Skip to content

Stress signals · Q2 2026 · Jul 2026

Stress signals from funding and trade together

Seven signals, each a money quantity from money-flow set against a trade quantity from trade-flow, scored every quarter for the 36 board economies with nothing read from later periods. Each one fires at a threshold; the count says how many fired. They are shown beside money-flow’s , and every one carries its test against past stress.

No composite is published. Under rule G19 a signal joins a composite only if it beats the better of its two single-side inputs, with the 90% interval of the gain above zero. None of the seven did on this build’s funding episodes, so the page shows the seven signals beside V and the count, as the plan said it would.

How each signal is made

net reserves ÷ (12-month imports ÷ 12), months

The oldest reserve-adequacy rule: three months of imports. Lower is worse.

Money input
Net reserves (IRFCL Section I plus the predetermined drains of Section II, as money-flow's A2 uses) money-flow board res_net
Trade input
12-month imports trade-flow series/W TOTAL m
Threshold
≤ 3 months · Threshold by convention
single-side inputs as tested
Net reserves, % of GDP · 12-month imports, % of GDP

(short-term external debt + current-account deficit) ÷ 12-month exports

The financing the economy needs in the next year, set against what it earns abroad in a year: can it earn its way out?

Money input
Short-term claims on the economy (BIS, board st) and the current-account deficit (board CA × GDP) money-flow board st
Trade input
12-month exports trade-flow series/W TOTAL x
Threshold
≥ 50 % of exports · Threshold chosen by validation
single-side inputs as tested
Financing need, % of GDP · 12-month exports, % of GDP
Threshold candidates (training quarters)
CandidateHit rateFalse rateHit − false
15012%15%−3.8
12515%19%−3.5
10015%23%−7.9
7539%30%+8.4
5058%42%+15.5

fires when F < 0 and export momentum is at or below its own 20th percentile

Funding leaving in the same quarter that exports fall, the two-axis at-risk test generalised.

Money input
Funding flow score F (money-flow board), own-history percentile money-flow board flow
Trade input
Export momentum (12-month exports against a year earlier), own-history percentile trade-flow series/W TOTAL x
Threshold
Fires by definition · Fires by definition
single-side inputs as tested
Funding flow F, inverted own percentile · Export momentum, inverted own percentile

Σ partners’ V weighted by their share of exports

Crises travel along trade links (Glick and Rose 1999; Kaminsky and Reinhart 2000): buyers in trouble before anything shows at home.

Money input
Partners' vulnerability score V (money-flow board); the United States counts as V = 0 money-flow board v
Trade input
Destination shares of 12-month exports (trade-flow snapshots) trade-flow snap/W/TOTAL w12
Threshold
≥ 50.85 V points · Threshold chosen by validation
single-side inputs as tested
Average V of the other board economies, unweighted · Concentration of export destinations (HHI)
Threshold candidates (training quarters)
CandidateHit rateFalse rateHit − false
50.85 (P90)28%5%+22.5
49.71 (P85)28%9%+18.5
48.61 (P80)28%14%+14.0
47.76 (P75)28%18%+9.6

Σ min(lender share, destination share)

When the same partner lends and buys, a shock to it arrives through both channels at once.

Money input
Creditor shares of bank claims on the economy (BIS LBS, by quarter) money-flow links bis.LBS
Trade input
Destination shares of 12-month exports trade-flow snap/W/TOTAL w12
Threshold
≥ 43.5 % overlap · Threshold chosen by validation
single-side inputs as tested
Concentration of bank lenders (HHI) · Concentration of export destinations (HHI)
Threshold candidates (training quarters)
CandidateHit rateFalse rateHit − false
43.5 (P90)4%13%−8.6
38.54 (P85)8%18%−10.5
33.64 (P80)12%24%−12.4
30.41 (P75)12%30%−18.0

−(tercile slope of export growth on the dollar, two quarters on) × 4-quarter dollar change

Bruno and Shin (BIS WP 819): after a dollar rise, exporters reliant on dollar credit lose credit and exports.

Money input
FX credit A1 (money-flow board) for the tercile; the broad dollar index (money-flow signals) money-flow board a1
Trade input
Export growth two quarters after each dollar move (the slope) trade-flow series/W TOTAL x
Threshold
≥ 4 % of exports · Threshold chosen by validation
single-side inputs as tested
A1 × rise of the dollar index · Export momentum, inverted own percentile
Threshold candidates (training quarters)
CandidateHit rateFalse rateHit − false
462%40%+21.6
362%47%+15.0
262%50%+11.3
162%62%−0.8

commodity share of exports × buffer weakness ÷ 100

Commodity exporters with thin buffers meet a terms-of-trade fall and a funding stop together.

Money input
Buffer half of V (net IIP and current account, money-flow board buf) money-flow board buf
Trade input
Commodity share of exports, HS chapters 25–27 (trade-flow annual) trade-flow nodes/A HS-25..27
Threshold
≥ 9.1 points · Threshold chosen by validation
single-side inputs as tested
Buffer weakness (100 − buffer score) · Commodity share of exports
Threshold candidates (training quarters)
CandidateHit rateFalse rateHit − false
19.67 (P90)19%9%+10.4
14.78 (P85)35%12%+22.5
10.98 (P80)50%16%+34.3
9.1 (P75)62%19%+43.1

Rules

  • G6 · A trade indicator enters a published score only after validation says so.
  • G7 · The validation is shown with the score: positives and calm quarters printed.
  • G19 · A joint signal enters the composite only if it adds skill over the better single-side input, with a bootstrap interval that excludes zero; otherwise it is shown as context. If nothing passes, the page ships the signals beside V without a composite.

Point in time: a quarter uses its own board row; the 12-month trade window ending in its last month (or the newest complete one up to six months earlier, marked partial); BIS bank positions of the quarter; the commodity share of the year before; own-history percentiles and scales from earlier quarters; level ranks over every board economy-quarter up to it; the dollar slope from points whose outcome is not later than it.

One change after the first run: the dollar slope (J6) is fitted on export growth clipped to ±40 points. Pooled growth reached +883% at the 99th percentile where an economy’s reporting coverage jumps, and a handful of such points set the slope.

Coverage

Quarters scored per signal and economy, and the first quarter scored. J1 is not scored for euro members and reserve-currency issuers (as A2 is not); J4 needs half of exports to go to covered partners.

EconomyJ1J2J3J4J5J6J7
Argentina82 200682 200670 200927 200682 200674 200886 2005
Australia82 200682 200670 200954 200682 200674 200886 2005
Austria080 200662 200980 200680 200674 200886 2005
Belgium082 200670 200982 200682 200674 200886 2005
Brazil82 200682 200670 200957 200682 200674 200886 2005
Canada082 200670 200982 200682 200674 200886 2005
Chile81 200681 200670 200955 200681 200673 200883 2005
China34 201734 201724 201934 201734 201738 201738 2017
Colombia80 200680 200668 200980 200680 200672 200884 2005
Czechia79 200679 200669 200979 200679 200673 200880 2005
Germany082 200670 200982 200682 200674 200886 2005
Denmark82 200682 200670 200982 200682 200674 200886 2005
Spain082 200670 200982 200682 200674 200886 2005
Finland082 200670 200962 200682 200674 200886 2005
France082 200668 200982 200682 200674 200886 2005
UK082 200670 200956 200682 200674 200886 2005
Greece082 200670 200919 200682 200674 200886 2005
Hungary64 200671 200653 200971 200671 200673 200885 2005
Indonesia66 200866 200849 201058 200866 200873 200873 2008
India68 200968 200960 20116 200968 200968 200968 2009
Israel80 200680 200660 200973 200680 200674 200886 2005
Italy082 200670 200982 200682 200674 200886 2005
Japan082 200670 200976 200682 200674 200886 2005
Korea79 200679 200670 200938 201779 200674 200879 2006
Mexico82 200682 200670 200982 200682 200674 200886 2005
Malaysia80 200680 200662 200955 200680 200674 200886 2005
Norway82 200682 200670 200982 200682 200674 200886 2005
New Zealand82 200682 200670 200964 200682 200674 200886 2005
Poland81 200681 200669 200981 200681 200673 200885 2005
Portugal082 200670 200982 200682 200674 200886 2005
Russia75 200675 200650 200969 200675 200674 200880 2006
Saudi Arabia40 201540 201525 202026 201840 201542 201542 2015
Sweden82 200682 200670 200982 200682 200674 200886 2005
Thailand82 200682 200669 200957 200682 200674 200886 2005
Türkiye82 200682 200668 200914 200682 200674 200886 2005
South Africa82 200682 200669 200957 200682 200674 200886 2005

References

  • Glick, R. and A. K. Rose (1999), “Contagion and trade: why are currency crises regional?”, Journal of International Money and Finance.
  • Kaminsky, G. L. and C. M. Reinhart (2000), “On crises, contagion, and confusion”, Journal of International Economics.
  • Bruno, V. and H. S. Shin, “Dollar and exports”, BIS Working Paper 819.
  • IMF, “Assessing Reserve Adequacy” (2011, 2016): listed as a reference, not drawn (it would repeat A2).
  • money-flow, vulnerability method memos 1.3 and 1.4 (V, F, A1, A2, the episode set). money-flow →

The method memo with every number of this build: docs/methodology/stress-v1.0.md in the repository.

money-flow board (BIS LBS, CBS and DSS; IMF IIP, BOP and IRFCL), BIS LBS by creditor, the Federal Reserve’s broad dollar index; UN Comtrade 12-month windows and annual HS chapters via trade-flow. Calculations: datasteg global-network.

Sources

Board quarters to Q2 2026; trade windows to Jul 2026. Calculations: datasteg global-network (scripts/stress.py).