Sunday, February 9, 2020

Software Can Eat the World Even if its Direct GDP Contribution is Small

Some argue that information technology now changes everything; others argue IT cannot be that big a deal, as it represents such a small percentage of total U.S. gross domestic product, for example.  But that matters when observers try and estimate the impact of industries on the overall economy. 

In principle, one can argue that electricity, the internal combustion engine and information technology or communications are important only as their contribution to GDP suggests. Others might argue that misses the point. Innovations such as electricity are essential inputs to most other parts of the economy, and therefore have importance far beyond direct revenues.

But measurement alone is complicated, as there are methodological issues, for example.

Industries have every reason to inflate their economic impact. Furthermore, if one adds up all the claimed economic impact, the number is greater than the actual stated GDP. The reason is that multiple inputs (electricity, fuel, transportation, warehouse operations, information technology, communications, marketing and advertising all contribute to the defined output of any single “industry.”  And each industry can rightfully claim that economic activities in the infrastructure are part of the total economic contribution made by each industry.

In other cases all we can measure are expenditures by customers and suppliers, which means there is some risk of double counting, as, in principle, all producer costs are recouped by retail sales to actual users and customers. 

Finally, it is next to impossible to measure quality improvements. Computing, communications and device prices might go down, even as value and capability go up. We cannot measure qualitative changes using our quantitative methods. 

The U.S. electrical energy industry represents, in some analyses, six percent of gross domestic product. That seems too high. The U.S. Energy Information Administration estimates total expenditures on energy--including natural gas, fuel and electricity--of close to seven percent. So it is doubtful electricity as such represents more than a few percent of GDP. 

We see the same issue with information technology, said by some to represent about 2.3 percent of U.S. GDP (six percent of real inflation-adjusted GDP, some say). Other estimates have U.S. “telecommunications revenue” at about 3.5 percent of GDP.  

Likewise, the U.S. Bureau of Economic Analysis suggests “digital economy” category that represents perhaps six percent of U.S. GDP. That definition includes computing hardware and software, communications equipment and services, data centers and internet of things, collectively. 


By definition, none of those contributing industries can represent more than a couple percent of total. But industry size does not capture total economic value. In other words, it is still possible to argue that  software is eating the world and yet still attribute relatively little direct value to the software or computing industries.

No comments:

Many Winners and Losers from Generative AI

Perhaps there is no contradiction between low historical total factor annual productivity gains and high expected generative artificial inte...