Help us direct you to the right place to sign up
SKIP AHEAD TO
Simple rent averages can lie. When Manhattan’s office market appeared to be recovering sharply in 2022 and 2024, the raw data showed net effective rents spiking, but that surge was driven almost entirely by a disproportionate share of trophy building leases, not a broad market rebound. The Columbia CompStak Rent Index (CCRI) strips out these compositional distortions, revealing that actual quality-adjusted rents showed only modest improvement during those periods.
This constant quality rent index, developed jointly by Columbia Business School and CompStak, represents a methodological leap forward for CRE professionals who need reliable rent benchmarks that hold asset quality constant over time and across markets.
Why Simple Rent Averages Mislead
The core problem with tracking average rents is straightforward: the mix of properties leasing in any given quarter isn’t constant. Coming out of COVID, leasing activity skewed heavily toward trophy office buildings and the newest industrial assets, properties that naturally command premium rents. When most market activity concentrates in the high end, average rents rise even if underlying market conditions haven’t improved.
This compositional bias cuts both ways. In weak markets, averages may understate true rent levels if only distressed properties are transacting. In recovering markets, averages overstate strength when high-profile deals dominate. The effect is especially pronounced at the end of a data sample, when the first leases to hit the database tend to be the most newsworthy trophy deals, creating the illusion of a sharper recovery than actually exists.
According to CompStak data from Manhattan, the two periods of apparent rent recovery (Q2-Q3 2022 and Q2-Q4 2024) corresponded exactly with above-average concentrations of prime Class A lease signings. Once the CCRI methodology adjusts for this quality mix, those “recoveries” flatten considerably.
See the Columbia CompStak Rent Index in action. Download the national indices or specific market indices today (for free!).
Hierarchical Geographic Fixed Effects: The Methodological Core
The CCRI’s innovation lies in its hierarchical geographic fixed effects (HGFE), a technique that controls for unobserved quality dimensions that no list of variables can fully capture. Building finishes, corner versus mid-block location, lobby quality, specific floor attributes: these factors matter for pricing but are difficult or impossible to observe systematically.
The HGFE approach works by estimating building-specific rent baselines wherever data density allows. For a Manhattan office tower with 100 leases in the CompStak database, the model estimates a building fixed effect that captures all quality attributes—observed and unobserved—unique to that property. When building-level data is sparse, the model falls back to census block group, then census tract, then county.
The results are striking: 53% of observations support building-level fixed effects, and an additional 33% support block group-level effects. That means nearly 90% of the index’s lease observations benefit from highly granular quality controls. Compared to a rich hedonic model without HGFE, the constant quality approach explains approximately 20 percentage points more of cross-sectional rent variation, a substantial improvement in explanatory power.
What the Index Reveals About Manhattan Office
The practical difference between raw averages and quality-adjusted rents is stark in the Manhattan office market. According to CompStak data, raw net effective rents fell from roughly $73 to $58 per square foot during COVID, then appeared to surge during the 2022 and 2024 “recovery” periods.
The CCRI tells a different story. Quality-adjusted rents show a more muted trajectory, stable to modestly improving during those periods rather than sharply rebounding. Only in the final quarters of the dataset do both methods converge on the same conclusion: Manhattan office rents have now exceeded their pre-pandemic levels for the first time since COVID.
This distinction matters enormously for underwriting, portfolio allocation, and market timing. A simple average might signal “buy” when the market is merely experiencing a temporary concentration of trophy deals, not a fundamental recovery.
Applications Beyond Trend Analysis
The CCRI methodology extends beyond simple rent tracking. Portfolio managers can use the index to construct efficient frontiers, mapping MSA-level rent growth (return proxy) against volatility (standard deviation) to optimize geographic allocation. Early analysis suggests the index leads appraisal-based indices like NCREIF’s MPI by approximately four quarters—making it a potentially valuable leading indicator for investors tracking market momentum.
The hedonic regression framework also supports automated valuation applications. By holding all property characteristics constant except age, analysts can estimate fair market rents for vacant spaces or upcoming renewals without relying solely on traditional comparable-based appraisal methods, which can produce high-variance estimates depending on comp selection.
KEY TAKEAWAYS
- Compositional bias distorts simple rent averages: Manhattan’s apparent 2022 and 2024 office “recoveries” were largely driven by above-average trophy building lease concentrations—quality-adjusted rents showed only modest improvement.
- Hierarchical geographic fixed effects explain 20 percentage points more rent variation than standard hedonic models, enabling far more precise quality adjustment.
- Nearly 90% of CCRI observations use building-level or block group-level controls, ensuring granular quality adjustment even for unobserved property characteristics.
- The CCRI leads appraisal-based indices by approximately four quarters, with a 54% correlation to NCREIF MPI when lagged—suggesting value as a market leading indicator.
- Manhattan office rents have now exceeded pre-pandemic levels for the first time since COVID, according to both raw data and the quality-adjusted CCRI.