Blog home

Improving Economic Forecasting with Alternative Data

January 30, 2021

The past year has been a breakthrough year for the use of alternative data in large financial institutions. We recently partnered with economists from Goldman Sachs on a webinar highlighting how alternative data helped them capture both the initial COVID-19 collapse and the recovery in a better way than official economic data did. Here’s a recap.

How Alternative Data is Changing Economic Forecasting

Alternative data includes credit card spending data, geospatial data, and other datasets that can be applied to financial forecasting and modeling. Alternative data is often much larger than the datasets produced by governmental statistics agencies (known as traditional data).

There are several advantages of using alternative datasets. For example, they can be significantly more timely than traditional data. Alternative data providers typically update their datasets more frequently than traditional data providers, and in a year as tumultuous as 2020-2021, these updates proved crucial to ensuring the accuracy of financial models.

Using alternative data related to cargo ships, Goldman Sachs created an export tracker that gave them a more direct read on the value of exports on a much shorter timeline compared to traditional data. Additionally, alternative datasets tend to have a greater geographic variation than traditional datasets (which can be limited by the governing bodies that produce them) and provide more detailed information. This greater granularity helps economists to isolate the impacts of economic shocks or show sudden economic shifts in the economy.

There are also disadvantages of alternative data: for example, data can be noisy which limits its usability. A lot of alternative datasets come with significant startup costs as teams learn some of the quirks of a new dataset and adjust to them. And of course, large quantities of alternative datasets require big data cloud infrastructure to access and manage the data, which also requires significant startup costs. To learn more about ways to work efficiently with large alternative datasets, check out our webinar and blog post with tips and tricks from Jefferies Head of Data Strategy and Equity Research. 

The Impact of COVID-19 on Economic Forecasting

The pandemic has accelerated the drive towards alternative data in many organizations, causing a demand for more timely and accurate data. Because of this, Goldman Sachs developed the following tools addressing the economy during the pandemic using a combination of existing alternative datasets, traditional data, and newly created datasets from which insights about the direction of the economy were extracted:

Business bankruptcies tool

A bankruptcy-tracking tool compares existing alternative data on legal filings of large companies with data from the American Bankruptcy Institute..

Urban-rural divide tool

Combining alternative data with traditional data, this new tool uses mobility data with census data on county population densities to show how mobility differs across small, medium, and large cities or towns.

Consumer spending tool

Consumer spending is an important economic forecaster. Collecting multiple alternative data series from third parties based on foot traffic or credit card spending, this tool measures consumer spending on a high-frequency basis with a short time lag.

Measuring the impact of unemployment benefits

This tool tracks the spending habits of people receiving unemployment benefits, compared to spending habits of people who did not get unemployment benefits, giving an overall picture of how the pandemic affected consumer spending.

 Measuring sensitive segments to the economy

Using SafeGraph data, this tool places geofences around four large universities to keep track of businesses inside them and compare that to business activity in the same timeframe and same state to analyze the impact of COVID-19 in college towns without students.

Consumer sentiment analysis

A Twitter sentiment index using economy-related tweets shows that consumer sentiment correlates extremely well with consumer spending during the pandemic. 

Sourcing the Right Alternative Data

In addition to being timelier and more detailed, alternative data can make forecasts more accurate. But to source the right alternative data, it needs to be back tested against traditional data. For economic forecasting, a standard baseline would be Bloomberg’s Consensus which is the median forecast from a panel of professional forecasters. 

It’s possible to test multiple alternative data series derived from completely different methods, which will result in more confident forecasts about the economy. To increase the confidence in these measures, the dataset can be back tested at a more granular level using some additional variation that is common with alternative datasets (for example, at the industry level or different geographical scales).

How Geospatial Alternative Data Gives Economists an Edge

Because everything has a location, alternative data is often geospatial data, like SafeGraph Places data. SafeGraph is a founding partner of Placekey, which is a universal identifier for every place in the world and serves as a join key which makes it easier for economists to tie different data sources together.

SafeGraph data provides the building blocks for spatial analytics. SafeGraph is a leader in location data, counting over 8 million POIs in the US, Canada, and the UK. By tracking both a combination of branded and non-branded places, SafeGraph aims to include every place that may impact a business strategy, such as stores, warehouses, and parks. The company also provides geometry data with precise building footprints for each POI. SafeGraph’s mobility products have been particularly helpful during COVID-19, showing how people interact with the physical world through aggregated and anonymized mobility data at different scale levels.

To browse SafeGraph and other alternative datasets for economic forecasting, visit the SafeGraph Shop

Browse the latest

Questions? Get in touch with our team of data experts.