Our forecasting model features three key components whose interaction yields a characterization of GDP growth over alternative phases of the business cycle, and enables us to anticipate movements in the economy between phases.

The first component of the model is a variable referred to as a "tension index". This is defined as a sum of past discounted discrepancies between actual GDP growth rate and an estimated "long-term sustainable" growth rate. Long-term sustainable growth corresponds to the growth rate of potential GDP, and is currently estimated as approximately 3% annually.

It so happens that actual GDP growth tends to alternate between periods during which it consistently falls short of its sustainable rate (sluggish-growth period), and periods during which it consistently exceeds its sustainable rate (robust-growth period). In turn, the tension index tends to undergo sustained periods of general contraction and expansion. It is also the case that movements between sluggish and robust-growth regimes are fairly predictable: the farther the index drifts away from zero, the more likely a transition between regimes becomes. Finally, the onset of sluggish-growth regimes for the tension index systematically precede the occurrence of recessionary phases of the business cycle, thus making the tension index a useful leading indicator of the occurrence of recessions.

To view the behavior of the tension index over time, click here. The thin vertical lines in the figure denote peaks and troughs in the business cycle (as defined by the NBER); periods between peaks and troughs coincide with recessions.

The second component of the model is designed to exploit predictable patterns of behavior observed in the tension index in order to anticipate movements in GDP growth and the occurrence of business-cycle turning points. Specifically, it is designed to anticipate transitions in the tension index between sluggish and robust-growth regimes regimes. It does so by mapping the absolute value of the tension index into probabilities of the occurrence of regime changes: as the tension index increases in absolute value, the probability of a regime change also increases. The identification of regime changes serves to signal anticipated changes in the general pattern of GDP growth, and shifts between phases of the business cycle.

To view the occurrence of regime changes identified by the model, click here.

The third component of the model is designed to capture the evolution of GDP growth within any given tension-index regime. This evolution is determined by the interaction of two conflicting forces. The first force, referred to as an "Error Correction Mechanism" (ECM), tends to keep GDP growth in line with its long-term sustainable rate. In each period, past divergences from the sustainable rate are partially corrected for (in some fixed proportion of past divergence as well as of the past value of the tension index). The second force consists of a progressive "stochastic drift" away from the sustainable growth rate. The characteristics governing the speed at which this drift operates are specific to each tension-index regime.

To view the entire history of the stochastic drift component (solid line), click here. The tendency of actual GDP growth to track the stochastic drift component over time is depicted by the corresponding line. (Technically, this line represents deviations of GDP growth from the ECM component.) To view its more recent behavior click here.

In a nutshell, within any given regime, the model characterizes GDP growth as tracking initially its long-term sustainable rate (according to the ECM force). However, over time the stochastic drift component deviates exponentially from the long-term sustainable rate, and this behavior ultimately dominates the ECM force. In turn, as the regime evolves the tension index ultimately grows in absolute value to the point at which a regime change is triggered, thus launching GDP growth along a new trajectory.