In global shallow GCMT earthquakes (1977-2026), faulting style changed the b-value only at Mw 6.5 and larger; at Mw 5.8-6.5 and within 5-degree cells after 2005 the rake effect was absent

We tested the claim that the Gutenberg-Richter b-value depends on faulting style (Schorlemmer et al. 2005: normal highest, thrust lowest). We used shallow (centroid depth 50 km or less) Global CMT earthquakes of Mw 5.8 and larger. We fitted on 1977-2004 (n = 3,687) and tested on 2005 to June 2026 (n = 3,320). We used an exponential magnitude likelihood with ln b = alpha + gamma sin(rake), 5-degree cell fixed effects and a cluster bootstrap. Globally, gamma stayed negative but became half as large: -0.117 (95% CI -0.176 to -0.064), then -0.061 (-0.110 to -0.006). After 2005, normal b was not clearly higher than thrust b. Within 5-degree cells, gamma was -0.070 (-0.157 to +0.006), then +0.022 (-0.038 to +0.092). For 5.8 <= Mw < 6.5, the rake slope was zero or positive in both periods. For Mw 6.5 and larger (post hoc), gamma was -0.16 in both periods, globally and within cells (intervals exclude 0). Thrust b was 0.97 and 0.87; strike-slip b was 1.23 and 1.12. Class b-values fitted on 1977-2004 did not improve the 2005-2026 magnitude likelihood (+0.8 millinats per event, -1.8 to +3.4). Cell b-values did improve it (+7.0, +3.2 to +11.2), and the faulting class added nothing beyond the cell b-values. Thus the faulting-style effect is in the large-event tail and is not a general change of slope. This agrees with rupture-width saturation more than with a stress control that acts at all magnitudes, but we did not test the mechanism. Limitations: the GCMT method changed in 2004, the Mw 6.5 split is post hoc, and the within-cell intervals are wide.

b-value · Gutenberg-Richter · faulting style · rake · Global CMT · stress meter · out-of-sample test · replication · earthquake magnitude distribution · seismic hazard