1986. www.ausstats.abs.gov.au, read 20 November 2026.
  24. Australian Bureau of Statistics, Census 86: Summary Characteristics of Persons and Dwellings, New South Wales, cat. 2479.0, released 1 February 1989, Table C28. NSW median household income about $22,056 a year, interpolated by the author from the income bands. www.ausstats.abs.gov.au, read 20 November 2026.

Figure status

Read from source means copied as printed in the source. Calculated means my arithmetic on figures read from source. Broker pricing means my own current pricing, which is not a published series.

FigureStatus
Rates: 15.5% (1986), 17% (1989), 9.02% (October 2026), 6.22% (September 2026), ANZ 9.14%Read from source
Rates actually paid: 6.9% and 7.0%Broker pricing
Test rates: 9.9% and 10.0% (broker rate plus APRA's 3 points)Calculated
Average full-time wage: $427.20 a week (1986), $2,083.70 a week (2026)Read from source
Average full-time wage a year: $22,214 and $108,352 (weekly × 52)Calculated
Average new loan: $736,000 (2026)Read from source
ABS lending total $9,011.3 million and 197,492 loans (1986-87)Read from source
Average new loan: about $45,600 (1986, ABS total ÷ number of loans)Calculated
Median house prices: Sydney $98,325, Melbourne $82,000, Perth $58,000 (1986); Sydney $1,529,308, Melbourne $936,528, Perth $1,073,500 (July 2026)Read from source
NSW sales median $1.55m and quartiles $1.19m and $2.28m (March 2026 quarter)Read from source
Demographia median multiple 14.0; Cotality 13.3, 11.0 and 16.7 yearsRead from source
Sydney ranks second worst of 96 cities (Demographia, 2025)Read from source
Median household income: $22,000 Australia (1986); $35,544 Sydney and $30,445 NSW (1991), all medians onlyRead from source
NSW 1986 household income about $22,056; 1986 median multiples about 4.5 and 3.8Calculated
CPI index numbers and inflation rates (9.7%, 3.9%)Read from source
Pay at the quarter, half and three-quarter marks (1986 and 2025), except the 2025 median of $1,741 a weekCalculated
2025 median full-time pay, $1,741 a weekRead from source
Top-tenth to bottom-tenth pay (2.8 and 3.4 times)Calculated
Gold and silver prices: October 2026 close; gold fell 8.2% in OctoberRead from source
Gold and silver prices: 1986 (average of the twelve monthly averages)Calculated
Home ownership, 30 to 34 year olds (64%, 50%); values 6.7% below peakRead from source
Every repayment, share of wage, what's left each week, years of pay, years to save, real change, ounce figure, real rate and debt-erosion figureCalculated

Issue 1’s score against the Standard

Scored against the Healthy Wealthy Investor Standard v1.1 on the live page. Scorer 1 is Claude (Anthropic), scoring for HWI, so this is a self-score. Scorer 2 is Codex (OpenAI, a different model family), which scored the 2:46 pm (Perth) page blind, without seeing scorer 1’s card. The two agreed on 20 of 27 scored criteria (74.1%). Where they disagreed, the evidence was checked line by line and the lower score was kept in six of seven cases. The last column is the score after the fixes made up to 3:22 pm (Perth) today, re-scored by scorer 1 on the criteria those fixes touched.

IDCriterionScorer 1, 2:46 pmScorer 2, 2:46 pmNow (3:22 pm)
C1Who produced each numberMetMetMet
C2Why this measureMetMetMet
C3How old each number isMetPartlyPartly
C4How exact each number isMetMetMet
C5Where sources disagreeMetMetMet
C6What each term meansMetPartlyPartly
C7Every source linked (gate)PartlyPartlyMet
C8Recognised measure firstMetPartlyPartly
C9Own measures labelledMetMetMet
C10Version named and linkedPartlyPartlyPartly
C11Figure status shown to readers (gate)PartlyPartlyMet
C12Nothing second-handMetMetMet
C13Spread beside every median (gate)PartlyPartlyMet
C14Work and hard moneyMetMetMet
C15Like for like, then the reality lineMetPartlyPartly
C16Getting in and carrying kept apartMetMetMet
C17Real and hard money side by sideMetMetMet
C18One wage and twoMetPartlyPartly
C19Anchor yearsMetMetMet
C20Same definitions every timeN/AN/AN/A (first issue)
C21Counter-measure as sound as the headlineMetMetMet
C22Source line on every chart or panel (gate)PartlyMetMet
C23Honest construction (gate)MetMetMet
C24Every chart readable without sightMetPartlyPartly
C25Forecasts and illustrations kept apartMetMetMet
C26General information and interestMetMetMet
C27Corrections log (gate)PartlyPartlyMet
C28Reproducibility packNot metNot metMet
Total46 of 54 (85.2%)41 of 54 (75.9%)47 of 54 (87.0%)

Result now: pass. All seven gate criteria (C1, C7, C11, C13, C22, C23, C27) are met and the score is above 80%: 20 met, 7 partly, none not met, 1 not applicable. Seven criteria are still only partly met: C3 (one 2026 price is dated only in the notes), C6 (two terms undefined), C8 (my own measure comes before the recognised one in one table and one panel), C10 (the Standard is named but not yet published to link), C15 (panel 01’s headline sets 1986’s standard variable rate beside today’s rate actually paid), C18 (one share has no two-wage figure) and C24 (two panel figures are in prose, not a table). They are listed so you can weigh them yourself.

What changed between the two scores: status rows were added for the 1986 lending inputs and Sydney’s ranking (C11); every median now says where it is a median only (C13); the corrections log carries a time and time zone for every entry (C27); release dates were added to the source notes that lacked them and the broker pricing note now points here (C7); the downloads were rebuilt with a source line on every panel (C22); and this page was published (C28).

The script

This is the full calculation, also in the zipped download above. It needs Python 3.8 or later and nothing else. Save it as hwi_issue1_calc.py and run python3 hwi_issue1_calc.py.

# hwi_issue1_calc.py
# Rebuilds every calculated figure in "What it took to own a home: 1986 vs 2026"
# (What the Numbers Leave Out, issue 1, Healthy Wealthy Investor, 20 November 2026)
# from the figures read from source, then checks each result against the figure
# printed in the article, to the article's rounding.
#
# Run:  python3 hwi_issue1_calc.py
# Needs Python 3.8 or later and nothing else. Every input is typed in below with
# its source note number from the article. Change an input and re-run to test it.
#
# Licence: CC BY 4.0. Credit "Healthy Wealthy Investor, What the Numbers Leave Out".
# Data and working page:
# https://healthywealthyinvestor.com.au/what-it-took-to-own-a-home-data/

# ---------------------------------------------------------------- inputs
# Figures read from source (note number in brackets) or broker pricing (note 17).
SV_1986 = 15.5          # [1] RBA F5 standard variable, from April 1986, %
SV_2026 = 9.02          # [1] RBA F5 standard variable, October 2026, %
PAID_2026 = 6.9         # [17] broker pricing, November 2026, %
PAID_SE_2026 = 7.0      # [17] broker pricing, self-employed or alt-doc, %
BUFFER = 3.0            # [11] APRA serviceability buffer, percentage points
AWE_1986_WK = 427.20    # [4] ABS 6350.0 Table 3, November 1986, $ a week
AWE_2026_WK = 2083.70   # [5] ABS AWE, May 2026, $ a week
LEND_1986 = 9011.3e6    # [9] ABS 5609.0 Table 2, 1986-87 commitments, $
LOANS_1986 = 197492     # [9] ABS 5609.0 Table 2, 1986-87 number of loans
LOAN_2026 = 736000      # [10] ABS Lending Indicators, Sep quarter 2026, $
HOUSE_1986 = {"Sydney": 98325, "Melbourne": 82000, "Perth": 58000}           # [6]
HOUSE_2026 = {"Sydney": 1529308, "Melbourne": 936528, "Perth": 1073500}      # [7]
CPI_DEC86, CPI_JUN26, CPI_SEP26 = 30.85, 102.31, 103.64                      # [8]
INFL_1986, INFL_2026 = 9.7, 3.9   # [8] year-ended CPI, Dec qtr 1986, Sep qtr 2026
# [12] Perth Mint monthly AUD bid averages, Jan to Dec 1986
GOLD_1986_MONTHS = [489.46, 482.05, 486.52, 470.03, 467.24, 493.42,
                    545.54, 608.52, 664.80, 660.46, 611.35, 589.06]
SILVER_1986_MONTHS = [8.21, 7.90, 7.54, 6.93, 6.62, 7.10,
                      7.48, 7.96, 8.53, 8.32, 8.16, 7.72]
GOLD_2026, SILVER_2026 = 5511.99, 69.87   # [13] Perth Mint, October 2026 close
HH_INC_AUS_1986 = 22000                   # [22] ABS 2506.0, median household income
SYD_1991, NSW_1991 = 35544, 30445         # [23] ABS 2710.1, 1991 medians
NSW_SALES_2026 = {"q1": 1194000, "median": 1550000, "q3": 2275000}           # [20]
FT_MEDIAN_2025_WK = 1741                  # [19] ABS 6337.0, published full-time median
FT_MEDIAN_1986_WK = 369                   # [18] ABS 6310.0 Table 9, printed median

# [18] ABS 6310.0 (1986) Table 10: full-time persons aged 20 and over,
# thousands in each weekly pay band ($ lower, $ upper, count).
BANDS_1986 = [(0, 120, 43.0), (120, 160, 44.8), (160, 200, 75.1), (200, 240, 194.4),
              (240, 280, 436.5), (280, 320, 626.5), (320, 360, 593.1), (360, 400, 472.7),
              (400, 440, 378.5), (440, 480, 284.4), (480, 520, 269.1), (520, 560, 210.0),
              (560, 600, 140.7), (600, 640, 118.5), (640, 680, 69.7), (680, 720, 69.1),
              (720, 760, 47.8), (760, 800, 70.2), (800, None, 104.9)]
# [19] ABS 6337.0 Table 7 (Data 7), Australia, total employees, full-time persons,
# thousands in each $100 weekly pay band, from $0-99 up to $4,900-4,999, then $5,000+.
FT_1989 = [14.409, 210.521, 399.781, 1459.421, 895.415, 952.484, 377.022, 372.221,
           103.265, 66.476, 40.262, 30.847, 18.08, 12.087, 7.587, 4.352, 3.924, 5.239,
           3.486, 2.297, 3.207, 0.603, 1.648, 0.275, 0.417, 0.915, 0.16, 0.207, 0.18,
           0.15, 0, 0, 0, 0, 0, 0.351, 0.286, 0.074, 0, 0, 0.286, 0, 0.276, 0, 0, 0.124,
           0, 0, 0, 0, 0.456]
FT_2025 = [5.495, 8.794, 14.816, 15.504, 44.768, 60.15, 69.297, 135.38, 148.542, 303.17,
           405.824, 464.819, 468.424, 505.482, 387.443, 575.635, 351.148, 468.431,
           286.646, 344.856, 381.476, 221.975, 354.652, 202.733, 155.675, 245.806,
           123.397, 261.319, 99.331, 75.589, 138.914, 64.754, 153.555, 53.931, 62.04,
           74.826, 55.002, 35.846, 37.116, 35.983, 48.472, 27.682, 26.476, 22.516,
           19.141, 27.327, 16.988, 9.744, 112.557, 9.941, 230.656]

# ---------------------------------------------------------------- formulas
def annual_repayment(loan, rate_pct, years):
    """Principal and interest, monthly compounding. Yearly = 12 x monthly."""
    i, n = rate_pct / 100 / 12, years * 12
    return 12 * loan * i / (1 - (1 + i) ** -n)

def percentile(bands, share):
    """Linear interpolation inside the band where the share falls (the ABS method)."""
    total = sum(c for _, _, c in bands)
    target, cum = share * total, 0.0
    for lo, hi, c in bands:
        if cum + c >= target:
            return lo + (target - cum) / c * (hi - lo)
        cum += c

def hundred_bands(counts):
    return [(100 * k, 100 * (k + 1) if k < 50 else None, c) for k, c in enumerate(counts)]

# ---------------------------------------------------------------- working
pay86, pay26 = AWE_1986_WK * 52, AWE_2026_WK * 52
loan86 = LEND_1986 / LOANS_1986      # $45,629
loan86_run = 45600                   # the article runs the 1986 repayment on the rounded $45,600
rep = {
    "1986 at 15.5%": annual_repayment(loan86_run, SV_1986, 25),
    "2026 test at 9.9%": annual_repayment(LOAN_2026, PAID_2026 + BUFFER, 30),
    "2026 self-employed test at 10.0%": annual_repayment(LOAN_2026, PAID_SE_2026 + BUFFER, 30),
    "2026 at 9.02%": annual_repayment(LOAN_2026, SV_2026, 30),
    "2026 at 6.9%": annual_repayment(LOAN_2026, PAID_2026, 30),
}
cpi_to_sep26 = CPI_SEP26 / CPI_DEC86     # used for "what's left" (2026 dollars)
cpi_to_jun26 = CPI_JUN26 / CPI_DEC86     # used for real wage and real house change
gold86 = sum(GOLD_1986_MONTHS) / 12
silver86 = sum(SILVER_1986_MONTHS) / 12
nsw86 = 22056   # [24] interpolated from Table C28: 820,080 sits just above the
                # 818,409 households under $22,000, so the median is about $22,056.
syd86_income = nsw86 * SYD_1991 / NSW_1991
b86, b89, b25 = BANDS_1986, hundred_bands(FT_1989), hundred_bands(FT_2025)
p = {k: percentile(b, s) for k, (b, s) in {
    "86p10": (b86, .1), "86p25": (b86, .25), "86p50": (b86, .5), "86p75": (b86, .75),
    "86p90": (b86, .9), "89p10": (b89, .1), "89p90": (b89, .9), "25p10": (b25, .1),
    "25p25": (b25, .25), "25p50": (b25, .5), "25p75": (b25, .75), "25p90": (b25, .9)}.items()}
# Rounded to the dollar a week, as the article does, then x 52.
yr = {"86": [round(p["86p25"]) * 52, FT_MEDIAN_1986_WK * 52, round(p["86p75"]) * 52],
      "25": [round(p["25p25"]) * 52, FT_MEDIAN_2025_WK * 52, round(p["25p75"]) * 52]}
syd86, syd26 = HOUSE_1986["Sydney"], HOUSE_2026["Sydney"]
wk_rep86 = rep["1986 at 15.5%"] / 52

# (label, calculated value, figure printed in the article, rounding step)
checks = [
    ("1986 average loan, $", loan86, 45600, 100),
    ("1986 pay a year, $", pay86, 22214, 1),
    ("2026 pay a year, $", pay26, 108352, 1),
    ("1986 loan in years of one wage", loan86 / pay86, 2.1, .1),
    ("2026 loan in years of one wage", LOAN_2026 / pay26, 6.8, .1),
    ("1986 yearly repayment, $", rep["1986 at 15.5%"], 7222, 1),
    ("1986 share of one wage, %", 100 * rep["1986 at 15.5%"] / pay86, 32.5, .1),
    ("1986 share of two wages, %", 100 * rep["1986 at 15.5%"] / (2 * pay86), 16.3, .1),
    ("2026 test yearly repayment, $", rep["2026 test at 9.9%"], 76855, 1),
    ("2026 test, share of one wage, %", 100 * rep["2026 test at 9.9%"] / pay26, 70.9, .1),
    ("2026 test, share of two wages, %", 100 * rep["2026 test at 9.9%"] / (2 * pay26), 35.5, .1),
    ("2026 self-employed test, share of one wage, %",
     100 * rep["2026 self-employed test at 10.0%"] / pay26, 71.5, .1),
    ("2026 at 9.02% yearly, $", rep["2026 at 9.02%"], 71191, 1),
    ("2026 at 9.02%, share of one wage, %", 100 * rep["2026 at 9.02%"] / pay26, 65.7, .1),
    ("2026 at 9.02%, share of two wages, %", 100 * rep["2026 at 9.02%"] / (2 * pay26), 32.9, .1),
    ("2026 at 6.9% yearly, $", rep["2026 at 6.9%"], 58168, 1),
    ("2026 at 6.9%, share of one wage, %", 100 * rep["2026 at 6.9%"] / pay26, 53.7, .1),
    ("2026 at 6.9%, share of two wages, %", 100 * rep["2026 at 6.9%"] / (2 * pay26), 26.8, .1),
    ("Loan grew, times", LOAN_2026 / loan86, 16, 1),
    ("Pay grew, times", pay26 / pay86, 4.9, .1),
    ("1986 repayment, $ a week", wk_rep86, 139, 1),
    ("Left 1986, one wage, 2026 $ a week", (AWE_1986_WK - wk_rep86) * cpi_to_sep26, 969, 1),
    ("Left 1986, two wages, 2026 $ a week", (2 * AWE_1986_WK - wk_rep86) * cpi_to_sep26, 2404, 1),
    ("Left 2026 at 9.02%, one wage", AWE_2026_WK - rep["2026 at 9.02%"] / 52, 715, 1),
    ("Left 2026 at 9.02%, two wages", 2 * AWE_2026_WK - rep["2026 at 9.02%"] / 52, 2798, 1),
    ("Left 2026 at 6.9%, one wage", AWE_2026_WK - rep["2026 at 6.9%"] / 52, 965, 1),
    ("Left 2026 at 6.9%, two wages", 2 * AWE_2026_WK - rep["2026 at 6.9%"] / 52, 3049, 1),
    ("Prices rose (CPI, Dec qtr 1986 to Jun qtr 2026), times", cpi_to_jun26, 3.3, .1),
    ("Real wage rise, %", 100 * (pay26 / pay86 / cpi_to_jun26 - 1), 47, 1),
    ("Real wage rise a year, %", 100 * ((pay26 / pay86 / cpi_to_jun26) ** (1 / 39.5) - 1), 1, 1),
    ("Sydney house rose, times", syd26 / syd86, 15.6, .1),
    ("Sydney house rose after inflation, times", syd26 / syd86 / cpi_to_jun26, 4.7, .1),
    ("Median multiple 1986, Australian income", syd86 / HH_INC_AUS_1986, 4.5, .1),
    ("Sydney 1991 income above NSW, %", 100 * (SYD_1991 / NSW_1991 - 1), 17, 1),
    ("Sydney 1986 income estimate, $", syd86_income, 25750, 50),
    ("Median multiple 1986, Sydney estimate", syd86 / syd86_income, 3.8, .1),
]
for city, r86, r26 in [("Sydney", 4.4, 14.1), ("Melbourne", 3.7, 8.6), ("Perth", 2.6, 9.9)]:
    checks += [(f"{city} house, years of one wage, 1986", HOUSE_1986[city] / pay86, r86, .1),
               (f"{city} house, years of one wage, 2026", HOUSE_2026[city] / pay26, r26, .1)]
checks += [
    ("Years to save 20% Sydney deposit, 1986", .2 * syd86 / (.15 * pay86), 5.9, .1),
    ("Years to save 20% Sydney deposit, 2026", .2 * syd26 / (.15 * pay26), 18.8, .1),
    ("1986 P10 check against printed 246", p["86p10"], 246, 1),
    ("1986 P90 check against printed 619", p["86p90"], 619, 1),
    ("1986 P25 pay, $ a week", p["86p25"], 297, 1),
    ("1986 P75 pay, $ a week", p["86p75"], 486, 1),
    ("2025 P25 pay, $ a week", p["25p25"], 1293, 1),
    ("2025 P75 pay, $ a week", p["25p75"], 2472, 1),
    ("2025 median by this method (published 1,741; about 1% high)", p["25p50"], 1756, 1),
    ("1986 pay a year: lower, middle, higher ($15,400)", yr["86"][0], 15400, 100),
    ("($19,200)", yr["86"][1], 19200, 100), ("($25,300)", yr["86"][2], 25300, 100),
    ("2025 pay a year: lower, middle, higher ($67,200)", yr["25"][0], 67200, 100),
    ("($90,500)", yr["25"][1], 90500, 100), ("($128,500)", yr["25"][2], 128500, 100),
]
for k, (lab, r86, r26, rise) in enumerate([("Lower-paid", 6.4, 23.1, 3.6),
                                           ("Middle", 5.1, 17.1, 3.3),
                                           ("Higher-paid", 3.9, 12.1, 3.1)]):
    y86, y26 = syd86 / yr["86"][k], NSW_SALES_2026["median"] / yr["25"][k]
    checks += [(f"{lab} worker, Sydney house in years of pay, 1986", y86, r86, .1),
               (f"{lab} worker, years of pay, 2026", y26, r26, .1),
               (f"{lab} worker, rise, times", y26 / y86, rise, .1)]
checks += [
    ("Lower-quartile house / lower-paid worker, years", NSW_SALES_2026["q1"] / yr["25"][0], 18, 1),
    ("Upper-quartile house / higher-paid worker, years", NSW_SALES_2026["q3"] / yr["25"][2], 18, 1),
    ("Top tenth / bottom tenth pay, 1989", round(p["89p90"]) / round(p["89p10"]), 2.8, .1),
    ("Top tenth / bottom tenth pay, 2025", round(p["25p90"]) / round(p["25p10"]), 3.4, .1),
    ("NSW quartile, $ million (1.19)", NSW_SALES_2026["q1"] / 1e6, 1.19, .01),
    ("NSW quartile, $ million (2.28)", NSW_SALES_2026["q3"] / 1e6, 2.28, .01),
    ("Gold 1986 average, $ an ounce", gold86, 547.37, .01),
    ("Silver 1986 average, $ an ounce", silver86, 7.71, .01),
    ("Wage in gold, 1986, oz", pay86 / gold86, 40.6, .1),
    ("Wage in gold, 2026, oz", pay26 / GOLD_2026, 19.7, .1),
    ("Wage in silver, 1986, oz", pay86 / silver86, 2880, 10),
    ("Wage in silver, 2026, oz", pay26 / SILVER_2026, 1550, 10),
    ("Sydney house in gold, 1986, oz", syd86 / gold86, 180, 1),
    ("Sydney house in gold, 2026, oz", syd26 / GOLD_2026, 277, 1),
    ("Sydney house in gold, % more", 100 * ((syd26 / GOLD_2026) / (syd86 / gold86) - 1), 54, 1),
    ("Average loan in gold, 1986, oz", loan86 / gold86, 83, 1),
    ("Average loan in gold, 2026, oz", LOAN_2026 / GOLD_2026, 134, 1),
    ("Gold rose faster than consumer prices, times", (GOLD_2026 / gold86) / cpi_to_jun26, 3, 1),
    ("$100,000 debt after 10 years at 9.7%, $", 100000 / (1 + INFL_1986 / 100) ** 10, 39600, 100),
    ("$100,000 debt after 10 years at 3.9%, $", 100000 / (1 + INFL_2026 / 100) ** 10, 68200, 100),
    ("Real rate 1986, points", SV_1986 - INFL_1986, 5.8, .1),
    ("Real rate 2026 standard variable, points", SV_2026 - INFL_2026, 5.1, .1),
    ("Real rate 2026 at 6.9%, points", PAID_2026 - INFL_2026, 3.0, .1),
    ("Test rate, %", PAID_2026 + BUFFER, 9.9, .1),
    ("Self-employed test rate, %", PAID_SE_2026 + BUFFER, 10.0, .1),
]

if __name__ == "__main__":
    bad = 0
    for label, calc, printed, step in checks:
        ok = abs(calc - printed) <= step / 2 + 1e-9
        bad += not ok
        print(f"{'OK  ' if ok else 'DIFF'} {label}: calculated {calc:,.4f}, article {printed:,}")
    print(f"\n{len(checks) - bad} of {len(checks)} figures match the article to its rounding.")
    raise SystemExit(1 if bad else 0)

General information only. It does not take your personal situation into account. Juan Jeffery, Healthy Wealthy Investor.