A single participant's dot now sets the Fed's 2029 funds-rate median

The trim costs six dots in every column. The central tendency excludes the three highest and three lowest projections for each variable in each year ^[6]. That leaves 2026 and 2027 summarising 12 of the 18 submissions ^[3]. For 2028 and 2029, where one participant filed nothing ^[3], it is 11 of 17 ^[4].

The median works differently. It is the middle projection, and with an even count it is the average of the two middle ones ^[5]. Eighteen is even and seventeen is not, so the 2026 and 2027 medians average the ninth and tenth dots while the 2028 and 2029 medians are the ninth dot alone ^[2]. In June, one participant skipped 2028 but did file 2029 ^[4], which means June's 2029 median averaged two numbers and September's does not ^[5].

Figure 2 rounds each participant's judgement to the nearest eighth of a percentage point ^[9], so adjacent positions sit 12.5 basis points apart and two positions make a quarter point ^[6]. Move one dot by a single position in the 2029 column and the median moves the full 12.5 basis points. Make the same move in an even-count column and the two-dot average moves 6.25 ^[7].

The units differ from the ones most fund models carry, too. GDP growth and both inflation measures are fourth-quarter-over-fourth-quarter changes, and unemployment is the fourth-quarter average ^[11]. The funds-rate figure is the midpoint of the projected appropriate target range at the end of the calendar year ^[8]. A 2029 number is a year-end reading.

In my view the missing submission is a small statistical fact with a real reporting consequence. The June-to-September change in the 2029 median compares a two-dot average against one participant's number. That is a different thing from seventeen or eighteen people moving together. If the absent participant sits near the middle of the distribution, the eighteen-count median would land within 12.5 basis points of the seventeen-count one and the distinction is bookkeeping. The published table settles it: an unchanged 2029 median from June would mean the parity and the trim both wash out.

These are individual projections. The Committee did not make them. Each set rests on that participant's own assessment of appropriate policy, which the release defines as the path each deems most likely to satisfy his or her individual interpretation of the statutory mandate ^[12]. The longer-run values are where each expects the variable to converge in the absence of further shocks ^[13]. Longer-run core PCE is not collected ^[10], so a terminal real policy rate built from this release has to use the headline PCE longer-run number. The release text sets out those definitions, the participant counts and the rounding convention; the values themselves sit in Table 1 and in the dot chart ^[1].