Multipliers and bad statistics, once again

Economists, a large group of them, want multipliers and adders to be greater than one, such that a positive increase in government debt yields a greater positive increase in total GDP.

Does an increase in government spending add or multiply? What do I mean? In a fully functioning positive growth economy, the government sector and the private sector share transaction space efficiently, in that scenario, any exogenous shock to their external terms of trade will be equally felt by government and private sectors. The multiplier is one. In an economy that has suffered a sudden crash, the assumption is that government and the private sector have become disconnected, any shock to gdp will effect the two sectors independently as their external terms of trade dominate their internal terms of trade. In this scenario, the economy is operating way below output, the government and private sector can be treated independently.

Shocks can be additive when operating below output capacity, they assume. Thus  the DSGE assumption that adding log growth is the same as multiplying growth. So, two regimes, one way below output capacity and we can assume log linear; and another near output capacity when the log linear assumption is invalid.

When is the phase shift between the two regimes? Evidently, the phases shift is placed where they get confirmation bias. So we end up with charts that contain both positive growth economies and zero growth economies. They make assumptions about the latter and apply them to the former. When does the phase shift occur? Right after the dead cat bounce, about four years ago. Watch the yield curve. it recovers its nominal shape about 8 months after the crash.

The log linear  assumption includes negative and positive multipliers, in a disconnected economy, under the log linear assumption, both negative and positive issues with government spending become visible.  But the Keynesians mix the two modes to confirm their bias.  Hence, their multiplier studies can  never include modelling errors, the stability of their measurements are way off in the low growth mode, and their assumptions patently wrong in the high growth mode.