Skip to Content Skip to Navigation

Index of Coincident Economic Indicators (ICEI)

Technical Notes

In general, when the Index of Coincident Economic Indicators (ICEI) demonstrates a prolonged upward trend, the New York State economy is believed to be expanding. Conversely, a sustained downward trend indicates a probable recession.

Index Construction

National Bureau of Economic Research economists Arthur Burns and Wesley Mitchell conducted early work in the area of business cycle measurement in the 1940s. James Stock of Harvard University and Mark Watson of Princeton University, who built on the work of Burns and Mitchell using techniques of modern time-series analysis, developed the basic model for constructing a coincident index for the U.S. economy. The basic assumption underlying the department's ICEI is that each of the four economic indicators contains some useful information about the economy. However, no single indicator provides a clear and immediate signal with all of the information required to determine where the state's economy is within the business cycle.

The department's monthly ICEI is based on a "state space" model. In the "state space" model, the common co-movements in the four economic indicators typically share the influence of a single, unobserved factor, referred to as the state of the economy; this is what the coincident index is attempting to measure.

The "state space" model relies upon a statistical technique called the Kalman filter, which estimates the optimal weights of the four economic indicators in any given month. In the Stock and Watson model, weights for the four economic indicators are based on their past statistical relationship with the unobserved state of the economy element and the inter-correlations among the four indicators over time. Before the advent of Stock and Watson's formal statistical model, coincident models such as those published by the U.S. Department of Commerce, did not attempt to estimate optimal weights each month. Instead, economic indicators that displayed greater volatility received less weight in the model, and all model weights were forced to sum to 1.

Alan Clayton-Matthews of the University of Massachusetts-Boston developed the software program used in the estimation of the department's monthly ICEI. The Clayton-Matthews program employs maximum likelihood and Kalman filter techniques to filter each constituent of the coincident indicator to eliminate idiosyncratic noise. Filtering produces an index that best estimates the common co-movements of all four economic indicators. This index of the common trend is the estimate of the index of coincident economic indicators.

Estimated Model

All four economic indicators comprising the ICEI model are seasonally adjusted using the X-11 ARIMA procedure developed by the U.S. Census Bureau. Sales tax data are deflated using the U.S. Consumer Price Index. The tax data are also smoothed with a Kolmogorov-Zurbenko filter, which is similar to a moving average with polynomial weights.

It is important to note that the entire history of the ICEI may be revised each month. There are three sources that contribute to such revisions:

  1. The first source is new data that were not available at the time of the ICEI's release.
  2. The second source of revision comes from revisions to existing economic data used in the ICEI model.
  3. The third source of revision is inherent to this type of model. As noted above, the ICEI is derived from the Kalman filter, meaning that the estimate of the index (i.e., the state of the economy) in any particular month is the model's best assessment given all past and future observations.

Thanks for the feedback! It will help us improve your experience.