MSCI Indonesia: the cost of a narrower benchmark

What is lost when a benchmark becomes narrower

An equity benchmark is expected both to describe a broad market and to be a portfolio investors can hold. Those goals usually point in the same direction. Doubt about ownership information can put them in conflict. Retaining broad coverage may include shares that are difficult to acquire in practice. Restricting coverage to the portion that can be held with confidence may measure a much narrower group of companies than the market its name suggests. Choosing which error to reduce affects the opportunity set offered to international capital.

MSCI’s 15 October 2026 proposal would move Indonesia from Emerging to Standalone. Increases in float factors and index share counts and new additions remain restricted; authority-flagged high-concentration securities can still be deleted and ownership data used to adjust float. This is consultation, not an exchange trading suspension. [1] [2]

The central problem is that decisions protecting a benchmark’s dependability can make more companies disappear from its field of view. Better information about the survivors does not establish improvement across the market, including omitted firms. SG Group places more weight on the evidence that would permit broader inclusion, and whether that evidence works repeatedly across the market, than on the classification label alone. Investors can use a cautious measure properly only when they understand what it has stopped measuring.

A place to trade and a place in an index

Exchange listing, legal eligibility for foreign ownership, practical access to sufficient shares and inclusion in a particular index are different conditions. Satisfying one does not automatically satisfy the others. A company omitted from a benchmark still has a business and shareholders; its revenue and factories do not disappear. Yet continued listing alone does not guarantee the same attention from international managers. Benchmark membership is one route through which a company enters investors’ comparison set.

For companies, this exposes questions that compliance with listing rules alone cannot answer. Identifying registered holders may not tell a manager how much stock could become available to the market. Frequent trading does not remove the need to explain ownership either: turnover and the stock of accessible shares measure different things. Investor communication has to connect formal eligibility with practical availability.

The guide to equity-index weighting and ETFs explains the inputs to ordinary index calculations. The issue here sits outside that arithmetic: the same formula can describe a different market when the admissible inputs change. A narrow set that is easier to replicate and a broad representation of economic activity may therefore need different uses.

Asymmetric acceptance leaves a history

Consider a system that cannot incorporate expansions and contractions on equal terms. It may demand additional assurance before accepting evidence of greater availability, while reducing its estimate when earlier assumptions look optimistic. The resulting benchmark reflects both the current market and a history of evidence accepted or deferred. A current list of companies cannot reconstruct that history. It also takes information about improvements that were not admitted at the time.

Suppose, hypothetically, that a company’s accessible ownership first narrows and later broadens again. If the contraction is incorporated while evidence of recovery awaits stronger assurance, identical starting and ending ownership structures need not produce identical index treatment. This is a conditional example, not the history of any Indonesian security. It shows why the route by which information is accepted can matter even when the underlying market returns to the same endpoint.

Time alone need not close the gap. More information can arrive without satisfying the conditions for using it. Managers need to understand the evidence required to expand the admissible set, and the range of securities for which it is persuasive, rather than simply counting reports. As conservative treatment persists, explaining deferred information becomes increasingly important to understanding market representation.

Caution carries two kinds of cost

Including shares that investors cannot reliably acquire can impose acquisition costs on managers seeking to replicate the index. A smaller available quantity may require split orders or substitute holdings. Omitting shares that are in fact accessible creates a different cost: an incomplete comparison set. That omission is less visible while the index continues to calculate normally each day. Operational continuity and broad market representation can therefore produce different assessments of success.

Which error matters more depends on the user. Overstating accessibility is an immediate problem for someone trying to reproduce a portfolio. Missing companies may matter more to someone studying changes in domestic industry. Giving both users the same measure for the same purpose invites conflicting expectations. Assessing a benchmark requires asking which use it serves and which error its design seeks to reduce, rather than ranking it by constituent count alone.

There is a strong defence of strict standards: retaining poorly understood exposures to create the appearance of breadth transfers the difficulty to actual investors. But that defence does not establish that a narrower index also describes the entire market well. Supporting a cautious investability standard still leaves a need to observe omitted firms separately. Making the cost of caution visible clarifies the benchmark’s appropriate uses; it does not invalidate caution itself.

Why the survivors can give a misleading answer

The surviving constituents may have clearer disclosures than the earlier group, making market transparency look better. Yet if difficult-to-explain companies have already been removed, the group average can improve without any individual company changing. Selection itself moves the result. This can happen independently of rising or falling share prices. Comparing only the documents of current members risks crediting changes in membership to improvements in information.

One way to address this bias is to ask the same questions of three groups: retained firms, omitted firms and potential entrants. Can ownership be explained? Does the explanation hold at another date? Can a correction be traced? Looking for where the explanation fails across those groups is more useful than compressing them into a single completion rate. It identifies obstacles to broader inclusion. This is a proposed way to test improvement, not a finding about the present condition of each group.

Evidence about omitted firms is particularly valuable. If managers stop researching them because they sit outside the benchmark, improvements may have a weaker route into investment decisions. Companies may also find the benefit of continued explanation less visible after exclusion. Reporting that tracks changes among omitted firms can interrupt this loop more effectively than a report completed entirely within the surviving group. Treating exclusion as a reason to stop looking makes the original judgement harder to challenge.

From ownership records to accessible shares

Ownership information raises interpretative questions beyond numerical accuracy. Consistent names and holdings on a given date do not establish the conditions under which those holders might sell. There may also be reasons to examine whether formally separate holders behave in a related way. Similar patterns alone, however, do not prove wrongdoing or coordinated trading by a particular firm. Caution in estimating accessible shares and proof of illegal conduct require different standards of evidence.

Additional data also increases the need for a coherent explanation. A register, change notices, corporate capital actions and practical availability may refer to different dates. Greater numerical detail will not make those fragments describe the same state. When records conflict, users need to know which value changes and how earlier judgements are affected. A system that detects, corrects and preserves the history of errors can be more dependable over time than a promise that errors never occur.

The article on Eramet’s ownership and control illustrates how a holding and the rights attached to it can differ. The present question concerns how much stock outside investors can access, rather than who can determine corporate decisions. An explanation sufficient to understand control may still be insufficient to estimate market availability. The usefulness of a disclosure depends on the decision its reader is trying to make.

Read a smaller index against three different objects

Three reference sets help interpret a smaller benchmark: current constituents, the same companies followed over time and the whole market. Current members define today’s benchmark opportunity set. A fixed-company comparison removes some effects of changing composition. The whole market includes businesses outside the benchmark and is relevant to broader economic coverage. Each answers a different question. Substituting one for another obscures what the shrinkage means.

The two counts in the chart do not identify contributions to the change or describe firms outside the index. Explaining the reduction requires the individual changes. Before extending a conclusion about the survivors to the entire market, examine the effect of changing membership.

The same issue arises when profitability or information quality is compared across changing membership. An aggregate cannot reveal whether companies improved or difficult comparisons disappeared. The guide to index rebalancing provides the basic mechanics; placing fixed-company and current-membership comparisons alongside each other can then narrow the interpretation. A fixed-company approach still has to handle firms that delist along the way. Changing the comparison method does not eliminate every selection problem.

The comparison can change without an investor trading

A manager’s holdings can stay unchanged while a revised benchmark changes how far those holdings appear to depart from the market. Continuing to own a formerly included company turns that exposure into an off-benchmark position under the new comparison. Apparent distance can rise without any change in the manager’s view of the firm. Performance evaluation should avoid attributing changes caused by the comparison set to investment trades.

Where a mandate is tied to a specified benchmark, this change in distance can become an operational constraint. Responses depend on permitted off-benchmark exposure, explanation requirements and client reporting. It does not establish simultaneous selling. Inferring every manager’s action from an index change would treat different mandates as one contract. Estimating the effect on a company requires evidence of which investors hold it and under what terms.

Some users may only change a comparison; others may have to reconsider the investment policy itself. The distinction depends in part on whether the index defines the portfolio’s admissible universe or is used to assess results. The guide to global index families and coverage explains that starting point. Even when a classification label changes, the benchmark specified in each product or mandate is needed to identify whose decision would change.

A weight cannot be worked backwards into cash

A country’s benchmark weight depends on the value assigned to its companies and the value of other markets in the denominator. Within the country, prices, membership and the shares admitted to the calculation may all change. A lower weight alone cannot distinguish an investor outflow from rising values elsewhere or changes in coverage. Multiplying the weight by assets associated with a benchmark does not supply the missing explanation.

Even an actual sale has a buyer who pays for and receives the shares. Cross-border cash movement cannot be established without knowing the residence of the parties and the origin of the funds. Nor does a decline in market value mean that the same amount of cash was removed. Repricing, a transfer of holdings and a balance-of-payments flow need different evidence. Combining them into one loss figure can misidentify both the cause of a funding problem and its remedy.

Useful tests align membership and share-adjustment records, price changes, and actual holding or transaction evidence over the same period. Incomplete evidence can still narrow a conclusion if the explained and unexplained parts remain separate. Knowing that a company was removed, for example, does not establish that every former holder subsequently sold it. Keeping each observation in its proper role makes the market transmission more concrete.

Three connections that make better evidence useful

SG Group proposes three connections for assessing improved evidence: from record to object, across dates, and from evidence to a changed judgement. The first asks whether the holders and shares in a document correspond to the companies and share universe being assessed. It also tests whether evidence about one part is being used to reach a conclusion about the whole market. A document rich in detail may still be a weak basis for wider inclusion if that correspondence is unclear.

The second asks whether the explanation lasts. Even a coherent register at one date becomes stale if subsequent changes cannot be followed. The third asks whether sufficient evidence can alter the treatment. If errors can lead to a narrower set but the practical route for recognising improvement is unclear, those producing better information may struggle to see the benefit of their effort. Explaining how corrections can work in both directions supports trust in the process.

These tests are more demanding than counting submissions, but they do not depend on a single document format. Different records can be compared if their objects, continuity and connection to decisions are explained. Conversely, universal use of one form may be insufficient if corrections and exceptions are opaque. Transparency as a communications achievement requires different evidence from transparency that allows investment judgements to be updated. The latter is the more useful standard here.

Corrections also raise a question of scope. When a record is corrected at one firm, the appropriate review depends on whether the problem affects other firms using the same system or belongs to that firm alone. A shared problem warrants wider investigation. Generalising an isolated error to the whole market creates unnecessary burdens. The speed of correction and the evidence for extending it to other objects both matter if caution is to avoid excessive contraction.

Companies do not bear the same cost

If explaining a company to benchmark users becomes harder, additional work may follow: organising ownership data, reconciling documents and answering investor questions. Companies in the same market differ in how easily they can do this and in the investors they already reach. Firms with coherent records should not automatically be assigned the same costs as firms with unresolved gaps. The extent to which market-wide doubts spill over into company-specific assessments determines how the burden is distributed.

If investors narrow their research lists, smaller firms and potential new entrants may find it harder to establish contact. This is a conditional transmission channel where explanation has fixed costs, not a measured finding about company size. Established names benefit from accumulated research; newcomers have to answer the same questions from the beginning. A narrow benchmark gateway may allow that research advantage to persist.

That does not establish an immediate reduction in investment or employment. The timing of new financing, support from existing owners, borrowing and operating cash generation intervene. The guide to business revenue, profit and cash flow explains why equity valuation and payment capacity are different quantities. A funding analysis needs to trace exclusion from comparison, changed investor decisions and eventual corporate financing terms as separate steps.

Managers’ obligations and clients’ understanding

Managers must both maintain an achievable portfolio and explain its results to clients. Narrower coverage can make the first task easier while complicating the second. Avoiding hard-to-access shares may help execution, but clients expecting exposure to a whole country need an explanation of the businesses actually held. A country name in a product title does not establish exposure to every part of its domestic economy. The explanation has to focus on holdings rather than labels.

Clients may also use the index to judge domestic companies generally. Strength or weakness among omitted firms may receive little representation in that number. Different business coverage can sometimes explain why a product fails to rise with a strong economy, or remains resilient during weakness. Explaining any particular result still requires measured earnings and price evidence. Representation alone cannot be assigned responsibility for every market move.

Changing a category and fixing the underlying problem

A category change can alter which regional or market group provides the comparison. A different label does not itself repair incomplete ownership explanations. Equally, improved records at one firm may not restore market-wide treatment where a broader assessment is required. Classification compresses several judgements into one label. Using that label as the only measure of success conceals how far underlying improvement has progressed.

Company-level information and the continuing suitability of a market require different evidence. Strong firms can support a favourable market assessment without explaining how exceptions and corrections work throughout the system. Conversely, a need for market-wide review is no reason to ignore good company evidence. The assessment needs both accumulated improvements at individual firms and tests of whether the framework also works in difficult cases.

A binary question about retaining or changing a category therefore does not capture every investor concern. Representation can remain an issue under an unchanged category if coverage does not recover. Better company information can still have value to direct researchers after a category changes. Tracking the formal outcome and the usefulness of the evidence separately prevents a single announcement from carrying too much explanatory weight. It identifies where the actual improvement occurred.

SG Group View: explain both retention and restoration

Improvement may not map directly onto restoration, creating another difficulty. A firm can provide a clear explanation yet see little change in its treatment if concerns about other companies or the wider market persist. If firms stop explaining problems they cannot solve alone, wider improvement becomes harder. Restoration cannot be promised, but identifying which obstacles a company can address and which depend on market-wide conditions can help preserve the incentive to keep providing evidence.

The most easily overstated claim is that a single classification outcome determines investment in an entire country. Business values, investor constraints and existing holdings differ. An underappreciated possibility is that research habits change before a formal decision. If fewer investors investigate potential entrants, better documents may take longer to attract attention. Those habits can persist through a channel separate from the formal category itself.

Concern about deteriorating representation must itself be open to challenge. A narrower index may remain highly useful if omitted firms are also practically inaccessible to its users and the survivors capture the opportunities those users need. Better evidence across the wider market that changes treatment would reduce the gap discussed here. The seriousness of the concern depends on what omitted firms represent and who needs them, rather than on the count of omissions alone.

Testing competing explanations

One alternative is that a smaller index mainly reflects company values and market prices, with little effect from the treatment of information. Testing it requires comparing values for the same companies with values that include membership changes. A lower constituent count alone does not identify the cause. Price and coverage effects need to be separated, and simultaneous changes cannot be attributed to just one channel without a decomposition.

A second explanation is that managers continue to investigate off-benchmark firms, so the altered comparison does not materially reduce their opportunities. Research coverage, holdings and client communication can help test this; product names and categories cannot answer it. A third is that stricter standards improve confidence in the surviving group enough to offset the cost of narrower coverage. Evidence would have to connect that confidence to actual investment terms or valuations.

Dropping objects with missing evidence adds another layer of selection. Treating untraceable ownership as unchanged, or excluding every firm with incomplete records, can move the comparison in different directions. Record which groups contain the gaps and keep results confined to the objects they actually explain. If later evidence about missing firms differs markedly from that of the survivors, it can warrant revising the earlier market assessment. This preserves room to change a conclusion rather than filling missing observations with zero.

How the effect could reach international capital

There is also a possible feedback in the other direction. More specific investor questions may lead companies to improve ownership explanations. If that information is shared, other investors’ research costs may fall. An initial expense can produce a reusable benefit. The effect is more limited if each company provides separate explanations only in private meetings. Comparability and continuing updates determine whether a local response can contribute to wider improvement.

Guides to recording an investment thesis and building conditional scenarios can support this work. The useful step is to specify in advance which observation would change a judgement, rather than accumulating documents in support of a desired conclusion. Evidence that an explanation also works for omitted firms speaks more directly to representation than additional documents about survivors. It keeps the object of the assessment clear when the next announcement arrives.

Three conditional paths

The first path is that convincing explanations work across a broad range of companies, supporting reconsideration of conservative treatment. Attention should then extend beyond the category outcome to assessments of new and previously omitted firms. Wider coverage may require managers to restart research. There is still work and judgement between better information and restored research or holdings. The benefit is unlikely to reach every firm immediately upon announcement.

The second path is uneven improvement: some companies provide better explanations without changing the wider assessment. That has company-level value while leaving the gap in representation unresolved. The third is that new doubts or corrections warrant stricter estimates. Dependence on the remaining set could increase, but prices need not move in the same direction. These are paths for evidence and coverage, not a three-way forecast of rising, flat or falling share prices.

Across all three paths, constituent count alone does not measure diversification. Relevant questions include whether the survivors share industries or revenue drivers, and which activities omitted firms represent. The guide to index breadth and concentration explains that distinction. Even where wider membership is desirable, adding companies need not diversify the underlying dependencies. Coverage has to be assessed through both count and composition.

What remains uncertain and what would change the view

Major uncertainties concern which objects will be judged sufficiently explained, how revised treatment would reach practice, and how users would respond. These questions need not be resolved simultaneously by one date. Company assessments and mandates can produce different outcomes even after a classification decision. Investors need to identify what changes in the object they actually use, as well as whether an announcement has been made.

Useful evidence includes security-level treatment changes, the history of corrections to ownership data, explanations that also work for omitted firms, and managers’ statements about benchmarks and holdings policies. Aligned dates can distinguish improved information without changed treatment from changed treatment that research has not yet caught up with. Price moves placed beside a large headline cannot make that distinction. The evidence must show how far a change has travelled.

SG Group’s concern would weaken if durable explanations across a broad company set led to revised treatment that users could understand. A comparison including omitted firms could also show that the current benchmark already captures the market characteristics its users need, reducing the cost of narrower representation. If only survivors’ documents improve while omitted firms remain unexplained, the basis for extending that improvement to the entire market is weaker.

Final assessment: know what a dependable measure leaves out

The decisive question is whether evidence supporting expansion can become as explainable as evidence supporting contraction. A clear route can bring cautious implementation and broad representation closer over time. Without it, the index can protect the portion it understands while retaining a risk of drifting away from the wider market. In subsequent documents, changes to the explanation and treatment of particular companies and shares will be more informative than the forcefulness of the headline.