Manhattan Housing & Economic Regimes Dashboard

Total Transactions

--

Manhattan Database

Economic Regimes

--

K-Means Corrected

Predictive R²

0.82+

RF + FinBERT Sentiment

Final Project Conclusion

Question: "How do Manhattan apartment prices respond to macroeconomic conditions across different economic regimes, and does incorporating textual sentiment from FOMC communications via NLP improve predictive accuracy?"

Conclusion: Yes, significantly. We isolated distinct business cycles ("Regimes"), extracted state-of-the-art sentiment probabilities from Federal Reserve meeting minutes using FinBERT, and proved Granger-causal feedback loops between central bank signaling, interest rates, and housing. Incorporating FinBERT Hawkishness and Regime classes into non-linear Random Forest architectures significantly outperformed classical macroeconomic benchmarks.

Apartment Price Appreciation by Segment and Regime (MoM %)

Homebuyer Insight: This chart illustrates historical Month-over-Month price appreciation across different apartment sizes during distinct macroeconomic business cycles. Notice how larger apartments (3BR/4BR+) experience higher volatility during Expansion cycles, while smaller units (Studios/1BR) maintain steadier, tighter appreciation ranges even amid aggressive Fed tightening.

Regime Distribution Sequence

Economic Fingerprints by Regime (Including FinBERT)

Target Segment:

Median Sale Prices by Cycle & Bed Count (Log10)

Granger Causality Heatmap (p-values)

The FOMC Feedback Loop

FinBERT
Net Hawkishness

↓ Granger Causes (p < 0.05)

FEDFUNDS
Interest Rate

↓ Granger Causes (p < 0.05)

Manhattan
Housing Dynamics

Our VAR modeling confirms a unidirectional transmission mechanism: FOMC rhetorical sentiment directly predicts future shifts in the Federal Funds Rate, which in turn structurally shifts Manhattan housing volatility. The central bank's "jawboning" is a leading indicator for the real estate market.

Interactive Data Synthesis Dashboard