Why It Matters
When you see "New York-Jersey City" or "Los Angeles-Long Beach-Glendale" in federal data and wonder why it doesn't match your mental model of the metro, you're looking at a Metropolitan Division. OMB splits the 11 biggest MSAs into MDs because internal commuting within those metros isn't unified — people who live in Oakland commute to a different job center than people who live in San Francisco, and treating the Bay Area as one labor market blurs real economic differences. The practical consequence for investors: FHFA doesn't publish a parent-MSA HPI for those 11 metros. You either aggregate MD data yourself, or you use Freddie Mac's FMHPI as a parent-level substitute.
At a Glance
- What it is: A sub-unit of an MSA defined by OMB for the 11 largest MSAs where internal commuting structure is complex.
- Why it matters: FHFA doesn't publish parent-MSA HPI for these 11 metros — you get MD-level data or you switch to Freddie FMHPI as a substitute.
- Which MSAs split: New York (5 MDs), Los Angeles (2), Chicago (2), Dallas-Fort Worth (2), Washington DC (2), Philadelphia (2), Miami (2), Atlanta (2), Boston (2), San Francisco (2), Detroit (2), Seattle (2), Tampa (2). Total: 13 MSAs across different OMB vintages; 11 active under the current 2023 standards.
- Commuting threshold: An MSA splits into MDs when it has 2.5M+ population AND its internal commuting structure has distinct employment cores.
- 5-digit codes: MDs have their own CBSA codes separate from the parent MSA.
How It Works
When OMB splits an MSA into MDs. Most MSAs are single functional labor markets — you commute to a downtown job center, the suburbs serve that core, and one set of demographic and economic data describes the whole thing. But once an MSA gets past 2.5 million people AND contains multiple employment cores that don't share unified commuting patterns, OMB breaks it into Metropolitan Divisions. The most dramatic example: New York-Newark-Jersey City MSA splits into 5 MDs — Nassau County-Suffolk County (Long Island), Newark (northern NJ), New York-Jersey City-White Plains (the urban core), and two others. A software engineer in Newark commutes to a different set of jobs than one in Long Island, even though they're technically in the same MSA. The MD structure captures that. Full glossary at Census's metro-micro reference; broader context in Wikipedia's MSA overview.
The 11 MSAs that split. Under the current OMB 2023 standards, 11 MSAs contain Metropolitan Divisions: New York (5 MDs), and 10 others with 2 MDs each — Los Angeles, Chicago, Dallas-Fort Worth, Washington DC, Philadelphia, Miami, Atlanta, Boston, San Francisco, Detroit, Seattle. The LA MSA splits into Los Angeles-Long Beach-Glendale and Anaheim-Santa Ana-Irvine. Chicago splits into Chicago-Naperville-Evanston and a smaller Gary-Hammond IN MD. Each MD has its own 5-digit CBSA code, its own demographic data, and its own economic indicators. Federal statistical agencies report data at both the MSA level AND the MD level for these metros, which is why the same news event can get you slightly different numbers depending on which footprint the reporter used.
Why investors care — the HPI problem. FHFA publishes its House Price Index at either the MSA or MD grain, but not both. For the 11 MD-split metros, FHFA reports only at MD grain — no parent-MSA rollup. The full FHFA HPI datasets page documents the methodology. If you're trying to underwrite "appreciation in NYC" over 5 years, FHFA gives you 5 separate MD readings, not one number for the metro. You either aggregate them yourself (weighting by population or housing unit count) or you substitute Freddie Mac's FMHPI, which does report at parent-MSA grain. Most metro-hub analyses on market-data platforms that cite "FHFA HPI for NYC" are actually Freddie FMHPI values relabeled. Our own metro hub pages use Freddie FMHPI for those 11 metros specifically to get a parent-MSA reading.
The commuting logic that makes MDs real. OMB's methodology for defining MDs uses the same commuting data as the parent MSA definition, but applied differently. The parent MSA includes counties where 25%+ of workers commute to (or from) the urban core. Within an MSA that's eligible for splitting (2.5M+ population), OMB then looks for counties that share 25%+ commuting WITH EACH OTHER but NOT with other clusters in the same MSA. Those clusters become MDs. Effectively: MDs are "the sub-regions where people actually live and work together." The practical test: ask a local where they live and where they work. In Tampa, everyone says "Tampa" — one MSA, no MDs. In the Bay Area, the answer depends on whether you live in Oakland, SF, or Silicon Valley — three distinct commuting circuits, captured by 2 MDs plus a separate San Jose MSA.
Real-World Example
Diego Ramírez tries to pull 5-year HPI for "NYC" and discovers MDs.
Diego is evaluating a Jersey City 3-unit as his next acquisition. He wants to know NYC metro appreciation over 5 years. He pulls FHFA HPI for CBSA 35620 (New York-Newark-Jersey City MSA). No data.
Looking closer at the FHFA dataset, he sees 5 separate rows for the NYC metro:
- New York-Jersey City-White Plains (MD 35614) — the urban core, includes Manhattan and Jersey City
- Nassau County-Suffolk County (MD 35004) — Long Island
- Newark (MD 35084) — northern NJ minus Jersey City
- Dutchess County-Putnam County (MD 20524) — exurban Hudson Valley
- New Brunswick-Lakewood (MD 35154) — central NJ
His Jersey City property sits in MD 35614 (New York-Jersey City-White Plains). He pulls HPI for that MD: +26% over 5 years. If he'd used Long Island (MD 35004), he'd have gotten +34%. If he'd naively averaged the 5 MDs, he'd have gotten a number that doesn't describe any actual submarket.
He also runs Freddie FMHPI for the parent MSA — +29% over 5 years, a weighted average across the 5 MDs. Useful for a top-level comparison, but for his specific property he sticks with the 35614 number since that's where Jersey City actually is.
The MD structure tells him something his friend hadn't mentioned: Jersey City (inside the NYC core MD) has appreciated at a markedly different rate than Newark (a separate MD a few miles away), and both differ from Long Island. "NYC metro appreciation" as a single number hides all of that.
Pros & Cons
- MDs capture real sub-metro differences that a parent-MSA number averages away
- 5-digit MD codes are unambiguous — no confusion about which sub-region a data point covers
- FRED distributes most MD-level series, so pulling sub-metro data is straightforward
- OMB updates MD definitions alongside MSA definitions, so vintage consistency is maintained
- HUD publishes Fair Market Rent at MD grain for the 11 split metros, giving you more precise voucher-level rent data
- FHFA's no-parent-MSA-HPI rule for the 11 split metros forces either aggregation or substitution
- Aggregating 5 NYC MDs into a single "NYC metro" number requires weighting choices — population, housing units, or transactions — and different choices give different answers
- MD boundaries can change between OMB vintages; 2023 Bulletin 23-01 adjusted several MD footprints
- Some commercial real estate platforms report "NYC metro" as their own aggregation, so two sources can both say "NYC" and mean different geographies
- MDs only exist for 11 MSAs — smaller metros don't have sub-units, so analysis techniques that work for NYC don't transfer to Cleveland
Watch Out
- FHFA HPI parent-MSA doesn't exist for these 11: Your first encounter with "why is NYC HPI missing" is always confusing. The answer: it's by design, and you use Freddie FMHPI as the parent-level substitute.
- MD names can be similar to MSA names: "Los Angeles-Long Beach-Glendale" (MD 31084) is NOT the full LA MSA. The full MSA is "Los Angeles-Long Beach-Anaheim" (CBSA 31080). Always check the 5-digit code.
- 2023 vintage moved some MD boundaries: Philadelphia's MDs, for example, had county movements in OMB Bulletin 23-01. Historical MD-level comparisons need vintage verification.
- Not every large metro has MDs: Phoenix, Houston, and Atlanta are all large MSAs, but only Atlanta splits into MDs. The threshold is population AND commuting-structure complexity — size alone doesn't trigger it.
- Commercial data platforms may not carry MDs: Some paid listing or underwriting platforms report only at MSA grain and silently roll up MD data. If you're comparing platform data to federal FHFA data, boundary differences can move your numbers 2-5%.
Ask an Investor
The Takeaway
Metropolitan Divisions are how OMB handles the reality that New York, LA, and a handful of other very large metros aren't single labor markets. If you're underwriting property in one of the 11 MD-split metros, learn the MD your property sits in, pull FHFA HPI at MD grain, and use Freddie FMHPI if you need a parent-MSA comparison. In the other ~380 MSAs, you can ignore MDs entirely — there aren't any.