Our team is pleased to announce the launch of a new initiative, the Adobe Digital Economy Project. The project’s core goal is to add insight to current economic discussions based on dynamic, responsive data about the digital economy. The Adobe Digital Index, a group of data scientists and researchers within Adobe, publishes research on digital marketing and the economic landscape based on anonymous data aggregated from thousands of websites worldwide. With this latest project, we are focused on shedding light on an increasingly large driver of the economy: online spending.
Exploring the Digital Economy Project
The Digital Economy Project has three key components:
- Digital Price Index (DPI): The Digital Price Index looks at inflation through the lens of digital commerce. It is based on massive aggregate and anonymized data sourced through the Adobe Marketing Cloud. Our first iteration, comprised of electronics and grocery products, includes data from 8 billion website visits and 1.4 million products sold online between January 2014 and January 2016. The methodology, discussed in more depth below, implements the gold standard of price indices: the Fisher Ideal Price Index. This implementation is made possible due to the unique capability of Adobe Marketing Cloud to amass huge amounts of data where the quantity of products sold is available as well as sale price.
- Digital Housing Index (DHI): The Digital Housing Index looks at trends in the housing market based on online behaviors. The initial DHI is based on an analysis of aggregated data from 2 billion visits to US housing-search websites between December 2014 and December 2015.
- Job Seeking Index (JSI): The Job Seeking Index enhances insights in to national employment data based on digital job-search behaviors. The initial JSI is based on analysis of data aggregated by Adobe Marketing Cloud from 1 billion visits to US employment-search websites and top employer-career pages from December 2014 through December 2016. The JSI is a significant index given that Adobe Marketing Cloud currently powers 20 of the top-30 US employers.
The dataset for each index is updated daily and is based on transactions processed and behavior measured through the Adobe Marketing Cloud. Future findings, as well as additional product categories, will be released regularly. All data remains fully anonymous and confidential and is analyzed only in aggregate.
Ensuring Methodological Integrity
Quantifying the digital economy is challenging, and Adobe recognizes an opportunity to add another perspective to the existing conversations regarding inflation and the economy. Consequently, ensuring the methodological integrity of our data was an overarching objective. We worked with two of the leading economists in the United States to create the indices. Our goals in collaborating with them were to take in to consideration as many factors as possible, to ensure methodological soundness, and to provide a deeper level of insight in to digital transactions than is currently publicly available.
The experts we worked with are Dr. Austan Goolsbee and Dr. Pete Klenow who recognize the potential found in the vast dataset Adobe possesses. Dr. Goolsbee is the Robert P. Gwinn Professor of Economics at The University of Chicago’s Booth School of Business. He served as President Obama’s Chairman of the Council of Economic Advisers. Dr. Klenow is a Professor in the Department of Economics at Stanford University. He has served as both a Senior Economist and Visiting Scholar at the Federal Reserve Bank, and he is a member of the editorial boards of publications such as Econometrica, American Economic Review, and the Quarterly Review of Economics.
A Closer Look at the Digital Price Index
Let us explore how the Adobe Digital Economy Project will be valuable by taking a closer look at the Digital Price Index. Currently, the most frequently cited source of economic data on inflation is the US Bureau of Labor Statistics Consumer Price Index (CPI), which is the standard for measuring inflation. The CPI methodology is based on a finite dataset comprised of approximately 83,000 products. The list of products — and their respective weights within the CPI — is obtained through consumer surveys administered every four years for each product category. Then, each month, US Bureau of Labor Statistics research assistants visit retail outlets to manually record the current prices for those products, which is generally a handful of products in each outlet.
When the Adobe Digital Economy Project team set out to develop the Digital Price Index, our goal was to leverage, in-depth, Big Data on both product categories and quantities sold through online channels as tracked through the Adobe Marketing Cloud. As stated above, the Fisher Ideal Price Index is considered the gold standard for calculating inflation. Nonetheless, the index has been largely aspirational, as it requires measurement of the quantity sold of each product in every time period instead of through surveys administered once every four years. This data has been difficult to capture consistently — and at the immense scale needed — in an efficient, affordable way. Now, with the accuracy and massive data-gathering capabilities of Adobe Marketing Cloud, we are able to aggregate that data efficiently and accurately. Today, we track 1.2 million products across categories like electronics and groceries. Due to the dynamic nature and direct measurement of this data, we are able to accurately and speedily identify pricing trends, inflation, and more.
The Digital Price Index also allows us to responsively track rapidly shifting consumer preferences. In the category of electronics, for example, we estimate that 80 percent of online spending goes toward products that have been in existence for less than 12 months. Even in the category of groceries, 16 percent of monthly online spend goes to products that have been on virtual shelves for less than a year. In other words, the capability of the DPI to do daily data collection and updates enables us to track changes closer to the timeframe in which they are happening. These insights can help drive business and public-policy decisions.
The DPI can also demonstrate how prices shift during specific time periods. Consider trend data from the category of computers. Between January 2015 and January 2016, the Adobe DPI shows cumulative deflation of 13.07 percent. Yet, for the same period, the CPI has reported cumulative deflation of 7.09 percent. In a closer look at month-over-month price changes in the November and December timeframes each year, the DPI is able to show price decreases informed by Black Friday, Cyber Monday, and other holiday-related promotions. As seen in the graph below, the DPI highlights trends that previously went undetected: a trend toward deeper holiday-season discounts, year-over-year, is reflected. With its dynamic data, Adobe can pinpoint discreet, seasonal price fluctuations to better identify both macro and micro trends.
There are caveats to note about the Adobe Digital Economy Project. It is intended as an addition to existing economic-data sources not a replacement. And while the Project’s indices incorporate billions of data points, that information represents only one very specific slice of the economy: digital transactions. In specific categories, such as groceries, digital transactions represent a very small but rapidly growing segment of the market. The customers who purchase these products online tend to skew affluent and urban, so they cannot be taken to represent a median demographic. However, in the proper context, our aim is that the Digital Economy Project bring additional insights about the online economy to the collective knowledge base.
Moving forward, the Adobe Digital Index team will periodically release findings from the Project. We also anticipate expanding the indices further by, for example, developing additional categories for the DPI.
Luiz Maykot, Tyler White, Data Scientists, Adobe Digital Index