Do not Forecast what you can Nowcast

Then, we forecasted. Now, we nowcast. It enables us to understand economic crises much better than ever before.

Forecasting is a different thing to different people. Ask an astrologer, and it is the palm of your hand. Ask Google, and it “is a technique that uses historical data as inputs to make informed estimates that are predictive in determining the direction of future trends.” Ask an FX-market forecaster, and it is “something of a joke: you’re forecasting one thing based on another thing.”

Consider, however, what Saeed Amen opines: (why do) “we all want to forecast what’s around the corner? (..) Perhaps a more pertinent question, which we might seek to ask is the following, is this actually a corner?”

When a man with over a decade’s worth of experience in running systematic trading models at Lehman Brothers and Nomura (essentially, a forecaster), asks this, it’s not just that irony abounds. It is, perhaps, the signalling of an era of change, i.e. a transformation from the era of forecasting to the era of nowcasting.

The Era of Now

Nowcasting, born as a response to the dot-com bubble and the 2008 financial crisis, revolves around understanding the ‘now’ and places stress on data indicators that explain the present over anything else. It’s roots were crucial in disseminating the overreliance on past economic data, which is often subject to publication lags of several months, exposing organisations to several risks or missed opportunities.

Nowcasting has, once again, proved to be especially relevant over the past fifteen months or so – a time where most models using past data have heavily lagged the gyrations of the economy caused due to the pandemic and ensuing lockdowns.

Nowcasting models essentially make use of millions of recently-observed data points in the form of unstructured datasets to measure/observe a target variable either directly (for example, a basket of goods to measure inflation) or indirectly (like using parking lot occupancy as a proxy for estimated revenue) to make short-term predictions.

In finance, such techniques have proved to be much more reliable and accurate compared to most current econometric forecasting models, according to experts. Cornell University’s Professor Marcos Prado, for example, opines that nowcasting inflation indicators “based on web-scraping millions of online prices every day is much more accurate than forecasts derived from convoluted econometric models.” Similarly, other financial metrics such as retailer earnings can be now casted using variables such as satellite images for parking lot occupancy, email receipts, etc, and industrial production using cargo shipments, electricity consumption or engineering datasets.

Nowcasting in the face of Economic Uncertainty – Less is More

The ability to nowcast becomes particularly important, as McKinsey puts it, in the face of major economic uncertainty: “the ability to gather and interpret information quickly is crucial for decision makers, especially when a crisis turns into a recovery, or vice versa. Those able to understand and react to the evolving situation quickly and appropriately will not only survive but will also create a more resilient organization.”

It is, however, prudent to mention that nowcasting isn’t bereft of econometric complexities either. In fact, typical economic nowcasting models incorporate up to 50 indicators for the economy, based on a variety of data and assumptions. These bring with it complex inter-relationships between variables as well. The use of alternative high-frequency variables such as air pollution levels, online searches or footfall, currently in use in modern nowcasting models, instead enable much more robust decision-making.

Along the same lines, reducing the number of variables by choosing only the most significant ones – a combination of key performance indicators (KPIs) and high-frequency explanatory variables – may prove to be the most consistent way to analyse the economy. McKinsey analysts opine that “this new approach to nowcasting makes it easier to interpret estimates, understand structural breaks, and provide up-to-the-moment information”

Thus, given that nowcasting provides real-time analyses, it enables us to understand economic crises much better than ever before. In fact, a modified approach to modelling, stressing on country- and industry-specific variables allow not only to make a model lighter but also more accurate.