“simple” Phillips curve relationship does not take into account

The labour-market reforms
of the Agenda 2010 have rendered the labour market much more flexible and
probably caused a flattening of the Phillips curve, which means that any decline
in unemployment will lead to lower wage increases than in former years.
Structural unemployment probably declined as well. The EC estimates that the
NAWRU (non-accelerating wage rate of unemployment), i.e. The level of
unemployment which must be undershot for wage growth to pick up, has
dropped from more than 9% at the beginning of the millennium to less than 4%
by now.

Actual unemployment was above NAWRU in the last few years and
only converged to it in 2017, dropping to an annual average of 3.8% (ILO
definition, in % of the total workforce). Such estimates are to some extent
tautological: If, under this approach, wage moderation cannot be explained by
other, identifiable shocks, the unemployment threshold must have dropped.

In reality, however, unusually low inflation and the related gains in purchasing
power probably contributed to wage moderation as well.