Cookies on Macro by Mark.We use essential cookies for sign-in, locale, and security. With your permission, we also use optional cookies for preferences, for analytics, and for advertising measurement with the Meta (Facebook) Pixel. Read our cookie policy.
Regime-switching macro system where latent states govern recession and expansion dynamics through transition probabilities.
Graph workspace
The defining output of a Markov-switching model is the smoothed regime probability. Plotting against observed data shows how the model classifies each period and how closely the statistical regimes match economic narratives.
Detected recessions
Number of quarters classified as recession (smoothed probability above 0.5).
64.00
quarters
Expected recession duration
Average number of quarters a recession lasts under the transition probability set on the page. The toy filters with the true parameters; it estimates nothing.
5.00
quarters
Ergodic recession fraction
Long-run fraction of time spent in recession. For the US, typically 10-20%.
0.2000
proportion
Regime separation (Mahalanobis)
How distinguishable the two regimes are. Below 1 signals weak separation and unreliable classification.
0.4351
standardized distance
Mean growth gap (recession - expansion): -1.2
The gap between average growth in recessions and expansions; a wider gap makes recessions deeper and clearer.
Wider gap produces sharper regime separation - smoothed probabilities snap more decisively to 0 or 1.
Volatility ratio (recession / expansion): 2.4
How much more the data bounces around in recessions than in expansions.
Higher ratio produces wider scatter in recession episodes and more confident recession classification.
Recession persistence: 0.8
The chance the economy is still in recession next quarter, given it is in one now.
Higher produces longer recession episodes and wider smoothed-probability spikes.
Expansion persistence: 0.95
The chance the economy is still expanding next quarter, given it is expanding now.
Higher produces longer gaps between recessions. Lower makes the economy cycle rapidly.
How many quarters of data the simulation produces.
Longer samples show more recession episodes and reveal the stationarity of the regime process.