Mapping portfolio exposure by micro-market from the register

Every real estate credit book has geographic limits: so much for MMR, so much for Pune, a ceiling per city. Few have catchment limits, and catchments are where losses correlate. When one micro-market turns, it takes every project inside it down the same sales curve at the same time, whichever entity borrowed against them. The register makes a catchment map computable, within limits worth stating plainly. This piece is the method, the limits and the numbers.

The logic: where correlation actually lives#

Portfolio theory for a real estate book starts with an uncomfortable observation: the diversification unit that matters is not the borrower, the entity or the city, but the buyer pool. Projects within a catchment sell to the same households in the same price band; their sales rise and fall together; a stall among them anchors pricing for the rest, which is why the map of Maharashtra's lapsed projects shows them lapsing in clusters rather than at random. Two exposures in one catchment are, for stress purposes, closer to one exposure of double size than to two independent risks.

City aggregates hide this because cities contain their own dispersion. The first half of 2026 read well at city level in Mumbai, which logged 80,221 property registrations, up 6 percent and its best first half since 2013 (IGR Maharashtra data via Knight Frank India, June 2026). Pune's read was mixed: unsold stock up 19 percent to 57,879 units as launches rose 17 percent, while sales rose 2 percent to 24,890 homes (Knight Frank, India Real Estate H1 2026). Neither number says which of the city's micro-markets are clearing and which are filling, and a city cap treats every lane inside it as one.

What the register can supply, and what it cannot#

Placing each exposure. The pincode is filed on 19,646 of the 19,650 registrations active on the register as updated on 21 September 2026, and usable coordinates on 14,207 of them, 72.3 percent. Pincode is the practical catchment unit: it is filed everywhere and it is how buyers search. Where coordinates exist, a 2 kilometre radius is sharper, and a desk can use both.

Forward supply. Every registration files a completion date, and the building table, filed on 77.1 percent of active registrations against about a third of all registrations, because older filings rarely carry it, files sanctioned units. Stack a catchment's active projects by filed completion year and you have the supply your borrowers will be selling against, with the caveat that filed dates slip.

Sales pace. The building table files sold and unsold units as of today. Divide sold units by months on the register and you have an average pace per project; divide a catchment's unsold stock by its combined pace and you have months of inventory at that pace. It is a cross-section, many projects of different ages photographed on one day, and it should be labelled as one. What the register does not keep is the history: MahaRERA shows the current filing, not last quarter's, so a pace trend exists only where someone has been recording. ReraGenie began recording every project's filed figures on 31 July 2026, and a desk that starts its own quarter-end snapshot now has a trend within a year.

Distress share. Status is filed on every registration. Take the registrations whose original completion date has passed, the ones that have had their chance to finish, and count the share that lapsed, were revoked or were de-registered, the three ways a registration ends without a completion. A lapse covers both projects that stalled and projects that finished without filing their completion, so read the share as a delivery-and-discipline signal rather than a count of unfinished buildings. Either way it measures promoters who stopped keeping their filings current.

What is not there. The register files no flat configuration, no carpet area per flat and no sale price. A map built from it segments by place, size and record, never by ticket size, and any price-band overlay has to come from transaction data. The limits of the catchment method are set out in the micro-market method for developers, which reads the same tables for a launch decision.

One district, measured#

Here is the distress share for Thane district, across the pincodes with enough history to measure.

One district, one regulator and one market cycle, and a spread of more than 40 points: across all 21 measurable pincodes the share runs from 14.8 percent, in Ghansoli on the Navi Mumbai side, to 56.2 percent. The two Dombivli pincodes, one town, sit 17 points apart. The same spread appears in every large district we checked: Pune's 46 pincodes with 60 or more such registrations run from 17.2 to 60.6 percent around a district figure of 37.5 percent, and Mumbai Suburban's 32 run from 21.1 to 62.3 percent around 33.2 percent. A lender that knows its book is 30 percent Thane district knows almost nothing about its distress exposure until it knows which of these pincodes the 30 percent sits in.

Counting only registrations whose original date has passed holds age roughly constant, but not entirely: a pincode full of small 2017 filings will still read differently from one of recent towers. That is one more reason to read each catchment's own history rather than borrow the district's.

Worked example: the book that looked diversified#

Illustrative numbers throughout. A 12-exposure book, Rs 380 crore committed, spread across MMR and Pune, city limits comfortably respected: 55 percent MMR, 45 percent Pune. Map each exposure's project to its catchment and the picture changes. Three Thane-corridor loans, written in different years to different sponsors, sit within four kilometres of each other: Rs 88 crore. Two Pune west loans share a catchment: Rs 64 crore. Together, five loans and Rs 152 crore, exactly 40 percent of the book, sit in two buyer pools.

Attach the metrics. The Thane catchment reads healthy: 14 months of inventory at its average filed pace, and a distress share of 16 percent, near the best in the district. The Pune west catchment reads heavy: 31 months of inventory, and a distress share of 48 percent, three times the Thane figure. The desk's own quarter-end snapshots, kept since it started mapping, add the one thing the register cannot: the Pune catchment's sold units have barely moved in four quarters.

The Rs 64 crore there is not two performing loans; it is one concentrated bet that the catchment clears before the weaker sponsor's liquidity runs out. Sized that way, the desk caps new origination in the catchment, tightens the certificate gate on the weaker name now, while goodwill is cheap, and prices the next Pune west proposal to a different number than the city cap suggested. Nothing in the borrower files changed; the map did.

The stress test the map enables#

The map's second use is scenario work. Take the same book and shock the Pune west catchment: its pace halves from an already flat base, inventory months double towards 60, and one of its projects stalls outright. The loan-level consequences follow mechanically. The stronger borrower's exit window stretches past facility maturity, turning a performing loan into a restructuring candidate on timing alone. The weaker borrower's cash thins with its bookings: the Transaction account's 30 percent falls, and the 70 percent account can release only what buyers have paid into it, so interest cover breaks through the plumbing before any default, as the three-account arithmetic shows. The two exposures fail together, through the same buyer pool, and the Rs 64 crore behaves as the single bet it always was. The Thane loans, under the same shock applied district-wide, ride it out on their catchment's 14 months and its low distress share: same state, same cycle, opposite outcomes.

The shock parameters need not be invented either. The register does not archive old quarters, but it carries each catchment's completed record in the form that matters: how many of its registrations since 2017 finished, how many ended without a completion, and how far their filed dates moved. A catchment's own documented record is a more defensible stress assumption than a committee's round number.

From map to policy#

A map that stays a slide changes nothing; the last step turns it into limits and pricing. Three instruments follow from the metrics.

Catchment caps. A ceiling on committed exposure per catchment, expressed as a share of the real estate book, exactly as sector caps work. The cap can move with the data: more months of inventory, lower cap. A catchment under 12 months at its filed pace might carry a 10 percent ceiling; one past 24 months, half that, with existing exposure grandfathered and new origination stopped.

A pricing grid. Months-of-inventory bands map naturally to spread adjustments, because they price the borrower's exit risk: under 12 months at card rate, 12 to 24 at a premium, past 24 priced as the concentrated bet it is. The grid makes the map self-enforcing, since origination teams route around expensive catchments without being told to.

A refresh calendar. Progress and sales tables move on the quarterly filing cycle, so the map refreshes at quarter close, beside the account-level reviews, and each refresh is also the snapshot that builds the pace history the register does not keep. City series such as Mumbai's registration data show which way the tide is running; the filed catchment data says which of your corridors defy it, and the five early-warning signals score the individual accounts inside each one.

The policy step is where most attempts stall, usually on a fair objection: catchment boundaries are judgment. They are, and so are industry codes in sector limits, which nobody abandons for that reason. Draw the boundaries once, document them, and let consistency do the work precision cannot.

The analogy for committee: sector limits exist because steel loans default together. A residential catchment is a sector a few kilometres wide. Nobody runs a corporate book with industry exposure invisible; residential books run exactly that way whenever monitoring stops at the city line.

The Bhandarkar family office (illustrative, as ever) ran the mapping on its structured-credit allocations before its last commitment and found what most first maps find: an accidental concentration, two funds' deals in one Navi Mumbai corridor, invisible in every fund-level report because each fund was itself diversified. The commitment went ahead, into a different fund whose pipeline sat in catchments the office did not already own. The rest of that fund-level check, from the sponsor's own register record to the questions worth adding to a DDQ, is in committing to a real estate AIF.

Building the map by hand means aggregating every registration in each catchment, which is the assembly ReraGenie sells. The area market report, a flat Rs 2,999 per pincode, carries one pincode's filed supply by vintage, its status ledger and slip record, a sales cross-section, an inventory table of its active registrations with their pace measures, its operators, the lenders holding declared charges there, and the complaints and cases on record; it can be ordered from the reports page. The Rs 2,999 project analysis places a single exposure against its competitive set, and the free district pages show each district's registrations by status, where its possession promises land and how far they have shifted, as the first cut.

The one-line summary#

Losses correlate where buyers overlap: map every exposure to its pincode, attach filed supply, pace and distress share, start recording the register every quarter because it keeps no history, and cap and price at that level, where the concentrations your city limits were designed to prevent actually sit.

Methodology and sources#

• Register figures: ReraGenie analysis of every published MahaRERA registration, as updated on 21 September 2026. Distress share counts registrations with status Lapsed, Revoked or De-Registered among those whose original filed completion date had passed by that date; pincodes are shown where 60 or more such registrations exist. Because it counts only registrations whose original date has passed, it runs higher than a share of all registrations, which includes projects still within their dates.
• City figures: Knight Frank India, Mumbai registrations from IGR Maharashtra data (June 2026) and India Real Estate H1 2026, as reported by Punekar News.
• RBI's large exposures framework for banks: single counterparty and connected-group limits as a share of Tier 1 capital.

This article is educational and not credit or legal advice.