Tot.
10.01
1.625
2.89
2010s
1
Apple
29.6
25.8
41.8
2.54
0.061
0.24
2
ExxonMobil
8.3
34.4
36.7
1.91
0.052
0.87
3
Microsoft
32.8
18.4
23.0
1.68
0.073
0.07
4
Alphabet
27.7
23.2
43.3
1.56
0.036
0.09
5
Berkshire Hathaway
15.2
13.2
6.6
1.43
0.216
0.58
Average
22.7
23.0
20.8
Tot.
9.11
0.438
1.84
Notes: Based on US-headquartered companies in Compustat. All quantities in percentage points. Cost of goods sold (COGS) adjusted for firm export shares. MV share is market value of equity divided by total US stock market value. Emp share is employment divided by total US civilian employment. MV / Emp ratio is ratio of market value share over employment share. AT&T COGS missing in 1950s, value input from 1960. Current names of firms are used for historical data (ExxonMobil, AT&T).
Table 13.2 tells a fascinating story of how the economy has changed over time. The 1950s were dominated by manufacturing and petroleum. International Business Machines, better known as IBM, appears in the 1960s. GM drops out in the 1990s just as Microsoft and Walmart enter. Google (Alphabet) and Apple appear in the 2010s. To see Amazon and Facebook you need to zoom in at the end of the 2010s, as shown in Tables 13.1 and 13.3.
You might wonder why the banks are conspicuously absent in the 2000s. Citigroup is indeed a star in the early 2000s, but it crashes so badly in 2008 that its decade average barely makes the cut. JPMorgan Chase and Bank of America appear in the top ten in 2017, as shown in Table 13.3.
ExxonMobil is the only company that has remained in the top five for seventy years.a As the mother of the Princess of Parma reminds her daughter in Proust’s In Search of Lost Time, “God in his bounty has decreed that you should hold practically all the shares in the Suez Canal and three times as many Royal Dutch as Edmond de Rothschild,” and thankfully “nothing can alter the antiquity of blood, while the world will always need oil.” ExxonMobil has changed a lot over time, however. Its employment weight has shrunk more than four times.
Do the GAFAMs Make Too Much Money?
Do the GAFAMs make excess profits? Let us focus on their profit margins. You can see in Table 13.2 that the pretax operating profit margins of the top twenty-five firms have typically been around 20 percent. Today’s pretax margins are not out of line with historical norms. What has changed, however, is the average tax rate that these companies pay. AT&T used to have a profit margin of 25 percent, but it paid a tax rate of 45 percent. In the 2010s, Apple’s margin was almost 30 percent and its tax rate was less than 26 percent. The after-tax margin has increased dramatically. But is this a general phenomenon, or is this specific to the GAFAMs?
TABLE 13.3
Current Stars at the End of 2017
Profitability (%)
MV /
Emp ratio
Share of the Economy (%)
Rank
Company
Op. Inc. / Sales
Taxes* / Op. Inc.
MV share
Emp share
COGS / GDP
1
Apple
24.9
26.4
36.5
2.92
0.080
0.37
2
Alphabet
16.9
19.7
47.3
2.46
0.052
0.15
3
Microsoft
16.8
13.9
27.6
2.22
0.081
0.09
4
Amazon
28.7
35.0
5.2
1.90
0.367
0.42
5
Facebook
12.7
18.4
105.8
1.73
0.016
0.01
6
Berkshire Hathaway
30.9
25.4
6.7
1.65
0.245
0.70
7
Johnson & Johnson
25.3
15.4
14.5
1.26
0.087
0.05
8
JPMorgan Chase
16.3
19.1
7.5
1.23
0.164
0.08
9
ExxonMobil
13.5
−43.4
26.4
1.19
0.045
0.75
10
Bank of America
12.9
17.9
7.5
1.02
0.136
0.06
11
Wells Fargo
24.6
24.0
5.9
1.00
0.171
0.05
Average
1–5
20.0
22.7
18.8
Tot.
11.23
0.596
1.03
GFAM (4)
17.8
19.6
40.8
9.32
0.229
0.61
6–10
19.8
6.9
9.4
6.35
0.677
1.64
Top 10
19.9
14.8
13.8
17.58
1.273
2.68
Notes: Based on US-headquartered companies in Compustat. All quantities in percentage points. COGS adjusted for firm export shares. MV share is market value of equity divided by total US stock market value. Emp share is employment divided by total US civilian employment. MV / Emp ratio is ratio of market value share over employment share. GFAM removes Amazon and does the calculations for the remaining four firms. *Tax rate as of 2016 because of tax changes in 2017.
FIGURE 13.1 Pretax operating profit margins
Figure 13.1 shows the profit margins of the top twenty firms in the United States (ranked by market value), the margins of the GAFAMs, and the margins of the top twenty excluding the GAFAMs. Operating margins have increased and settled at a higher level in recent years. The margins of the GAFAMs are significantly higher than those of the other top twenty firms. But the GAFAMs are not large enough to change the average much. The margins of the top twenty are rather similar with or without the GAFAMs. In all cases, we see the sharp increase in profit margins around 2000 that we have discussed earlier in the book. This increase happens with or without the GAFAMs.
The GAFAMs have extremely high profit margins, but so did many stars of the past. Their average profit margin in 2017 is 20 percent (Table 13.3), but the next five firms have an average margin of 19.8 percent. The profit margin of Apple was 25 percent in 2017, but IBM had the same margin in the 1960s and 1970s, and AT&T had a higher average margin for thirty years. The stars make money; that’s why they are stars. But the stars of today are not making much more than the stars of the past. They just keep more of it.
Even more su
rprising—and contrary to many of the commentaries—there seems to be little new about the stock market value of the GAFAMs. The market value shares of the GAFAMs are not much higher than those of past stars. The GAFAMs account for 11.2 percent of the market cap of US stocks in 2017. In the 1980s, General Electric (GE), GM, IBM, AT&T, and Exxon accounted for 9.95 percent of the market. Apple might be almost 3 percent of the market, but IBM was more than 3 percent throughout the 1980s, and ExxonMobil was 2.5 percent in the 2000s.
What the data tell us here is that the assumption that tech firms are somehow thoroughly different from dominant companies of previous generations doesn’t stand up.
This brings us to the second assumption—that the GAFAMs are so integral to the health of the US economy that they must be protected.
This is where we’ll point out something that does set these new firms apart from their predecessors: they employ few people and interact little with the rest of the business sector. Actually, I should say the GFAM, because these points do not really apply to Amazon, as we shall see.
Why Footprints Matter
Let us now look at the footprints of the stars. Footprints are important in theory, because they affect the extent to which the performance of a company affects the performance of the whole economy.
A firm is deeply integrated when it buys many of its inputs from other firms in the economy. When a firm is deeply integrated, what is good for that firm tends to be good for the economy. The old saying, slightly restated in 1953 by Charles Wilson, GM’s CEO, during confirmation hearings for Secretary of Defense, is true in that case: a star firm produces a star economy.
Box 13.1 explains why footprints matter. The punch line is that a star firm that does not interact with the rest of the economy matters less than a star firm that sits at the center of the economy.
Box 13.1. Inputs, Outputs, and Economic Footprints
A simple example illustrates why footprints matter (see Figure 13.2). Imagine two economies. Each has three firms. All firms produce output, and the GDP is the sum of their outputs. (We are using a simplified example in which relative prices do not enter.) In the first economy, firm 1 produces x1 units and firm 2 produces x2 units. Firm 3 produces q units, and total output is x1 + x2 + q. Let us use some simple numbers: x1 = 2, x2 = 1, and q = 1. GDP is equal to 4. Now suppose the productivity of firm 3 increases by 10 percent, from 1 to 1.1. What happens? GDP rises from 4 to 4.1, a 2.5 percent improvement. That’s because firm 3 accounts for one-quarter of GDP, and its productivity increases by 10 percent. The impact on the economy is one-quarter of 10 percent. It’s good but not great.
FIGURE 13.2 Why footprints matter
Now look at the second economy. In that economy, firm 2 produces intermediate inputs for firm 3. Firm 3 purchases x2 inputs from firm 2 and turns them into qx2 units of output. The value added of firm 3 is qx2 − x2 because it consumes the intermediate inputs. Let us imagine that x1 = 3 and q = 2, so the starting value of GDP is still 4, the same as it was in the first economy. The GDP share of firm 3 is still one-quarter. So the second economy looks just like the first. But now imagine that firm 3 becomes 10 percent more productive. You can see that output increases by 5 percent. That’s twice as large as before. How is that possible? It’s because the ratio of sales over GDP of firm 3 is now one-half. Even though its share of GDP is the same—and its share of market value would also be the same—it is now more integrated with the rest of the economy. As a result, improvements in firm 3 matter more in the second economy than in the first.
How should we measure the footprint of a company? It is difficult because it depends on the sector. In manufacturing, the cost of goods sold (COGS) is the right measure. Outside manufacturing, it does not work as well. In finance, it is useless. For lack of a better proxy, I will use employment. The correlation between the share of civilian employment and COGS / GDP in Table 13.2 is 86 percent. It’s not perfect, but it’s good enough for our simple analysis.
Figure 13.3 shows the employment share of the top twenty firms since 1950. The labor footprint of the stars has decreased over time. The recent pickup is only due to Walmart. We have already discussed the efficiency of the US retail sector. It is a particularly competitive and efficient industry. But if we exclude the retail sector, the footprint of the stars actually decreases from 4.5 percent to 2 percent over the past seventy years.
If we zoom in on the top five, the footprint shrinks even more dramatically. Table 13.2 shows that the employment share of the top five has decreased from 2.59 percent to 0.44 percent between the 1950s and 2010s, which is almost six times smaller.
Rise of the Recluse Companies
I define the MV / Emp ratio as the market equity weight of a company relative to its employment weight. To understand the idea, imagine a world where all workers are identical and all firms are equally productive and use the same capital to labor ratio. In such a world, firms would differ only by size, and size could be equivalently measured by number of employees, profits, or market values. Profits and market values would be directly proportional to the number of employees. The MV / Emp ratio would be 1 for all firms.
Of course, in reality, firms differ in the productivity and skill of their employees. Companies have high MV / Emp ratios when they create a lot of market value per employee. This can be because they are capital intensive, technologically advanced, and employ a highly skilled labor force.
FIGURE 13.3 Labor footprint of the stars
When we focus on star firms, we select a group that is likely to have higher productivity than the rest of the economy. They may also hire more highly skilled employees than other firms, and they might use a high share of capital (machines, computers, software). We therefore expect the stars to have high MV / Emp ratios, and indeed this has always been the case. The average MV / Emp ratio for the top firms was between 7.5 and 15 from the 1950s to the 1980s.
Out of the GAFAMs, Amazon looks the most like a regular company. Its market value share is 1.9 percent in 2017, its employment share is 0.37 percent, so its MV / Emp ratio is 1.9 / 0.37 = 5.2, which is similar to General Motors in the 1950s and 1960s.
Starting in the 1990s, however, the MV / Emp ratio starts to increase dramatically. It is above 25 at Microsoft, Apple, and Google. If we look at the GFAM (minus Amazon), we see that these four companies account for 9.3 percent of the stock market but only 0.23 percent of employment. That’s an MV / Emp ratio of 40.8. The most extreme example is, of course, Facebook, with an MV / Emp ratio of 105.8. Facebook hires only highly skilled workers and builds everything in house. It buys essentially nothing from other firms.
Fading Stars
The notion that the biggest tech firms are somehow the pillars of the US economy is false on its face. The defining feature of the new stars is not how much money they make or how high their stock market values are. If we exclude Amazon, the defining feature of the new stars is how few people they employ and how little they buy from other firms. As Larry Page, co-founder of Google, said, “You don’t need to have a hundred-person company to develop that idea.”
Because their footprint is small, whatever happens to the GAFAMs does not matter a lot for the overall productivity of the US economy. If GM’s productivity had doubled in 1960, people would have noticed the difference. Cars would have become cheaper, safer, and more fuel efficient, and the entire supply chain of GM consequently would have become more productive.b If Facebook’s productivity were to double overnight, you would not notice much difference. The ads you see when you browse the app might be better targeted, but no other firm would become significantly more productive as a result.
FIGURE 13.4 Contribution of stars to US growth
Germán Gutiérrez and I (2019a) have studied the fifty-year history of the contribution of stars to overall economic growth in the US. Figure 13.4 shows that the superstars of today contribute less to productivity growth than their counterparts in previous decades: the contribution of superstar firms to US productiv
ity growth has decreased by over 40 percent over the past twenty years.
We define superstar firms as the top twenty firms by market value in any given year (“economy-wide stars”) or the top four firms by market value within each industry (“industry stars”). Stars—or any firm for that matter—can make two contributions to growth: they can increase the productivity of their current workers (within contribution), or they can be more productive in the first place and hire more workers (reallocation contribution). There are theorems along the lines of Box 13.1 that help us do the accounting correctly. We find that the within contribution has collapsed while the reallocation contribution has become quite significant since the mid-1990s. Nonetheless, when we add them up, we get Figure 13.4: stars used to bring about seventy basis points of labor productivity growth each year (using the industry stars definition), but now it’s only forty basis points.
Our results challenge the common wisdom about the stars of the new economy and shed light on the debate between Erik Brynjolfsson and Andrew McAfee (2014), who view digital technologies as “the most general purpose of all,” and Robert J. Gordon (2016), who is skeptical about the impact of recent innovations. Our results are perhaps less surprising for students of history. History students are well aware of biased thinking that today is different—to paraphrase Reinhart and Rogoff (2009)—and that our current stars are exceptional. But there have always been star firms in the US economy, and they have always been large and productive. In our data, we find that today’s stars are no match (so far) for yesterday’s stars.
Ask More from the Stars
Facebook, Apple, Google, and Microsoft are smaller than the star companies of previous decades. When their productivity increases, it has less of an effect than a similar productivity increase at GM once had. Perhaps the battery life of your cell phone improves, perhaps your laptop runs a bit faster, perhaps you can more easily watch a movie in the subway. These welfare gains are meaningful, of course, and economic measurement has to take them into account. But they are not going to move the needle much as far as GDP or life expectancy is concerned.
The Great Reversal Page 27