Using ADP Data to Forecast Private NFP Growth

Using the correlation between the log first difference of BLS nonfarm payrolls and ADP, I get the following forecast or “nowcast” for tomorrow’s BLS data release:

figure 1: BLS nonfarm payrolls (black) and forecast (light blue), +/- 1 standard deviation (gray line), Bloomberg consensus 4/6, both sa Source: BEA, ADP via FRED, Bloomberg and Author’s calculation.

I estimate this specification during 2021M06-2023M02:

Δpnfp_bls_Ton = 0.001 – 0.593 pnfp_adp_Ton + you_Ton

Adjustment. R² = 0.63 SER = 0.0010, n = 21, DW = 2.38. (coefficients significant at 5% msl, using HAC robust standard errors bold italic.)

Note that a simple autoregression of first-difference private NFP yields a rather low adjusted R2 (0.19). The lagged private NFP from the BLS does not enter significantly if entered into the above equation.

In terms of numbers (not percentages), the Bloomberg consensus is for a net gain of 215k, while the model implies 281k.The Bloomberg Consensus is the average of forecasters using various methods, some of which involve disaggregation into categories, forecasting, and then aggregation (e.g. astotal NFP approach). I used a simpler method (I don’t think it’s necessarily better).

Below is the underlying data (by level).

figure 2: BLS nonfarm payrolls series (black) and ADP series (light brown), both 000’s, sa Source: BEA, ADP from FRED.