Fig 1.
An example of a rule, modeling the production of a good at a certain rate.
Fig 2.
Stylized stock-flow structure following Caiani et al. [45].
Arrows point from paying to receiving agent.
Fig 3.
Note: The Figure shows the provenance of the composed macroeconomic agent-based model “Mak(h)ro_0” including entities from theory, empirics and simulation models from literature. Section references are labeled in parentheses and citations in brackets. For a short description of the entities and activities in this Figure, see S1 Table.
Fig 4.
Update rule of firms for prices and production quantity in ML3.
Note: The Figure shows the supply mechanism rule, modeling firm production behavior as shown in Eq 1 at the transition rate defined in Eq 2.
Fig 5.
Household consumption rule in ML3.
Note: The Figure shows the demand rule, modeling household consumption or expenditure of financial wealth in exchange for homogeneous goods from firms. “Mercury” is an auxiliary agent, which otherwise has no interactions with other agents.
Table 1.
Stylized facts replicated by the model.
Fig 6.
SF1: Deviations of real GDP and real investment time series from trend: Bandpass-filtered (6,32,12).
Note: The Figure depicts deviations of average log real GDP (solid line) and average log real firm investment (dashed line) from trend. The simulation output is a representative run and shows an extract to keep the line profile more identifiable without the warm-up phase of 200 simulation periods. The grey shaded area represents the 90% confidence interval.
Fig 7.
SF2: Exponential fit of recession durations.
Note: Graphical analysis of the frequency of recession duration plotted as a cumulative probability distribution in log-log scale according to Wright [73] and Dosi et al. [69]. The points present the simulated outcome and the solid line depicts the exponential fit to the data. The recession duration is defined as the length of periods in which real GDP growth percentage change is less than zero. The simulation output is a representative run without the warm-up phase of 200 simulation periods.
Table 2.
Cross-correlation of macro variables—SF3.
Fig 8.
SF4: Pro-cyclical aggregate firm investment.
Note: The Figure depicts average real GDP as a dashed line and aggregated total investment as a solid line. The simulation output is a representative run and shows an extract to keep the line profile more identifiable without the warm-up phase of 200 simulation periods. The grey shaded area represents the 90% confidence interval.
Table 3.
Cross-correlation of credit variables—SF5.
Fig 9.
SF6: Firm debt and loan losses time series.
Note: The Figure depicts average firm debt as a dashed line and average loan losses as a solid line. The simulation output is a representative run and shows an extract to keep the line profile more identifiable without the warm-up phase of 200 simulation periods. The grey shaded area represents the 90% confidence interval.
Table 4.
Distribution related stylized facts.
Fig 10.
SF9: Individual firm investment time series.
Note: The Figure depicts investment frequencies for three randomly chosen firms. The dashed line shows the firm with identification number (id) 10, the solid line the firm with id 19 and the dotted line the firm with id 27. The simulation output is a representative run and shows an extract to keep the line profile more identifiable without the warm-up phase of 200 simulation periods.
Fig 11.
SF10: Real GDP and firm bankruptcies time series.
Note: The Figure depicts average real GDP as a dashed line and firm bankruptcies as a solid line in bars. The simulation output is a representative run and shows an extract to keep the line profile more identifiable without the warm-up phase of 200 simulation periods. The grey shaded area represents the 90% confidence interval.
Fig 12.
SF11: Inflation in relation to unemployment.
Note: The Figure shows a scatter plot of average inflation percentage change and unemployment rate (minus natural unemployment rate) with following regression feature: Natural unemployment rate (0.012−0.154x + ϵ);R2 = 0.369. The natural unemployment rate is assumed to be 0.05. The simulation output comprises 20 representative runs to increase the sample with a total number of 12,000 period observations without the warm-up phase of 200 simulation periods. The grey shaded area represents the 90% confidence interval.
Fig 13.
SF12: GDP growth in relation to unemployment growth.
Note: The Figure shows a scatter plot of aggregated real GDP and unemployment percentage change with following regression features: Unemployment rate (0.007−0.913x + ϵ); R2 = 0.295. The natural unemployment rate is assumed to be 0.05%. The simulation output comprises 20 representative runs to increase the sample with a total number of 12,000 period observations without the warm-up phase of 200 simulation periods. The grey shaded area represents the 90% confidence interval.
Fig 14.
Note: The Figure depicts average interbank rate as a dashed line and excess reserves as a solid line. The simulation output is a representative run with an interbank experiment between period 50 and 100. The grey shaded area represents the 90% confidence interval.
Table 5.
Related literature in comparison to Mak(h)ro_0.