Participants from across the state in a variety of industries respond to a questionnaire and report the change in a variety of indicators from the previous month. Respondents also state the likely direction of these same indicators six months ahead. April 2002 is the first report, although survey data date back to July 2001.

The survey is sent on the first day of each month to the same pool of about 200 manufacturing executives in New York State, typically the president or CEO. About 100 responses are received. Most are completed by the tenth, although surveys are accepted until the fifteenth.

For demonstration only:
Sample survey 1 page / 44 kb

Respondents come from a wide range of industries from across the New York State. No one industry dominates the respondent pool.

The survey's main index, general business conditions, is not a weighted average of other indicators—it is a distinct question posed on the survey. Each index is seasonally adjusted.

Revisions
Each January, all data undergo a benchmark revision to reflect new seasonal factors.

Seasonal Adjustment
The Empire State Manufacturing Survey seasonally adjusts data based on the Census X-12 additive procedure utilizing a logistic transformation. Since this procedure requires a lengthy data history, and the survey dates back only to July 2001, data prior to that date are obtained from the Federal Reserve Bank of Philadelphia's Business Outlook Survey. The two surveys pose identical questions during the same period to a similarly diverse group of manufacturers.

The "increase" and "decrease" percentage components of the diffusion indexes are each tested for seasonality separately and adjusted accordingly if such patterns exist. If no seasonality is detected, the component is left unadjusted. The "no change" component contains the residual, computed by subtracting the (adjusted) increase and decrease from 100. Seasonal factors are forecast in December for the upcoming year.

Data are adjusted using a logistic transformation. The not-seasonally adjusted series, expressed in decimal form (referred to as "p"), is transformed using the following equation:

X = log(p/(1-p))

The seasonal factor is then subtracted from X:

adjX = X - seasonal factor

The result is then transformed using the following equation:

SA Series = exponential(adjX)/(1+exponential(adjX))

Seasonal factors are available here
Data definitions 2 pages / 15 kb

Contacts
Tech help: ny.piwebteam@ny.frb.org or (212) 720-7737

Questions about survey/data: richard.deitz@ny.frb.org or (716) 849-5025; jason.bram@ny.frb.org or (212) 720-5651