The Rate–Premium Relationship Over Time
Rolling 15-year Pearson correlation between the year-end market premium and Q4 10-year Treasury yield · window ending year · 1973–2025. The rolling relationship was still strongly negative in 2021 (r = -0.79), weakened toward zero by 2023 (-0.15), and turned positive in 2024–2025 (0.46 latest). The flip is recent—and positive windows have appeared before.
What does it show?
The rolling relationship was still strongly negative in 2021 (r = -0.79), weakened toward zero by 2023 (-0.15), and turned positive in 2024–2025 (0.46 latest). The flip is recent—and positive windows have appeared before.
Methodology
Use calendar-year-end observations from Federal Reserve Financial Accounts table S.11.1.b for U.S. nonfinancial corporate business. Total equity at market value (LM103181105.Q) is compared with net worth, defined as assets minus non-equity liabilities (FL102090005.Q). The market premium is the arithmetic residual: total equity minus recorded net worth. Intellectual-property products already recorded at current cost (LM105013765.Q) are retained as context but remain inside recorded net worth. Values are end-of-period and not seasonally adjusted. For each ending year from 1973 onward, calculate the Pearson correlation between the two level series over the trailing 15 annual observations. The window length is fixed before inspecting sign changes. Correlation is descriptive and does not establish causality or a structural break. Limitations: This is the U.S. nonfinancial corporate sector, not the S&P 500; it includes a broader corporate and foreign-direct-investment equity universe. The residual is not a direct valuation of patents, brands, software, data, organizational capital, or other intangible assets. It also reflects expectations, risk premia, sentiment, and measurement differences. Recorded net worth already includes some intellectual-property products at current replacement cost, so 'recorded' is not synonymous with 'tangible.' Federal Reserve Financial Accounts history is revised; the source release is pinned so the prototype remains auditable. Interest rates and liquidity are economy-wide monetary conditions, not treatments assigned independently of corporate valuations; the comparisons are descriptive, not causal. The 10-year yield is the average of October, November, and December monthly observations; the market premium is measured at calendar year-end. M2/GDP is the reciprocal of FRED's M2 velocity series at Q4. The 2020 redefinition of M1 and M2 creates a monetary-aggregate break that must remain visible in interpretation. Rolling windows overlap by 14 of 15 observations, so adjacent points are highly dependent and must not be treated as independent tests. A different defensible window length would change the timing and magnitude of sign crossings; the chart is a sensitivity view, not a breakpoint estimator. Correlating levels remains vulnerable to common trends. Annual-change models are shown separately in the multivariate diagnostic. Each window holds 15 annual observations; at n = 15 the 95% interval on the latest r = 0.46 spans roughly -0.07 to +0.79, so the recent positive reading is not statistically distinguishable from zero.