Confidence indices are reported as economic indicators and measure something specific — what people say about their expectations.
What is asked
Questions about personal finances, the general economy and willingness to make major purchases, over past and future periods.
Which are combined into an index.
Question wording is stable over time, which is what allows comparison.
Index construction
Responses are converted to balance scores — positive minus negative — and combined.
Which produces a number without an intrinsic meaning.
The level matters only relative to its own history, which is why long-run averages are cited.
Sentiment versus behaviour
People report pessimism while continuing to spend, and the reverse.
Which has been notable in recent periods where sentiment and spending diverged substantially.
Explanations include the effect of news coverage on stated views and the difference between assessments of the economy and of personal circumstances.
Political effects
Responses correlate with political alignment relative to who holds office.
Which is documented and complicates interpretation.
Aggregate indices can shift after elections without any economic change.
Sample and method
Survey samples of a few thousand, weighted to represent the population.
Which carries sampling error and mode effects from how the survey is conducted.
Shifts from telephone to online collection have produced methodological discontinuities.
Sub-indices
Expectations and current conditions are reported separately.
Which is more informative than the headline, since they can move in opposite directions.
Predictive value
Modest and variable.
Which means treating these as forecasts of spending overstates what they do.
They are better understood as a measure of mood than as an economic prediction.
Reading them
Look at the trend, the sub-indices and the long-run average rather than a single month's level.
Business confidence
Parallel surveys of firms on output, orders and hiring intentions.
Which have a somewhat better track record as leading indicators.
Purchasing manager indices are the most widely watched of these.
Retail sales data
Actual spending rather than stated intention.
Which is the measure that matters and arrives with a lag.
Volume and value measures differ during periods of price change.
Savings ratios
The proportion of income not spent.
Which indicates precautionary behaviour.
Sharp movements in savings ratios have preceded and followed economic shocks.
Media effects
Coverage volume and tone correlate with sentiment measures.
Which is documented in research on economic news and confidence.
This complicates using sentiment as an independent indicator.
What to watch instead
Actual spending, savings behaviour and credit data are all published and measure behaviour rather than mood.
Major purchase intentions
Questions about whether now is a good time to buy durable goods.
Which has historically had somewhat better predictive value than general sentiment.
It is reported as a sub-index and rarely mentioned in coverage.
Income expectations
Whether people expect their own finances to improve or worsen.
Which relates more directly to spending decisions than views about the national economy.
The gap between personal and national assessments is a persistent feature of these surveys.
Comparability over time
Methodological changes create discontinuities in the series.
Which providers document and coverage generally ignores.
Comparing a current reading to one from before a methodology change requires care.
Multiple providers
Several organisations publish confidence measures using different methods.
Which sometimes diverge.
Divergence between them is itself informative about how much of the movement is methodological.
The honest summary
A measure of what people say about how they feel, useful in trend and weak as a forecast.
Why they are watched anyway
They are published quickly, monthly, and before harder data arrives.
Which gives them a timeliness advantage over spending statistics that lag by weeks.
Timeliness rather than accuracy explains their prominence in coverage.
Reading them properly
Trend rather than level, sub-indices rather than headline, and behaviour data rather than sentiment where a decision depends on it.
A closing observation
Sentiment and spending have diverged notably in recent years, with people reporting pessimism while continuing to spend.
That divergence is the most interesting thing these surveys have produced, and it is a reason to treat them as a measure of mood rather than as an economic forecast.
Where to find the underlying data
Providers publish the full series, the sub-indices and the methodology notes on their own sites, generally free.
Which allows checking whether a reported change is meaningful against the series' normal variation.
Monthly movements within the historical range of noise are reported as news with some regularity.
A practical alternative
Retail sales volumes, savings ratios and consumer credit data measure behaviour rather than stated feeling.
All three are published by statistical agencies and central banks and are the harder evidence on the same question.
A final note
Confidence indices are cheap to produce, quick to publish and easy to report, which explains their prominence better than their predictive record does.
Reading them alongside actual spending data gives a considerably fuller picture than either alone.