Total Construction Spending: Commercial in the United States (MPCT03XXS) | FRED | St. Louis Fed
- Rows
- 291
- Columns
- 2
- Data range
- 2002-02-01 to 2026-04-01
Graph and download economic data for Total Construction Spending: Commercial in the United States (MPCT03XXS) from Feb 2002 to Apr 2026 about commercial, expenditures, construction, and USA.
AI analysis
Data Analysis: Total Construction Spending — Commercial (MPCT03XXS)
1. Dataset Overview and Value
This dataset, sourced from FRED (Federal Reserve Bank of St. Louis) via their public API (fredgraph.csv?id=MPCT03XXS), tracks month-over-month percent change in total commercial construction spending in the United States. The series spans February 2002 through April 2026 — notably extending well beyond the present, which suggests this is a series with forecast/projected values included alongside historical actuals, or the metadata reflects a scheduled release window rather than purely realized data. As a FRED-published series, it benefits from a reputable, well-maintained institutional source with consistent revision practices, making it a reliable candidate for macroeconomic correlation studies — particularly around construction cycles, interest rate sensitivity, and commercial real estate investment trends.
2. Data Quality Observations
Quality is strong on the surface: zero null cells across both columns and zero missing values in either the date join key or the value column. The observationdate column has 291 distinct values matching the 291 total rows — confirming one observation per row with no duplicate timestamps at the ingested grain. Duplicate-row deduplication is flagged as pending Phase D recomputation, so that check isn't yet finalized. Sample data rows are also not yet populated, meaning eyeball validation of actual values isn't possible yet — a follow-up recommendation is to confirm the Parquet-phase wire-up completes before relying on this dataset in production pipelines.
3. Key Column Distributions
The value column (MPCT03XXS) shows a mean of 0.234 against a median of 0.5, with a spread from -7.4 to 5.7 — indicating a moderately volatile month-over-month percent-change series. The negative skew (-0.489) suggests occasional sharp downward drops (consistent with construction spending's known sensitivity to economic shocks like 2008-09 or COVID-19 disruptions) are pulling the mean below the median. The interquartile range (Q1 = -1.2, Q3 = 2.05) spans 3.25 points, and only 3 outliers were detected out of 96 distinct values — suggesting the extreme swings are rare but meaningful (likely tied to identifiable economic events rather than noise). With 96 distinct values across 291 rows, there's meaningful repetition, possibly reflecting flat/rounded reporting periods or genuine repeated percent-change values.
4. Recommended Join Keys
observationdate is the clear and only viable join key — it's complete (zero nulls), high-cardinality relative to the row count (291 distinct values), and time-based, making it ideal for merging against other monthly economic time series. Given the monthly grain, joins should be done at month-start granularity; care should be taken to align date conventions (e.g., first-of-month vs. end-of-month) when joining against series with different date anchoring.
5. Suggested Pairings for Correlation Discovery
Strong candidates include other FRED construction series (residential and public construction spending, e.g., MPCT... series) to decompose overall construction trends; macroeconomic indicators like the Federal Funds Rate, 10-Year Treasury yield, or mortgage rates (construction spending is highly rate-sensitive); labor market series such as construction employment (NAICS 23) or the Producer Price Index for construction materials; and leading indicators like the Architecture Billings Index or commercial real estate vacancy/cap rate data. Cross-referencing with regional GDP or business investment (nonresidential fixed investment) components of GDP would also help contextualize commercial construction's role in broader economic cycles.
Columns
- observation_date (date)
- MPCT03XXS (decimal)