Employment Composition (CPS)

Monthly composite of labor market health from CPS: Participation, Employment–Population Ratio, and Unemployment Rate. Robust z‑scores, momentum diagnostics, 0–100 scaling, and categorical regimes.

Why: Higher participation and employment with lower unemployment indicate a tighter, healthier labor market; the inverse signals rising slack.

Abstract

We construct a monthly index from three CPS (household survey) series: Participation, Employment–Population, and Unemployment. Each is standardized with a robust rolling z‑score. We then build a weighted composite (participation 0.4, emp‑pop 0.4, unemployment −0.2), apply smoothing, scale it to 0–100, and classify regimes.

1. Data (BLS CPS Identifiers)

LNS11300000 — Labor force participation rate (%)
LNS12300000 — Employment–population ratio (%)
LNS14000000 — Unemployment rate (%)

Inputs are monthly. If any required ID is missing from the initial subset, the pipeline augments from long_df (or many) before pivoting.

2. Data Handling & Validation

Types & dates: Coerce value to numeric; parse date to month‑start timestamps.
Pivot & grid: Wide pivot by series_id → monthly start grid via asfreq('MS').
Gaps: Bounded forward‑fill ffill(limit=2); larger gaps remain NaN.
Fail‑fast: raise if any of the three core series is entirely missing after pivot.

3. Feature Engineering

Participation_vs_EmpRatio = Participation / EmpPop

Labor_Slack_Index = (1 − Unemployment/10) × Participation

Momentum diagnostics: 3‑month differences for each core rate (diff(3)).

4. Standardisation (Robust Rolling z‑scores)

Each series is scaled with a median/MAD z‑score using an adaptive window.

z_t(x) = (x_t − median_W(x)) / (1.4826·MAD_W(x)),\ W = min(36, max(8, ⌊0.8·N_valid⌋))

Unemployment enters negatively via unemp_z_neg = −z(unemployment).

5. Weighting & Composite Construction

Weights reflect economic intuition: participation and employment carry equal weight; unemployment detracts.

{
  "part_z": 0.40,
  "empr_z": 0.40,
  "unemp_z_neg": 0.20
}

Employment_Composition_z = 0.40·part_z + 0.40·empr_z + 0.20·unemp_{z_neg}

6. Smoothing, Scaling & Regimes

Headline smoothing: 3‑period EMA on the composite z‑score.
Score (0–100): min–max scaling over observed history; returns NaN if flat.
Regimes: thresholds on the unsmoothed composite z‑score.

HOT (healthy/tight): > +0.75
NEUTRAL: −0.75 to +0.75
COOL (slack rising): < −0.75

7. Output Panel

[
  "participation_rate","emp_pop_ratio","unemployment_rate",
  "Participation_vs_EmpRatio","Labor_Slack_Index",
  "part_mom_3m","empr_mom_3m","unemp_mom_3m",
  "part_z","empr_z","unemp_z_neg",
  "Employment_Composition_z","Employment_Composition_0_100",
  "Employment_Composition_Smoothed","Employment_Regime"
]

8. Implementation Notes (Python)

# Expect columns: date, series_id, value
REQ = ["LNS11300000","LNS12300000","LNS14000000"]
# Merge in any missing IDs from long_df/many prior to pivot; align to MS grid; ffill(limit=2)
# Apply robust_z(), momentum, composite, EMA smoothing, regime, and 0–100 scale as specified above.

9. Interpretation & Use

A HOT employment composition reading aligns with historically tight labor conditions and potential wage pressure; a COOL reading indicates rising slack. Use alongside wage, inflation, and liquidity signals for cross‑validation.