The market claims that U.S. data is unreliable, yet reacts strongly to "better-than-expected" results, especially inflation data

Despite increasing doubts about the reliability of U.S. economic data, financial markets have shown an unusually strong reaction to any "better-than-expected" signals, especially inflation data.

This paradox is challenging traditional market analysis frameworks in the face of a data-dependent central bank. According to reports from the Wind Trading Desk, a recent report by Goldman Sachs analyst Jan Hatzius and his team indicates that the quality of several top U.S. official surveys, including non-farm employment and CPI, is being eroded due to a continuous decline in survey response rates.

However, the decline in data quality has not diminished its market impact. The report shows that the market's sensitivity to unexpected data has remained high since it sharply increased three years ago. This reaction is particularly pronounced in inflation data, with the bond market's sensitivity to inflation surprises still reaching 2.7 times the normal level, while the stock market's sensitivity is 1.4 times the normal level.

This decline in data quality has directly raised the standard error, expanding the confidence intervals of data point estimates. Goldman Sachs analysis shows that the current average standard error is 26% higher than the 2015-2019 period, with the standard error for job vacancy data increasing by 87,000, equivalent to a 90% confidence interval expanding to about 700,000 jobs; the monthly standard error for CPI is expected to rise to nearly twice the average level of 2015-2019.

Declining Response Rates, Deteriorating Data Quality

Goldman Sachs has found that the reliability of U.S. economic data is facing severe challenges, rooted in a structural decline in response rates for government statistical surveys.

The report points out that, compared to the average levels from 2015-2019, the response rate supporting the non-farm payroll report (NFP) has dropped by 18 percentage points, the unemployment rate survey has decreased by 16 percentage points, the CPI survey has fallen by 10 percentage points, while the response rate for the Job Openings and Labor Turnover Survey (JOLTS) has plummeted by as much as 30 percentage points.

The decline in response rates has directly led to two consequences: an increase in statistical error and a rise in the magnitude of data revisions.

First, the reduction in sample size has raised the "standard error," which is the expected difference between the sample estimate and the true population value. The report shows that among the ten government surveys examined, eight have seen an increase in standard error, averaging 26% higher than the 2015-2019 period. This issue is particularly severe in JOLTS and employment reports.

Goldman Sachs estimates that the recent decrease in the frequency of CPI price collection may push its monthly standard error close to twice the average level of 2015-2019.

Secondly, lower response rates may also lead to significant revisions of the initial published values of data in the future.

However, the report also notes that not all indicators are experiencing increased revision magnitudes. In recent years, the number of indicators that have undergone significant revisions (17) is almost equal to those that have seen smaller revisions (15). Some indicators' large revisions have other reasons, such as the revisions of initial jobless claims and the New York Fed manufacturing index, primarily aimed at correcting seasonal distortions introduced during the pandemic

"Bad" Data and "Strong" Reactions: Inflation Remains the Market's Most Sensitive Nerve

Despite the decline in data reliability, the market's reaction to unexpected data remains strong.

According to the report, the response intensity of the U.S. Treasury market to standardized growth surprises has fallen back to historical average levels, while the stock market's response is 1.5 times the normal level. Notably, the U.S. Treasury market's response to standardized inflation surprises has reached 2.7 times the normal level, while the stock market's response is 1.4 times.

This indicates that among the key variables determining the Federal Reserve's policy direction, the market places far greater importance on inflation signals than on other data.

This strong reaction in the financial market creates a core contradiction: if market participants generally believe that data quality is declining, theoretically, they should reduce the intensity of their reactions to single data "surprises." However, the reality is quite the opposite, especially regarding inflation issues, where the market's "knee-jerk reaction" has not only not weakened but has been significantly amplified.

The Logic Behind the Contradiction: High Uncertainty and the "Data-Dependent" Federal Reserve

The market's seemingly contradictory behavior is driven by two key factors.

First is the high uncertainty regarding economic prospects.

The report shows that the current dispersion of predictions for growth, inflation, and interest rates over the next year—i.e., the degree of disagreement among different forecasters—is, on average, about 35% higher than the levels seen in 2019.

Secondly, the Federal Reserve's "data-dependent" policy stance amplifies the importance of each data release.

The report states that because the Federal Reserve has explicitly linked future interest rate decisions to economic data, investors and traders have no choice but to closely monitor these indicators and attempt to interpret their potential impact on monetary policy ahead of others.

In this environment, even if the quality is questionable, any new incremental information will be viewed by the market as a valuable clue for judging future directions