Why more information can make macroeconomic expectations less accurate: Global evidence from 47 countries

Households do not simply choose how much macroeconomic information to acquire; they choose among sources that differ in representativeness and credibility. Using a survey of 46,285 consumers in 47 countries, this column shows that reliance on shopping, utility bills, and other local signals is associated with larger macroeconomic perception and expectation errors. Households that use official reports, newspapers, and television for information make smaller errors. Crucially, distrust of governments and central banks helps explain why more representative aggregate information is dismissed.

Inflation has fallen substantially from its post-pandemic peaks, but household expectations remain elevated relative to both current inflation and professional forecasts (BIS 2026). At the same time, central banks face scarce public attention and fragile institutional trust, making effective communication increasingly important (Ehrmann et al. 2022, van der Cruijsen et al. 2015). Standard models assume that more information produces more accurate beliefs. But what if the information people encounter most is precisely what makes their beliefs less accurate?

In a new paper, we provide the first homogeneous global evidence on how consumers form macroeconomic perceptions and expectations. We show that households often rely on vivid, personally relevant local signals, from shopping trips and utility bills to conversations with family and friends, that can be systematically unrepresentative of the aggregate economy. More information can therefore produce not more accurate, but more systematically biased, beliefs (D’Acunto and Weber 2026).

A global laboratory for belief formation

We designed a homogeneous survey of 46,285 consumers in 47 countries, fielded in April and May 2023. The countries account for roughly 90% of world GDP and span economies with very different inflation histories, institutions, and media systems. Respondents reported their perceptions of current inflation and their expectations for inflation, house prices, and interest rates over the following 12 months. We can compare their inflation beliefs with realised outcomes.

We also asked which sources respondents considered important for forming economic beliefs. We classify official government and central-bank reports, television, and newspapers as aggregate sources. Utility bills, shopping, acquaintances, work-related interactions, and social media are local sources. The distinction concerns the signal, not the medium: aggregate sources are designed to describe the economy-wide distribution, whereas local sources expose households to selected prices from limited portions of their consumption baskets. In fact, some of these signals are even excluded from core measures of inflation that central banks around the world monitor and target due to their volatility.

Local signals are universal, but not representative

Figure 1 maps inflation perception and expectation errors across the 47 countries in our sample. Upward bias is a global phenomenon: average inflation perceptions and expectations exceed subsequently observed inflation in every country, although the size of the error varies substantially within and across regions. This evidence extends earlier findings that household expectations are upward-biased and dispersed (Weber et al. 2022, D’Acunto et al. 2022).

Figure 1 Global inflation perception and expectation errors

Notes: Errors are ratios of reported to realised inflation; a value of one denotes no error. Darker shading indicates larger positive errors. The survey covers 46,285 consumers in 47 countries in April-May 2023. Source: D’Acunto and Weber (2026), Figure 1.

Figure 2 shows that these biases are closely linked to the information sources. More than 80% of respondents identify utility bills and shopping as important inputs. About 70% report using official sources, but access to official information does not crowd out local signals. Consumers typically combine several sources, and the most commonly used sources are also those most closely tied to salient personal prices. This pattern is consistent with evidence that grocery-price exposure shapes inflation expectations (D’Acunto et al. 2021) and that supermarket prices influence learning about inflation (Cavallo et al. 2017).

Figure 2 Aggregate and local sources used to form inflation beliefs

Notes: Bars show the share of respondents who consider each source ‘very important’ or ‘somewhat important’. Blue bars denote aggregate sources; red bars denote local sources. Source: D’Acunto and Weber (2026), Figure 2.

The source of information matters

Figure 3 summarises our central result: information acquisition does not always improve accuracy, contrary to standard rational inattention frameworks. Within countries, and conditioning on a broad set of demographics, respondents who use official reports have inflation expectations more than two percentage points lower than otherwise similar respondents who do not consider official reports important. Associations for television and newspapers are smaller but point in the same direction. By contrast, those who rely on shopping and utility bills report inflation expectations almost two percentage points higher. The same pattern holds for inflation perceptions: local sources are associated with perceiving higher inflation than is realised in aggregate.

The ordering is similar for expected house-price growth and interest-rate changes: aggregate sources are associated with lower expectations, local sources with higher ones. The pattern also holds across most countries. Moreover, among consumers using the same local source, more frequent exposure to grocery prices or gas stations is associated with larger inflation expectation errors. More exposure to an unrepresentative signal can therefore amplify, rather than attenuate, bias.

These are conditional correlations, not causal effects of assigning consumers to information environments. Yet their consistency across outcomes, countries, and both the extensive and intensive margins is difficult to reconcile with models in which information sources provide unbiased signals that differ only in precision.

Figure 3 Information sources and inflation and house-price expectations

Notes: Bars report estimated differences in 12-month expectations associated with use of each source, controlling for demographics and country fixed effects; whiskers show 95% confidence intervals. Blue bars denote aggregate sources; red bars denote local sources. Estimates are correlational. Source: D’Acunto and Weber (2026), left and middle panels of Figure 4.

Demographics and sorting across sources

Previous research documents expectation gaps by gender, education, income, financial literacy, and shopping responsibility (Conrad et al. 2022, D’Acunto and Weber 2024). Our global evidence suggests a unifying interpretation. Women, main grocery shoppers, frequent gas purchasers, and more impulsive consumers are more likely to use local sources. College-educated, financially literate, numerate, and employed respondents are more likely to use aggregate sources. Once we condition on source use, demographic differences in expectation errors become substantially smaller. Demographics appear to matter primarily because people sort into different information environments, not because demographic groups systematically process the same signal differently.

Institutional trust is key for information selection

The results raise a puzzle: why do households rely on local information that requires acquiring, recalling, and processing hundreds of signals when aggregate statistics are readily available at virtually no monetary cost, for example, by simply googling ‘inflation’? And why don’t they learn over time that their perceptions are systematically wrong, either from realisations or from the consequences of their economic choices? We find that trust in the institutions that produce aggregate information (governments, statistical agencies, and central banks) is key to resolving this apparent contradiction with standard rational inattention frameworks. Consumers who distrust these institutions discount their data and turn to information they can observe directly. They also fail to learn from macroeconomic realisations because they do not regard the reported numbers as truthful representations of reality.

Across every country for which the relevant questions are available, distrust in governments and central banks is negatively associated with using official reports to form macroeconomic perceptions and expectations. As Figure 4 shows, greater distrust is also associated with larger inflation perception and expectation errors. These relationships survive controls for trust in private economic institutions (e.g. banks and the stock market), political alignment, demographics, and country fixed effects, suggesting they do not simply capture generalised distrust or opposition to incumbent governments.

Historical economic instability also matters. Across countries, distrust is related to inflation volatility over the previous decade, but not systematically to its average level. A history of unstable outcomes may erode trust in economic institutions and their ability to manage the economy, and, in turn, the credibility of the official information they produce, even after economic conditions return to normal.

Figure 4 Inflation errors rise with distrust in public economic institutions

Notes: Plots show binned averages of inflation perception errors (left) and expectation errors (right) over a composite measure of distrust in the government and central bank. Higher ratios indicate larger upward errors. Source: D’Acunto and Weber (2026), top panels of Figure 9.

From publishing information to building credibility

The policy implication is that central banks’ communication cannot be limited to reports, website information, or aggregate messages on social media. Information about inflation targets and policy can shift household expectations (Coibion et al. 2022), and recent work shows gains from communication designed for non-experts (Ehrmann et al. 2025). But communication policy must also recognise that households choose among signals that differ in both objective representativeness and perceived credibility.

First, credibility is a policy input. Statistical integrity, consistent explanations, and transparent accounting for forecast errors can determine whether households use official information at all. Simpler language and more accessible charts cannot substitute for trust. Second, communication should bridge aggregate statistics and households’ local economic environments. Explaining why food or energy prices may rise faster than the overall consumption basket, while reminding households that less frequently purchased goods and services may experience lower and less volatile inflation, could make aggregate figures more relevant than simply reporting the headline number. Third, segmentation should reflect information environments, not merely demographics. Consumers who rely on grocery prices, utility bills, or social networks may require different messages and intermediaries that feel closer and more credible than “experts, from organisations with acronyms, saying that they know what is best and getting it consistently wrong”.

The broader lesson is that models and policy analysis should distinguish between the amount of information, its quality (its representativeness of the aggregate distribution of a macroeconomic variable), and trust in its producer. When local signals are salient and official sources are discounted, more information can produce more confident but less accurate beliefs. Effective communication therefore requires not only reaching the public, as earlier work and VoxEU columns have stressed (Coibion et al. 2019, D’Acunto et al. 2024, Meeks and Monti 2024), but also influencing which information the public finds credible enough to use.

Source : VOXeu