1Q26_ Quarterly Outlook Report_Final_EN - Flipbook - Page 114
T H E P LUMB LI N E | A RETU RN TO F I RS T PRI N CI PL ES
As far as our intra-asset class positioning is concerned, we are making a tactical asset allocation
shift from sovereign debt to domestic equities (Canada). Upside risk to government bond yields
has increased as governments globally embark on ambitious fiscal spending programs,
particularly in infrastructure and defense, which are likely to widen deficits and increase bond
supply, placing downward pressure on global sovereign debt valuations. This adjustment brings
us to a lesser underweight in Canadian equities where prior rate cuts should continue to support
household consumption, growth is expected to remain positive, and government fiscal initiatives
aimed at infrastructure and defense spending provide additional support for economic activity
and corporate earnings. Within fixed income, this shift results in a reduction in sovereign debt
exposure (though still close to neutral), while we maintain a slight overweight to investment‑grade
corporate credit at the expense of high yield credit.
While not pessimistic by any means, our optimism on risk assets is measured relative to the
forecasting community (hence our neutral posture). Bottom-up estimates as at December 31
point to the S&P 500 finishing the 2026 calendar year at 7,944, implying a calendar year price
return of 18% in USD terms. Historically, year-ahead consensus forecasts, led by 12-months in the
chart below, tend to align with the index’s actual value quite well.
Bottom-up consensus forecasts point to another strong year for equities ahead
10000
S&P 500 index value
8000
6000
4000
2000
0
2005
2008
2011
2014
12-mth fwd forecasted index value
2017
2020
2023
2026
Index value (actual)
Bloomberg Finance LP, Scotia Wealth Management
However, the chart masks underlying forecast errors. Historically, the median forecasting error
was -1.7%, implying that forecasts have been slightly optimistic relative to the realized index
value. While this bias appears modest, a more informative measure of forecasting performance
is the median absolute forecasting error. Taking the absolute value removes the effect of positive
and negative errors offsetting one another, allowing for a clearer assessment of the typical
magnitude of forecast deviations regardless of direction. The median absolute forecasting error
is roughly 7%, indicating that the consensus forecast tends to deviate somewhat from the actual
outcome.
A look at the distribution of median errors shows that errors tend to be fairly modest in a +/- 5%
range about 40% of the time; widening that band to +/- 10% captures two-thirds of all available
113
