How the Canadian Vigie Index is calculated

In short

Four signals are tracked. Each gets a score out of 100 from a scale: Canadian history for the gap and cover, France’s scale for non-residents and ratings. The two “markets” signals are averaged, then the two “fundamentals” signals. The index is the average of these two parts. It is the same calculation as for France.

Index = ( Markets + Fundamentals ) ÷ 2

Markets      = ( Gap + Cover ) ÷ 2
Fundamentals = ( Non-residents + Ratings ) ÷ 2

Index = ¼·Gap + ¼·Cover + ¼·Non-residents + ¼·Ratings

Each signal counts for a quarter of the index. The Canadian index is read against Canada’s own history: a Canadian 40 does not compare with a French 40.

1. The four signals

Signal	What is measured	Source	Frequency
Gap	Canada’s 10-year yield (benchmark bond) minus the US 10-year yield, at the close, in basis points (100 bp = 1%), on days published on both sides only	Bank of Canada, Federal Reserve	every day since 2001; every month since 1982 for the benchmarks
Cover	Amount bid ÷ amount sold at government 10-year bond auctions (reopenings of older bonds set aside)	Bank of Canada	about every month since 2020, four to six times a year before
Non-residents	Federal bonds held by foreign investors ÷ all marketable government bonds, at month end	Statistics Canada	every month since 1995
Ratings	Lowest rating from S&P, Moody’s and Fitch, counted in notches lost from AAA	agencies, two matching sources	at each decision

There is no Canadian equivalent of TARGET2: Canada has its own currency. A fifth signal, pressure on the Canadian dollar, is under study and is not part of the index.

2. Each signal becomes a score out of 100

Each signal has five benchmarks: the score is 0 at the first, 25 at the second (watch), 50 at the third (stress), 75 at the fourth (crisis) and 100 at the last. Between two benchmarks, the score moves in a straight line, as for France. The benchmarks come from two rules.

Gap and cover: the percentiles of their history. The score is 0 at the 25th percentile (a better level was seen only one time in four), 25 at the median, 50 at the 75th percentile, 75 at the 90th and 100 at the 99th (a worse level was seen only one time in a hundred). For the gap, the history starts in 1982, in monthly averages (Statistics Canada, Federal Reserve): it includes Canada’s 1994-1995 debt crisis, when the gap was 150 to 200 bp. For cover, it starts in 2001. France has applied the same rule to its cover since 9 November 2026 (its auctions since 2004): the thresholds differ from one country to the other, the rule is the same.
Non-residents and ratings: France’s scale. The agencies’ scale is the same for every country. The risk of debt held abroad, sellers who can leave all at once, depends on the share held, not on the country. In Canada, this share has mostly followed demand for debt seen as safe: 33% during the 1995 crisis, 14% in 2009, 47% in 2026. Percentiles would have put it in “crisis” at its record.

Score	0	25 watch	50 stress	75 crisis	100
Gap	-36 bp	2 bp	86 bp	142 bp	218 bp
Cover	2.38	2.28	2.16	2.05	1.85
Non-residents	40%	50%	55%	60%	70%
Ratings	AAA	AA-	A+	BBB+	BBB-

For cover, a low value is bad: the scale goes down. Over the 2001-2026 sessions, the gap was in “calm” 73% of the time, in “watch” 24%, in “stress” 4% and never in “crisis”; the non-resident share never reached 50%. These benchmarks are fixed: they do not move with new data.

3. The two parts and the index

The “markets” part is the average of the gap and cover; the “fundamentals” part, that of non-residents and ratings. A part is calculated only when both its signals are up to date, never with one alone: a signal that is too old makes its part, and therefore the index, unavailable.

Signal	Kept for up to	Why
Gap	7 days	a long holiday weekend in Canada or the United States, plus a margin
Cover	5 months	at most 127 days between two 10-year auctions since 1998
Non-residents	4 months	published about two and a half months after month end
Ratings	no limit	a rating holds as long as it has not changed

With these durations, the index could have been calculated at every one of the 6,339 sessions since 2 January 2001.

4. Full example: 8 November 2026

Signal	Value	Date	Score
Gap	-117 bp	5 November 2026	0 (below the first benchmark, -36)
Cover	2.51	29 October 2026	0 (above the first benchmark, 2.38)
Non-residents	45.9%	end of August 2026	14.7 (between 40% and 50%: 25 × 5.88 ÷ 10)
Ratings	AA+ (Fitch)	since 24 June 2020	8.3 (1 notch lost: 25 × 1 ÷ 3)

Markets      = ( 0 + 0 ) ÷ 2       = 0
Fundamentals = ( 14.7 + 8.3 ) ÷ 2  = 11.5
Index        = ( 0 + 11.5 ) ÷ 2    = 5.8    →  calm

5. What is not in the index

The Bank of Canada’s financial stress index, published monthly, is shown next to the index for comparison: it measures strain across all Canadian markets (equities, currency, credit), not only on federal debt. It is converted to 100 by the percentile rule, over 2001-2026: 0.03; 0.06; 0.13; 0.20; 0.47. Its record, 0.55, dates from February 2009.

6. Known limits

The gap with the United States is not a pure risk premium. The two countries share neither a currency nor a central bank: the gap also follows their policy rates. Its high, +272 bp in April 1990, and its record since 2001, +134 bp on 9 October 2002, came mostly from monetary policy, not a debt crisis. It is kept because it is the only free daily market measure.
A high non-resident share can also reflect confidence. This is why it follows France’s scale rather than Canadian history: its 2026 record (46.9% in July) stays “calm”, below the 50% threshold.
Non-resident numerator and denominator, checked on 8 November 2026. Both count bonds only: no Treasury bills, no Crown corporations, no retail savings products. They differ on two points. Bonds held abroad are at book value (accrued interest included), the total at nominal value. And they include foreign-currency bonds, 2.8% of foreign holdings in August 2026 (7.8% at the end of 2015): without them, the share would be 44.6% instead of 45.9%. Before 1995, the series also counted repurchase agreements: it therefore starts in January 1995.
The Canadian index moves little. Over 2001-2026, it never went above 31 (on 18 December 2008, with its “markets” part at 61) and was in “watch” in only 4% of sessions. It does not go back to the 1995 crisis: 10-year bond auctions are only published from 1998.
The benchmarks were reviewed on 8 November 2026. The first version drew all four signals except ratings from 2001-2026 percentiles: the gap entered “crisis” from 43 bp, far from the 1995 crisis, and the non-resident record put the index in watch (27 on 8 November 2026, against 5.8 now).