Labour hoarding resolves DSGE puzzle on procyclical productivity
Integrating variable labour utilisation and employment adjustment costs into the Riksbank's MAJA model allows dynamic stochastic general equilibrium frameworks to replicate procyclical labour productivity following monetary policy shocks, according to a research memo published in September 2026.
Breaking the one-for-one link
Standard macroeconomic models assume firms adjust labour purely along the extensive margin of headcount, implying hours worked move in lockstep with output.
By introducing quadratic employment adjustment costs alongside flexible labour utilisation into the MAJA model, intermediate firms absorb short-term demand variations by altering worker effort rather than payrolls.
Following a one-percentage-point monetary policy rate hike, hours worked respond more gradually than gross domestic product.
The calibrated specification yields a peak labour productivity drop of 0.68 percent, closely tracking the 0.56 percent decline estimated by a Swedish Bayesian vector autoregression for the 1995 to 2024 period.
Tighter fit for Okun dynamics
Bayesian estimation over quarterly Swedish, euro area and US data yields a posterior mode of 0.056 for labour utilisation costs and 13.9 for employment adjustment costs.
This dual structure lowers the model-implied Okun coefficient from 1.67 in baseline MAJA to 0.94, bringing the ratio of peak unemployment to output response closer to the empirical Swedish benchmark of 0.60. Additionally, accounting for labour hoarding sharply reduces the estimated persistence of temporary technology shocks from 0.81 to 0.14.
Structural fix with empirical limits
The framework repairs a major theoretical blind spot in central bank business cycle modeling.
Yet treating unobservable worker effort as a cost-absorbing sponge remains an empirical shortcut.
Until micro-level effort data can be verified, general equilibrium calibrations will remain somewhat speculative.
Source: Staff Memo: A DSGE Model of Labour Hoarding
IN: