What is in the BlockPhi Global Liquidity Index?

What does a global liquidity measure contain?

What a global liquidity measure contains depends entirely on who built it. The term covers at least six incompatible measures, and each one produces a different series and a different answer.

The Bank for International Settlements has the strongest claim to the term and uses it for credit rather than for money. Its global liquidity indicators track credit to non-bank borrowers, cross-border bank loans and international debt securities, and it defines global liquidity as the ease of financing in global financial markets. That is a credit aggregate. It is not a money supply aggregate and it is not a measure of central bank balance sheets.

Most research in this market means something narrower: the sum of national broad money across a set of economies, converted into US dollars. The set varies by publisher. Lyn Alden and Sam Callahan use eight economies, the bgeometrics chart states twenty-one major central banks, and the arXiv preprint in the table uses eighteen. Different coverage produces a different series, so two publishers can measure the same idea honestly and still disagree.

A third meaning is US net liquidity, the Federal Reserve balance sheet less the Treasury General Account and reverse repurchase balances. It is daily, free and widely charted, and it is not global. A fourth is a proprietary composite. CrossBorder Capital's Global Liquidity Index measures a flow of financial capital across central bank liquidity, private sector bank credit and cross-border flows. Coinbase Institutional describes its own index as a blend of broad money across eight economies weighted for signal power rather than for size. BlockPhi's series belongs to that fourth group.

How is the BlockPhi Global Liquidity Index built?

As a composite of central bank balance sheets, private sector credit, cross-border capital flows and short-term credit spreads, read as a rate of change rather than as a level.

The series belongs to the composite family rather than to the broad money family. Its inputs are four monetary transmission channels, and each is in because it carries funding the other three do not see. They are taken in the order the money moves: central bank balance sheets, private sector credit, cross-border capital flows and short-term credit spreads.

The first channel is central bank balance sheets: quantitative easing by the Federal Reserve, the European Central Bank, the Bank of Japan and the People's Bank of China. This is where the impulse starts, and it is the channel most often mistaken for the whole measure. On its own it records what a central bank has created and nothing about what the rest of the system then does with it.

The second is private sector credit: commercial bank lending, which multiplies the base money central banks create. A central bank creates base money; the banking system decides how far it travels. A measure that stops at the central bank has read the first step of a two-step process and called it the total.

The third is cross-border capital flows: growth in international reserves, and the FX swap lines that carry US dollar liquidity beyond the United States. Dollar credit created outside the United States sits outside the Federal Reserve's balance sheet and outside any single country's money supply statistics, and the strength of the dollar governs how much of it reaches borrowers abroad. Hélène Rey's work on the global financial cycle, cited in full below, is the standard statement of why a measure built on one country's aggregates is incomplete.

The fourth is short-term credit spreads, taken as the price of liquidity rather than its quantity: tight spreads read as abundant funding, wide spreads as stress. It is the one that separates this series from a money aggregate, and it is the subject of the next section.

Wherever BlockPhi publishes a figure taken on the index, this sentence travels with it. Global Liquidity has no single standard definition, and this figure is measured on BlockPhi's own Global Liquidity Index, which is not directly comparable to global M2, to the BIS global liquidity indicators, to CrossBorder Capital's Global Liquidity Index, or to US net liquidity.

Why build a composite instead of using a money supply aggregate?

Because a money aggregate counts deposits, and most of the funding that moves markets is not a deposit. A composite reads the parts of the system a deposit count cannot see.

A money supply aggregate is a stock of deposits at banks in a set of countries. Modern funding is largely not that. A large share of lending is secured against collateral rather than funded out of deposits, a large share of dollar credit is created outside the United States, and a large share of short-term funding is intermediated by institutions that take no deposits at all. Each of those is liquidity under any working definition of the word, and none of them appears in a national money supply series.

That is the case for a composite. The cost belongs beside it. A composite cannot be rebuilt from a single public data table the way a money supply sum can, which makes it harder to reproduce, easier to get wrong, and impossible for a reader to check without the weights. What it buys is that no single channel dominates the reading. Whether that trade is worth making is a judgement, and this page states it as one rather than as a finding.

The credit-spread channel is the one that needs the most defence, because a spread is not a quantity of anything. That is precisely why it is in. The first three channels count how much money and credit exist. A spread reads how easily that funding can actually be obtained, and the two move apart: a system can hold a record stock of money at the moment the market for funding it seizes. Brunnermeier and Pedersen's work on market liquidity and funding liquidity, cited in full below, is the standard treatment of that interaction and is the footing for putting a price into a measure otherwise built from quantities.

It is also the most substantive reason a reading taken on this series need not agree with one taken on global broad money. Three of the four channels overlap with what a money aggregate counts. The fourth does not overlap with it at all.

Is the index used as a level or as a rate of change?

As a rate of change. Levels enter no model, because two series that both trend upward correlate on levels almost regardless of what happens between them.

Granger and Newbold established in 1974 that a regression between two trending series reports an apparently strong relationship between variables that are unrelated in the short run. Bitcoin and every global money and credit aggregate have trended upward across this sample, so a levels regression between them is the textbook case rather than an edge case. The transformation is therefore not a preference. It is the condition under which a regression is admissible at all.

The model is fitted on the weekly change in the series, not its level and not the six-week smoothed rate of change, which failed its stationarity test and is used in charts only.

Whether a transformation is admissible is decided by a test rather than by argument. The augmented Dickey-Fuller test asks whether a series drifts arbitrarily over time, and a series that does cannot be used in a regression without being transformed first. Bitcoin's weekly log return passes it comfortably. The weekly change in the index passes it at the ten per cent level rather than at the conventional five, and that is reported here rather than rounded away: central bank aggregates are volatile and heavily revised, and a marginal pass is a marginal pass.

Rate of change is also the harder test of the relationship itself, which is a separate point from admissibility. A levels correlation flatters the relationship and hides its behaviour between the endpoints. Rate of change produces the smaller number, and the smaller number is the honest one.

Which transformation of the index failed its stationarity test?

The six-week smoothed rate of change. It did not pass, so it appears in charts and in no model, and it is reported here rather than left out.

Smoothing a series makes it easier to look at and harder to use. A rolling average introduces autocorrelation that was not in the underlying data, and the augmented Dickey-Fuller test reads that induced persistence as drift. The six-week smoothed rate of change fails the test on those grounds, and not marginally. It is excluded from every model on this site.

It still appears in charts, because a smoothed line is easier to read than a weekly one and a chart is a picture rather than a regression. That is a legitimate use and a narrow one. Every figure BlockPhi publishes on the index is estimated on the weekly change, and none is estimated on the smoothed transformation.

This is published because the alternative is worse. A firm that reports only the transformations that passed is reporting a selection, and a reader has no way to judge how large the discarded set was. Here the discarded set is one transformation, it is named, and its test statistic is held to the same standard as every other figure on this site: it is published when the date the estimates were last re-run is published, and not before.

What is not published about the index?

The component weights, the country and institution coverage, the re-estimation date, and the current reading. The categories are published; the recipe is not.

Publishing the ingredient list does not reproduce the number. Rebuilding the index would need the weights, the source tables, the central bank and statistical feeds behind them, and the vintage and revision handling that decides what the index knew on any past date. The weights carry the construction, so the weights stay unpublished.

The estimation sample is published, to the year. The series is estimated on approximately four years of weekly observations running from 2022 to 2026, which BlockPhi's two source documents agree on independently of each other. The exact first and last observation dates are not published, because neither document states them, and a day invented to fill the gap would make the window look checkable while leaving it unverifiable.

The date the estimates were last re-run is not published, and it is the largest remaining gap between what BlockPhi discloses and what the rest of the field discloses, because every competing figure in the tables on this site carries a date range. Until that date exists, no figure measured on the index renders in this research estate. That is an arrangement rather than an omission: every study in the published record that tested the stability of a liquidity lead found that the lead moves, so an undated figure is not a weaker version of a dated one but a different and unfalsifiable claim.

Beyond the live headline level on the home page, the current reading of the index, its direction and anything derived from them are not published openly on this domain; the gated Monitor carries them for members. That is a product boundary and it is also a disclosure one. A current reading is the part of this work that sits closest to a recommendation, and the openly published layer is deliberately everything but that.

What are the limitations of this work?

Four limitations that BlockPhi states about its own work, and one the external record adds. They are published in full, because a limitation stated by the author is worth more than one found by a reader.

The index is proprietary, and that is a limitation rather than a disclosure preference. Nobody outside BlockPhi can rebuild it, so nobody outside BlockPhi can independently check a figure estimated on it. Every result BlockPhi publishes on the index inherits that, and a reader is entitled to weigh it against results taken on measures that are public.

The sample is short. It runs to approximately four years of weekly observations, which is about two Bitcoin cycles, and two cycles is a small number on which to establish a cyclical relationship. A relationship that holds across two cycles has not been tested against a third.

The model is linear and Bitcoin is not. Bitcoin's crash dynamics are faster than its rally dynamics, so a specification carrying one coefficient in both directions will describe the tails less well than the centre. A threshold extension, which fits different coefficients in different states, could improve the tail modelling. It has not been fitted, and until it is, the tails are the part of this work with the weakest support.

The lag may shorten. Bitcoin's institutional depth has grown across the sample, and an asset reachable through the same accounts that hold Treasury bonds sits fewer intermediary steps away from central bank liquidity than one that is not. If that continues, the transmission window compresses, and a lag fitted on this sample would overstate the lag that follows it.

One further limitation, which BlockPhi's own documents do not state and the external record does: every independent study in the published reconciliation that tested the stability of a liquidity lead found that the lead moves. That is not a criticism of any single estimate, including this one. It is the reason a lead figure means little without the date it was last re-estimated attached to it.

What methodological literature does this rest on?

Five works: the framework, two on the channels that need an external footing, and two on the statistics that decide how the index may be used.

The works below are cited by BlockPhi's own working paper on Global Liquidity transmission and are listed here with full bibliographic detail so that a reader can go and check them. Each identifier was resolved through Crossref on the date in the caption rather than reproduced from the paper's reference list, which carries no identifiers.

They are the methods and the framework this construction uses. None of them is evidence for a figure BlockPhi publishes, and none is cited here as corroboration of one.