The Mengerian Stress Index: From Spec to Live Dashboard

The third essay of this series proposed a quantifiable extension of Menger's saleability spectrum. The proposal was that every transformation between a directly-held physical asset and its derivative paper substitutes imposes a measurable saleability haircut, that the cumulative haircut is invisible in normal conditions but observable as spreads under stress, and that those spreads — tracked across five specific market proxies — together constitute a Mengerian dashboard that would lead conventional crisis indicators by several weeks.

The proposal was theoretical. This essay is its public-facing engineering pass: precise definitions for each of the five proxies, the composite Mengerian Stress Index (MSI) that aggregates them, and the marketability half-life concept that turns instantaneous readings into a regime-classification tool. The live MSI itself runs at /toolkit/mengerian-stress-index/, updated continuously. The specific weights, calibration windows, and operational details that drive that live index are not described here — they sit behind the dashboard, where they belong, and will be refined as more crisis data arrives.

Menger gave us the concept. Fekete gave us the first instance for a single market. The work of generalization has been done abstractly across the rest of this series. What follows is the public definition layer of the live framework.

Premise: what the index measures, and what it does not

Before specifying the components, it is worth restating precisely what the MSI is designed to capture and what it deliberately ignores.

The MSI measures the spread between paper substitutes and their underlying physical or near-physical referents, aggregated across multiple asset classes, normalized to a stress-period baseline. It is a measure of substitute-layer integrity, not a measure of liquidity, not a measure of volatility, and not a measure of credit risk in any conventional sense. A market can have low volatility, narrow bid-ask spreads, abundant liquidity, and stable credit conditions, while the MSI is rising — because the index is measuring a different variable than any of those.

The relationship between the MSI and conventional risk metrics is asymmetric. When the MSI rises, conventional risk metrics typically remain calm for a period (often weeks to months) before catching up. When the MSI is calm, conventional risk metrics may be elevated for reasons unrelated to substitute-layer integrity — geopolitical tension, idiosyncratic credit events, sentiment-driven volatility — without indicating any structural saleability stress. The MSI is, in this sense, a complement to standard risk dashboards rather than a substitute for them.

The framework's claim is that the MSI's signature is causally upstream of major financial crises in the post-1971 monetary regime. The 2008 crisis showed visible signatures in the gold basis, the FX basis, and the ETF NAV-discount data weeks before the equity market broke. The 2020 COVID liquidity event showed the same pattern in a compressed timeframe. The 2023 banking stress (SVB, First Republic, Credit Suisse) was preceded by repo-haircut dispersion widening over the prior six months. None of this is observable in retrospect to anyone who was not specifically tracking these proxies; it is fully observable to anyone who was. The MSI is the formalization of "what would have led these crises if anyone had been watching the right variables."

Component 1: the paper-physical premium in precious metals (PPP)

The first proxy is the most direct successor to Fekete's original gold basis concept, generalized to include silver and to use spot-versus-physical retail data rather than spot-versus-futures.

Definition. The paper-physical premium is the spread between the spot market price of a precious metal and the delivered retail price of physical bullion at major institutional dealers, expressed as a percentage of the spot price.

Formula.

$\text{PPP}_t = \frac{P^{\text{phys}}_t - P^{\text{spot}}_t}{P^{\text{spot}}_t} \times 100\%$

where $P^{\text{phys}}_t$ is the median delivered price across a panel of major bullion dealers for a standard one-ounce sovereign coin, and $P^{\text{spot}}_t$ is the corresponding LBMA fix or COMEX continuous-contract reference. Silver is computed identically against its own retail and spot references.

Normalization. PPP is normalized to a rolling-baseline Z-score; what counts as a soft, hard, or regime-shift signal is calibrated empirically against the historical record and lives inside the live dashboard. The pre-2008 baseline runs in the low single digits for gold and roughly two to three times that for silver, reflecting fabrication and dealer markup costs that are essentially fixed in normal conditions.

Current readings, baselines, and component charts: /toolkit/paper-physical-premium/.

Component 2: the on-the-run / off-the-run Treasury spread (OTROFF) — planned

The second proxy measures the saleability differential between the most recently-issued Treasury at a key maturity (the on-the-run benchmark) and the immediately preceding issue (the off-the-run), which carries near-identical credit and duration characteristics. Any observable yield differential between the two reflects the market's pricing of saleability rather than any other factor.

Definition. The OTROFF spread is the yield difference between the on-the-run 10-year Treasury note and the immediately preceding off-the-run 10-year, both adjusted to identical maturity through standard term-structure interpolation.

Formula.

$\text{OTROFF}_t = y^{\text{off}}_t - y^{\text{on}}_t \quad \text{(in basis points, maturity-adjusted)}$

The maturity adjustment is necessary because the off-the-run is typically a few months shorter in remaining maturity than the on-the-run; the adjustment uses the slope of the local term structure to project both onto a common maturity reference. A companion metric — the ratio of on-the-run daily volume to off-the-run daily volume — confirms that any observed spread reflects saleability rather than other factors, since under stress, flow concentrates dramatically in the most-current issues.

Status. OTROFF is in scope for the framework but not yet live in the dashboard. The bottleneck is data: there is no freely-available per-CUSIP secondary-market yield feed, and the bench-quality OTROFF reading requires inputs that currently sit behind paid fixed-income subscriptions. The component placeholder lives at /toolkit/otroff-spread/ and will be wired into the composite MSI once a viable data path is secured.

Component 3: repo haircut dispersion (RHD)

The third proxy is the most analytically valuable and the most operationally difficult to measure. It captures the distribution of haircuts that dealers demand against different classes of collateral in tri-party repo markets. In calm conditions, haircuts on similar-quality collateral cluster tightly. Under stress, the distribution widens dramatically as dealers price differential saleability into otherwise-similar collateral.

Definition. The RHD is the cross-sectional standard deviation of haircuts across asset classes within a given quality tier in tri-party repo, computed at the highest frequency the available data supports.

Formula. For a set of $n$ asset classes within a quality tier (e.g., AAA-rated long-duration paper):

$\text{RHD}_t = \sqrt{\frac{1}{n-1} \sum_{i=1}^{n} (h_{i,t} - \bar{h}_t)^2}$

where $h_{i,t}$ is the median haircut for asset class $i$ on day $t$ and $\bar{h}_t$ is the cross-sectional mean.

Normalization. Pre-2008 RHD across investment-grade tiers ran in a tight band; the September 2008 reading widened by roughly an order of magnitude. The live component normalizes against this historical reference and against a rolling recent baseline. The composition of the dispersion is often more informative than the level: when haircuts on agency MBS widen relative to comparable Treasuries, that asymmetry is a duration-stress signature consistent with the diagnosis advanced in Article 8.

Current reading and dispersion decomposition: /toolkit/repo-haircut-dispersion/.

Component 4: FX cross-currency basis (CCB)

The fourth proxy measures deviations from covered interest parity (CIP) — the no-arbitrage condition that should tie spot exchange rates, forward exchange rates, and interest rate differentials together. Persistent deviations indicate that one party's synthetic dollar funding (assembled through derivatives) costs differently than its direct dollar funding (obtained in cash markets), which is a saleability signal in the dollar's own payment infrastructure.

Definition. The cross-currency basis at maturity $\tau$ for currency pair (USD, X) is the spread that must be added to the non-dollar floating leg of an FX swap to make it equivalent to direct dollar borrowing.

Formula. Following the standard literature:

$b^{\tau}_t = r^{X,\tau}_t - r^{\text{USD},\tau}_t - \frac{1}{\tau}\left(\ln F^{\tau}_t - \ln S_t\right)$

where $r^{X,\tau}$ and $r^{\text{USD},\tau}$ are the relevant interest rates at maturity $\tau$ in currency $X$ and USD, $F^{\tau}$ is the forward exchange rate, and $S$ is the spot rate. A negative basis indicates that synthetic dollar borrowing is more expensive than direct — a signal of dollar scarcity in the swap market.

Normalization. The pre-2008 baseline was within a few basis points of zero, consistent with CIP holding. Post-2015 the baseline shifted to a structurally negative regime that the live component treats as the new normal, with stress detected against that shifted baseline rather than against zero. The dashboard computes the basis across major USD pairs at short maturities; the dispersion across pairs is often more informative than any single reading.

Current readings across pairs: /toolkit/cross-currency-basis/.

Component 5: ETF NAV deviation in stress windows (ENV)

The fifth proxy measures the spread between an exchange-traded fund's market price and its calculated net asset value (NAV) during periods of underlying-market stress. Under normal conditions, the creation/redemption arbitrage mechanism keeps ETF prices within a few basis points of NAV. Under stress — especially for ETFs whose underlyings become difficult to trade — the spread can widen substantially, and the depth and duration of those discounts directly measure how the ETF wrapper's marketability has decayed relative to its constituents.

Definition. The ENV is computed for a panel of stress-relevant ETFs spanning investment-grade credit, high-yield credit, agency MBS, emerging-market debt, leveraged loans, and a Treasury control, as the absolute discount of market price to NAV.

Formula.

$\text{ENV}^{i}_t = \frac{|P^{\text{mkt},i}_t - \text{NAV}^i_t|}{\text{NAV}^i_t} \times 100\%$

with $i$ indexing the panel and $\text{NAV}^i_t$ taken at the official close. Intraday discounts can be computed against the indicative NAV (iNAV) for additional resolution when the underlying market is open.

Normalization. Normal-period absolute discounts are tight across the panel. The interesting signal is differential: when credit-quality-sensitive ETFs trade meaningfully wider than government-paper controls, that asymmetry is a substitute-layer stress signature even when the panel aggregate looks calm.

Current panel readings: /toolkit/etf-nav-deviation/.

The composite Mengerian Stress Index

The components above are aggregated into a single composite index through a weighted Z-score combination.

General form.

$\text{MSI}_t = \sum_{k} w_k \cdot Z^{(k)}_t$

with the components being PPP, RHD, CCB, ENV — and OTROFF, once its data path is solved.

The weights are not equal. They are calibrated empirically from the historical signal-quality of each component during the stress episodes for which we have full data (2008, 2011, 2020, 2023). The framework treats the weight vector as a tunable internal parameter that should evolve as more data accumulates, not as a published constant; the live dashboard carries the current calibration. Conceptually, components with cleaner lead-time relative to subsequent crisis events carry more weight than components whose signal-to-noise is dominated by idiosyncratic events.

Current MSI reading and component breakdown: /toolkit/mengerian-stress-index/.

Marketability half-life: regime classification

The instantaneous MSI reading is one signal. A more diagnostic signal is the marketability half-life — the time required for an MSI elevation to compress back to half its peak deviation from the pre-stress baseline. This metric distinguishes between two different classes of substitute-layer stress that look similar in their initial readings.

Definition. Given an MSI peak of $\text{MSI}^*$ achieved at time $t^*$ and a pre-stress baseline of $\text{MSI}^0$, the marketability half-life $\tau_{1/2}$ is the time required for the MSI to return to:

$\text{MSI}_{t^* + \tau_{1/2}} = \frac{\text{MSI}^* + \text{MSI}^0}{2}$

A short half-life (weeks) indicates a transient liquidity event whose substrate properties were not actually impaired. A long half-life (multiple quarters) indicates structural impairment that monetary intervention has compressed but not resolved. An infinite or indeterminate half-life — where the MSI does not return to half-peak before a new stress event begins — indicates a permanent regime shift in substitute-layer integrity.

Historical pattern. Across the four documented post-2008 episodes, the half-life trajectory tells a consistent story: each successive episode has a shorter half-life than the previous one, but the baseline MSI level steps up between episodes. The framework's interpretation is that the post-2008 monetary regime has produced progressively shorter MSI half-lives because central-bank intervention has become more aggressive and more rapid — but the half-life compression is purchased at the cost of permanently elevated baseline stress. The mathematical limit of that trajectory is a vanishing half-life with a permanently elevated baseline: the system flickers in and out of acute stress on shorter and shorter timescales while never returning to pre-2008 normalcy.

The single most important reading the dashboard provides is therefore not the instantaneous MSI but the baseline drift — the rolling mean of the MSI across non-acute periods. That drift is the framework's quantification of the secular saleability decay in the dollar substitute layer, decoupled from any specific event-driven stress.

Current half-life decomposition: /toolkit/marketability-half-life/.

What the dashboard reveals

A live dashboard implementing the above definitions, run continuously, produces several observations that are difficult to surface from any single conventional risk metric.

The structural baseline is elevated and has been drifting. The MSI baseline is not a function of any specific crisis; the drift is the framework's quantification of the gradual saleability decay across the dollar substitute layer described qualitatively in the previous essays of this series. The drift is observable in multiple components simultaneously, which is the signature of a regime shift rather than a component-specific phenomenon.

The components are not perfectly correlated. PPP and CCB tend to move together; OTROFF and ENV (where available) tend to move together; RHD tends to lead the others by a meaningful interval. This decomposition is informative because it allows the dashboard to distinguish between equity-market-driven stress (which manifests first in OTROFF and ENV), currency-driven stress (PPP and CCB), and funding-market-driven stress (RHD). The framework's most consistent prediction is that the next major event will be funding-market-driven, and RHD is the component to watch for early warning of that scenario.

The cross-asset signal is consistent when the diagnosis is right. When the framework's diagnosis is correct, the components elevate together (potentially with characteristic phase lags). When the diagnosis is wrong — when stress is actually idiosyncratic rather than systemic — the components elevate independently. Synchronized elevation across the panel is the configuration the framework treats as confirmation that the underlying phenomenon is substitute-layer decay rather than any specific market-segment dysfunction.

Limitations and what's still uncertain

The framework is honest about what the MSI does and does not capture, and where the live implementation has known weaknesses.

The component weights are empirically estimated from four crisis episodes. This is a small sample. The weights may not generalize to crisis configurations not represented in the historical record. In particular, a cryptographic-substrate failure of the kind described in Article 6 would manifest in components the historical record does not reflect, and the MSI as currently specified might miss it. The weights should — and will — evolve as more episodes accumulate.

The RHD component is operationally weak for non-institutional implementations. The best repo-market data is behind paid subscriptions. Public approximations (Fed monthly releases, FHLB advances) are reasonable but lagged and aggregated. A genuinely real-time RHD requires institutional access. Future versions of the framework should explore on-chain repo equivalents (which are publicly observable) as a substitute or supplement.

OTROFF is not yet live. Per-CUSIP secondary-market yield data is paid-only, and the framework will not publish numbers it cannot defend with reproducible inputs. The component re-enters the composite when a viable data path arrives.

The dashboard does not capture saleability stress in instruments without paper-physical analogs. Equity markets, for example, do not have a "physical share" analog in the same sense that gold has bullion. The framework's claims about saleability still apply to equity, but the proxies above will not detect equity-specific stress until it propagates into the credit-market or funding-market components.

Thresholds are reference points, not laws. The threshold levels embedded in the live dashboard are calibrated against historical episodes whose underlying conditions may not exactly recur. The framework treats the MSI as a trajectory indicator rather than a level indicator: the rate of change and the persistence of elevations matter more than whether a specific threshold is crossed.

The framework's predictions are not falsifiable in any clean Popperian sense. The MSI predicts that elevated readings precede crisis events, but the timing and specific trigger of those events are not predicted. A skeptic can always argue that an elevated MSI was a false positive that simply preceded a crisis happening for unrelated reasons. The framework's response is that across multiple historical episodes, the MSI lead-time is consistent enough that the alternative explanation (that all of these events happened to coincide with elevated MSI readings by chance) is implausible. But the framework does not claim more than this.

The dashboard is the artifact

The previous essays of this series produced the conceptual apparatus — Menger's spectrum, Fekete's basis, the decay function, the substitute-layer diagnosis. This essay defines the public face of the instrument that operationalizes that apparatus. The instrument itself — the calibrated weights, the alert thresholds, the half-life decompositions — lives at /toolkit/mengerian-stress-index/ and the component pages linked above, updated continuously and refined as more data accumulates.

What the framework treats as proprietary is not the concept — Menger and Fekete have a century's claim on that — but the calibration: which weights, against which baselines, with which thresholds, produce a useful early-warning signal in 2026 conditions rather than 2008 conditions. That calibration is the work product, and the dashboard is where it lives.

---

With this essay the series has now produced both the diagnostic apparatus (the framework, the proxies, the live dashboard) and the constructive proposals (the housing trilogy, the on-chain assessment). Future essays will return to specific applications as conditions warrant — the next event in the global financial system, the next regulatory development, the next AI-mediated structural change. The framework exists to make sense of those events as they arrive. The dashboard exists to read their early signals. The work continues.