Portfolio rebalancing: the tolerance band at Norway's pension fund

In 2018 Norway's sovereign pension fund was worth about NOK 8,500bn, more than twice its value six years earlier, and the Ministry of Finance asked its managers whether the rule for bringing the fund back to its target equity share needed changing; the answer was a public letter (Norges Bank Investment Management, 2018). An owner holding a reserve of a few billion dong in shares and bonds meets the same question in miniature: shares have risen and the mix has drifted from plan, so how far must it drift before you sell some? The Norwegian fund answers with a number, and it set out how it weighed the trade-off.

What does the 4-point rule say?

The full name is the Government Pension Fund Global, managed by Norges Bank Investment Management (NBIM) under a mandate from the Ministry of Finance. The Ministry sets a benchmark index of global equities and bonds with a strategic target for the equity share, 70 per cent at the end of 2025 (Norges Bank Investment Management, 2018; Norges Bank Investment Management, 2025). Because equity and bond prices move differently, the equity share of the index rises when equities outperform bonds and falls when they do not (Norges Bank Investment Management, 2018).

According to Norges Bank's letter, the Ministry sets how far the share may deviate from target without rebalancing, and issues detailed rules on how it is brought back. The bank's job is to manage the portfolio under those rules (Norges Bank Investment Management, 2018). The rule in force when the letter was written (2018) was to rebalance if the equity share of the index deviated by more than 4 percentage points from target, in place since 2012; since then the target equity share had been raised from 60 to 70 per cent (Norges Bank Investment Management, 2018). That is threshold-based rebalancing rather than calendar-based.

One caution on the numbers. The 2018 letter is the library's document, and the rule has since changed. NBIM's benchmark-index page, which I read on 3 November 2026, says rebalancing takes place when the equity share of the actual benchmark index differs by more than 2 percentage points from the strategic weight on the last trading day of the month. I could not find the date of the change in the library's documents. The library's case file on Norway's pension fund gives 4 points and should be read with this note.

Does rebalancing earn extra return?

The fund's manager does not think so, at least for this kind of simple rule. The 2018 letter recalls that in 2012 the Ministry gave two aims: stop the composition drifting too far from the strategic allocation, and possibly raise returns by selling when prices are high and buying when low, weighed against transaction costs (Norges Bank Investment Management, 2018). NBIM tested the second idea: would a rule based on realised returns hold, on average, a higher equity share than target when expected returns are high and a lower one when they are low? Its conclusion was that this is unlikely, so other considerations should take priority when designing the rule (Norges Bank Investment Management, 2018). To exploit variation in the equity risk premium it would need indicators beyond realised returns, and the bank would have to be given more responsibility for the level of equity risk, which changes the division of duties with the Ministry (Norges Bank Investment Management, 2018).

This is worth remembering. Many people picture rebalancing as a disciplined way to "buy low, sell high". The manager of a fund worth thousands of billions of kroner says other considerations should take priority when designing the rule, chiefly keeping the composition from straying far from the strategic allocation, and states that it is not discussing whether to rebalance, only how the rule should be designed (Norges Bank Investment Management, 2018). The 70 per cent equity target is a risk choice in the sense of Modern portfolio theory and the efficient frontier: a point on the risk and expected-return frontier that the owner can bear. Rebalancing stops the portfolio drifting to another point (that link is mine, not NBIM's).

Wide or narrow band, fast or slow selling?

Each choice has a price, and the 2018 letter shows both directions using the fund's historical data. A wider band before triggering lowers transaction costs but lets the equity share stray further from target; a narrower band does the reverse. Spreading trades over several months in smaller amounts lowers costs at any band width, but prolongs the deviation (Norges Bank Investment Management, 2018). The main finding: a narrower band combined with more gradual adjustment can reduce both transaction costs and deviation compared with the rule then running; rebalancing would then occur more often and over longer periods, becoming part of ordinary portfolio management (Norges Bank Investment Management, 2018).

Their recommendation was more gradual adjustment and a narrower band, which could be set at ±2 percentage points (Norges Bank Investment Management, 2018). On disclosure, the band width should be part of the public mandate, while how and how fast the allocation is brought back would stay unpublished, set by the Ministry after consulting the bank (Norges Bank Investment Management, 2018). My understanding of the reason is that a published band tells the public when the fund must trade, while the method stays private to avoid being front-run; the letter does not say so directly, so that is my guess.

How does the fund's size make the problem heavier?

The larger the fund, the dearer each trigger. The 2018 letter says the fund grew from about NOK 3,400bn to about NOK 8,500bn since 2012, so much larger trades are needed whenever the rule runs; since the financial crisis it had found large trades more challenging to carry out without moving prices, and corporate bonds in the index present a number of challenges for rebalancing (Norges Bank Investment Management, 2018). My multiplication: 4 percentage points of NOK 8,500bn is NOK 340bn, if the fund is brought back exactly to target.

The 2025 report shows the new scale: NOK 21,268bn at year-end, made up of 71.3 per cent equities, 26.5 per cent fixed income, 1.7 per cent unlisted real estate and 0.4 per cent unlisted renewable infrastructure, against a strategic benchmark of 70 per cent equities and 30 per cent bonds (Norges Bank Investment Management, 2025). The fund warns that with a fund this large and about 70 per cent in equities, it must be prepared for a significant fall in value; expected absolute volatility is 10.9 per cent, meaning that in one year out of three its value can swing by more than NOK 2,300bn (Norges Bank Investment Management, 2025). If a 2-point band applied, one trade from the band edge back to target on NOK 21,268bn would be about NOK 425bn (my multiplication), a very large amount; the 2018 letter ties larger trades to the fund's growth and recommends more gradual rebalancing to cut costs (Norges Bank Investment Management, 2018).

What can a small business owner use?

The three questions the fund must answer apply to a company's reserve too: how wide a band, how fast to sell, and who decides. A hypothetical example, every number mine. A company holds a reserve of VND 10bn, targeting 70 per cent in a share fund and 30 per cent in bonds. The share fund rises 20 per cent while bonds stay flat, so the total becomes VND 11.4bn with VND 8.4bn in shares, or 73.7 per cent. Assume a dealing cost of 0.2 per cent of the amount traded.

If shares keep rising, the ±4 band triggers later and sells more in one go; the ±2 band sells earlier in several smaller steps. That is exactly the trade-off in Norges Bank's letter: a narrow band keeps the mix closer to target, a wide one saves fees (Norges Bank Investment Management, 2018). The right choice depends on how far off-target the company can tolerate being when shares fall, and what dealing costs really are.

This fits if you decided your risk level in advance and want a rule that spares an argument when markets are hot. It fits poorly if the reserve is so small that fixed fees outweigh the benefit, or if you may need the money for operations at any moment. Keep three lines on record: the band, the speed, and who can change them. The Norwegian fund separates the rule-setter (the Ministry) from the executor (the bank) (Norges Bank Investment Management, 2018); a small company might separate the owner from the chief accountant in the same way (my suggestion).

Going deeper: what did I read?

I read the August 2018 letter from Norges Bank to the Ministry of Finance and the sections on key figures, risk and the mandate in the fund's 2025 annual report. The two enclosures to the letter (on different rebalancing rules and on the detailed rules) are marked as exempt from publication, so I did not read them. The 2025 report does not state the current rebalancing band; the 2-point figure comes from NBIM's website, which is not a library source, and I do not know when it changed.

The post does not use anything from Markowitz or Tobin: portfolio theory appears only at the concept level on the framework page. Nor does it assess how the Norwegian fund performed against others, or say whether 70 per cent or any other level is the right one, and the VND 10bn company is wholly hypothetical.