FIPI and LIPI: Reading the Money Flows That Move the PSX

Every trading day, after the close, the National Clearing Company of Pakistan publishes something more useful than the closing price: a list of who bought and who sold. Foreign funds, local mutual funds, banks, insurance companies, ordinary individuals — each category's net position, in dollars, every session. It is the closest thing the PSX has to an X-ray. Most retail investors never look at it. Of those who do, most read it as a smart-money signal, which is the one thing it reliably is not.

What the clearing company sees

NCCPL sits underneath the exchange: every PSX trade clears and settles through it, so it does not sample the market the way a broker or a journalist does — it sees all of it. From that position it publishes two daily series.

FIPI — Foreign Investor Portfolio Investment — is the net buying or selling by foreign corporates, foreign individuals and overseas Pakistanis. LIPI — the local equivalent — splits domestic money into categories: individuals, companies, banks and DFIs, NBFCs, mutual funds, insurance companies, brokers trading their own books, and a catch-all of other organisations. The figures are net values, quoted in US dollars, with running totals for the month and the fiscal year.

One property of these numbers does more work than any other: they sum to zero. Every share bought was sold by someone, so if foreigners are net sellers of $5 million, local categories between them were net buyers of exactly $5 million. "Money left the market" is a phrase the data does not support. What actually happened is that foreigners handed stock to locals at prices both sides accepted. Whether that was foreigners being wise or locals getting a bargain is precisely the question the report cannot answer — and everything useful about reading it follows from keeping that in mind.

Why a few million dollars moves this market

On a large exchange, one fund's rebalancing disappears into the flow. The PSX is not a large exchange. The free float — the portion of shares actually available to trade, rather than locked up with sponsors, group companies and the state — is a fraction of an already modest market capitalisation. On many sessions the entire exchange turns over less value than a single large American stock trades before lunch. And the KSE-100's construction concentrates that further, because a handful of heavyweight names carry most of the index's movement.

The mechanics are the same ones that make pump groups profitable in illiquid small-caps — a thin order book means modest, persistent buying moves the price, which is a large part of why most stock tips lose money — except here they operate at the scale of the whole market. A single foreign fund deciding Pakistan should be 0.3% of its portfolio instead of 0.5% is not background noise in Karachi. On a quiet day it can be a meaningful fraction of everything that traded. That is why FIPI is watched here with an intensity that would look eccentric in Mumbai or New York: on a thin market, category flows are not commentary on the price. They are frequently the thing setting it.

How to read the report without fooling yourself

The daily table looks simple, which is how it misleads. Four disciplines keep it honest:

• Net figures hide gross churn. FIPI of −$0.4 million could mean foreigners bought $19.8 million and sold $20.2 million. That is a very different market from one where no foreigner showed up at all, and the headline number cannot tell you which you are looking at.
• One day means nothing. A single session's flow can be one fund rebalancing, a block trade crossing, or an index-tracking adjustment. Reacting to it is reacting to noise with a category label attached.
• Cumulative trends mean something. Weeks or months of flow in one direction tells you a category has genuinely changed its stance. That is the unit of information here: the month, not the day.
• Categories are aggregates. "Individuals" nets the decisions of hundreds of thousands of people who disagree with each other into one line. A small net figure can conceal enormous, evenly matched conviction on both sides.

The dollar quotation adds one more wrinkle: across years of rupee depreciation, a flow of the same dollar size represents a growing share of a rupee-denominated market. Long-run comparisons need that adjustment before they mean anything.

The patterns that keep repeating

The modern FIPI story has one dominant arc. MSCI upgraded Pakistan to emerging-market status in 2017, and foreign money famously bought the story in advance and sold the fact. From 2017 until the demotion back to frontier status in late 2021 — and for a while after — foreigners were persistent net sellers, year after year, through good news and bad.

Someone absorbed all of it, because someone must. The buyers were overwhelmingly local: mutual funds and individuals took the bulk, with insurance companies as the steady, unhurried bid underneath. Insurers are structurally suited to that role — long-dated rupee liabilities, no redemption pressure, and a mandate measured in decades — so they accumulate on weakness and almost never sell in size. For years, the shape of the market was foreign supply meeting patient local demand at ever-lower valuations.

Then the direction of travel changed where it usually changes: at a macro inflection. After the IMF Stand-By Arrangement in mid-2023, with default risk repricing and the currency stabilising, foreign buying returned — and the index more than doubled over the following eighteen months. That clustering is the pattern to internalise: foreign money treats Pakistan as a macro instrument, not a collection of companies. It moves around IMF milestones, reserves and currency turns, not around cement dispatch numbers. A foreign fund does not buy LUCK because it admires the kilns. It buys Pakistan risk, and LUCK is how you buy it.

Sentiment of categories, not an oracle

Here is the part the morning-show framing gets wrong. Foreign selling is routinely narrated as "smart money leaving", and foreign buying as validation. The record does not support either. Foreign selling ran straight through some of the cheapest markets Pakistan has ever offered — anyone who took persistent FIPI outflows as a verdict sat out the recovery that followed. And foreign buying has arrived late to rallies as often as early.

The reason is that each category trades its own situation, not the market's future. Foreigners face redemptions at home, index-weight rules and frontier-exposure caps; their selling often says something about their investors and nothing about HBL's deposit franchise. Mutual funds are the mirror image: when unit holders pile in, funds must buy something, and when unit holders panic, funds must sell regardless of view. Bank treasuries compete equities against government paper — when the policy rate peaked at 22% in 2023-24, a guaranteed 22% made them predictably scarce buyers, and their return as rates fell was arithmetic, not insight. Read each line as the circumstances of a category and the table becomes legible. Read it as a forecast and you have adopted a tipster with a spreadsheet.

A field guide to the categories

Using it with one portfolio

The genuinely practical use is context, not prediction. Suppose your holding is down 8% this month. The first question is not "what is wrong with the company" but "who is selling" — and the flow data can answer it. If FIPI shows sustained foreign selling in financials and HBL, UBL and MCB have all fallen in step — check the sector pages and the individual charts, say HBL's page, against the KSE-100 — then your stock's weakness may need no company-specific explanation at all. A category is exiting, it sells what it owns, and forced sellers eventually finish. That reading changes what you do: it argues for checking your original thesis rather than abandoning it.

The second use is divergence. Persistent selling that has stopped pushing prices down is telling you the supply is being absorbed; persistent buying that no longer lifts prices says demand is exhausting itself. Both are worth noticing. Neither is worth acting on alone, because flows tell you who and never why — and on their own they have no record as a timing signal that survives honest testing.

NCCPL publishes the tables daily on its website, free as of mid-2026, and nearly every brokerage summarises them in its morning note. Which makes the last observation the telling one: the flow data is public, official and costs nothing, yet nobody forwards it on WhatsApp — because there is no commission in it and no urgency to sell you. Keep your own monthly tally of FIPI and the two or three big local lines instead. The month a persistent seller finally goes quiet will tell you more than any target price forwarded by a stranger.