Q3 in Review

 

In this note we set out to explain three core topics. It reviews what moved in the quarter and why, drawing on our decomposition of the Treasury sell-off. It sets out our current asset allocation signals across equities, bonds and cash. And it addresses two questions that clients have raised most often in recent months: whether the rise in US yields marks a durable shift, and whether the concentration of equity indices in a small number of large companies is a risk that should change how portfolios are built.
 
Our answers follow the four pillars we apply to every asset class: valuations, economics, financial conditions and sentiment. The conclusion that emerges is a constructive view on fixed income, particularly in South Africa, a cautious view on developed and emerging market equities, and no change to the systematic process that converts these signals into positions.
 
The quarter in review

Bonds carried the quarter's losses and equities absorbed them. The 10-year Treasury yield rose 82 bps between 30 June and 30 September, the largest quarterly increase in the benchmark since 2009, and the Bloomberg US Aggregate Bond Index returned about -3.5%. Long-dated bonds fell furthest. Credit outperformed government paper because the move was a rates move, not a credit move: spreads on investment grade and higher-quality high yield were broadly stable, with widening confined to CCC-rated issuers.
 
The driver was energy. Brent crude rose roughly 40% over the quarter to above $100 a barrel as the US-Iran conflict persisted, and core PCE inflation held at 3.0% in August, above the Federal Reserve's 2.0% objective for a sixty-fifth consecutive month. Unemployment at 4.1% sits below the FOMC's own longer-run estimate of 4.2%. Against that backdrop the September increase in the federal funds rate was the expected outcome rather than a surprise.

Equities diverged beneath a flat surface. The S&P 500 returned 2.3% and the Nasdaq Composite 2.6%, but the equal-weighted S&P 500 fell about 1.5% and US small caps fell 7.3%. Earnings did the work: consensus expects S&P 500 earnings growth of roughly 29% year on year for the third quarter, the third consecutive quarter above 25%, with semiconductor earnings growing at more than double that rate. Developed markets outside the US slipped 0.7% in dollar terms and emerging markets fell 2.4%, though both remain well ahead year to date.
 
South Africa followed its own sequence. The rand strengthened through R16.00 to the dollar in September for the first time in six months, supported by gold and platinum prices, a softer dollar and foreign demand for local bonds, with roughly R64 billion of inflows into the debt market this year. Headline CPI eased to 4.3% in July, yet the SARB raised the repo rate to 7.25% on 23 September, citing the oil price and the Federal Reserve's own move. Second-quarter GDP data showed the economy close to stalling, with consensus for a 0.1% quarterly contraction after six consecutive quarterly expansions.

The JSE All Share fell 6.7% in September as a correction in precious metals weighed on resources shares, with gold down 6.3%, platinum 4.5% and palladium 11.5% over the month. The benchmark R2035 government bond ended the quarter yielding around 8.6%, which, with CPI near 4.3%, implies a real yield above 4.0%. That spread between what the bond pays and what inflation takes is the measurement behind our fixed income view in the sections that follow.
 
Core topic 1: What the US Treasury sell-off is, and is not

The second Treasury sell-off of 2026 is a repricing of the Federal Reserve's policy path, not a loss of the inflation anchor. That distinction determines how long it lasts, and it is only visible once the yield is decomposed.
 
The yield on a 10-year nominal bond can be written as the sum of four parts: the average short-term real rate investors expect over the bond's life, the real term premium they demand for holding duration rather than rolling short-dated paper, the average inflation they expect, and an inflation risk premium. The first two sum to the real yield on an inflation-linked bond; the last two sum to breakeven inflation. Our bond team estimates each component daily using the term structure regression framework of Adrian, Crump and Moench (2013).
 

 
The result is unambiguous. Average expected short-term real rates supplied 56.8% of the rise and the real term premium a further 26.2%, so 83.0% of the move ran through real yields. Expected inflation over the decade rose by only 5 bps despite a 42% increase in the oil price. The market revised its view of how high the Federal Reserve will hold real rates, not its view of where inflation will settle.
 
The sequence within the quarter confirms the reading. In July the term premium led: its largest one-day rise, 12 bps, came on 29 July when the FOMC held rates with inflation above target and oil rising, which the market read as tolerance of inflation. In September the lead passed to expected real rates. A strong payroll report, a firm CPI print and the 16 September rate increase each lifted that component by about 10 bps, and on each of those days the term premium and breakeven inflation fell. A central bank seen to be leaning against inflation lowers the compensation investors need for inflation risk even as it raises the expected path of real rates.
 
The spring episode, examined in our July note, was different: it ran mainly through the real term premium, which is the more reactive and more reversible component. The two sell-offs of 2026 therefore have different half-lives.
 
Core topic 2: Index concentration is an allocation question, not a stock-picking problem

Equity indices are more concentrated than at any point in modern records, and this is not a reason to abandon capitalisation weighting. The evidence points the other way: the cost of fighting the index has been high, the index runs little additional risk for its concentration, and the risk that does exist is best managed at the asset allocation level.
 
The Herfindahl-Hirschman Index squares each constituent's weight and sums the result; its inverse gives the effective number of equally sized holdings a portfolio behaves like. On this measure the S&P 500 behaves like 46 stocks, the lowest in 55 years and below the dot-com trough. Nvidia's 8.2% peak weight in 2025 was the largest single-stock weight since records began in 1981. The ten largest constituents hold close to 40% of US large-cap value, double their share of 2010. The FTSE/JSE All Share carries 48.5% of its weight in ten names across 119 constituents, an effective breadth of about 35 stocks. But: The weight follows profit. The ten largest S&P 500 companies generate roughly a third of index earnings, double their share of the mid-1990s, and the seven largest technology companies accounted for 52% of index net income growth in 2024 and 30% of total capital expenditure in 2025.
 
Concentrated prices are, to a large degree, a reflection of concentrated fundamentals rather than a departure from them. The current quarter repeats the pattern: a 2.3% index return built on 29% earnings growth, led by the same companies that carry the weight.
 
The United States has run the controlled experiment. Equal weighting the S&P 500 is the purest form of down-weighting the winners, and over the ten years to September 2026 it returned 12.0% a year against 15.3% for the capitalisation-weighted index, a gap of 3.4 points a year that widens to 5.8 points over three years. The equal-weight version also carried about 4 percentage points more volatility since 2003 at a beta above one. Over 150 years of US data, periods of high concentration delivered marginally higher average returns, and rules that de-risked whenever concentration was high underperformed buy-and-hold by a wide margin. Concentration is not a timing signal. South Africa ran the same experiment with the Capped SWIX, created in 2016 to limit any single share to 10% at the height of Naspers' ascent. The cap cost roughly 4.9% a year through that period. No share is now near the cap, and capped and uncapped returns have converged. A permanent cap is a standing bet against whichever company succeeds next.
 
Decades of attribution research, from Brinson, Hood and Beebower onward, find that asset allocation policy explains roughly nine tenths of the variability of a diversified portfolio's return. Security selection, the level at which the concentration debate is conducted, explains under 5%. The Top 40 concentrates the All Share's 119 names into 41 at a cost of about one percentage point of annualised volatility and no difference in the decade's worst drawdown. Name count is a poor proxy for risk; correlation, factor exposure and currency carry the risk, and those are set at the allocation level.
 
Our response is to keep the equity beta whole and to earn excess return elsewhere. The capitalisation-weighted index is a rules-based active strategy with built-in momentum, continuous position trimming of losers and no manager risk. Portable alpha preserves that engine in full and adds excess return from fixed income, where it is structural and repeatable. The one concentration decision that matters is how much equity the portfolio holds in the first place, and that is a question for the allocation framework in the next section.
 
House view and asset allocation

Our allocation signals favour South African cash and bonds over global equities, with developed and emerging market equities at strong negative. Each signal is the output of the four-pillar scoring of valuations, economics, financial conditions and sentiment, updated systematically as data arrive. The table gives the current reading for the main asset classes.
 
The gradient runs from short to long in South Africa: cash is the strongest signal, the 20-year bond the strongest duration signal, and the belly is neutral. The curve is paying investors to hold either very little or a great deal of duration, but not the middle, where issuance pressure is heaviest.
The United States gives the reverse ordering. The front end scores positive because the repricing of expected real rates is largely complete on a Taylor rule basis; the long end is neutral because the term premium carries reversal risk in both directions.

Lastly, the equity signals are the most negative in the framework, and South Africa is the exception. Developed market valuations have expanded while real discount rates have risen, which is the combination that historically precedes weak forward returns. South African equities retain a valuation discount and a positive currency pillar, and the moderate positive signal reflects that, tempered by resources exposure to a precious-metals correction.
 
Themes for the quarter ahead

The fourth quarter opens with South African real yields above 4% at the 10-year point and a real cash rate above 2.5%, both at the upper end of their post-2010 ranges. The SARB's September increase was pre-emptive rather than reactive: headline CPI at 4.3 percent, just above the target, but the move protects the rand ahead of further Federal Reserve tightening. That combination, elevated real rates and a central bank with credibility to spare, is the setting in which bond carry is most reliably earned.
 
In the United States, the decomposition says most of the repricing is in the price. Expected real rates at 2.5% are consistent with a Federal Reserve that holds policy near the rule-consistent level, and the 0.9 point Taylor gap leaves limited room for further upward revision. We favour the front end, where the signal is moderate positive, and hold the long end at neutral: the term premium added 31 bps after 22 September and the two tightening precedents both saw it retrace most of its overshoot within three months of the yield peak.
 
In the equity market, concentration is not the risk, valuation is. Developed market equities enter the quarter with expanded valuations, a real discount rate that has risen 67 bps in three months, and earnings growth that has so far justified both. The signal is strong negative because the valuation and financial conditions pillars have moved against the asset class at the same time, not because of index structure. The earnings season beginning in October is the test: consensus expects the third consecutive quarter of growth above 25%, concentrated in semiconductors and the largest technology companies. The index will continue to own that growth in full. The question is what price has been paid for it.
 
Emerging markets face the same discount-rate headwind with the added cost of dollar funding and a September correction in precious metals that weighed directly on the JSE's resources weight. South African equities are the exception in our framework, at moderate positive, because the valuation discount remains and the currency pillar has turned supportive. Within equity portfolios our approach is unchanged: hold the index in full, do not cap or down-weight the winners, and source excess return from fixed income through portable alpha.
 
The quarter demonstrated the point the attribution literature has made for forty years. Equities and bonds responded to the same oil shock in opposite directions, and the return to a diversified portfolio was decided almost entirely by how much of each it held. Our balanced portfolios enter the fourth quarter tilted toward South African cash and bonds, underweight developed and emerging market equities, and neutral in global bonds, with the tilt sized by the strength of each signal rather than by conviction about any single event.
 
Two scenarios bound the quarter. If core inflation moderates and oil stabilises, the term premium retraces, long bonds rally and the equity signal improves as financial conditions ease; the framework would reduce the cash tilt and add duration first. If oil rises further and core inflation holds at 3%, expected real rates stay high, the equity signal stays negative and cash continues to earn a real return without drawdown. The portfolio is positioned to perform well in both and to adjust as the data choose between them.
 
Concluding observations

The third quarter rewarded investors who measured rather than reacted. A rise in the 10-year Treasury yield to a nineteen-year high read, on the surface, as a loss of control over inflation; decomposed, it was the opposite, a market pricing a central bank that will hold real rates high enough to keep expectations anchored. An index at record concentration read as fragility; measured, it was the mechanical result of concentrated earnings, and the alternatives to holding it cost between 3 and 6 percentage points a year over the past decade.
 
Our positioning follows from the measurement. We are constructive on fixed income, with the strongest signals in South African cash and the long end of the local curve, and moderate positive at the front of the US curve where the repricing is most complete. We are cautious on developed and emerging market equities on valuation and financial conditions, with South Africa the exception. We hold equity exposure in the capitalisation-weighted index without caps or down-weighting and earn excess return on top of it through fixed income.
 
None of these views depends on forecasting the oil price, the next FOMC decision or the next earnings print. They depend on the four pillars being scored consistently and the portfolio being moved in proportion to what the scores say. That process has carried our clients through the first three quarters of 2026, through two Treasury sell-offs, a 40% move in oil and a rotation that reversed the previous quarter's leadership, and it is the process that will set the allocation for the fourth.

Disclaimer:

Prescient Investment Management (Pty) Ltd is an authorised Financial Services Provider (FSP 612) in terms of the Financial Advisory and Intermediary Services Act, 2002 (FAIS).

The information in this document is provided for general information purposes only and is not intended to constitute financial advice (as defined in FAIS), investment advice, a recommendation, or an invitation/offer to issue, sell, subscribe for, or purchase any financial product. Any views or opinions expressed are those of the author (unless otherwise stated) and may change without notice. Past performance (if referenced) is not necessarily indicative of future performance, and no guarantee is given as to future returns. While reasonable care has been taken in preparing this document, no representation or warranty (express or implied) is made as to the accuracy, completeness, or fairness of the information, and Prescient Investment Management (Pty) Ltd and its affiliates disclaim liability for any loss, damage, cost, or expense (whether direct, indirect, or consequential) arising from reliance on this information. This document may contain proprietary material and is protected by copyright law. For more information visit www.prescient.co.za.