Saturday, October 22, 2022

Nowcasts - Helpful or just extrapolation?

“The investment management industry thrives on the expedient of forecasting the future by extrapolating the past.”

“Is this a forecast or a nowcast?” If the former, and if it is correct, it may have material value to an investor because current market prices may not reflect the insight. If it is a nowcast, it will be of no use if it is correct—the nowcast is already reflected in current prices—and will hurt us if it is not correct."

- from "Where’s the Beef?" by  of 

Arnott provided a harsh critique of nowcasting, and it is appropriate. Nowcast has become a buzz word on Wall Street and with central banker for its humble roots with the field of meteorology.

Nowcasting is a combination of two terms: “now” and “forecasting.” Economic indicators are often released with a considerable delay, but there is usually intense interest in knowing what a given indicator is before its official release. Nowcasting refers to the prediction of the present, the very recent past, and the very near future—in essence, trying to fill in that missing data point with an accurate estimate. Often, nowcasts take advantage of higher-frequency data that are informative about some other economic indicator to predict what the next release of the indicator of interest will say. -from Cleveland Fed 

It is important to first determine whether a macro forecast is just an extrapolation given another name. There is nothing wrong with employing extrapolative forecasts. Many managers use them all the time. Macro data do show trends and should be expected to have some memory. An extrapolation or some form of exponential moving average is a good prior no different than a view of no change can be a good prior. These simple views should be acknowledged as base cases and as an extrapolation. An extrapolative can be called a nowcast, but not something unique or special.

Nowcasts that are just extrapolative should generally be embedded in market prices. This is the fundamental difference between adaptive and rational expectations. Nowcast as forecasts should weigh or include other high frequency data to provide insight on the future. For example, PPI and market inflation breakevens can be used to predict the next CPI number along with a trend. This can be considered as a simple nowcast for the next CPI. However, this approach has been used for decades and is not something new. It can be called a nowcast, but it is just another name for a short-term forecast. 

Like in meteorology, the critical issue associated with any nowcast is the short-term and local nature of the forecast. A nowcast is not a large model looking to explain many economic phenomena. Rather the nowcast is a short-term immediate forecast of a specific event. In the case of macro data, another example could be trying to forecast the weekly initial jobless claims. It will likely be extrapolative and try to include recent data that has arisen since the last report. Given the limits and delays with data, this will be a simple focused model.

Using the word nowcast is somewhat a fad and forecasters should be careful with how they use the term, but it is a useful tool that can add value when thinking about how to forecast short-term macro variables. Just beware of false gifts given fancy names.

No comments: