Liberty Street Economics

« | Main | »

October 6, 2022

Computer Saturation and the Productivity Slowdown

One of the current puzzles in economics is the recent worldwide slowdown in productivity, compared to the late 1990s and early 2000s. This productivity loss is economically large: if productivity growth had stayed at the same level as in 1995-2004, American GDP would have increased by trillions of dollars. In this post, I discuss a new paper that links this productivity slowdown to saturation in electronics adoption across most industries. I show that most of the productivity growth from electronic miniaturization is concentrated between 1985 and 2005.

Declining Productivity and Moore’s Law

In the U.S., average labor productivity growth in the 1995-2004 period was 2.85 percent per year. This productivity growth significantly declined in the following decade, to an average of 1.27 percent per year in the 2005-2015 period. This pattern is not unique to the United States. For a panel of OECD countries, Syverson (2017) shows that labor productivity growth fell from 2.3 percent in the 1995-2004 period, to 1.1 percent over the 2005-2015 period.

I propose that saturation of electronics in other industries may explain the dynamics of productivity growth. When computers first were introduced, they were very large and heavy, and were only used in very specialized applications. Due to constant improvements in semiconductor manufacturing technology, electronic transistors shrank by 50 percent every eighteen months, a trend known as Moore’s Law. As computers and electronics decreased in size and weight, they were adopted as inputs by more and more industries—including virtually all manufacturing and service industries—leading to increased productivity across the economy. In Azar (2022), I use a new data set of products’ weights to estimate the effect of Moore’s Law on productivity growth.   

Electronic Miniaturization Leads to New Product Combinations

To estimate how electronic miniaturization affects productivity growth, I develop a new model where firms have different methods or  “recipes’’ to make their products, and the productivity of each method depends on the combination of inputs used in production. For example, there are many different ways to produce a car, depending on which materials are used. A car from the 2020s with thousands of microchips and a carbon-fiber alloy frame will be much safer, comfortable, and efficient than a car from the 1950s that has no electronic components and has a steel frame.

In this model, there are physical limits on which inputs can be combined to make a product. The reason why cars or industrial machines from the 1950s did not rely on electronics isn’t because computers didn’t exist, but because they were too large and heavy to be practically used in production. This is illustrated below.

Note: When manufacturing a car, the total sum of component weights cannot exceed the weight of the car, which in this figure is 4,000 pounds. Small electronic components, such as modern GPS devices (0.5lbs), cameras (0.06 lbs.) and Electronic Control Units (2 lbs.) all fit comfortably inside the car. However, state-of-the-art supercomputers (5,500 lbs.) cannot be incorporated into the car without violating the weight constraint.

I use a new data set of products’ weights from IHS Markit to determine the median weight of each industry’s product. I combine this with supply chain tables from the Bureau of Economic Analysis (BEA) to determine how much each input’s weight contributes to the final product’s weight. With this combined data set, I can simulate what happens to the supply chain as the size of electronics and semiconductors decreases.   

As electronics and computers decrease in size, the number of possible input combinations used by different manufacturing and service industries increases. The chart below shows the log-change in the number of possible input combinations for the average manufacturing and the average non-manufacturing industry. From the chart, we can see that the number of new possible combinations for manufacturing starts increasing in the 1960s, peaks in the 1970s, and drops off slowly starting the 1980s through the early 2000s. In contrast, the number of new possible combinations for non-manufacturing industries starts rising in the early 1980s, peaks in the late 1990s and early 2000s, and declines since then—with a small bump in the early 2010s.

Number of New Combinations

Source: Author’s computation from PIERS data set.

These patterns showing the adoption of new input combinations arising from electronic miniaturization match the historical adoption of computers and electronics in different industries. Manufacturing industries, such as airplane manufacturers or industrial machine producers, were early adopters of computers and electronic components. Even though computers of the time were large, the machines being produced at the time were large enough to fit them.

At the same time, computers were only widely adopted in non-manufacturing industries—such as finance, services and retail—after the introduction of the IBM and Apple personal computers (PCs) in the 1980s. PCs allowed a large number of desk workers to have access to their own device, instead of having to rely on a company-wide mainframe. The introduction of laptops, smartphones and tablets, allowed service providers on the field to carry a computing device on them at all times. Furthermore, the miniaturization of electronics led to formerly “dumb” devices such as cameras and medical equipment to have their own computational abilities.

New Product Combinations Lead to Increased Productivity

So far, we have seen that electronic miniaturization has led to an increase in the number of feasible combinations of inputs, both for manufacturing and non-manfacturing industries. But does this increase productivity? I show that this is indeed the case, by regressing industry-level productivity on the number of feasible input combinations made possible by electronic miniaturization. The regression results imply that a 1 percent increase in input combinations leads to a 0.004 percent increase in industry productivity.

I then use this estimated regression coefficient—together with the time series on the number of feasible combinations—to estimate the effect of Moore’s Law on each industry’s productivity. Finally, I aggregate the industry-specific estimates to obtain an estimate of how electronic miniaturization affected the entire U.S. economy over time. The results are shown in the next chart. The yellow line shows the realized productivity trend, while the blue line shows the productivity attributable to Moore’s Law. The red line shows another measure, which does not take into account productivity spillovers between different industries. In total, about 11.74 percent of all productivity gains between 1960 and 2020 can be attributed to electronic miniaturization.

Productivity Growth Attributable to Electronic Miniaturization

Source: Author’s calculations based on Port Import/Export Reporting Services, Bureau of Economic Analysis, and NBER-CES Manufacturing Industry Database.

Most of these productivity gains are concentrated in the 1985-2005 period, when computers and electronics were incorporated into virtually every industry. During this period, Moore’s Law accounts for 14.22 percent of all productivity growth. An important observation from the chart is that, after the mid-2000s, productivity contributions from Moore’s Law become negligible and overall productivity declined. Before 1985, computers were still too large to make a significant contribution to the performance of most services and products. After 2005, computers had become so small that they had already become essential in almost every industry, and any subsequent adoption since then has only led to incremental improvements in productivity. This suggests that a large part of the post-2005 productivity slowdown can be explained by the saturation of electronics adoption. Once electronics were small enough to be used in almost all industries, their effect on productivity vanished.

Photo: portrait of Pablo Azar

Pablo Azar is a financial research economist in Money and Payments Studies in the Federal Reserve Bank of New York’s Research and Statistics Group.

How to cite this post:
Pablo Azar, “Computer Saturation and the Productivity Slowdown,” Federal Reserve Bank of New York Liberty Street Economics, October 6, 2022, https://libertystreeteconomics.newyorkfed.org/2022/10/computer-saturation-and-the-productivity-slowdown/.


Disclaimer
The views expressed in this post are those of the author(s) and do not necessarily reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System. Any errors or omissions are the responsibility of the author(s).

About the Blog

Liberty Street Economics features insight and analysis from New York Fed economists working at the intersection of research and policy. Launched in 2011, the blog takes its name from the Bank’s headquarters at 33 Liberty Street in Manhattan’s Financial District.

The editors are Michael Fleming, Andrew Haughwout, Thomas Klitgaard, and Asani Sarkar, all economists in the Bank’s Research Group.

Liberty Street Economics does not publish new posts during the blackout periods surrounding Federal Open Market Committee meetings.

The views expressed are those of the authors, and do not necessarily reflect the position of the New York Fed or the Federal Reserve System.

Economic Research Tracker

Image of NYFED Economic Research Tracker Icon Liberty Street Economics is available on the iPhone® and iPad® and can be customized by economic research topic or economist.

Economic Inequality

image of inequality icons for the Economic Inequality: A Research Series

This ongoing Liberty Street Economics series analyzes disparities in economic and policy outcomes by race, gender, age, region, income, and other factors.

Most Read this Year

Comment Guidelines

 

We encourage your comments and queries on our posts and will publish them (below the post) subject to the following guidelines:

Please be brief: Comments are limited to 1,500 characters.

Please be aware: Comments submitted shortly before or during the FOMC blackout may not be published until after the blackout.

Please be relevant: Comments are moderated and will not appear until they have been reviewed to ensure that they are substantive and clearly related to the topic of the post.

Please be respectful: We reserve the right not to post any comment, and will not post comments that are abusive, harassing, obscene, or commercial in nature. No notice will be given regarding whether a submission will or will
not be posted.‎

Comments with links: Please do not include any links in your comment, even if you feel the links will contribute to the discussion. Comments with links will not be posted.

Send Us Feedback

Disclosure Policy

The LSE editors ask authors submitting a post to the blog to confirm that they have no conflicts of interest as defined by the American Economic Association in its Disclosure Policy. If an author has sources of financial support or other interests that could be perceived as influencing the research presented in the post, we disclose that fact in a statement prepared by the author and appended to the author information at the end of the post. If the author has no such interests to disclose, no statement is provided. Note, however, that we do indicate in all cases if a data vendor or other party has a right to review a post.

Archives