Lesson 5 of 8 · 15 min
Accuracy, reliability and efficacy of indicators
Evaluate how far GDP, the CPI and the labour force measures can be trusted in a past scenario: their accuracy, their reliability and how well each one measures the objective it is used for.
What you will be able to do
- Distinguish the accuracy, reliability and efficacy of an indicator.
- Explain the main limitations of real GDP, the CPI and the unemployment and participation rates.
- Evaluate an indicator in a past scenario, such as the labour market in 2020, using other indicators as evidence.
In April 2020, as the pandemic restrictions took hold, the unemployment rate rose from 5.2% to 6.3%. That is a big monthly rise, but it badly understated what happened: hours worked fell by over 10% in a month. To use an indicator well you have to know what it measures, how it is measured, and when it misleads.
Definition
Three tests for an indicator
- Accuracy: how close the published figure is to the true value. Sampling error, measurement problems and later revisions all limit accuracy.
- Reliability: how consistently the indicator is measured: the same definitions and methods over time and across places, so that one period can be compared with another.
- Efficacy: how well the indicator measures the objective it is used to judge. The unemployment rate is an accurate count of people who meet the ABS definition, but is that definition a good measure of "full employment"?
Percentage change from March to April 2020, seasonally adjusted. Hours: monthly hours worked. Source: ABS Labour Force via RBA table H5.
The figure shows why the unemployment rate understated the shock. Two definitions did it:
- Employed means at least one hour of paid work in the survey week, or being away from a job you still have. Hundreds of thousands of people stood down with zero hours, many of them kept attached to their employer by the JobKeeper Payment, were counted as employed, so the fall in hours was far bigger than the fall in employment.
- Unemployed means actively looking for work and available to start. Many people who lost jobs could not search while businesses were closed, so they left the labour force instead: the participation rate fell from 65.5% in March to 62.2% in May. They became hidden unemployment, invisible in the rate.
Which indicator gave the most accurate picture of the fall in labour use between March and April 2020?
| Indicator | Accuracy and reliability | Efficacy |
|---|---|---|
| Real GDP | Estimated from many surveys and tax data, and revised for years afterwards. Real GDP growth through the year to the June quarter 2025 was first reported as 1.8%; the ABS figure is now 1.9%. | Measures market output, not living standards: it leaves out unpaid work, leisure, the environment and how income is shared. Total GDP grows with population: real GDP per person was lower than a year earlier in every quarter from June 2023 to March 2025, while total real GDP grew. |
| CPI (headline inflation) | A fixed basket priced each period; quality changes and new goods are hard to price. The index is rebased from time to time: inflation to the December quarter 2022 was first published as 7.8% and reads 7.9% on today's index. | Overstates the cost of living when households switch to cheaper substitutes (substitution bias). One-off and volatile prices (fuel, fruit, government rebates) move it; the trimmed mean shows the underlying trend the RBA targets. |
| Unemployment rate | From a monthly sample survey of households, so it has sampling error: a monthly move of 0.1 or 0.2 points may be noise. Seasonally adjusted figures are revised as new months arrive. | Counts anyone with one hour of work as employed and leaves out discouraged job seekers, so it misses underemployment and hidden unemployment. The underutilisation rate (unemployment plus underemployment) is a wider measure of spare labour. |
| Participation rate | From the same survey, with the same sampling error. | Falls as the population ages even when the labour market is strong, so a fall is not always a sign of weakness. Rose to records near 67% in 2023–25. |
Match each criticism to the test of an indicator it is about.
Common mistake
Saying an indicator is "wrong"
The ABS's figures are not wrong: they measure exactly what they define. The problem comes when the definition does not match the objective you are judging. Say what the indicator leaves out and which other indicator fills the gap (hours worked, underemployment, the trimmed mean, GDP per person).
Exam tip
Evaluate with a second source
The syllabus asks you to evaluate indicators "in a past scenario". The strongest move is to test one indicator against another from the same period: "The unemployment rate rose 1.1 points in April 2020, but hours worked fell 10.4% and participation fell 2.2 points, so the rate understated the fall in labour use."
Summary
Accuracy: closeness to the true value (sampling error, revisions). Reliability: consistent measurement over time. Efficacy: how well the indicator measures the objective.
Real GDP is revised and leaves out living standards, unpaid work and the environment; per person figures can tell a different story.
Headline CPI is moved by volatile and one-off prices and has substitution bias; the trimmed mean shows the underlying trend.
The unemployment rate misses underemployment and hidden unemployment; in 2020 hours worked and participation showed the true fall in labour use.
Judge an indicator by testing it against other indicators from the same period.
Finished the lesson?
Try the 2 quick checks above and the lesson completes itself, or mark it complete now.