The Great Reversal Page 12
FIGURE 6.1 Cumulative growth of GDP per capita in the US, the euro area, the EU, and selected EU countries. Source: OECD
Even if one argues that the lack of competition is also bad for innovation—and indeed, I believe this is the case—that effect by itself does not predict permanent differences in growth rates between regions such as Europe and the US that trade and share ideas. In a globalized world, technology flows across countries, and the average growth rates of productivity tend to be similar among advanced economies. As a result, the long-run growth rate of GDP per capita in a particular country may not depend much on the degree of competition in that country. But when domestic competition is weak, many goods and services not exposed to foreign competition will be too expensive.
Figure 6.1, then, is consistent with the idea that the two regions use more or less the same technologies and therefore enjoy roughly the same per-capita growth rates in the long run. There are success stories (such as Germany) and growth disasters (Italy) because of specialization, comparative advantages, and policy choices. The same is true inside the US: some states grow quickly, some struggle. It is simply not true that the US is growing systematically faster than Europe on a per-capita basis.
Fundamental economic similarities in tastes and technologies make Europe a natural comparison group for the US, so let’s analyze Europe in the same way we analyzed the US in Chapters 3, 4, and 5.
Profits and Concentration Are Not Rising in Europe
Figure 6.2 compares profit margins in the US and in the EU. We have seen that this is an important part of the puzzle in the US. Before 2000, the margins are lower in the US than in the EU. After 2000, we see that profit margins increase in the US, while they remain stable or decline in Europe. By the end of our sample, margins are higher in the US than in Europe.
Figure 6.3 shows the evolution of concentration in the US and in the EU over the past fifteen years. The solid line shows increasing concentration in the US. In the EU, concentration is roughly flat, although it appears as slightly decreasing in some data sets and slightly increasing in others.
There is quite a bit of work required to make sure that the data are comparable across regions.a The sample of EU countries includes only ten countries for which we have good firm-level data: Austria, Belgium, Germany, Spain, Finland, France, Great Britain, Italy, Netherlands, and Sweden. Then there is an interesting question: should we treat the EU as one market? Or as ten markets?
If we treat the EU as one market, we get the line with triangles by using one data set and with squares by using another. In these graphs, we start from EU-wide market shares, we compute HHI for each EU industry, and then we take the weighted average of these industry HHIs. For instance, we compute the market share of Peugeot or Volkswagen in EU car production. Then we compute EU HHI for the car industry. We do the same for the pharmaceutical industry, and so on. Then we take the average of these HHIs, weighted by the size of the industry at the EU level. We use the same process for the US.
FIGURE 6.2 Profit margins in the US and EU. Shown are profit rates for the nonagriculture business sector, excluding real estate. The line with circles weighs by EU country × industry gross output. The line with triangles first aggregates across EU countries, within industries, using EU country × industry output as weights, then across EU industries using US industry output as weights. Data source: OECD Database for Structural Analysis (STAN)
If we treat each country as a separate market, we get the line with circles. For example, in assessing the telecom industry in France, we start from the market share of various French service providers: Orange, SFR, Free, and so on. We do the same for business services in France, and then compute French HHIs. We repeat this process for other countries and, finally, take the average across countries. Of course, the level of this measure is higher than that of the previous measures, since it is based on national market shares instead of EU-wide market shares. But you can see that their evolutions over time are quite similar.
FIGURE 6.3 Concentration in the US and in the EU. The figure reports the real gross-output weighted average of absolute changes in an eight-firm concentration ratio (CR) across industries, from 2000. Country series treat each country as an independent market. Aggregate series treat the EU as a single market. To ensure consistency, all CRs follow the EU KLEMS segmentation and are averaged across industries using the US share of sales in each industry and year. CRs are adjusted for database coverage using gross output from OECD STAN. EU concentration includes Austria, Belgium, Germany, Spain, Finland, France, Great Britain, Italy, Netherlands, and Sweden. See Gutiérrez and Philippon (2018a) for details. Data sources: US CR, Compustat. EU CRs, consolidated financials from Compustat (squares) and unconsolidated financials from ORBIS (circles and triangles), using the data of Kalemli-Ozcan et al. (2015)
Which line is the relevant one? There is no simple answer. For cars, EU-wide shares are probably more relevant. For personal services, national shares might be more relevant.
Measuring concentration in Europe is more complicated than in the US. Another data set from the OECD suggests mildly increasing concentration in Europe (Bajgar et al., 2019). They take into account that some firms are part of larger business groups. When they measure concentration at the business group level within two-digit industries they find a moderate increase in concentration in Europe, with the unweighted average CR8 increasing from 21.5 percent to 25.1 percent. In North America, CR8 increases from 30.3 percent to 38.4 percent. Our main conclusion—that concentration has increased in the US more than in the EU—therefore holds regardless of the measure that we use. Moreover, as EU integration progresses, we can expect more intra-Europe competition. Even if national market shares remain constant, the effective concentration is likely to decrease. This would reinforce our conclusion.
TABLE 6.1
Profit Margins and Profit Rates
US
EU
1997–99
2013–15
Δ
1997–99
2013–15
Δ
Operating margin
9%
13%
4%
8%
7%
−1%
Operating profit rate
13%
16%
3%
9%
8%
−1%
Data source: EU KLEMS data for Nonfinancial Corporate Business Sector
There are other data sources and definitions we can use to assess the evolution of competition. Table 6.1 summarizes some of these measures. The profit margin compares profits to sales (revenues). The profit rate compares profits to the stock of capital. These two measures increase in the US between the late 1990s to the present, while they are roughly stable in the EU.
In a paper I co-wrote with Germán Gutiérrez, we presented many more measures as well as adjustments for the cost of capital (Gutiérrez and Philippon, 2018a). In all cases, the indicators point toward an increase of concentration and profits in the US and a stability or a small decrease in the EU.
The Labor Share of Income
This discussion brings us to another controversial topic: the evolution of the labor share. The fundamental idea in economics is that firms combine labor and capital to produce goods and services. Ideas matter as well, of course, but they are embedded in either physical capital (patents, for instance) or in human capital (people’s heads). Firms also use intermediate inputs, but we net them out when we think about value added. For instance, a coffee store needs to buy coffee beans and milk to make cappuccinos. These are intermediate inputs. The value added of the coffee store is the value of cappuccinos minus the cost of milk, coffee, and electricity. The value added is then split between the owners of the capital (machines, tables, real estate) and wages for the baristas. The ratio of wages to value added is called the labor share. If we subtract taxes, it is also 1 minus the capital share.
FIGURE 6.4 US lab
or share. Data source: FRED
The labor share is the ratio of the compensation of labor relative to value added.b Over the past fifteen years, labor has lost 5 percentage points in its share of value added in the US. In Europe, by contrast, the labor share has remained roughly constant.
If we focus on the US, we can study the labor share over a long period, from 1947 to today, with consistent data. Figure 6.4 shows labor’s share of the value added in the US nonfarm business sector. When we teach economics, we use models in which the theoretical value of the labor share is two-thirds.c You can see why we like these models. The labor share has been fairly close to 0.66 for much of the postwar period. In the 2000s, however, it declined and stabilized around 0.6. Labor has lost about five or six points of value added since 2000.
Broadly speaking, there are two possible interpretations of this evolution. The first interpretation is that the competitive capital share has increased because capital has truly become more important in the production of goods and services. This could presumably be explained by changes in technology, including automation, or by international trade flows. If capital becomes more important, it will be compensated relatively more, and labor relatively less.
The second interpretation is that payments to capital include rents. These rents can reflect market power in the market for goods and services (monopoly), or in the market for labor (monopsony). Monopsony reflects the idea that employers have discretion in setting wages because employees have few alternatives. Monopoly rents are the ones we have discussed in previous chapters. It is not entirely obvious that monopoly rents increase the capital share. These rents can also be appropriated by key employees. The finance industry and the health care industry provide examples of this. On average, however, we expect monopoly rents to accrue disproportionately to profits, and thus to increase the capital share.
Computing labor’s share of value added requires a framework for national accounts. To compute the value added we need to take a stand on what are intermediate inputs (expensed) and what are investments (capitalized). To compute the total compensation of labor we need to estimate the wages of the self-employed. All of the assumptions going into the calculations differ across countries, which makes direct international comparison difficult. Fortunately, the KLEMS database project has created comparable measures.
FIGURE 6.5 Labor shares for the market economy. Euro area includes eleven original countries plus Greece. Data source: KLEMS
Figure 6.5 compares labor shares of the market economy in the US and the euro area. The figure focuses on the last fifteen years because this is the period when we observe a large decline in the US, and also because the euro area did not really exist before 2000. Over this period, the US labor share has declined by about five points. It has not declined in Europe, however. In fact, it is exactly the same at the beginning and at the end of the sample. The rise in profits, the rise in concentration, and the decline in the labor share are thus phenomena specific to the US. Since they do not happen in Europe, and since Europe broadly uses the same technology as the US, this casts doubt on the technological interpretation. Since Europe also trades with China and other emerging markets, this casts doubt on the international trade interpretation.
Europe Is Different
Europe offers an interesting contrast to the US, casting doubt on the technology and trade explanations for rising profits. In most sectors, technologies are similar in Europe and in the US. Europe is also exposed to the same trade flows as the US. Yet, it does not have the same increase in profits, increase in concentration, and decrease in labor shares. This suggests that we need to look at explanations based on differences in policy, not in technology or external factors.
Before we do so, however, we would like to be more confident with our diagnosis. Are European markets really more competitive than US markets? When I present my results, people are usually more convinced by direct price comparisons. I think this is fair. After all, if I am right, prices should be lower in Europe. Unfortunately, comparing prices across countries is a lot more complicated than asking, say, if a bottle of Coca Cola is cheaper in Atlanta than in New York. Our next chapter, then, is all about comparing prices around the world.
* * *
a We need to make sure that we have consistent industry classification and that we cover the same sample of firms in both regions. To ensure consistency, HHIs follow segmentation in the EU KLEMS database and are averaged across industries using the US share of sales in each industry and year. We use the top fifty firms in each industry. We use industry data to compute the market shares: share = firm sales / industry sales. We get industry sales from the KLEMS database and firm sales from Compustat for the US and Amadeus for Europe.
b For workers on payrolls, compensation consists of their wages and salaries plus employer contributions to pension and insurance funds and to social insurance. It’s more complicated for self-employed workers since their income includes returns on the business assets that they own. Their labor share is usually imputed assuming that their hourly wage is the same as for payroll employees.
c The slight decrease from the 1950s to the 1980s comes from the decrease in self-employment and the assumption of equal wages. Other methods make the labor share even more stable. The decline after 2000 is clear in all measures. See Elsby, Hobijn, and Sahin (2013) for an excellent discussion.
CHAPTER 7
Are US Prices Too High?
ECONOMISTS SPEND of a lot of time comparing prices. We compare prices over time to estimate real growth rates. We compare prices across countries to estimate differences in living standards.
Here, we want to compare prices to see if a relative lack of competition in US markets has led to an increase in what US consumers pay for goods and services. This is a tough question. The data we are going to use are not specifically designed for this purpose. And, of course, we need a comparison group, so we are going to use Europe again.
To improve our analysis of Europe and the US, we would like to be able to directly compare the prices of goods and services in the two regions. Comparing prices across countries, however, is a lot harder than you might think. Should we expect similar goods to be sold at similar prices in different regions? In the field of economics, this is what we call the law of one price, or LOOP.
The LOOP says that identical goods sold in different countries must sell for the same price (when those prices are expressed in terms of the same currency and after we take into account shipping and distribution costs). Retail prices naturally depend on distribution costs. We know from the literature on international trade that distribution costs are high and that they depend on local wages (and taxes), so we need to be careful when we compare prices across countries.
Suppose that a pair of shoes costs €50 in Brussels, and that the euro / dollar exchange rate is 1.2 dollars for one euro. Then, according to the LOOP, the shoes should cost $60 in Chicago. Why would the LOOP hold? Let us think about how the law would be enforced. If the price in Chicago was more than $60, then you could buy the shoes in Brussels, ship them to Chicago, and sell them there. The profit would be the price difference minus the transportation cost and the cost of reaching customers in Chicago. You can see the issue immediately. If the transport costs are high, then that would not be profitable.
Imagine that the manufacturing cost for the shoes is €20, the shipping cost to Brussels is €5, the distribution cost is €10, and the profit margin is €15. This adds up to the price you pay in the store: 20 + 5 + 10 + 15 = €50 in Brussels.
Now imagine that the transport cost to Chicago is $6, the distribution cost is $15, and the profit margin is $18. The manufacturing cost is $24, as before. The cost of the shoes in the store in Chicago is thus 24 + 6 + 15 + 18 = $63.
You can see that the retail price of a good has four components: one is the manufacturing cost, for which the LOOP should hold. The other components are transport costs, distribution costs, and profit margins. In our
example, the price in the US is higher by $3 because the retail cost is $3 higher, but the profit margin is the same, since at an exchange rate of $1.2 per euro, $18 = €15. In this case the higher cost in the United States reflects distribution costs instead of higher markups.
Of Haircuts and Ferraris
Comparing prices around the world is complicated for two reasons. One reason is practical, the other theoretical. In the Introduction, we discussed the relative prices of cell phone plans and broadband internet between the US and Europe. In doing so we relied on detailed work by industry specialists who worked hard to compare equivalent contracts in various countries. It is not easy to replicate such an analysis for a broad basket of goods and services. That is the practical issue.
But there is another, deeper—and more interesting—issue. If I told you that haircuts cost more in the US than in Cambodia, you would not think that this has anything to do with oligopoly rents in the US. And you would be right. The cost of a haircut basically depends on the wages of hairdressers. According to United Bank of Switzerland (UBS) analysts, who used data from 2015, haircuts for women cost $95.04 in Oslo, Norway; $83.97 in Geneva, Switzerland, but only $4.63 in Jakarta, Indonesia, and $9.27 in Beijing, China. Hairdressers are paid more in rich countries. Hence, haircuts cost more in rich countries than in poor countries. That’s pretty obvious, but it has deep implications, as we shall see. It even has a name: the Balassa-Samuelson effect, after Hungarian economist Béla Balassa and Nobel Prize winner Paul Samuelson, who described the phenomenon in the 1960s.